Sixth Canadian Edition

bel65796_fm_i-xxii i 01/13/22 12:49 PM

 Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries.

Published in Canada by
Oxford University Press
8 Sampson Mews, Suite 204,
Don Mills, Ontario M3C 0H5 Canada

[Link]

Copyright © Oxford University Press Canada 2023

The moral rights of the authors have been asserted

Database right Oxford University Press (maker)

First Edition published in 2005

All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence, or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Permissions Department at the address above
or through the following url: [Link]/permission/permission_request.php

Every effort has been made to determine and contact copyright holders.
In the case of any omissions, the publisher will be pleased to make
suitable acknowledgement in future editions.

Library and Archives Canada Cataloguing in Publication

Title: Social research methods / Edward Bell, Alan Bryman, Steven Kleinknecht.
Names: Bryman, Alan, author. | Bell, Edward, 1955- author. | Kleinknecht, Steven W. (Steven
William), author.
Description: Sixth Canadian edition. | Includes bibliographical references and index.
Identifiers: Canadiana (print) 20210320427 | Canadiana (ebook) 20210320451 | ISBN 9780190165789
(spiral bound) | ISBN 9780190165796 (softcover) | ISBN 9780190165772 (EPUB)
Subjects: LCSH: Social sciences—Research—Textbooks. | LCSH: Social sciences—Methodology—Textbooks.
| LCGFT: Textbooks.
Classification: LCC H62 .B78 2022 | DDC 300.72—dc23

Cover image: Alexander Spatari/Getty Images

Cover design: Laurie McGregor
Interior design: Laurie McGregor

Oxford University Press is committed to our environment.

This book is printed on Forest Stewardship Council® certified paper
and comes from responsible sources.

Printed and bound in the United States of America

1 2 3 4 — 23 22 21 20

bel65796_fm_i-xxii ii 01/13/22 12:49 PM

 Dedication

To my teachers, who continue to shape me.

Edward Bell

To my mom and dad, for always being in

the stands cheering me on.
Steven Kleinknecht

bel65796_fm_i-xxii iii 01/13/22 12:49 PM

Brief Contents
Guide to the Book viii
Special Features of the Book x
Acknowledgments xviii
Preface xx

PART I Fundamental Issues in Social Research

1 General Research Orientations 2
2 Research Designs 29
3 Research Ethics 53

Appendix to Part I An Overview of the Research Process 73

PART II Quantitative Research

4 The Nature of Quantitative Research 78
5 Survey Research: Interviews and Questionnaires 100
6 Structured Observation 139
7 Quantitative Sampling 152
8 Quantitative Data Analysis 177

PART III Qualitative Research

9 The Nature of Qualitative Research 206
10 Ethnography and Participant Observation 225
11 Interviewing in Qualitative Research 249
12 Content Analysis 281
13 Qualitative Data Analysis 314

PART IV Bringing It All Together: Revisioning Quantitative

and Qualitative Research, and Some Practical
Advice
14 Bridging the Quantitative/Qualitative Divide 334
15 Conducting a Research Project 359

Appendix Using IBM SPSS Statistics and NVivo Software 374

Glossary 404
References 412
Index 430

bel65796_fm_i-xxii iv 01/13/22 12:49 PM

Contents
Guide to the Book viii Acknowledgments xviii
Special Features of the Book x Preface xx

PART I Fundamental Issues in Interactive Classroom Activities 72

Relevant Websites 72
Social Research
Appendix to PART I A n Overview of
1 General Research Orientations 2
Chapter overview 2
the Research
Introduction 3
Process 73
Theory and research 4
PART II Quantitative Research
Deductive and inductive approaches 7
Epistemological considerations 10 4 The Nature of Quantitative
Ontological considerations 15 Research 78
General orientations: Quantitative and qualitative Chapter overview 78
research 17 Introduction 79
Influences on the conduct of social research 19 The main steps in quantitative research 79
Key Points 25 Concepts and their measurement 81
Questions for Review (r) and Creative Reliability and measurement validity 86
Application (a) 25
The main goals of quantitative researchers 89
Interactive Classroom Activities 26
Critiques of quantitative research 94
Relevant Websites 27
Key Points 97
Questions for Review (r) and Creative
2 Research Designs 29 Application (a) 97
Chapter overview 29 Interactive Classroom Activities 97
Introduction 30 Relevant Websites 98
Research designs 32
Bringing research orientation and design 5 Survey Research: Interviews and
together 49
Questionnaires 100
Key Points 51
Chapter overview 100
Questions for Review (r) and Creative
Application (a) 51 Introduction 101

Interactive Classroom Activities 52 Open or closed questions? 102

Relevant Websites 52 Types of questions 104

Rules for designing questions 106
Issues related to conducting interviews 116
3 Research Ethics 53
Questionnaires 122
Chapter overview 53
Secondary analysis of survey data 128
Introduction 54
Key Points 135
General ethical principles 56
Questions for Review (r) and Creative
Conclusions 70
Application (a) 136
Key Points 71
Interactive Classroom Activities 137
Questions for Review (r) and Creative
Relevant Websites 138
Application (a) 71

bel65796_fm_i-xxii v 01/13/22 12:49 PM

vi Contents

6 Structured Observation 139 Theory and concepts in qualitative research 210

Chapter overview 139 Criteria for evaluating qualitative research 210
Introduction 140 The main goals of qualitative researchers 215
Problems with survey research 140 Critiques of qualitative research 219
So why not just observe behaviour directly? 140 Some contrasts between quantitative and
qualitative research 221
The observation schedule 141
Key Points 222
Strategies for observing behaviour 142
Questions for Review (r) and Creative
Issues of reliability and validity 143
Application (a) 223
Field experiments as a form of structured
Interactive Classroom Activities 223
observation 145
Relevant Websites 224
Criticisms of structured observation 149
Key Points 150
Questions for Review (r) and Creative 10 Ethnography and Participant
Application (a) 150 Observation 225
Interactive Classroom Activities 151 Chapter overview 225
Relevant Websites 151 Introduction 226
Access 227
7 Quantitative Sampling 152 Roles for ethnographers 233
Chapter overview 152 Field notes 236
Introduction 153 The rise of visual ethnography 238
Sampling error 155 Institutional ethnography 240
Types of probability sample 156 Sampling 242
The qualities of a probability sample 160 The end 244
Sample size 160 Key Points 246
Types of non-probability sampling 164 Questions for Review (r) and Creative
Limits to generalization 170 Application (a) 246
Content analysis sampling 171 Interactive Classroom Activities 247
Reducing non-response 171 Relevant Websites 247
Key Points 174
Questions for Review (r) and Creative 11 Interviewing in Qualitative
Application (a) 174 Research 249
Interactive Classroom Activities 175 Chapter overview 249
Relevant Websites 176 Introduction 250
Differences between structured and qualitative
8 Quantitative Data Analysis 177 research interviews 251
Chapter overview 177 Unstructured and semi-structured interviewing 251
Introduction 178 Focus groups: An introduction 262
A small research project 178 Conducting focus groups 263
Key Points 202 Size of groups and selecting participants 264
Questions for Review (r) and Creative Asking questions and level of moderator
Application (a) 202 involvement 264
Interactive Classroom Activities 203 Group interaction in focus group sessions 268
Relevant Websites 204 Limitations of focus groups 270
Online interviews and focus groups 271
Qualitative interviewing (without immersion in a
PART III Qualitative Research social setting) versus ethnography 273
Key Points 278
9 The Nature of Qualitative Questions for Review (r) and Creative
Research 206 Application (a) 278
Chapter overview 206 Interactive Classroom Activities 279
Introduction 207 Relevant Websites 280

bel65796_fm_i-xxii vi 01/13/22 12:49 PM

 Contents vii

12 Content Analysis 281 Introduction 335

Chapter overview 281 The natural science model and qualitative
research 336
Introduction 282
Quantitative research and interpretivism 338
Personal documents 283
Quantitative research and constructionism 339
Government documents 287
Research methods and epistemological and
Official documents from private sources 288
ontological considerations 340
Mass media outputs 289
Problems with the quantitative/qualitative
Virtual outputs and the Internet as objects of contrast 340
analysis 290
Mutual analysis 342
What things need to be analyzed? 292
Quantification in qualitative research 343
Coding 293
Multi-strategy research 345
Content analysis without a pre-existing coding
Two positions in the debate over quantitative and
scheme 298
qualitative research 346
Readers and audiences—active or passive? 299
Approaches to multi-strategy research 346
Two approaches to the study of language 300
Reflections on multi-strategy research 353
Advantages of content analysis 310
Key Points 356
Disadvantages of content analysis 310
Questions for Review (r) and Creative
Key Points 311 Application (a) 356
Questions for Review (r) and Creative Interactive Classroom Activities 357
Application (a) 311
Relevant Websites 358
Interactive Classroom Activities 312
Relevant Websites 313
15 Conducting a Research
Project 359
13 Qualitative Data Analysis 314
Chapter overview 359
Chapter overview 314
Introduction 360
Introduction 315
Know what is expected by your institution 360
General strategies of qualitative data
analysis 315 Identifying research questions 360
Key Points 330 Interacting with a supervisor 361
Questions for Review (r) and Creative Managing time and resources: Start thinking early
Application (a) 330 about the research area 362
Interactive Classroom Activities 330 Searching the existing literature 364
Relevant Websites 331 Preparing for research 365
Writing up research 366
Interactive Classroom Activities 372
PART IV Bringing It All Relevant Websites 373
Together: Revisioning
Quantitative and Appendix Using IBM SPSS Statistics and
Qualitative Research, NVivo Software 374
and Some Practical Glossary 404
Advice References 412
Index 430
14 Bridging the Quantitative/
Qualitative Divide 334
Chapter overview 334

bel65796_fm_i-xxii vii 01/13/22 12:49 PM

Guide to the Book
Who would benefit from of the book and the way it approaches issues and
reading this book? methods. Since both perspectives are crucial in de-
veloping an understanding of social phenomena,
This book was written for undergraduate students both receive full-blown treatment and analysis.
taking a research methods course in social science The Preface that begins this new edition has
disciplines such as sociology, criminology, social two purposes: to provide an entrée into the world
work, politics, history, and education. It covers a of social research methods, and to make the case
wide range of methods, approaches to research, and that research methods are something to get excited
ways of carrying out data analysis. about. The rest of the text is divided into four parts,
Research methods are not tied to any particular which are followed by an appendix.
nation, and the principles underlying them tran- PART I comprises two scene-setting chapters
scend national boundaries. The same is true of this that deal with basic ideas about the nature of social
book. The original text by Alan Bryman was writ- research, and a chapter on research ethics. It also
ten with the needs of UK post-secondary students includes an appendix that outlines the stages of
in mind, but it was widely adopted in Europe and research.
Canada as well. Feedback from adopters and re-
viewers suggested that the book could be made even • Chapter 1 examines issues such as the nature of
more useful for Canadian instructors and students the relationship between theory and research
through the addition of Canadian and, more broadly, and the degree to which a natural science ap-
North American examples, sources, and research proach is an appropriate framework for the
studies. Edward Bell’s adaptations have preserved study of society. It’s here that the distinction
the qualities that contributed to the book’s initial between quantitative and qualitative research
success—its clarity, comprehensiveness, and pre- is first encountered: the two are presented as
sentation of social research methods in an interna- different research orientations with different
tional context—while expanding on those strengths ways of conceptualizing how people and so-
by incorporating elements that are integral to North ciety should be studied. This chapter also in-
American, and especially Canadian, courses in the cludes a discussion of research questions: what
social sciences. Social Research Methods has always they are, why they are important, and how
offered comprehensive coverage of qualitative, quan- they are formulated.
titative, and mixed methods. The sixth Canadian • Chapter 2 introduces the idea of a research
edition offers more in-depth coverage of qualitative design, along with the basic frameworks
research through the contributions and expertise of within which social research is carried out (ex-
its new co-author, Steven Kleinknecht. perimental, cross-sectional, longitudinal, and
case study designs).
Structure of the book • Chapter 3 deals with research ethics for all
types of social research.
In social research, an important distinction is made • The Appendix to Part I outlines the stages of
between the quantitative and qualitative approaches research in an ideal scenario (with the caveat
to inquiry. This distinction lies behind the structure that real-world research is never quite so

bel65796_fm_i-xxii viii 01/13/22 12:49 PM

 Guide to the Book ix

straightforward). These first three chapters and that is the source of some of the best-known
appendix provide the basic building blocks for studies in social research.
the rest of the book. • Chapter 11 examines the kinds of interview
that qualitative researchers conduct (typically
PART II consists of five chapters concerned with semi-structured or unstructured) and focus
quantitative research. groups, in which groups of individuals are
­interviewed on a specific topic.
• Chapter 4 presents the fundamentals of quan- • Chapter 12 applies qualitative approaches to
titative research and provides the context for content analysis, a method used in the study
later chapters. of “documents” ranging from books, letters,
• Chapter 5 focuses on structured interviewing and newspapers to movies, chat lines, and
and the design of questionnaires. It delves into television shows. It also examines two ways to
how to write questions for both questionnaires analyze language: conversation analysis and
and interviews. It also discusses how to com- discourse analysis.
pose a self-completion questionnaire, using • Chapter 13 explores some approaches to
data from already-completed questionnaires the analysis of qualitative data, including
and interviews. grounded theory and coding.
• Chapter 6 covers structured observation, a
method developed for the systematic observa- PART IV moves beyond the quantitative/qual­itative
tion of behaviour. division to explore what the two approaches have
• Chapter 7 deals with quantitative sampling: in common, how they may complement each other,
how to select a sample and the considerations and how they may be combined in the same research
involved in assessing what can be inferred project.
from different kinds of samples.
• Chapter 8 presents a range of basic non-techni- • Chapter 14 proposes that the distinction be-
cal tools for quantitative data analysis. The em- tween quantitative and qualitative research
phasis is on how to choose a method of analysis may be less fixed than is sometimes supposed,
and how to interpret findings. In order to keep and presents some ways in which they can be
the focus on methodological concepts and in- combined to produce multi-strategy research.
terpretations, formulae are not discussed. • Chapter 15 offers advice on conducting a re-
search project, taking readers through the
PART III presents five chapters on aspects of quali- main steps involved, and provides guidance on
tative research. writing up research, an often-neglected area in
the teaching of the research process.
• Chapter 9 plays the same role for Part III that
Chapter 3 does for Part II. It provides an over- Finally, the Appendix presents an easy-to-access re-
view of the nature of qualitative research and source for successful research.
hence the context for the other chapters in
this part. • The appendix explains how to use IBM SPSS
• Chapter 10 discusses ethnography and partici- Statistics software (SPSS) and QSR NVivo soft-
pant observation. The two terms are often used ware to perform, respectively, the quantitative
interchangeably to refer to the immersion of data analyses described in Chapter 8 and the
the researcher in a social setting, a technique qualitative data analyses discussed in Chapter 13.

bel65796_fm_i-xxii ix 01/13/22 12:49 PM

Special Features of the Book
Several features make this sixth Canadian edition especially helpful to students:

2 Research Designs 51

Key Points
• There is an important distinction between a gen- • Replicability, validity (measurement and exter-
eral research orientation (quantitative versus nal), and the ability to establish causation are
qualitative) and a research design. important criteria for evaluating the quality of
• The nomothetic approach to explanation involves quantitative social research.
discovering general laws and principles, and is • Four key research designs are experimental,
usually associated with quantitative research. cross-sectional, longitudinal, and case study.
• Nomothetic explanations must satisfy three cri- • Except for the experimental design, each one of
teria of causation: correlation, time order, and these can be used with either a quantitative or

2
non-spuriousness. a qualitative research orientation. Experiments
• Qualitative researchers usually take the idiographic are used with a quantitative approach.
approach to explanation, which entails creating a • Threats to the establishment of causation are

Research Designs rich description of a person or group based on the

perceptions and feelings of the people studied.
of particular importance in non-experimental,
quantitative research.

Questions for Review (R) and Creative Application (A)

Criteria for the evaluation of social among 15-year-olds at summer camp. You decide
Chapter Overview research to conduct an experiment, and have followed
A research design is a framework for the collection and analysis of the data that is used to answer research R1 Explain the time order criterion of causation. proper ethics protocols. How could you conduct the
questions. It must satisfy certain criteria, and the form it takes depends on the research questions being a1 A survey researcher finds that people with high experiment? Provide as much detail as possible.
asked. This chapter will discuss four prominent research designs: self-esteem make more money than people with R6 What is a quasi-experiment?
low self-esteem. You are tempted to conclude A6 You want to know whether the legalization of
• experimental and related designs (such as the quasi-experiment) from this that self-esteem influences earning marijuana in Canada will affect national crime
• cross-sectional designs, including those involving survey research and some forms of qualitative power. But can time order be established using rates. How could you use the quasi-experimental
interviewing this design? Explain. method to research this issue? Explain.
r2 What is a spurious correlation?
• longitudinal designs, such as panel and cohort studies and various forms of qualitative research Cross-sectional design
a2 You are at a dance party where a lot of alcohol
• case study designs R7 What is meant by a cross-sectional research
is being served. You abstain from drinking, but
design?
notice that the people with the craziest dancing
A7 How could you use a cross-sectional design to
style are the most likely to go to the washroom
Key Learning Outcomes and vomit. Should you conclude that dancing
determine whether there is an association be-
tween the amount of time spent studying and
Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After crazily induces vomiting? Explain.
grades? Assess the degree to which your method
r3 What is a nomothetic explanation?
completing this chapter, you should be able to: can establish causality.
a3 Come up with a nomothetic explanation for why
1. Discuss how and why some research designs are used to produce nomothetic explanations, while others students sometimes drop out of university. Longitudinal design(s)
are used to come up with idiographic explanations. r4 What is an idiographic explanation? How do R8 Why might a longitudinal research design be su-
2. Explain the structure and logic of the experimental method, and describe how a knowledge of this qualitative researchers produce them? perior to a cross-sectional one?
method can aid in the understanding other research designs. a4 Assume that your best friend just dropped out of A8 How could a qualitative researcher use a longi-
3. Identify the strengths and weaknesses of cross-sectional designs. university. Come up with an idiographic expla- tudinal design to study people active in a local

NEW! List of numbered learning outcomes. Learning

4. Describe the purposes of using a longitudinal design. nation of how that happened. environmental movement?
5. Explain how the case study approach is used in research.
Research designs Case study design
Experimental design R9 What is a case study?
R5
outcomes aligned with main headings and end-of-chapter
How are true experiments able to establish
causal connections between variables? Explain.
A9 Pick a particular case (it can be any person,
group, or event) and explain how a qualita-
tive researcher could study it. Then describe
A5 You want to know whether the amount of time

key points to guide students’ reading.

▲ FatCamera
spent on social media affects loneliness levels how a quantitative researcher could gather

bel65796_ch02_029-052 29 10/29/21 12:40 PM

bel65796_ch02_029-052 51 12/11/21 07:51 PM

5 Survey Research: Interviews and Questionnaires 101

• various prerequisites of structured interviewing, including establishing rapport with the interviewee;
asking questions exactly as they appear on the interview schedule; keeping to the question order as it
appears on the schedule; and recording exactly what is said by interviewees;
• problems with survey research, including the influence of the interviewer on respondents and the
possibility of systematic bias in answers (known as response sets);
• the advantages and disadvantages of the questionnaire compared to the structured interview;
• researcher-driven diaries as a form of survey research; and
• secondary analysis of survey data.

Key Learning Outcomes

5
Survey Research: Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

Interviews and 1.
2.
Identify the advantages and disadvantages of open and closed questions.
Describe the types of questions used in survey research, and outline the functions of each one.

Questionnaires
3. List the rules to be followed in designing survey questions, and explain the rationale for them.
4. Discuss key issues involved with structured interviews, including those pertaining to interview formats
and how researchers should conduct themselves.
5. Summarize the use of questionnaires in social research, including matters such as their pros and cons
Chapter Overview compared to structured interviews, the administration of online questionnaires, how to design a ques-
tionnaire, and how researcher-driven diaries can take the form of a questionnaire.
Survey research is one of the most commonly used data-gathering techniques in the social sciences. Surveys
6. Explain the role of secondary analyses of survey data in social research, and discuss their limitations.
can take the form of either interviews, in which participants are asked questions orally, or questionnaires,
which respondents read and fill out themselves. One of the strengths of survey research is that it allows for
standardization in the asking of questions and the categorization of the answers given. “I never do surveys,” said Anna-Lise as she hung up they do the ‘On a scale from 1 to 5’ thing because
This chapter explores: the phone. they want to compare responses between indi-
• the reasons for the widespread use of surveys, including a consideration of the importance of stan- “Why?” I asked. viduals in a way that is meaningful, or they may
dardization to the process of measurement; “For one thing, they always call at dinnertime, want to put a number on what people in gen-
when I’m tired and hungry.” eral think of the issue, by calculating a mean or
• issues associated with asking questions in surveys:
“Maybe that’s because most people are home something.”
» whether to use open or closed questions at dinnertime,” I said. “They turn phone conversations into numbers?
» different kinds of questions that can be asked in structured interviews and questionnaires “That doesn’t make it any more comfortable. Very weird.”
And they always ask questions in the weirdest ways, “Ever step on a scale to weigh yourself? Hey,
» rules to bear in mind when designing questions
like: ‘On a scale from 1 to 5, with 1 being strongly there’s another kind of scale! Or how about the
» optimal question order dislike and 5 being strongly like, how do you feel numbers that appear on the gas pump when
» how questionnaires can be designed to minimize error and make answering easier for respondents about blah, blah, blah.’ Why can’t they just ask you you’re filling up, or numbers showing the time
» projection or vignette questions in which respondents are asked to reflect on a hypothetical sce- your opinion on something and let you say what in the bottom right-hand corner of your com-

Chapter-opening vignettes. At the beginning of each

nario presented to them you want? That would be way more interesting.” puter screen? Same idea—turning things into
“Well, sometimes they give you a statement to numbers.”
» the importance of pre-testing questions
respond to rather than a question. And I guess “I still say the whole thing is very weird.”

chapter, the topics to be addressed are introduced in an

» the use of questions taken from previous research
• the different ways to do survey research, such as using more than one interviewer, conducting
interviews by telephone, or using the Internet or email to administer questionnaires;
Introduction
informal and provocative way to help students grasp the
questionnaires, which have become familiar aspects
of everyday life. There are job interviews, media
A common method of acquiring data in quantitative interviews, and police interviews, and one is often

real-world relevance of key issues.

▲ JHVEPhoto/Shutterstock
research involves the use of interviews and asked to complete a questionnaire before getting

bel65796_ch05_100-138 100 12/22/21 06:28 PM

bel65796_ch05_100-138 101 12/22/21 06:28 PM

bel65796_fm_i-xxii x 01/13/22 12:49 PM

 Special Features of the Book xi

10 Ethnography and
Participant Observation
Chapter Overview
Ethnography and participant observation require extended involvement in the activities of the people
under study. This chapter explores:
Chapter overviews. Each chapter opener includes
• the problems of gaining access to different settings and ways of overcoming them;
• whether covert research is practicable and acceptable;
an overview that serves as a route map, alerting
• the role of key informants;
readers to what they can expect to learn.
• the different roles that ethnographers can assume in the course of their fieldwork;
• the function of field notes and the forms they can take;
• the role of visual materials in ethnography; and,
• bringing an ethnographic study to an end.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to: 58 PART I Fundamental Issues in Social Research
1. Describe the methodological and ethical issues involved in gaining and maintaining access to field sites
and participants, conducting overt and covert research in open and closed research settings, and work-
ing with key informants.
2. Identify the different roles ethnographers can take on while collecting data, including complete partic-
Oral History Consent Form
ipant, participant-as-observer, observer-as-participant, and complete observer, as well as how active
or passive they will be in these roles. Consent to Participate in an oral history interview being conducted by students enrolled in [course] at
3. Discuss strategies for writing different types of field notes and analytic memos while conducting ethno- ____________ University. This is to state that I agree to participate in a program of research being con-
graphic research. ducted by 45 students under the supervision of Dr. ________ of the Department of _________________
University ([telephone number], email: [email address]).

▲ Chase Clausen/Shutterstock
PURPOSE: I have been informed that the purpose of the project is to explore the history of those parts of
the _________ area that have been demolished or otherwise “lost” to us due to urban change. Students have
formed into teams of 3 to 5 and will be exploring specific sites of memory such as ________, stories of im-
migration and displacement, and the former [site]. We will be working closely with [organization] and the
[organization] which are developing exhibitions on the [site] and [topic] in [year], based in part on these
student projects.
bel65796_ch10_225-248 225 12/22/21 08:38 PM
PROCEDURES: The interview will be conducted at participants’ homes, at facilities provided by ________
University or at another appropriate place. Interviewers will record participants’ life stories using video, or
Examples. In general, undergraduates lack both audio depending on the preference of the interviewee. Participants can choose to discuss any aspect of their
lives and they may refuse to answer any questions. Interviews normally take about 1.5 hours, but participants
may take as long as they would like and are free to stop at any time.
the time and the resources required to carry out RISKS AND BENEFITS: Describing difficult experiences can be upsetting and emotionally difficult. As the stu-
dent projects will contribute to two exhibitions, as well as a website, with your permission, your story will be heard.
a full-scale research project. This makes it all CONDITIONS OF PARTICIPATION: Please review the following conditions and options with the inter-
viewer. Feel free to ask questions if they appear unclear.

the more important to include examples of how ___I understand that I am free to withdraw my consent at any time during the interview and discontinue
from that point forward.

professional researchers have done their work In terms of identification and reproduction of my interview, I agree to (please choose one):
___Open public access: My identity may be revealed in any publications or presentations that may result from
and the lessons they have learned in the process. this interview.
___I agree to the possible broadcasting and reproduction of sound and images of my interview by any method

Most of the major topics discussed in the text are

and in any media by participants of this research project. I consent3 that my interview, or portions
Research Ethics 59
of it, be made
available on the Internet through Web pages and/or online databases of the project.
___I agree that transcripts and/or recordings of my interview will be stored at a local archive for long-term
illustrated with several examples from published preservation.
Experiments are particularly Your interview
problematic when may be accessed camps.
concentration by researchers
It appeared andthat
the ordinary,
public byev- viewing it at the Centre for
it comes to informed _______ and/or at
consent because a local ________-area
researchers eryday people archive
wereholding
willing to theinflict
preservation [Link]
serious harm
research, both in-text and in numbered boxes want the participants OR to behave naturally and au-
thentically. If participants
innocent others if ordered to do so by an authority
___were to be fullyMy
Anonymity: informed
identity willfigure—a
be known finding
only that
to themay have important
interviewer and theimpli-
course instructor; others will
throughout each chapter. regarding the purposenot
be difficult to avoid the
of the
dent
gainexperiment,
access to my
projecteffects
reactive
it would
is completed
cations
identity unless
in [month/year],
that occur
forgain
they
and ultimately
the understanding
special permission
the audio
their and/or
of human
videoalthough
prevention,
atrocities
from me,
recordingthat
the interviewee. Once the stu-
willin-
be destroyed by the student
when people know they(though a copy
are being of theFor
observed. interview
this may be given of
terpretation to his
you).findings has come into question
In cases
reason, experiments usually where
involve family
some photographs
deception. or documents
(Perry, are scanned
2012). Nonetheless, or photographed:
the fact remains that the
For example, Milgram ___(1963)
I agreedesigned
to let theastudent
series of studycopy
researcher involved
familyaphotographs
number of and ethically questionable
documents for use in the student project only.
studies in an effort to understand the brutality that elements. For one thing, the “teachers”/participants
OR
occurred in Nazi concentration camps during the themselves suffered at least short-term harm; most
Second World War. In ___ I agreehetowanted
particular, let the tostudent
know researcher
felt verycopy family photographs
uncomfortable and documents
administering what theyfor use in their assignment
how a person could beand forcause
led to theirextreme
being archived
harm towithbelieved
the interview
to be recording. I likewise
powerful shocks, andgive permission
while the ex- to let future researchers
use these images in their publications.
an innocent human being, and whether being or- periment was in progress many participants came
I HAVE CAREFULLY
dered to do so by an authority figure had anything STUDIEDtoTHE ABOVE
doubt AND UNDERSTAND
the experimenter’s claim thatTHIS AGREEMENT. I FREELY AND
the “learn-
VOLUNTARILY AGREE TO PARTICIPATE IN THIS STUDY.
to do with it. Milgram devised a laboratory exper- ers” were not suffering any serious harm. Some were
INTERVIEWEE:
iment in which a supposedly reputable researcher apparently traumatized by the experience, and ex-
(the authority figure) NAME
asked participants
(please print) to act as pressed serious misgivings about taking part in the
“teachers” who wouldSIGNATURE
punish “learners” by giving study decades later (Perry, 2012).
them increasingly severe electric shocks whenever Another issue was the deception that the partic-
Date and Birthplace
they gave incorrect answers to questions. Neither (optional) ipants experienced: they were told that the experi-
the shocks nor the howls INTERVIEWER:
of pain coming from the ment was investigating how people learn, not how
“learners” were real, butNAMEthe “teachers”/participants
(please print) people come to commit cruel acts, and they were
did not know this. Some of them were further dis- led to believe that the shocks were real. Thus to the
SIGNATURE
advantaged in that they could not see the “learners”: extent that the participants were not fully informed
they could only hear the DATE:
shrieks of pain supposedly about the purpose of the experiment, they were lied
coming from them. If at any time you have questions about your rights
to, although onceasitawasresearch
over participant, please contact _________, Chair
they were debriefed
In all instances, theof“teachers”
the Researchwere Ethics
told that Committee,
the Department
(given previouslyofwithheld
_______,information
_______ University
about the at [phone number] or by
email at [email address].
shocks were part of the study and would not cause study) in that they were informed by the “learner”
any permanent harm, despite the increasingly shrill that the shocks had not really been as painful as he
cries of pain. TheFIGURE
experiment 3.2 A Consent
continued Form had let on, and that he had not been harmed by them.
until the
“teacher”/participant refused
Adapted from to administer
consent form frommoreConcordiaHowever,
University: in keeping with the procedural norms of
[Link]
shocks. In the version of the study in which the
Urban%20Change%[Link] the era, the debriefing provided was rather cursory
“teachers” could not see the “learners,” a substantial and did not fully disclose the deception involved in
majority (62 per cent) of the former administered the the study. In most cases the participant left the lab
strongest shocks possible. Further, in one variant of without being told that the shock machine was not
the experiment, the research volunteers were given real and that the “learner” was an accomplice of the
the option of ordering another person (actually an experimenter rather than a research participant. In
accomplice of Milgram’s) to administer the shocks; some cases, participants were not given a full expla-
in this situation, more than 90 per cent of them or- nation of how the experiment was conducted until
dered the greatest shocks possible. Many of the par- almost a year later (Perry, 2012).
ticipants were visibly upset as they carried out the Most people would agree that in everyday life,
bel65796_ch03_053-076
experimenter’s orders, but they58still did what the au- lying and deception should be avoided. On the other 10/29/21 09:21 AM
thority figure told them to do. hand, the experiment would not have worked had
Milgram saw these results as shedding light on the volunteers known the truth from the outset.
the circumstances that led to the horrors of the Nazi Imagine the outcome if participants had been told,
bel65796_fm_i-xxii xi 01/13/22 12:49 PM
 if, for a variety of reasons, a person’s response not necessarily score high on other dimensions (e.g.,
does not reflect his or her real position on fiscal honesty or continuing education), so that for
the matter examined. For example, someone each respondent one can have a multidimensional
may “agree” with an item or set of items just “profile.”
to appear cooperative or friendly. Multiple However, in much quantitative research, there
indicators can be used to test for this. For in- is a tendency to rely on a single indicator for each
stance, if on a particular item someone agrees concept. This is quite adequate for some pur-
xii Special Features of the Book that unemployment is the most important poses, in particular when one is measuring an
problem facing Canadians today, but later uncomplicated variable such as age. Some stud-
40 PART I Fundamental Issues
also agrees in Social
that Canada Research
has bigger problems ies, like Hay’s (2014) research on secularization
than unemployment, that’s an indication of in Canada (see Box 4.2), employ both single- and
Boxes. Special feature boxes provide response set. In such situations the researcher multiple-indicator measures of concepts. What is
may chose to exclude the person’s responses crucial is that the measures be reliable and valid
in-depth examples of how the various Box 2.3 from Evaluation
the data [Link] representations of the concepts they are supposed
to be gauging.
research methods discussed in the book Dimensions of concepts feasible or ethical to randomly assign research
Coding unstructured data
participants to the two groups, such studies are
have been used in real research situations. As was suggested earlier, the concept of interest may
usually quasi-experimental. For instance, data
have multiple facets or dimensions, which are often Many forms of social science data are essentially
The boxes also list the advantages and gathered from people before a change may be
revealed in the theories and research associated with unstructured and unorganized, including answers
compared with data acquired after; the “before”
it. In developing a measure for a concept, its different to open questions in interviews and questionnaires,
disadvantages of a particular method, people become the control group, the “after”

Adam Melnyk/Shutterstock
aspects or components should be considered. Bryman people theand the content group.
experimental of newspaper articles. To make sense
This approach
summarize important points, discuss and Cramer (2001) demonstrated this approachhas with of the
the added information,
advantage researchers
that the two groupsmust
are go through it
reference to the concept of “professionalism.”basically
The all, same,
the deriving themes
making or categories
random assignmentof behaviour to
methodological controversies, and offer unnecessary.
Such a design was used to evaluate the effect
practical advice. As a researcher studying the effects of a music of a community arts program on the well-being
program, what questions would you ask partic- of older adults in the Vancouver area (Phinney,
ipants before and after the program in order to
Box 4.2 A multiple-indicator measure
gauge its impact on their health? Would you focus
of another
Moody, & Small, concept
2014). Over three years, four
groups of participants took part in the collective
on their objective health indicators, or the subjec-
In Hay’s (2014) study of secularization (seecreation
Box ofrelating
a physical work of artoforreligious
to frequency a perfor-attendance,
tive perception of their own health? What might be
1.3),ofreligious pluralism was measured usingmance
a thatwaswas measured
presented towith
the public.
the Baseline
question, “Do you
the benefit doing both?
single, five-point Likert item that formed quantitative
part measures
currently of well-being
attend church were taken
temple or mosque?”
of an extensive survey of Canadians’ value in the firstRespondents
year of the program, with the same
who answered “yes” were then
A key question asked in evaluation research is measures administered again at its completion.
systems. However, secularization (the depen- asked: “How often?” The response choices
whether a new policy initiative or organizational The researchers also did a qualitative analysis of
dent variable in Hay’s analysis) was measured were: “once a week or more” (given a code of
change achieved its goals. Ideally, to answer that the project by conducting five focus groups at the
with several different indicators in order to tap 5 after reverse-coding), “monthly” (4), “every
question the design would have one group that is conclusion of the program. The quantitative results
into different dimensions of the concept. One few months” (3), “once or twice a year” (2), and
exposed to the treatment—the new initiative—and indicated that the participants had higher levels of
dimension, religiosity, was measured by av- “never” (1). A third dimension, concerned with
a control group that is not. Since it is often not perceived overall health and sense of community,
eraging the responses to three 10-point items the participants’ belief in the religion of their
indicating the importance respondents placed parents, had the response categories “believe
on: (a) “believing in God;” (b) “obeying God, all of it” (4), “-believe most of it but not all” (3),
are ignored (Truth and Reconciliation Commission
doing what he wishes;” and (c) “relating to experiments reveals. some
“believe At the heart
partsofbut
the experimental
disbelieve others” (2),
of Canada, 2015). Evaluation research is only be-
God in a personal way.” A second dimension, design is a logic“don’t
and of comparison
believe which
any ofisit”
used
(1). to iso-
ginning to come to terms with these sorts of issues. late the causal influence of a particular variable—a
Increasing attention is now being paid to the larger goal to which researchers using other designs may
structures of power, both historic and contemporary, also aspire. At the very least, an experiment allows
that give rise to various program designs, and it is the researcher to compare the results obtained from
increasingly recognized that Indigenous ontologies an experimental group with those from a control
and epistemologies must be respected (Shepherd & group. In the case of the Fisher and Ma (2014) exper-
Graham, 2020). iment in Box 2.1, the research compared the effects
of different levels of a child’s physical attractiveness
Significance of experimental
bel65796_ch04_077-099 84 design and distress on the level of empathy felt by an ob-10/29/21 11:29 AM
2 Research Designs 43
Although, as noted at the outset of this section, server. The advantage of such comparisons is that
true experiments are rare in sociology for a vari- they permit a better understanding of the phenom-
ety of reasons, it is important to draw attention to enon in question than would be possible if it were
Methods in Motion Applications to Canadian Society
a significant general lesson that an examination of examined under one condition alone. The argument

Why are some Canadians politically conservative and others liberal?

Bell, Aitken Schermer, & Vernon (2009) did a be- “nature” and “nurture” interact in the formation of
havioural genetic study of political attitudes political attitudes. Methods in Motion: Applications
and behaviours that addressed the issue of why These sorts of studies are uncommon in so-
some Canadians tend to have conservative or ciology, political science, or anthropology, al- to Canadian Society. These boxes
right-leaning attitudes, while others are more though they are conducted more frequently in
liberal or left oriented in their political views.
bel65796_ch02_029-052 40
psychology and the health sciences. They have
highlight recent Canadian 10/29/21
research12:40 PM
that
The methodology used in the study (the details far-reaching implications for the social sciences, illustrates how the methods discussed
of which go beyond the scope of this book) was in particular for disciplines in which biological
a kind of natural experiment comparing mono- influences are seldom examined. The standard in the chapter have been used to
zygotic (identical) twins, who are essentially paradigm in the latter fields of study assumes
genetic clones of each other, with dizygotic (fra- that some combination of socialization, culture, study Canadian society.
ternal) twins, who share only about 50 per cent of and situational factors can explain the origins
the genetic material that varies between human of most attitudes and behaviours. Researchers
beings. The researchers examined how similar using behaviour genetic methods, on the other
the identical twins were in their political attitudes, hand, acknowledge that those sorts of influences
and compared that with how similar the frater- are crucial, but argue that a more complete
nal twins were on that score. If identical twins are understanding of the human condition can be
more similar on the characteristic under examina- achieved by considering both social and biolog-
tion than the fraternal twins, that would indicate ical factors.
that genetic factors or other biological influences If behaviour genetics and other approaches
have some impact on the trait in question. The re- that consider both social and biological influences
sults of the study showed that the identical twins were to be more widely adopted in the social sci-
were in fact much more similar in their political ences, that could have a major impact on the
attitudes than the fraternal twins, which suggests theoretical perspectives that are developed
that genetic factors play a role in the formation and rise to prominence in academe. For exam-
of these attitudes. The study also found evidence ple, a blend of sociological conflict theories with
that social factors (such as educational influences Darwinian evolutionary theory is one possibility
and media effects) affected political orientations, (e.g., Sanderson, 2001). Such approaches have
although it did not test for specific social influ- the potential to promote consilience or the unity
ences. Taken as a whole, the results indicate that of knowledge (Bell & Kandler, 2017).

temporally prior to other variables. For example, The current discussion of the cross-sectional
in a relationship between ethnic status and alcohol design places it firmly in the context of quanti-
consumption, the latter cannot be the independent tative research. But qualitative research can also
variable because it occurs after ethnicity. Ethnicity use a form of cross-sectional design. For example,
still cannot be said with certainty to be the cause, Beardsworth and Keil (1992) carried out a study of
however: it is just a possible cause. Nonetheless, the dietary beliefs and practices of vegetarians. They
even though researchers are unable to manipulate administered “relatively unstructured interviews,”
things like ethnic status or gender, causal infer- which were “guided by an inventory of issues,” with
ences can still be cautiously drawn from cross-sec- 76 vegetarians and vegans (1992, p. 261). The inter-
tional data. views were taped and transcribed, yielding a large

bel65796_fm_i-xxii xii 01/13/22 12:49 PM

 Special Features of the Book xiii

148 PART II Quantitative Research

Research in the News

The effect of social trust on Canada’s reaction to the COVID-19 pandemic
collective action,” Wu explained in a CBC inter-
view (Weber, 2020). In a pandemic, trust in fellow
citizens, politicians, the health-care system, and
perceived outsiders could be affected, he main-
tained. Without a high level of social trust, it’s
hard to get people to conform to public health
sockagphoto/Shutterstock

directives such as those regarding physical dis-

tancing and mask wearing.
Both Wu and Kennedy are embarking on
social research programs involving surveys in an
effort to determine what effect the pandemic has
As a researcher choosing between structured
had on social trust in Canada. So far, their results
Research in the News. These boxes outline a
observation and survey-based measures of be-
suggest that levels of social trust have remained
haviour, which method would help you better
understand the intentions behind participants’ high during the crisis. “By and large, Canadians
social distancing behaviours? are reporting taking a lot of the steps that they’ve
been called upon to take,” Kennedy explained. He
research story that has appeared in a major
Public health during the COVID-19 pandemic
also reported that Canadians are more trusting of
public officials than is the case in other countries.
media outlet, illustrating how social research can
have real impacts on our everyday lives.
may depend on things that go beyond discover- “Canadians are more likely to follow social dis-
ies in medical science such as treatment proto- tancing requirements. People perform better in
cols and vaccines. According to Cary Wu, a York terms of complying with policy matters,” he said
University sociologist, and Eric Kennedy, a his- (Weber, 2020). Both Wu and Kennedy are hoping
torian at York, social trust must also be present that their research will help those in charge of
for a nation to cope with the global epidemic. “If promoting public health in Canada to enhance
people do not trust, it’s very difficult to promote levels of social trust as the pandemic wears on.

Methods in Motion Applications to Canadian Society

Systematic social observation of the police
Schulenberg (2014) describes her experiences and the personal characteristics of the people ob-
in using systematic social observation (SSO) to served affect the phenomenon of interest. Since
study decision-making processes and the use of SSO is a method that yields both quantitative
discretion among members of a Canadian re- and qualitative data it has the potential to pro-
gional police service. With this method, structured vide highly meaningful information on a topic, but 114 PART II Quantitative Research
observation is employed to gather behavioural it is rarely used because of the heavy burdens it
information on the people studied, while partic- places on the researcher.
care for ailing parents, the wife should be the of choices outlined in the scenarios. Often what people
ipant observation techniques are used to collect In Schulenberg’s case, she was interested
one to give up her job. say they would do in a particular situation turns out
field notes and develop narratives that provide in how police officers dealt with the anti-social
detailed descriptions of how situational factors and criminal behaviour they encountered while
Strongly agree _____ to be very different from what they actually do. People
Agree _____ are not necessarily being dishonest—predicting be-
Undecided _____ haviour, whether our own or someone else’s, is often
Disagree _____ more difficult than we would imagine.
Strongly disagree _____
Run a pilot study
The advantage of the vignette over this kind of
It is always desirable to conduct a pilot study
attitude item is that it anchors the choice in a more
before collecting data from respondents: not just
realistic situation and thus reduces the risk of an
bel65796_ch06_139-151 148 10/29/21 05:02 PM to ensure that individual questions operate well
unreflective reply. Finch (1987) also argued that on
but also to ensure that the research instrument as
a sensitive topic like this, some respondents may
a whole is appropriate. Pilot studies may be par-
feel uncomfortable or fear that they will be judged
ticularly crucial in research using questionnaires,
on their replies. The fact that the vignette is about
since there will be no interviewer present to clear
other people (and imaginary ones at that) creates a
up any confusion.
certain distance between the questioning and the re-
Here are some specific uses of pilot studies in
spondent, and hence the chance for a more candid
survey research:
answer. One obvious requirement of the vignette
technique is that scenarios must be believable. • If the study is going to employ mainly closed
Finch also pointed to some limitations with this questions, a researcher can pose open ques-
style of questioning, however. It may be impossible to tions in pilot qualitative interviews and then
establish what assumptions are being made by the re- use them to generate the fixed-choice answers.
spondents about the characters in the scenario (such • Piloting an interview schedule can give inter-
as their age, ethnicity, and number of children at viewers some experience in using it and help
home) and the significance of those assumptions for them develop confidence.
the validity and comparability of people’s replies. It is • If everyone (or virtually everyone) who an-
also difficult to establish how far respondents’ answers swers a question gives the same answer, the re-
reflect their own normative views or indeed how they sulting data are not likely to be useful. A pilot
themselves would act when confronted with the kinds study gives you an opportunity to identify

Practical Tip | Getting help in designing questions

Practical tips. Most chapters include at
When designing questions, be empathetic: put Then try out the questions on some friends
least one set of practical tips on how to yourself in the position of the people who will or classmates, as in a pilot study (see the next
be answering them. This can be difficult, be- section). Ask them to be critical and to consider
approach regular tasks or avoid common cause the questions may not always apply to how well the questions connect to each other.
the person producing them—for example, to a Also, look at some questionnaires and struc-
mistakes. young student doing a survey of retired people. tured interview schedules designed by experi-
Even so, you need to think about how you would enced researchers. Those researchers may not
reply. This means concentrating not just on the have asked questions on your particular topic,
questions themselves but also on the links be- but the way they have asked their questions
tween the questions. For example, do filter ques- should give you an idea of what to do and what
tions work in the expected way? to avoid.

bel65796_ch05_100-138 114 10/29/21 04:16 PM

bel65796_fm_i-xxii xiii 01/13/22 12:49 PM

xiv Special Features of the Book

End-of-chapter questions. Questions designed 4 The Nature of Quantitative Research 97

to test understanding of key concepts have been Key Points

expanded to include both review and application • Figure 4.1 depicts quantitative research as a
linear series of steps moving from theory to con-
• Quantitative research has the following key
goals: measurement, the establishment of cau-
clusions, although this is an ideal from which sality, generalization, and replication.
questions. These different question types have there are many departures.
• Examining the reliability and validity of measures
• Quantitative research has been criticized by quali-
tative researchers. These criticisms tend to revolve

been indicated with R and A icons, and are grouped

is important for assessing their quality. around rejecting the view that a natural science
model is appropriate for studying the social world.

under headings that mirror the structure of the Questions for Review (R) and Creative Application (A)

chapter. The main steps in quantitative research

r1 What are the main steps in quantitative research?
The main goals of quantitative
researchers
a1 Assume you are conducting a study to deter- r4 Outline the main goals of quantitative research-
mine which variables influence poverty. Name ers. What is to be gained by achieving them?
one variable that might have an impact on the a4 Assume you have to design a study that will
chances that a person will live in poverty, and examine why some people vote in national
use it to derive a specific hypothesis that can be elections while others do not. Name two inde-
tested with survey data. pendent variables that would be appropriate
to use in this study. What steps can be taken to
Concepts and their measurement ensure that your findings will be applicable to
r2 Why may multiple indicators of a concept be some larger population, i.e., to people who will
preferable to using a single indicator? not be participating in your study?
a2 Describe three different ways to measure the
concept “homophobia.” The critique of quantitative research
r5 Explain how the techniques used by quantitative
Reliability and validity researchers may produce findings that are not
r3 Name and define four different kinds of measure- applicable to everyday life.
ment validity, providing an illustration of each one. a5 You are using surveys to conduct research on al-
a3 How might a researcher measure the concept cohol consumption among stay-at-home moms.
“satisfaction with university student housing”? Might there be a gap between stated and actual
How could the test–retest method be used to es- behaviour on this topic? Why? What could you
tablish reliability? What drawbacks might there do to reduce or eliminate the gap?
be to the use of this method in the example you
have provided?

Interactive Classroom Activities

1. Divide the class into groups of seven or more. c. Explain how each operational definition has
Each group is to do the following: face validity.
a. Come up with a causal statement that in- d. Explain, giving concrete examples, how con-
volves two variables, for example, “Being a current, construct, and convergent validity
victim of racism affects one’s self-esteem.” could be established for both of the opera-
b. Provide a nominal and an operational defini- tionalizations proposed in point b.
tion for each variable (in this example, “racist e. Explain, giving concrete examples, how the op-
victimization” and “self-esteem”). erationalizations could be tested for reliability.

bel65796_ch04_077-099 97 10/29/21 11:29 AM

52 PART I Fundamental Issues in Social Research

information on the case that is relevant to the Visit our online resource center for free assess-
findings of the qualitative investigator. ment: [Link]

Interactive Classroom Activities

1. This activity involves some friendly competition. the fire for the fire trucks and damage example
Divide the class into small groups. Each group is given above).
given 15 minutes to come up with as many exam- 2. Divide the class into small groups. Each group is
ples of spurious correlations as they can (e.g., the to
correlation between the number of fire trucks at
a fire and the amount of fire damage). Examples
a. pick a social phenomenon or experience of
interest (e.g., sex work, homelessness);
Interactive classroom activities. Appearing
at the end of each chapter, these offer students
given in this chapter or in previous classes are not b. provide a nomothetic explanation for it; and
allowed. After the time is up, a spokesperson from c. provide an idiographic explanation for it.
each group presents the group’s spurious cor-

and instructors the perfect opportunity to put the

relations to the rest of the class, and the instructor The groups are to go into as much detail as
judges whether each one is in fact an example of possible. When that is completed, each group is
paired with another group (A and B, C and D, E
concepts learned in each chapter into practice in
a spurious correlation. The instructor also keeps
a record of all groups’ accepted correlations. The and F, etc.) such that Group A presents its results
group with the most accepted correlations wins. to the rest of the class, while Group B has the task

the classroom.
Once that has been done, the class exam- of providing a critical evaluation (enumerating
ines each correlation individually. The instructor both strengths and weaknesses) of Group A’s
asks for volunteers to name an antecedent third arguments. Then the reverse: Group B presents
factor that might be causing the two variables and Group A critiques. Repeat for Groups C and
to vary together (e.g., it would be the severity of D, E and F, and so on.

Relevant Websites
The National Longitudinal Survey of Children and [Link]/kb/[Link]
Youth site provides information on an ongoing, lon- The Human Genome Project site offers some basic
gitudinal study in Canada. information on this 13-year project and its implica-
[Link]/census-recensement/2011/
ref/92-135/surveys-enquetes/nationalchildren-
tions for understanding human behaviour.
[Link]
[Link]
Genome/[Link]
The Online Dictionary of the Social Sciences site,
The Behavior Genetics Association, an interna-
maintained by Athabasca University, “has 1000 en-
tional organization, provides a forum for the dis-
tries covering the disciplines of sociology, criminology,
semination of behavioural genetic research and
political science and women’s study with a commit-
offers methodological workshops.
ment to Canadian examples and events and names.” [Link]
[Link]
(Websites accessed 4 September 2021.)
The Research Methods Knowledge Base provides
a detailed discussion of various types of experi-
mental design.

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student Study Guide • Flash cards • Videos • Activities

• Student self-quiz • Audio clips • Web links

bel65796_ch02_029-052 52 10/29/21 12:40 PM

bel65796_fm_i-xxii xiv 01/13/22 12:49 PM

 Special Features of the Book xv

1 Research Designs 51

Key Points 138 PART II Quantitative Research

• There is an important distinction between a gen- • Replicability, validity (measurement and exter-
eral research orientation (quantitative versus nal), and the ability to establish causation arewith the question “What did you learn
Key points. Each chapter(e.g., assistedconcludes
that starts designed to measure attitudes toward a contro-
qualitative) and a research design. important criteria for evaluating the quality of
from this exercise?” versial topic suicide, the removal
3. The instructor reads each question, one at a time, of statues commemorating historical figures,
• The nomothetic approach to explanation involves quantitative social research.
withis read,
from Box 5.5. After each question a summary
the class animalof itsetc.)most
rights, significant
and to gather basic socio-de-
discovering general laws and principles, and is • Four key research designs are experimental,
is asked to point out its flaws. All members of the mographic data. Once that has been done, each
usually associated with quantitative research. class
cross-sectional, longitudinal, and case study. are then given two points. Relevant websites. A list
minutes to come up with group puts its questions up on the of
board/screen,
a revised version of the question. Individual stu- and the rest of the class provides a critique. Once
• Nomothetic explanations must satisfy three cri-
teria of causation: correlation, time order, and
• Except for the experimental design, each one
dents
these can be used with either a quantitative
areof
websites offering
called on to produce their revised ver-
or are written on the board/screen. The
sions, which
further information
all groups have had their turn, the groups get
together again and revise their questions to ad-
non-spuriousness. or elaborationdress
class is then asked to comment on whether the
a qualitative research orientation. Experiments
revised versions solve all the problems contained
is provided at the end of
the concerns raised by their classmates.
Each group presents its questions to the class a
• Qualitative researchers usually take the idiographic are used with a quantitative approach. in the original question, or if further revision would
each chapter.
second time, with commentary on how the previ-
approach to explanation, which entails creating a • Threats to the establishment of causation are
be appropriate. ous weaknesses have been corrected. The class
4. The class is divided up into small groups. Each again offers a critique to each group.
rich description of a person or group based on the of particular importance in non-experimental,
group is asked to come up with 10 survey questions
perceptions and feelings of the people studied. quantitative research.
Relevant Websites
Questions for Review (R) and Creative Application (A) This Canadian Election Study site provides access The General Social Survey—Social Identity, 2020,
to structured interviews and questionnaires and the provides the questionnaires used by Statistics
numerical data derived from them for the Canadian Canada to examine how Canadians identify with
Criteria for the evaluation of social among 15-year-olds at summer camp. elections
You decide
of 2011, 2015, and 2019. their social and cultural environments.
research to conduct an experiment, and have followed
[Link] [Link]
R1 Explain the time order criterion of causation. proper ethics protocols. How could you conduct the pl?Function=getSurvey&SDDS=5024
This YouTube video explains how to make an online
a1 A survey researcher finds that people with high experiment? Provide as much detail assurvey
possible.
using Google Docs. The Inter-University Consortium for Political and
self-esteem make more money than people with R6 What is a quasi-experiment? [Link]/watch?v=DEhD0m1fhAQ
Social Research is an international association of
about 700 academic institutions and research or-
low self-esteem. You are tempted to conclude A6 You want to know whether the legalization of
In the General Social Survey on Time Use, 2015 ganizations that “provides leadership and training
from this that self-esteem influences earning marijuana in Canada will affect national crime you can find an example of a re-
Questionnaire, in data access, curation, and methods of analysis
power. But can time order be established using rates. How could you use the quasi-experimen-
search diary that forms part of a larger structured for the social science research community.” Click on
this design? Explain. tal method to research this issue? Explain.
interview. “Find Data” to access over 500,000 digital files for ex-
r2 What is a spurious correlation? [Link] amples of questions used in a wide variety of social
Cross-sectional design pl?Function=assembleInstr&lang=en&Item_ research projects.
a2 You are at a dance party where a lot of alcohol Id=217656#qb218258
[Link]/icpsrweb/ICPSR
is being served. You abstain from drinking, but R7 What is meant by a cross-sectional research
The General Social Survey—Canadians’ Safety, 2019,
notice that the people with the craziest dancing design? contains the questionnaires used by Statistics Canada
The UK Data Service provides a searchable data-
A7 How could you use a cross-sectionalto design to bank of questions used on social surveys in the UK.
style are the most likely to go to the washroom assess perceptions of our criminal justice system and
[Link]
and vomit. Should you conclude that dancing determine whether there is an association
experiencesbe-
of criminal victimization in Canada.
tween the amount of time spent studying and
(Websites accessed 4 September 2021.)
crazily induces vomiting? Explain. [Link]
pl?Function=getSurvey&SDDS=4504
r3 What is a nomothetic explanation? grades? Assess the degree to which your method
a3 Come up with a nomothetic explanation for why can establish causality. More resources are available on Dashboard.
students sometimes drop out of university. Longitudinal design(s) Visit [Link] for:
r4 What is an idiographic explanation? How do • Student Study Guide • Flash cards • Videos • Activities
R8 Why might a longitudinal research design be su-
qualitative researchers produce them? • Student self-quiz • Audio clips • Web links
perior to a cross-sectional one?
a4 Assume that your best friend just dropped out of
A8 How could a qualitative researcher use a longi-
university. Come up with an idiographic expla-
Glossary. Learning a new subject usually involves
nation of how that happened.
tudinal design to study people active in a local
environmental movement?
learning some new terminology, orCase at least new
Research designs

meanings for familiar words and phrases. To study?

Experimental design R9 What
study design
is a case help
Glossary
bel65796_ch05_100-138 138 12/22/21 06:28 PM

R5 How are true experiments able to establish A9 Pick a particular case (it can be any person,
youcausal
keep track between
of new terms and concepts, key and explain how a qualita-
Terms in italic type are defined elsewhere in the Glossary.
connections variables? Explain. group, or event) action research Same as participatory action research. code, coding In quantitative research, codes are the
A5 You want to know whether the amount of time tive researcher could study it. Then describe
termsspent are bolded the first time they appear in
tags used to assign the data on each variable to a cat-
adjacency pair Two kinds of talk activity that are
egory of the variable in question. Numbers are usually
on social media affects loneliness levels how a quantitative researcher could gather
linked together, such as an invitation and a response.
assigned to each category to allow easier computer

each chapter and are defined in the glossary near

analytic induction An approach to the analysis of processing. In qualitative research, coding is the pro-
qualitative data in which the collection of data con- cess in which data are broken down into component
tinues and the hypothesis is modified until no cases parts, which are then assigned names.

the end of the book. inconsistent with it are found.

arithmetic mean What everyday language refers to
coding frame or coding manual A list of the codes to
be used in the analysis of a particular set of data. For
as the “average”: the sum of all the scores divided by answers to a structured interview schedule or ques-
the number of scores. Also known simply as the mean. tionnaire, the coding frame delineates the categories
biographical method See life history method. used for each open question. With closed questions,
the coding frame is essentially incorporated into the
bel65796_ch02_029-052 51 10/29/21 12:40 PM analysis Examination of the relationship
bivariate fixed answers from which respondents must choose;
between two variables, as in contingency tables; hence the term “pre-coded question.”
correlation.
concept A general or abstract idea; a category that
CAQDAS An abbreviation of “computer-assisted (or serves to organize observations and ideas about some
computer-aided) qualitative data analysis software.” aspect of the social world.
case study A research design that entails detailed concurrent validity A type of validity that is tested by
and intensive analysis of either a single case or (for relating a measure to an existing criterion or a differ-
comparative purposes) a small number of cases. ent indicator of the concept to see if one predicts the
causality A connection between variables in which other; one of the main forms of measurement validity.
one variable changes as a result of a change in connotation A term used in semiotics to refer to the
another, as opposed to a mere correlation between meanings of a sign associated with the social context
them. within which it operates: a sign’s connotations are
cells The areas in a table where the rows and col- supplementary to its denotation and less immediately
umns intersect and data are inserted. apparent.
census A count of an entire population; by contrast, a constant An attribute on which cases do not differ;
sample counts only some units of a population. compare with variable.
chi-square test Chi-square (χ2) is a test of statistical constructionism, constructionist An ontological
significance used to establish confidence that a find- position (the antithesis of objectivism) according to
ing displayed in a contingency table can be general- which social phenomena and their meanings are con-
ized from a probability sample to the population from tinually being created by social actors; also known as
which it is drawn. constructivism.
closed, closed-ended question A question in an in- construct validity (1) Same as measurement validity;
terview schedule or questionnaire that presents the (2) a type of measurement validity that is established
respondent with a fixed set of possible answers to by determining whether the concepts being measured
choose from; also called a fixed-choice question. relate empirically in a manner that would be predicted
by relevant theories.
cluster sampling A procedure in which the re-
searcher first samples sets of cases (“clusters”) and content analysis An approach to the analysis of
then samples units within them, usually using a proba- documents and texts that seeks to quantify content
bility sampling method. in terms of predetermined categories in a systematic

bel65796_glos_374-381 374 10/31/21 07:20 AM

bel65796_fm_i-xxii xv 01/13/22 12:49 PM

xvi Special Features of the Book

Online resources edition is supported by an outstanding array

of ancillary materials for both students and
Textbooks today do not stand on their own: they instructors, including an enhanced e-book with
are only the central elements in a complete integrated videos, self-grading quizzes, and
learning and teaching package. Social Research additional resources.
Methods is no exception. This sixth Canadian

Additional Materials

For Students
(Available at [Link]/he/ • FAQs for each part of the research process
SocialResearch6Ce) • List of “Dos and Don’ts” of research
In addition to the enhanced e-book, OUP Canada • Set of vignettes highlighting potential prob-
offers these resources free to everyone using the lems when conducting research
textbook: • Examples of research projects
• Audio clips
• Interactive flash cards for students • Activities
• Self-grading quizzes for students • Videos
• Printable checklists • Web links
• Researcher’s Toolkit

At the end of each chapter, you may notice the logo along with a
list of materials; this will let you know what additional material on
this topic is available on Oxford Learning Link.

bel65796_fm_i-xxii xvi 01/13/22 12:49 PM

 Special Features of the Book xvii

For Instructors
Online Instructor’s Manual edition—summarize key points from each
This revised online resource includes chapter and can be edited to suit individual
comprehensive outlines of the text’s chapters, instructors’ needs.
additional assignments, classroom activities
Online Test Bank
designed to encourage student engagement,
A comprehensive electronic test bank that gives
and teaching aids that will enhance the learning
instructors a wide array of options for sorting,
experience.
editing, importing, and distributing questions—
PowerPoint Slides provides approximately 1500 questions in
Hundreds of slides for classroom presentation— multiple-choice, short-answer, and true/false
newly updated and enhanced for this formats.

How to use this book

Social Research Methods can be used in many be found in the Appendix. Even if the module is
ways. Some instructors, for reasons of time or pref- taught without actual computer applications,
erence, may not want to include all chapters or all exposure to them will reinforce the textual ma-
sections of a specific chapter. Following is an over- terial and may be useful for later work.
view of the major topic areas and the parts of the • The quantitative/qualitative distinction is used
book where they are addressed: in two ways: to organize the discussion of re-
search methods and data analysis, and to intro-
• Wider philosophical and methodological duce some wider philosophical issues that have
issues are discussed at some length in Chapter 1. a bearing on social research. Chapter 1 reviews
Instructors who do not wish to use this contextual the main areas of difference between quantita-
material can largely ignore the chapter, except tive and qualitative research, while Chapter 14
for the section on formulating a research ques- explores ways of integrating the two. If time is a
tion. Those who do want to emphasize issues of concern, the latter chapter can be skimmed.
context should also see Chapter 14. • Writing up research is as much a part of the re-
• Practical issues involved in doing quantitative search process as data collection and analysis.
research are the subject of Part II. Chapter 2 Chapter 15 discusses a variety of issues related
is a useful introduction to this topic because it to writing and should be drawn to students’ at-
maps out the main research designs used in both tention even if it is not discussed in class.
quantitative and qualitative research. • Specific advice on doing a research project.
• Practical issues involved in doing qualita- As we have already noted, the whole book is
tive research are the subject of Part III. Again, relevant to student projects, but Chapter 15 ad-
Chapter 2 outlines the most common designs. dresses this subject directly. If time is a factor, the
• Data analysis is covered in Chapters 8 (quan- Appendix to Part I briefly overviews the stages of
titative) and 13 (qualitative), and a guide to the research for students.
use of computer software for these purposes can

bel65796_fm_i-xxii xvii 01/13/22 12:49 PM

Acknowledgments
The previous edition of this book was written shortly to Kevin Gosine, Brock University; Alicia Horton,
after the death of Alan Bryman, who created the University of the Fraser Valley; Oral Robinson,
original version for a UK readership several years University of British Columbia; Meridith Burles,
ago. Alan was, probably unknowingly, a mentor to University of Saskatchewan; Heather Mair, University
me. Working with him had a major influence on of Waterloo; Efe Peker, University of Ottawa; Deidre
my approach to social research methods and to the Rose, University of Guelph; Rochelle Tucker, Simon
social sciences in general, and he opened my eyes to Fraser University; Sarah Knudson, St. Thomas More
many important issues in the philosophy of science. College at the University of Saskatchewan; Neda
His DNA is still all over this book. James Teevan Maghbouleh, University of Toronto Mississauga; and
produced the first Canadian edition, and his efforts several anonymous reviewers.
are greatly appreciated. I’d also like to thank my students, who over the
This sixth edition of the book introduces Steven course of many years have been the source of much
Kleinknecht as a co-author. Steve’s expertise in- inspiration and influence on how I see the world,
cludes a wealth of knowledge and experience in both as a researcher and as a fellow human being.
doing qualitative research, and he is no stranger to Producing this book would not have been possible
the world of quantitative research methods. Steve’s without the love and devotion of my family. To my
many contributions allowed us to probe more deeply wife Jennifer, children Ted and Angelica, and soon
and thoroughly into a broad range of topics. It was to be daughter-in-law Brooke: thanks for everything.
a pleasure exchanging chapter drafts with him for
Edward Bell
mutual feedback as we went through the process of
creating this latest edition of the book together. My respect and gratitude go out to those I’ve
Amy Gordon did a stellar job launching the proj- worked with on this text and those who have in-
ect as Developmental Editor, and was a valuable formed my understanding of research methods.
sounding board as we presented our ideas for the Given Alan Bryman’s passing in 2017, I did not have
new edition. Mariah Fleetham took over in that role the privilege of working with him on this project.
after the project was underway, and was a resource- I value his strong understanding of the art and sci-
ful and encouraging presence as she oversaw the ence of research which are clearly on display in his
completion of the book. Tara Tovell was an efficient writing. I am thankful that my friend and colleague,
and effective copyeditor, with a keen eye for style, Ed Bell, invited me to join him on this project. I’ve
detail, and contemporary usage. Senior Acquisitions learned much from Ed in working with him at Brescia
Editor Ian Nussbaum always seemed to be in our and now as co-authors. When we began contemplat-
corner when we needed him, providing unflagging ing ideas for the sixth edition, little did we know the
support in tough times. world was headed for a pandemic. Amidst much
Many thanks go to the academic reviewers who social upheaval, we wrote. We are fortunate, though,
provided valuable comments and criticisms, and who to have completed this task with considerably less
filled us in on how their students typically react to the disruption to our everyday lives than so many others
content and features presented in the book. Your de- who have experienced significant hardships over the
tailed assessments have allowed us to make many im- last several months. I am thankful for the frontline
provements and have saved us from going down the workers and health-care professionals whose dedica-
wrong path on more than one occasion. Thank you tion and sacrifice allowed people like Ed and me to

bel65796_fm_i-xxii xviii 01/13/22 12:49 PM

 Acknowledgments xix

continue to do our jobs in relative safety. I am also a great gift. I’ve applied many of the lessons learned
thankful for having learned the craft of qualitative re- through the conference to my writing in this text.
search from great mentors like Billy Shaffir, Dorothy Thank you to those who participate in our studies.
Pawluch, and Bob Prus. In 2001, they introduced me The time you share not only teaches us about soci-
to the professional side of academia by way of the ety, it also provides us with a practical understand-
Canadian Qualitative Analysis Conference. My life ing of the research process. These insights have been
and career have benefited immensely from partic- invaluable in my writing and teaching. Thank you to
ipating in and co-organizing the conference with my students, who inspire me with their energy, en-
wonderful colleagues like Tony Puddephatt, Carrie thusiasm, and curiosity for learning about the social
Sanders, Lisa-Jo and Jeff van den Scott, and Deana world. A special thank you to my wife, Mandy, and
Simonetto. Planning, laughing, and problem-solv- our children, Lincoln, Lauren, and Darren. Your love
ing alongside people I enjoy working with has been and support mean the world to me.

Steven Kleinknecht

bel65796_fm_i-xxii xix 01/13/22 12:49 PM

Preface
Why Study Research understanding where social knowledge comes from
will help you distinguish between valid claims and
Methods? fatuous ones.
At this point in your life you probably haven’t devoted There are some answers to the “Why are you
much thought to social research methods. In all like- here?” question that we seldom get from students.
lihood you are reading this book in conjunction with Consider the following. The acquisition of knowl-
the first methods course you have ever taken. That edge through research always goes beyond coming
means you are about to experience something new. up with new ideas or getting more information or
Since you are going to expend considerable time and developing novel ways of looking at society, how-
effort doing something you’ve never done before, it ever important those things may be. To produce or
would make sense to pause for a moment and ask acquire knowledge is a political act. It is tied in with
yourself the following existential questions: Why am the exercise of power. Social research always has po-
I here? Why should I read this book? Why should I litical ramifications, because it always implies, subtly
study social research methods? What’s the point of it or not so subtly, that some ways of organizing society
all? Pondering those questions will make what is to are better than others. Some researchers are happy
come much more meaningful for you. to leave their political involvement at that level: an
So why are you here? When we ask our students implication in their work that certain social struc-
that question (usually on the first day of classes) tures or practices should change or be preserved.
they often tell us that a solid background in social Others go so far as to engage in advocacy, taking part
research methods is indispensable if they are to de- in public campaigns to persuade governments and
velop a sophisticated understanding of the topics the public at large that some sort of action should be
they are passionate about. taken. Whichever position the researcher takes, the
What are you passionate about? Does it matter to politics of research cannot be avoided. What varies is
you that just under half the world’s population strug- how loudly the trumpet is sounded, which is a matter
gles to survive on less than $5.50 per day (World for the individual researcher to decide. Debates
Bank, 2018), while others are so wealthy that they about the wisdom of social researchers becoming
can’t relate to ordinary people? Do you have an in- advocates are part of the storied history of the social
terest in gender equity, environmental sustainabil- sciences, some of which will be told in this book.
ity, crime, single parenthood, ethnic tension, cyber We often hear that “knowledge is power,” and in
bullying, racism, the living conditions of Indigenous many ways that is true. Dictators know this best;
people, changing notions of the family, sexual they all try to limit access to knowledge in order to
mores? In order to come up with informed, thought- preserve their control over others. In fact, the pow-
ful analyses of these and other social issues, and to erful in any society may try to inhibit free inquiry
be capable of evaluating the claims made by others and the flow of ideas. But the “knowledge is power”
on these topics, it is crucial to be familiar with the maxim leaves several questions unanswered. What
various research methodologies used in the social is knowledge? How can it be acquired? How are we
sciences to investigate these subjects. Just as learn- to tell the difference between a sound idea and one
ing how clothes are made can help you tell the differ- that should be ignored? How can we gather informa-
ence between a good pair of jeans and a shoddy pair, tion that will help us understand our subject matter,

bel65796_fm_i-xxii xx 01/13/22 12:49 PM

 Preface xxi

and what sorts of information should we seek? How Another way in which methodology can be power
should that information be analyzed and evaluated? is that it can help to make you a producer of ideas.
Are there some things that we will never fully under- Start thinking of yourself not only as someone who
stand, regardless of how hard we try? takes in ideas and evaluates them, but also as someone
This book explores a variety of answers to those who projects ideas onto the world. Start thinking of
questions, which will make it somewhat different yourself as a researcher: someone who has the poten-
from your other readings and textbooks. In most tial to examine the human condition and contribute
courses in the social sciences, students spend their to our understanding of it. That’s far more interesting
time absorbing, and sometimes challenging, claims and impactful than just reading other people’s work,
made about the social world. For example, you may although reading is an essential part of the process.
learn that men are more likely to commit violent When can you start being a researcher? At vari-
crimes than women. You may come to understand ous points in your university career you will proba-
what it feels like to be homeless. Or perhaps you’ll bly be called on to do some small research projects.
encounter a theory explaining why some countries The knowledge you will gain by learning about social
are rich and powerful while others are not. That research methods can put you at an advantage with
sort of endeavour—learning about and reflecting that sort of assignment, and it could even be an
on claims to knowledge—is clearly worthwhile, but entrée to more sophisticated work if you decide to go
it will not be our central concern. Instead, we will on and develop your research capabilities in gradu-
focus on how those types of claims are generated ate school or in some other way.
and how they gain acceptance as knowledge. In Although it is important to acknowledge that
other words, instead of exploring the body of facts, social research and its attendant methodologies have
concepts, theories, and interpretations that social political ramifications, we shouldn’t lose sight of
scientists use to make sense of the world, we’ll look the fact that it is the height of folly to pursue one’s
at how all those things are acquired or created. One political goals with bad social science. To reduce
of the most profound questions that can be asked of suffering and promote human flourishing we need
someone making a claim to knowledge is, “How do rigorous social research, which is dependent on
you know that?” That simple question is at the heart sound research methods. Learning how to take good
of this book. And like so many simple questions, it field notes, getting instruction on how to construct
has no simple answer. effective survey questions, using appropriate sam-
To produce knowledge, we need a methodology— pling procedures, and a host of other methodologi-
a way to get it. And if you accept the premise that cal issues are not trivial matters. In fact, doing those
knowledge is power, then learning about research sorts of things properly are prerequisites to acquir-
methods means learning about how to become pow- ing the knowledge needed to achieve one’s broader
erful, or at least more powerful. You could even say social goals. Without good methodologies we have
that “methodology is power,” and you wouldn’t be very little to offer that is not available elsewhere.
far off the mark. If you want to learn about how knowledge is cre-
How could methodology be power? For one thing, ated in the social sciences, or if you aspire to make
if you know how knowledge is generated, you won’t an impact on the world by doing social research
be fooled as easily as someone who doesn’t know. yourself, reading this book is a good way to start.
Another way to put this is that learning about re- We hope it will at least pique your interest in how
search methodologies is an excellent way to enhance social researchers do what they do, and what the im-
your critical thinking skills—skills that are essential plications of social research are. To paraphrase Gore
in any situation where power is exercised, which is Vidal: the more one learns, the more interesting con-
everywhere. sciousness becomes.

bel65796_fm_i-xxii xxi 01/13/22 12:49 PM

bel65796_fm_i-xxii xxii 01/13/22 12:49 PM
 PART I
Fundamental Issues in
Social Research

The first two chapters of this book will acquaint you with a number
of fundamental methodological concepts that feature prominently
in the social sciences, in particular the notion of a general re-
search orientation and the idea of a research design. In Chapter 1,
two general orientations are identified—the quantitative and
qualitative approaches—along with a variety of considerations
that affect the practice of social research as a whole. Chapter 2
presents various kinds of research design and identifies some cri-
teria used to evaluate social research. Together these chapters
provide the basic conceptual building blocks that will be drawn
upon throughout the book. Chapter 3 features a discussion of re-
search ethics, which are a primary concern for every kind of social
research and affect every stage of the research process. Part I
concludes with an appendix that offers an overview of how social
research is done.

bel65796_ch01_001-028 1 01/13/22 12:09 PM

 1 General Research
Orientations
Chapter Overview
The aim of this chapter is to examine the fundamental assumptions upon which social research is based.
An important distinction commonly drawn by practitioners of social research—between the quantitative
and qualitative approaches—is explored in relation to those considerations. We will consider:
• the relationship between theory and research—in particular, whether theories and the hypotheses
derived from them are tested by gathering data (the deductive approach) or whether data gathering
is used as a means to create theory (the inductive approach);
• epistemological issues, such as whether a natural science model like the one used in physics or biol-
ogy is suitable for the study of the social world;
• ontological issues, such as whether the social world should be regarded as a reality external to in-
dividuals over which they have little or no control, or as something that social actors may fashion
into their personal realities;
• how values and practical issues impinge on the research process; and
• how these issues relate to both quantitative and qualitative research; a preliminary discussion, fol-
lowed up in Chapter 14, suggests that although the quantitative and qualitative orientations are dif-
ferent, they complement each other.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Explain the relationship between theory in the broad sense and data gathering; in particular, you should
be able to recognize that the purpose of collecting data may be to test theories or to create theories.

▲ LeonWang/Shutterstock

bel65796_ch01_001-028 2 01/13/22 12:09 PM

 1 General Research Orientations 3

2. Discuss the positivist, interpretivist, and critical approaches to social science, including their key assump-
tions regarding the purposes of social research and how it should be conducted.
3. Describe the distinctions between qualitative and quantitative research.
4. Identify how different factors, including values, politics, and issues related to the research question(s)
may influence social research.

Soon-Yi wants to find out why Indigenous people Soon-Yi wonders about the ethical implica-
in Canada are more likely to live in poverty than tions of each of these approaches. She also puz-
other Canadians, but doesn’t know where to begin. zles over what she should do if it turns out that
Should she start by examining the history of colonial- Indigenous people take positions on her research
ism and conflict between Indigenous peoples and topic that differ significantly from her own. Should
settler-colonizers, familiarizing herself with issues
­ she try to determine whose perspective is more
such as disputes over land claims and treaties, res- realistic, and if so, how could she do that? Can
idential schools, or anti-Indigenous prejudice? Or those sorts of issues ever be resolved? Are there
how about gathering aggregate data on present “right” and “wrong” answers on such matters? As
conditions like residence patterns, economic activ- she ponders these issues, she begins to reflect
ities, the age structure, or educational trajectories? on the nature of social reality itself. Is it, as many
The list of topics she could investigate seems endless. social researchers put it, “out there,” indepen-
Maybe rather than beginning her study by ac- dent of the person who perceives it, or is it socially
cumulating information, it would be better to start constructed?
out with some hunches and then gather data to see Then there are the more practical questions:
whether they are supported by evidence. For ex- What, if anything, can be done about the issue of
ample, perhaps the discrepancy in economic con- poverty among Indigenous Canadians? Can soci-
ditions is a manifestation of a centuries-old world ety be changed so that Indigenous poverty is elimi-
system of dominance and exploitation. Similarly, nated, or at least reduced? Considering the matter
it could have arisen through a clash of values and more broadly, are the social forces that affect our
cultures. Then there is the question of how to gather lives so strong that we have no choice but to con-
the information to address those theories. Should form to them, like leaves in a windstorm?
she consult mainly historical sources? How about A great many social thinkers and research-
living in an Indigenous community for a while and ers have confronted the same issues that Soon-Yi
recording her interactions and observations? Or is contemplating. In fact, there are a number of
would it be better to stay where she is and select a rich intellectual traditions that can be drawn on to
few Indigenous individuals for interviews in which make sense of these matters and to assist people in
they could share their life stories? Or maybe she making the kinds of methodological and practical
should design a detailed questionnaire on prej- decisions that she has to make. In this chapter you
udice and send it out to large random samples of will be introduced to that body of ideas and the de-
Indigenous and non-Indigenous people? bates surrounding them.

Introduction appropriate, and how to use them. But the practice

of social research does not exist in a vacuum, sealed
This book is about social research. It would be easy off from philosophical and political debates and con-
to “cut to the chase” and get on with explaining what tested assumptions. As we will see, the explanations
the various research procedures are, when each is of social phenomena that scholars offer and their

bel65796_ch01_001-028 3 01/13/22 12:09 PM

4 PART I Fundamental Issues in Social Research

choice of research methods often depend on the po- research is personal experience (Lofland & Lofland,
sitions they take on those issues. 1995). Sugiman (2004), who is a “sansei” or third-­
The way in which social research is done may generation Japanese, examined Japanese-Canadian
also be affected by what is motivating the re- women’s experiences of internment during the
searcher to conduct the study. A variety of motives Second World War after hearing about the personal
can come into play. Quite often the goal is to assess histories of her family and friends.
the adequacy of a particular social theory, such as Regardless of the motivation for doing research,
Ellis et al.’s (2020) use of minority threat theory the data gathered are usually viewed in relation to
to make sense of Black Muslim immigrants’ inter- theories of some kind. That’s because theories are an
actions with police forces in Minneapolis, Boston attempt to “make sense of it all,” to find order and
and Toronto. The study found, consistent with meaning in a seemingly infinite mass of informa-
the theory, that Somali immigrants perceived tion. How is that done?
the negative treatment they received from the
police to be related to their “racial,” religious, and Theory and research
ethnic status rather than to their actions. In other
cases, the aim is to gather information to create The connection between theory and research is
theories; for example, a researcher may pose as not straightforward. There are several issues at
a street person to find out how the homeless are stake here, but two stand out: first, the form of the
treated by the public, as Orwell did in Paris and theory; and second, the relationship between data
London in the 1920s (Orwell, 1933). Sometimes and theory.
simple “fact-finding” or exploratory work is car-
ried out. For instance, Milgram’s (1963) famous Degree of abstraction
study of obedience was done partly to see how The term “theory” is used in a variety of ways, but its
far participants would go in obeying an author- most common meaning is an explanation of observed
ity figure’s commands, and the results were as- regularities or patterns. For example, one could try to
tounding—many people appeared to be willing to come up with a theory to explain why schizophre-
inflict severe pain on innocent others. Similarly, nia is more common in the working class than in
Bell’s (2007) study of the western Canadian sepa- the middle class, or why more men than women are
ratist movement was motivated in part by a desire alcoholics. Theories are composed of interrelated
to know how much public support the movement and usually verifiable statements or propositions.
had, and it produced evidence that the movement The statements and propositions come in varying
was more popular than previously thought. forms, and different types may be combined in the
In other instances, research is driven by what is same theory. Here are three common components of
seen as a pressing social problem. In fact, the disci- a theory:
pline of sociology came into being in the eighteenth
and nineteenth centuries partly as a way of under- 1. Definitions specify what the key terms in the
standing the social crises and societal upheavals theory mean; for example, “Attachments are
of the day, and that tradition has continued to the stable bonds between people who are fond of
present time. Nagra and Maurutto (2016), for ex- each other.”
ample, investigated the difficulties young Canadian 2. Descriptions outline the characteristics of
Muslims experience at border crossings and air- the phenomena of interest; for example,
ports, while Lyon and Frohard-Dourlent (2015) “Attachments to parents and other family
explored how same-sex partners in common-law members tend to decline in adolescence. Later,
relationships in Canada experience the relatively often after age 25, new attachments develop as
new socio-legal environment in which formal a person marries or lives with a romantic part-
marriage is an option. Yet another stimulus for ner, becomes a parent, or gets a steady job.”

bel65796_ch01_001-028 4 01/13/22 12:09 PM

 1 General Research Orientations 5

3. Relational statements connect two or more evidence. For instance, Durkheim’s (1897/1952)
variables, so that knowing the value of one theory of suicide, which maintains that suicide
variable conveys information about the other; is a function of the level of social integration, is
for example, “As the proportion of people aged a theory of the middle range. One way to test it
15–25 decreases, the crime rate also decreases.” would be to compare suicide rates for married
Relational statements come in two forms: people with those for single, divorced, or widowed
a. deterministic, which means the two vari- individuals. Merton’s (1938) anomie theory, which
ables always go together in a particular way; suggests that crime is more common when a so-
for example, “as the number of people aged ciety instills a desire for wealth in everyone but
15–25 decreases by 1 per cent, the crime provides insufficient means for all to achieve it, is
rate also shrinks by 1 per cent.” If research another theory of the middle range. Such theories
uncovers an instance in which the variables represent attempts to understand and explain a
are not related in this way, the relational limited aspect of social life.
statement must be modified. Grand theories, by contrast, are general and
b. probabilistic, which means the two vari- abstract. They include theories such as structural-­
ables go together with some degree of regu- functionalism, symbolic interactionism, critical
larity, but the relationship is not inevitable; theory, post-structuralism, feminism, and so on.
for example, “regions of the country where Grand theories generally offer few direct indications
the number of people aged 15–25 is decreas- of how to collect evidence to test them, but they pro-
ing are more likely to experience a decline vide ways of looking at the world that can be the
in the crime rate than regions in which the inspiration for a wide variety of research programs.
number of people in that age category is For example, standpoint theory was developed by
increasing.” Here, finding a case that does Dorothy Smith (2004; 2005) and others from a gen-
not fit the pattern does not mean that the eral feminist perspective. This theory maintains that
theory must be modified; this could simply the way we view the world and make our way in it is
be one of the times when the variables are largely determined by our placement in various hier-
not related in the usual way. archies of status and power. One offshoot of stand-
point theory has been an increased willingness to
There are different types of theories. One distinc- examine the views and perceptions of poor or mar-
tion that is sometimes made is between theories ginalized groups in society. Similarly, Box 1.1 shows
of the middle range (Merton, 1967) and grand how an abstract theory like Giddens’s structuration
theories. The former are more limited in scope, theory (1984) can be applied to a specific issue and
and can be tested directly by gathering empirical yield some important insights.

Box 1.1 Grand theory and social research

Giddens’s (1984) structuration theory attempts to empirical research by Neves, Waycott, & Malta
bridge the gulf between notions of social structure, (2018), who examined the use of digital technol-
which is said to shape or constrain human activi- ogies among older adults (65+) in Canada and
ties, and agency, which refers to freedom of action Australia. The authors were trying to determine
in social life. This theoretical issue is explored in whether there was any validity to the common

(Continued)

bel65796_ch01_001-028 5 01/13/22 12:09 PM

6 PART I Fundamental Issues in Social Research

idea that older people are averse to the use of A major motivator was that in many cases it
digital technologies, and whether adopting these proved successful in reducing participants’ lone-
technologies could enhance seniors’ connected- liness and boredom, which led to a desire to keep
ness with friends and relatives and in so doing help using the app beyond the period of the study.
to reduce loneliness and feelings of isolation. The social structures that had an impact on
The researchers used a mixed-methods design the use of digital technology in some cases facil-
(see Chapter 14) that included semi-­structured itated its adoption but in other cases hindered it.
interviews, focus groups, attitude scales and The willingness of relatives to assist in the learn-
field observations. The participants were frail ing process was an important factor in making it
institutionalized older adults (Canadian sample) possible for people in the study to use the tech-
and similarly indisposed older adults living with nology. Some non-institutionalized participants
assistance in the community (Australian sample). stated that geographical isolation from their
Each participant was given a portable electronic close relatives prompted them to use the app,
device and was encouraged to use a communi- while one resident in a care home limited her use
cations app that was installed in it. The results in- of the app when other residents were around
dicated that the common idea of an age-based because of feelings of distrust and alienation to-
digital divide did not adequately describe the wards them. In keeping with structuration theory,
social complexity surrounding the use of tech- agency was described as taking place within the
nology among people in this age group. While context of these sorts of social structures, and as
many participants required assistance to learn evolving through a complex interplay between
how to use the app, most did use it, although with personal perceptions and desires on the one
varying degrees of sophistication and creativity. hand, and the social context on the other.

Although theory plays a crucial role in the social possible to tell whether the authors had the concept
sciences, not all studies make reference to it. For ex- in mind when they collected their data, their book
ample, some qualitative writers focus on providing offers real-life examples of Goffman’s (1963) notion
a rich description of the experiences of a group of of “stigma” and the way stigmatized individuals,
people without trying to come up with a compre- in this case sex workers and their clients, manage
hensive theory that would explain those experiences. a spoiled identity. Their analysis also sheds light on
Nonetheless, as Box 1.2 shows, social scientists are Hochschild’s (1983) concept of “emotional labour,”
often under pressure from their peers to relate their a term she coined to refer to what flight attendants
work to theories of some kind. do when they feign friendliness in order to deal with
As Box 1.2 indicates, some social scientists will difficult passengers. Similarly, several other non-­
reject research that has no direct connection to theoretical studies provide data that could eventu-
theory in either the grand or the middle-range sense ally be used to evaluate or devise a theory.
of the term. However, non-theoretical work can pro- Our discussion of what theory is and its impor-
vide insights that are useful or revealing in their own tance invites consideration of another question:
right. McKeganey and Barnard’s (1996) research on What is the relationship between theory and re-
British sex workers and their clients is a case in point. search? Up to this point we have focused primarily
The authors related their research findings to investi- on how theory can guide research, in particular on
gations of people in the sex industry in several other how the collection and analysis of data can be used
countries, and what they describe offers good illus- to test theories. But this notion of research as essen-
trations of ideas that form an important part of the tially “theory testing” does not provide a complete
sociologist’s conceptual toolkit. Although it is not picture of what social scientists do. Theory may also

bel65796_ch01_001-028 6 01/13/22 12:09 PM

 1 General Research Orientations 7

Box 1.2 The need for theory

Teevan and Dryburgh (2000) collected data from
57 Canadian male adolescents concerning their
participation in deviant activities, such as truancy,
theft, vandalism, and fighting, and asked them why
they acted the way they did. Then the boys were read
specially adapted sociological explanations of such
behaviour and asked what they thought of them. In
effect, they were asked to evaluate the sociologists’
theories as if they were sociologists themselves. The
idea was to give marginalized people a voice and
to see the issue in question from the point of view of
those studied—two common approaches in quali-
tative research. Some journal reviewers liked the
idea but thought that the findings’ relevance for so-
ciological theory needed to be explored. Eventually
the authors brought a variety of social theories to
bear on the data they had gathered, and in doing
so provided a critical assessment of existing expla-
nations for delinquency and conformity.

A young person painting graffiti on a wall. Some

call this behaviour vandalism, while others consider
it art. What social theories might you invoke to ex-
sunflowerey/123RF

plain what this person is doing? How might the young

person explain it? Do you think his explanation would
have anything in common with your social theories?

follow upon or arise from the collection and analysis by coming up with a theory that seeks to explain a
of data. Here we begin to see two different ways to go particular phenomenon, and then deduces specific
about acquiring knowledge: the deductive and induc- hypotheses from it that are tested with empirical
tive approaches. data (see Box 1.3 for a concrete example) and then
are either confirmed or rejected. In the case of the re-
Deductive and inductive search described in Box 1.3, which examined several
theories simultaneously, some hypotheses were sup-
approaches ported while others were not. If the data gathered do
The deductive method is the most common approach not support the researchers’ hypotheses, the theory
to social research, although by no means the only one. may have to be revised or rejected.
The sequence of steps taken in deductive research is It is also important to bear in mind that when
depicted in Figure 1.1. Note that the researcher starts the deductive approach is put into operation, the

bel65796_ch01_001-028 7 01/13/22 12:09 PM

8 PART I Fundamental Issues in Social Research

researcher does not necessarily follow the exact Deductive approach

linear sequence shown in Figure 1.1. For example, Theory
a new hypothesis may come to mind during the
­data-gathering stage; or the relevance of the data for
a second theory may become apparent after the data Observations/Findings
have been collected. Although the sequence outlined
in Figure 1.1 is a common one, it is only a general
model, one that is not strictly followed in all cases.
Inductive approach
In fact, some research makes no attempt to follow
the sequence outlined in Figure 1.1. Some investiga- Observations/Findings
tors prefer an inductive approach in which theories
and interpretations are the outcome of research. In
inductive social science, the researcher begins not by Theory
coming up with a theory to be tested, but by gather- FIGURE 1.2 Deductive and inductive approaches
ing or examining data relevant to the phenomenon to the relationship between theory and research
being investigated; see Figure 1.2, which illustrates
the difference between induction and deduction. The
Research in the News box on page 10 provides an ex-
ample: Bikos (2017) offered her interpretations and comes first, and the effort to create concepts and
conclusions regarding police culture after she had theories out of it comes later. After some theoretical
gathered qualitative information on the topic. reflection a researcher may decide to collect more
When the inductive method is used, data are data to establish the conditions under which the
gathered not to test a theory, but to come up with the newly developed theory does or does not hold. This
information required to construct a theory or inter- strategy of moving back and forth between data and
pretation. That’s why with induction, data gathering theory is often described as iterative. The practice
of deriving theories from qualitative data is some-
11. Theory times referred to as grounded theory, discussed in
Chapters 9 and 13.
In actual research situations it is impossible to
2. Hypotheses
conduct a study that is purely deductive or purely in-
2
ductive. Just as deduction always entails an element
of induction (theories do not emerge from a pristine
mind unaware of previous findings) so the inductive
3. Data collection
3 process always entails a modicum of deduction (no
researcher will be totally unaware of theories and
perspectives that might be applicable to the phenom-
4 Findings
4. enon he or she is observing). Often some combina-
tion of both can be found in the same research.
Although some researchers using induction un-
5 Hypotheses confirmed or rejected
5. doubtedly try to develop theories, sometimes the
results of their research are little more than empir-
ical generalizations, however useful they may be.
6 Substantiation, revision, or rejection
5. An example is a Statistics Canada finding that the
of theory number of people in same-sex marriages in this
country increased from 42,030 in 2011 to 48,740 in
FIGURE 1.1 The process of deduction 2016 (Statistics Canada, 2017), which a researcher

bel65796_ch01_001-028 8 01/13/22 12:09 PM

 1 General Research Orientations 9

Box 1.3 A deductive study

Hay (2014) has taken a deductive approach
to this issue. As in all deductive research, his
goal was “theory testing.” He started by propos-
ing some theories, derived testable hypotheses
from them, and then set out to determine which,
if any, of the hypotheses were supported by the
Guenter Guni/iStockphoto

data. He reviewed seven well-known theories of

secularization, and deduced specific hypotheses
from each one. For instance, the deprivation–­
compensation theory holds that the existential se-
curity that affluent people in developed societies
Women wearing headscarves in Iran, a society that enjoy through health care and education, long life
has experienced far less secularization than many expectancy, and the absence of military conflict
Western countries. Do you think that some of the the- shields many of us from the traumas that draw
ories examined in Hay's (2014) research can help to people toward religion. The hypothesis he de-
explain the high levels of religiosity found in Iran? For rived from this theory was that people who have
example, could the acceptance of religious beliefs experienced real tragedy in their lives are more
and practices there be related to cultural factors that
likely to be religious than people who have no
favour societal cohesion over personal autonomy?
direct exposure to such things. Hay then analyzed
Could low levels of religious pluralism be a factor?
contemporary survey data that included mea-
sures of religious beliefs and practices, as well as
For millennia, religion has been the basis for be- a number of variables that provided indicators of
liefs about our place in the cosmic order, what it factors that were relevant to each of the theories.
means to be human, how we should treat each There was limited support for deprivation–
other, and whether anything exists beyond the compensation theory, but there was strong sup-
material, physical universe. However, in many port for three other theoretical positions. One
Western countries a process of secularization theory that gained support involves the idea that
has occurred in which fewer and fewer people a growing acceptance of personal autonomy in
embrace religious beliefs and practices, a pro- spiritual and moral matters has led to increased
cess that became especially marked after the levels of secularization. A second position that
middle of the twentieth century. To be sure, reli- was consistent with the results holds that the re-
gion has not disappeared in Western countries, ligious pluralism found in wealthy countries cre-
including Canada. In fact, about 26 per cent ates a situation in which it is very difficult for a
of Canadians attend religious services once a particular faith group to make the claim that it
month or more frequently (Statistics Canada, is the “one true religion,” which casts doubt on
2013), and 29 per cent say religion is very im- all religious doctrines and so increases levels of
portant to them (Pew Research Center, 2019). secularization. A third theory that was substan-
Nonetheless, several indicators suggest that the tiated by the data maintains that increasing ac-
proportion of Canadians who are firmly reli- ceptance of human evolution and the belief that
gious is considerably lower than it was several observable phenomena are purely material or
decades ago. What can account for this pro- physical in nature has reduced levels of religiosity
found social change? in Canada.

bel65796_ch01_001-028 9 01/13/22 12:09 PM

10 PART I Fundamental Issues in Social Research

Research in the News

Police culture: An inductive study
Police forces have faced severe criticism re- Her interviews with officers unearthed many
cently for their workplace culture and practices, accounts of sexual assault and intimidation in the
some of which have sparked protest move- workplace, along with racism, homophobia, and
ments around the globe. The Civilian Review and sexism. Police culture in general was commonly
Complaints Commission for the RCMP and the described by her informants as being charac-
national Auditor General’s Office have published terized by a “high-school mentality.” She deter-
damning reports on sexual harassment and mined that the toxic work environment that the
other forms of abuse that officers themselves police find themselves in “damages many of its
have endured on the job. In an article that ap- officers physically, mentally, and spiritually.”
peared in the Globe and Mail, Lesley Bikos (2017), Bikos concluded that a crucial factor that
a former member of the London, Ontario, police perpetuates the toxic work environment in police
service, described her research into noxious as- departments is a lack of protection for officers
pects of police culture across a variety of police who want to speak out against the infractions
organizations. that they have experienced or witnessed. The
Bikos conducted in-depth interviews with of- “small percentage of bad-apple officers” who
ficers from 23 different police services. Her re- wreak havoc in the workplace would not be such
search was inductive in that her goal was not a problem if better mechanisms for reporting
to test a pre-existing theory. As a former officer and investigating objectionable behaviours were
she had some familiarity with the subject matter in place. Other factors that she identified that
before she began her study, but her interpreta- could contribute to a change in police culture in-
tions and conclusion largely took shape after she cluded greater training and professionalization
had collected some rich, detailed interview data, in the force, civilian involvement in promotional
which she used to make sense of the social con- decisions and internal investigations, and better
text in which police culture evolves. communication across ranks.

could use in the context of a discussion of the chang- context is whether the social sciences should follow
ing forms of the family in Canada. the same principles and procedures as the natural
The next section examines some epistemological sciences. Three broad positions on these matters
issues that affect the conduct of social research. have emerged: the positivist, interpretivist, and criti-
cal approaches to social science.
Epistemological Positivism
considerations One epistemological position that affirms the impor-
Those who do social research base their work on a tance of following the natural sciences is positivism.
number of epistemological assumptions—notions Although definitions of the term vary, positivism is
of what can be known and how knowledge can be generally taken to entail the following:
acquired. A related epistemological issue pertains
to the question of what should be regarded as ac- 1. Only phenomena confirmed by the senses
ceptable knowledge. A fundamental debate in this (sight, hearing, etc.) can be accepted as

bel65796_ch01_001-028 10 01/13/22 12:09 PM

 1 General Research Orientations 11

knowledge: this is the principle of empiricism. A common mistake is to treat positivism as syn-
Ideas must be subjected to the rigours of em- onymous with science and the scientific. In fact,
pirical testing before they can be considered philosophers of science and social science differ
knowledge. quite sharply over how best to characterize scientific
2. A key purpose of theory is to generate hypoth- practice, and since the early 1960s positivism has ac-
eses that can be tested and thereby allow expla- quired some negative connotations. One reason for
nations of observed laws and principles to be dissatisfaction was the fact that in the past, certain
assessed (deduction). researchers in the positivist tradition ignored some
3. Knowledge can also be arrived at through the fundamental differences between human beings
gathering of facts that provide the basis for and the often inanimate or not fully conscious enti-
generalizations or laws (induction). ties studied by natural scientists. Unlike subatomic
4. Science must (and presumably can) be particles or plants, for example, we humans have
“­value-free.” That is, it must be conducted in thoughts, feelings, and values—perhaps even some
such a way that different researchers, given capacity for volition. Those aspects of human be-
the same data, will always reach the same con- haviour were often not addressed in the leading pos-
clusions, no matter how different their values itivistic theories of the day. The positivist idea that
might be. In the past this “value-free” quality science can or should be value-free was also a source
was called “objectivity”; today it is more likely of dissatisfaction. Critics pointed out that “neutral”
to be called intersubjectivity. social scientists often took moral positions on social
5. There is a clear distinction between scientific issues, at least implicitly. For example, theories im-
statements, which describe how and why cer- plying that social equilibrium or harmony is normal
tain social phenomena operate the way they seemed to suggest that social change was not needed
do, and normative statements, which outline or not desirable. Critical social science, discussed
whether certain acts or social conditions are below, claims that it is the duty of the researcher to
morally acceptable. Only scientific statements help bring about social change in order to create a
have a place in the domain of science; norma- better world.
tive statements belong in the realm of philos-
ophy or religion. This idea is implied by point Interpretivism
1 above, because the truth of moral claims Interpretivism to some extent grew out of the epis-
cannot be confirmed by the senses. temological critique of positivism, and provides an
alternative to the sort of social science typically done
It is possible to see in these five points a link with by positivists. Interpretive researchers maintain that
some of the issues already raised about the rela- it is the role of social scientists to grasp the subjective
tionship between theory and research. Positivism meanings of people’s actions. They make the point
assumes a fairly sharp distinction between theory that people act on the basis of the meanings that
and research and includes elements of both deduc- they attribute to their acts and to the acts of others.
tion and induction. One role of research is to test Individuals use their own common-sense constructs
theories and gather the information necessary for to interpret the reality of their daily lives, and it is
the development of scientific laws. Positivism also those interpretations that motivate their behaviour.
implies that it is possible to collect observations Interpretivists claim that it is the job of the social
without any reference to pre-existing theories, scientist to gain access to the “common-sense think-
and to develop new theories purely on the basis of ing” of the people they study and hence to understand
those observations. Finally, theories and proposi- people’s actions and their social world from the point
tions not directly testable through empirical ob- of view of the actors. Thus, any thoughts constructed
servation are often not considered to be genuinely by the social scientist to grasp this social reality must
scientific. be founded on the perceptions and lived experience

bel65796_ch01_001-028 11 01/13/22 12:09 PM

12 PART I Fundamental Issues in Social Research

of those they study: people living their daily lives their environment (including the actions of others)
within their own social world (Schutz, 1962, p. 59). and acting on the basis of that imputed meaning
In order to gain access to other people’s perspectives, (cf. Collins, 1994). In research terms, according to
interpretivists commonly immerse themselves in the Blumer (1962, p. 188), “the position of symbolic in-
social environments frequented by the people they teraction requires the student to catch the process
study, or at least conduct lengthy interviews with of interpretation through which [actors] construct
them. The insights gained by the researcher com- their actions.”
monly follow from the information derived in this Taking an interpretative stance can result in
way, so the process used tends to be inductive. surprising findings, or at least findings that appear
Many interpretive social scientists argue that the surprising if the researcher’s or reader’s everyday ex-
subject matter of the social sciences—people, groups, perience does not include the particular social con-
and institutions—is fundamentally different from text being studied. Box 1.4 provides an interesting
that of the natural sciences. For them it follows that example.
the study of the social world requires a different logic Of course, as the example in Box 1.4 suggests,
and research procedure, one that reflects what they when social scientists adopt an interpretive stance,
see as the distinctiveness of humans as against other they are not simply revealing how members of a
living things or inanimate objects. This clash reflects social group interpret the world around them. The
a division between the positivist emphasis on the ex- social scientist almost certainly aims to place those
planation of human behaviour and society, and the interpretations into a social scientific framework.
interpretivist preference for an empathetic under- Thus, there is a double interpretation going on: the
standing and interpretation of human existence. This researcher is interpreting others’ interpretations.
division, which precedes the emergence of modern Indeed, there is even a third level of interpretation,
social science, finds expression in Max Weber’s because the researcher’s interpretations have to be
(1864–1920) notion of Verstehen (which means “em- further interpreted in terms of the concepts, the-
pathetic understanding”). Weber described sociol- ories, and literature of the social sciences. Thus, in
ogy as a “science which attempts the interpretive Box 1.4 the idea that Riverside was not perceived as
understanding of social action in order to arrive at a high-crime area by residents was Foster’s interpre-
a causal explanation of its course and effects” (1947, tation of the residents’ interpretations. She then had
p. 88). Weber’s definition seems to embrace both ex- the additional job of placing her findings into a social
planation and understanding, but the crucial point scientific framework, which she accomplished by re-
is that the task of “causal explanation” is undertaken lating them to existing concepts and discussions in
with reference to the “interpretive understanding criminology: concepts such as informal social con-
of social action.” This is a different emphasis from a trol, neighbourhood watch schemes, and the role
more Durkheimian view in which the external forces of housing style as a possible influence on criminal
that affect behaviour may not be perceived by those activity.
involved, or at least may have no meaning for them.
Symbolic interactionism is an example of a so- Critical approaches to social science
ciological perspective that falls under the heading Like interpretivism, critical social science devel-
of interpretivism. The ideas of the founders of sym- oped in part as a reaction to positivism. Social sci-
bolic interactionism—in particular George Herbert entists who adopt a critical approach use a diversity
Mead (1863–1931), who maintained that the indi- of research methods, including those used by pos-
vidual’s self-concept emerges through an apprecia- itivists and interpretivists, and may use a deduc-
tion of the perceptions of others—have been hotly tive or inductive approach. But they disagree with
debated. Symbolic interactionists argue that interac- the positivist notion that researchers should take a
tion takes place in such a way that individuals are value-neutral stance regarding their subject matter.
continually interpreting the symbolic meaning of In fact, they maintain that research and knowledge

bel65796_ch01_001-028 12 01/13/22 12:09 PM

 1 General Research Orientations 13

Box 1.4 Interpretivism in practice

Foster (1995) conducted ethnographic research social control worked to keep the level of crime
using participant observation and semi-­ contained.
structured interviews in a housing estate in East Informal social control has several aspects. One
London, UK, referred to as Riverside, a residen- is that neighbours often look out for each other. In
tial complex experiencing a high level of crime the words of one of Foster’s interviewees: “If I hear
according to official statistics. However, she a bang or shouting I go out. If there’s aggravation
found that residents did not perceive the estate I come in and ring the police. I don’t stand for it”
to be a high-crime area; nor were they overly (Foster, 1995, p. 575). Another aspect of informal
anxious about becoming victims of crime. Those social control is that people often feel secure be-
perceptions could be attributed to a number cause they know each other. A second respondent
of factors, but a particularly important one said: “I don’t feel nervous . . . because people do
was “informal social control,” which was used generally know each other. We keep an eye on
in conjunction with more formal methods such each other’s properties. . . . I feel quite safe because
as policing. People expected a certain level of you know your neighbours and you know they’re
crime, but felt fairly secure because informal there . . . they look out for you” (Foster, 1995, p. 575).

should not be considered as ends in themselves, but in feminist, post-colonial, anti-poverty, anti-­racism,
as means to be used to rid the world of suffering and and queer studies. Another example is participa-
oppression (Neuman, 2003). tory action research, which is discussed in the sec-
Marxists, for example, argue that those who own tion below and examined again in later chapters.
the means of production deceive, constrain, and Institutional ethnography and critical discourse
exploit the weak. The masses could be free if social analysis, examined in Chapters 10 and 12, respec-
scientists, by asking embarrassing questions and tively, are further examples of critical social science.
making pointed arguments, would uncover exploita-
tion, expose hypocrisy, and reveal to the general Participatory action research (PAR)
populace the nature and extent of their oppression. Of all the approaches to social science, PAR is the
This would transform the masses from what Marx one most closely associated with social and political
called a Klasse an sich (a class in itself, an objective activism. The origins of participatory action research
reality) into a Klasse für sich (a class for itself, one (sometimes referred to as “action research,” “emanci-
with an awareness of its exploitation). patory research,” or “participatory research”) can be
Critical social scientists also believe that research found in the immediate post–Second World War era
should be action oriented. It should involve praxis: in the work of Lewin (1946) and others who sought
putting one’s theoretical and academic positions into to bring social science to bear directly on specific
practice. The idea of praxis is contained in Marx’s social problems. Lewin is said to be the originator
famous dictum that “philosophers have only inter- of the adages, “no research without action, no action
preted the world in various ways; the point, however, is without research,” and “nothing is so practical as a
to change it [emphasis in original]” (Marx, 1845/1998, good theory” (Ring, 1967, p.114).
p. 574). This exhortation applies not only to positivism, In the early 1970s, PAR began to take on a more
but also to interpretive perspectives on social life. explicitly activist orientation when people in Latin
A critical perspective has been adopted by a America, Africa, Asia, and elsewhere became disillu-
wide variety of scholars, including those engaged sioned with the ability of conventional social science

bel65796_ch01_001-028 13 01/13/22 12:09 PM

14 PART I Fundamental Issues in Social Research

to improve social conditions in their countries. These would arrive in communities, collect data, and
individuals expanded the methodology of PAR such rarely, if ever, return. … In elementary school,
that local people affected by the issue at hand acted I became one of many subjects in IQ and
as partners and co-equals with professional social achievement research on Indigenous children.
scientists and government officials. All three groups It remains uncertain what these researchers
worked together to devise research questions, choose did with the data they collected, how it was in-
research methods, gather and analyze data, and un- terpreted, and, more importantly, how it may
dertake a course of action to ameliorate the problem have contributed to misinformation about my
(Frisby et al., 1997; Smith, 1997). community.
Now, working for social justice is built into the
methodology of PAR itself. Participants (including As a corrective, Peltier (2018) advocates an approach
community partners) first reflect on a social problem that combines PAR with Indigenous methodologies
and how it might be studied and rectified, and then through what is called “two-eyed seeing,” a way of
use the results of that process as the basis for social knowing in which both Indigenous epistemologies
action. Once action has been taken, participants and Western approaches to knowledge are valued.
discuss the consequences of the action and how In her study of Indigenous access to plural systems
those consequences might be used to generate more of health care, which comprise Indigenous healing
knowledge and inform further action. In this way a and Western medicine, she first sought community
social problem is dealt with by the people directly af- sanction for the research by consulting local Elders
fected by it, in collaboration with others. and participating in spiritual ceremonies. She then
The PAR process has the potential to transform reached out to health directors in three Indigenous
the participants personally, intellectually, and even communities to determine whether her topic was of
spiritually. Debbink and Ornelas (1997), for instance, relevance to those communities, which allowed her
describe how some Alberta dairy farmers worked to develop her social relationships with the groups
together with Mexican campesinos (low-income she would study. She then ensured that her project
farmers) and an activist intellectual to bring donated adhered to both national and local research ethics
Alberta cattle to an impoverished, rural Mexican protocols.
community. Similarly, Frisby, Crawford, & Dorer A local advisory committee was set up to ensure
(1997) outline how some low-income women in accountability to the community. It directed the
British Columbia teamed up with social scientists hiring of an Indigenous research assistant and the
and local recreation directors to help women with recruitment of participants, and helped to develop
limited means gain access to physical activity ser- the research instruments, thus contributing to the
vices. In both cases, people directly affected by a co-creation of knowledge. Participants were asked
problem and researchers from outside the commu- to recount their personal experiences of how cancer
nity interacted as equals to produce knowledge and had an effect on their emotional and spiritual lives
address a social issue. and on their relationships with family and their com-
PAR has a special relevance for Indigenous people munity. Part of the narrative involved a description
in Canada. Peltier (2018, p. 1) lays out in rather stark of how Indigenous healing or Western medicine, or
terms the shortcomings of conventional research some combination thereof, was part of their lived
methods: experience. Each participant had an opportunity to
read and reflect on a copy of the story that came out
In the past, people who have had limited knowl- of the interaction with the researcher. All the sto-
edge about Indigenous peoples, worldviews, ries gathered were used to create a collective, teach-
or communities have conducted research on ing story, which was developed cooperatively using
Indigenous peoples. These researchers often qualitative data coding (see Chapter 13). The final
employed a ‘helicopter approach’ where they results of the study were shared with the participants

bel65796_ch01_001-028 14 01/13/22 12:09 PM

 1 General Research Orientations 15

and with members of other Indigenous communi- (1883/1968, p. 267). Such people maintain that there
ties, and were disseminated at academic conferences is no objective social reality against which our
and through academic publications. conceptions and views of the world may be tested.
A middle-ground or “soft constructionist” position
Summary is also possible, and is held by many. It maintains
that there may be an objective social reality, but that
The aim of this section has been to outline how epis- many of our ideas do not reflect it: instead, those
temological considerations are related to research ideas are constructed to justify or rationalize various
practice. As noted, a key concern is whether a nat- forms of domination. Box 1.5 provides two illustra-
ural science or positivist approach can supply legiti- tions of the less stringent form of constructionism.
mate knowledge of the social world. We’ve seen that A second debate revolves around these questions:
positivists often use the deductive method, although Is social reality akin to the physical world as most
they concede that it is possible to generate theories people see it—largely fixed and “out there,” some-
using induction. By contrast, we’ve discerned that thing that individuals and groups have to confront
interpretive researchers typically adopt an induc- but over which they have little or no control, like
tive strategy, and that critical social scientists may a snowstorm? Or is social reality not necessarily
use either one. Another important issue involves pre-existing and fixed, but rather created through
the purpose of social research, in particular whether our actions? A “yes” to the first question indicates
social scientists, in both their academic work and support for a variant of objectivism, while a “yes” to
their roles as citizens, should be actively engaged in the second affirms a kind of constructionism. In this
issues of social justice. chapter we’ll focus mainly on this second debate:
It is important not to overstate the connections whether social reality can be created.
between epistemological issues and research prac- Some social scientists suggest that social phenom-
tice: they represent tendencies only. Particular epis- ena confront individuals as external facts beyond
temological principles and research practices do not their reach or influence. For example, an organiza-
necessarily go hand in hand in a neat, unambiguous tion has rules and regulations and adopts standard-
manner. Often hybrid approaches are taken, which ized procedures for getting things done. A division
combine different positions and approaches. This of labour assigns people to different jobs. There is a
point will be made again on several occasions. hierarchy of authority, a mission statement, and so
on. Objectivists see any organization as possessing a
Ontological considerations reality external to any of the specific individuals who
inhabit it; they may leave, but it will stay. Moreover,
There are two ontological debates that are of particu- the organization represents a social order in that it
lar interest to social scientists. The first is concerned exerts pressure on individuals to conform to orga-
with the following questions: Do social phenomena nizational requirements. People learn and apply the
have an objective reality, independent of our percep- rules and regulations and follow the standardized
tions? Or is what passes for reality a set of mental procedures. They do the jobs to which they are ap-
constructions? If you answer “yes” to the first ques- pointed. If they do not do these things, they may
tion, you are in the objectivist camp. People on this be reprimanded or even fired. The organization is
side of the debate maintain that there is such a thing therefore a constraining force that acts on and inhib-
as social reality, and that it is the job of social scien- its its members. To a large extent, this is the “classic”
tists to discover what that reality is. An affirmative way of conceptualizing an organization.
answer to the second question means that you agree An alternative ontological position challenges
with the constructionist position. People holding the suggestion that things such as organizations
this view are in sympathy with Nietzsche’s famous are external realities confronting social actors who
aphorism that there are no facts, only interpretations have limited power to influence or change them.

bel65796_ch01_001-028 15 01/13/22 12:09 PM

16 PART I Fundamental Issues in Social Research

Box 1.5 Constructionism in action

Lantz and Booth (1998) examined media treat- that the idea that young women’s lifestyles cause
ment of the apparent rise in the incidence of breast cancer was constructed as a social fact by
breast cancer that began in the early 1980s popular-magazine writers.
and found that its depiction as epidemic can be Similarly, Hallgrimsdottir, Phillips, & Benoit
treated as a social construction. They analyzed (2006, p. 266) argue that the media “contributes
a variety of popular magazines and noted that to constructing, reproducing and deepening the
many of the articles drew attention to the life- social stigmas associated with working in the
styles of modern women, such as delaying first sex industry.” The authors compared media de-
births and having careers. The authors also pictions of sex-trade workers in Victoria, British
argued that the articles ascribe blame: “Women Columbia, from 1980 to 2004 with accounts pro-
are portrayed as victims of an insidious disease, vided by street sex workers, escorts, and others
but also as victims of their own behaviours, many in this sector. Whereas the media portrayed
of which are related to the control of their own such people as wicked and blameworthy in the
fertility” (1998, p. 915). earlier years, and as exploited, trapped, and in-
This article concludes that, as a social cat- nocent girls more recently, the sex workers them-
egory, the breast cancer epidemic was rep- selves interpreted their work lives very differently.
resented in popular magazines in a particular Although there was considerable heterogeneity in
way—one that blamed the victims and their life- their experiences and attitudes toward their trade,
styles, particularly in the case of young women. many saw sex work as a largely mundane matter
Yet, in fact, fewer than 20 per cent of cases of of earning a living. Because they are marginalized
breast cancer were in women under the age of and stigmatized, however, their voices are seldom
50. Lantz and Booth’s study is fairly representa- heard; most people have their ideas about the sex
tive of a constructionist ontology in suggesting trade constructed for them by the media.

Strauss, Schatzman, Ehrich, Bucher, & Sabshin For instance, the official rules may say that only
(1973), for example, carried out research in a psy- a doctor can increase medication; however, some
chiatric hospital and proposed that its organization nurses are routinely given this power, even though
was best conceptualized as one of “negotiated order.” this is never actually stated in the regulations. The
Instead of viewing order as a pre-existing character- social order is in a constant state of change because
istic, they argued that it is worked at and created to the hospital is “a place where numerous agreements
some extent, and that the rules are far less extensive are continually being terminated or forgotten, but
and less rigorously imposed than might be supposed also are continually being established, renewed, re-
from an objectivist account of organizations. viewed, revoked, [and] revised. . . . In any pragmatic
Indeed, Strauss and colleagues saw rules more as sense, this is the hospital at the moment [emphasis
general understandings than as commands (1973, added]: this is its social order” (Strauss et al., 1973, pp.
p. 308). Precisely because relatively little of the ac- 316–317). The authors argued that a preoccupation
tivity of doctors, nurses, and other personnel is with the formal properties of organizations (rules,
specifically set down or prescribed, the social order organizational charts, regulations, and roles) makes
of a hospital is an outcome of agreed-upon pat- it hard to recognize the degree to which order in or-
terns of action that are themselves the products of ganizations has to be established in everyday interac-
negotiations among the different parties involved. tion. This informal organization arises because there

bel65796_ch01_001-028 16 01/13/22 12:09 PM

 1 General Research Orientations 17

cannot be rules for every possible contingency, and researcher who sees organizations as objective social
because the existing rules are sometimes problematic. entities that act on individuals is likely to emphasize
However, this is not to say that the formal properties their formal properties. Alternatively, a researcher
of organizations have no effect on individual action. who is interested in the negotiated, changing nature
Although Strauss and colleagues stressed the of organizations is likely to focus on the active in-
active role of individuals in the social construction volvement of people in reality construction. In each
of reality, they did not push the argument to an ex- case, a different research design is required.
treme. For example, they did not claim that nurses
can negotiate their roles to the point where they are
allowed to operate on patients. But not all writers General orientations:
adopting a constructionist position are similarly pre- Quantitative and qualitative
pared to acknowledge the existence or importance of research
an objective reality. It is precisely this split between
seeing the social world as an objective reality and Many writers on methodological issues distinguish
seeing it as a subjective reality in a continuous state between quantitative and qualitative research. We
of flux that Giddens sought to straddle in formulat- have done so above without explicitly describing
ing his idea of structuration (recall Box 1.1). the difference between the two. The most basic dif-
The constructionist perspective that maintains ference is that quantitative research uses numbers
that social reality can be negotiated also suggests and statistics in the collection and analysis of data,
that the concepts people employ to help them under- while qualitative research relies mainly on words
stand the natural and social world are social prod- and other non-numerical symbols. Some writers see
ucts whose meaning is constructed in and through the distinction between the two types of research as
social interaction. For example, a concept such as fundamental; others believe it to be no longer useful
“masculinity” is treated as a social construction. This or even “false” (Layder, 1993, p. 110). Although this
implies that masculinity is not a distinct, timeless, issue has been vigorously debated, we’re convinced
and universal entity, but something whose mean- that there is a meaningful difference between quan-
ing is built up through interaction. That meaning titative and qualitative research. That difference will
is likely to be ephemeral, in that it will vary over be a recurring theme in this book, both because it is
time and place. The construction of one’s masculin- a useful way of classifying various research methods
ity may even have to be “recuperated” from time to and because it is a helpful reference point for a range
time, such as when Canadian politicians endeavour of issues in the practice of social research.
to restore their public gender identities (Maiolino, On the surface, it seems that the main difference
2015). This sort of social construction can be seen between quantitative and qualitative research is that
particularly well in discourse analysis, examined in quantitative researchers rely more on formal and
Chapter 12. As Potter (1996, p. 98) observed: “The mathematical measurement and analysis techniques
world . . . is constituted [emphasis in original] in one than qualitative researchers do. But many sociologists
way or another as people talk it, write it, and argue and others suggest that the differences are deeper
it.” This sense of constructionism frequently sparks than merely the amount of quantification. For many
an interest in how social phenomena are represented. writers, quantitative and qualitative research differ
in their epistemological foundations and in other
Relationship to social research respects too. Indeed, when we look at the areas that
Questions of social ontology cannot be divorced from were the focus of the last three sections—the rela-
issues concerning the conduct of social research. tionship between theory and research, epistemolog-
Ontological assumptions and commitments affect ical considerations, and ontological issues—we can
both the way research questions are formulated and see quantitative and qualitative research as forming
the way the research is carried out. For example, a two distinctive general orientations to the conduct of

bel65796_ch01_001-028 17 01/13/22 12:09 PM

18 PART I Fundamental Issues in Social Research

TABLE 1.1 Fundamental differences between quantitative and qualitative

research strategies
Quantitative Qualitative
Role of theory in research Mainly deductive; testing of theory Mainly inductive; generation of theory
Epistemological orientation Natural science model; positivism Interpretivism
Ontological orientation Objectivism Constructionism

social research. Table 1.1 outlines the differences be- Although it is useful to contrast the two general
tween quantitative and qualitative research in terms orientations, it is important not to hammer a wedge
of the three areas. between them. For example, although quantitative
Thus, quantitative research: research tends to be deductive and qualitative work
inductive, there are notable exceptions to that gen-
• usually entails a deductive approach to the eral rule. It may seem perverse to introduce a basic
relationship between theory and research in set of distinctions and then suggest that they are
which theory testing is a prime objective; problematic, but a recurring theme of this book is
• incorporates the practices and norms of the that discussing the nature of social research is just as
natural science model and of positivism in complex as conducting research itself. We can out-
particular; and line the typical philosophical assumptions and re-
• generally embodies a view of society as an ex- search practices of the two general orientations, but
ternal, objective reality. the full reality is messier than those neat categories
would suggest. Issues become more complicated the
By contrast, qualitative research:
deeper we delve into them.
• takes a predominantly inductive approach to For example, we’ve seen that qualitative research
the relationship between theory and research, is typically described as being more concerned with
in which the generation of theories and inter- generating theories than testing them. However, there
pretations is the main goal; are many studies in which qualitative research is used
• rejects the use of the natural science and positiv- to test rather than generate theories. Ellis et al.’s (2020)
ist models in social research and replaces them study of interactions between Black Muslim immi-
with methodologies that seek to determine how grants and the police, discussed above, used a quali-
individuals interpret their social world; and tative approach, but an important goal of the research
• embodies a view of social reality as a constantly was to determine whether a pre-­existing theory—mi-
shifting and emergent property of individuals’ nority group threat theory—was supported by the
creations. qualitative data that was gathered. Another example
is Hier’s (2002) investigation of Toronto rave scenes,
There is even more to the quantitative/qualitative with their all-night dancing and amphetamine use.
distinction than this contrast suggests. The nature Hier wanted to show that the regulation of raves was a
of quantitative and qualitative research respectively contest between a city that feared increased drug use
will be outlined in greater detail in Chapters 4 and and rave supporters, who eventually won the day by
9. Then in Chapter 14 their contrasting features will arguing that banning raves would drive the drugs un-
be further explored as we examine the effects of the derground with even worse consequences. Similarly,
commitment to a positivist epistemology in quanti- although Wilson’s (2002) study of the same topic is
tative research and the rejection of that epistemology broadly interpretivist, with its examination of how
by qualitative researchers. ravers view their social situation, it includes some

bel65796_ch01_001-028 18 01/13/22 12:09 PM

 1 General Research Orientations 19

objectivist overtones. For example, in exploring the and objective in their studies. Research that simply
effects of technology, including the Internet, on the reflects the personal views of its practitioners
scene, he was describing a world that is “out there” would be biased and invalid, and thus unscientific.
and as having a formal, objective quality. And, of Durkheim (1858–1917) wrote that social facts are ob-
course, there are quantitative studies that incorporate jects whose study requires that all “preconceptions
elements of the qualitative orientation, such as stud- must be eradicated” (1938, p. 31). Since values are a
ies of social attitudes that explore people’s perceptions form of preconception, his point implies that they
of societal issues in considerable depth (e.g., Kandler, should be suppressed when conducting research. But
Bell, & Riemann, 2016). is that humanly possible? Social researchers never
The point here is that, in general, quantitative work in a moral or evaluative vacuum: they are
and qualitative research represent different research always influenced by value presuppositions that have
orientations, and that the two approaches may be implications for the conduct of social research. This
quite different in terms of the role of theory, episte- view is increasingly accepted among social research-
mological issues, and ontological concerns, but the ers. Indeed, it is now recognized that values can play
distinction is not hard and fast. Also, it is becoming a role at any or all points in the process of social re-
more commonplace for research projects to combine search, including the following:
both within a single research project, as we will see
in Chapter 14. • choice of research area
Finally, lest there be any doubt about it, the position • formulation of the research question
taken in this book is that both general ­orientations— • choice of method
the quantitative and the qualitative—are invaluable • formulation of the research design and data
in the quest for knowledge and understanding. And collection techniques
both have profound social and political importance. • data collection
• analysis of data
Influences on the conduct of • interpretation of data
• conclusions
social research
You can now see how social research is influenced This can have both positive and negative conse-
by a variety of factors. Figure 1.3 summarizes the in- quences. Values can help us act ethically and may
fluences examined so far, but adds three more: the motivate us to be a positive influence in the world.
impact of values, politics, and issues related to the re- At the same time, they may shade off into ideolog-
search question. ical bias or observational blind spots if they lead
us to ignore theoretical perspectives we don’t like
Values or evidence that does not match our expectations.
How might the values, personal beliefs, and feelings Similarly, problems can arise if researchers become
of researchers affect their work? Perhaps one would too emotionally involved with the people they
expect social scientists to be completely value-free study.
One way of dealing with the problem of values
Issues related to the and bias is to recognize that research cannot be
Theory research question Epistemology
value-free, and to try to ensure that values in the
research process are acknowledged and made ex-
Social research plicit. This is part of a larger process of reflexivity
or self-reflection that researchers are encouraged to
carry out. As Turnbull (1973, p. 13) put it at the be-
Values Politics Ontology ginning of his book on an African society known
FIGURE 1.3 Influences on social research as the Ik:

bel65796_ch01_001-028 19 01/13/22 12:09 PM

20 PART I Fundamental Issues in Social Research

The reader is entitled to know something of of objectivity, and some of the more categorical pro-
the aims, expectations, hopes, and attitudes nouncements on the subject, such as Durkheim’s,
that the writer brought to the field [in his case, have fallen into disfavour. At the same time, giving
Western values about the family], for these will free rein to one’s political beliefs and value positions
surely influence not only how [the writer] sees can be problematic. Researchers today still have to
things but even what [the writer] sees. fight the all-too-human propensity to demonize
those whose values are different from their own, and
Researchers are increasingly prepared to forewarn they still struggle with the temptation to summarily
readers of their biases and assumptions and to ex- reject research findings when the researcher’s ideo-
plain how these may have influenced their findings. logical or moral positions are not compatible with
Since the mid-1970s many researchers have pub- their own.
lished “insider” accounts of what doing research is
really like, as opposed to the generalities presented Politics in social research
in social research methods textbooks (like this one). At various places in this book we take the position
These accounts frequently function as “confessions” that social research has political implications. Here
of personal biases and reveal the pride that research- are some examples of the ways in which social re-
ers take in telling readers how open they are in re- search may be political:
vealing them.
Still another approach is to argue for consciously • Social researchers often take sides. There are
value-laden research. Some writers on social re- many ways this can happen. To consider just
search celebrate what Mies (1993, p. 68) called a a few: feminist researchers may focus on the
“conscious partiality.” For example, Tastsoglou and disadvantages that women face and the possi-
Miedema (2003) clearly adopted a feminist, anti-­ bilities for improving the position of women
racist approach in studying immigrant women in in society; some social scientists may favour
the Maritimes. A similar perspective allowed Pratt increased government intervention in eco-
and Valverde (2002) to describe a large Canadian nomic affairs, while others defend the free
newspaper as a “notorious tabloid,” “obsessed” with market; sociologists and political scientists in
what it called bogus refugees. It is also exemplified in Quebec may be split between sovereigntists
Hallgrimsdottir et al.’s (2006) condemnation of the and federalists.
media’s role in stigmatizing sex workers in British • A related issue involves research funding. Much
Columbia. In fact, some feminist researchers would social research is funded by organizations such
consider it inappropriate (as well as difficult) to do as private firms or government departments
research on women in an objective, value-neutral that may have a vested interest in the outcomes
way because that would be incompatible with their of the research. The very fact that these orga-
values. Instead, many feminist researchers argue for nizations fund some research projects but not
research that exposes the conditions of women’s dis- others opens the door to political influence.
advantage in a male-dominated society, as Demaiter Such organizations may seek to invest in stud-
and Adams (2009) did in their study of women in ies that will be useful to them or supportive
the IT sector, featured in the Methods in Motion box of their operations and world views. They will
on page 21. Some feminist writers argue that only re- often call for researchers to tender bids for an
search on women intended for women is consistent investigation in a certain area. When social
with women’s wider political needs. researchers take part in such exercises, they
There are, then, different positions that can be enter a political arena, since their research may
taken in relation to values and value-free research. be designed to please the funding body. As a
Few writers today believe it is possible to be truly result, as Hughes (2000) observed in relation to
objective. There is a greater awareness of the limits research in the field of crime, an investigation

bel65796_ch01_001-028 20 01/13/22 12:09 PM

 1 General Research Orientations 21

Methods in Motion Applications to Canadian Society

Women in male-dominated occupations: How do they fare?
Demaiter and Adams (2009) did a qualitative The authors found that, by and large, the
study of how Canadian women cope in the male-­ women they interviewed did not perceive their
dominated world of information technology (IT) gender to be a barrier to their career advance-
workplaces. As is typical of qualitative research, ment. One database analyst, for example, stated:
the study involved only a relatively small number “Yes, I’ve never felt that I was at a disadvan-
of people: interviews were conducted with 11 suc- tage (or an advantage) because I am a woman”
cessful women in eight IT organizations located in (Demaiter & Adams, p. 41). Yet when asked to
four different provinces. The authors acknowl- elaborate on their experiences, many of the
edge that the “small sample size and the unique- women suggested that gender was in fact an issue
ness of study participants prevent generalization” in their workplace. For example, one interviewee
(Demaiter & Adams, p. 39), which is commonly said, “I mean you have to be super smart, super
the case with this sort of research. However, as intelligent, like way above average to go beyond a
we will see in later chapters, it is not necessarily certain level because the glass ceiling is definitely
the goal of qualitative research to come up with there” (Demaiter & Adams, p. 43). Demaiter and
findings that can be generalized to some larger Adams concluded that the women in their study
population. Instead, qualitative researchers seek may have been successful in their careers partly
to find out how the people in the study perceive because they were, to a certain extent, oblivious
their world by allowing them to speak for them- to the gendered nature of their workplace—they
selves. The sorts of insights that can be derived forged ahead undaunted by the gender barriers
from research of this kind are difficult to attain in they faced.
quantitative research, which normally requires In addition to providing a window from which
research participants to choose from a fixed set to view women’s experiences in male-dominated
of responses. occupations, this study illustrates a point raised
For the interviews, Demaiter and Adams used in the Preface to this book, namely that issues of
a semi-structured format that included open- power and politics inevitably enter the picture
ended questions. The interviews were conducted whenever social interaction takes place. What
in three different ways: eight by telephone, two in differs from study to study is the extent to which
person, and one by email. Here again the authors the political implications are made explicit. In the
were open about the limitations of their meth- case of this project, the political elements of the
ods: “We acknowledge the fact that the mixed study are never far from the surface. For instance,
format [for interviews] . . . is not ideal” (Demaiter the authors write that the women’s “tendency
& Adams, p. 40). The authors are not alone in to downplay the significance of gender” in their
having to settle for a methodology that wasn’t career histories may serve to “prevent mean-
exactly what they would have preferred. In fact, ingful change” (Demaiter & Adams, p. 31), which
virtually all studies in the social sciences involve suggests that the researchers see their work as
methodological compromises of one sort or an- informing a larger discussion on gender relations
other, partly because the resources available to in Canadian society, an issue that has important
do the research are always limited. Nevertheless, political dimensions. Again, virtually all research
the results can still be highly informative if wise in the social sciences is to some degree relevant to
methodological decisions are made. politics in the larger sense of the word.

bel65796_ch01_001-028 21 01/13/22 12:09 PM

22 PART I Fundamental Issues in Social Research

of gun crimes among the “underclass” is more home has to be brought onside, then the staff,
likely to receive funding than one concerned and then the actual adolescents. Frequently, one
with wrongdoing committed by agents of the of the staff is then given the responsibility of
state. Morgan (2000) pointed out that research dealing with the fieldworkers. A suspicion that
funded by government is typically empirical the researchers are really working for manage-
and quantitative; it tends to be concerned with ment then has to be overcome. And it is unwise
the short-term costs and benefits of a partic- to assume that simply because gatekeepers have
ular policy or innovation; and it is generally granted access, a smooth passage will ensue in
uncritical in the sense that the underlying subsequent dealings with the people to be stud-
government policies are not questioned: all ied. Perhaps the most powerful of the boys will
the government wants to know about is the turn out to be the key gatekeeper. Researchers
effectiveness of their implementation. Political may find themselves used as pawns if subgroups
issues often arise when the funding agency attempt to enlist them in advancing a particular
itself is trying to secure a continuous stream of goal. Research participants who doubt the util-
government funding. ity of social research may even try to obstruct
• Gaining access to research participants can also the research process. For example, in Beagan’s
be a political process, especially in the case of or- (2001) study of “everyday inequalities” among
ganizations. Access to organizations is usually Canadian medical students, some of the stu-
mediated by gatekeepers concerned not only dents refused to complete a survey when they
about the researcher’s motives but also about heard she was asking about the treatment of
what the organization stands to gain from the gays and lesbians, and one student who did par-
investigation, what it will lose by participating ticipate nevertheless reported annoyance with
in terms of staff time and other costs, and the being asked about sexual orientation.
potential risks to its image. Often, gatekeepers • There may be pressure to restrict the publica-
seek to influence how the investigation will tion of findings. Hughes (2000) cited a study of
take place: what kinds of questions can be plea bargaining in the British criminal justice
asked; who can and cannot be a focus of study; system as a case in point. The researchers had
the amount of time to be spent with each re- uncovered what were deemed at the time to be
search participant; the interpretation of the disconcerting levels of informal plea bargain-
findings; and the form the reports will take, ing, and they concluded that the formal judi-
even to the point of asking to approve drafts. cial process was being weakened. The English
• Public institutions, such as police departments, legal establishment sought to thwart the dis-
schools, and hospitals, as well as most com- semination of the findings and was persuaded
mercial firms, are concerned with how they to allow publication only when a panel of aca-
are going to be represented in publications. demics confirmed the validity of the findings.
Consequently, gaining access is almost always Similarly, the editors of academic journals may
a matter of negotiation and as such inevitably refuse to publish pieces that do not conform to
turns into a political process. The product of their own ideological or political preferences.
such negotiations is often referred to as “the re-
search bargain,” and in many cases there is more This is just a small handful of the ways in which pol-
than one bargain that has to be struck. Once itics intrudes in the research process.
in the organization, researchers often discover
layers of gatekeepers. For example, let’s say a Issues related to the research
provincial government grants a research team question
permission to talk to the boys in a group home. Some important determinants of how research is
Before the research can begin, the head of the conducted follow from the research question one is

bel65796_ch01_001-028 22 01/13/22 12:09 PM

 1 General Research Orientations 23

trying to answer. Specifically, the choice of research stimulating than a simple declarative statement. A
orientation, design, or method has to match the spe- question arouses curiosity and challenges the re-
cific research question being investigated. For in- searcher to find ways to answer it. Here are some
stance, if a researcher is interested in measuring the examples of research questions taken from a quan-
impact of various possible causes of a social phenom- titative study of voluntary associations in Canada:
enon, a quantitative strategy is probably appropriate. “Does individual income level affect whether an indi-
Alternatively, if the focus is on the world views of vidual is a voluntary association member? Does this
members of a certain social group and how those effect differ between neighbourhoods?” (Duncan,
views develop, a qualitative research strategy—one 2010, p. 578). Here are two research questions taken
sensitive to the way participants interpret their social from a qualitative study of older immigrant women
world—may be the way to go. If a researcher is in- in Quebec: “What does it mean to older women to
terested in a topic on which little or no research has grow old in a land of immigration? What are the
been done, a quantitative strategy may be difficult to effects of immigration on their living conditions,
employ because there is little prior literature from especially when they have immigrated after 50?”
which to draw leads about possible causes. A more (Charpentier & Quéniart, 2017, p. 437).
qualitative, exploratory approach may be preferable The process of formulating and assessing re-
because that type of investigation is typically associ- search questions is something of an art, but here are
ated with the generation of theory rather than theory some general thoughts. As you may have noticed in
testing (see Table 1.1) and with a relatively unstruc- the research questions quoted above, the form the
tured approach to the research process. question takes may differ depending on whether the
A related dimension that has important implica- study is quantitative or qualitative in orientation.
tions for how the research will be conducted involves Quantitative questions usually ask whether a partic-
the nature of the topic and of the people being in- ular variable or set of variables has an influence on
vestigated. A researcher who wants to study individ- the phenomenon of interest. Duncan’s (2010) ques-
uals involved in illicit activities—for example, price tion, for example, reveals that her study was designed
fixing, shoplifting, or drug dealing—may find it dif- to inquire into whether personal income affects the
ficult to develop the rapport with them that is needed likelihood that one will join a voluntary association,
to conduct a social survey. It is not surprising, there- and if so, whether that relationship varies depending
fore, that researchers in these areas tend to use a on the neighbourhood in which one lives. Implied
qualitative strategy. On the other hand, it’s unlikely in her questions is a certain causal model that may
that the hypotheses in Box 1.3 on secularization apply to some larger population, one that extends
could have been tested using a qualitative approach. beyond the people studied. This is the deductive
method at work.
Formulating a research question Qualitative research questions tend to be less
Creating a research question is like picking a desti- specific, and are generally not designed to evaluate
nation for a hike: where you will start out, the route causal models of this kind. They tend to be “open-
you will take, and what you will experience along ended, evolving, and non-directional” (Creswell,
the way are largely determined by the endpoint you 2013, p. 138). The reason for this is that qualitative
are trying to reach. What you stand to accomplish researchers usually want to allow multiple interpre-
with a particular research study and how you will tations and perspectives to emerge from the people
accomplish it are profoundly affected by the goal you and settings they are examining. Also, since qualita-
have in mind, which is expressed in your research tive social scientists often use the inductive method,
question. they don’t have a fully preconceived notion of what
A research question states the purpose of the they expect to find, making it pointless to ask a spe-
study in the form of a question. This is useful be- cific question at the beginning of a study. Flexibility
cause a question can often be more evocative and in exploring whatever one encounters in the course

bel65796_ch01_001-028 23 01/13/22 12:09 PM

24 PART I Fundamental Issues in Social Research

of doing the research is essential if one is to uncover to mind. These questions could be something like
the meanings and interpretations held by the people the following: “How do the people under study
studied, and the processes by which they develop. For deal with those who violate group norms? In cases
instance, Charpentier and Quéniart’s (2017) research where people are punished for their violations, how
question, “What does it mean to older women to grow do the offenders make sense of the punishment they
old in a land of immigration?” was purposely open- receive, and how does it affect how they feel about
ended because their goal was to give the women they the other members of the group?”
studied a voice, not to march them through specific No single study can answer all the research ques-
topics and issues that they as researchers may have tions that will occur to the researcher. Only a small
considered to be important. The data they collected number of them can be selected. The need to narrow
allowed them to discern certain themes and patterns the topic is not only a matter of the time available
of response offered up by the participants in their and the cost of doing research. It is also a reflection
study, which is the goal of the inductive method. of the need for a clear focus.
Quantitative research may start with the choice As suggested above, the research questions may
of a general area of interest: for example, male ho- change as the study progresses, for a number of rea-
mosexuality. At this stage, a very general research sons. The discovery of a new data source may change
question might be “How do people in general feel the focus a bit, as might some of the initial findings.
about gays?” This broad research area would have For instance, if in the study of male homosexuality
to be narrowed down: for example, to “How does the researchers find that having a gay relative in the
the Canadian adult population react to the por- immediate family makes a large difference in peo-
trayal of gays in television dramas?” But even that ple’s attitudes, the research question and the atten-
is too broad, so the next level of specification might dant methodology and theoretical orientation may
be something like “Do young straight men react be revised. The research question may also change
differently than young straight women to the por- because of limitations in time and other resources
trayal of gay male romance on television program available to the researcher. Box 1.6 offers some tips
X? If so, do those differences reflect a more fluid on developing research questions.
sexual identity on the part of young women as Research questions set realistic boundaries for re-
compared to young men?” Such questions could be search. A poorly formulated research question can
linked to larger theories of sex roles, sexual orien- result in unfocused and substandard research. No
tation acquisition, socialization, and the tolerance matter how well designed a questionnaire is or how
of difference in society. skilled qualitative interviewers are, clear research
Research questions in qualitative inquiries questions are required to avoid going off in unneces-
may become more specific over time too, but that sary directions and tangents. Research questions are
usually occurs primarily after the researcher has crucial because they guide:
started to gather data. A general research ques-
tion may be retained throughout the study, but • the literature search;
more specific sub-questions normally evolve as the • decisions about the kind of research design to
research unfolds. For example, ethnographic re- employ;
search may begin with a general question such as • decisions about what data to collect and from
“What is going on here?” (Wolcott, 2008, p. 73), but whom;
as the researcher learns more about the people to • the analysis of the data; and
be analyzed, more specific questions usually come • the writing up of the findings and conclusions.

bel65796_ch01_001-028 24 01/13/22 12:09 PM

 1 General Research Orientations 25

Box 1.6 Considerations when developing research questions

A good research question will: • be linked to the other research questions in the
• be as clear as possible so that it will be under- study, so that you can develop a single argu-
standable to others; ment or at least a set of related arguments;
• be researchable: it must allow for the develop- this is hard to do with unrelated research
ment of a research design and the collection questions; and
of data; this means that extremely abstract • be neither too broad (no research project can
terms may not be suitable; do justice to all aspects of a topic) nor too
• relate in some way to existing studies that sug- narrow (unable to make a meaningful contri-
gest how your question may be approached. bution to an area of study).
Even with a topic that has not been widely If you are unsure about how to formulate re-
researched, there will probably be some rel- search questions (or about other aspects of
evant literature (for example, on related or research), look at journal articles or research
parallel topics). Establishing connections with monographs to see how other researchers have
existing studies will help to show how your re- handled them.
search can make a contribution to the existing
knowledge on the topic;

Key Points
• Theory can either precede research and data important dimensions of the quantitative/
gathering (the deductive method) or emerge out qualitative contrast.
of it (induction). • Values can impinge on the research process in
• Epistemological considerations loom large in the various ways; as a consequence, research often
choice of a research strategy. To a great extent, has political dimensions.
the issues revolve around the advantages and • The political dimensions of research relate to
disadvantages associated with the positivist, in- the exercise of power at different stages of an
terpretivist, and critical approaches to science. investigation.
• Quantitative and qualitative research constitute • Issues related to the research question can also
different approaches to social investigation and affect decisions about research methods. Clear
carry with them important epistemological and research questions improve the chances of
ontological assumptions. success.
• Ontological considerations, such as objec-
tivism versus constructionism, also constitute

Questions for Review (R) and Creative Application (A)

Theory and research than men), then provide a social theory that
R1 What is a “theory,” as the term is used in this might explain it. Be sure to comment on how the
chapter? theory would explain the phenomenon.
A1 Name a social regularity or pattern (e.g., women R2 What is the purpose of gathering data if the de-
are more likely to suffer from eating disorders ductive method is used?

bel65796_ch01_001-028 25 01/13/22 12:09 PM

26 PART I Fundamental Issues in Social Research

A2 Name one specific hypothesis that can be de- relationship between theory and data, episte-
rived from the theory you mentioned to address mological considerations, and ontological issues.
the previous bullet. Provide an example of a hy- A6 Illustrate your answer by describing how quanti-
pothetical research finding that would support tative and qualitative researchers could conduct
the hypothesis. a study on academic achievement among trans
R3 What is the inductive method, and how can it be high school students.
used to come up with theories or interpretations
of social phenomena? Influences on the conduct of social
A3 How could you use induction to analyze in-class research
interactions between professors and students? R7 How might someone’s personal values influence
What sorts of theories or interpretations might you the topics they choose to research?
come up with if you were to use induction in that A7 Explain how values might have affected
setting? Demaiter and Adams’s (2009) decision to do
research on women in the information technol-
Epistemological considerations ogy sector (see the Methods in Motion box on
R4 What is meant by the terms “positivism,” “inter- page 21).
pretivism,” and “critical social science”?
A4 How could you use each of these three approaches Research questions
to social science to do a study on how social media R8 What are the main characteristics of a good re-
usage shapes a person’s self-image? search question?
A8 Imagine that you are curious about the connec-
Ontological considerations tion between growing up in an economically
R5 What do the terms “objectivism” and “construc- poor neighbourhood and joining a street gang.
tionism” mean? Produce a research question that would be ap-
A5 Provide an objectivist account of why women propriate for a quantitative study on this topic.
are more likely to pursue a career in nursing Then construct a research question that a quali-
than men. Then compare and contrast it with a tative researcher might use.
constructionist view of this issue.
Visit our online resource center for free assessment:
General research orientations: [Link]
Quantitative and qualitative research
R6 Outline the main differences between quanti-
tative and qualitative research in terms of the

Interactive Classroom Activities

1. Divide the class into small groups of students. Each drawing on the knowledge they’ve gained in
group is to select a social phenomenon, human class and from reading this chapter:
behaviour, or experience that they think is espe-
cially interesting or important to understand. The a. Have your social or cultural values influenced
groups then come up with a research question or your choice of subject matter? Explain.
questions that could be used as a focal point for b. Can your topic be researched scientifically?
research. For example, “What causes eating disor- Explain. (Hint: Think of the different ways in
ders?” Or, “What is it like to be a victim of domestic which social researchers define “science.”)
violence?” The students can pick any topic they like. c. Explain how a positivist, an interpretive re-
Each group then explains how they would re- searcher, and a critical social scientist would
search the topic they have chosen. They are to conduct research on the topic you’ve chosen.
do that by answering the following questions, Provide as much detail as possible.

bel65796_ch01_001-028 26 01/13/22 12:09 PM

 1 General Research Orientations 27

d. Make up some results that researchers in group can choose either “a” or “b.” Each group is
each of the three approaches to science given 20 minutes to prepare its case.
might produce. In other words, pretend that a. The social researcher’s most important duty
you did the research in each of the three ways is to explain the social world, not to change
and came up with some findings. You may it. The search for practical applications such
have a different set of findings for each of the as positive social change is best left to other
three approaches. Use your imagined find- people because it may create political bias on
ings to generate three answers to your re- the part of the researcher, such as when find-
search question(s), one for each of the three ings are ignored or rejected simply because
approaches to science. they come from people who do not share the
Once those tasks have been completed, each researcher’s political views.
group is to select a spokesperson(s) to report b. The social researcher’s most important duty
their group’s work to the rest of the class. Those is to make the world a better place by ridding
listening are to ask questions and offer a critique society of things such as racism, economic
of the group’s presentation. inequality, and sexism. Research that does
not have positive social change as its primary
2. The class is divided into three groups in order to goal is not worth doing.
conduct a debate. The first group must make the
After all groups have made their arguments,
case for position “a” below; the second group
each one is allowed time for rebuttal. When that
argues in favour of position “b”; and the third
is finished, the class votes to decide which group
made the strongest case.

Relevant Websites
A good way to become familiar with how research [Link]/core/journals/canadian-journal-
is done is to read articles published in academic of-political-science-revue-canadienne-de-science-
politique
journals. You may not be able to fully understand
the methodologies and data analyses at this stage Canadian Social Work Review
of your academic career, but reading articles will [Link]
help you become familiar with formal research
Native Social Work Journal
presentations. Below are the Web addresses for
[Link]
some leading Canadian and international social
science journals. You may not be able to access Canadian Journal of Education
articles in a particular journal directly; it may be
[Link]
necessary to sign on to your institution’s library
server first. American Journal of Sociology
[Link]/toc/ajs/current
Canadian Journal of Sociology
American Sociological Review
[Link]
[Link]
Canadian Review of Sociology
Social Forces
[Link]/canadian-review/
[Link]
Canadian Political Science Review
Qualitative Sociology
[Link]
[Link]/social+sciences/journal/11133
Canadian Journal of Political Science

bel65796_ch01_001-028 27 01/13/22 12:09 PM

28 PART I Fundamental Issues in Social Research

British Journal of Sociology The websites below offer information and guidance
[Link] on Indigenous research methods:

Sociology National Indigenous Research and Knowledges

[Link] Network

International Journal of Sociology [Link]

[Link]/loi/mijs20#.VVtagvlVhBc Indigenous Research Methodologies

Here are some websites you can visit to learn more [Link]

about participatory action research: Indigenous Research Resources, compiled by

Lakehead University
International Institute for Environment and
[Link]
Development research-services/resources/indigenous
[Link]/participatory-learning-action-pla
FAQ on Doing Indigenous Research in a Good Way,
Community-Based Research Canada offered by Memorial University
[Link] [Link]

The Tavistock Institute (Websites accessed 4 September 2021.)

[Link]

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch01_001-028 28 01/13/22 12:09 PM

 2 Research Designs

Chapter Overview
A research design is a framework for the collection and analysis of the data that is used to answer research
questions. It must satisfy certain criteria, and the form it takes depends on the research questions being
asked. This chapter will discuss four prominent research designs:
• experimental and related designs (such as the quasi-experiment)
• cross-sectional designs, including those involving survey research and some forms of qualitative
interviewing
• longitudinal designs, such as panel and cohort studies and various forms of qualitative research
• case study designs

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Discuss how and why some research designs are used to produce nomothetic explanations, while others
are used to come up with idiographic explanations.
2. Explain the structure and logic of the experimental method, and describe how a knowledge of this
method can aid in the understanding other research designs.
3. Identify the strengths and weaknesses of cross-sectional designs.
4. Describe the purposes of using a longitudinal design.
5. Explain how the case study approach is used in research.

▲ FatCamera

bel65796_ch02_029-052 29 01/13/22 12:14 PM

30 PART I Fundamental Issues in Social Research

“Would you tell me, please, which way I ought to It’s not hard to feel a little like Alice when you start
go from here?” out on the journey that is a research project, es-
“That depends a good deal on where you want pecially if the voyage is one you are somewhat re-
to get to,” said the Cat. luctant to undertake (this is not at all uncommon).
“I don’t much care where—” said Alice.
Where should I be going with this study? How can
“Then it doesn’t matter which way you go,” said
I get there? How can I tell if I’m on the right track?
the Cat.
Such questions flood the mind at the beginning of
“—so long as I get somewhere,” Alice added as
an explanation. the trip. As Alice learned, finding the right way starts
“Oh, you’re sure to do that,” said the Cat, “if you with knowing where you want to end up. In research,
only walk long enough.” a lot depends on what you want to accomplish with
Alice’s Adventures in Wonderland the project, which in turn depends on the sort of
(Carroll, 1865/2009, p. 56) research question you have posed. In this chapter
we’ll explore what kinds of destination are possible,
some ways of getting to them, and some techniques
for determining whether you are on the right path.

Introduction may explain that result with a neo-Darwinian theory

of erotic attraction and sexual disgust. This sort of
Once a general research orientation has been chosen, explanation is supposed to apply to humanity in
the next step is to select a research design. A research general, not just to the people who participated in
design is a broad structure that guides the collection the study. This approach to explanation is called
and analysis of data. Choosing a research design in- nomothetic.
volves decisions about what you want to accomplish Nomothetic explanations have to satisfy three cri-
with the study. Is one of your goals to describe pos- teria of causation. In other words, three conditions
sible causal connections between variables? Will have to be satisfied before the explanation is consid-
you explore how a social phenomenon changes over ered acceptable. Suppose you are trying to explain
time, and how it may be linked to other events and why some people are more violent than others, and
situations? Will it be important to search out the your explanation involves the idea that the propen-
meanings that the research participants attach to sity to be violent is influenced by watching violent
certain things, and how they act on those meanings? movies. For this to be accepted as a causal explana-
Do you want to be able to generalize your findings to tion, it would have to meet the following criteria:
people and groups that were not part of the study?
How these questions are answered will affect your 1. Correlation. The proposed cause and the
choice of research design. proposed effect have to vary together. In our
Another important consideration in choosing a example, as the number of violent movies
research design is the kind of explanation you would watched changes, the level of violence would
like to develop. Quantitative researchers often ex- have to change as well. You would have to ob-
plain a phenomenon in terms of proposed causes and serve that, in general, the greater the number
effects that are expressed in laws and principles. Such of violent movies watched, the higher the level
laws and principles are usually fairly general, and are of violence.
meant to apply to people who were not part of the 2. Time order. The proposed cause must precede
study. For example, a study that finds that men are the effect in time. Assume that you do find a
more likely than women to approve of pornography correlation between watching violent movies

bel65796_ch02_029-052 30 01/13/22 12:14 PM

 2 Research Designs 31

and real violence. You would also have to show researchers, on the other hand, aim for different
that the increase in the watching of violent kinds of explanations. They usually seek to pro-
movies came before the increase in violence. If duce a rich description of a person or group based
you found that the increase in the watching of on the perceptions and feelings of the people
violent movies occurred after the level of vio- studied, rather than to discover general laws and
lence went up, your explanation would have to principles. They may also try to get a sense of how
be rejected. those phenomena develop over time and how they
3. Non-spuriousness. Alternative explanations for can result from interpersonal interactions or one’s
the correlation observed have to be ruled out. position in a power hierarchy. These sorts of ac-
“Spurious” literally means false or illegitimate. counts are called idiographic explanations, and
A common source of spuriousness is a variable they usually involve a detailed “story” or descrip-
that influences both the proposed cause and the tion of the people studied that is based on empa-
effect. Suppose our data reveal that people who thetic understanding. Bowen (2015), for instance,
are violent tend to watch a lot of violent movies, used this approach to determine how off-street sex
and that people who reject violence watch very workers in Vancouver viewed the experience of
few violent films, i.e., that the two variables transitioning from the sex industry into “square”
are correlated. Can we conclude from this that (non-sex) occupations and lifestyles. On the basis
watching violent movies causes people to be of the insights gained from the study, she revealed
more violent? In order to answer that question, that people who are in the process of leaving the
we would try to determine whether some third industry or who are contemplating leaving are
factor is the cause of the correlation between often stigmatized and humiliated for their past be-
the number of violent movies watched and the haviours, and that the voices of such people must
level of violence. For instance, could it be that be heard if government transition programs are
people with low impulse control are drawn to to be effective. Qualitative researchers will choose
violent movies and tend to commit acts of vi- research designs that will produce these sorts of
olence? If so, then the level of impulse control idiographic explanations.
affects both the watching of violent movies (the Once a design has been selected, a specific method
original proposed cause) and the level of vio- for collecting data has to be chosen. There are many
lence (the effect). If that is the case, the correla- different ways of gathering data. One way is to
tion between watching violent movies and the use a preset instrument, such as a self-­completion
level of real violence is spurious, and your ex- questionnaire or a structured interview schedule.
planation does not hold up. As a further illus- Another is to utilize a less formalized method like
tration of spuriousness, imagine that you live participant observation or ethnography in which
in a summer resort community where there is the researcher takes part in the activities of a group
ample opportunity for swimming. Throughout of people, sometimes even living among them
the calendar year, there is probably a correla- for a time.
tion between the number of ice-cream cones Consider one of the research designs to be covered
sold on a particular day and the number of in this chapter: the case study. It entails a detailed
people swimming: as one increases the other exploration of a specific case, which could be a com-
also increases. But does purchasing ice-cream munity, an organization, a person, or an event. Once
cones cause people to go swimming? Of course you’ve selected a case to investigate, how will you get
not. Those two things are influenced by a third data on it? Do you use participant observation? Do
factor: warm weather. you observe from the sidelines? Do you conduct in-
terviews? Do you examine documents? Do you ad-
Quantitative researchers choose research designs minister questionnaires? Any or all of these methods
that help to satisfy these three criteria. Qualitative can be used.

bel65796_ch02_029-052 31 01/13/22 12:14 PM

32 PART I Fundamental Issues in Social Research

Research designs whether the country came into existence through

revolution (for example, Lipset, 1990). In order to
Experimental design test this idea experimentally, revolution would have
When people hear the word “research,” the use of to be induced in some randomly selected areas to
experiments often comes to mind. However, true ex- produce new countries, while in others no revo-
periments are fairly rare in sociology, though they lution would be fomented. The researcher would
can be found in areas such as social psychology and then compare the national characters of the revolu-
studies of organizations. Researchers in social policy tionary countries with the non-revolutionary ones.
may also use them to assess the impact of reforms or Needless to say, such experimental manipulations
new policies. The main reason for discussing exper- would usually be impossible to carry out.
iments here is that a true experiment involves a sys- This example also illustrates a second reason why
tematic comparison of what happens when one set of experiments are so rare in sociology: ethical concerns
participants has a particular experience or treatment often preclude them. Suppose you are interested in
while another does not, a logic that can also be found the effect of poverty-induced poor nutrition on chil-
in other research designs. The greatest strength of dren’s academic performance. In order to examine
experiments is that they can in some instances be this issue experimentally, you could select a number
effective in establishing causation (sometimes re- of six-year-olds and randomly assign them to one of
ferred to as internal validity), especially when they two groups. Those in the first group would be under-
are used to isolate the effects of a small number of fed for six months, while those in the second would
factors. This does not mean that the experimental be given ample, nutritious diets. Then you would test
method is necessarily superior to the others. As we the two groups to see if they differed in a­ cademic
shall see below, experiments, like other methodolog- achievement. Although sound from a purely meth-
ical approaches, have their strengths and weaknesses odological standpoint, such a study would be hid-
when they are used in social research, and are better eous from an ethical perspective. (If conducting
suited to some research topics than others. such a study sounds farfetched, Chapter 3 examines
Canadian experimental research from the 1940s
and 1950s in which serious malnutrition among
Manipulation Indigenous people was allowed to continue for two
If experiments may be strong in establishing years in order to establish baseline measurements
causation, why do social researchers not make for a study on the effectiveness of various dietary
more use of them? The reason is simple: to con- interventions.) In fact, as we will see in Chapter 3,
duct a true experiment, it is necessary to do some- ethical issues may arise even when the experimental
thing to people and observe the effects. To put the manipulation appears to be innocuous and is of very
matter more formally, an experiment manipulates short duration.
an independent variable (a proposed cause of Another reason why experiments are uncommon
some phenomenon) to determine its influence on a in sociology is that many of the things of interest to
dependent variable (the proposed effect). Typically, sociologists—gender roles, political preferences, the
some people are allocated to a “treatment” group formation of social movements, and so on—have
in which the independent variable is changed or complex, long-term causes that cannot be easily
manipulated, while others are placed in a “control” simulated in experiments. Could the experimental
group where no manipulation takes place. The de- method be used to explain the rise of second-wave
pendent variable is then observed and measured. feminism in the 1960s? Unfortunately not: exper-
The problem is that many of the independent vari- iments are generally limited to relatively simple,
ables of concern to social researchers cannot be ma- short-term manipulations of independent variables
nipulated. For example, some sociologists maintain (see Brannigan, 2004). A further reason is that
that a country’s national character is affected by even where social scientists are successful in using

bel65796_ch02_029-052 32 01/13/22 12:14 PM

 2 Research Designs 33

experiments to identify causal variables, the percep- that the teachers’ expectations that the spurters
tions and feelings of the participants—which pro- would show superior academic performance ac-
vide information vital to a full understanding of the tually caused those students to do better than the
phenomenon in question—are usually not examined others, presumably because they received differ-
in depth (Dobash & Dobash, 2000). ential treatment from the teachers. But here again,
Before moving to a more complete discussion of ethical issues arise. Was the study fair to the stu-
experimental designs, it is important to introduce a dents designated as non-spurters? Should their
basic distinction between laboratory and field exper- teachers have been given false information indicat-
iments. The former take place in artificial settings, ing that they were not especially gifted students?
whereas the latter occur in real-life surroundings Did their participation in the study impede their
such as classrooms and factories. The Rosenthal academic development?
and Jacobson (1968) study described below is a well- The Rosenthal and Jacobson study includes
known example of a field experiment. most of the essential features of what is known as
the classical experimental design. The people in
Classic experimental design the study (in this case, students) are randomly as-
Rosenthal and Jacobson (1968) tried to determine signed to two groups. The experimental manipu-
what effect teachers’ expectations had on their lation (different levels of the independent variable,
students’ academic performance. In addition to il- here heightened teacher expectations) is carried
lustrating the classical experimental design, their out on the experimental group or treatment group
study is a good example of research that gets a lot (the spurters). The other group is not given the
of attention, is then subject to intense method- treatment and thus forms a control group. The
ological scrutiny (see Brannigan, 2004, pp. 80–89 dependent variable—academic performance—is
for critiques), and ultimately provides the impetus measured before the experimental manipulation to
for further research designed to improve on the make sure that the two groups really are, on aver-
original. age, equal at the start (see Figure 2.1). If they are
The research was conducted in a US school equal, and because of random assignment they
where many poor and minority-group children should be, the researchers can feel confident that
were enrolled. In the spring all the students were any differences in student performance found be-
given a test and told it was designed to identify tween the two groups after the manipulation were
“spurters”—that is, students who were likely to ex- due to the treatment. Everything else about the two
perience a sudden improvement in their academic groups is presumed to be the same, leaving differ-
performance. At the beginning of the following ac- ences in teacher expectations as the only possible
ademic year, the teachers were given the names of explanation for any differences that are found be-
the spurters in their classes. But this was a ruse: the tween the spurters and non-spurters.
students identified as spurters were simply selected
at random; they were not chosen on the basis of their
test performance. The test was r­e-­ administered 8 months
T1 T2
eight months after the original one, allowing the Experimental
group Obs1 Exp Obs2
authors to compare the so-called spurters with spurters IQ Teacher IQ
the other students on things such as IQ scores, expectancies
reading ability, and intellectual curiosity. Since Random
assignment
Rosenthal and Jacobson believed that there was no
initial difference in ability between the spurters Control Obs3 No Exp Obs4
and the others, any improvements were attributed group IQ No teacher IQ
non-spurters expectancies
to the fact that the teachers had been led to expect
the spurters to perform better. The authors report FIGURE 2.1 Classical experimental design

bel65796_ch02_029-052 33 01/13/22 12:14 PM

34 PART I Fundamental Issues in Social Research

To capture the essence of the classical experimen- • History. This refers to events other than the
tal design, the following simple notation is used: manipulation of teacher expectations that
might have caused the spurters’ scores to rise.
• Obs: an observation made of the dependent For example, suppose that the school’s princi-
variable. There may be more than two obser- pal had taken steps to raise standards in the
vations but, to simplify, the design shows the school. Without a control group, we could not
most common configuration: pre-test and be sure whether it was the teachers’ expecta-
post-test; here, IQ test scores before the exper- tions or the principal’s action that produced
imental manipulation and after. the increase in spurters’ grades. With a control
• Exp: the experimental treatment (indepen- group, we can say that the principal’s action
dent variable); here, the creation of teacher should have had an effect on the people in that
expectancies. No Exp refers to the absence of group too, and therefore that the differences
an experimental treatment and represents the between the experimental and control groups
experience of the control group. can be attributed to the effect of teacher expec-
• T: the timing of the observations made in re- tations alone.
lation to the dependent variable; here, the time • Testing. This threat refers to the possibility that
when an IQ test is administered participants may become more experienced at
taking a test or sensitized to the aims of the
What is the purpose of the control group? Surely it experiment as a result of the pre-test. The pres-
is what happens to the spurters (the experimental ence of a control group, which presumably
group) that really matters. For this study to be a would also experience the same things, dimin-
true experiment, however, it must control for (in ishes this possibility.
other words, eliminate) rival explanations of its • Instrumentation. This threat refers to the possi-
causal findings, leaving teacher expectations as the bility that changes in the way a test is adminis-
only factor that could have created any differences tered can account for an increase (or decrease)
in performance between the two groups. Of course, in scores between a pre-test and post-test; for
student performance is a complex phenomenon, example, perhaps the teachers know their stu-
with many causes, but the present study wanted to dents better or are more friendly the second
examine only one of those causes (teacher expec- time they give the test. Again, if there is a con-
tations). The presence of a control group and the trol group, the people in that group should be
random assignment of participants to the experi- affected as well.
mental and control groups help to eliminate rival • Mortality. A particular problem for studies
explanations for differences in academic perfor- that span a long period of time is the risk that
mance, which in this case are any explanations some people will leave the experiment before it
other than different teacher expectations. To see is over: for example, some students might move
this, consider some threats that would pose seri- to a different school, or experience a long-term
ous challenges to the study’s conclusions if there illness. Since this problem is likely to affect
were no control group or random assignment. The the control group too, it may not make a dif-
following list is based on a book written by Cook ference to the results. However, experimenters
and Campbell (1979). In each situation, the pos- should try to determine whether mortality has
sibility of a rival interpretation of Rosenthal and affected the experimental and control groups
Jacobson’s findings is proposed, but the presence differently.
of both a control group and random assignment • Maturation. Quite simply, people change over
greatly reduces the threat. As a result, confidence time and the ways in which they change may
in the finding that teacher expectations influenced have implications for the dependent variable.
student performance is enhanced. The spurters might have improved anyway as

bel65796_ch02_029-052 34 01/13/22 12:14 PM

 2 Research Designs 35

they got older, regardless of the effect of teacher the identification of some schoolchildren as spurt-
expectancies. But the control group would also ers enough to create the conditions needed for the
mature, so maturation effects cannot explain hypothesis about teacher expectations to be tested?
the differences between it and the experimen- The study depended on the teachers’ remembering
tal group. the “information” they were given about the spurters
• Selection. When participants are not assigned for the duration of the experiment, but it is possible
randomly to the experimental and control that as time went on some of them came to doubt or
groups, variations between them in the post- forget it; in that case, the manipulation would have
test may be due to pre-existing differences be- been contaminated.
tween the two groups. For example, if all the A second set of issues concerns the generaliz-
best students were given the spurter label, it ability of a study’s findings. In other words, do the
might have been their pre-existing academic study’s conclusions apply to other people, settings,
ability rather than teacher expectations that or time periods? These matters pertain to the study’s
caused them to do better. However, since a external validity. Cook and Campbell (1979) iden-
random selection process was employed here, tified five major threats to the external validity and
the “selection” risk is greatly reduced. With hence the generalizability of findings derived from
random assignment, the top students would an experiment:
make up roughly the same proportion of the
spurter group as the non-spurter group, can- • The representativeness of the study participants.
celling out the effect of pre-existing academic To what social and psychological groups can a
ability. However, even with the use of random finding be generalized? Can it be generalized to
assignment, if the number of people in each a wide variety of individuals who differ in eth-
group is relatively small, there is still a risk of nicity, social class, religion, gender, and type
pre-­existing differences between the experi- of personality? In the Rosenthal and Jacobson
mental and control groups. study, the students were largely from poorer
groups and a large proportion from ethnic mi-
Even if all these threats have been overcome, fur- norities. This may limit the generalizability of
ther issues may arise. First there is the question of the findings.
whether the variables used in the study have been • The effects of the setting. Can the results of a
adequately measured. This refers to the matter of study be applied to other settings (in Rosenthal
measurement validity, a topic addressed in detail and Jacobson’s case, to other schools)? There is
in Chapter 4. In the case of the Rosenthal and also the wider issue of whether expectation ef-
Jacobson (1968) study, there are potentially two as- fects can also be discerned in non-educational
pects to this question. First, has academic perfor- settings.
mance been adequately measured? Reading scores • History effects. This threat raises the question
seem to correspond to what they are supposed to of whether the findings can be generalized to
be measuring. However, given the controversy sur- the past and into the future. The Rosenthal and
rounding IQ tests and what they measure, we may Jacobson research was conducted more than
feel unsure that gains in IQ test scores can be re- 50 years ago. Would the findings still apply
garded as strongly indicative of academic perfor- today? Also, their investigation was conducted
mance. Similarly, to take another of the authors’ at a particular juncture in the school year.
measures, is intellectual curiosity a valid measure Would the same results have been obtained if
of academic performance? Does it really measure the research had been conducted at different
what it is supposed to measure? points in the year?
Another question is whether the experimental • The effects of pre-testing. As a result of being pre-
manipulation really worked. In other words, was tested, people in an experiment may become

bel65796_ch02_029-052 35 01/13/22 12:14 PM

36 PART I Fundamental Issues in Social Research

sensitized to the experimental treatment, with and the timing of the creation of teacher expectan-
the result that their responses are affected as cies was different from the original study. Clairborn
they become more test-wise. Consequently, the failed to replicate Rosenthal and Jacobson’s findings,
findings may not be generalizable to groups casting doubt on the external validity of the original
that have not been pre-tested, and, of course, research and suggesting that the first two threats to
in the real world people are rarely pre-tested. external validity referred to earlier in this chapter
This may have occurred in the Rosenthal and may have limited the applicability of the findings.
Jacobson research, since all students were pre- Virtually all experiments in the social sciences
tested at the end of the previous academic year, involve deception of some kind. In the Rosenthal
so the students new to the school in the fall and Jacobson study, for example, the experimenters
(when the teachers were told about spurters) told the teachers that certain students were spurters,
presumably were not pre-tested. which simply was not true. But could the experi-
• Reactive effects produced by the experimen- ment have been carried out without any deception?
tal arrangements. People are frequently, if not For example, could the experimenters have told the
invariably, aware of the fact that they are par- teachers that they were testing the effects of teacher
ticipating in an experiment. Their awareness expectations on student achievement? Could they
may influence how they respond to the experi- have given the teachers a list of students and said,
mental treatment; for example, they may react “Pretend that these students are gifted, although
by behaving in a socially acceptable manner they are not any more gifted, on average, than the
rather than sincerely and spontaneously, students not on the list. We want to see whether you
which could affect the generalizability of the would treat such children differently, and whether
findings. Since Rosenthal and Jacobson’s study differential treatment affects their academic per-
participants do not appear to have been aware formance.” Surely, that sort of approach would not
that they were in an experiment, this problem have produced authentic behaviour on the part of the
is unlikely to have been significant. The issue teachers, and the teachers might well have refused to
of reactivity and its potentially damaging ef- participate under those conditions for ethical and
fects is a recurring theme in many types of practical reasons. Clearly, some form of deception
social research. was necessary for the experiment to work.
But, as suggested above, deception raises ethical
Then there is the question of replicability. A study concerns—it is basically a form of lying. Chapter 3
is replicable if others are able to repeat it and get the will discuss the ethical implications of using decep-
same results. Sometimes replications are conducted tion in social research, as well as other sorts of ethical
to make sure that the original research was carried dilemmas facing researchers in the social sciences.
out properly; this is especially important if a study’s
results do not match prior findings on the topic. For The laboratory experiment
replication to be possible, the initial researcher must One of the main advantages of laboratory over field
spell out all the research procedures in great detail. experiments is the researcher’s greater control over
Rosenthal and Jacobson laid out their procedures the research environment. In particular, it is easier
and measures in detail, and anyone carrying out a to randomly assign participants to different experi-
replication could obtain further information from mental conditions in the laboratory than in a real-life
them. Consequently, their research is replicable, al- situation, which enhances the researcher’s ability
though there has never been an exact replication. to establish nomothetic causation. For example,
Clairborn (1969) conducted one of the earliest rep- Walsh, Hickey, & Duffy (1999) were able to tell some
lications and followed a procedure very similar to randomly assigned university students in Eastern
Rosenthal and Jacobson’s, although the study was Canada that previous results on the mathematics
carried out in three middle-class suburban schools, test they were about to take showed that women

bel65796_ch02_029-052 36 01/13/22 12:14 PM

 2 Research Designs 37

perform less well than men on such tests. Other stu- that help the researcher establish causation. Several
dents, also randomly assigned, were not told this. different types of quasi-experiments have been iden-
The data showed that women scored lower than tified (Cook & Campbell, 1979). Not all of them can
men when informed of this “fact.” When told that be covered here, but a particularly interesting type
the test was to compare Canadians and Americans, are “natural experiments,” in which experiment-like
there was no gender difference. Without the chance conditions are produced by naturally occurring
to randomly assign participants, there would always phenomena or changes brought about by people not
be some doubt that the treatment (being told that doing research. When that occurs, researchers can
women score lower than men on the test) rather gather data in much the same way they do in experi-
than a pre-­existing difference between the two sets ments. For example, if an earthquake hit a particular
of participants actually caused the women to do less city, but a city of comparable size and composition a
well than the men. Also, laboratory experiments are short distance away was spared, the conditions for a
more easily replicated because they are less bound to natural experiment would be present. The effects of
a natural milieu that could be difficult to reproduce. the natural disaster on civic pride or attitudes toward
However, laboratory experiments also suffer from local political leaders could be measured by compar-
a number of limitations. Many of those shortcom- ing the two cities on those variables. However, in
ings arise from low external validity. For instance, natural experiments, it is usually impossible to ran-
the experimental setting may not mirror real-world domly assign participants to experimental and con-
experiences and contexts, despite the fact that the trol groups. The absence of random assignment casts
participants are very involved in most experiments doubt on any causal inferences, since the groups
and take them very seriously. In addition, the treat- may not have been equivalent on all relevant char-
ment effects may be unique to the people in the acteristics before the independent variable was in-
study; others may not react the same way. In the case troduced. For instance, in the earthquake example,
of Fisher and Ma’s (2014) study described in Box 2.1, there may have been some pre-existing differences
for example, the participants were drawn from an between the two cities before the earthquake struck
online panel and hence may not be representative of that contributed to post-earthquake differences. So
the general population, if only because their willing- if the stricken city showed an increase in civic pride,
ness to go online and their facility with the Internet perhaps that was not because of the earthquake, but
may have made their responses to the experimental because it had a more charismatic mayor than the
stimuli distinctive. They were also volunteers, who neighbouring city. However, the results of such stud-
generally differ from non-volunteers (Rosnow & ies are still compelling because they are real rather
Rosenthal, 1997, ch. 5). In addition, the fact that they than artificial interventions in social life, making
were given an incentive to participate (they were paid them high in external validity.
a small fee) might have further distinguished them St Helena in the South Atlantic provided a fas-
from others since not everyone would be equally mo- cinating natural laboratory for the examination
tivated to earn the reward given to those who partici- of various claims regarding the effect of television
pate. There was no effect of pre-testing because, as in violence on children when TV was introduced to
many experiments, the participants described in Box the island for the first time in the mid-1990s. The
2.1 were not pre-tested. However, it is quite possible findings—from video footage showing young chil-
that reactive effects occurred: the people in the study dren at play during school breaks, from diaries kept
knew they were in an experiment, and that may have by about 300 of the children, and from ratings by
affected their behaviour. teachers—suggested that the introduction of televi-
sion was not followed by an increase in the number of
Quasi-experiments aggressive acts observed (Charlton, Coles, Panting,
Quasi-experiments have some characteristics of the & Hannan, 1999; Charlton, Gunter, & Coles 1998).
experimental model but lack some of the features The researchers suggest that in environments such

bel65796_ch02_029-052 37 01/13/22 12:14 PM

38 PART I Fundamental Issues in Social Research

Box 2.1 A laboratory experiment

It is commonly recognized, both in scientific cir- picture was from a village in Africa that had just
cles and in everyday life, that there are tangible been struck by a mudslide, making her a home-
benefits associated with being physically attrac- less orphan. People experiencing the “low sever-
tive. It is also well known that attractive people ity” manipulation were told of the mudslide, but
are thought to possess more positive attitudes were informed that the young girl’s home and
and behavioural traits than the less attractive; parents were not harmed by it. The results indi-
hence the “beautiful is good” stereotype. Yet two cated that participants in the low severity con-
researchers at the University of Alberta (Fisher & dition who were presented with the “attractive”
Ma, 2014) found that under certain experimental version of the girl’s photograph showed lower
conditions, participants felt less empathy toward levels of empathy toward her than those shown
an “attractive” child. the less attractive image: the opposite of the
The researchers wanted to know whether the “beautiful is good” stereotype. In the high se-
attractiveness of children in need affects the em- verity condition, physical attractiveness had no
pathy that non-relatives feel toward them. In one effect on the level of empathy. The researchers
version of the study, the perceived attractiveness reasoned that when severe harm is not imminent,
of a young girl was manipulated by showing one people perceive attractive children to be more
group of participants a picture of her that had socially competent and hence more capable of
not been altered. A second group was shown taking care of themselves. But when children
a version of the picture that had been digitally are in grave danger or distress, those superficial
modified to make the girl appear less attractive. considerations are overridden by strong feelings
A “high severity” of need condition was created of compassion and a desire to help, regardless of
by telling some participants that the child in the how attractive the child may be.

as St Helena, where children are closely watched by genetic studies indicate that social and political atti-
the community and are expected to avoid violence or tudes are affected not only by societal and situational
aggression, television may have little effect on their factors, but also by genes (Alford, Funk, & Hibbing,
behaviour. 2005; Bell, Kandler, & Riemann, 2018). Additional
Another type of natural experiment can be found examples of opportunities for quasi-experiments are
in twin studies, which are commonly used in be- presented in Box 2.2.
havioural genetic research. Monozygotic (popularly Quasi-experimental designs have been partic-
known as “identical”) twins share 100 per cent of ularly prominent in evaluation research studies,
their genetic structure, whereas dizygotic (“frater- which examine the effects of organizational innova-
nal”) twins have in common only about 50 per cent tions such as a longer school day or greater worker
of the genes that vary between human beings. This autonomy in a plant (see Box 2.3). Sometimes the
natural difference allows researchers to estimate the results are surprising. A quasi-experimental investi-
strength of environmental and genetic influences gation on the effect of support for people who take
on a variety of attitudes and behaviours (Knopik, care of the elderly (Demers, 1996) showed that the
Neiderhiser, DeFries, & Plomin, 2017). The presence extra support made caregivers feel less depressed,
of greater behavioural or attitudinal similarities but more burdened. Demers was not sure why, al-
among monozygotic than dizygotic twins suggests though she speculated that perhaps the help was seen
a genetic influence. For example, recent behavioural as something else to be coordinated and managed.

bel65796_ch02_029-052 38 01/13/22 12:14 PM

 2 Research Designs 39

Box 2.2 Quasi-experiments

A common type of quasi-experiment compares the real social world, and as such have to meet
data collected before and after a policy shift by goals that are very different from those of experi-
government or industry. For example, the num- mental research. That makes it next to impossible
bers of car accidents before and after the low- for governments to randomly assign people to
ering of a speed limit can be compared. If the different conditions. Therefore, they usually have
number goes down after the speed limit has to make do with quasi-experimental evaluations
been lowered, the policy would appear to be a of policy changes. A criminologist may want to
success, but one cannot be sure because there randomly assign criminals to jail or home custody
is no control group not experiencing the change. and then compare the two forms of detention, but
Perhaps the change was caused by something the potential for some of the at-homes to reoffend
other than the reduction in the speed limit, such would be seen as too great a risk. Because minor
as increased media coverage of car accidents. criminals are more likely to get home custody
Another example might involve the installa- than are more serious criminals, a fair test is not
tion of cameras to detect speeding. If the cam- possible without random assignment. Similarly,
eras are placed only in randomly selected places when Canada abolished the death penalty, it
and not in others, the research changes from a was not for the purpose of conducting an exper-
quasi-experiment to a real one, since the loca- iment: abolition was a policy decision. But data
tions without cameras would constitute a control could be examined as if the change were part of
group. Governments find it difficult, however, to a ­quasi-experiment, for example, by comparing
subject some people to one condition and others the murder rate before and after capital punish-
to another. The prison system, schools, and other ment was abolished, even though the legislation
institutions controlled by governments are part of was not intended for that purpose.

The situation that Demers found herself in—not This division was not perfect, since the men in the
being sure how to interpret her findings—is actually first group may have differed in some relevant way
quite common. It’s one thing to come up with a set of from those in the second; the fact that some in the
findings, but another to explain them. The solution latter category received treatment is also an issue.
usually involves doing more research. In any case, the results showed that, after an average
Quasi-experiments are also used to evaluate the 12-year follow-up period, there were equal rates of
effectiveness of institutional policies. For example, reoffending in both groups: about 20 per cent for sex
Hanson, Broom, & Stephenson (2004) wanted to crimes, which was not very encouraging for the pro-
know whether treating sex offenders after they are fessionals who treat offenders.
released from prison has any effect on their chances A major challenge in evaluation research, par-
of reoffending. A total of 724 convicted male sex ticularly in Canada, is finding appropriate ways to
offenders were divided into two groups: those who assess government programs and services that affect
underwent mandatory treatment after release to Indigenous people. As noted in Chapter 1, research
community supervision, and those who were re- involving Indigenous participants has been fraught
leased prior to the implementation of the mandatory with exploitation and cultural insensitivity, and
treatment program. (Some of the latter group had re- the residential schools system in Canada is an ex-
ceived treatment during or prior to incarceration, al- ample of a program that illustrates in a very tragic
though the extent of that treatment was not known.) way what can go wrong if Indigenous perspectives

bel65796_ch02_029-052 39 01/13/22 12:14 PM

40 PART I Fundamental Issues in Social Research

Box 2.3 Evaluation research

feasible or ethical to randomly assign research

participants to the two groups, such studies are
usually quasi-experimental. For instance, data
gathered from people before a change may be
compared with data acquired after; the “before”
people become the control group, the “after”
Adam Melnyk/Shutterstock

people the experimental group. This approach

has the added advantage that the two groups are
basically the same, making random assignment
unnecessary.
Such a design was used to evaluate the effect
As a researcher studying the effects of a music of a community arts program on the well-being
program, what questions would you ask partic- of older adults in the Vancouver area (Phinney,
ipants before and after the program in order to Moody, & Small, 2014). Over three years, four
gauge its impact on their health? Would you focus groups of participants took part in the collective
on their objective health indicators, or the subjec-
creation of a physical work of art or a perfor-
tive perception of their own health? What might be
mance that was presented to the public. Baseline
the benefit of doing both?
quantitative measures of well-being were taken
in the first year of the program, with the same
A key question asked in evaluation research is measures administered again at its completion.
whether a new policy initiative or organizational The researchers also did a qualitative analysis of
change achieved its goals. Ideally, to answer that the project by conducting five focus groups at the
question the design would have one group that is conclusion of the program. The quantitative results
exposed to the treatment—the new initiative—and indicated that the participants had higher levels of
a control group that is not. Since it is often not perceived overall health and sense of community,

are ignored (Truth and Reconciliation Commission experiments reveals. At the heart of the experimental
of Canada, 2015). Evaluation research is only be- design is a logic of comparison which is used to iso-
ginning to come to terms with these sorts of issues. late the causal influence of a particular variable—a
Increasing attention is now being paid to the larger goal to which researchers using other designs may
structures of power, both historic and contemporary, also aspire. At the very least, an experiment allows
that give rise to various program designs, and it is the researcher to compare the results obtained from
increasingly recognized that Indigenous ontologies an experimental group with those from a control
and epistemologies must be respected (Shepherd & group. In the case of the Fisher and Ma (2014) exper-
Graham, 2020). iment in Box 2.1, the research compared the effects
of different levels of a child’s physical attractiveness
Significance of experimental design and distress on the level of empathy felt by an ob-
Although, as noted at the outset of this section, server. The advantage of such comparisons is that
true experiments are rare in sociology for a vari- they permit a better understanding of the phenom-
ety of reasons, it is important to draw attention to enon in question than would be possible if it were
a significant general lesson that an examination of examined under one condition alone. The argument

bel65796_ch02_029-052 40 01/13/22 12:14 PM

 2 Research Designs 41

and lower levels of physical pain, when the pro- there are several ways in which CCTV deters
gram was over. The themes that emerged from car crime. For instance, it serves as a direct
the focus groups included the conclusions that the deterrent to offenders (fewer people are will-
program provided the seniors with structure and ing to commit crimes if they think they will be
discipline, facilitated coping, required hard work filmed and caught), which encourages greater
and effort, brought out their artistic side, pro- usage of parking lots, which in turn increases
moted social involvement, and made a positive personal surveillance, which itself deters crime.
contribution to the community. Examples of contexts relevant to the interven-
Quantitative quasi-experimental designs in tion include time (such as when the parking lot
evaluation research go back a long way, but as fills up and empties during rush-hour periods,
the Phinney et al. (2014) study indicates, eval- or slow times during the day), and the avail-
uations based on qualitative research have ability of other nearby venues for offenders to
also emerged. Although there are differences commit car crimes. The kind of evaluation re-
of opinion about how qualitative evaluation search advocated by Pawson and Tilley maps
should be carried out, there is consensus on the the different combinations of cause and con-
importance of, first, understanding the context text in relation to different outcomes.
in which an intervention occurs and, second, Evaluation research is not limited to examining
hearing the diverse viewpoints of the stakehold- whether an initiative has achieved its stated goals.
ers (Greene, 2000). For example, Pawson and In fact, some approaches are “goal free” (Scriven,
Tilley (1997) advocate a pluralistic methodology 1991) in that the people doing the evaluation are
that examines not only the context but also the not told what the purpose of the program is. This
mechanisms that allow programs to work. Tilley allows evaluators to assess its overall impact, in-
(2000) provided an example of the approach in cluding unforeseen consequences, rather than
an evaluation of the use of closed-circuit tele- limit their attention to how well it achieved spe-
vision (CCTV) in parking lots. He observed that cific, preconceived goals.

that in certain situations greater physical attractive- Cross-sectional design

ness in children evokes less empathy is much more Many people associate cross-sectional designs
persuasive when the empathy felt for attractive chil- with questionnaires and structured interviewing.
dren can be compared with that elicited by those However, other data-gathering techniques may
perceived to be less attractive. also be used in cross-sectional research, includ-
That same logic of comparison is invoked in other ing structured observation and analysis of official
research designs. For example, cross-sectional de- statistics or diaries. These will be covered in later
signs of the kind associated with survey research, chapters, but we will outline the basic structure of
discussed in the next section, compare groups of the cross-sectional design here. In cross-sectional
people that differ in certain ways, and can be used studies, observations are taken at one point in
to evaluate causal hypotheses. For example, men and time—there are no before-and-after comparisons.
women may be compared in terms of their average Also, cross-sectional designs do not involve any
incomes to assess the hypothesis that after taking manipulation of the independent variable: they are
other influences into account, gender affects eco- like snapshots taken of a group or phenomenon at
nomic inequality. one point in time.

bel65796_ch02_029-052 41 01/13/22 12:14 PM

42 PART I Fundamental Issues in Social Research

A cross-sectional design entails the collection of Replicability, causal inferences,

data (usually quantitative) on more than one case. and external validity
Researchers are interested in variation between dif- How does cross-sectional research measure up
ferent people, families, nation-states, and so on, and in terms of replicability, the ability to establish
variation can be established only when more than causation, and external validity?
one case is examined. Researchers employing this
design usually select many cases, for at least two • Replicability characterizes most cross-sec-
reasons. For one thing, a larger number makes it tional research, so long as the researcher spells
more likely that variation will be encountered in all out the procedures for selecting respondents,
the variables of interest. A second reason is that cer- administering research instruments (struc-
tain statistical techniques are likely to require large tured interview or self-completion question-
sample sizes (see Chapter 7). naire, etc.), and the analysis of data.
In quantitative studies, data are collected on • Establishing causality can be problematic.
two or more variables, which are then examined to As was just suggested, it may be difficult to es-
detect patterns of association. This approach some- tablish causal direction from the resulting data.
times makes it difficult to show cause and effect Cross-sectional research designs can identify
because the independent and dependent variables associations that are indicative of causation,
are measured simultaneously, making any demon- but other designs may have to be employed to
stration of time order (showing that the cause ac- substantiate causal inferences.
tually precedes the effect in time) hard to prove. • External validity is strong when the sample is
For example, there is a well-supported negative re- a random one. When non-random methods
lationship between social class and serious forms of sampling are employed, external validity
of mental illness: poor people are more likely to becomes questionable, an issue addressed in
be mentally ill than rich people. But there is also a Chapter 7.
debate about the nature of that relationship: Does
being poor lead to stress and therefore to mental ill-
ness? Or does being mentally ill lead to difficulties Variables that cannot be manipulated
in holding down a job and thus poverty? Or is it As noted in the section on experimental design, in
a bit of both? To take another example, a study of much (if not most) social research it is not possible
1000 men found that those who had two or more to manipulate the variables of interest. This is a key
orgasms a week exhibited a 50 per cent lower mor- reason why most quantitative social research employs
tality risk compared with men who had on average a cross-sectional design rather than an experimental
less than one orgasm per week. It may be tempting one. To more or less all intents and purposes, things
to conclude that male orgasm leads to longer life ex- like ethnicity, age, and socioeconomic status are
pectancy, but it is also possible that the causal arrow “givens” and not really amenable to the kind of ma-
points in the other direction: men who are ill (and nipulation necessary for a true experimental design.
thus at greater mortality risk) are less likely to be Fictitious manipulations are possible, such as when
sexually active in the first place (Houghton, 1998, p. an experimenter digitally alters photographs to pro-
14). This finding and the preceding one are similar, duce different ages or ethnicities, perhaps to see the
showing what Blaxter (1990) called “an ambiguity effects on job offers, but the manipulation is limited
about the direction of causal influence.” There is to the external signs of age and ethnicity, missing the
only an association between the two variables—no more subjective and experiential aspects.
clear causal link. However, as will be shown below However, the very fact that certain variables are
and in Chapter 8, there are a number of ways for givens provides a clue as to how to make causal
researchers to draw cautious inferences about cau- inferences in cross-sectional research. Many of
sality using cross-sectional designs. the variables of interest can be assumed to be

bel65796_ch02_029-052 42 01/13/22 12:14 PM

 2 Research Designs 43

Methods in Motion Applications to Canadian Society

Why are some Canadians politically conservative and others liberal?
Bell, Aitken Schermer, & Vernon (2009) did a be- “nature” and “nurture” interact in the formation of
havioural genetic study of political attitudes political attitudes.
and behaviours that addressed the issue of why These sorts of studies are uncommon in so-
some Canadians tend to have conservative or ciology, political science, or anthropology, al-
right-leaning attitudes, while others are more though they are conducted more frequently in
liberal or left oriented in their political views. psychology and the health sciences. They have
The methodology used in the study (the details far-reaching implications for the social sciences,
of which go beyond the scope of this book) was in particular for disciplines in which biological
a kind of natural experiment comparing mono- influences are seldom examined. The standard
zygotic (identical) twins, who are essentially paradigm in the latter fields of study assumes
genetic clones of each other, with dizygotic (fra- that some combination of socialization, culture,
ternal) twins, who share only about 50 per cent of and situational factors can explain the origins
the genetic material that varies between human of most attitudes and behaviours. Researchers
beings. The researchers examined how similar using behaviour genetic methods, on the other
the identical twins were in their political attitudes, hand, acknowledge that those sorts of influences
and compared that with how similar the frater- are crucial, but argue that a more complete
nal twins were on that score. If identical twins are understanding of the human condition can be
more similar on the characteristic under examina- achieved by considering both social and biolog-
tion than the fraternal twins, that would indicate ical factors.
that genetic factors or other biological influences If behaviour genetics and other approaches
have some impact on the trait in question. The re- that consider both social and biological influences
sults of the study showed that the identical twins were to be more widely adopted in the social sci-
were in fact much more similar in their political ences, that could have a major impact on the
attitudes than the fraternal twins, which suggests theoretical perspectives that are developed
that genetic factors play a role in the formation and rise to prominence in academe. For exam-
of these attitudes. The study also found evidence ple, a blend of sociological conflict theories with
that social factors (such as educational influences neo-Darwinian evolutionary theory is one pos-
and media effects) affected political orientations, sibility (e.g., Sanderson, 2001). Such approaches
although it did not test for specific social influ- have the potential to promote consilience or the
ences. Taken as a whole, the results indicate that unity of knowledge (Bell & Kandler, 2017).

temporally prior to other variables. For example, The current discussion of the cross-sectional
in a relationship between ethnic status and alcohol design places it firmly in the context of quanti-
consumption, the latter cannot be the independent tative research. But qualitative research can also
variable because it occurs after ethnicity. Ethnicity use a form of cross-sectional design. For example,
still cannot be said with certainty to be the cause, Beardsworth and Keil (1992) carried out a study of
however: it is just a possible cause. Nonetheless, the dietary beliefs and practices of vegetarians. They
even though researchers are unable to manipulate administered “relatively unstructured interviews,”
things like ethnic status or gender, causal infer- which were “guided by an inventory of issues,” with
ences can still be cautiously drawn from cross-­ 76 vegetarians and vegans (1992, p. 261). The inter-
sectional data. views were taped and transcribed, yielding a large

bel65796_ch02_029-052 43 01/13/22 12:14 PM

44 PART I Fundamental Issues in Social Research

body of qualitative data. The research was not pre- Obs1 Obs2 Obs3 … Obsn
occupied with quantitative criteria such as estab- Case1
lishing causation, external validity, replicability,
Case2
measurement validity, and so on. Nonetheless, the
Case3
conversational interview style made the study more
Case4
externally valid than research using more formal in-
struments of data collection. Case5
The study was concerned with the factors that in- …
fluence food selection, like vegetarianism. The very Casen
notion of an “influence” carries a strong connotation
FIGURE 2.2 The data rectangle in cross-sectional
of causality, suggesting that qualitative researchers
research
can also be interested in the investigation of causes
and effects, although they do not use the language
of quantitative research with its talk of independent of causal influence” that plagues cross-sectional
and dependent variables. As well, the emphasis was designs. Because potential independent variables
much more on understanding the experience of can be identified at T1, the researcher is in a better,
something like vegetarianism than is often the case if not perfect, position to infer that the effects
with quantitative research. identified at T 2 or later occurred after changes oc-
This qualitative research bears many similari- curred in the independent variables. In all other
ties to the cross-sectional design in quantitative re- respects, the points made above about cross-sec-
search. It entailed interviewing quite a large number tional designs are the same as those for longitu-
of people at a single point in time. And, as with many dinal designs. In spite of its heightened ability to
quantitative studies using a cross-sectional design, show cause and effect, the longitudinal design is
the examination of people’s past and current eating not frequently used in quantitative social research
habits was based on the participants’ retrospective because of the additional time and cost involved.
accounts of factors that affected their past and pres- Longitudinal studies can take different forms.
ent behaviour. For example, Goyder, Guppy, & Thompson (2003)
examined how gender-influenced evaluations of
Structure of cross-sectional designs occupational prestige had changed over 25 years.
A cross-sectional design collects data on a series of Some of the earlier male advantage had disappeared;
variables (Obs1, Obs2, Obs3, Obs4, Obs5, . . . Obsn) for indeed, some occupations showing a female incum-
different cases (people, households, cities, nations, bent were rated more highly than the same occupa-
etc.) at a single point in time. The effect is to create tion with a male incumbent. Baer, Curtis, & Grabb
a data set that comprises variables Obs1 to Obsn and (2001) did a 15-nation study examining whether
cases case1 to casen, as in Figure 2.2. Each cell in the membership in clubs and associations had changed
matrix has data in it. over time. Kerr (2004) and Kerr and Michalski
(2007) investigated hyperactivity in Canadian chil-
Longitudinal design(s) dren as they grew older and its sources in poverty
With a longitudinal design, cases are examined and family structure.
at a particular time (T1) and again at a later time There are two basic types of longitudinal design:
or times (T 2 , T3, and so on), but without the ma- the panel study and the cohort study. With the former,
nipulation of an independent variable that char- the same people, households, or other groups are
acterizes experiments (see Box 2.4). When used studied on at least two different occasions. A panel
in quantitative research, it allows insight into the study, especially one at the household level, needs
time order of variables and is better able to deal rules for handling new entrants (e.g., as a result of
with the problem of “ambiguity about the direction marriage or elderly relatives moving in) and exits

bel65796_ch02_029-052 44 01/13/22 12:14 PM

 2 Research Designs 45

Box 2.4 Longitudinal research and the case study

Case study research frequently includes a longi- A longitudinal element also occurs when a
tudinal element. The researcher can be a partici- case that has been studied is returned to at a
pant observer in an organization for an extended later time. A particularly interesting example of
length of time, or may do ethnographic research this occurred in “Middletown,” a pseudonym for
with a community for many months or years, an American Midwest town first studied by Lynd
or may conduct structured or qualitative inter- and Lynd (1929) in 1924–5 and restudied in 1935
views with individuals over a prolonged period. during the Depression to see what changes had
Moreover, the researcher may be able to inject occurred (Lynd & Lynd, 1937). In 1977, the com-
an additional longitudinal element by analyzing munity was again restudied, this time in a post–
archival information and asking respondents Vietnam War setting (Bahr, Caplow, & Chadwick,
to recall events that occurred before the study 1983), using the same research instruments but
began, thus discovering some history. with minor changes.

(e.g., as a result of marital break-up or children leav- human geography, quantitative research using these
ing home). designs usually takes the form of repeated survey
One example of a panel study is the National research using a self-completion questionnaire or
Longitudinal Survey of Children and Youth structured interview. Used in this way, panel and
(NLSCY), described in Box 2.5, in which the same cohort studies can both illuminate social change and
Canadian children were studied in successive years. improve the understanding of causal influences.
Another is a three-wave study of family struc- Panel and cohort studies also share similar prob-
ture and children’s socioeconomic attainment in lems. First, there is the problem of sample attrition:
Canada conducted by Seabrook and Avison (2015), some participants may die, some may move away,
which found that mothers’ education and family and some may simply choose to withdraw at later
income were key to the children’s success. The stages of the research. For instance, in a study by
Research in the News box (page 46) provides an ex- Dinovitzer, Hagan, & Parker (2003) on the educa-
ample of a panel study that had to be re-designed tional attainment of Canadian immigrant youth, re-
rather suddenly. searchers were able to talk to only 65 per cent of those
In a cohort study people sharing the same experi- originally surveyed 19 years earlier. Comparing those
ence, such as being born in the same year or gradu- who participated in the second round with those
ating from a particular school at the same time, are who were lost from the study, the authors found no
studied over time, but the same people may not be significant differences. That sort of comparison is a
studied each time. For example, Walters (2004) used common practice; when no difference is found, the
the 1982, 1986, 1992, and 1995 National Graduates losses are treated as random and thus acceptable to
Surveys to examine trends in the economic fortunes ignore. The main problem with attrition is that those
of Canadians graduating from post-secondary insti- who leave the study may differ in some important
tutions. The same people were not selected for the respects from those who remain, so that the latter do
sample each year, but the information was still useful not form a representative group. However, there is
in understanding each graduating cohort, as well as some evidence from panel studies that attrition de-
the similarities and differences between cohorts. clines with time (Berthoud, 2000); in other words,
Panel and cohort studies share similar features. those who do not drop out after the first wave or two
In social sciences such as sociology, social policy, and of data collection tend to stay on the panel.

bel65796_ch02_029-052 45 01/13/22 12:14 PM

46 PART I Fundamental Issues in Social Research

Box 2.5 The National Longitudinal Survey of Children and Youth

This panel study involved a long-term effort to mon- by Human Resources Development Canada. The
itor Canadian children’s development and well-­ purpose of the study was to contribute to the de-
being as they matured from infancy to adulthood. velopment of policies that help children live healthy,
It began with a representative sample of children active, and rewarding lives (see Michaud, 2001).
11 years of age or younger in 1994–5 being inter- The data for the final cycle (Cycle 8), which began
viewed, with follow-ups every two years. Statistics in September 2008, were released in November
Canada collected the data, with direction provided 2010 (Statistics Canada, 2010).

Second, there are few guidelines for determining per cent of respondents attributed a change in the
the best timing for further waves of data collection. way they cared for relatives to their participation in
Finally, there is evidence of a panel conditioning the research.
effect, whereby continued participation in a longitu- As suggested above, qualitative research may
dinal study affects respondents’ behaviour. Menard also incorporate elements of a longitudinal design.
(1991) cited a study of family caregiving in which 52 For example, it is used in ethnographic research

Research in the News

The effects of COVID-19 on physical and mental health, family life, and employment
in Canada
Sociologist Scott Schieman, along with col- the next five years. Online questionnaires along
leagues and students at the University of with qualitative interviews are being used to
Toronto, had plans in the fall of 2019 to start a gather the data.
panel study of 2500 people that would exam- Some preliminary results have been surprising.
ine the participants’ employment and financial The data indicate that for certain groups, work-
experiences each year for the next decade. But life conflict actually decreased as the pandemic
things changed when the pandemic struck a few wore on. “We see the largest decrease, on aver-
months later. “We decided to change the design age, among people who don't have any children,”
of the study,” Schieman explained. “We realized Schieman said. But others had different experi-
that it was important to try to map and explain ences. “[A]mong those with children, we find the
changes in people’s experiences with employ- largest increase in work-life conflict among those
ment, work and economic life as the pandemic who have kids younger than six and between the
was unfolding, especially given the sweeping ages of six to 12. This seems to have unfolded
shifts in the broader economy and social interac- similarly for both men and women" (Canadian
tion” (Canadian Government News, 2020). Government News, 2020). He emphasizes that
A decision was made to add an additional over time, people’s lives may be impacted by
2500 people to the study, and to analyze the COVID-19 in ways that are still unknown, given
larger sample’s COVID-related experiences for that the pandemic has not yet run its course.

bel65796_ch02_029-052 46 01/13/22 12:14 PM

 2 Research Designs 47

when the ethnographer is in a location for a lengthy There is a tendency to associate case studies with
period of time or when interviews are carried out qualitative research, such as Chowdhury’s (2018)
on more than one occasion in order to examine study of Bangladeshi immigrants in Toronto.
change. In an example of the latter, Smith, Lister, Exponents of the case study design often favour
& Middleton (2004) described a study of the ex- methods like participant observation and unstruc-
periences of citizenship for 110 young people. tured interviewing, which are viewed as particularly
They were interviewed in depth in 1999 and then helpful in generating an intensive, detailed exam-
re-­
interviewed at two-year intervals to examine ination of a case. However, case studies can also use
changes in their lifestyles, feelings, opinions, and quantitative methods, or some combination of quan-
ambitions in relation to citizenship issues. Only titative and qualitative.
64 people participated in all three waves of data col- With a case study, the case is an object of interest
lection, illustrating the high rate of sample attrition in its own right and the researcher aims to provide
in this style of research. an in-depth elucidation of it. Sometimes it is diffi-
cult to distinguish case studies from other research
Case study design designs because almost any kind of research can be
The basic case study design entails a detailed and in- construed as a case study. Even research based on
tensive analysis of a single case. A case may be: a national, random sample of Canadians could be
considered a case study of Canada.
• a single community, as in Hughes’s (1943) What distinguishes a true case study is the goal
classic study of Drummondville, a textile of finding and revealing the features of the case.
town in Quebec; Chowdhury’s (2018) work on Collecting in-depth, often qualitative data that may
Bangladeshi immigrants in Toronto; or Pratt be unique to time and place is characteristic of this
and Valverde’s (2002) research on Toronto’s sort of research. Case studies are often idiographic in
Somali community; nature, seeking to provide a rich description of the
• a single family, as in Lewis’s (1961) study of the subject matter, like Shalla’s (2002) study of how Air
Sánchez family in Mexico; Canada’s customer sales and service agents became
• a single organization, such as the automobile victims of airline restructuring.
factory studied by Rinehart (1996), or a group When the predominant research orientation is
within an organization, like the nurses in a qualitative, a case study tends to take an inductive
Hamilton hospital researched by White (1990); approach to the relationship between theory and re-
• a person, as in Nemni and Nemni’s (2006) search. If a mainly quantitative strategy is adopted,
study of Pierre Trudeau or Foran’s (2010) book the research is often deductive, guided by specific
on Mordecai Richler; such research is charac- research questions derived from social theories.
terized by use of the life history or biographi-
cal approach (see Chapter 11); Measurement validity, causal inference,
• a single event, like the fight against locating a external validity, and replicability of
home for recovering addicts in a Richmond, case studies
British Columbia, neighbourhood (Huey, The question of how the case study fares on the
2003), or the Alberta election analyzed by Bell, research design criteria of measurement valid-
Jansen, & Young (2007); ity, establishing causation, external validity, and
• a state or province, as in Laplante’s (2006) replicability depends in large part on whether the
study of the rise of cohabitation in Quebec; or researcher feels these criteria are appropriate for
• a sector of the economy, such as the seafood their work. Writers of qualitative case study re-
processing industry in rural New Brunswick search tend to play down or ignore the salience of
examined in Knott’s (2016) research on differ- these factors (cf. Stake, 1995). Those influenced
ent types of mobile labour. by the quantitative research tradition see them as

bel65796_ch02_029-052 47 01/13/22 12:14 PM

48 PART I Fundamental Issues in Social Research

more significant, and usually try to develop case unique case and thus could challenge the then
studies that meet the criteria. popular nature-over-nurture hypothesis. She
One question that has generated a great deal of reported that, unlike adolescents in most other
discussion concerns the external validity or general- societies, Samoan youth did not suffer a period
izability of case study research. How can a single case of anxiety and stress in their teenage years. She
possibly be representative of other cases? For exam- explained this by their culture’s strong, consis-
ple, would the findings from a study of the Toronto tently enforced standards of conduct and mo-
police department be generalizable to all large urban rality. These factors were of interest because
police departments in Canada? The answer, of course, many readers thought they might contain les-
is that the findings probably cannot be applied to sons for the West.
other police departments. Case study researchers do • The revelatory case. The basis for the revelatory
not usually delude themselves into thinking that it is case exists “when an investigator has an oppor-
possible to identify typical cases that can represent tunity to observe and analyse a phenomenon
a class of objects, whether factories, mass media re- previously inaccessible to scientific investiga-
porting, police services, or communities. In other tion” (Yin, 1984, p. 44). This can happen when
words, they typically do not think that a case study previously unavailable evidence becomes ac-
is a sample of one. cessible, as was the case with certain KGB files
after the fall of the Soviet Union.
Types of case
Yin (1984) distinguished three types of case, each of Cases may also be chosen for mundane reasons, such
which relates to the issue of external validity: as convenience, and still provide an adequate context
for answering certain research questions or examin-
• The critical case. Here the researcher has a ing key social processes. To take a concrete example,
clearly specified hypothesis, and a case is Russell and Tyler (2002) studied a Girl Heaven store
chosen on the grounds that it will allow a in the UK (which caters to 3- to 13-year-old girls) not
better understanding of the circumstances because it was a critical or unique case, or because it
under which the hypothesis does or does not offered a context never before studied, but because of
hold. The classic study by Festinger, Riecken, its capacity to illuminate the links between gender
& Schachter (1956) of a doomsday/UFO cult and consumption and the commodification of child-
is an example. The fact that, contrary to their hood in modern society. Indeed, often it is only at a
dire predictions, the world did not end allowed very late stage in the research that the singularity and
the researchers to test propositions about how significance of the case becomes apparent (Radley &
people respond to thwarted expectations. Chamberlain, 2001).
What did cult members do when, after quit- As we have mentioned, one of the standard crit-
ting their jobs, leaving their homes, and wait- icisms of the case study is that its findings cannot
ing on a mountaintop, nothing happened? Did be generalized. Case study researchers argue stren-
they sneak down and move to another town? uously that this is not the purpose of their craft.
No: they decided that their faith had saved A valid picture of one case may be more valuable
humankind and that their new role was to tell than a potentially less valid picture of many. Their
others of that miracle so more people could be aim is to generate an intensive examination of
converted to their religion. a single case, which may or may not be used for
• The extreme, even unique, case; a common theoretical analysis. Pratt and Valverde (2002)
focus in clinical studies. Margaret Mead’s studied only Somalis and expressed a hope that
(1928) well-known (albeit controversial) study others would study other immigrant groups in
of growing up in Samoa seems to have been other places. Their central concern was the rich-
motivated by her belief that it represented a ness of the data and the quality of the theoretical

bel65796_ch02_029-052 48 01/13/22 12:14 PM

 2 Research Designs 49

reasoning that the case allowed. As we have noted, As illustrated in Box 2.6, comparative case study
sometimes case studies are primarily inductive, work is often quantitative in orientation.
used as information to generate theories. Other The strength of comparative designs is that they
times they may be deductive in nature, providing highlight the similarities and differences between
the data required to assess theories. cases, which can be used to assess or generate theo-
Problems can arise when the research involves ries. They exhibit certain features similar to experi-
comparison of two or more cases. Dyer and Wilkins ments and quasi-experiments, which also rely on the
(1991), for example, argued that a multiple–case capacity to establish comparisons.
study approach means that less attention is paid to
the specific details of a particular case, and more to
the ways in which multiple cases can be contrasted. Bringing research orientation
Furthermore, the need for comparison often leads and design together
the researcher to choose an explicit focus at the
outset, when it might be advantageous to adopt a Finally, we can bring together the two general re-
more open-ended approach. A preference for contex- search orientations covered in Chapter 1 with the
tual insight and a less structured research approach research designs outlined in this chapter. Table 2.1
is associated with a qualitative research strategy. shows the typical form associated with each

Box 2.6 Comparative research: Cross-cultural studies

Phenomena such as voting behaviour or crime different in terms of ridership, safety, cleanliness,
victimization in two or more countries can be and so on.
compared using the same research instruments, A strength of cross-cultural research is that it
seeking similarities and differences and a deeper helps to illustrate how social scientific findings may
understanding of social reality in different na- be culturally specific. For example, Wilson’s (2002)
tional contexts. At the very least such research examination of Ontario raves made frequent
supplies a replication. comparisons to the earlier rave scene in Britain.
Cross-cultural research is more expensive However, the UK scene was primarily an outgrowth
than other approaches. It also presents other of working-class struggles, whereas Canadian
problems. When using existing data such as offi- raves appealed more to middle-class, cultur-
cial statistics or survey evidence, the researcher ally alienated youths. Similarly, Baer et al. (2001)
must ensure that the variable categories and found that joining clubs and voluntary organiza-
data-collection methods are comparable. When tions increased toward the end of the last century
new data are being collected, the researcher in the US, West Germany, and the Netherlands,
must check that data-collection instruments (for was stable in Canada and 10 other countries, but
example, questionnaires and interview sched- decreased in Spain. Finally, Young and Dugas
ules) are translated properly. Even when transla- (2012) compared Canadian print media cover-
tion is carried out competently, there may still be age of climate change issues in English-language
a problem with insensitivity to specific national publications with those in written in French, and
and cultural milieus. For example, the London found that the different environmental and media
“tube,” the Toronto “subway,” and the Montreal cultures in the two language communities con-
“métro” differ in more than name: public tran- tributed to important differences in how climate
sit experiences in the three cities may be very change topics were narrated.

bel65796_ch02_029-052 49 01/13/22 12:14 PM

50 PART I Fundamental Issues in Social Research

TABLE 2.1 Research strategy and research design

GENERAL RESEARCH ORIENTATION
Research design Quantitative Qualitative
Experimental Typical form: Most experiments involve quantitative Experiments are not used in qualitative
comparisons between experimental and control research, although they may inspire
groups on the dependent variable. Example: the or be inspired by qualitative findings.
Rosenthal and Jacobson (1968) study discussed in For example, a qualitative study of
this chapter. the teachers who participated in the
Rosenthal and Jacobson (1968) study
discussed in this chapter would be
enlightening.
Cross-sectional Typical form: Survey research and structured Typical form: Qualitative interviews or
observation on a sample at a single point in time focus groups at a single point in time
are two forms; content analysis of a sample of are two forms; qualitative content
documents is another. Example: Box 1.3. analysis of a set of documents relating
to a single period is another. Example:
the Beardsworth and Keil (1992) study
of vegetarianism discussed in this
chapter.
Longitudinal Typical form: Survey research on a sample on more Typical form: These include
than one occasion, as in panel and cohort studies, is ethnographic research over a long
one form; content analysis of documents relating to period, qualitative interviewing on
different time periods is another. Example: Box 2.5. more than one occasion, or qualitative
content analysis of documents
relating to different time periods. Such
research is longitudinal when the main
purpose is to map change. Example:
the Lynd and Lynd (1929; 1937)
studies of “Middletown” discussed in
this chapter.
Case study Typical form: Survey research is conducted on Typical form: The intensive study done
a single case with a view to revealing important by qualitative interviewing of a single
features about its nature. Example: the Bell et al. case, which may be an organization,
(2007) study of the 2004 Alberta provincial election. person, family, or community.
Example: Box 1.4.

combination of research orientation and research contain elements of both designs. Such studies are per-
design, along with a number of examples. haps better conceptualized as longitudinal case studies
The distinctions are not always perfect. In particular, rather than as belonging to one category or another.
in some qualitative research it is not obvious whether a A further point is that there is no typical form in the
study is an example of a longitudinal design or a case qualitative research orientation/experimental research
study design. Life history studies, research that concen- design cell. Qualitative research in the context of true
trates on a specific issue over time, and ethnography, experiments is very unusual; a q ­uasi-experimental
in which the researcher charts change in a single case, design is a more realistic alternative.

bel65796_ch02_029-052 50 01/13/22 12:14 PM

 2 Research Designs 51

Key Points
• There is an important distinction between a gen- • Replicability, validity (measurement and exter-
eral research orientation (quantitative versus nal), and the ability to establish causation are
qualitative) and a research design. important criteria for evaluating the quality of
• The nomothetic approach to explanation involves quantitative social research.
discovering general laws and principles, and is • Four key research designs are experimental,
usually associated with quantitative research. cross-sectional, longitudinal, and case study.
• Nomothetic explanations must satisfy three cri- • Except for the experimental design, each one of
teria of causation: correlation, time order, and these can be used with either a quantitative or
non-spuriousness. a qualitative research orientation. Experiments
• Qualitative researchers usually take the idiographic are used with a quantitative approach.
approach to explanation, which entails creating a • Threats to the establishment of causation are
rich description of a person or group based on the of particular importance in non-experimental,
perceptions and feelings of the people studied. quantitative research.

Questions for Review (R) and Creative Application (A)

Criteria for the evaluation of social among 15-year-olds at summer camp. You decide
research to conduct an experiment, and have followed
R1 Explain the time order criterion of causation. proper ethics protocols. How could you conduct the
a1 A survey researcher finds that people with high experiment? Provide as much detail as possible.
self-esteem make more money than people with R6 What is a quasi-experiment?
low self-esteem. You are tempted to conclude A6 You want to know whether the legalization of
from this that self-esteem influences earning marijuana in Canada will affect national crime
power. But can time order be established using rates. How could you use the quasi-experimental
this design? Explain. method to research this issue? Explain.
r2 What is a spurious correlation?
Cross-sectional design
a2 You are at a dance party where a lot of alcohol
R7 What is meant by a cross-sectional research
is being served. You abstain from drinking, but
design?
notice that the people with the craziest dancing
A7 How could you use a cross-sectional design to
style are the most likely to go to the washroom
determine whether there is an association be-
and vomit. Should you conclude that dancing
tween the amount of time spent studying and
crazily induces vomiting? Explain.
grades? Assess the degree to which your method
r3 What is a nomothetic explanation?
can establish causality.
a3 Come up with a nomothetic explanation for why
students sometimes drop out of university. Longitudinal design(s)
r4 What is an idiographic explanation? How do R8 Why might a longitudinal research design be su-
qualitative researchers produce them? perior to a cross-sectional one?
a4 Assume that your best friend just dropped out of A8 How could a qualitative researcher use a longi-
university. Come up with an idiographic expla- tudinal design to study people active in a local
nation of how that happened. environmental movement?

Research designs Case study design

Experimental design R9 What is a case study?
R5 How are true experiments able to establish A9 Pick a particular case (it can be any person,
causal connections between variables? Explain. group, or event) and explain how a qualita-
A5 You want to know whether the amount of time tive researcher could study it. Then describe
spent on social media affects loneliness levels how a quantitative researcher could gather

bel65796_ch02_029-052 51 01/13/22 12:14 PM

52 PART I Fundamental Issues in Social Research

information on the case that is relevant to the Visit our online resource center for free assess-
findings of the qualitative investigator. ment: [Link]

Interactive Classroom Activities

1. This activity involves some friendly competition. the fire for the fire trucks and damage example
Divide the class into small groups. Each group is given above).
given 15 minutes to come up with as many exam- 2. Divide the class into small groups. Each group is
ples of spurious correlations as they can (e.g., the to
correlation between the number of fire trucks at a. pick a social phenomenon or experience of
a fire and the amount of fire damage). Examples interest (e.g., sex work, homelessness);
given in this chapter or in previous classes are not b. provide a nomothetic explanation for it; and
allowed. After the time is up, a spokesperson from c. provide an idiographic explanation for it.
each group presents the group’s spurious cor-
relations to the rest of the class, and the instructor The groups are to go into as much detail as pos-
judges whether each one is in fact an example of sible. When that is completed, each group is
a spurious correlation. The instructor also keeps paired with another group (A and B, C and D, E
a record of all groups’ accepted correlations. The and F, etc.) such that Group A presents its results
group with the most accepted correlations wins. to the rest of the class, while Group B has the
Once that has been done, the class exam- task of providing a critical evaluation (enumer-
ines each correlation individually. The instructor ating both strengths and weaknesses) of Group
asks for volunteers to name an antecedent third A’s arguments. Then the reverse: Group B pres-
factor that might be causing the two variables ents and Group A critiques. Repeat for Groups C
to vary together (e.g., it would be the severity of and D, E and F, and so on.

Relevant Websites
The National Longitudinal Survey of Children and The Human Genome Project site offers some basic
Youth site provides information on a longitudinal information on this 13-year project and its implica-
study in Canada. tions for understanding human behaviour.
[Link]/census-recensement/2011/ [Link]
ref/92-135/surveys-enquetes/nationalchildren- Genome/[Link]
[Link]
The Behavior Genetics Association, an interna-
The Online Dictionary of the Social Sciences site, tional organization, provides a forum for the dis-
maintained by Athabasca University, “has 1000 en- semination of behavioural genetic research and
tries covering the disciplines of sociology, criminology, offers methodological workshops.
political science and women’s study with a commit-
[Link]
ment to Canadian examples and events and names.”
(Websites accessed 4 September 2021.)
[Link]

The Research Methods Knowledge Base provides a de-

tailed discussion of various types of experimental design.
[Link]/kb/[Link]

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch02_029-052 52 01/13/22 12:14 PM

 3 Research Ethics 53

3 Research Ethics

Chapter Overview
Before research involving humans can begin, researchers in Canada and many other countries must
demonstrate that what they are about to do is ethical. This is no small task, given that ethical considerations
pertain to all research methods and come into play at every stage of research, from the recruitment of
participants to the publishing of the results. This chapter examines some general ethical principles and their
applications to real research situations. The backdrop for the discussion is the Tri-Council Policy Statement:
Ethical Conduct for Research Involving Humans (TCPS2) (2018), which governs most research practices
in Canada. The TCPS2 was prepared by the Canadian Institutes of Health Research (CIHR), the Natural
Sciences and Engineering Research Council of Canada (NSERC), and the Social Sciences and Humanities
Research Council (SSHRC). It identifies three core principles, which are complementary and mutually
reinforcing:
• Respect for persons stipulates that people’s autonomy and their freedom to choose what will happen
to them must be ensured. Research participants must therefore be made aware of what the study will
entail, so that their consent is not only freely given but fully informed. Consent must also be ongoing;
participants should be allowed to withdraw from the study at any time.
• Concern for welfare covers all aspects of a person’s life as well as the welfare of groups and commu-
nities. This includes the physical, mental, emotional, spiritual, social, and economic facets of life. To
protect the welfare of individuals and groups, the proper balance between the risks and benefits of
research must be sought.
• Justice requires that people be treated fairly and equitably, so that no segment of the population
bears an undue burden of risk associated with research, and no part of society is excluded from the
benefits that research may bestow.

Each of these principles provides valuable guidance and can be used to identify research activities and
practices that should be avoided for moral reasons.

▲ The Canadian Press/Nathan Denette

bel65796_ch03_053-076 53 01/13/22 12:19 PM

54 PART I Fundamental Issues in Social Research

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Articulate why it is necessary to consider the ethical implications of all aspects of the research process.
2. Explain how the ethical principles of respect for persons, concern for welfare, and justice can be imple-
mented in social research.
3. Review and assess the ethical dimensions of any research project.

You may be surprised to learn that there are In fact, the history of the social sciences is riddled
stringent, formal policies in place to ensure that with studies in which well-meaning researchers
research in Canada is conducted in an ethical harmed the people they studied in some way. In
manner. But are such policies necessary? Surely many cases the studies seemed to be quite ac-
researchers are decent people, not mad scien- ceptable when they were done, and it was only
tists who take pleasure in abusing the persons with the passage of time that the consequences
who participate in their studies. While it is true for participants were recognized. And in some
that those who do research are generally hon- instances, what was done was later deemed to
ourable individuals, every research situation has be not only unethical, but egregious. How is that
an ethical dimension to it, and sometimes the eth- possible? And why are ethical dilemmas still en-
ical implications of a procedure are not obvious. demic to social research?

Introduction any other goal. The same principle lies at the heart of
research ethics in the social sciences. What it means
As noted in Chapters 1 and 2, choosing a general re- is that the welfare of research participants should
search orientation and selecting a research design take priority over everything else, including the ac-
are matters that are normally dealt with very early in quisition of knowledge, even if that knowledge would
the research process. Another issue to be addressed expand our understanding of humanity or lead to
in the initial stages of a study, and one that should be improvements in the human condition. To some
kept in mind in every phase of the project, is research degree this principle conflicts with the concept of ac-
ethics. How the ethics of a study are handled affects ademic freedom. But minimizing the risks of partic-
the integrity of the project as well as the reputation ipation and striking an appropriate balance between
of the social sciences as a whole, so understanding the risks and benefits of research is compatible with
research ethics is crucial. If you have read the first social scientists’ freedom to choose the topics of their
two chapters of this book you will have a basic un- research and, within reasonable bounds, the meth-
derstanding of how research is conducted and will be ods to be used.
in a good position to grapple with the ethical dilem- It may seem obvious that research participants
mas that researchers typically face. should not be harmed by taking part in research
The first priority of a social researcher should activities, whether they be experiments, field re-
be to ensure that the people being studied are not search studies, or any other kind of social scientific
harmed by their participation. This idea is similar to investigation. As with many moral rules, however,
a fundamental precept among physicians and sur- knowing how to put this principle into practice is
geons: “First do no harm.” Notice the word “First”: anything but simple. For example, defining “harm”
it means that the prevention of harm should override and outlining its limits is more complicated than

bel65796_ch03_053-076 54 01/13/22 12:19 PM

 3 Research Ethics 55

you might think. We all know that research partic- in Canada use the TCPS2 in making their decisions
ipants should not be killed in the process of being and recommendations.
studied, but is any and all harm unjustifiable? What REBs may approve a research project, request
about milder forms of harm, such as short-term modifications, or reject it outright. If the research
anxiety or embarrassment? Is it morally acceptable is to be conducted over an extended period of time,
to subject people to temporary discomfort if the the REB may also hold periodic reviews, or at least
research has the potential to improve our under- require that the researchers file an annual report.
standing of society or human behaviour in general? In addition, REB members themselves are required
Similarly, most people would say that telling lies is to follow procedures outlined in the TCPS2. For ex-
generally not a good thing, in part because it may ample, they must avoid any conflict of interest, such
cause harm to the person who is lied to. But what as having a financial stake in a project that is up for
about situations in which some temporary decep- approval.
tion is necessary in order to create a research en- Van den Hoonaard (2001) has suggested that
vironment in which people feel free to behave in it may be more difficult to get ethics approval for
a natural or unguarded way? Typically, it is these qualitative research than for quantitative studies.
sorts of “grey areas” that are debated in the discus- Qualitative research is sometimes considered less
sion of research ethics, although very serious ethi- scientific than the quantitative kind, and ethics com-
cal violations have been known to occur. mittees may prefer the epistemology of the latter,
Each method of social research comes with its with its derived hypotheses and specific plans. This
own set of ethical issues. However, it’s worth con- can be a fatal problem, because if the REB rejects
sidering some general principles that pertain to all the research proposal, the work can neither receive
social scientific research. In Canada, most research funding nor be carried out. Also, whereas in quanti-
must comply with the TCPS2 (CIHR, NSERC, & tative research data are usually collected in one-on-
SSHRC, 2018). Different academic organizations one meetings, in qualitative research a whole group
may also have their own ethical policies and proce- may be under observation, including some people
dures. For example, sociologists in Canada are en- who might not want the attention. One REB allowed
couraged to follow the Code of Ethics of the Canadian a researcher to go ahead with her ethnographic work
Sociological Association. Research in Canada is also but required that she turn her head when someone
subject to Canadian law, in particular the Charter of who had not explicitly agreed to be part of the study
Rights and Freedoms. (See “Relevant Websites” at the entered her field of vision. This directive is an exam-
end of this chapter.) ple of what some researchers would consider to be an
Before a research project involving humans overzealous attempt to prevent harm.
begins—even before anyone is asked if they would Another example of possible REB overreach
like to participate in the study—a research ethics involving ethnographic research occurred in
board (REB) must review and approve the project. Vancouver, where researchers sought permission to
REBs are normally made up of researchers from a do observational fieldwork and in-depth interviews
variety of disciplines and usually include a com- with intravenous drug users (Small, Maher, & Kerr,
munity member from outside the institution under 2014; see Box 15.2). The ethnographers were told by
whose auspices the research will take place (which their REB, which cited safety concerns, that they had
in the case of social research is usually a university). to provide police with the names of the observers
If the institution with which the researchers are af- who would have contact with the drug users, a move
filiated administers research funds issued by any of that would have made it all but impossible to conduct
the three agencies that created the TCPS2 (CIHR, the study.
NSERC, & SSHRC), then its REB must abide by the Indeed, some researchers claim that the advent of
TCPS2 in making its judgments. What this means REBs has had serious, deleterious consequences for
in practice is that REBs at virtually every university the social sciences as a whole (e.g., van den Hoonaard

bel65796_ch03_053-076 55 01/13/22 12:19 PM

56 PART I Fundamental Issues in Social Research

& Tolich, 2014). The fact that TCPS2 acknowledges before they are placed in a study, but should also be
the right of researchers to appeal REB decisions is a informed of all the risks and benefits to which they
tacit admission that at times they can be overly re- would be exposed. To assess those risks and benefits,
strictive. In any event, a strong case can be made that prospective participants need a basic idea of what the
REBs perform a valuable function in protecting the study will entail. Special arrangements for consent
rights of research participants, as the discussion to have to be made if the person being considered for
follow will show. inclusion is not able to exercise sound judgment due
to their young age, cognitive impairment, mental
General ethical principles health, or physical infirmity. In such cases a trusted,
authorized third party is required to make the deci-
The first thing to note regarding general principles sion on their behalf.
is that researchers themselves may disagree on the In most quantitative studies and in some forms
ethical acceptability of certain research practices. of qualitative research, people who are weighing the
Indeed, there are disputes from the 1960s and 1970s pros and cons of participation are given an informa-
that are still being rehashed today, as we shall see. tion sheet or letter of information outlining what the
Also, although the TCPS2’s codes and guidelines research is about, how it will be conducted, and what
provide some direction, they often allow for a degree the risks and benefits of participation are. The doc-
of flexibility. Nonetheless, there is little disagreement ument normally includes the names of the people in
that recognizing the ethical implications of doing charge of the research, their institutional affiliation,
social research is crucial if informed decisions are to and contact information. It also typically contains
be made. assurances of confidentiality along with a descrip-
The TCPS2 outlines three core principles for the tion of how the data will be stored and the form in
conduct of research involving human participants: which the results are to be published. A sample from
respect for persons, concern for welfare, and justice. a research study in which one of the authors of this
Although they are listed separately, there is consid- book participated is shown in Figure 3.1.
erable overlap between them. Prospective participants should be encouraged to
think about the project; they should also be given a
Respect for persons chance to ask questions and seek clarification if nec-
The principle of respect for persons maintains that essary. If they have all the information required and
humans should not be treated as mere “objects” or are willing to participate, a consent form is some-
as a means to an end. Even if the end is a worthy times provided for them to sign. This form may in-
one, it cannot justify the use of unethical means. As clude a brief synopsis of the study and state explicitly
we noted above, even if the knowledge a researcher that research volunteers are free to leave the study
is seeking will be of benefit to all humankind, the at any time without penalty. Consent must be freely
welfare of the individuals and groups under study is given and cannot be coerced. Sometimes the consent
of paramount importance. Researchers must recog- form and the information document are merged into
nize that people have rights, including the right to one, as illustrated in Figure 3.1. A consent form taken
be treated with dignity by social researchers. Respect from a Canadian study involving qualitative inter-
for persons is arguably the most fundamental of all views conducted by university students is shown in
the principles considered here. Figure 3.2.
The starting point of respect for persons in re- In some forms of qualitative research, such as
search is the stipulation that the people being stud- ethnography and participant observation, provid-
ied must give free, informed, and ongoing consent ing information sheets and consent forms may be
to participate in the research. Most researchers impractical; in some cases, it might even defeat the
now adhere to the principle that potential partici- purposes of the study. In cases where a formal agree-
pants should not only be asked for their permission ment between the researchers and the informants

bel65796_ch03_053-076 56 01/13/22 12:19 PM

 3 Research Ethics 57

Letter of Information: A study of personality in adult twins

Dear _________:
You recently participated in a research project conducted by Dr. _________ at the University of _________,
in which you completed a number of questionnaires. We are very grateful for your participation in our re-
search. At this time, we would like to in invite you to take part in a further research study examining relation-
ships between different measures of personality.
If you agree to participate in this study, you will be asked to complete the following questionnaires: the
Toronto Alexithymia Scale, the Deadly Sins Scale, the AQ-9 (a measure of attitudes toward mental illness),
the Moral Foundations Questionnaire, the Political Issues Questionnaire, and the Self-Report Altruism Scale.
Each of these is a paper-and-pencil personality questionnaire, and each will take between approximately 5 to
15 minutes to complete.
Most of the items in the questionnaires require you to indicate the extent to which different statements are
descriptive of you or your twin. Please feel free to complete the items at your own pace, in your own home.
It is not necessary to fill out all of the questionnaires in one sitting. We do, however, ask that you fill out the
questionnaires by yourself rather than with your twin, as we are interested in your unique responses.
There are no known risks involved in participating in this study. Participation in this study is voluntary. You
may refuse to participate, and to not answer any questions that you do not feel comfortable addressing. You
may withdraw from the study at any time.
Once you have completed as many of the questionnaires as you wish to complete, please mail them back to us
in the provided self-addressed stamped envelope. The questionnaires will then be stored in a locked cabinet in
Dr. _________’s office until they have been scored. At that point, they will be shredded. Only Dr. _________
and his research assistant will have access to the questionnaires and to the response sheets. Your name will
never be given out nor will it appear in any written report about the study. All of your responses will be kept
completely confidential.
Once we have processed your questionnaires, you will be sent $25.00 and your name will be entered into a
draw for a chance to win one of ten $100.00 prizes. At the completion of the study, we will send you a summary
of the study’s findings, if you so request.
If you have questions regarding any aspect of this study, you are welcome to contact Dr. _________ by tele-
phone (xxx-xxx-xxxx) or by email (xxxxxx@[Link]). If you have any further questions about the manner
in which this study is being conducted or about your rights as a research participant, you may contact the
Director of the Office of Research Ethics at the University of _________ by telephone (xxx-xxx-xxxx) or via
email (xxxxxx@[Link]).
Completion and return of the questionnaires indicates your consent to participate in the study.
Thank you for your interest in this study. This letter is yours to keep for future reference.
Dr. _________, Professor of _________, University of _________

FIGURE 3.1 A Letter of Information

might signal a lack of trust, less formal approaches their involvement. Relatively minor transgres-
such as verbal assurances and ad hoc explanations sions of the principle of informed consent are
may be employed. Whichever method they choose, pervasive in social research: for example, not
researchers must provide REBs with documentation giving all the details about the study for fear of
outlining how informed consent will be obtained. contaminating people’s answers to questions.
The principle of free and informed consent seems • In ethnographic research, the researcher is
straightforward, but putting it into practice can be likely to come into contact with a wide spec-
difficult and usually involves some ethical trade- trum of people, and ensuring that absolutely
offs. Homan (1991, p. 73) observed that abiding by everyone has the opportunity to give informed
this principle “is easier said than done.” At least two consent is not practicable. Even when all par-
major points stand out here: ticipants in a certain setting are aware that
they are being observed, it’s unlikely that all
• It’s extremely difficult to give prospective par- of them will have been similarly (let alone
ticipants absolutely all the information re- identically) informed about the nature of the
quired to make an informed decision about research.

bel65796_ch03_053-076 57 01/13/22 12:19 PM

58 PART I Fundamental Issues in Social Research

Oral History Consent Form

Consent to Participate in an oral history interview being conducted by students enrolled in [course] at
____________ University. This is to state that I agree to participate in a program of research being con­
ducted by 45 students under the supervision of Dr. ________ of the Department of _________________
University ([telephone number], email: [email address]).
PURPOSE: I have been informed that the purpose of the project is to explore the history of those parts of
the _________ area that have been demolished or otherwise “lost” to us due to urban change. Students have
formed into teams of 3 to 5 and will be exploring specific sites of memory such as ________, stories of im-
migration and displacement, and the former [site]. We will be working closely with [organization] and the
[organization] which are developing exhibitions on the [site] and [topic] in [year], based in part on these
student projects.
PROCEDURES: The interview will be conducted at participants’ homes, at facilities provided by ________
University or at another appropriate place. Interviewers will record participants’ life stories using video, or
audio depending on the preference of the interviewee. Participants can choose to discuss any aspect of their
lives and they may refuse to answer any questions. Interviews normally take about 1.5 hours, but participants
may take as long as they would like and are free to stop at any time.
RISKS AND BENEFITS: Describing difficult experiences can be upsetting and emotionally difficult. As the stu-
dent projects will contribute to two exhibitions, as well as a website, with your permission, your story will be heard.
CONDITIONS OF PARTICIPATION: Please review the following conditions and options with the inter-
viewer. Feel free to ask questions if they appear unclear.
___I understand that I am free to withdraw my consent at any time during the interview and discontinue
from that point forward.
In terms of identification and reproduction of my interview, I agree to (please choose one):
___Open public access: My identity may be revealed in any publications or presentations that may result from
this interview.
___I agree to the possible broadcasting and reproduction of sound and images of my interview by any method
and in any media by participants of this research project. I consent that my interview, or portions of it, be made
available on the Internet through Web pages and/or online databases of the project.
___I agree that transcripts and/or recordings of my interview will be stored at a local archive for long-term
preservation. Your interview may be accessed by researchers and the public by viewing it at the Centre for
_______ and/or at a local ________-area archive holding the preservation copy.
OR
___ Anonymity: My identity will be known only to the interviewer and the course instructor; others will
not gain access to my identity unless they gain special permission from me, the interviewee. Once the stu-
dent project is completed in [month/year], the audio and/or video recording will be destroyed by the student
(though a copy of the interview may be given to you).
In cases where family photographs or documents are scanned or photographed:
___ I agree to let the student researcher copy family photographs and documents for use in the student project only.
OR
___ I agree to let the student researcher copy family photographs and documents for use in their assignment
and for their being archived with the interview recording. I likewise give permission to let future researchers
use these images in their publications.
I HAVE CAREFULLY STUDIED THE ABOVE AND UNDERSTAND THIS AGREEMENT. I FREELY AND
VOLUNTARILY AGREE TO PARTICIPATE IN THIS STUDY.
INTERVIEWEE:
NAME (please print)
SIGNATURE
Date and Birthplace (optional)
INTERVIEWER:
NAME (please print)
SIGNATURE
DATE:
If at any time you have questions about your rights as a research participant, please contact _________, Chair
of the Research Ethics Committee, Department of _______, _______ University at [phone number] or by
email at [email address].

FIGURE 3.2 A Consent Form

Adapted from consent form from Concordia University: [Link]
Urban%20Change%[Link]

bel65796_ch03_053-076 58 01/13/22 12:19 PM

 3 Research Ethics 59

Experiments are particularly problematic when concentration camps. It appeared that ordinary, ev-
it comes to informed consent because researchers eryday people were willing to inflict serious harm on
want the participants to behave naturally and au- innocent others if ordered to do so by an authority
thentically. If participants were to be fully informed figure—a finding that may have important impli-
regarding the purpose of the experiment, it would cations for the understanding of human atrocities
be difficult to avoid the reactive effects that occur and ultimately their prevention, although that in-
when people know they are being observed. For this terpretation of his findings has come into question
reason, experiments usually involve some deception. (Perry, 2012). Nonetheless, the fact remains that the
For example, Milgram (1963) designed a series of study involved a number of ethically questionable
studies in an effort to understand the brutality that elements. For one thing, the “teachers”/participants
occurred in Nazi concentration camps during the themselves suffered at least short-term harm; most
Second World War. In particular, he wanted to know felt very uncomfortable administering what they
how a person could be led to cause extreme harm to believed to be powerful shocks, and while the ex-
an innocent human being, and whether being or- periment was in progress many participants came
dered to do so by an authority figure had anything to doubt the experimenter’s claim that the “learn-
to do with it. Milgram devised a laboratory exper- ers” were not suffering any serious harm. Some were
iment in which a supposedly reputable researcher apparently traumatized by the experience, and ex-
(the authority figure) asked participants to act as pressed serious misgivings about taking part in the
“teachers” who would punish “learners” by giving study decades later (Perry, 2012).
them increasingly severe electric shocks whenever Another issue was the deception that the partic-
they gave incorrect answers to questions. Neither ipants experienced: they were told that the experi-
the shocks nor the howls of pain coming from the ment was investigating how people learn, not how
“learners” were real, but the “teachers”/participants people come to commit cruel acts, and they were
did not know this. Some of them were further dis- led to believe that the shocks were real. Thus, to the
advantaged in that they could not see the “learners”: extent that the participants were not fully informed
they could only hear the shrieks of pain supposedly about the purpose of the experiment, they were lied
coming from them. to, although once it was over they were debriefed
In all instances, the “teachers” were told that the (given previously withheld information about the
shocks were part of the study and would not cause study) in that they were informed by the “learner”
any permanent harm, despite the increasingly shrill that the shocks had not really been as painful as he
cries of pain. The experiment continued until the had let on, and that he had not been harmed by them.
“teacher”/participant refused to administer more However, in keeping with the procedural norms of
shocks. In the version of the study in which the the era, the debriefing provided was rather cursory
“teachers” could not see the “learners,” a substantial and did not fully disclose the deception involved in
majority (62 per cent) of the former administered the the study. In most cases the participant left the lab
strongest shocks possible. Further, in one variant of without being told that the shock machine was not
the experiment, the research volunteers were given real and that the “learner” was an accomplice of the
the option of ordering another person (actually an experimenter rather than a research participant. In
accomplice of Milgram’s) to administer the shocks; some cases, participants were not given a full expla-
in this situation, more than 90 per cent of them or- nation of how the experiment was conducted until
dered the greatest shocks possible. Many of the par- almost a year later (Perry, 2012).
ticipants were visibly upset as they carried out the Most people would agree that in everyday life,
experimenter’s orders, but they still did what the au- lying and deception should be avoided. On the other
thority figure told them to do. hand, the experiment would not have worked had
Milgram saw these results as shedding light on the volunteers known the truth from the outset.
the circumstances that led to the horrors of the Nazi Imagine the outcome if participants had been told,

bel65796_ch03_053-076 59 01/13/22 12:19 PM

60 PART I Fundamental Issues in Social Research

“We’re going to see whether you are willing to inflict basis. Some writers argue that if all ruses were dis-
severe pain on an innocent person, and by the way, allowed, a great deal of useful information—about
the ‘shocks’ you will be asked to give will not be real.” terrorists, cults, drug gangs, white-collar criminals,
But does the need for unguarded behaviour justify and so on—would be inaccessible, because finding
the deception? Was the knowledge gained worth the out about their activities often requires some kind of
psychological distress the participants experienced disguised observation. Some social reformers might
when they believed they were causing innocent per- add that exposing the sins of the powerful would also
sons to suffer? This is the sort of question that re- justify the use of deception, but that argument opens
searchers and REB members must ponder. the door to its use in other situations. Would it be
It must be emphasized that Milgram is not alone acceptable to use deception to examine the trans-
in his use of deception: in fact, most social research gressions of the underclass and the less powerful? Is
experiments involve some kind of pretense, although lying to a priest to uncover pedophilia more accept-
the level of deception used in his day is now viewed able than lying to an animal rights activist to reveal a
by many as excessive—it is unlikely that an REB in plan to destroy a mink coat factory?
North America would allow a similar study to pro- Sometimes it may be beyond the power of a qual-
ceed today. For that reason, Milgram’s experiment itative researcher to obtain informed consent. For
has rarely been replicated, except in parts of the example, Van Maanen (1991) used his experience as
world where research ethics codes are less stringent. a ride operator in Disneyland as the basis for a study
As you think about the ethics of Milgram’s study, he produced many years after he had been employed
keep in mind that REBs are not infallible and that there during school vacations. He never went back
REB members sometimes disagree among them- and asked permission from the people he described
selves about the ethics of a particular study. As a in his work; to do so would have been close to impos-
student, you need to make up your own mind about sible. Is the absence of consent acceptable in such a
ethical issues. Don’t feel obliged to accept the ethical situation? Is it a form of deception?
claims of more experienced social researchers; his- Although the use of deception can be controver-
tory has shown that empirical, theoretical, or tech- sial in both quantitative and qualitative research,
nological expertise is no guarantee of moral wisdom. there is consensus on a number of criteria regard-
But do listen carefully to the arguments of experi- ing its use. First, deception should be used only as
enced researchers. It is incumbent on you to consider a last resort. Second, if deception is used, that use
what is at stake in any research situation, especially should be as sparing as possible. Finally, where fea-
the probable harms and benefits to the research par- sible, anyone deceived should be fully debriefed as
ticipants, and to think about the effects—both good quickly as possible; specifically, the researcher must
and bad—that the knowledge produced by the study reveal the deception and explain why it was used.
would generate for the world at large. Milgram’s research illustrates the potential for harm
Deception can be a controversial issue in qual- to participants if the debriefing is not comprehensive
itative research as well. Punch (1994, p. 91), for and done in a timely manner.
example, observed that “some dissimulation is in- Another issue regarding informed consent is
trinsic to social life and, therefore, to fieldwork.” whether the researcher is obliged to reveal the names
He quoted Gans (1962, p. 44) in support: “If the re- of all those who have sponsored the project, whether
searcher is completely honest with people about his individuals or organizations. At one Ontario univer-
activities, they will try to hide actions and attitudes sity, ethical rules prohibit any reference to the names
they consider undesirable, and so will be dishonest. of research sponsors on the information sheet. This
Consequently, the researcher must be dishonest to rule reflects the possibility that some people might
get honest data.” feel pressured or coerced if they knew that the re-
Goode (1996) argued that judgments about de- search was sponsored by (for example) the Canadian
ception should be considered on a case-by-case Cancer Society. But another university considers

bel65796_ch03_053-076 60 01/13/22 12:19 PM

 3 Research Ethics 61

it unethical not to supply that information, on the the case of historical work or of studies dealing with
grounds that it is necessary to enable the participant public figures. In quantitative research, participants
to give informed consent. For instance, some people are usually identified by code numbers rather than
might not want the government to know anything names when the data are processed; their actual
about them, so if a department such as National identities are stored in a secure location. In such
Defence were funding the study, they would want to studies research participants are normally told on
be informed so they could decide not to participate. the information sheet that the data they provide will
Yet even if the sponsor’s name is revealed, partici- form part of a statistical aggregate (as in “50 per cent
pants’ ability to give informed consent may not be of respondents expressed some dissatisfaction with
guaranteed, since the name may be virtually un- their romantic partner’s listening skills”) and there-
known to the general public. Ethical issues are never fore they will not be personally identifiable.
simple. Researchers have to be especially resourceful in
As noted above, only people competent to make situations where participants prefer to remain anon-
a decision about participating in a research study ymous. One method that is sometimes employed is
should be asked to give their consent. That category the randomized response technique, which was orig-
normally includes most adults, although for reasons inally designed for interview situations in which the
of illness, other incapacitation, or language difficul- respondent is asked about controversial or illegal
ties some adults may not be capable of making an in- activities. One reason for developing this technique
formed decision and must have a competent person was to protect researchers as well as respondents
make the decision on their behalf. Similarly, young from the risk that police would examine the data and
children are not ordinarily considered capable of then bring the researchers into court as witnesses.
giving informed consent. Parents or legal guardians (Can some harm be anticipated here, as when the
are usually given the information sheet as well as an still-free criminals repeat their crimes?)
opportunity to ask questions and seek clarifications The random response process begins with re-
about the research before signing the consent form spondents flipping a coin but not revealing the re-
for their child. Quite often a parent or guardian is re- sults. About half of them should get heads, half tails.
quired to be present when the child interacts with the All respondents are then instructed: “For the follow-
researcher, especially if the study involves children ing question, if you have heads you must say yes, re-
who are very young. Exceptions include situations in gardless of whether it is true or not. With tails you
which older children fill out questionnaires that the should answer the question truthfully.” Then comes
parents or guardians have seen and approved. the question: “Have you ever used cocaine?” Even if
no one had in fact used it, roughly 50 per cent should
Concern for welfare say yes because they got heads.
“Welfare” in this context covers all aspects of the Assume that 60 per cent of respondents say yes.
well-being of the person, group, or community af- The excess over the 50 per cent (heads who had to
fected by the research, including physical, mental, say yes) can be used to calculate how many people
emotional, spiritual, social, and economic well- used cocaine. Here the excess is 10 per cent, presum-
being. A very basic “welfare” concern in social research ably representing people who got tails and have used
involves respect for privacy and confidentiality. REBs the drug. Then, since one would expect roughly the
normally require that the identities and records of same number of cocaine users in the heads group,
the people being studied remain confidential to pre- that figure must be doubled to get the percentage of
vent any embarrassment or harm, especially harm to the total sample who had used cocaine.
their reputations or personal relationships. This in- One of the authors of this book did this as a
junction means that the research participants should class exercise and had students raise their hands to
not be identifiable when the findings are published, say yes or no. When the behaviour in question was
although exceptions may occasionally be made in shoplifting after the age of 16, 80 per cent answered

bel65796_ch03_053-076 61 01/13/22 12:19 PM

62 PART I Fundamental Issues in Social Research

Box 3.1 What about research using the Internet?

Any situation that does involve direct contact
with research participants, including online sur-
veys and interviews, is subject to the same ethics
protocols as in-person research. The same goes
for gaining access to online materials in situa-
tions where there is a reasonable expectation
of privacy, as in the case of social media that
Michael Simons/123RF

may limit access to materials unless the cre-

ator’s permission is granted, such as Facebook,
Instagram, and LinkedIn. Internet chat room
discussions, email, and the postings of organi-
As a researcher, consider how you could conduct zations having restricted memberships, such as
online research in an ethical way. What kinds self-help groups, fall into this category as well.
of steps would you need to take to secure the in- In these sorts of situations, both researchers and
formed consent of your participants and to protect research participants have to be especially vigi-
your data? lant, since it is relatively easy to conduct covert
A great deal of research is now done online (see research online. It would be quite easy, for exam-
Chapters 5, 7, 8, 10, 11, and 12). All the ethics ple, for a researcher to pose as a regular partici-
guidelines discussed in this chapter apply to pant in a self-help chat room for sex addicts, and
online research, with a handful of exceptions. use the material as data for a qualitative study.
One exception involves material that is publicly But doing so would be equivalent to joining an
available and protected by law, such as Statistics in-person discussion group and not revealing
Canada public-use files or data in public ar- one’s identity as a researcher. (The ethical impli-
chives. In such cases the researcher is not re- cations of covert research will be discussed later
quired to obtain informed consent, although he in this chapter.)
or she must obey the relevant laws and regula- Sending research data over the Internet is
tions. Another exception applies to online infor- another concern. If data containing identifiable
mation that is publicly accessible and does not information are to be transferred via the Internet,
involve any expectation of privacy. Such material the data should be encrypted to ensure the se-
would include films, newspaper materials, offi- curity of the information. Even when sensitive
cial publications, third-party interviews, and any data are merely stored on a computer that is
other online research that does not entail direct connected to the Internet, encryption should be
interaction with research participants. considered.

“yes,” meaning a 60 per cent (30 per cent plus an- in the difference; some “heads” students were afraid
other 30 per cent) shoplifting rate. When the topic their classmates would think a raised hand was an
was post-puberty homosexual behaviour, however, admission; or perhaps some “tails” students who had
“yes” responses came out at only 50 per cent. Here engaged in the activity did not want to give any in-
there are several possibilities: no one had engaged in dication that they had, fearing that a “yes” response
the activity; a few had, but because heads and tails would be interpreted as an admission. In any case,
rarely come out exactly 50–50, they were concealed although randomized response techniques have not

bel65796_ch03_053-076 62 01/13/22 12:19 PM

 3 Research Ethics 63

been used extensively in social research, there is evi- general rule, but it is often broken in dealing with
dence that they can be effective in studying sensitive dangerous people such as pimps and other crimi-
topics (Blair, Imai, & Zhou, 2015). nals. Where such permission has not been obtained,
In qualitative research, dealing with privacy and it is absolutely crucial to maintain the anonymity of
confidentiality issues is quite different, especially the people observed.
if the study involves in-depth analysis of a small One difficulty with the arguments justifying the
number of people. Pseudonyms are typically used use of covert methods is the assumption that it is
to protect people’s identities, but quite often the impossible to obtain the data using other means.
detailed description of the physical and social set- Covert observers sometimes base their judgments
ting, which qualitative researchers usually include on the anticipated difficulty of gaining access to a
to provide the needed context for the study, can setting rather than the actual experience of being
offer enough clues to identify some of the partici- denied entry. For example, Homan justified his use
pants. The study of an American town that Vidich of covert participant observation of a religious sect
and Bensman (1968) called “Springdale” (not its real on the grounds that sociologists were viewed very
name) is instructive in this regard. Their book was negatively by group members; hence “It seemed
uncomplimentary about the town and several of its probable that the prevalence of such a perception
leaders, and many people felt that its tone was pa- would prejudice the effectiveness of a fieldworker de-
tronizing. To make matters worse, it was possible to claring an identity as sociologist” (Homan & Bulmer
identify individuals in the published account. The 1982, p. 107).
town’s inhabitants responded with a Fourth of July One of the most famous and controversial cases
parade in which people wore badges citing their of covert research can be found in Humphreys’s
book pseudonyms, and an effigy of Vidich was set Tearoom Trade (1970), a study that Desroches (1990)
up so that it was peering into manure. The towns- partially replicated in Canada using less contentious
people also announced their refusal to cooperate in methods. The researchers (Humphreys himself and
any more social research; they were clearly upset by police acting for Desroches) observed homosex-
the publication, and to that extent were harmed by it. ual encounters in public washrooms (“tearooms”),
Another area of controversy surrounding privacy taking the roles of “watch queens”—people who
and confidentiality in qualitative studies pertains to watch out for possible intruders while other men
covert research: investigations in which the people meet and engage in sexual activity. Such “voyeur-
being observed are not informed that they are part ism” was offensive to some critics.
of a study. In spite of the widespread insistence on As part of his research, Humphreys recorded
informed consent and privacy rights, and the rec- some of the participants’ car licence numbers. He
ognition that covert observation is especially prone was then able to track down their names and ad-
to violations of those principles, covert studies still dresses, thus further compromising their privacy,
appear from time to time. The defence is usually that and ended up with a sample of 100 active tea-
the benefits of the research outweigh any harm that room-trade participants. To reduce the risk of being
may come to the people being studied. Virtually all recognized, Humphreys waited a year before con-
codes of ethics allow covert research as a last resort, tacting the respondents whose licence numbers he
but most state that it should be avoided “as far as had written down, and also changed his hairstyle.
possible.” Those who use this method argue that it After this deception, he then conducted an interview
is necessary to prevent research participants from (Desroches did not do this) with a sample of the men
changing their behaviour because they know they about their health issues (a ruse), including some
are being studied. Still, where informed consent has questions about marital sex. He did not tell them he
not been obtained prior to the commencement of the knew about their activity, nor did he debrief them
research, permission to use the data gathered should when the study was finished. Humphreys learned
be obtained post hoc whenever possible. That is a that many of the men he observed were actually

bel65796_ch03_053-076 63 01/13/22 12:19 PM

64 PART I Fundamental Issues in Social Research

married, respected members of the community—a researchers admitted their invasion of client privacy
finding that allowed him to challenge a popular ste- and failure to obtain consent from them, but were
reotype of the day. “untroubled” by the intrusion because the clients
Was his research ethical? Those who say “yes” were anonymous to O’Connell Davidson and she was
argue that he had no choice but to do covert research not “in a position to secure, store or disclose infor-
because very few people engaging in gay sex in those mation that could harm them” (O’Connell Davidson
days were willing to talk about it; in fact, sexual acts & Layder, 1994, p. 214). The fact that there was no
between men were against the law and the people harm to participants was regarded as the litmus test
committing them were subject to prison sentences. of the ethical status of the research. They offered a
Humphreys’s supporters also maintain that the further defence by saying that ethical transgression
knowledge his studies produced led to a better un- is pervasive: “Virtually all social research is intru-
derstanding of human sexuality and the treatment sive and exploitative to some degree” (O’Connell
of stigmatized individuals. His detractors argue that Davidson & Layder, 1994, p. 215). There was an ac-
the knowledge gained did not justify the deception knowledged commitment to attend to the concerns
and invasion of privacy that his research participants of Desiree and her receptionists, as well as consider-
experienced. The debate continues to this day. able sympathy for them, but these sentiments did not
Regardless of one’s position on the ethics of extend to her clients:
Humphreys’s work, the public reaction when it was
I have . . . no personal liking and no real sym-
first published illustrates what can happen to the
pathy for them. I have a professional obligation
reputation of social researchers as a whole if there is
to preserve their anonymity and to ensure that
a perception that one of their kind has committed
they are not harmed by my research, but I feel
serious ethical violations. The following is an excerpt
no qualms about being less than frank with
from an article written in the Washington Post in
them, and no obligation to allow them to choose
1970, shortly after the controversy erupted:
whether or not their actions are recorded.
We’re so preoccupied with defending our pri- (O’Connell Davidson & Layder, 1994, p. 215)
vacy against insurance investigators, dope
Although an effort was made to follow ethics proto-
sleuths, counterespionage men, divorce de-
cols, it appears that not all the participants received
tectives and credit checkers, that we overlook
the same level of ethical protection, and that the re-
the social scientists behind the hunting blinds
searchers believed that not all research participants
who’re also peeping into what we thought were
deserved the complete range of ethics coverage. This
our most private and secret lives. But they are
sort of arrangement is more common than one might
there, studying us, taking notes, getting to
think and has far-reaching implications. It also illus-
know us, as indifferent as everybody else to the
trates how difficult it can be to resolve ethical dilem-
feeling that to be a complete human involves
mas in social research. Do sex-trade clients have a
having an aspect of ourselves that’s unknown.
right to choose to be excluded from research studies?
(Von Hoffman, 1970)
What if they are abusing the sex workers? What if they
The ethical quandaries surrounding privacy and are not? Does it matter whether such studies contrib-
confidentiality inherent in qualitative research are ute to a better understanding of the sex industry—for
also illustrated in the work of O’Connell Davidson instance, by revealing the physical risks that people in
and Layder (1994). These researchers conducted a the sex trade take? Such questions must be weighed
small-scale ethnographic project on a sex worker, carefully and are not easily answered.
whom they called Desiree, and her clients and as- Sometimes a researcher may feel obliged to dis-
sistants. While Desiree and the women who served regard concerns about privacy and confidentiality
as her receptionists were fully aware of O’Connell if the research uncovers evidence of serious harm
Davidson’s status, the clients were not. Both or criminal activity. In his ethnographic research,

bel65796_ch03_053-076 64 01/13/22 12:19 PM

 3 Research Ethics 65

Totten (2001) wanted to know how youth gang in the province (Le Devoir, 2016). Ultimately, in 2017,
members interpreted the violence they perpetrated the Superior Court of Quebec ruled that she would
against their girlfriends, and how they accounted for not have to give up her data to the court. The decision
their racist and homophobic activities. One underly- was hailed as a victory for research participants, the
ing factor he observed was their mental construction researchers who rely on them, and the general public
of masculinity, which included the belief that such that benefits from open inquiry and scientific discov-
activities are acceptable for men. Most had lived on ery (Kondro, 2016; Castonguay, 2017).
the street, were themselves victims of severe child Another aspect of the “concern for welfare” cri-
abuse and neglect, and had witnessed their mothers terion, one that can affect the larger community or
being beaten. They saw their own violence as affirm- even the society as a whole, involves sponsored re-
ing their masculinity. search. Most social research in Canada is funded
That was the sociology, but what about the wel- either by SSHRC or by the university with which the
fare of the people concerned? Here the researcher researcher is affiliated, but other government agen-
was in some instances compelled to act. By law, he cies, private individuals, corporations, unions, and
had to report any abuse suffered by children under non-profit organizations can also sponsor research.
the age of 16, regardless of whether they were gang Ethical issues arise if the sponsor places conditions
members or their victims. Second, those deemed at on the methodology to be used or the publication
risk of suicide had to be referred to mental health– of the results, or has certain expectations regarding
care providers. Finally, some of the gang members the findings. In the United States, a social researcher
were on probation or living in treatment facilities, was asked by the owner of an apartment building to
so the relevant authorities also had to be informed. do a study of the behaviour of tenants in “singles”
Totten handled the confidentiality issue by studying apartments after neighbours complained about bad
only the boys who agreed to those limitations on behaviour (Bailey, 1994). The sponsor in this case
confidentiality. made it clear that the research was expected to show
Legal issues can affect confidentiality for other that the tenants were in fact good people and desir-
reasons. When Marie-Ève Maillé was a graduate stu- able neighbours. Perhaps the most famous Canadian
dent in Quebec in 2010, she interviewed 93 people af- case of a sponsor trying to influence the research
fected by a large wind farm in the province that was process involved Dr Nancy Olivieri, who took part
operated by a private company. When she conducted in research on a drug designed to treat thalassae-
the interviews, she followed good research practice mia, a blood disease. She claimed that the drug was
by telling the interviewees that their identities would not effective and had harmful effects, and wanted
remain private. The data collection was particularly to inform the research participants. A controversy
sensitive because many people in the area surround- arose when the pharmaceutical company that spon-
ing the wind project were bothered by its construc- sored the research denied her claims and cited a con-
tion and operation, and a few years later filed a class fidentiality agreement that she had signed that would
action lawsuit against the company. be violated if she were to contact the participants.
Maillé volunteered to act as an expert witness for Another famous case involved three Canadian sci-
the people who launched the lawsuit. But as the legal entists who claimed that their supervisors at Health
proceedings unfolded, she was ordered by a judge to Canada had tried to coerce them into approving cer-
provide the company that owned the wind farm with tain drugs for use on cows and pigs even though the
the raw data from her study, declaring it to be evi- scientists objected that the drugs could have harmful
dence relevant to the case. To protest the judge’s order, effects on humans; the scientists were subsequently
over 200 Quebec researchers signed an open letter fired, although one was later reinstated. In 2010, the
published in Le Devoir stating that if Maillé had to head of Statistics Canada resigned over the federal
turn over her confidential data, that would set a prec- government’s decision to discontinue the mandatory
edent that would seriously impede scientific inquiry long-form census and replace it with a voluntary

bel65796_ch03_053-076 65 01/13/22 12:19 PM

66 PART I Fundamental Issues in Social Research

version, which in his view would not be adequate to research process must be upheld. Any conflict of in-
fulfill Statistics Canada’s mandate. (The decision was terest, including a financial conflict of interest involv-
reversed by a later government.) ing a relationship with a sponsor, must be reported
People conducting social research should be to the appropriate REB. Most fundamentally, the
mindful of the perils of sponsored research. The public good, in particular the welfare of the research
TCPS2 states that financial considerations do not participants, “takes precedence over the interests
reduce the need to abide by the principles that under- of researchers, institutions, and sponsors” (CIHR,
lie its policies, and that the scientific integrity of the NSERC, & SSHRC, 2018, p. 143).

Methods in Motion Applications to Canadian Society

The effect of ethics reviews on qualitative research
Did the implementation of formal ethics re- forms to sign. Such difficulties are often not ad-
views stifle social research in Canada? Van den dressed in national ethics regulations. Indeed,
Hoonaard and Connolly (2006) examined this the authors point out that the original TCPS (1998)
issue by gathering information on Canadian mas- contained 476 paragraphs, of which only four per-
ter’s theses completed in anthropology in the 10 tained to “natural observation research.”
years from 1995 to 2004. The mid-point in that Van den Hoonaard and Connolly (2006) found,
time span—2001—marked the year that the first contrary to expectation, that the number of an-
edition of the Tri-Council Policy Statement on thropology master’s theses actually increased
Ethics in Human Research (TCPS) (CIHR, NSERC, in the 10-year period under investigation, and
& SSHRCC, 1998) was implemented on a national that the number of research projects involving
basis. The authors chose anthropology theses research participants (as opposed to archival
in part because a large portion of the research documents, secondary data sources, and so on)
done in that discipline is qualitative, and qualita- also increased. This suggested that ethics reviews
tive researchers have been the most vocal about did not put a brake on anthropological research,
the negative consequences that stringent ethics although the authors noted that the data did not
policies may have on social research. include cases in which research was abandoned
One of the major complaints of qualitative because ethics rules were too strict. More om-
researchers is that ethics policies in Canada and inously, however, they noted that the proportion
elsewhere have been designed according to a of theses involving only qualitative interview-
bio-medical model that calls for standard re- ing rather than full-blown ethnography had in-
search protocols, signed consent forms, and other creased since 2001. They suggested that this
elements that are not compatible with qualitative trend might be a consequence of overly stringent
and especially ethnographic studies. In qualita- ethics regulations on ethnographic research.
tive research, topics and even methodologies may More recent versions of the Tri-Council Policy
emerge out of the research process itself, so infor- Statement were created with substantial input
mation sheets on the goals of the research, consent from qualitative researchers. The current version,
forms outlining the risks and benefits of participa- the TCPS2 (CIHR, NSERC, & SSHRC, 2018), has an
tion, and other features of the bio-medical model entire chapter on qualitative research, and the
can in many cases seriously interfere with the re- new guidelines recognize that requiring formal
search: imagine how members of an outlaw mo- consent documents and other elements of the
torcycle gang would respond to a researcher who bio-medical model may not be appropriate in all
handed them information sheets and consent research situations.

bel65796_ch03_053-076 66 01/13/22 12:19 PM

 3 Research Ethics 67

Justice that they need “help” of some kind, or the assumption

Justice in the context of research ethics includes the that the researcher has a superior sort of wisdom or
idea that the burdens and benefits of research should political insight. Some Western anthropologists have
be spread evenly across society. No person or group been barred from re-entering the countries where
they did their fieldwork because the locals found them
should be exploited in the research process or system-
to be intrusive and disrespectful.
atically excluded from its benefits. This is a major con-
Justice in a research context also includes the
cern in health research, in which care has to be taken
principle of minimizing harm to participants. As we
that vulnerable people are not treated as guinea pigs for
suggested above, it is a rule of thumb when harm is
the benefit of others, whether some interest group or
expected that it must not outweigh the potential ben-
the larger society. Unfortunately, there is a history of
efits of the research. But as the examples discussed
this sort of abuse in Canada, such as what happened in
earlier have shown, harms and benefits can be diffi-
Indigenous communities and residential schools in the
cult to judge correctly, and assessments of them ul-
1940s and 1950s. Nutritional experiments were con-
timately involve value judgments as well as scientific
ducted, apparently without informed consent, on chil-
decisions. Most social scientists are not curing cancer
dren and adults who were found to be suffering from
or ending racism. How much suffering or harm, if
serious malnourishment. In one study, rather than
any, can be justified in social scientific research? That
provide emergency food aid, the researchers set up ex-
is a key ethical question facing both researchers and
perimental and control groups to test the effectiveness
the REBs that oversee their work.
of vitamin and other dietary supplements; those in the
Research likely to harm participants is regarded
control group continued with their meagre diets. In
by almost everyone as unacceptable. But what is
some cases, Indigenous children were prevented from
“harm”? Harm can mean a number of things, in-
receiving dental care for extended periods of time on cluding physical harm, loss of self-esteem, stress, or
the grounds that it would interfere with an assessment even a mild sense of shame. And what does “likely”
of their nutritional status, and experimenters in some mean: a 1 per cent chance? 10 per cent? 50 per cent?
cases prolonged the existence of inadequate diets for Researchers should do their best to anticipate and
two years in order to establish baseline measurements. guard against consequences for research participants
Moreover, the experimenters were aware of the link be- that are expected to be harmful or disturbing. But
tween malnutrition and the high disease and mortality it’s not always easy to know where to draw the line.
rates found among Indigenous people (Mosby, 2013). For example, in the Rosenthal and Jacobson (1968)
At the same time, medical researchers should not arbi- study of teacher expectations discussed in Chapter 2,
trarily exclude from their studies certain categories of it is at least possible that the pupils not identified as
people (such as members of specific ethnic, gender, or “spurters” (students expected to excel quickly) were
age groups), especially if such people stand to benefit adversely affected in their intellectual development by
from experimental drugs or other treatments. the increased attention given to the “spurters.” After
With regard to social research, inclusiveness is all, teachers have limited time and energy, and time
usually not a pressing issue, partly because social spent with the “spurters” meant time away from ev-
scientific studies lack the immediate life-and-death eryone else. Was that a violation of research ethics?
consequences that are common in medical research, Was the harm suffered by the “non-spurters” out-
and also because vulnerable and exploited groups are weighed by the knowledge gained by the researchers?
often sought out for study, particularly by sociologists Similarly, in Festinger, Riecken, & Schachter’s (1956)
and anthropologists. However, there is always a risk study of a doomsday/UFO cult, the researchers joined
that members of disadvantaged groups will be abused, the group at a crucial time (close to the projected end
mistreated, or disrespected by the researchers who of the world), thereby deluding group members into
study them. For example, some street people might thinking that they had succeeded in winning new
resent an ethnographer’s implication, however subtle, converts. Was that ethical? Was the harm to the cult

bel65796_ch03_053-076 67 01/13/22 12:19 PM

68 PART I Fundamental Issues in Social Research

members less important than the understanding of harm to participants, and if the risks of the research
cults gained through this research project? are greater than the risks of everyday life, informed
One of the problems with trying to balance harms consent is a minimum requirement.
and benefits is that it is not possible to identify all Clearly, the principle of justice in social research
the circumstances in which harm is likely—although is closely tied in with the precepts discussed earlier,
that fact should not be taken to mean that there’s no in particular the notions of informed consent, pri-
point in seeking to protect research participants. vacy, and confidentiality. People are certainly less
Kimmel (1988) noted this in connection with the likely to be harmed if they know what to expect in a
1939 Cambridge–Summerville Youth Study: an ex- research situation, and if they know that their partic-
periment conducted on 506 boys aged 5 to 13 who ipation is voluntary. Thus, it is particularly import-
were identified either as likely to become delinquent ant to inform participants of their right to withdraw
or as having an average likelihood in that regard. The from the study at any time—and to honour that right.
boys were randomly assigned either to an experi- A good illustration of what can go wrong when re-
mental group in which they would receive preventive search participants are pressured into remaining in a
counselling, or to a no-treatment control group. In study can be found in the Stanford mock prison study
the mid-1970s the boys (by then men) were exam- (Haney, Banks, & Zimbardo, 1973). In this experiment,
ined and the results were quite shocking: psychologically healthy male university students were
Treated subjects were more likely than controls randomly selected to be either “guards” or “prison-
to evidence signs of alcoholism and serious ers” in a simulated prison set up in the basement of
mental illness, died at a younger age, suffered a university building. “Guards” were given uniforms,
from more stress-related diseases, tended to be wooden batons, and reflective sunglasses (to avoid eye
employed in lower-prestige occupations, and contact with the “prisoners” and thus to dehumanize
were more likely to commit second crimes. them), and were told to instill fear in the “prisoners.”
(Kimmel, 1988, p. 19) The “prisoners” were given humiliating smocks and
headgear to wear, and had to address the “guards”
In other words, the treatment itself may have had un- in a respectful manner regardless of how they were
intended negative consequences, although given the being treated. The aim of the study was to see whether
complex and long-term nature of the forces that cause taking on the role of “guard” would create feelings of
delinquency it seems likely that non-experimen- omnipotence and lead to belligerent behaviour, and
tal factors contributed to the results. Nonetheless, whether being treated like a prisoner would produce
the findings do illustrate the difficulty of anticipat- feelings of subservience, self-alienation, resentment,
ing harm to participants. Might a questionnaire on and hostility. In short, the aim of the study was to see
marital happiness lead a respondent to question and whether placing ordinary people in these roles would
eventually leave a marriage? Could asking a grade lead to stereotypical prison behaviour.
eight boy in gym class about steroid use encourage The study was designed to last two weeks.
him to experiment with a drug that until then was Zimbardo describes what happened:
unknown to him? Some interviewees may be un-
comfortable answering any question; some may find At the end of only six days we had to close down
the cut and thrust of a focus group discussion stress- our mock prison because what we saw was fright-
ful, especially if they inadvertently reveal more than ening. It was no longer apparent to us or most of
they had intended. Does it follow that such studies the subjects where they ended and their roles
should not be conducted? Is the risk of harm, even began. The majority had indeed become “pris-
mild harm, worth the potential increase in knowl- oners” or “guards” no longer able to clearly dif-
edge that this kind of research may make possible? ferentiate between role-playing and self. There
That is the crux of the issue when a researcher seeks were dramatic changes in virtually every aspect
a proper balance between harms and benefits. Most of their behavior, thinking and feeling; human
ethical codes maintain that if there is any prospect of values were suspended, self-concepts were

bel65796_ch03_053-076 68 01/13/22 12:19 PM

 3 Research Ethics 69

challenged, and the ugliest, most base, patho- considered unethical to offer payment if that would
logical side of human nature surfaced. We were lead volunteers to take risks they would not oth-
horrified because we saw some boys (“guards”) erwise accept. Similarly, payment should not put
treat other boys as if they were despicable an- undue pressure on people to take part or remain in a
imals, taking pleasure in cruelty, while other study, and for that reason should not exceed the sorts
boys (“prisoners”) became servile, dehuman- of benefits to which the participants are accustomed.
ized robots who thought only of escape, of their The participants in a study of suburban Toronto im-
own individual survival, and their mounting migrants conducted by Dinovitzer, Hagan, & Parker
hatred of the guards. (Zimbardo, 1971; quoted in (2003) were offered $5 to participate. It seems unlikely
Aronson, 1992, pp. 10–11; emphasis added) that anyone answered the questionnaire they were sent
simply for the small amount of money involved. Still,
A whole host of ethical issues come into play in this ex- some REBs do not allow the people being studied to be
periment, but let’s focus on the right to leave a study at paid for their participation, except to cover the costs
any time. Many of the “prisoners” wanted to leave long of taking part (parking, etc.). The purpose of this rule
before the study was terminated. Yet they were given is to prevent poor people from “selling themselves” to
every indication that they had no choice but to stay in researchers. Few would be critical of Beagan’s (2001)
the study, and were convinced that for all practical pur- offer to Canadian medical students of a chance at a
poses they were in a prison. A few were allowed to leave $200 draw (which technically got around the payment
early, but only after showing signs of extreme stress and rule) for taking part in her study of discrimination. But
even hysteria. It’s hard to see how this was not a viola- would offering $10,000 for testing a promising drug
tion of research ethics. Looking back on what he had with potentially dangerous side effects be acceptable?
created, Zimbardo himself was appalled. What about selling a kidney for research purposes?
Another point this experiment illustrates is the Fortunately, the risk of harm to participants in social
need for close oversight of ethical issues on the part scientific research is usually not as severe as the risks as-
of REBs. Like the Milgram (1963) study, Zimbardo’s sociated with medical research, so those sorts of ques-
research was conducted before the era of stringent tions rarely have to be considered.
ethics reviews. REB oversight is absolutely necessary Benefit in the social sciences is typically indi-
because even well-meaning researchers like Zimbardo rect; in most cases it follows from an increase in
and Milgram may be completely unaware that their knowledge. However, direct benefits can accrue, for
work raises ethical problems—even if those problems instance in cases where research findings inspire
seem quite obvious to others. For example, it simply positive social reform or contribute to valuable so-
did not occur to Zimbardo that his experiment was cietal conservation. One way to maximize the ben-
causing his research participants significant harm. It efit derived from social research is to conduct it in
was not until a student research assistant told him she a methodologically sound manner. It is considered
thought the treatment of the “prisoners” was awful unethical to put research volunteers through the
that he relented and brought the study to an end. rigours of participation if the study is not designed
Presumably, a strict REB review would have spared or conducted properly and therefore is of little or
the “prisoners” considerable agony. no value. As you will learn as you make your way
In recent years, the Stanford prison study has also through this book, researchers can disagree over
been criticized on methodological grounds. Some of what constitutes a sound methodology. Nonetheless,
these criticisms are discussed in the Research in the each methodological perspective assumes that there
News box on pages 70–71, where it is apparent that are optimal ways of doing the research, and if these
methodological controversies can spill over into the are ignored it’s doubtful that anyone, including the
realm of research ethics, an idea that is explored in researcher, will get the maximal benefit from it.
more detail below. Maximizing benefit is also related to the inclusive-
Another issue related to harm to participants ness principle discussed above: no group or individ-
concerns payment for participation. In general, it is ual should be arbitrarily excluded from research.

bel65796_ch03_053-076 69 01/13/22 12:19 PM

70 PART I Fundamental Issues in Social Research

Finally, justice in the conduct of social research in- evidence that the excessively formal and legalistic
cludes fairness in the adjudication of research proposals letters of introduction that some universities re-
by REBs. The procedures and protocols that REBs use quire researchers to send to prospective participants
have to be reasonable and equitable, and their decisions lowers participation rates (Charbonneau, 2005).
unbiased. In order to achieve those ends, REBs must be Low response rates may result in sample sizes that
independent of the institutions in which they function. are too small for analysis and can lead to additional
This means, for example, that universities cannot over- expenses for follow-up.
turn the decisions of their REBs.
As we noted above, sometimes REBs can be Conclusions
overcautious in trying to protect the interests of re-
search participants. This is often a concern in survey These ethical principles, and the larger discussion of re-
research, which normally involves minimal risk to search ethics contained in the TCPS2 (CIHR, NSERC,
participants. J. Paul Grayson, a sociologist at York & SSHRC, 2018), provide an invaluable guide for the
University, conducted a study in which he found conduct of social research. Their implementation can

Research in the News

Methodological controversies in the Stanford prison experiment
Zimbardo’s Stanford prison experiment is one of other by number only, that they were not allowed
the most famous studies ever done by a social sci- to refer to their situation as an experiment or a
entist, and as such has garnered a lot of media at- simulation, and that failure to abide by these sorts
tention in the half-century since it was conducted. of rules could result in punishment. “I was given
The ethics of the study have long been debated, but the responsibility of trying to elicit ‘tough-guard
more recently it is the methodology that has come behaviour,’” the research assistant explained.
under fire, even in the popular media. An investiga- This created demand characteristics,
tion into the study published in Medium GEN by Ben which are cues picked up by research partici-
Blum (2018) has revealed some serious shortcom- pants that lead them to act in certain ways in
ings in the study (see also Le Texier, 2019). order to avoid “ruining” the study or to other-
Blum questions the general narrative taken wise please the researcher. Often participants
by Zimbardo that psychologically “normal” believe that it is their responsibility to exhibit
people can be transformed into either abusive, the sorts of behaviours that the researcher is
sadistic guards or cowering, deeply disturbed expecting to find, and so do not behave in a
prisoners by simply assigning them these roles genuine and authentic way. Demand char-
randomly and letting the power of the prison acteristics can be unintentional and subtle,
situation spontaneously run its course. A crucial but in the case of the Stanford prison study
methodological flaw revealed by Blum is that the they were quite explicit. One of the “guards,”
“guards” were given extensive, explicit instruc- who had studied acting, even faked a south-
tions on how to treat the “prisoners,” in particular ern accent in order to appear more intimi-
how to break them down psychologically. An un- dating to the “prisoners.” “I took it as a kind
dergraduate research assistant told the “guards” of improv exercise,” he said. The “prisoners”
the day before they came in contact with the also appear to have been affected by demand
“prisoners” that the latter were to address each characteristics. One of them, who during the

bel65796_ch03_053-076 70 01/13/22 12:19 PM

 3 Research Ethics 71

study appeared to have a hysterical emotional have received only scant attention and discussion
breakdown, told Blum: “Anyone who is a clini- by Zimbardo over the years. The reluctance to
cian would know that I was faking.” The par- fully address these issues also brings to light an
ticipant stated that he staged the breakdown ethical dimension of the study that had not been
in order to be released from the study. widely considered before—namely, the duty of the
These are serious methodological flaws that researchers to provide a candid and comprehen-
cast doubt on the study’s main conclusions, and sive account of how the research was conducted.

be challenging, however, especially when acquiring particular practice, such as the covert methods used
informed consent might lead participants to behave by Humphreys (1970) in his Tearoom Trade study.
less naturally than they otherwise would. Balancing There is no simple answer to these ethical dilem-
risks and potential benefits can also be problematic. mas, especially if there is no way to measure harms
But be that as it may, wherever the procedures outlined and benefits directly. However, it is incumbent on
above are not followed, the onus is on the researcher researchers to be vigilant in protecting the rights
to convince the REB that it would be appropriate to go of research participants, and to maximize the ben-
ahead with the study. efits and minimize the harm that is created by their
Another kind of challenge arises when profes- research.
sional researchers disagree on the ethical status of a

Key Points
The TCPS2 outlines three core principles for the Most social research in Canada has to abide by
ethical conduct of research: those principles. In most cases, before social research
• Respect for persons demands that research par- can be conducted it has to be approved by a Research
ticipants be asked for their free, informed, and Ethics Board (REB) that uses the TCPS2 to guide its
ongoing consent. decisions. It may be difficult to measure harms and
• Concern for welfare includes all aspects of the benefits in social research, so the implementation of
well-being of individuals, groups, and communi- these principles is rarely simple and straightforward.
ties affected by the research. Researchers may disagree on how the three
core principles are to be implemented, but they still
• Justice covers all dimensions of fairness in the
treatment of research participants and the re- must do everything they can to comply with them.
searchers themselves.

Questions for Review (R) and Creative Application (A)

r1 Explain what is meant by “harm to participants.” lived racism? Would the harm be so severe that
Does it refer to physical harm alone? the study should not be conducted? Explain.
a1 What harm might be suffered by a person r2 What are some of the difficulties involved in im-
completing an online questionnaire on the plementing the principle of informed consent
topic of illegal drug use? Would the harm be so when doing ethnographic research?
severe that the study should not be conducted? a3 Assume you are doing ethnographic research
Explain. at a homeless shelter. Is it necessary to get in-
a2 What harm may be suffered by a person partic- formed consent from every person you interact
ipating in qualitative interviews on the topic of with or observe? Explain.

bel65796_ch03_053-076 71 01/13/22 12:19 PM

72 PART I Fundamental Issues in Social Research

r3 Can deception in social research ever be justi- r4 What is meant by “balancing risks and potential
fied? Explain. benefits”?
a4 Assume you are participating in survey research a5 List the risks and potential benefits of Milgram’s
that seeks to determine the degree to which (1963) “electric-shock” experiments. Did the po-
participants possess sexist attitudes. Would tential benefits outweigh the risks? Explain.
some deception regarding the purposes of the a6 List the risks and potential benefits of Humphreys’s
research be justified? Explain. (1970) Tearoom Trade study. Did the potential
benefits outweigh the risks? Explain.

Interactive Classroom Activities

1. After reviewing the three core principles for eth- the third question, and divides the class into
ical research with the class, the instructor will three groups. Group 1 must argue that the study
show one of two films: Milgram’s Obedience was ethical; Group 2 must argue that the
or Zimbardo’s Quiet Rage: The Stanford Prison study was not ethical; and Group 3 can choose
Experiment. Before viewing the film, students either of these positions. All groups are to justify
are given the following questions to think about their positions as best they can, and are encour-
as they watch it: aged to allude to the material discussed in this
a. What scientific knowledge did this study pro- chapter. Once all three groups have argued their
duce? In other words, what can we conclude case, a vote is held to determine how the class as
from this study? a whole views the ethics of the study, i.e., whether,
b. Can the knowledge generated by this study on balance, they find it to be ethical or not.
be used to explain real-life situations? 2. Using the classroom screen, the instructor can walk
c. Was this study ethical? the class through the TCPS2 tutorial available at
[Link]
After the film has been shown, an open class dis- [Link] (accessed 4 September 2021). It in-
cussion is held to address the first two questions. cludes multidisciplinary examples, some interactive
Once that is completed, the instructor informs features, audio and video downloads, and a quiz.
the class that a debate will be held to answer

Relevant Websites
The TCPS2 (CIHR, NSERC, & SSHRC, 2018) is avail- Visit the Canadian Sociological Association web-
able online. site to read its Code of Ethics.
[Link] [Link]/code-of-ethics
eptc2_2018.html
This link leads to the Canadian Charter of Rights
This TCPS2 tutorial, which provides an excellent in- and Freedoms.
troduction to the document and gives users the op-
[Link]
portunity to apply ethical principles to case studies,
is available at Quiet Rage is a documentary film about the
[Link] Stanford mock prison study that can be found at
html [Link]
(Websites accessed 4 September 2021.)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch03_053-076 72 01/13/22 12:19 PM

APPENDIX TO PART I
An Overview of the Research
Process
You now have some notion of how and why social of inquiry may begin with simple curiosity about
research is done. Before moving on to the rest of the what’s out there: as one wag put it, if you walk by a
book, which expands on many of the points raised in house with its curtains closed and wonder what goes
the previous three chapters, let’s step back and take a on inside, you’d probably make a good social scien-
look at the big picture. tist. With these sorts of motivations, researchers may
What follows could be described as the “ideal” be inspired to test a theory or interpretation of some
stages of research. They are ideal not because they social phenomenon, or to develop a new theory or
represent the best way to do things, but rather to in- interpretation.
dicate that they are an abstraction, a simplification Not all research begins with lofty motivations.
of a more complex reality. There is no single best way In fact, the reasons for conducting a study are often
to do research, any more than there is a single best quite mundane. For example, research may be done
way to build a house or paint a portrait. Sometimes to enhance the career of the researcher, especially if
the order of the tasks presented in the overview of it is supported by substantial funding. Sometimes
the research process on page 74 can be rearranged. researchers compete with one another to see who
Also, it’s not uncommon to go from one step to the can produce the most publications or get the most
next and then backtrack to adjust something in a attention for their work. More generally, research
previous step. Nonetheless, some general points and may be motivated by “sheer egoism,” a reason that
observations can be made that outline how research topped the list of motivations for writing drawn up
may progress from its initial stages to its completion. by George Orwell (1946/2004). Graduate students
(and some undergraduates) may do research simply
In the beginning: Motivation to fulfill degree requirements. In most instances, the
motivations for doing research are some combina-
Where does research begin? tion of the exalted and the mundane.
Sometimes research begins with a sense of
wonder, a feeling of amazement with the social world The research question(s)
that is accompanied by a strong desire to understand
and explain it. Sometimes it begins with a sense of Once the decision has been made to do research in
crisis, as when a society is strained to the breaking a particular area, the first task is to come up with
point and scholars are looking for solutions. This a research question, or a set of research questions,
motivation has been around since the beginnings of that will give the study a focus or goal. A good re-
social science in the eighteenth century. Since then, search question should allow you to contribute to the
the challenges posed by urbanization, industrializa- existing knowledge of the topic, if only in a modest
tion, ethnic conflict, and gender inequality (to name way. As we have mentioned earlier, research ques-
only a few) have provided the impetus for count- tions in quantitative studies tend to be more precise
less research projects. In other cases, the process and explicit than those found in qualitative research,

bel65796_ch03_053-076 73 01/13/22 12:19 PM

74 PART I Fundamental Issues in Social Research

An Overview of the Research Process

Motivation

Research question(s)

Literature review

Choose general orientation

Quantitative Qualititative
(choose design) (choose design)

Experiments Ethnography
Surveys Participant observation
– Structured interviews Unstructured interviews
– Questionnaires Semi-structured interviews
Structured observation Life history
Case studies Oral history
Narrative analysis
Ethics review Focus groups
Content analysis
Gather data – Ethnographic
– Semiotics
– Hermeneutics
Analyze data
– Conversation analysis
– Discourse analysis
Findings

Ethics review
Conclusions
– Do the findings support the hypotheses?
– Implications for existing theories? Gather data
– New hypotheses, concepts, or theories
generated? Analyze data
– Implications for further research?
– Relevance to public policy?
Findings

Conclusions
– Do the findings provide a new interpretation
of a person, group, setting, event, or text?
– Implications for existing theories?
– New hypotheses, concepts, or theories
generated?
– Implications for further research?
– Do the findings illustrate group processes
or social conflict in a new way?
– Relevance to public policy?

bel65796_ch03_053-076 74 01/13/22 12:19 PM

 APPENDIX TO PART I An Overview of the Research Process 75

although qualitative researchers must have at least a Of course it’s also possible to combine quantita-
general idea of what they are trying to accomplish. tive and qualitative approaches in a single study, as
With both types of inquiry, it’s fairly common to the arrows below “Choose general orientation” in the
revise research questions as the project proceeds, es- figure suggest. In practice, however, most studies are
pecially in the early stages. predominantly of one sort or the other.

Literature review Picking a research design

from the ones available
Chances are you won’t be the first person to investi-
gate your topic or to use the methodology you have within a general research
in mind. Reading up on what others have done will orientation
help you understand your subject matter more fully
Once your research orientation has been decided,
and will prevent you from venturing into territory
a number of research designs are available for con-
that has already been thoroughly explored. A good
sideration. Some of the designs listed in the figure
knowledge of the literature may also reveal contra-
were not discussed in the first three chapters of
dictions or gaps in the existing understanding of
the book, but will be described later. Consult the
your topic, which can form the basis of your project.
Glossary for brief descriptions of any designs you
Sometimes a review of the literature can lead to sub-
are not familiar with.
stantial revisions in research questions; for example,
The choice of the research design depends on a
if you encounter a finding that lacks a compelling ex-
number of things, but your first concern should be
planation and decide to make it the focus of your in-
to find a design that’s appropriate for your research
quiry. Reviewing the literature may also spark ideas
questions. For example, if you’ve chosen to explore
about how various methodologies could be applied
political attitudes using quantitative methods and
to your area of interest.
want to be able to apply your findings to some
larger population, a survey design would be ap-
Choosing a general research propriate. Similarly, if you want to do a qualita-
orientation: Quantitative tive study of how professional jazz dancers cope
and/or qualitative with the pressures of auditions, then unstructured
or semi-structured interviews would be a reason-
Once you’ve come up with your research question(s) able choice of design. Another consideration is
and reviewed the existing literature in the area, you whether a cross-sectional or longitudinal design
can decide on your general research orientation. would be most appropriate to answer the research
However, it’s not uncommon to decide on a particu- questions.
lar orientation even before reviewing the literature or Other factors that affect the choice of design in-
formulating any questions. For example, researchers clude things like the availability of resources such
specializing in qualitative studies are likely to start as funding and time. If you can’t afford to select a
with the assumption that their research will be qual- large probability sample, a full-blown survey analy-
itative rather than quantitative in nature. Likewise, sis will be out of reach. Likewise, ethnography might
a researcher who has done a number of quantitative be your design of choice, but time and travel con-
survey projects may start by designing a study with straints could rule that out. In most cases, some sort
such surveys in mind, and may not even review the of compromise has to be made between the best pos-
pros and cons of approaching their topic with a qual- sible research design and what is feasible under the
itative orientation. circumstances.

bel65796_ch03_053-076 75 01/13/22 12:19 PM

76 PART I Fundamental Issues in Social Research

Ethics review associated with age and ethnicity. Or a qualitative

study of heroin addicts in Vancouver’s downtown
As we emphasized at various points in Part I of this east side might find that they resent the people from
book, ethics are an important consideration at every whom they purchase their drugs.
stage of a research project, including the writing up
of results. Once a research design has been picked, Conclusions
researchers in most cases have to present an ethics
committee with a proposal that outlines how ethical Finally, what does it all mean? At this point an effort
considerations will be handled in each phase of the is usually made to relate the findings of the study to
study. Note that this step can usually be skipped if existing theories, knowledge, or interpretations of
the research does not involve human or animal par- the topic. This is where you enter the debate about
ticipants; for instance, if the goal is to offer a new the subject matter with other scholars.
interpretation of a body of theoretical work without Quantitative social scientists will discuss whether
gathering empirical data. their findings support their hypotheses (if they had
specific hypotheses to begin with—not all of them do).
Gather data Support for a hypothesis could provide substantiation
for a particular theory. For instance, findings indicating
Data gathering can begin as soon as the ethics com- that economic development leads to greater similarity
mittee grants its approval. Sometimes a pilot study is in gender roles would support a version of modern-
conducted to determine whether the chosen method ization theory. If the hypotheses are not supported
of data collection needs any refinement. For exam- or if the findings are not consistent with the theory, a
ple, a pilot study can be used to find out whether case can be made that the theory has to be revised or
the questions prepared for use in a survey are clear abandoned. Sometimes the findings will generate new
enough for respondents to understand, or whether hypotheses, concepts, or theories. In virtually all cases,
potential participants in a qualitative study react the researcher will be able to suggest ways in which
strongly to the presence of the observer. further research could be beneficial—perhaps different
variables should be examined, or maybe the key vari-
Analyze data ables need to be measured differently.
Qualitative researchers will discuss whether their
In quantitative studies, data analysis usually begins findings suggest a new interpretation or conceptual-
after all the data have been gathered. In qualitative ization of the subject matter, or whether they provide
research, by contrast, the process is often iterative, a new illustration of group processes or social con-
with preliminary data analysis leading to a change flict. For example, findings derived from qualitative
in either the kind of information to be sought or the interviews with 16-year-old mothers might reveal
way it is to be gathered. That is why there is a two- something new about how young people view their
headed arrow between “Gather data” and “Analyze sexuality and their relationships with their peers,
data” in the qualitative section of the figure. At this which may be quite different from what is currently
stage software may be used (for example, SPSS for believed. Similarly, qualitative researchers can usu-
quantitative analysis or NVivo for qualitative). ally outline the ways in which further research would
be desirable; for instance, by expanding the study of
Findings 16-year-old mothers to include 16-year-old fathers.
And both quantitative and qualitative studies may
The findings are the results of the data analysis. For include discussions of how their conclusions may
instance, a quantitative researcher might find that be relevant to issues of public policy, such as public
support for a particular government policy is strongly funding for daycare.

bel65796_ch03_053-076 76 01/13/22 12:19 PM

 PART II
Quantitative Research

Part II of this book is concerned with quantitative research.

Chapter 4 sets the scene by exploring its main features. Chapter
5 focuses on structured interviews and questionnaires—two of
the most widely used data-gathering techniques in quantitative
research—and includes guidelines on how to ask questions using
those instruments. Chapter 6 discusses structured observation, a
method that takes a systematic approach to the study of people in
their natural surroundings. Quantitative sampling—the selection of
people or other units of analysis to be included in the sorts of study
discussed in Part II—is the topic of Chapter 7, and an overview of
quantitative data analysis is provided in Chapter 8. These chap-
ters describe the essential tools for doing quantitative research,
from basic methodological principles to practical applications.

bel65796_ch04_077-099 77 01/13/22 12:19 PM

 4 The Nature of
Quantitative Research
Chapter Overview
This chapter is concerned with quantitative research, one of the two principal investigative orientations
taken in the social sciences. The emphasis here is on what quantitative research typically entails, though
departures from this ideal type will be outlined later on. This chapter explores:
• the main steps of quantitative research, presented as a linear succession of stages;
• the importance of measurement in quantitative research, and the ways in which measures are
devised for concepts;
• procedures for checking the reliability and validity of those measures;
• the primary goals of quantitative researchers: measurement, establishing causality (sometimes
referred to as internal validity), generalization, and replication; and
• some criticisms of quantitative research.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Identify the main steps taken in quantitative research.

2. Explain the importance of concepts and their measurement.
3. Describe what is meant by reliability and measurement validity, and explain why researchers go to con-
siderable lengths to achieve them.
4. Discuss the key goals of quantitative research.
5. Outline some shortcomings of quantitative research.

▲ Anna Kraynova/Shutterstock

bel65796_ch04_077-099 78 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 79

For his class project, Mario wants to ask some fellow people be offended by the question? And when they
students questions about a topic that’s always fas- do answer, how will he know if they’re telling the truth?
cinated him: drug and alcohol use, and how it might Is there a second method he could use, such as ob-
affect a person’s self-image. But how will he get the in- serving people in a local pub, to see if their behaviour
formation he needs? He quickly jots down a few notes, is consistent with the survey data? And once he’s asked
but soon realizes that coming up with effective ques- the basic questions about consumption, how will he
tions is more difficult than he imagined. His first ques- approach the self-image part of his research? What
tion reads: “Do you take drugs?”; but he quickly realizes exactly should he be asking about? Should the infor-
that “drugs” is too vague, since for some people that mation be recorded in numeric form? If so, how do
would include things like Aspirin and cough medicine. you do that? When he has finished analyzing the data,
And what about alcohol? Isn’t alcohol a kind of drug will he be able to say that his findings apply to some
too? Now he begins to think about other potential larger population, such as all university students? In
problems. What if some people refuse to say what their this chapter we will explore how researchers deal with
level of drug and alcohol consumption is? Would some the kinds of issues that Mario is facing.

Introduction 11. Theory

In Chapter 1, quantitative research was outlined as

2. Hypothesis
2
a distinctive research strategy involving particu-
lar epistemological and ontological assumptions.
It should be clear by now that much more than the 3. Research design
3
presence of numbers distinguishes a quantitative
from a qualitative research orientation. In very broad 4 Devise measures of concepts
4.
terms, quantitative research entails the collection of
numerical data, a deductive relationship between
5 Select research site(s)
5.
theory and research, a preference for the natural sci-
ence approach to inquiry (positivism in particular),
and an objectivist conception of social reality. This 6
5. Ethics review
chapter spells out the main steps in quantitative re-
search, and outlines how some of its main concerns,
7 Select research participants
5.
such as measurement validity, are addressed.

8 Administer research instruments/collect data

5.
The main steps in quantita-
tive research 9 Process data
5.

Figure 4.1 outlines the main steps in quantitative re-

search. The steps and the order in which they appear are 10
5. Analyze data

very much an ideal account of the process: research is

rarely as linear or straightforward as the figure implies. 11
5. Findings/conclusions
However, it represents a useful starting point for coming
to grips with the main facets of quantitative methods.
12
5. Write up findings/conclusions
The fact that the model starts off with a theory sig-
nifies a broadly deductive approach to the relation-
ship between theory and research. That a hypothesis FIGURE 4.1 The process of quantitative research

bel65796_ch04_077-099 79 01/13/22 12:19 PM

80 PART II Quantitative Research

is derived from the theory and then tested is also Step 8 is the administration of the research instru-
indicative of the deductive method. This approach ments. In experimental research, this usually means
was used by Logie, Lys, Okumu, & Fujioka (2019) pre-testing participants, manipulating the indepen-
in a study of sexually transmitted infection (STI) dent variable for the experimental group only, and
prevention strategies adopted by Indigenous and post-testing. In cross-sectional research using survey
non-Indigenous youth in the Northwest Territories. research instruments, it involves interviewing the
The research showed that depression, resilience, and members of the sample with a structured interview
food insecurity were related to STI prevention prac- schedule (a list of questions or statements to which the
tices. The authors first presented a theoretical frame- person being interviewed responds) or distributing
work and then several hypotheses that followed from a self-completion questionnaire. In research using
it, which were tested using survey data. However, a structured observation, this step involves watching
great deal of quantitative research does not specify a the setting of interest and the people in it, and then
hypothesis beforehand; instead, there may simply be recording the types of behaviours observed.
a loosely defined set of concerns in relation to which Step 9 simply refers to the fact that the informa-
the social researcher collects data. This was the case tion collected must be systematically recorded so
for Gazso-Windlej and McMullin’s (2003) study of that it can be analyzed. With some information this
how time, resources, and patriarchy contribute to can be done in a relatively straightforward way. For
the unequal spousal sharing of domestic labour in example, for information relating to matters such as
Canada. people’s ages, incomes, and the number of years they
In step 3 a research design is selected, a topic ex- spent at school, the information does not have to be
plored in Chapter 2. This choice has implications for transformed: the scores can simply be recorded in
a variety of issues, such as the external validity of the a computer file. For other variables, processing the
findings and the ability to impute causality. Step 4 en- data entails coding the information—that is, trans-
tails devising measures of concepts, a process often re- forming it into numbers to facilitate quantitative
ferred to as operationalization, a term derived from data analysis. This consideration leads into step 10:
physics to refer to the operations performed to mea- analysis of the data. In this step, the researcher
sure something (Bridgman, 1927). Further aspects of chooses among various statistical techniques to test
this issue are explored later in this chapter. for relationships between variables, tries to deter-
The next three steps involve choosing a research mine whether the measures are reliable, and so on.
site or sites, arranging for an ethics review (dis- Next, in step 11, the researcher must interpret the
cussed in detail in Chapter 3), and the selection of results of the analysis; it is at this stage that the “find-
participants or other units of analysis to be included ings” emerge. Here the researcher must consider the
in the project. In the study of delinquency among reasons why the research was done in the first place,
male adolescents described in Box 1.2, Teevan and in particular how the results may be used to answer
Dryburgh’s (2000) first choice as a research site was the research question(s). If a hypothesis was presented
a school where they had connections. When they had prior to the collection of data, was it supported? What
to arrange for permission slips to be signed, however, are the implications of the findings for the theoretical
they decided to restrict their study to boys old enough ideas that were the background to the research? Do
to give their own permission. Finally, because of a the findings support the theory? If not, should the
less-than-eager response, all boys who were willing theory be revised? Should it be abandoned entirely?
to participate became part of the project. In bigger Then the research must be written up in step 12. Until
studies, especially those in which the findings are ex- it enters the public domain in some way, as a paper to
pected to be applicable to some larger population, an be presented at a conference, a report to the agency that
elaborate sampling process (discussed in Chapter 7) funded the research, or as a book or journal article to
may be used to select research participants, although be read by other researchers or students, it cannot have
this approach is rare in experimental research. an impact beyond satisfying the researcher’s personal

bel65796_ch04_077-099 80 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 81

curiosity. In writing up their findings and conclusions, A concept can be either an independent or a de-
researchers must do more than simply relay their results pendent variable. In other words, a concept can be
to others: they must convince readers that the research presented as a possible cause of a certain aspect of the
conclusions are important and the findings robust. social world (independent), or it can represent some-
A significant part of doing research is convincing thing needing an explanation (dependent). Often
others of the relevance and validity of one’s findings. the same concept will be an independent variable in
Once the findings have been published, they one context and a dependent variable in another. The
become part of the stock of knowledge in their concept “social mobility,” for example, can be used in
domain. Thus, there is a feedback loop from step 12 either capacity: as a possible explanation of certain
back to step 1. The presence of both an element of attitudes (are there differences between the down-
deduction (steps 1 and 2) and induction (the feed- wardly mobile and others in their political attitudes?)
back loop) is indicative of the positivist foundations or as something to be explained (why are some people
of quantitative research. The emphasis on the trans- upwardly mobile and others not?). Equally, concepts
lation of concepts into measures (step 4) is also a fea- may be used for descriptive or comparative purposes.
ture of positivism, and it is to that important phase For instance, a researcher might be interested in
of the research that we now turn. As we will see, changes in the amount of social mobility in Canada
certain considerations follow from the need for mea- over time, or in variations among nations in their
surement in quantitative research—in particular, levels of social mobility.
finding valid and reliable ways to measure concepts.
Why measure?
There are three main reasons for the attention given
Concepts and their to measurement in quantitative research:
measurement
• Measurement can allow for the delineation of
What is a concept? fine differences between people in terms of the
Concepts are ideas or mental representations of characteristic in question. This is very useful
things. They are the building blocks of theory and since, although it is often easy to distinguish
represent the points around which social research is between people in terms of extreme categories,
conducted. Just think of the numerous concepts that finer distinctions are much more difficult to
have already been mentioned so far in this book: recognize. Clear variations in levels of job satis-
faction—people who love their jobs and people
emotional labour, secularization, hyperactiv-
who hate their jobs—are easy to see, but small
ity, academic achievement, teacher expecta-
differences are much more difficult to detect.
tions, crime, research ethics, gatekeepers, drug
• Measurement can provide a consistent device
and alcohol use, and self-image
or yardstick for gauging such distinctions. This
Each represents a label given to elements of the consistency relates both to the time at which
social world that seem to be significant and to have the measure is administered and to the people
common features. As Bulmer (1984) succinctly puts who are administering it. A measure’s depend-
it, concepts “are categories for the organization of ability should not be affected by either of those
ideas and observations” (p. 43). One item mentioned things. With regard to time, this rule means that
in Chapter 2, but missing from the list above is IQ. It the measure should generate consistent results
has been omitted because it is not, strictly speaking, unless the phenomenon or characteristic being
a concept. It is a measure of the concept of intelli- measured has changed. For example, measures
gence. This is a rare case for social science, a measure of worker morale administered at different times
so well known that the measure and the concept are should indicate the same level of morale unless
almost synonymous. the morale itself has changed. Similarly, the data

bel65796_ch04_077-099 81 01/13/22 12:19 PM

82 PART II Quantitative Research

produced by the measure should be the same • through the use of official statistics (e.g., those
regardless of who did the measuring. Whether from Statistics Canada); and
a measure actually possesses this consistency • by developing classification schemes to an-
pertains to the issue of reliability, which will be alyze the content of written material (e.g.,
examined later in this chapter. Sobel’s [2015] assessment of Canadian citizen-
• Measurement provides the basis for estimates of ship guides issued between 1947 and 2012 as
the nature and strength of the relationship between indicators of how the federal government de-
variables (for example, through correlation anal- fined what it means to be a Canadian and how
ysis, which will be examined in Chapter 8). naturalized citizens were expected to behave).

Indicators Indicators can be derived from a wide variety of

Concepts used in quantitative research are given sources and methods. Very often a researcher has
two types of definition. One is a nominal definition, to consider whether one indicator of a concept suf-
which describes in words what the concept means, fices. Instead of relying on a single indicator, the re-
much as a dictionary definition does. For example, searcher may use several to tap a concept. One way
a nominal definition of “crime” might be: “any vio- to do that is to combine a number of Likert items
lation of the Criminal Code of Canada.” By contrast, (see Box 4.1).
a concept’s operational definition (from the idea
of operationalization, discussed above) spells out Using multiple-item measures in
the operations the researcher will perform to mea- survey research
sure the concept. For example, one way to measure The main advantage in using a multiple-indicator
the incidence of crime is to use statistics provided measure of a concept is that there are potential prob-
by police forces; another is to ask a sample of people lems in relying on just a single indicator:
whether they have been victims of certain crimes.
To measure a concept it is necessary to have an • A single indicator may misclassify some indi-
indicator or indicators that stand for or represent the viduals if the wording of the question leads to
concept. Sometimes the term indicator simply refers a misunderstanding of its meaning. If there are
to a measure in the ordinary sense: for example, the several indicators, however, it is possible that
total income reported on a tax return is an indicator of the other items will allow proper classification
a person’s wealth. However, sometimes it refers to an of the person.
indirect measure of a concept that cannot be tapped • A single broad, general indicator may not cap-
easily or directly. For example, absenteeism may be ture all the meaning in the underlying concept.
used as an indirect measure of employee morale. If you were to ask people how satisfied they are
Income declared for tax purposes may be a direct with their work, for instance, you might miss
measure of personal income, but if used as an indi- the complexity of the situation. People who
cator of social class, it becomes an indirect measure. answer “not satisfied” may like some parts of
There are a number of ways to devise indicators; their work, just as those who claim to be “satis-
for example: fied” may dislike specific aspects of their work.
• Alternatively, a more specific question may
• through questions that are part of a structured cover only one dimension of a particular
interview schedule or self-completion ques- concept. For example, to measure job satis-
tionnaire. The questions can be concerned faction, is it sufficient to ask people how satis-
with the respondents’ attitudes (e.g., toward a fied they are with their pay? Almost certainly
social issue like immigration), their personal the answer is “no,” because for most people
experiences (e.g., stress), their behaviours (e.g., there is more to job satisfaction than just sat-
leisure pursuits), and so on; isfaction with pay. In this case, a single indi-
• by developing criteria for classifying observed cator will miss things like satisfaction with
behaviour (e.g., pupil behaviour in a classroom); benefits, with the work itself, and with other

bel65796_ch04_077-099 82 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 83

aspects of the work environment. By asking a concept in question. Bell, Aitken Schermer, &
number of questions, the researcher can get Vernon (2009), for example, used factor anal-
access to a wider range of issues covered by ysis to develop measures of political attitudes.
the concept. Cluster analysis allows researchers to deter-
• Multiple indicators of a concept allow for so- mine whether people or other units of study
phisticated data analyses, such as factor analy- can be grouped together on the basis of shared
sis and cluster analysis. Factor analysis provides characteristics. Kennedy, Krogman, & Krahn
an indication of the extent to which (a) mul- (2013), for instance, identified four clusters of
tiple measures really are measuring a single Edmonton residents based on the sustainable
concept as opposed to more than one concept; consumption behaviours in which they en-
and (b) individual items are measuring the gaged (see Box 4.1).

Box 4.1 Likert items

The investigation of attitudes is a promi- “Rarely,” “Sometimes,” “Often,” or “Always” as
nent area in survey research. One of the most their response.
common techniques for investigating attitudes There are several points to remember in com-
is the Likert scale. Named after Rensis Likert, bining Likert items. The following are particularly
who developed the method, it is typically a important:
multiple-indicator measure of the intensity of • The items must be statements, not questions.
feelings about a particular topic. In its most • The items must all relate to the same topic (in
common format, it comprises a series of state- the example above, sustainable consumption).
ments (known as “items”) on an issue or theme. • The items that make up the scale should be
Usually, each respondent is asked to indicate his interrelated.
or her level of agreement with the statement. • It is useful to vary the phrasing so that some
The format for indicating level of agreement items imply a positive view of the phenomenon
is normally a five-point scale from “strongly of interest and others a negative one. For ex-
agree” to “strongly disagree,” but seven-point ample, in the 2007 Alberta Survey conducted
and other formats are used as well. There is by the Population Research Laboratory at the
usually a middle position of “neither agree nor University of Alberta, attitudes toward body
disagree” or “undecided” indicating neutrality checking in minor hockey were measured
on the issue. A respondent’s reply on each item using five-point Likert items. One item read:
is scored and then the scores are aggregated to “Body checking is the main cause of injuries for
form an overall score. children in organized minor hockey.” A second
Variations on the typical format of agreement was: “Body checking is an essential part of
are scales referring to evaluation (for exam- the game of organized minor hockey.” The
ple, “very poor” to “very good”) and frequency. first implies a negative view of body checking,
Kennedy et al. (2013) used the latter in a study the second a more positive one. Under these
of sustainable consumption that was conducted conditions, it would be logically inconsis-
in Edmonton. Respondents were presented with tent to answer every question the same way,
several items such as “Leave vehicle at home so if someone has done that it indicates that
to avoid driving,” “Buy local food,” and “Actively response set may be a problem. A respondent
encourage others to reduce their consump- who either agreed or disagreed with all the
tion,” 12 of which were selected for analysis. items could not be answering truthfully, and
Participants in the study chose either “Never,” might have to be excluded from the study.

bel65796_ch04_077-099 83 01/13/22 12:19 PM

84 PART II Quantitative Research

• Multiple indicators can help to weed out idea is that people scoring high on one dimension
response sets (see Chapter 5), which can of the concept (e.g., respecting confidentiality) may
occur if, for a variety of reasons, a person’s not necessarily score high on other dimensions (e.g.,
response does not reflect his or her real po- fiscal honesty or continuing education), so that for
sition on the matter examined. For example, each respondent one can have a multidimensional
someone may “agree” with an item or set of “profile.”
items just to appear cooperative or friendly. However, in much quantitative research, there
Multiple indicators can be used to test for is a tendency to rely on a single indicator for each
this. For instance, if on a particular item concept. This is quite adequate for some pur-
someone agrees that unemployment is the poses, in particular when one is measuring an
most important problem facing Canadians uncomplicated variable such as age. Some stud-
today, but later also agrees that Canada has ies, like Hay’s (2014) research on secularization
bigger problems than unemployment, that’s in Canada (see Box 4.2), employ both single- and
an indication of response set. In such situa- multiple-indicator measures of concepts. What is
tions the researcher may chose to exclude the crucial is that the measures be reliable and valid
person’s responses from the data set. representations of the concepts they are supposed
to be gauging.
Dimensions of concepts
As was suggested earlier, the concept of interest may Coding unstructured data
have multiple facets or dimensions, which are often Many forms of social science data are essentially
revealed in the theories and research associated with unstructured and unorganized, including answers
it. In developing a measure for a concept, its different to open questions in interviews and questionnaires,
aspects or components should be considered. Bryman and the content of newspaper articles. To make sense
and Cramer (2001) demonstrated this approach with of the information, researchers must go through it
reference to the concept of “professionalism.” The all, deriving themes or categories of behaviour to

Box 4.2 A multiple-indicator measure of another concept

In Hay’s (2014) study of secularization (see Box relating to frequency of religious attendance,
1.3), religious pluralism was measured using a was measured with the question, “Do you
single, five-point Likert item that formed part currently attend church temple or mosque?”
of an extensive survey of Canadians’ value Respondents who answered “yes” were then
systems. However, secularization (the depen- asked: “How often?” The response choices
dent variable in Hay’s analysis) was measured were: “once a week or more” (given a code of
with several different indicators in order to tap 5 after reverse-coding), “monthly” (4), “every
into different dimensions of the concept. One few months” (3), “once or twice a year” (2), and
dimension, religiosity, was measured by av- “never” (1). A third dimension, concerned with
eraging the responses to three 10-point items the participants’ belief in the religion of their
indicating the importance respondents placed parents, had the response categories “believe
on: (a) “believing in God;” (b) “obeying God, all of it” (4), “believe most of it but not all” (3),
doing what he wishes;” and (c) “relating to “believe some parts but disbelieve others” (2),
God in a personal way.” A second dimension, and “don’t believe any of it” (1).

bel65796_ch04_077-099 84 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 85

form the basis for codes (the labels or titles given experiences with either the Ontario or the National
to the themes or categories): for example, “hostile Sex Offender Registry. People on these registries
to outsiders,” or “not hostile to outsiders.” Next, the have to report annually to their local police and
researcher usually assigns numbers to the codes. receive a yearly visit from the police at their resi-
This may be a largely arbitrary process in the sense dence, but unlike in the United States, the registries
that the numbers are simply tags to allow the mate- themselves are not available for public viewing. The
rial to be stored quantitatively. Then it is necessary responses to the open-ended questions were an-
to go through the information again to look for in- alyzed to discern whether there was consensus or
cidences of the theme or category, and to record the disagreement among participants and to identify
appropriate numbers on a computer spreadsheet. recurrent themes. About two-thirds of the offend-
This approach is sometimes called post-coding. Post- ers who commented on the matter stated that being
coding can be an unreliable procedure because there on a sex offender registry was not a great burden or
may be inconsistencies in the judgments of different even an intrusive experience, and about half said
coders, which leads to both measurement error and that having to register was “fair,” given the crimes
lack of validity. Box 4.3 provides an example of this they had committed. However, some offenders ob-
kind of coding. jected to the manner in which the police conducted
When Schuman and Presser (1981; see Box 5.1) their annual visits.
asked an open question about the features of a job When coding, you should observe three basic
that people like, the answers were to be grouped into principles (Bryman & Cramer, 2001):
11 codes: pay; feeling of accomplishment; control of
work; pleasant work; security; opportunity for pro- • The categories must not overlap. If they do, the
motion; short hours; working conditions; benefits; numbers assigned to them will not denote dis-
satisfaction; other responses. Each of these 11 cate- tinct behaviours or types of responses.
gories was assigned a number: 1 for pay; 2 for feeling • The list of categories must be exhaustive,
of accomplishment; 3 for control of work; 4 for pleas- covering all possibilities. If it does not, some
ant work, and so on. material cannot be coded. That is why many
Murphy and Fedoroff (2013) used a combination classification schemes, such as those used to
of fixed-response and open-ended questions to ex- code open questions, include the category
plore how 30 registered sex offenders viewed their “other.”

Box 4.3 Coding an open question

Foddy (1993) asked a small sample of his stu- with the difficulty of coding an open question; it
dents to state their fathers’ occupations; then he is also due to a lack of specificity in the question.
requested three details: nature of business, size If Foddy had asked, “How many employees are
of business, and whether their father was the (were) there in your father’s place of employ-
owner or an employee. The replies to the “size ment?” a more meaningful set of answers would
of business” question were particularly varied, have been forthcoming. Whether his students
including “big,” “small,” “very large,” “3000 would have known this information is, of course,
acres,” “family,” “multinational,” “200 people,” yet another issue. However, his experience il-
and “Philips.” The problem here is obvious: lustrates some potential problems with asking
these categories do not provide a useful mea- an open question, particularly one that lacks a
sure of size. The problem has only partly to do clear reference point.

bel65796_ch04_077-099 85 01/13/22 12:19 PM

86 PART II Quantitative Research

• There should be clear rules about how codes low, the measure is unstable, implying that it cannot
are to be applied, with examples of the kinds be relied upon.
of answers that may be subsumed under a par- However, there are a number of problems with
ticular category. Such rules are meant to pro- this approach to evaluating reliability. Respondents’
mote a coder’s consistency over time in how answers at T1 may influence how they reply at T2: for
the material is assigned to categories and, if instance, perhaps answering a question on designer
more than one person is coding, that the vari- goods increases some respondents’ interest in them.
ous coders are consistent with each other. Yet giving an answer at T1 is not supposed to affect
later measurements. Second, events may intervene
Coding is also used in qualitative research, but its between T1 and T2 that influence the degree of consis-
role and significance there are somewhat different tency. For example, if a long span of time is involved,
(see Chapter 13). changes in the economy or in respondents’ personal
financial circumstances can influence their views
Reliability and measurement about designer goods and their predilection for them.
validity There are no clear solutions to these problems,
other than introducing a complex research design
Reliability and so turning the investigation of reliability into
Reliability is concerned with the consistency of mea- a major project in its own right. Perhaps for these
sures. The term has at least three different meanings. reasons, many, if not most, research projects do not
include tests of stability over time.
Stability over time When such tests are done, they may indicate that there
“Stability” refers to whether the results of a measure is no problem, even when a substantial amount of time
fluctuate as time progresses, assuming that what is has elapsed. Berthoud (2000), for example, was pleased
being measured is not changing. This means that that an index of illness achieved high test–retest reliability
if one administers a measure to a group and then even though the tests were conducted a year apart.
re-administers it perhaps an hour later, there should
Internal reliability (or internal consistency)
be little variation in the results. Most thermometers
have this kind of reliability. The key issue here is whether multiple measures that
The most obvious way of testing for the stability are administered in one sitting are consistent—in
of a measure is the test–retest method. This involves other words, whether respondents’ scores on any
administering a test or measure on one occasion and one indicator tend to be related to their scores on
then re-administering it to the same sample on an- the other indicators. For example, on a scale cre-
other occasion, that is: ated to measure attitudes toward liberal democracy,
people who agree with the statement that voting is
T1 T2 an important right should also agree that freedom of
Obs1 Obs2
speech is an important right. Cronbach’s alpha coef-
One would expect to find a high correlation be- ficient is a commonly used test of internal reliability.
tween Obs1 and Obs2. For example, those who score Its value varies from 1 (denoting perfect internal reli-
high on the first observation should also score high ability) to 0 (denoting none). The value .80 is typically
on the second, and those who score low on the first employed to mark the minimum acceptable level of
should score low on the second. Imagine a multi- internal reliability, though many writers work with
ple-indicator measure that is supposed to tap a con- a lower value. Berthoud (2000, p. 169), for instance,
cept called “designerism” (a preference for buying wrote that a minimum level of .60 is “good.” In the
goods and especially clothing with “designer” case of George and Chaze’s (2014) battery of items
labels). The measure would be administered to a to measure discrimination against foreign trained
sample of respondents and then later re-adminis- engineers in Canada, alpha was .92 for both the
tered. If the correlation between Obs1 and Obs2 is international and locally trained subsamples. In a

bel65796_ch04_077-099 86 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 87

study of environmental values among Chinese- and Measurement validity

Anglo-Canadians by Deng, Walker, & Swinnerton As we noted in Chapter 2, measurement validity
(2006), the alphas ranged from .50 to .89 for various refers to whether an indicator (or set of indicators)
variables. The study of domestic labour by Gazso- devised to gauge a concept really does so. When
Windlej and McMullin (2003) could only achieve people argue about whether an IQ score really
alphas averaging .50 for its Likert-style items mea- measures or reflects someone’s level of intelligence,
suring gender ideology, although the study was only they are raising questions about the measurement
exploratory. The use of Cronbach’s alpha has grown validity of the IQ test in relation to the concept of
as a result of its incorporation into computer soft- intelligence. Whenever students and their teachers
ware for quantitative data analysis. debate whether multiple-choice examinations pro-
Another way of testing for internal reliability is vide an accurate measure of academic ability, they
the split-half method. Take the Likert items used too are raising questions about measurement valid-
by Kennedy et al. (2013) in their study of sustain- ity. Establishing face validity, concurrent validity,
able consumption (see Box 4.1). The 12 indicators construct validity, and convergent validity provides
would be divided into two halves of 6, allocated on evidence that you have measured what you wanted
a random or an odd-numbered item/even-numbered to measure.
item basis. The degree of correlation between the
scores on the two halves for all respondents would Face validity
then be calculated. If the 12 items are consistent, a At the very minimum, a researcher who develops
respondent’s score on the two groups of indicators a new measure should establish that it has face va-
should be similar, perhaps high on both, or low on lidity—that is, that it appears to reflect the content
both. A perfect positive correlation and therefore of the concept in question. Face validity can be es-
complete internal consistency would yield a correla- tablished by asking those with expertise in a field to
tion coefficient of 1; no correlation and therefore no act as judges to determine whether, on the face of it,
internal consistency would produce a coefficient of 0. the measure seems to reflect the concept concerned.
The meaning of correlation and correlation coeffi- Alternatively, researchers can assess their own mea-
cients will be explored in much greater detail later sures for face validity, as Kennedy et al. (2013) did in
on. The chief point at this stage is that the correla- their study of sustainable consumption (see Box 4.1).
tion coefficient establishes how closely respondents’ Establishing this sort of validity is essentially an in-
scores on the two groups of indicators are related. tuitive process.

Inter-observer consistency Concurrent validity

When more than one researcher is taking part in an The researcher can also seek to gauge the concur-
activity involving subjective judgment, such as the rent validity of the measure by examining a crite-
recording of observations or the translation of data rion relevant to the concept in question at about
into categories, their judgments may differ. This the same time as the measure of interest is ad-
problem can arise in a number of contexts: for ex- ministered. Assuming that as job satisfaction goes
ample, if answers to open-ended questions have to down, absenteeism goes up, a researcher seeking
be categorized, or in structured observation when to establish the concurrent validity of a new mea-
observers have to decide how to classify participants’ sure of job satisfaction might look to see if people
behaviour. Is the person under observation “afraid,” who, according to the new measure, are satisfied
“concerned,” or “just thinking” when reading about with their jobs are less likely to be absent from
the spread of COVID-19? Problems arise if all ob- work than those who are not satisfied. If there is
servers do not classify a particular behaviour in the no difference in absenteeism between those who
same way. Cramer (1998, ch. 14) provides a very de- are satisfied and those who are not, doubt is cast
tailed treatment of inter-observer consistency and on whether the new measure is really gauging job
the techniques that can be used to maximize it. satisfaction.

bel65796_ch04_077-099 87 01/13/22 12:19 PM

88 PART II Quantitative Research

Construct validity lot of variety. If so, the measures used may be invalid,
Some writers advocate that researchers estimate the or the deduction that is made from the theory may be
construct validity of a measure, which involves seeing misguided, or the theory itself may be in need of revi-
whether the concepts used in the research relate to sion. Whatever the problem may be, it’s probably best
each other in a way that is consistent with what their to seek another measure and try again.
theories would predict. For example, suppose that a
Convergent validity
researcher wants to establish the construct validity
of the new measure of job satisfaction. Drawing on a In the view of some methodologists, the validity of a
theory maintaining that job satisfaction is influenced measure ought to be gauged by comparing it to mea-
by the stimulation that comes from performing sev- sures of the same concept developed through other
eral different activities, he or she may anticipate that methods. For example, if a questionnaire asks man-
people who do routine jobs are less satisfied with their agers how much time they spend on various activi-
jobs than those who have greater opportunities for ties (attending meetings, touring their organization,
variety, complexity, and creativity. If these variables informal discussions, and so on), its validity may be
correlate in the expected way, then the measure in determined by directly observing the managers to see
question has construct validity. On the other hand, how much time they actually spend on those activities.
some caution is required if the relationship is weak or An interesting instance of convergent invalidity
non-existent—for example, if those who do routine is described in Box 4.4. Crime surveys were con-
jobs are just as satisfied as those with jobs involving a sciously devised to serve as a check on official police

Box 4.4 A case of convergent invalidity: Crime statistics

show more crime than the official police reports
because many Canadians do not report their ex-
periences with crime to the police, for a variety of
reasons (inconvenience, embarrassment, etc.). To
take one example, for robbery the victimization
rate reported in the surveys is almost three times
Steve Jolicoeur/Shutterstock

the official rate. But for auto theft, the police data
show 50 per cent more thefts, because the GSS
does not gather data on the theft of company cars
or cars taken from car dealerships (Silverman et
al., 2000, p. 58). So there will probably always be a
discrepancy between police statistics and victim-
Police gathering evidence at a crime scene. The
ization surveys on the amount of crime that exists.
crime will definitely be entered into the official
crime statistics. Why will it become part of the offi-
Another illustration of a lack of concurrence be-
cial record while other crimes might not? tween two methods of data gathering pertains to
murder rates. The police report 15 per cent more
Official police reports and national crime victim- homicides in Canada than does the Mortality
ization studies often provide very different indi- Database drawn from death certificates (Gabor,
cations of the amount of crime in society. As part Hung, Mihorean, & St-Onge, 2002); some officials
of the General Social Survey (GSS), Canadians in charge of the latter may be reluctant to clas-
are asked about their victimization experiences. sify certain deaths as homicides because doing so
It is widely known that the victimization data might upset the relatives of the victims.

bel65796_ch04_077-099 88 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 89

statistics. The two sets of data are collected in quite 11. A criminal offence
different ways: official crime statistics are collected
as part of the bureaucratic processing of offenders in
the criminal justice system, whereas crime victim- 2. Member of the public
2
is witness to or victim
ization surveys provide data from interviews with of a crime
members of the general public. In the case reported
in Box 4.4, a lack of convergent validity was found.
This illustrates a problem with the convergent ap-
proach to testing validity: it’s not easy to establish 3. Recognition of offence
3
as criminal by a member
which of the two measures provides the more accu- of the public
rate picture. In the case of crime statistics, the two
methods are really measuring somewhat different 5 Police decisions
5.
about patterns of
things. Victimization surveys measure the crimes surveillance
4 Reporting of offence
4.
that people experience, as well as some things that
by a member of the
may not be crimes; they also omit certain crimes public
(for example, when a stolen item is presumed to
6 Police discretion
be “lost”) and rely on respondent memory, which
can be faulty (see Chapter 5). Police statistics, on
the other hand, measure the crimes that people are
willing to call the police about, plus the crimes the 7 Recording of offence
by police
police themselves discover; therefore, not all crimes
end up in official police statistics. Figure 4.2 illus-
trates some of the problems associated with crime
rates that are based on police statistics. At any step 8 Notifiable crimes
along Figure 4.2, some potential crimes drop away. recorded by the
police—”the crime
For example, if a criminal act goes unnoticed, or is rate”
noticed but not recognized as criminal, or is noticed FIGURE 4.2 The social construction of crime statis-
and recognized as criminal but is not reported to tics: Eight steps
the police, it does not enter the official statistics. As Source: Adapted from a figure in Beardsworth, Bryman, Ford, & Keil (n.d.).

you can see, a substantial amount of crime—often

referred to as “the dark figure”—is not included It should also be remembered that, although reli-
in crime statistics (Silverman, Sacco, & Teevan, ability and validity are analytically distinguishable,
2000). The “true” volume of crime at any one time they are related: if a measure is not reliable, it cannot be
is almost always a contested notion (Reiner, 2000). valid. For example, a multiple-indicator measure lack-
ing internal reliability is probably measuring two or
Reflections on reliability and validity more different things and therefore is not a valid indica-
There are, then, a number of ways to investigate the tor of the single concept it is supposed to be measuring.
merit of measures devised to represent social scien-
tific concepts. However, the discussion of reliability
and validity is potentially misleading because not The main goals of
all new measures of concepts are submitted to the ­quantitative researchers
rigours just described. In fact, the reliability and
validity of many measures are never tested. When Measurement
a new multiple-indicator measure is devised, there To understand the social order one must be able to
may be an examination of face validity and a test for go into the world and measure things like prejudice,
internal reliability, but in many cases no further test- the incidence of homelessness, attitudes toward the
ing takes place. federal government, or the distribution of wealth in

bel65796_ch04_077-099 89 01/13/22 12:19 PM

90 PART II Quantitative Research

society. It would be awfully hard to make sense of suggested in Chapter 2, establishing causality in the
such phenomena or to evaluate theories and inter- social sciences is very difficult, and that correlation
pretations of them without acquiring data on them: alone is not sufficient to demonstrate causation.
that is, without measuring them. It’s not surprising, When an experimental design is employed, the
then, that the reliability and validity of such mea- independent variable is the one that is manipulated,
surement are matters of concern. and there is little ambiguity about the direction of
any causal influence. However, with cross-sectional
Establishment of causality designs of the kind used in most social survey re-
(internal validity) search, there is often ambiguity about the direction
Most quantitative research involves a search for of the purported causal influence because the data
causal explanations. Quantitative researchers are on all variables are collected simultaneously: this
rarely satisfied with merely describing how things means that in many instances we cannot say with
are: they are keen to find out why things are the way full confidence that a particular independent vari-
they are—an emphasis also found in the natural sci- able preceded the dependent one in time, one of the
ences. Researchers examining racism, for instance, three criteria of nomothetic causation mentioned in
may want not only to describe it but also to explain it, Chapter 2. To refer to independent and dependent
which means searching for its causes. They may seek variables in the context of cross-sectional designs,
to explain racism in terms of personal characteristics we must infer the temporal sequence of variables
(such as a low level of education) or social character- based on common sense or prior theory, as in the ex-
istics (such as low social mobility). In the resulting re- ample concerning level of education and racism in
ports, racism is the dependent variable, the one to be the previous paragraph. However, there is always the
explained, and levels of education and social mobility risk that the inference is wrong, that the variable pre-
are independent variables, the ones tested as possible sumed to be the cause did not precede the dependent
causal influences on racism. Keep in mind that, as variable in time (see Box 4.5).

Box 4.5 The case of displayed emotions in convenience stores

Following a review of the literature, Sutton and Sutton and Rafaeli (1992, p. 124) considered re-
Rafaeli (1988) hypothesized a positive relation- stating their hypothesis to make it seem that
ship between the display of positive emotions by they had found what they had expected (which
staff to retail shoppers (smiling, friendly greet- would have raised ethical concerns) but fortu-
ing, eye contact) and the level of retail sales. In nately resisted the temptation. Instead, they con-
other words, when retail staff are friendly and ducted a qualitative investigation of four stores in
give time to shoppers, higher sales follow. Sutton order to understand what was happening. They
and Rafaeli had data from 576 convenience used several methods: unstructured observa-
stores in a US national retail chain. Structured tion of interactions between staff and customers;
observation of the retail workers provided data semi-structured interviews with store manag-
on the display of positive emotions, and quan- ers; casual conversations with store managers,
titative sales data provided information for the supervisors, executives, and others; and data
other variable. gathered through posing as a customer. The
The hypothesis was not supported; indeed, qualitative investigation suggested that the rela-
stores in which retail workers were less inclined tionship between the display of positive emotions
to smile, be friendly, and so on, had better sales. and sales was indeed negative, but that sales

bel65796_ch04_077-099 90 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 91

were likely to be a cause rather than a conse- More displayed positive emotions →
quence of the display of emotions. In stores with More retail sales
high levels of sales, staff were under greater but
More retail sales → Fewer displayed
pressure and encountered longer lineups at
positive emotions
checkouts. Therefore, they had less time and in-
clination for the pleasantries associated with the This study illustrates the difficulty associated
display of positive emotions. with inferring causal direction when using a
Thus, the causal sequence was not cross-sectional research design.

Good quantitative research inspires confidence findings will be externally invalid. Measurement
in the researcher’s causal inferences. Research of the validity and plausible evidence of causation cannot
more experimental type is often better able to estab- guarantee this form of external validity.
lish causality than cross-sectional research because A second concern with external validity is
with the former it is easier to demonstrate the direc- whether the results of a study can be generalized
tion of causal influence and to control for other inde- beyond the people or cases analyzed by the re-
pendent variables. However, as we saw in Chapter 2, searcher. The researchers conducting the study in
experiments are not appropriate for many of the Box 1.2 asked 57 adolescent male volunteers about
topics of interest to social scientists. Instead, quan- their delinquency and examined whether any of
titative researchers who employ cross-sectional de- the currently popular sociological theories of delin-
signs often use techniques that allow them to make quency could explain their behaviour. How far can
at least tentative causal inferences, such as statistical the findings of this study be generalized? To other
controls (see Chapter 8). The rise of longitudinal re- male adolescents who attended the same school? To
search almost certainly reflects a desire on the part all Canadian male adolescents? If the findings are
of quantitative researchers to improve their ability to fully externally valid, they can be applied to any
generate findings that permit a causal interpretation. population of interest. If they possess no external
validity, the findings apply only to the 57 boys stud-
Generalization of findings ied and to no one else. Usually, external validity falls
In quantitative research, researchers usually want to somewhere between these extremes, and it is in this
generalize their findings beyond the people who partic- context that how people are selected to participate in
ipate in the study and the particular context in which research becomes crucial. If a representative sample
the research is conducted. These concerns pertain to of people is selected (see Chapter 7), the researcher
external validity, an issue introduced in Chapter 2. can be confident that the results of the study may be
A particularly important matter in this regard is applied to the population from which the sample was
whether a study’s findings are applicable in settings drawn. The goal of arriving at generalizable findings
outside the research environment—in particular, in this way can be seen as a parallel to the goal of
everyday or natural social settings. Sometimes the developing scientific “laws” in the natural sciences.
research setting is so artificial or so different from Probability sampling, explored in Chapter 7, is
real life that we may question whether the results ac- normally the first choice among researchers seek-
tually tell us anything about the sorts of things that ing a representative sample. This procedure largely
people normally experience. The more the social sci- eliminates bias by using a random selection process.
entist intervenes in natural settings or creates unnat- Random selection does not guarantee a representa-
ural ones, such as a laboratory (or even just a special tive sample, but it does improve the chances of get-
room for interviews), the greater the chance that the ting one. Here it is important to emphasize that even

bel65796_ch04_077-099 91 01/13/22 12:19 PM

92 PART II Quantitative Research

a perfectly representative sample would be represen- Replication

tative only of the population from which it was se- The natural sciences are often depicted as seeking to
lected: strictly speaking, therefore, results obtained reduce to a bare minimum the contaminating influ-
from that sample cannot be generalized beyond that ence of the scientist’s biases, values, characteristics,
population. This means that, if the population from and expectations. Were those influences allowed to
which a sample is taken is all inhabitants of a partic- affect research, the claims of the natural sciences to
ular town, city, or province, generalizations should provide a definitive picture of the world would be
be made only to the inhabitants of that town, city, seriously undermined. Scientists also try to mini-
or province. It may be very tempting to see the find- mize if not eliminate routine errors in the conduct
ings as having a wider applicability, so that results of their research. To guard against such problems,
from a sample representing a city such as Vancouver scientists believe that they should be able to replicate
or Toronto could be considered relevant to similar one another’s research. If a certain finding cannot
cities. Even so, no inferences should be made beyond be reproduced after repeated tries, serious questions
the population from which the sample is selected. are raised about its validity. Likewise, researchers in
The concern with the generalizability of research the social sciences often regard the ability to repli-
findings is particularly strong among quantitative cate as an important criterion for judging research.
researchers using cross-sectional and longitudinal It’s easy to see why: the risk that the researcher’s
designs. Experimental researchers are concerned values will affect his or her findings would appear to
about generalizability too, as the discussion of exter- be all the greater in the case of the social world, and
nal validity in Chapter 2 suggested, but people doing routine errors are always possible. Consequently, re-
experiments usually give greater attention to estab- searchers usually take care to spell out all research
lishing causation than to external validity. procedures so that others can replicate them.

Methods in Motion Applications to Canadian Society

How the “field” may influence men’s bodies, their self-acceptance, and their views on
cosmetic surgery
Ricciardelli and Clow (2009) examined two rarely One of the strengths of the study was that it
studied topics: men’s attitudes toward their phys- combined quantitative and qualitative methods.
ical appearance and their views on whether it is For the former, the authors used 18 Likert items
appropriate for men to have cosmetic surgery. The (see Box 4.1 for examples of these sorts of mea-
purpose of the study was to determine whether sures) administered online. The items were used
men’s self-confidence, self-deprecation (feelings to gauge the men’s perceptions of their physical
of low self-esteem), and levels of comfort regard- appearance, their satisfaction with their bodies,
ing their bodies affected their assessment of their how others judged their bodies, and their views
physical appearance and their acceptance of on cosmetic surgery. The items were analyzed
cosmetic surgery. The people participating in the using multiple regression (a topic discussed in
research were 103 Canadian men who lived in the Chapter 8). The qualitative element of the study
greater Toronto area. Although the sample was was a single, open-ended question that asked
small in size and not selected at random (which the men to “please describe your feelings to-
meant that the results could not be generalized wards cosmetic surgery.” The answers were an-
to any larger population; see Chapter 7), the find- alyzed with the intention of identifying common
ings were informative nonetheless. themes.

bel65796_ch04_077-099 92 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 93

The quantitative findings indicated that the body modifications to be appropriate for women
men expressing high self-confidence were less but not for men. By using the third person, the au-
likely to believe that their physical appearance thors suggest, the men were seeking to protect
made other people think they were lazy, and their male gender identity.
that men with higher levels of self-deprecation Following Bourdieu (1987), the authors saw
were more likely to believe that their appear- their results as illustrating the impact of the
ance decreased their self-confidence. The re- “field”—the totality of the individual’s social
sults also showed that the men with high scores networks as well as the society at large—on
on the self-deprecation measure were more likely self-evaluation, in this case on judgments about
to dislike and be embarrassed by their bodies, the acceptability of one’s body and physical ap-
while lower self-confidence was associated with pearance. They also suggest that their study
a higher willingness to have cosmetic surgery. In shows social and self acceptability to be closely
the open-ended responses, those who were in associated with the body, and the field as influ-
favour of cosmetic surgery often used the third encing the value and meaning given to certain
person to describe their feelings: for example, physical attributes. They also conclude that it is
“if surgery could boost their esteem they should a mistake to think that only women are highly
do it” (Ricciardelli & Clow, 2009, p. 123). The au- conscious of their bodies and feel judged on the
thors speculated that such men were distancing basis of their physical characteristics; men too
themselves from the notion of cosmetic surgery face negative consequences if they have the
because the culture around them deemed such “wrong” kind of physical features.

Unfortunately, replication tends to be consid- study of occupational prestige conducted in the

ered pedestrian and uninspiring, although that is Kitchener–Waterloo area of Ontario. The authors
beginning to change, at least in disciplines such as replicated research done 25 years earlier (Guppy
psychology (Ritchie, 2020). Until recently, it has not & Siltanen, 1977), taking great pains to ensure
been considered a high-status activity in either the that the methodologies of the two studies were
natural or social sciences, and standard replications as similar as possible. The replication not only
are the exception rather than the rule in most aca- provided a check for biases and routine errors;
demic journals. A further reason for the low inci- given that a quarter-century had passed since
dence of published replications in the social sciences the original study, it also provided a measure of
is the difficulty of ensuring that the conditions are how much attitudes had changed with regard to
precisely the same as those in the original study. If occupational prestige and gender. As reported
there is room for doubt on that score, any differ- in Chapter 2, the authors concluded that the
ences in findings may be attributed to the design earlier male advantage in occupational prestige
of the replication rather than to a deficiency in the had disappeared by the time the replication was
original study. Nonetheless, it is crucial that the done; in fact, women in people-oriented jobs en-
methods used to generate findings are made explicit, joyed higher prestige than men did in the more
so that replication is at least possible. Providing all recent study. If the authors had not taken care to
the information needed to conduct a replication is ensure accurate replication of the original study,
often regarded as an important quality of quantita- it would have been impossible to know whether
tive research. the differences in the findings represented actual
An example of the benefits of replication can social change or were simply a by-product of
be found in Goyder, Guppy, & Thompson’s (2003) using different methodologies.

bel65796_ch04_077-099 93 01/13/22 12:19 PM

94 PART II Quantitative Research

Critiques of quantitative • The measurement process produces an artifi-

cial and false sense of precision and accuracy.
research For example, quantitative research presumes
Every approach to research has its strengths and that different individuals responding to the
weaknesses. Over the years quantitative research, same question on a survey are interpreting
along with its epistemological and ontological foun- its key terms in the same way. Yet a question
dations, has been the focus of criticism, particularly such as “What is your social class?” can be in-
from exponents of qualitative research. The criticisms terpreted in at least two ways: as a reference to
pertain to quantitative research as a general research current wealth, or as a reference to ancestry
strategy, and to specific methods and research designs going back generations. Many methodologists
with which quantitative research is associated. find significant variations in respondents’ in-
terpretations of such terms. Researchers often
attempt to solve this problem by using ques-
Six criticisms of quantitative research
tions with fixed-choice answers—“Are you
To give a flavour of the critique of quantitative re- upper class, middle class, working class?” and
search, six criticisms are covered briefly here: so on—but this approach provides “a solution
to the problem of meaning by simply ignoring
• Quantitative researchers fail to distinguish it” (Cicourel, 1964, p. 108). Quantitative social
people and social institutions from “the researchers counter that there are ways to test
world of nature.” Some critics object to the for shared meanings, although they would
idea of treating the social world as if it were concede that this issue deserves more attention
no different from the natural order. They than it usually gets.
argue that because people interpret the • The reliance on instruments and procedures
world around them, their actions and expe- produces a disjuncture between research and
riences cannot be studied using the methods everyday life. This issue relates to the ques-
employed in the natural sciences. They claim tion of external validity. For example, many
that science is applicable only to entities and methods of quantitative research rely heavily
processes lacking this sort of self-reflection, on research instruments such as structured in-
such as chemical elements, photosynthesis, terviews and self-completion questionnaires.
and the circulation of blood. But many quan- As Cicourel (1982) pointed out, researchers
titative researchers maintain that humans tend to assume that survey respondents have
(and other animals) really are part of the both the knowledge they need to answer the
world of nature, and that it is simply wrong questions and a sense that the topic matters
to think that our existence cannot be use- in their everyday lives. Can researchers who
fully analyzed using science. For instance, pose a set of questions designed to measure
they would claim that consciousness itself attitudes toward federal–provincial relations,
is amenable to the scientific method, as are for example, be sure that respondents have suf-
emotions, decision-making processes, and ficient knowledge and interest in the topic to
so on. Similarly, they hold that quantitative produce meaningful responses? Quantitative
research is indispensable in understanding researchers would respond that the judicious
social problems. For example, to understand use of filter questions (see Chapter 5) can ad-
poverty it is important to know about things dress this problem. Similarly, there may be an
like mean and median incomes, as well as important difference between what people say
how to analyze various social categories that to researchers and what they actually do in
may be statistically associated with poverty, real life (see Box 4.6). Another criticism, intro-
such as gender, sexual identity, ethnicity, duced in Chapter 2, is directed at experimental
and so on. The debate continues. research that relies on controlling situations

bel65796_ch04_077-099 94 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 95

Box 4.6 Gap between stated and actual behaviour

A study of racial prejudice conducted many years survey research. On the other hand, the results
ago by LaPiere (1934) illustrates that there may be might have been just another example of the
a difference between what people say and what widespread difference between holding a preju-
they actually do. LaPiere spent two years travel- diced attitude and engaging in a discriminatory
ling with a young Chinese student and his wife, act. An experimental study of prejudice among
observing from a distance to see if they were re- college students (Frazer & Wiersma, 2001) showed
fused entry at hotels and restaurants. Of 66 hotels that in hypothetical situations, students would
approached, only one turned the couple away; of hire Black and White applicants of varying abil-
184 restaurants and diners, none refused entry. ities equally, but a week later recalled the Black
LaPiere then allowed six months to elapse applicants as less intelligent than the Whites even
before sending questionnaires to the hotels and though the two groups were presented as having
restaurants visited. One question asked: “Will you the same qualifications and offered identical an-
accept members of the Chinese race as guests in swers to job interview questions. In the real world,
your establishment?” Of the establishments that peer pressure can make an unprejudiced person
replied, 92 per cent of restaurants and 91 per cent discriminate, to “go along,” while a prejudiced
of hotels said no. LaPiere’s simple though striking person may not discriminate for fear of social con-
study clearly illustrated a gap between reports demnation or a lawsuit.
of behaviour and actual behaviour. The fact that The gap is usually worst when predictions
the question asked was somewhat unclear is not of future behaviour are involved. In 2004 the
usually noted in connection with this widely cited Canadian Blood Services reported that 28 per
study. “Will you …?” can be interpreted in two cent of Canadians said they intended to give
ways: as a question about the future or as a ques- blood in the next year, but only 3.7 per cent ac-
tion about current policy. Why the more obvious tually did. And in any election many people say
formulation “Do you …?” was not used is not clear, that they intend to vote—and even reveal their
though it is unlikely that this point had a significant preferences to pollsters—but then do not follow
bearing on the findings and their implications for through.

to determine causal connections. According of events to individuals is ignored and that the
to Brannigan (2004), such research can pro- connection to everyday contexts is missing.
duce only small, short-term manipulations of Quantitative researchers recognize those risks,
independent variables; yet much of everyday but claim that their research does not pre-
life is affected by long-term, ongoing social clude a search for how people interpret their
processes. everyday existence. For example, quantitative
• The analysis of relationships between variables sociologists sometimes ask respondents what
promotes a view of social life that is remote they think about social inequality, and try to
from everyday experience. Blumer (1956) link the responses with variables such as age,
argues that studies designed to bring out the gender, class, and ethnicity. Nonetheless, more
relationships between variables omit “the pro- thorough and research participant–centred in-
cess of interpretation or definition that goes on quiries into questions of meaning are usually
in human groups” (p. 685). This symbolic in- undertaken by qualitative sociologists.
teractionist assessment incorporates the first • Explanations for findings in quantitative stud-
and third criticisms above, that the meaning ies may be provided without examining the

bel65796_ch04_077-099 95 01/13/22 12:19 PM

96 PART II Quantitative Research

perceptions of the people to whom the find- based on the perspectives and life experiences
ings purportedly pertain. A related issue arises of the women who are its subjects. … [S]urveys,
when valid quantitative findings are achieved, though they have meticulously tabulated the
but the explanation for them is not informed trend, have led us to a dead end when it comes
by an empathetic understanding of the people to fully understanding the forces behind it”
involved. For example, it’s empirically true (Edin & Kefalas, 2005, p. 5).
that in many developed countries today much • Quantitative researchers tend to assume an ob-
larger proportions of children are born to jectivist ontology. As we saw in Chapter 1, quan-
unwed parents than was the case in previous titative researchers often assume that a reality
decades. Many researchers conclude from this exists that may be independent of the observer
that marriage is losing its popularity in certain or of individual consciousness. They may also
sectors of society, such as among the poor in see the social order as fixed or given—at least at a
inner-city areas. However, in-depth field re- particular point in history—rather than created
search conducted by Edin and Kefalas (2005) by individuals through negotiation, although the
suggests that among the poor single mothers latter view is not universal among quantitative
they studied in Philadelphia, marriage is in researchers. People who do quantitative work
fact highly valued, but is expected to happen maintain that it is not a mistake to assume that
later in life, after the children have been born. some things may exist and have certain social
“What is striking about the body of social sci- effects regardless of whether or how we perceive
ence evidence,” they write, “is how little of it is them. Again, the debate is ongoing.

Research in the News

COVID-19 deaths in Ontario long-term care homes related to their profit status
It is well known that residents of long-term care basis (e.g., by a charity) or those run by the local
homes in Canada have been devasted by COVID- municipality. The researchers also provided ev-
19. As with many medical issues, socio-political idence that older design standards, which were
factors can help to explain how the pandemic more common in the for-profit group, as well as
ran its course in Canada. chain ownership among that group largely ac-
In Ontario, the fate of the residents of these counted for these differences.
homes may have been affected by whether the According to Nathan Stall, the lead author of
home was privately owned and operated. The the study, care homes built according to older,
Toronto Star (Saba, 2020) has highlighted a study pre-1998 provincial standards were more likely
published in the Canadian Medical Association to have smaller rooms that were shared by up
Journal (Stall, Jones, Brown, Rochon, & Costa, to four residents, as well as shared washrooms
2020) that used quantitative data analysis tech- and crowded common areas, all of which are
niques that showed that both the extent of a risk factors for the spread of the virus. Stall also
COVID-19 outbreak in a long-term care home explained that privately owned homes tended to
as well as the number of resident deaths were offer lower levels of care due to lower staffing
related to the profit status of the home. Among levels (Saba, 2020). The authors of the study state
care homes with an outbreak of COVID-19, for- that their findings will help to inform debates
profit homes had more extensive virus outbreaks concerning how care homes should be financed,
and more deaths than homes run on a non-profit operated, and regulated.

bel65796_ch04_077-099 96 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 97

Key Points
• Figure 4.1 depicts quantitative research as a • Quantitative research has the following key
linear series of steps moving from theory to con- goals: measurement, the establishment of cau-
clusions, although this is an ideal from which sality, generalization, and replication.
there are many departures. • Quantitative research has been criticized by quali-
• Examining the reliability and validity of measures tative researchers. These criticisms tend to revolve
is important for assessing their quality. around rejecting the view that a natural science
model is appropriate for studying the social world.

Questions for Review (R) and Creative Application (A)

The main steps in quantitative research The main goals of quantitative
r1 What are the main steps in quantitative research? researchers
a1 Assume you are conducting a study to deter- r4 Outline the main goals of quantitative research-
mine which variables influence poverty. Name ers. What is to be gained by achieving them?
one variable that might have an impact on the a4 Assume you have to design a study that will
chances that a person will live in poverty, and examine why some people vote in national
use it to derive a specific hypothesis that can be elections while others do not. Name two inde-
tested with survey data. pendent variables that would be appropriate
to use in this study. What steps can be taken to
Concepts and their measurement ensure that your findings will be applicable to
r2 Why may multiple indicators of a concept be some larger population, i.e., to people who will
preferable to using a single indicator? not be participating in your study?
a2 Describe three different ways to measure the
concept “homophobia.” The critique of quantitative research
r5 Explain how the techniques used by quantitative
Reliability and validity researchers may produce findings that are not
r3 Name and define four different kinds of measure- applicable to everyday life.
ment validity, providing an illustration of each one. a5 You are using surveys to conduct research on al-
a3 How might a researcher measure the concept cohol consumption among stay-at-home moms.
“satisfaction with university student housing”? Might there be a gap between stated and actual
How could the test–retest method be used to es- behaviour on this topic? Why? What could you
tablish reliability? What drawbacks might there do to reduce or eliminate the gap?
be to the use of this method in the example you
have provided?

Interactive Classroom Activities

1. Divide the class into groups of seven or more. c. Explain how each operational definition has
Each group is to do the following: face validity.
a. Come up with a causal statement that in- d. Explain, giving concrete examples, how con-
volves two variables, for example, “Being a current, construct, and convergent validity
victim of racism affects one’s self-esteem.” could be established for both of the opera-
b. Provide a nominal and an operational defini- tionalizations proposed in point b.
tion for each variable (in this example, “racist e. Explain, giving concrete examples, how the op-
victimization” and “self-esteem”). erationalizations could be tested for reliability.

bel65796_ch04_077-099 97 01/13/22 12:19 PM

98 PART II Quantitative Research

f. Describe a methodologically sound way in iii. had trouble falling or staying asleep, or
which participants could be selected for the sleeping too much?
study. iv. felt tired or had little energy?
g. Describe the extent to which the study would v. had poor appetite or overate?
have external validity, and in particular the vi. felt bad about yourself, or that you are a
extent to which the findings could be gener- failure or have let yourself or your family
alized to people who were not studied. down?
vii. had trouble concentrating on things, such
Once these steps are done, each group presents
as reading the newspaper or watching
its work to the rest of the class. A different person
television?
in the group is to speak to each of the seven
viii. been moving or speaking so slowly that
points mentioned above. After speaking to each
other people could have noticed? Or the
point, group members should pause to allow the
opposite, being so fidgety or restless that
rest of the class to comment on what was just
you have been moving around a lot more
presented, with both strengths and weaknesses
than usual?
to be discussed.
ix. had thoughts that you would be better off
2. The instructor engages the class as a whole in a dead, or of hurting yourself?
discussion of measurement validity by: The fixed responses for these questions are: (1)
a. Asking for a nominal definition of the psy- Not at all; (2) Several days; (3) More than half the
chological concept of “depression.” Student days; (4) Nearly every day.
feedback is sought until a working definition The final question is:
of depression is achieved.
b. Assessing the measurement validity of a de- x. How difficult have these problems made it
pression scale used by Statistics Canada in the for you to do your work, take care of things at
Canadian Community Health Survey—2020 home, or get along with other people?
(available at [Link] Its response categories are: (1) Not difficult at
imdb/[Link]?Func tion= as sembleIn- all; (2) Somewhat difficult; (3) Very difficult; (4)
str&lang=en&Item_Id=1,262,397#qb1263385, Extremely difficult.
accessed 27 July 2020.) The scale is made up The measurement validity of each individual
of the 10 items (variables dep_Q005 through item is to be assessed by the class. The class is
dep_Q050) given below: asked to come up with at least one strength and
Over the last two weeks, how often have you one weakness for each item.

c. Asking the class whether this list of items

i. had little interest or pleasure in doing
covers all aspects (dimensions) of depression
things?
contained in the class’s nominal definition. If
ii. felt down, depressed, or hopeless?
it does not, the class is to construct additional
items that would rectify the problem.

Relevant Websites
The Statistics Canada site provides access to a such as the inflation rate, the level of unemploy-
wide variety of quantitative information about ment, and Canada’s population.
Canada. It also offers up-to-date figures on things [Link]/[Link]

bel65796_ch04_077-099 98 01/13/22 12:19 PM

 4 The Nature of Quantitative Research 99

The Canadian Research Institute for Social Policy The Web Center for Social Research Methods
(CRISP), at the University of New Brunswick, con- provides an illustration of how to create a Likert
ducts policy research on Canadian children and scale.
youth, fosters training in quantitative research meth- [Link]/kb/[Link]
ods, and helps economically developing nations en- (Websites accessed 4 September 2021.)
hance their research output on child development.
[Link]/research/institutes/crisp/

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch04_077-099 99 01/13/22 12:19 PM

 5
Survey Research:
Interviews and
Questionnaires
Chapter Overview
Survey research is one of the most commonly used data-gathering techniques in the social sciences. Surveys
can take the form of either interviews, in which participants are asked questions orally, or questionnaires,
which respondents read and fill out themselves. One of the strengths of survey research is that it allows for
standardization in the asking of questions and the categorization of the answers given.
This chapter explores:
• the reasons for the widespread use of surveys, including a consideration of the importance of stan-
dardization to the process of measurement;
• issues associated with asking questions in surveys:
» whether to use open or closed questions
» different kinds of questions that can be asked in structured interviews and questionnaires
» rules to bear in mind when designing questions
» optimal question order
» how questionnaires can be designed to minimize error and make answering easier for respondents
» projection or vignette questions in which respondents are asked to reflect on a hypothetical sce-
nario presented to them
» the importance of pre-testing questions
» the use of questions taken from previous research
• the different ways to do survey research, such as using more than one interviewer, conducting
­interviews by telephone, or using the Internet or email to administer questionnaires;

▲ JHVEPhoto/Shutterstock

bel65796_ch05_100-138 100 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 101

• various prerequisites of structured interviewing, including establishing rapport with the interviewee;
asking questions exactly as they appear on the interview schedule; keeping to the question order as it
appears on the schedule; and recording exactly what is said by interviewees;
• problems with survey research, including the influence of the interviewer on respondents and the
possibility of systematic bias in answers (known as response sets);
• the advantages and disadvantages of the questionnaire compared to the structured interview;
• researcher-driven diaries as a form of survey research; and
• secondary analysis of survey data.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:
1. Identify the advantages and disadvantages of open and closed questions.
2. Describe the types of questions used in survey research, and outline the functions of each one.
3. List the rules to be followed in designing survey questions, and explain the rationale for them.
4. Discuss key issues involved with structured interviews, including those pertaining to interview formats
and how researchers should conduct themselves.
5. Summarize the use of questionnaires in social research, including matters such as their pros and cons
compared to structured interviews, the administration of online questionnaires, how to design a ques-
tionnaire, and how researcher-driven diaries can take the form of a questionnaire.
6. Explain the role of secondary analyses of survey data in social research, and discuss their limitations.

“I never do surveys,” said Anna-Lise as she hung up they do the ‘On a scale from 1 to 5’ thing because
the phone. they want to compare responses between indi-
“Why?” I asked. viduals in a way that is meaningful, or they may
“For one thing, they always call at dinnertime, want to put a number on what people in gen-
when I’m tired and hungry.” eral think of the issue, by calculating a mean or
“Maybe that’s because most people are home something.”
at dinnertime,” I said. “They turn phone conversations into numbers?
“That doesn’t make it any more comfortable. Very weird.”
And they always ask questions in the weirdest ways, “Ever step on a scale to weigh yourself? Hey,
like: ‘On a scale from 1 to 5, with 1 being strongly there’s another kind of scale! Or how about the
dislike and 5 being strongly like, how do you feel numbers that appear on the gas pump when
about blah, blah, blah.’ Why can’t they just ask you you’re filling up, or numbers showing the time
your opinion on something and let you say what in the bottom right-hand corner of your com-
you want? That would be way more interesting.” puter screen? Same idea—turning things into
“Well, sometimes they give you a statement to numbers.”
respond to rather than a question. And I guess “I still say the whole thing is very weird.”

Introduction questionnaires, which have become familiar aspects

of everyday life. There are job interviews, media
A common method of acquiring data in ­quantitative interviews, and police interviews, and one is often
research involves the use of interviews and asked to complete a questionnaire before getting

bel65796_ch05_100-138 101 01/13/22 12:19 PM

102 PART II Quantitative Research

medical treatment or after using a service of some Disadvantages

kind. Research interviews and questionnaires share Open questions also present problems for a survey
some common features with the everyday variety, researcher:
although greater care is usually taken to elicit infor-
mation that is valid and reliable. The information • It’s more time-consuming to record the an-
gathered often pertains to matters such as behaviour, swers in an open interview.
attitudes, norms, beliefs, and values. • The answers have to be coded (see Chapter 4),
which can also take a lot of time and effort.
Open or closed questions? • Many prospective questionnaire respon-
dents may be put off by the need to com-
One of the most significant considerations in pose an answer, because of the bother and
survey research is whether to ask a question in an inconvenience.
“open” or “closed” format. With an open question, • Because of the difficulty of writing down ex-
respondents can reply however they wish. With a actly what respondents say in an interview,
closed question, they are presented with a set of there may be more inaccuracies in the record-
fixed alternatives from which they have to choose ing of answers. One obvious solution is to use
an appropriate answer (see Boxes 4.1 and 4.2 in a recording device such as a digital audio re-
the previous chapter for examples of closed ques- corder, but this can make some respondents
tions, including Likert scale items). What, then, nervous, and transcribing recorded answers
are some of the advantages and limitations of the takes a long time. (Issues with transcription
two formats? will be discussed in Chapter 11.)
• They run the risk of intra-interviewer variabil-
Open questions
ity, whereby an interviewer is not consistent in
Open questions present both advantages and disad- asking questions or recording answers, either
vantages to a survey researcher. By and large, how- with different respondents or with the same
ever, problems associated with the processing of respondent.
answers tend to limit their use, especially in quan- • They also face inter-interviewer variability,
titative research. which may occur when there is more than one
interviewer (the usual case). If interviewers are
Advantages
not consistent with each other in the ways they
Open questions, by allowing respondents to answer ask questions and/or record answers, measure-
in their own terms, have certain advantages over ment error can occur.
closed ones:
Needless to say, these sources of error are not mu-
• They allow for replies that the survey researcher tually exclusive; they can coexist, compounding the
might not have contemplated and therefore problem even further. Closed questions reduce the
might not have offered as fixed-choice options. incidence of these problems.
This makes them useful for exploring new or
changing areas. Closed questions
• They make it possible to tap the participants’
The advantages and disadvantages of closed ques-
unprompted knowledge and understanding of
tions are in many respects implied in the consider-
issues.
ations relating to open questions.
• The salience of particular issues for respon-
dents can also be examined.
Advantages
• They can generate fixed-choice format answers
(a point we will return to later in this chapter). Closed questions offer the following advantages:

bel65796_ch05_100-138 102 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 103

• They minimize intra-interviewer variability The aim is to keep the error portion to a mini-
and inter-interviewer variability. mum (see Figure 5.1), since error reduces the valid-
• Some respondents may not understand what ity of a measure. Standardization in interviews and
a question is getting at; the available answers questionnaires means that two sources of variation
may provide some clarification. due to error—items 2 and 5 in the preceding list—are
• Because interviewers and respondents are not likely to be less pronounced.
expected to write or speak extensively and Variation due to error can come from several
instead are to select an answer from a range other sources.
of possible responses, closed questions can
be answered quickly and easily, reducing re- Disadvantages
sponse rate issues. They are also are easier to Closed questions also present certain disadvantages,
process. however:

All of the above points illustrate how closed questions • Respondents’ answers may lack spontaneity
promote standardization in both the asking of ques- and authenticity. It is also possible that the an-
tions and the recording of answers. Standardization swers provided do not cover all the potential
means that variations in replies are due to “true” or replies. One solution is to use an open question
“real” variations in the characteristic being mea- to generate the categories (see Box 5.1). Another
sured rather than extraneous factors. Still, there is is to include “other” as a response option and
always a chance that some responses to closed ques- category and then invite respondents who
tions will be inaccurately measured, and there are a choose it to elaborate on their choice. Closed
number of reasons for this. Some prominent sources questions may irritate respondents who are
of error in survey research are: unable to find an answer category that they feel
applies to them. In theory, all possible answers
1. poorly worded questions; should be provided.
2. interviewer error in asking a question; • Respondents may differ in their interpreta-
3. misunderstanding on the part of the tion of the questions or the answers provided;
interviewee; for example, the understanding of the word
4. interviewee lapses in memory; “soon” in a question can vary widely from
5. interviewer error in recording information; person to person.
6. mistakes in entering the data into a computer • In interviews, a large number of closed ques-
file; and tions reduces conversation and gives the
7. biases related to the characteristics (such as interview an impersonal feeling, reducing
gender or “race”) of the interviewers and/or rapport.
the interviewees.

To take a simple illustration, a closed question on al-

cohol consumption among students will show that True variation
students vary in the amount of alcohol they con-
sume. Most measurement will contain an element of Variation due to error

error, so it is helpful to think of the measured varia-

tion in alcohol consumption as made up of two com-
ponents: true variation and error. In other words: True variation Variation due
to error
measured variation = true variation + FIGURE 5.1 A measure with little error and one
variation due to error with considerable error

bel65796_ch05_100-138 103 01/13/22 12:19 PM

104 PART II Quantitative Research

TABLE 5.1 Open questions and closed questions: Advantages and disadvantages
OPEN QUESTIONS
Advantages Disadvantages
» May produce responses researcher did not » More time consuming than closed questions.
anticipate.
» Can provide knowledge that is unprompted by the » Answers must be coded.
researcher.
» Offer an indication of the salience of the topic to the » Participants may be perturbed with having to
participant. compose an answer.
» Are a source of information for creating fixed-choice » Possible inaccuracies in recording answers.
questions.
» Risk of intra- and inter-interviewer variability in
recording answers.
CLOSED QUESTIONS
Advantages Disadvantages
» Answers do not have to be coded. » Answers may lack spontaneity and authenticity.
» Minimize intra- and inter-interviewer variability in » Answers provided may not cover all possible replies.
recording answers.
» Answers provided may clarify, for respondents, what » Respondents may differ in their interpretation of the
the question is asking. questions asked and the answers provided.
» Easier for respondents to provide answers to » Can recreate an impersonal atmosphere, reducing
questions, which may boost response rate. rapport.
» Easier to process responses.

Types of questions Asking people about what they witnessed at

a riot—how many people were there, the ap-
Structured interviews and questionnaires generally proximate ages of people engaged in looting—
contain several different types of question. There are is an example. This sort of question can lead to
various ways of classifying them, but here are the problems because in everyday life most people
main forms: are not careful, systematic observers, and they
may have biases that affect their perceptions
• Personal, factual questions. These are ques- and recollections.
tions that ask respondents to provide personal • Questions about attitudes. A five-point Likert
information, such as their age, occupation, scale is one of the most common formats for
marital status, and income. This also includes measuring attitudes. Often the responses to a
questions about their behaviour, such as the number of Likert items are aggregated to form
frequency of their attendance at religious ser- a composite measure of an attitude.
vices or how often they go out to a movie. • Questions about beliefs. For example, in Hay’s
• Factual questions about an entity or event. (2014) study of secularization in Canada (see
Sometimes respondents will act as informants Box 1.3), one item used to measure moral au-
about some phenomenon or event with which tonomy pertained to the belief that “it is up
they are familiar, particularly when informa- to the individual to decide what is right and
tion is not readily available from other sources. wrong.” The response categories were: “Agree

bel65796_ch05_100-138 104 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 105

Box 5.1 A comparison of results for closed and open questions

Schuman and Presser (1981) conducted an ex- then revised the closed categories to reflect the
periment to determine whether responses to answers they had received to the open-ended
closed questions could be improved by asking question, and re-administered the questions to
the questions first in open-ended format and two large samples. The questions were: “This
then developing categories of reply from re- next question is on the subject of work. People
spondents’ answers. They asked a question look for different things in a job. Which one of the
in both open and closed formats about what following five things do you most prefer in a job?
people look for in a job, using different sam- [closed question] What would you most prefer
ples for each. They found considerable dis- in a job? [open question]” The answers they re-
parities between the two sets of answers. They ceived were as follows:

CLOSED FORMAT OPEN FORMAT

Answer % Answer %
Work that pays well 13.2 Pay 16.7
Work that gives a feeling of 31.0 Feeling of accomplishment 14.5
accomplishment
Work where there is not too much 11.7 Control of work 4.6
supervision and you make most
decisions yourself
Work that is pleasant and people who 19.8 Pleasant work 14.5
are nice to work with
Work that is steady with little chance of 20.3 Security 7.6
being laid off
96% 57.9%
of sample of sample
Opportunity for promotion 1.0
Short hours/lots of free time 1.6
Working conditions 3.1
Benefits 2.3
Satisfaction/liking a job 15.6
Other/dk/na 4.0 Other responses 18.3
100% 100%

With this revised form of the closed question, to the open question matched the categories
Schuman and Presser found that a much higher offered in the closed one. Also, the distributions
proportion of the answers to the open question were somewhat different: for example, twice as
corresponded to the closed categories. They many respondents cited a feeling of accomplish-
argued that the new closed question was supe- ment with the closed format as with the open one.
rior to its predecessor and also superior to the Nonetheless, the experiment demonstrated the
open question. However, it was still disconcerting desirability of generating forced-choice answers
to find that only 58 per cent of the answers given from open questions.

bel65796_ch05_100-138 105 01/13/22 12:19 PM

106 PART II Quantitative Research

strongly” (5), “Agree somewhat” (4), “Neither which would be a waste of your time as well as that
agree nor disagree” (3), “Disagree somewhat” of your respondents.
(2), and “Disagree strongly” (1).
• Questions about knowledge. Sometimes ques- What exactly do you want to know?
tions are designed to “test” respondents’ Rule of thumb number two relates to the first one.
knowledge in a certain area. For example, a It’s important to focus on exactly what you want to
poll sponsored by Historica Canada in 2020 know. Consider this seemingly harmless question:
estimated that almost half of Canadians knew
Do you have a car?
that Canadian scientists contributed to the de-
velopment of the polio vaccine, but only about What is this question seeking to tap? Car ownership?
29% were familiar with Viola Desmond and Access to a household car? Use of a company car?
her achievements (she appears on Canada’s $10 The ambiguity of the word “have” means that a “yes”
bill) (Historica Canada, 2020). answer could signify any of the three. If you want to
• Questions about others. It’s usually better to find out whether a respondent owns a car, you need
interview the person directly than to rely on to ask specifically about ownership. Be as specific as
the second-hand accounts of others, because possible.
respondents’ knowledge of others’ behaviour
may be distorted or inaccurate (Beardsworth How would you answer it?
& Keil, 1997). However, if the researcher is in- Rule of thumb number three is to put yourself in the
terested in the respondent’s own perceptions position of the respondent. Ask yourself the question
of how someone else acts or feels, questions and try to work out a reply. If you do this, there is
pertaining to such matters are appropriate. at least a chance that the ambiguities inherent in a
Responses to this sort of question can be used question like “Do you have a car?” will become evi-
as a check on the validity of self-reports, or may dent. For example, you may remember that you have
be relied on if the person in question is unable access to a car that you don’t actually own. Putting
or unwilling to describe their own thoughts or yourself in the position of the respondent can reveal
behaviours. the difficulty of answering poorly constructed
questions.

Specific rules when designing

Rules for designing questions questions
Over the years, numerous rules have been devised in Avoid ambiguous terms in questions
connection with the dos and don’ts of asking ques- If possible, avoid terms such as “often” and “regu-
tions. Yet mistakes persist. So here are three simple larly” as measures of frequency. They are ambiguous,
rules of thumb, followed by some more specific and different respondents will take them to mean
guidelines. different things. It may not always be possible to get
a specific number from respondents, but that should
General rules of thumb be your goal. Consider a question asked of people
Keep the research questions in mind who have said that they own a cellphone:
The questions you ask in a questionnaire or struc-
How often do you use your cellphone to place
tured interview should be designed specifically to
a call?
answer your research questions. Focusing on the
research questions will help protect you against the Very often _____
pain of realizing too late that a particular research Quite often _____
question was not addressed in your study. It will also Not very often _____
reduce the risk of pursuing issues that are irrelevant, Not at all _____

bel65796_ch05_100-138 106 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 107

The problem with this question is that, with the The problem here is obvious: people may be satis-
exception of “not at all,” all the terms in the response fied with one but not the other. Not only will the
categories are ambiguous. Instead, ask about actual respondent be unclear about how to reply, but the
frequency, such as: answer provided may pertain to pay, to conditions,
or to both.
How often do you use your cellphone to place The same rule applies to fixed-choice answers.
a call? In Box 5.1, one of Schuman and Presser’s (1981) an-
(Please tick whichever category comes closest to swers is:
the number of times you use your cellphone.) Work that is pleasant and people who are nice
More than 10 times per day _____ to work with.
5–9 times per day _____ While there may well be symmetry between the
1–4 times per day _____ two ideas in this answer—pleasant work and nice
A few times a week _____ people—there is no necessary correspondence be-
A few times a month _____ tween them. Pleasant work may be important for
Almost never _____ people who are indifferent to their co-workers. For
Alternatively, you could simply ask respondents to examples of triple- and quadruple-barrelled ques-
estimate the number of times per day they use their tions, see Box 5.2.
cellphone to make a call.
Avoid very general questions
It’s also important to bear in mind that certain
common words, such as “dinner,” mean different It may be tempting to ask a very general question
things to different people. For some, dinner is a light when what is needed is a response to a specific issue.
midday meal, whereas for others it is a substantial A question such as
evening meal. In such cases it is necessary to clearly How satisfied are you with your job?
define the terms that are being used.
seems harmless, but it lacks specificity. Does it
Avoid long questions refer to pay, working conditions, the nature of
Most methodologists agree that long questions the work, or all of these? Respondents are likely
are undesirable. Interviewees can lose track of to vary in their interpretation of the question,
long questions, and respondents completing a and this will be a source of error. A favourite gen-
questionnaire may be tempted to skim or even eral question comes from Karl Marx’s Enquête
skip over them. However, Sudman and Bradburn ouvrière, a questionnaire sent to 25,000 French
(1982) suggested that this advice is more appli- socialists and others (whether any were returned
cable to attitude questions than to those asking is unknown). The final question (the one-hun-
about behaviour. They argued that when the dredth!) reads:
focus is on behaviour, longer questions can have What is the general, physical, intellectual,
certain positive features in interviews: for exam- and moral condition of men and women em-
ple, they are more likely to provide memory cues ployed in your trade? (Bottomore & Rubel,
and facilitate recall because of the time it takes to 1963, p. 218)
ask them. By and large, however, it’s best to keep
questions short. If you find it difficult to come up with precise ques-
tions to replace very broad, general ones, you might
Avoid double-barrelled questions take inspiration from Blauner (1964), who provides
a very detailed and nuanced description of job sat-
How satisfied are you with your pay and work- isfaction among American factory workers based on
ing conditions? some 80 specific questions.

bel65796_ch05_100-138 107 01/13/22 12:19 PM

108 PART II Quantitative Research

Box 5.2 Matching questions and answers in closed questions

A publisher inserted a feedback questionnaire
Elegance:
into copies of a novel, including a series of Likert-
Very Poor Poor Average Good Excellent
style items regarding the book’s quality. In each
Seamlessness:
case, readers were asked to indicate whether
Very Poor Poor Average Good Excellent
a particular attribute was “poor,” “acceptable,”
“average,” “good,” or “excellent.” However, the Imaginativeness:
items were presented as questions that, strictly Very Poor Poor Average Good Excellent
speaking, demanded “yes” or “no” answers. For
example: This approach would also solve the additional
problem of the “triple-barrelled” question that
Was the writing elegant, seamless, asks about three distinct aspects of the writing,
imaginative? leaving the reader unable to express differing
None of the response options provided by the opinions of each.
publisher could be considered logical answers It could be argued that readers would under-
to the questions that were posed. The questions stand that they were being asked to rate the book
would have been better presented as state- in terms of each attribute. The problem is that the
ments, for example: impact of a disjunction between question and
answer is simply not known—which means that
Please indicate the quality of the writing in any conclusions drawn by the publisher might
each of the following categories: well be wrong.

Avoid leading questions using response categories on a seven-point scale

Leading or loaded questions push respondents in a “where 1= very poor and 7 = excellent.” Reviewers
particular direction; for example: providing evaluations for another prominent journal
are asked, “Does the paper address novel questions
Would you agree to cutting taxes, even though and provides [sic] novel insight? Does it explore im-
welfare payments for the neediest sections of portant but overlooked phenomena, a creative ap-
the population might be reduced? proach to a topic, new or seldom used designs and
Phrasing the question in this way is likely to make it methods, or understudied samples?” Reviewers are
difficult for some people to agree, even if they do feel to give a single response to this set of questions, using
that taxes are too high. the categories “inadequate,” “marginal,” “good,” and
Mismatched questions and answers, and request- “not applicable.”
ing a single response for a question that covers more
Avoid questions that actually ask two
than one issue, are not limited to publishers’ inserts
(or more) questions
in novels. They also appear in questionnaires that
academic reviewers are asked to complete as part of Some of the questions cited above clearly transgress
their evaluation of manuscripts that have been sub- this rule, but there are also more subtle versions of
mitted for publication in leading social science jour- the problem. A question such as
nals. Reviewers for one journal are asked to rate the Which party did you vote for in the 2019 fed-
“Importance of topic for public opinion scholars,” eral election?

bel65796_ch05_100-138 108 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 109

provides an illustration. What if the respondent did Ensure that respondents have the requisite
not vote in that election? The question is really two knowledge
questions, which should be asked separately: There is little point in asking about matters unfamil-
Did you vote in the 2019 federal election? iar to respondents. For example, it’s doubtful that
meaningful data regarding opinions on foreign af-
Yes fairs can be gained from people who do not follow
international political issues.
No
Ensure that the answers provided for a
closed question are balanced
If YES, for which political party did you vote?
It’s important that the range of answers provided for
a closed question be properly balanced between pos-
Avoid questions that include negatives itive and negative options. A set of responses like this
The problem with questions that include negative
Excellent _____
terms such as “not” is that some respondents may
Good _____
miss the negative and give answers opposite to what
Borderline _____
they intended. There may be situations when it’s im-
Poor _____
possible to avoid using a negative, but questions like
the following should be avoided as far as possible: is weighted toward a favourable response. “Excellent”
and “Good” are both positive; “Borderline” is a neu-
Do you agree with the view that university
tral or middle position; and “Poor” is a negative re-
students should not have to take out loans to
sponse. A second negative response choice, such as
finance their education?
“Very poor” is needed.
Instead, the question should be asked in a positive
form; for example: “Do you find it acceptable that Ensure answers provided for a closed
some university students have to take out loans to fi- question do not overlap
nance their education?” Questions with double neg- Care must be taken to ensure that answers to closed
atives are never appropriate, because it’s impossible questions are mutually exclusive. For example, a
to be sure that respondents have interpreted them question on the respondent’s age with a set of re-
correctly. The following provides an example: sponses like this
Would you rather not drink non-alcoholic 20–30 _____
beer? 30–40 _____
40–50 _____
Minimize technical terms 50–60 _____
Use simple, plain language and avoid jargon. Don’t would confuse anyone with an age of 30, 40, or 50,
ask a question like this: as these answers each fit into two of the supplied
options.
Do you sometimes feel alienated from your
work? Don’t overstretch people’s memories
The problem here is that many respondents may It might be useful to know exactly how many
not be familiar with the term “alienated,” and even times respondents have used their cellphones in
if they are, they may not understand it in the same the previous month, but few people will be able to
way. The same goes for acronyms and abbreviations: give you an accurate answer for such an extended
avoid them whenever possible. period of time. For this reason, it’s important to

bel65796_ch05_100-138 109 01/13/22 12:19 PM

110 PART II Quantitative Research

choose a suitable time frame, such as per day in job?” twice: once early on, and once again later, in
the case of cellphone use. the context of questions on contact with the police.
Almost a quarter (22 per cent) of those respondents
Carefully consider using “Don’t know” gave a more positive rating the second time. Mayhew
as a closed answer option suggests that as the interview wore on, respondents
One area of controversy regarding closed questions became more sensitized to crime-related issues and
is whether the answer set should include a “don’t more sympathetic to the pressures on the police. As
know” or “no opinion” option when asking ques- appealing as this explanation sounds, it cannot ac-
tions concerning attitudes. The main argument for count for the 13 per cent who gave a lower rating the
including a “don’t know” option is to ensure that second time around. That is quite a reliability issue.
respondents will not be forced to express views that (See also Box 5.3.)
they don’t actually hold. Many recommend offer- It’s hard to draw any general lessons from ques-
ing the “don’t know” option in the form of a filter tion-order research, at least in part because exper-
question to remove those respondents who have no iments do not always reveal clear-cut effects, even
opinion on the topic. Filter questions are used to in cases where effects might legitimately have been
determine whether it is appropriate to ask certain expected. Nonetheless, two general lessons have
questions of a particular respondent. The inter- emerged:
viewer could ask the filter question, “Do you have
an opinion on this issue?” of everyone and then ask • Unless the study is designed to test question-
the second question only of those who say that they order effects, all respondents should receive
do have an opinion. questions in the same order.
The alternative argument in connection with the • Researchers should be sensitive to the possible
“don’t know” option is that it offers an easy way out effects that a question could have on subse-
for respondents. A series of experiments conducted quent questions.
in the US found that many respondents who claim
to have no opinion on a topic do in fact hold one In addition, the following rules about question order
(Krosnick et al., 2002). Respondents with lower levels are sometimes proposed:
of education are especially likely to select the “don’t
know” option; and later questions in a questionnaire • Early questions should be directly related to
(see the next section for question-order effects) are the announced research topic. This removes
the most likely to elicit a “don’t know” response. The the risk that respondents will be put off by ap-
researchers concluded that data quality is not en- parently irrelevant questions asked at an early
hanced by the inclusion of a “don’t know” option and stage in the interview. It also means that per-
that it may prevent some respondents from express- sonal questions about age, social background,
ing an opinion that they do actually hold. and so on should not be asked at the beginning
of an interview.
Pay close attention to question order • Questions likely to be important or meaning-
Quite a lot of research has been directed at find- ful to respondents should be asked early in the
ing out how responses are affected by asking ques- interview, to stimulate their interest and at-
tions at different points in an interview schedule. tention. This suggestion may conflict with the
Results have varied and few if any consistent effects previous one if questions about the research
have been found. Mayhew (2000) tells an interest- topic are not of interest to respondents.
ing anecdote on question order in a crime survey. • If possible, questions that may cause embar-
By mistake, half of the respondents were asked the rassment or anxiety should be left for later in
question “Taking everything into account, would the schedule—although not to the very end.
you say the police in this area do a good job or a poor Respondents should not leave the interview (or

bel65796_ch05_100-138 110 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 111

complete the questionnaire) with misgivings • It is sometimes recommended that questions

or negative feelings about the research. dealing with opinions and attitudes precede
• With a long schedule or questionnaire, ques- questions about behaviour and knowledge.
tions should be grouped logically in related This is because questions that tap opinions and
sections, and should not jump back and forth attitudes may be affected by the latter sorts of
from one topic to another. questions. For instance, if a husband reports
• Within each group of questions, general ques- that he does only 20 per cent of the house-
tions should precede specific ones. If a specific work, making that statement can affect how he
question comes first, the aspect of the general responds when asked if housework should be
question that it highlights may be discounted shared equally between spouses.
by respondents who feel they have already cov- • If a respondent provides an answer to a ques-
ered it. For example, if a question asking how tion that is to be asked later in the interview,
people feel about their salary precedes a gen- the question should still be asked later on,
eral question about job satisfaction, some re- when the interviewer arrives at the scheduled
spondents may discount the issue of pay when point. A different answer may be offered at
responding to the job satisfaction question, the latter juncture because of question-order
since they have already addressed it. effects.

Box 5.3 About question order

Imagine that there are two pollsters: one who Set Two:
favours the federal New Democratic Party and 1. The Canada–United States–Mexico free trade
another who favours the Conservatives. The key deal will help create jobs in Canada.
question in both their surveys is: “If an election were 2. Canada’s mounting debt is mortgaging the
held today, for which party would you vote?” That future of its children.
question would be followed by an alphabetical list 3. Lower taxes would benefit everyone, espe-
cially hard-working Canadian families.
of the main political parties. Look at the two sets
4. Our laws should protect law-abiding citizens,
of Likert items that follow (assume there are five
not criminals.
response options ranging from “Strongly agree”
to “Strongly disagree”), and imagine that the items For those who do not follow politics closely, the
immediately preceded the voting question. Try to first set would probably increase the number
determine whether these items would influence re- of respondents saying they would vote NDP,
sponses to the voting question and, if so, how. while the second would likely do the same for
the number declaring their intention to vote
Set One: Conservative. Although most pollsters avoid
1. Canada must do more to reduce global such blatant biases, sometimes more subtle in-
warming. fluences on respondents go unnoticed. In any
2. The environment should be the federal gov- event, this is an interesting exercise because it
ernment’s top priority. illustrates the importance of question order in
3. Corporate taxes should be increased to pay survey research: if it is necessary to use these
for social assistance for Canada’s poorest
items, they ought to be presented after the
families.
voting question.
4. Canada must do more to protect Indigenous
rights.

bel65796_ch05_100-138 111 01/13/22 12:19 PM

112 PART II Quantitative Research

Avoid provoking response sets read each question. Similar procedures may be used
This issue (see also Chapter 4) is especially relevant to avoid the effects of laziness or boredom. For exam-
to multiple-indicator measures, where respondents ple, when visiting a doctor’s office for the first time,
reply to a battery of related questions or items. A re- patients usually have to fill out a form containing a
sponse set exists if responses are motivated by some- long list of diseases, indicating whether they have
thing other than the person’s actual feeling about the ever suffered from them. Someone may begin care-
items. They can result from psychological predispo- fully, reading the list and checking off “No, No, No,
sitions like acquiescence (a desire to be “cooperative” …,” but eventually tire of the task and just tick “No”
or to please the researcher) or laziness, and in such for the remaining items without reading them closely.
cases are evident when someone responds to a series Those who persist in choosing the same answer re-
of items in the same way, regardless of the content of gardless of the question may have to be removed from
the items. To avoid acquiescence, researchers using the study. Response sets may also stem from a desire
multiple-item measures may include some items that to be liked or respected by the interviewer, so in-
logically call for opposite positions, so that if respon- structing interviewers to avoid both becoming overly
dents are consistent in their attitudes, they can nei- friendly with respondents and appearing judgmental
ther agree with all the items nor disagree with all of about their replies can help. Using vignettes can also
them. This also requires the respondent to carefully help reduce answers motivated by social desirability.

Practical Tip | Common mistakes when asking questions

Over the years, the authors of this book have Attitudes are complex, and most respon-
read many projects and dissertations based dents will not be simply “satisfied” or “not satis-
on structured interviews and questionnaires. fied.” For one thing, people’s feelings about such
A number of mistakes recur regularly, among things vary in intensity. An improvement would
them the following: be to rephrase the item as:

• Excessive use of open questions. While re- How satisfied are you with opportunities
sistance to closed questions is understand- for promotion in your firm?
able, open questions are likely to reduce the
Very satisfied ____
response rate and cause analysis problems.
Satisfied ____
Keep them to a minimum.
Neither satisfied nor dissatisfied ____
• Excessive use of yes/no questions. Sometimes
Dissatisfied ____
students include lots of questions that call for
Very dissatisfied ____
a yes/no response (usually a sign of inade-
quate thinking and preparation). The world This sort of format also makes it possible to cal-
rarely fits into this kind of response. Take a culate some widely used statistics that are dis-
question like: cussed in Chapter 8.

Are you satisfied with the opportunities for • Too many questions that allow respondents
promotion in your firm? to choose more than one answer. Although
there are times when such questions are un-
Yes __________ No __________
avoidable, the replies they produce are often
difficult to analyze.

bel65796_ch05_100-138 112 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 113

Consider using vignette questions Box 5.4 outlines a vignette designed to tease out
when appropriate respondents’ norms concerning several aspects of
Closed questions that are used to examine eth- family obligations, including the nature of the as-
ical standards and beliefs may be employed in sistance required (direct involvement or simple pro-
conjunction with the vignette technique. It in- vision of resources); geographical considerations;
volves presenting respondents with one or more the choice between paid work and unpaid care; and
scenarios and asking them how they would re- among heterosexual couples, the gender question
spond if confronted with the circumstances de- (should it be the man or the woman who gives up
picted in the scenario. For example, Kingsbury paid work for unpaid care?). The specificity of the
and Coplan (2012) used vignettes to examine how situation facing Jim and Margaret increases as the
some Ontario mothers of preschool children re- vignette develops. The first question (a) does not
acted to hypothetical accounts of their child’s say whether they are prepared to move; the second
shyness and aggression. The researchers hypoth- (b) says that they are; and in the last question (d) they
esized that mothers would look more favourably have in fact moved and are facing a new dilemma.
on gender-congruent behaviours (such as shyness Many aspects of the issues tapped by the ques-
in girls) than gender-incongruent ones (physical tions in Box 5.4 can be accessed through attitude
aggression in girls), especially if they held more items. For example:
traditional attitudes toward sex roles. The find- When two heterosexual working spouses
ings provided some support for the hypotheses, decide that one of them should quit work to
although the results were mixed.

Box 5.4 A vignette to establish family obligations

Jim and Margaret Robinson are a married right now could have a bad effect on their
couple in their early forties. Jim’s parents, who children’s education.
live several hundred kilometres away, have What should Jim and Margaret do? Should
had a serious car accident and they need long- they move or should they stay?
term daily help. Jim is their only son. He and his • move
wife both could get transfers to work nearer his • stay
parents.
c. Why do you think they should move/stay?
a. What should Jim and Margaret do? d. Jim and Margaret do go to live near Jim’s par-
• move to live near Jim’s parents ents. A year later Jim’s mother dies and his fa-
• have Jim’s parents move to live with them ther’s condition worsens to the point where he
• give Jim’s parents money to help them pay needs full-time care.
for daily care Should one of either Jim or Margaret take an
• let Jim’s parents make their own extended leave from work to take care of Jim’s
arrange­ments father? IF YES, who should: Jim or Margaret?
• do something else (specify) • Yes, Jim should give up his job.
• don’t know • Yes, Margaret should give up her job.
b. In fact, Jim and Margaret are prepared to • No, neither should give up their job.
move and live near Jim’s parents, but teach- • don’t know/depends
ers at their children’s school say that moving Source: Adapted from Finch (1987, p. 108).

bel65796_ch05_100-138 113 01/13/22 12:19 PM

114 PART II Quantitative Research

care for ailing parents, the wife should be the of choices outlined in the scenarios. Often what people
one to give up her job. say they would do in a particular situation turns out
Strongly agree _____ to be very different from what they actually do. People
Agree _____ are not necessarily being dishonest—predicting be-
Undecided _____ haviour, whether our own or someone else’s, is often
Disagree _____ more difficult than we would imagine.
Strongly disagree _____
Run a pilot study
The advantage of the vignette over this kind of at-
It is always desirable to conduct a pilot study
titude item is that it anchors the choice in a more
before collecting data from respondents: not just
realistic situation and thus reduces the risk of an
to ensure that individual questions operate well
unreflective reply. Finch (1987) also argued that on
but also to ensure that the research instrument as
a sensitive topic like this, some respondents may
a whole is appropriate. Pilot studies may be par-
feel uncomfortable or fear that they will be judged
ticularly crucial in research using questionnaires,
on their replies. The fact that the vignette is about
since there will be no interviewer present to clear
other people (and imaginary ones at that) creates a
up any confusion.
certain distance between the questioning and the re-
Here are some specific uses of pilot studies in
spondent, and hence the chance for a more candid
survey research:
answer. One obvious requirement of the vignette
technique is that scenarios must be believable. • If the study is going to employ mainly closed
Finch also pointed to some limitations with this questions, a researcher can pose open ques-
style of questioning, however. It may be impossible to tions in pilot qualitative interviews and then
establish what assumptions are being made by the re- use them to generate the fixed-choice answers.
spondents about the characters in the scenario (such • Piloting an interview schedule can give inter-
as their age, ethnicity, and number of children at viewers some experience in using it and help
home) and the significance of those assumptions for them develop confidence.
the validity and comparability of people’s replies. It is • If everyone (or virtually everyone) who an-
also difficult to establish how far respondents’ answers swers a question gives the same answer, the re-
reflect their own normative views or indeed how they sulting data are not likely to be useful. A pilot
themselves would act when confronted with the kinds study gives you an opportunity to identify

Practical Tip | Getting help in designing questions

When designing questions, be empathetic: put Then try out the questions on some friends
yourself in the position of the people who will or classmates, as in a pilot study. Ask them
be answering them. This can be difficult, be- to be critical and to consider how well the
cause the questions may not always apply to questions connect to each other. Also, look
the person producing them—for example, to a at some questionnaires and structured in-
young student doing a survey of retired people. terview schedules designed by experienced
Even so, you need to think about how you would researchers. Those researchers may not have
reply. This means concentrating not just on the asked questions on your particular topic,
questions themselves but also on the links be- but the way they have asked their questions
tween the questions. For example, do filter ques- should give you an idea of what to do and
tions work in the expected way? what to avoid.

bel65796_ch05_100-138 114 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 115

such questions and modify them as necessary cause, missing data are undesirable and a
to elicit more varied answers. pilot study may be very helpful in identifying
• In survey interviews, it may be possible to problems.
identify questions that make respondents feel • A pilot study may also offer an opportunity to
uncomfortable and to detect any junctures consider the overall structure of the study. If
where interest tends to flag. the questions do not flow well, it may be neces-
• Questions that respondents tend to skip or sary to move some of them around.
have difficulty understanding (more easily
recognizable in an interview than in a ques- The pilot should not be carried out on people
tionnaire context) should become apparent. who may become members of the sample in the
If certain questions are frequently skipped, full study, since participation in the pilot could
it may be because of unclear or threatening affect their responses in the study proper. If the
phrasing, poorly worded instructions, or con- population you are investigating is small, you
fusing positioning in the interview schedule don’t want to be forced to exclude certain people,
or questionnaire (see Box 5.5). Whatever the since that could affect the representativeness of

Box 5.5 A bad questionnaire

How would you fix this questionnaire? 7. If one were planned, would you fight having
“Hello. I am taking a sociology course on re- an AIDS hospice on your street?
search methods and would like to ask you some Yes ______ No ______
questions. Would that be ok?” 8. How does your spouse feel about AIDS
1. Were you ever scared that you were victims?
HIV-positive? Compassionate ______ Sympathetic
Yes ______ No ______ ______
Concerned ______  Angry ______  Don’t

For questions 2 to 6, please answer Agree or
know ______
Disagree.
9. Elizabeth Taylor has raised much money for
2. As a religious person, I feel sorry for AIDS
AIDS research. Should people be doing more
victims.
fundraising in this area?
3. I think victims of any disease deserve
Yes ______ No ______
compassion.
4. I think the new legislation will be a boon to 10. Do you know any AIDS victims?
AIDS victims. Yes ______ No ______
5. The government should not allocate more
money to AIDS research. Finally, I need a few facts for comparison
6. People with AIDS and other sexual diseases purposes.
should be quarantined. 11. How old are you? ______
If you agree, has your viewpoint changed since 12. Did you graduate from high school?
the 1980s when AIDS first widely emerged? Yes ______ No ______
13. Did you ever have yourself tested for AIDS?
Yes ______ No ______ Yes ______ No ______
Now there are just a few more questions. Thank you for your cooperation.

bel65796_ch05_100-138 115 01/13/22 12:19 PM

116 PART II Quantitative Research

the sample. If this is an issue, do the pilot study Issues related to conducting
on respondents from a different but similar pop-
ulation. Beagan (2001) did this in her study of
interviews
medical students. In addition to crafting good questions, there are
other issues to be considered when doing research
Consider using existing questions
using structured interviews and questionnaires.
Finally, it is sometimes possible to use questions These are addressed below.
that have been employed by other researchers
for at least part of the questionnaire or interview Interview formats
schedule. This may seem like stealing, but it is a In a traditional interview, an interviewer stands
legitimate research practice so long as the original or sits in front of the respondent, asks a series of
source is cited. One advantage of using existing questions, and writes down or keys in the answers.
questions is that they may have already been pi- However, there are several possible departures from
loted and tested for their reliability and validity. this pattern.
Robinson, Shaver, & Wrightsman (1999), for ex-
ample, provide an extensive set of survey ques- More than one interviewer or interviewee
tions that have been used to measure political Multiple interviewers are rare in social research be-
attitudes, along with measures of their reliability cause of the considerable cost involved in dispatching
and validity. more than one person to conduct an interview. In the
A further advantage is that existing questions case of focus groups it is more common, but in that
allow for comparisons with other research. For instance there is more than one interviewee as well,
instance, in psychology it is quite common to and to administer a structured interview to a focus
use standardized personality and intelligence group would be very unusual. In most survey re-
tests; these permit comparisons between differ- search a specific individual is the object of question-
ent samples and allow different researchers to ing by one interviewer. Indeed, it is usually advisable
make consistent measurements. Similarly, some to discourage as far as possible the presence or intru-
of the questions used in Bell, Aitken Schermer, & sion of others during the interview. Studies in which
Vernon’s (2009) study of the heritability of polit- more than one person is being interviewed tend to be
ical orientations were taken from earlier studies. conducted by qualitative researchers, although that is
They were either used verbatim or slightly revised not always the case. For example, Pahl’s (1990) study
to make them suitable for the sample to which of patterns in the control of money among couples
they were administered. Walklate (2000, p. 194) employed structured interviewing first of the couples
described how in developing a survey instrument and then of husbands and wives separately.
for victims of crime, she and her colleagues used
“tried and tested questions taken from pre-exist- Alternatives to in-person interviews
ing criminal victimization surveys amended to It is customary in academic social research not in-
take account of our own more localized concerns.” volving national samples to use face-to-face inter-
Existing questions can also be used to investigate views. However, telephone interviewing is the norm
whether change has occurred over time by com- in fields such as market or government research.
paring current results with those arrived at in Telephone interviews have several advantages over
earlier time periods. At the very least, examining the face-to-face variety:
questions used by others may provide ideas about
how best to approach your own questions, even • They are far cheaper and quicker to adminis-
if you eventually decide to modify those used by ter than in-person interviews, which may re-
other researchers. quire the interviewers to spend a great deal of

bel65796_ch05_100-138 116 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 117

time travelling to meet with respondents; this ex-directory households, though it’s important
is a particular problem when a sample is geo- to note that in some areas it is prohibited by
graphically dispersed. law to contact cellphones in this way.
• Telephone interviews are easier to supervise • Respondents with hearing impairments are
than the personal kind. This is a particular ad- likely to find telephone interviews more diffi-
vantage when several interviewers are calling cult than personal interviews.
from the same location at the same time, since • A telephone interview is unlikely to be sustain-
supervisors may be able to detect interviewer able beyond 20 to 25 minutes, whereas per-
errors, such as rephrasing questions or probing sonal interviews can be conducted for longer
inappropriately. Face-to-face interviews can periods of time (Frey, 2004).
be recorded for the purpose of assessing data • There is some evidence that telephone inter-
quality, but the recording of interviews raises views fare less well than personal interviews in
issues of confidentiality and so must be treated obtaining data on sensitive issues such as drug
with caution. and alcohol use, income, tax payments, and
• Telephone interviewing can also reduce bias health. The evidence is not entirely consistent
arising from the characteristics of the inter- on this point, but when many questions of this
viewers or interviewees (for example, gender, kind are to be asked, a personal interview may
class, and ethnicity). The remoteness of the be superior (Shuy, 2002).
interviewer in telephone interviewing removes • Telephone interviewers cannot see the respon-
this potential source of bias to a significant dents and therefore cannot respond to signs of
extent, since the interviewers’ “race” and gen- puzzlement or unease on their faces. In a per-
eral appearance cannot be seen, although their sonal interview, the interviewer may respond to
gender is usually apparent. such signs by restating the question or attempt-
ing to clarify its meaning, though this has to be
On the other hand, telephone interviewing suffers handled in a standardized way as much as pos-
from certain limitations when compared to the per- sible. A further issue is that interviewers may be
sonal interview: able to collect subsidiary information, such as
whether the home in which the interview was
• People who do not have a telephone or other- conducted is in need of repair. That sort of infor-
wise cannot be contacted by phone cannot be mation cannot be acquired over the telephone.
part of the study. Since lack of a phone is most • Frequently, the target of the research is a spe-
likely to be a feature of poorer households, the cific individual in a household or firm: for ex-
potential for sampling bias exists. Also, many ample, the person in a certain role or position,
people choose to be ex-directory (they pay to or someone with particular characteristics. It
have their telephone numbers not listed in a is probably more difficult to ascertain by tele-
telephone book) or they use only cellphones phone interview whether the correct person is
and for that reason are not listed. This presents replying.
a problem if the researcher is using a directory • The telephone interviewer cannot employ visual
to select respondents. A solution to the latter aids such as “show cards” from which respon-
problem is random-digit dialling. With this dents select their replies (see Box 5.6). Similarly,
technique (used, for example, in the General diagrams or photographs cannot be used.
Social Survey administered by Statistics
Canada), a computer randomly selects tele- There is some evidence that the quality of the data
phone numbers within a predefined geograph- gathered through telephone interviews is inferior to
ical area. This method can give you access to that produced by face-to-face interviews. According to

bel65796_ch05_100-138 117 01/13/22 12:19 PM

118 PART II Quantitative Research

vote in the recent federal election?” Only those who

said yes were then asked, “Which party did you vote
Box 5.6 Two show cards for?” Once the response to the filter question has
been keyed in, the computer will automatically go to
Card 4 (age) Card 6 (for various items)
the next relevant question, eliminating inappropri-
a. Less than 20 1. Strongly agree ate items where necessary. This removes the possibil-
b. 20–29 2. Agree ity of interviewers inadvertently asking meaningless
c. 30–39 3. Undecided questions or failing to ask some that should be asked.
d. 40–49 4. Disagree In this way, computer-assisted interviewing im-
e. 50–59 5. Strongly disagree proves standardization in the asking and recording
f. 60–69 of questions.
g. 70 and over Computer-assisted interviewing makes it easier to
perform other tasks as well. For example, sometimes
it is beneficial to randomize the order in which cer-
Holbrook, Green, & Krosnick (2003), a series of experi- tain items are presented to the respondent in order
ments in the US using long lists of questions found that to determine whether the responses given depend
respondents interviewed by telephone were more likely on the order in which the items are listed. Automatic
to express no opinion or give “don’t know” as an answer; randomization was used in the Goyder, Guppy, &
give the same answer to a series of linked questions; ex- Thompson (2003) study of occupational prestige
press socially desirable opinions; be apprehensive about done in southwestern Ontario. As the authors put it:
the interview; and be dissatisfied with the time taken by “The deck was electronically shuffled for each new
the interviews (even though they were shorter than the respondent.” Another advantage of computer-as-
face-to-face kind). Also, telephone interviewees tend to sisted interviews is that if interviewers are out in the
be less engaged in the interview process. field all day, the data can be sent to the research office
electronically.
Computer-assisted interviewing
Today it is common to use computers in the inter- Using online (email) personal interviews
view process. There are two main formats for such Online interviews run a higher risk of respondent
interviews: computer-assisted personal interviewing dropout than other forms do. However, Mann and
(CAPI) and computer-assisted telephone interview- Stewart (2000, pp. 138–139) have suggested that it
ing (CATI). The use of CAPI has been growing lately, is possible to develop a relationship of mutual trust
mainly because of the increased portability and af- when using them. This can be accomplished by regu-
fordability of laptop computers and other electronic larly sending messages to respondents assuring them
devices, and the availability of quality software that their responses are helpful and significant, espe-
packages. cially since online interviewing is still an unfamiliar
With computer-assisted interviewing, the ques- experience for most people and takes longer than
tions in the interview schedule appear on the screen, other forms. It is worth the trouble because online
the interviewer keys in the reply, and the computer interviews make it relatively easy for the researcher
shows the next question. This approach is espe- to go back to interviewees for further information or
cially useful when filter questions are asked. In the reflections—something difficult to do with face-to-
2015 Canadian Election Study (funded by Elections face interviews.
Canada), respondents were asked the filter ques- A further issue for the online personal interviewer
tion, “The federal election was held on Monday, is whether to send all the questions at once or to con-
October 19. In any election, some people are not able duct the interview on a question-followed-by-reply
to vote because they are sick or busy, or for some basis. The problem with sending all at once is that
other reason. Others do not want to vote. Did you respondents may read through them and reply only

bel65796_ch05_100-138 118 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 119

to the most interesting ones. Bampton and Cowton Webcam technologies like Skype and Zoom
(2002) reported that conducting email interviews by may offer further possibilities for online personal
sending questions in small batches takes the pressure interviews, making the online interview similar to
off interviewees to make a quick reply, gives them the the telephone version and quite comparable to an
opportunity to provide considered replies (although in-person interview because those involved in the
there can be a loss of spontaneity), and gives the in- exchange can see each other. However, one of the
terviewers greater opportunity to respond to inter- main advantages of the email interview is lost: re-
viewee answers. spondents’ answers need to be transcribed.
There is evidence that prospective online inter-
viewees are more likely to agree to participate if agree- Conducting interviews
ment is solicited prior to sending the actual questions. Know the interview schedule
Another way to encourage respondents to participate Before interviewing anybody, an interviewer should
is for the researcher to use some form of self-disclo- have a thorough knowledge of the interview sched-
sure, such as directions to a website that contains ule. Interviewing can be stressful, and it is possible
detailed contact information. It also helps to provide that under strain standard interview procedures can
personal material about the researcher (such as a cause interviewers to get flustered and leave questions
photo), as well as information relevant to the research out or ask the wrong questions. All interviewers need
topic (Curasi, 2001; O’Connor & Madge, 2001, 2003). to be fully trained to reduce interviewer variability in
Such steps are necessary because unsolicited emails asking questions, a potential source of error. Fowler
(“spam”) are often seen as a nuisance and can result (1993) cited evidence to suggest that a training period
in an immediate refusal to take the message seriously. shorter than one full day is rarely sufficient.
Curasi (2001) conducted a comparison in which
24 interviews carried out through email correspon- Introducing the research
dence were contrasted with 24 parallel face-to-face Prospective respondents have to be offered a credible
interviews. The interviews were concerned with rationale for participating in the research, in particular
shopping online. She found that: for giving up their valuable time. This aspect of con-
ducting interviews is of special significance at a time
• face-to-face is better than online for maintain- when survey response rates appear to be declining.
ing rapport with respondents; The introductory rationale may be either spoken by
• because greater commitment and motivation the interviewer or printed for the respondent to read.
are required to complete an online interview, It is usually spoken if interviewers “cold call” potential
replies are often more detailed than in face-to- respondents at their homes, either in person or by tele-
face interviews; and phone. A written rationale is common to alert respon-
• online interviewees’ answers tend to be more con- dents that someone will be contacting them to request
sidered and grammatically correct because they an interview. In many cases, respondents may be pre-
have more time to ponder and tidy up answers. sented with both modes—for example, researchers will
Whether this is a positive feature is debatable. send respondents a letter inviting them to participate
There is the obvious advantage of a “clean” tran- in a study and then call or speak to them in person.
script, but there may be some loss of spontaneity.
Rapport
On the other hand, Curasi found that some online As suggested earlier, interviewers need to establish a
responses were short on detail, perhaps because rapport with respondents. A relationship must be forged
replies had to be typed, not just spoken. The full fairly quickly to encourage respondents to participate
significance of the difference between email and in and persist with the interview. While this injunction
face-to-face interviewing is still being explored by essentially invites the interviewer to be friendly with
researchers. respondents and put them at ease, it is important not

bel65796_ch05_100-138 119 01/13/22 12:19 PM

120 PART II Quantitative Research

Methods in Motion Applications to Canadian Society

Father involvement in parenting in Canada: The role of mental health
There is considerable evidence that the degree The results indicated that by and large, the fa-
to which fathers are involved in raising their chil- thers in the study had fairly high levels of involve-
dren has important, long-lasting effects on their ment with their children, with scores on various
children’s well-being. However, relatively little re- involvement scales clustering at the high end of
search has been conducted on what might enable the spectrum. On the whole, Canadian fathers
or impede fathers’ participation in bringing up seem to be good parents, offering responsible
their children, especially in the Canadian context. and affectionate support for their kids. Paternal
Shafer and Renick (2020) focus on one possible depression levels were, on average, compara-
factor: the fathers’ mental health—specifically, tively low.
the degree to which fathers exhibit symptoms of With regard to the association between de-
depression. The effect of paternal mental health pressive symptoms and fathers’ involvement in
on childrearing in Canada has to date received childrearing, higher levels of depression were re-
very little attention from social researchers. lated to lower levels of father–child relationship
To conduct their study, the authors used a quality, after accounting for various socio-de-
quota sample (see Chapter 7) drawn from an mographic factors. One of the key findings of the
online opt-in survey panel, which is made up of study was an association between high levels of
people who have joined a centralized database depression and the use of harsh disciplinary tech-
of respondents willing to participate in surveys, niques, which according to previous research
often for small amounts of money or other re- tends to create dysfunctional relationships be-
wards. The sample included biological fathers, tween fathers and their children and are often
social fathers, and stepfathers in Canada whose accompanied by physical and mental health dif-
children were from 1 to 18 years of age. A number ficulties on the part of the children. Nonetheless,
of data quality checks were performed, but given depressed fathers generally fulfilled the basic
the nature of the sampling process, Shafer and responsibilities of providing the necessities of life
Renick cautioned that their findings may not be for their children about as well as the fathers who
applicable to Canadian society as a whole. were not depressed.

to take it too far. Too much rapport can result in the in- • If you are a student doing research for a thesis,
terview going on too long and can also lead respondents make that fact clear.
to tailor their answers to please the interviewer. Thus, • Indicate what the research is about in broad
achieving the appropriate rapport between interviewer terms and why it is important, and give an
and respondent is a delicate balancing act. indication of the kind of information to be
collected.
Topics and issues to include in an • Indicate how the respondent has been selected
introductory statement (for example, by a random process, by conve-
In an introductory statement to a prospective nience, because of special characteristics).
interviewee: • Provide reassurance about the confidentiality
of any information given by the respondent.
• Clearly identify yourself. • Explain that participation is voluntary.
• Identify the auspices under which the research • Reassure respondents that they will not be
is being conducted (for example, a university, a identified or identifiable. This can usually
government agency). be achieved by pointing out that when the

bel65796_ch05_100-138 120 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 121

data are aggregated or analyzed at the group • When the interviewer needs to know about
level, individual participants cannot be something that requires quantification, such
identified. as the number of visits made to a doctor in
• Give the respondent the opportunity to ask the last four weeks or the number of banks in
questions; be sure to provide a contact tele- which the respondent has accounts, but the
phone number or email address. If in person, respondent answers in general terms (“quite
simply ask if the respondent has any questions. often” or “I have several”), the interviewer
needs to persist for a more precise answer.
These suggestions are also relevant to the covering With fixed-response surveys, this will usu-
letter that should accompany online and mailed ally entail repeating the response options. The
questionnaires. The latter should also include a interviewer should not suggest an answer on
stamped, pre-addressed return envelope. the basis of the respondent’s reply since the re-
spondent may not be comfortable with the idea
Probing of disagreeing with the interviewer.
A highly problematic area for researchers employing
a structured interview method is probing respon- In a face-to-face interview, the interviewer may use
dents who need help with their answers. This may “show cards” (see Box 5.6) instead of reading out a
occur if respondents do not understand the question series of fixed-choice alternatives. Sometimes called
and then struggle to provide an adequate answer, or “flash cards,” they display all the answers from which
if they do not provide a complete answer and have the respondent is to choose and are handed to the
to be probed for more information. For example, respondent at different points in the interview. There
in the 2013 Alberta Survey, respondents were asked are several situations in which it may be beneficial to
“What is your religion, if any?” If they answered use show cards:
“Christian,” the probe available for all interviewers
was: “Any particular denomination?” • Sometimes there is a very long list of possi-
A potential problem with some probes is that ble answers; for example, respondents may be
some interviewers may be more inclined to use it asked about which social media sites they use
than other interviewers, and the probe itself may most frequently. To read out a list of sites would
affect the response given, which could lead to reli- be tedious and it is probably better to hand the
ability problems. A bigger problem arises if different respondent a list from which to choose.
interviewers use different probes, although proper • Some people are not keen to divulge personal
organization and training should preclude that. details such as their age or income. One way of
Some general tactics with regard to probes are as reducing the impact of such questioning is to
follows: present respondents with age or income ranges
with a letter or number attached to each (see
• If further information is required, usually in Box 5.6). A similar approach may be used with
the context of an open-ended question, stan- other sensitive topics such as sexual practices.
dardized probes should be employed, such However, this procedure will obviously not be
as “Can you say a little more about that?” or appropriate if the research requires exact fig-
“Are there any other reasons why you think ures pertaining to sensitive topics.
that?” • Sometimes, during the course of an inter-
• If, with a closed question, the respondent re- view, respondents will be presented with a
plies in a way that does not match one of the series of statements or questions to which the
pre-designed answers, the interviewer should same possible responses apply—for example:
repeat the fixed-choice alternatives and make “strongly agree, agree, neutral, disagree, and
it clear that the answer must be chosen from strongly disagree.” It would be time-consum-
those provided. ing and off-putting to read out all five possible

bel65796_ch05_100-138 121 01/13/22 12:19 PM

122 PART II Quantitative Research

answers over and over again, but it might also the heritability of political attitudes. Sometimes
be expecting too much of respondents to ask respondents will be asked to deposit their com-
them to keep all the possible answers in their pleted questionnaires at a specified location: for
heads for the entire batch of questions to which a study conducted in a business organization, for
they apply. Providing a show card listing the instance, the location might be their supervisor’s
possible responses is an obvious solution. office. Researchers may hand out questionnaires
to students in a class and collect them there as well,
Prompting as Smith and McVie (2003) did in their longitudi-
Prompting occurs when the interviewer suggests a nal cohort study on crime. And, of course, ques-
specific answer to a particular interviewee. It is very tionnaires can also be filled out online, which was
rare and should be used only as a last resort. An ex- an option for the 2016 Canadian census. Online
ample of unacceptable prompting would be to ask platforms like Amazon’s Mechanical Turk and
an open-ended question and then suggest a possible software such as Survey Monkey are becoming
answer to a respondent who appears to be struggling increasingly popular among social researchers.
to think of an appropriate reply. In all situations, in- In many ways the questionnaire and structured
terviewers should do what they can to allow respon- interview are very similar. The obvious difference
dents to come up with their own replies. Otherwise, is that with questionnaires, respondents must
the data gathered will not be authentic—they may read the questions themselves and record their
also reveal more about the interviewer than the own answers. Because there is no interviewer to
interviewee. administer it, the research instrument has to be
especially easy to follow and the questions partic-
Assessing interviewers ularly easy to answer. As a result, questionnaires
Researchers use various ways to determine whether tend to differ from structured interviews in the
an interviewer has been trained properly, including: following ways:

• checking individual interviewers’ response • They have fewer open questions because closed
rates; questions are easier to answer.
• recording a sample of interviews; • They have easy-to-follow designs to minimize
• examining completed schedules to deter- the risk that a respondent will inadvertently
mine whether any questions are being left miss a question or part of one.
out and ensure that they are being completed • They are shorter to reduce the risk of “re-
properly; and spondent fatigue” (it’s much easier for a tired
• making call-backs on a sample of respondents respondent facing a long questionnaire to
(usually about 10 per cent) to make sure they abandon it than it is for a tired interviewee to
were interviewed and to ask about the inter- ask an interviewer to leave).
viewer’s conduct.

Advantages of the questionnaire

Questionnaires over the structured interview
A questionnaire is essentially a structured inter- Cheaper, quicker, more convenient to
view without an interviewer. Questionnaires can administer
be delivered and retrieved in several different Interviewing in person is expensive and time-con-
ways. One way is by mail. That is how Anderson et suming. The cost is obviously less with telephone
al. (2006) studied family physicians in southwest- and email interviews, but even there the question-
ern Ontario, and how Bell et al. (2009) distrib- naire enjoys cost advantages. A class of 400 students
uted questionnaires to twins in their research on can fill out questionnaires in one class period, and

bel65796_ch05_100-138 122 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 123

the number of people accessing a questionnaire question when they are on their own than when they
online is virtually unlimited. Even with a team of are with an interviewer; questions that respondents
telephone interviewers, it takes a long time to con- find boring or irrelevant are especially likely to be
duct personal interviews with such large sample skipped, although some software programs can be
sizes. However, it is important to remember that set up in such a way that participants cannot skip
questionnaires may not be filled out immediately certain questions. Questionnaire respondents are
(respondents usually complete them at their con- more likely than interview participants to become
venience). In many situations it is necessary to tired of answering questions that are not fully sa-
send out follow-up letters and/or duplicate ques- lient to them, and to abandon the project entirely.
tionnaires to those who fail to complete them in a Put positively, when a research issue is important to
timely manner. the respondent, a high response rate is possible. This
means that when the questions are highly relevant,
Absence of interviewer effects a questionnaire may be a good choice, especially be-
We have already noted that interviewer characteris- cause of its much lower cost.
tics may affect respondent answers. Obviously, those
sorts of effects are not an issue with questionnaires, Cannot probe
since there is no interviewer present. Similarly, with There is no opportunity to probe respondents to elab-
a questionnaire, no one is there to read the questions orate on an answer. However, this problem mainly
to the subject in the wrong order, to present them applies to open questions, which are not widely used
in different ways to different respondents, or to state in questionnaire research.
the items with varying emphases from person to
person; the problems those practices can cause are Difficult to ask a lot of questions
precluded. Long questionnaires are rarely feasible because of the
Probably of greater importance is the tendency risk of respondent fatigue. They may also discourage
for people to exhibit a social desirability bias when prospective respondents from participating at all.
an interviewer is present, giving “politically cor-
rect” rather than genuine responses. Research sum- Difficult to ask other kinds of questions
marized by Tourangeau and Smith (1996) strongly Long or complex questions should be avoided as far
suggests that respondents tend to report more drug as possible, since some respondents find them diffi-
use and alcohol consumption and higher numbers of cult to follow.
sexual partners and abortions when responding to
questionnaires than in structured interviews. Questionnaire can be read as a whole
When respondents are able to read the whole ques-
tionnaire even before answering the first question,
Disadvantages of questionnaires
none of the questions asked is truly independent of
versus structured interviews
the others. It also means that the researcher cannot
Cannot explain the question be sure that questions have been answered in the
It’s always important to ensure that the questions correct order, raising the possibility of question-or-
asked are clear and unambiguous, but this is espe- der effects, although some software packages can
cially true with questionnaires, since there is no in- minimize this problem.
terviewer to help respondents with questions they
cannot understand. Not appropriate for some kinds
of respondents
Greater risk of missing data Respondents whose literacy is limited or whose fa-
Incomplete questionnaires are more common. It is cility with the language used is restricted may not be
also easier for respondents to decide not to answer a able to answer a questionnaire. The second of these

bel65796_ch05_100-138 123 01/13/22 12:19 PM

124 PART II Quantitative Research

difficulties cannot be entirely overcome with an in- Knowing how to read and then return an attach-
terview, but the difficulties are likely to be greater ment requires a certain facility with online com-
with a questionnaire. munication that is still not universally possessed.
Also, the recipients’ operating systems or software
One last problem: Who filled out the may not be able to handle attachments, or respon-
questionnaire? dents may refuse to open them because of concerns
With mailed or online questionnaires, the researcher about viruses. On the other hand, since most email
can never be sure whether the questions were an- software allows only limited formatting, the ap-
swered by the designated respondent or someone pearance of embedded questionnaires tends to be
else. It is also impossible to prevent other members of rather dull and featureless, although this situation
a household (for example) from helping the respon- is rapidly changing. Furthermore, it is slightly easier
dent answer the questions. for respondents to type answers into an attachment
that uses well-known software such as Microsoft
Online social surveys Word because if the questionnaire is embedded in
As mentioned, the number of surveys being ad- an email the alignment of questions and answers
ministered online has grown considerably. There may be lost.
are both pros and cons with online surveys as
Web surveys
compared to mailed questionnaires (see Box 5.7).
There is also a crucial distinction between sur- Web surveys invite prospective respondents to visit
veys administered by email (email surveys) and a website where the questionnaire can be found and
surveys administered via the Web (Web surveys). completed online. The Web survey has an import-
In the case of the former, the questionnaire is sent ant advantage over the email survey in that it can
via email to a respondent, whereas with a Web use a much wider variety of embellishments, such
survey, the respondent is directed to a website in as “radio buttons” and pull-down menus of possible
order to answer it. answers. With open questions, the respondent is in-
vited to type directly into a boxed area.
However, the advantages of the Web survey go
Email surveys
beyond appearance and format. As with interview
With email surveys it is important to distinguish be- software, questionnaires can be designed so that
tween embedded and attached questionnaires. When when there is a filter question, it skips automatically to
the questions are embedded in the body of the email, the next appropriate question. The questionnaire can
there may be an introduction to the questionnaire also be programmed either to show only one question
followed by a graphic that partitions the introduction at a time or to allow the respondent to scroll down
from the questionnaire itself. Respondents may be and look at all the questions in advance. If a question
asked either to indicate their replies using simple nota- is missed, the software may provide a prompt telling
tion, such as an “X,” or to delete alternatives that do not the respondent that it was not answered and that a
apply to them. If a question is open, they are asked to response is needed. Finally, respondents’ answers
type in an answer. When finished, they simply hit the can be automatically programmed to download into
reply button to return the completed questionnaire. a database, eliminating the daunting task of coding
An attached questionnaire arrives as an attach- a large number of questionnaires. Software packages
ment to an email that introduces it. As with the em- designed to produce questionnaires with all the fea-
bedded questionnaire, respondents must select and/ tures just described are now widely available.
or type their answers. To return the questionnaire, it Web surveys are also advantageous in that they
must be attached to a reply email. may give researchers access to unique populations,
The chief advantage of the embedded question- such as people who share a particular interest.
naire is that it requires less computer expertise. Web surveys are particularly useful for reaching

bel65796_ch05_100-138 124 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 125

Box 5.7 Advantages and disadvantages of online surveys

compared to mailed questionnaires
This box summarizes the main advantages and online population still differs in significant
disadvantages of online surveys, both email and ways from the non-online population, it re-
Web varieties, compared to mailed question- mains a difficulty.
naires. Note that all three share one disadvantage • Confidentiality and anonymity issues. It is
relative to personal and telephone interviews: standard practice for survey researchers to
indicate that replies will be kept anonymous.
namely, that the researcher can never be certain
However, with email surveys the addresses of
of who is actually answering the questions.
the respondents are known, which may make
it difficult for people to believe that their iden-
Advantages tity really will be protected.
• Low cost. Even though mailed questionnaires • Multiple replies. With Web surveys, there is
are inexpensive to administer, online surveys a risk that some people may complete the
cost even less. questionnaire more than once, although soft-
• Faster response and processing. Online ware is available that can reduce that possi-
surveys can be returned considerably faster bility. There is much less risk of this with email
than mailed questionnaires. Researchers also surveys.
appreciate the automatic skipping of irrele-
vant questions that is possible when they use
filter questions, as well as the opportunity
for immediate downloading of replies into a
database.
• Fewer unanswered questions. There is ev-
idence that online questionnaires are com-
pleted with fewer unanswered questions than
mailed questionnaires.

cnythzl/iStockphoto
• Better response to open questions. Open
questions are more likely to be answered
online, and online replies are usually more
detailed.
Have you ever participated in an online survey,
Disadvantages either through email or via the Web? If so, were
• Low response rate. Typically, response rates you concerned about a possible lack of ano-
for online surveys are lower than those for nymity? Did this affect how you answered the
comparable mailed questionnaires. questions?
• Restricted to online populations. This re- Sources: Adapted from Cobanoglu, Ward, & Moreo (2001); Kent & Lee
(1999); Schaeffer & Dillman (1998); Sheehan & Hoy (1999); Sinclair, O’Toole,
striction should ease over time, but since the Malawaraarachchi, & Leder (2012); Tse (1998).

stigmatized populations who are difficult to reach integrity (i.e., cheating and plagiarism) in Canadian
offline, such as people with eating disorders or and American dental schools. It may be most effi-
extreme political views (Wright, 2005). cient to send the email invitation to someone who
Potential respondents need to be directed to the can forward it to large numbers of potential par-
website containing the questionnaire. One way to ticipants. In the Andrews et al. study the invitation
do this is to email them, as Andrews, Smith, Henzi, was sent to academic deans who then forwarded it to
& Demps (2007) did in their study of academic faculty members and students. For surveys in which

bel65796_ch05_100-138 125 01/13/22 12:19 PM

126 PART II Quantitative Research

participation must be restricted, it may be necessary

to set up a password system to filter out people for Box 5.8 Should a closed question
whom the questionnaire is not appropriate. have a vertical or a
horizontal format?
Designing the questionnaire
The following question is taken from the 2015
No matter the medium of delivery, there are im- Canadian Election Study—Mailback Survey:
portant considerations that must go into a question- Which party do you feel closest to?
naire’s visual design. These are discussed below.
Conservative Party 1
Clear presentation
Liberal Party 2
Make sure that the layout is easy on the eye and that it NDP 3
facilitates answering all the questions applicable to the Bloc Québécois 4
respondent. Using a variety of styles (different fonts
Green Party 5
and print sizes, boldface, italics, capitals, etc.) can en-
Other (specify) _________ 6
hance the questionnaire’s appearance so long as they
are used consistently. Thus, all general instructions And this is the same question in a horizontal
should appear in one style, all headings in a second, format:
all specific instructions (e.g., “Go to question 7”) in Which party do you feel closest to?
a third, and so on. Mixing print styles or uses (e.g.,
using a particular style for both general instructions Conservative 1  Liberal 2
and specific questions) may confuse respondents. NDP 3  Bloc Québécois 4
Green Party 5
Vertical or horizontal closed answers? Other (specify) _________ 6
Bearing in mind that most questions in a question- Source: Fournier, Cutler, Soroka, & Stolle, D. (2015), accessed 8
naire are likely to be closed, you will need to decide August 2020.

whether to arrange the fixed answers vertically or

horizontally. Usually, a vertical arrangement is used Should they circle it? Is it acceptable to choose more
unless respondents are asked to choose their answer than one answer? If not, this should be indicated in
from a row of numbers or abbreviations, as is often the instructions, for example:
done with Likert scales. Many writers prefer a verti-
(Please choose the ONE answer that best rep-
cal format because it reduces the risk of confusion.
resents your views by placing a tick in the ap-
With a horizontal format there is a significant risk,
propriate box.)
especially if the respondent is in a hurry, that the re-
quired tick or circle will be put in the wrong place If this is not made clear and some respondents
(for an example, see Box 5.8). Also, a vertical format choose more than one answer, their replies cannot
more clearly distinguishes questions from answers. be counted and will have to be treated as if they were
A further advantage of the vertical format is that it missing. If choosing more than one category is ac-
may facilitate the manual recording of answers, espe- ceptable, this too must be made clear, for example:
cially when pre-codes appear on the questionnaire.
(Please choose ALL answers that represent your
Clear instructions about how to respond views by placing a tick in the appropriate boxes.
You may tick more than one box for each ques-
On questionnaires in which the selection of re-
tion—choose as many as apply to you.)
sponses is not automated, always be clear about how
respondents should indicate their replies to closed It is a common error for such instructions to be
questions. Are they supposed to place a tick by the omitted and for respondents to be unsure about how
appropriate answer, or are they to underline it? to reply.

bel65796_ch05_100-138 126 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 127

Keep question and answers together from HIV infection they felt during the sexual en-
This simple rule may seem obvious, but it is often counter, and so on.
transgressed: a question, or the question and the an- Sullivan’s (1996) research on the domestic di-
swers that accompany it, should not be split between vision of labour provides another illustration of a
two separate pages. A common error is for a question structured diary. The kind of diary Sullivan used is
to appear at the bottom of a page and the responses often described as a “time-use” diary because it is de-
on the next page. The danger here is that the respon- signed for recording, more or less at the time of the
dent will either (a) fail to answer the question or (b) actual behaviour, how long people engage in activi-
pay insufficient attention to the question and give a ties such as food preparation, child care, or eating.
superficial answer (this problem is especially likely Sullivan also asked couples to record on a five-point
when a series of questions with a common answer scale the amount of enjoyment they derived from
format is being used, as with a Likert scale). those kinds of activities.
Recording events as they happen is often con-
Researcher-driven diaries as a form sidered a more accurate approach than recording
of questionnaire them at the end of the day, largely because there is
When the researcher is interested in precise accounts less risk of forgetting. However, it is more intrusive
of behaviour, what Elliott (1997) called the research- than end-of-day reporting or answering a question-
er-driven diary can be used as a kind of questionnaire. naire because it is a constant interruption and may
The researcher asks participants to record their per- even change people’s behaviour. For example, one
ceptions, feelings, or actions with regard to certain can become preoccupied with the behaviours under
experiences shortly after they occur. Laurenceau and study and thus engage in them in a more thoughtful
Bolger (2005) used the diary method to study mar- or premeditated way.
ital and family processes. Hessler et al. (2003) used Sometimes diaries can be constructed as part
researcher-driven diaries in an email study of ado- of a larger structured interview. This was the ap-
lescent risk behaviour that produced rich narratives proach taken in the 2015 Time Use study, which
of the participants’ everyday lives. Sometimes this was part of the General Social Survey conducted
method is supplemented by a personal interview in for Statistics Canada. A 24-hour diary was pro-
which the researcher will follow up on points raised duced by asking people what they were doing on
in the diary: for example, by asking the diarist what a certain day, starting at 4:00 a.m. The first ques-
he or she meant by certain remarks. Diary data may tion was: “Last [day of the week] at [4:00 a.m.], what
also be used in conjunction with information de- were you doing?” (Most people say “sleeping,” and
rived from surveys, as in Robinson and Lee’s (2014) are later asked how much time they spent sleeping,
examination of the effect of new information tech- and what time they went to sleep.) Sometimes ques-
nologies on people’s social lives and mass media use. tions such as “Where were you?” or “Who was with
Corti (1993) distinguished between “structured you?” are asked as well.
diaries” and “free-text diaries.” Either can be em- An example of a free-text diary is provided in
ployed by quantitative as well as qualitative re- Crook and Light’s (2002) study in which university
searchers. Structured diaries, such as those used by students were asked to keep a diary for a week, di-
O’Sullivan, Udell, & Patel (2006) to examine sexual vided into 15-minute intervals, of the different kinds
practices among college students, have the general of study and learning activity in which they engaged,
appearance of a questionnaire with largely closed where the activity took place, and the study resources
questions. At the end of each day the students in used (for example, the library). The various activities
the study indicated whether they had engaged in were grouped into three types: classes, private study,
sexual activity that day, and if they had, they used and social study (that is, study with a peer). They
a checklist to mark which sexual activity had been were able to show the very different amounts and
performed, whether they used a condom, how safe patterns of study typically undertaken during a day.

bel65796_ch05_100-138 127 01/13/22 12:19 PM

128 PART II Quantitative Research

Using free-text recording of behaviour carries On the other hand, the diary method may present
the same kinds of coding problems as those associ- the following problems:
ated with open-ended survey questions—namely, the
time-consuming nature of the exercise and the risk • Diaries tend to be more expensive than per-
of introducing error while coding answers. However, sonal interviews because of costs related to
the free-text approach is less likely to be problematic recruiting diarists and ensuring that they are
when, as in the case of Crook and Light’s (2002) re- using their diaries properly.
search, diarists are given instructions about what is re- • There is a risk of attrition as diarists tire of the
quired and the behaviours of interest are specific and task, which raises reliability issues.
concrete. It would be much more difficult to code free- • Failure to record details quickly enough can
text entries relating to more general behaviours such result in errors or omissions.
as the domestic activities studied by Sullivan (1996).
Corti (1993) recommended that people doing re- However, diary researchers such as Sullivan argue
search involving diaries should: that the information collected through diaries is
more accurate than that collected through inter-
• provide explicit instructions for diarists; views or questionnaires on the same topic.
• be clear about the time periods in which the
behaviour of interest is to be assessed (for ex- Secondary analysis
ample, daytime, certain 24-hour periods, par- of survey data
ticular weeks), and indicate that in the diary
template; Survey research, including questionnaires, can be
• supply a model of a completed section of a extremely time-consuming and expensive to con-
diary; and duct, and is usually well beyond the means of most
• provide brief checklists of “items, events, or students. This is where secondary analysis comes
behaviour” to jog the memory. in. Large amounts of quantitative data already
exist, collected by individual social scientists and
by organizations such as government departments
Advantages and disadvantages of the and university-affiliated research centres (see Box
diary as a method of data collection 5.9). The latter include the Institute for Social
The studies illustrating the use of diaries discussed Research at York University, and the Institute
above suggest potential advantages: of Urban Studies at the University of Winnipeg.
Sources such as these make raw statistical data
• When fairly precise estimates regarding fre- available on topics such as income, fertility, crime,
quency and/or amount of time spent in differ- and unemployment, as well as a host of other
ent forms of behaviour are required, the diary social and political issues. Using secondary data
may provide more valid and reliable data than rather than collecting your own has the additional
questionnaires. advantage of sparing an already over-surveyed
• When information about the sequencing of public yet another round of questions. For this
different types of behaviour is required, it is reason, secondary analysis should be considered
likely to perform better than questionnaires or not just by students but by all social researchers.
interviews. Indeed, some granting agencies require applicants
• A diary is useful for collecting data on be- proposing to collect new data to demonstrate
haviour that is personally sensitive, such as that relevant data are not already available in an
sexual activity. archive.

bel65796_ch05_100-138 128 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 129

Box 5.9 Existing data sets

Many important studies have been conducted after finishing their post-secondary edu-
through secondary analysis of existing data. cation; used by Walters (2004) and Marin &
Among the high-quality data sets used by re- Hayes (2017).
searchers in Canada are the following: 4. The Longitudinal Immigration Database
(IMDB) developed by Citizenship and
1. The National Longitudinal Survey of Children Immigration Canada and Statistics Canada
and Youth, which contains data on young to provide data for the study of immigrants’
people up to age 25; used by Brannigan, long-term earnings; used by Li (2003) and
Gemmell, Pevalin, & Wade (2002), Kerr Mata & Pendakur (2017).
(2004), Beran, Hughes, & Lupart (2008), 5. The Canadian Election Studies, which pro-
Ames, Rawana, Gentile, & Morgan (2015), vide data for all federal elections from 1965
and Comeau, Duncan, Georgiades, Wang, & to the present; they are available through the
Boyle (2020). Inter-University Consortium for Political and
2. The National Population Health Survey, Social Research; scores of publications have
begun in 1994–5 with data gathered every been produced using these surveys, for exam-
two years until 2011; it examines health and ple, Gidengil, Everitt, Blais, Fournier, & Nevitte
health care in Canada; used by Légaré, (2006), Goodyear-Grant & Croskill (2011),
Décarie, & Bélanger (2014) and Friedenreich Kevins & Soroka (2018), & Anderson and
et al. (2019). McGregor (2020).
3. National Graduates Surveys, which ask grad-
uates about their employment two years These are only a few of the available data sets.

Advantages of secondary analysis as anyone else, well-established procedures

There are several reasons for considering secondary are usually in place for following up with
analysis a serious alternative to collecting new data. non-respondents and thus keeping this prob-
Its advantages have been enumerated by Dale, Arber, lem to a minimum. Second, the samples are
& Proctor (1988): frequently national in scope, or at least cover
a wide variety of regions, which is a highly
• Cost and time. As we noted above, secondary desirable—and costly—feature. Third, many
analysis often offers good-quality data for a data sets have been generated by highly expe-
tiny fraction of the cost involved in collecting rienced researchers and, in the case of some of
new data. the large data sets, gathered by social research
• High-quality data. Most of the data sets organizations with strong control procedures
employed for secondary analysis are of ex- to check on data quality.
tremely high quality. First, the sampling pro- • Opportunity for longitudinal analysis. Another
cedures have often been rigorous, in many valuable feature of secondary analysis is that
cases producing samples that are as close it can provide an opportunity for longitudi-
to representative as is reasonably possible. nal research. Sometimes, as with the General
Although those responsible for these stud- Social Survey (GSS; see Box 5.10), a panel
ies face the same problems of non-response format (in which the same participants are

bel65796_ch05_100-138 129 01/13/22 12:19 PM

130 PART II Quantitative Research

Box 5.10 General Social Survey features

Introduction policy issues of current interest. The GSS is a

In 1985, Statistics Canada introduced the continuing research project that operates on an
General Social Survey (GSS), which covers annual cycle. Each annual survey classifies par-
major topics such as the health of Canadians ticipants by age, sex, education, and income.
and their levels of social support. The GSS Core content areas, however, cannot be treated
has two principal objectives: first, to gather adequately every year. Therefore, they are usu-
data regularly on Canadian social trends; and ally covered every five years. The content by
second, to provide information on specific cycle is as follows:

Cycle Year Topics

1 1985 Health, Social Support
2 1986 Time Use, Social Mobility, Language
3 1988 Personal Risk, Victim Services
4 1989 Education and Work
5 1990 Family and Friends
6 1991 Health (Various Topics)
7 1992 Time Use, Culture, Sport and Unpaid Activities
8 1993 Personal Risk, Alcohol and Drug Use
9 1994 Education, Work, Transition into Retirement
10 1995 Family Effects of Tobacco Smoke
11 1996 Social Support, Tobacco Use
12 1998 Time Use
13 1999 Victimization, Spousal Violence, Senior Abuse, and Public Perceptions of
Alternatives to Imprisonment
14 2000 Access to and Use of Information Communication Technology
15 2001 Family History
16 2002 Social Support and Aging
17 2003 Social Contact with Family, Friends, and Neighbours; Involvement in Formal
Organizations; Political Activities and Volunteer Work; Values and Attitudes; Trust
in Public Institutions
18 2004 Criminal Victimization
19 2005 Replication of Time-Use Study

examined at different times) is used to chart • Subgroup analysis. When samples are large (as
trends over time. Similarly, because certain in the GSS) there is an opportunity to study
interview questions are recycled and asked of subgroups. A sample of 100 people, for exam-
different samples each year, shifting opinions ple, might include only three people over age
or changes in behaviour can be identified. 85. No quantitative researcher wants to talk

bel65796_ch05_100-138 130 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 131

Cycle Year Topics

20 2006 Family Life in Canada
21 2007 Life transitions experienced by Canadians over the age of 45, such as those
relating to employment, family composition, social networks and housing
22 2008 Social Networks, Social and Civic Participation
23 2009 Victimization
24 2010 Time-Stress and Well-Being
25 2011 Family
26 2012 Caregiving and Care Receiving
27 2013 Giving, Volunteering and Participating; Social Identity
28 2014 Canadians’ Safety (Victimization)
29 2015–16 Time Use
30 2016 Canadians at Work and Home
31 2017 Family
32 2018 Caregiving and Care Receiving
33 2018 Giving, Volunteering and Participating
34 2019 Canadians’ Safety (Victimization)
35 2020 Social Identity

Collection Methods summary data, primarily in the form of tables

Telephone interviewing is the major form of data and charts, along with initial analyses and
collection for three reasons: low cost, ease of findings. Public-use microdata files, together
monitoring interviewers, and data quality. The with supporting documentation, are avail-
sample size for each cycle of the GSS is now ap- able for secondary analyses. These files con-
proximately 25,000 households, generally one tain individual records, screened to ensure
person per household. confidentiality.

Source: Statistics Canada (2019), accessed 8 August 2020.

Availability
GSS findings form the basis of a series of pub-
lications that present national and regional

about those three seniors, as in “only 33 per cent health. It is impossible to reach such a number
of the aged are in good health”: that amounts in a small study unless the group in question is
to one person. With a national sample of 2000, the specific focus of the research.
the numbers are increased twenty-fold to 60. • Opportunity for cross-cultural (interna-
Even though this number is still small, it will tional) analysis. It’s easy to forget that many
allow a more meaningful estimate of seniors’ findings may not apply to countries other

bel65796_ch05_100-138 131 01/13/22 12:19 PM

132 PART II Quantitative Research

than the ones in which the research was con- differences where national differences were widely
ducted. Cross-cultural research can address expected.
that issue. A study of religiosity by Kelley
and De Graaf (1997) is an example. The au- • More time for data analysis. Precisely because
thors described the process as follows: data collection is time-consuming, the analysis
of the data may be rushed. It might seem that
data collection would be the difficult phase and
Data are from the 1991 “Religion” module of
the analysis relatively straightforward, but this
the International Social Survey Programme
is not always the case. Working out what to
(ISSP) … a module containing exactly the
make of the data requires considerable thought
same questions, answer categories, and se-
and, often, willingness to learn unfamiliar sta-
quencing for all countries surveyed. … The
tistical techniques. Freedom from the task of
samples are all large, representative national
collecting fresh data means that data analysis
samples of adults. The most common pro-
can be better planned and executed.
cedure is to hold face-to-face interviews …
• Reanalysis can offer new interpretations. It’s
followed by a leave-behind self-completion
easy to think that once a set of data has been
questionnaire containing the ISSP module …
analyzed, the data have in some sense been
(1997, p. 642)
used up. In fact, data can be analyzed in so
Kelley and De Graaf’s results were based on a second- many different ways that there is almost always
ary analysis of data from 15 nations. Opportunities additional insight to be gained from them. For
for such cross-cultural analysis appear to be in- example, a secondary analyst may look at rela-
creasing. For example, common core questions may tionships between variables that have not been
be used in national surveys conducted in several previously considered, prompting a reconsider-
countries. Both the US and Britain have the equiv- ation of the data’s relevance (see Box 5.11). Also,
alent of our GSS. This allowed Grabb and Curtis new methods of quantitative data analysis, with
(2004) to create a body of literature comparing the the potential of allowing different interpreta-
US and Canada, revealing cross-national regional tions of the data, are continuously emerging. As

Box 5.11 Secondary analysis and new research questions

Secondary analysis can involve topics that in all that it may be related to feelings of guilt, irreg-
likelihood were not envisaged by those responsi- ular eating patterns, and concerns about body
ble for the data collection. In a secondary anal- shape. Similarly, Bell and Kandler (2017) did a
ysis of data on dietary choices, attitudes, and secondary analysis of American twin data that
practices, Beardsworth, Keil, Goode, Haslam, & showed that political orientations, party identifi-
Lancashire (2002) added gender as a focus that cation, and interest in politics could not be fully
had not been examined in previous analyses. The explained by sociological factors, and that those
study showed that women were more likely than political characteristics were still highly heritable
men to adopt a “virtuous” pattern of eating, one after the sociological variables were taken into
in keeping with Western ethical and nutritional account. Both of these studies provide examples
principles. The article also showed that adher- of profitable extensions of prior research using
ence to these principles had a negative side in existing data.

bel65796_ch05_100-138 132 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 133

awareness of such techniques spreads, and their • The ecological fallacy. This can be a problem
potential relevance is recognized, researchers if data gathered by region or neighbourhood
can apply them to existing data sets. (such as census data) are used to make state-
• It fulfills the wider obligations of the social re- ments about individuals. The term “ecological
searcher. For all types of social research, re- fallacy” comes from the common practice of
search participants give up some of their time, acquiring data for a geographical area, such
usually for no reward. It’s not unreasonable as the crime rate in a particular neighbour-
that the public should expect the data they hood. Coleman and Moynihan (1996) pro-
provide to be mined to the fullest. Making data vided an example of this fallacy as it relates
available for secondary analysis enhances the to the relationship between ethnicity and
chances that they will be put to maximal use. crime. They observed that findings showing a
higher incidence of crime in neighbourhoods
with high concentrations of ethnic minorities
Limitations of secondary analysis had been used to imply that members of such
The foregoing list of benefits sounds almost too good minority groups were more likely than others
to be true. But there are some drawbacks to analyz- to commit crimes. However, individual data
ing data gathered by others: are needed to examine this hypothesis, which
may or may not be true. For example, it may
• Lack of familiarity with the data. With data not be the members of the minority groups
collected by others, a period of familiariza- who are responsible for the high levels of of-
tion is necessary to come to grips with a wide fending, but their non-minority neighbours.
array of variables, the ways in which they were Similarly, people could be coming from ad-
coded, and various aspects of their organiza- joining neighbourhoods to commit crimes in
tion. With large data sets this period can be a community they perceive as more vulnera-
quite prolonged. ble than their own. Group data cannot evalu-
• Complexity of the data. Some of the best-known ate these possibilities. To avoid the ecological
data sets employed for secondary analysis, such fallacy, the unit of analysis (individual, group,
as the GSS, are very large both in the numbers neighbourhood, etc.) of the data must be
of cases and in the numbers of variables they the same as the unit of analysis of the state-
contain. Sometimes the sheer volume of data ment or hypothesis. For example, to test the
can present problems, and here too a period of hypothesis that people in group X are more
acclimatization may be required. Also, some likely to commit crimes than people not in
of the most prominent data sets employed for group X, you would need data on individuals,
secondary analysis are hierarchical, meaning not neighbourhoods or regions.
that the data are collected and presented at the • No control over data quality. The point has
level both of the household and of the indi- been made on several occasions that second-
vidual (and sometimes at other levels as well). ary analysis offers researchers the opportunity
Different data may apply to each level. Thus, at to examine data of far higher quality than they
the household level, data on variables such as could collect themselves. However, this point
the number of cars may be included, while at applies mainly to reputable data sets such as
the individual level, data on income and occu- the GSS and others. With lesser-known data
pation are found. The secondary analyst must sets, more caution may be necessary with
decide which level of analysis to use; if the de- regard to data quality, although certain fun-
cision is to analyze individual-level data, the damental checks on quality are usually un-
individual-level data must then be extracted dertaken by the archives in which the data are
from the data set. deposited.

bel65796_ch05_100-138 133 01/13/22 12:19 PM

134 PART II Quantitative Research

• Absence of key variables. Because secondary of activity. Durkheim’s (1897/1952) famous

analysis entails the analysis of data collected study of suicide, for example, was the result
by others for their own purposes, one or more of a comparative analysis of official statistics
of the secondary analysts’ key variables may from several countries.
not be present, or may be measured differ-
ently in different years. Walters (2004) had this However, as mentioned in the previous chapter (see
problem in a study of incomes earned by grad- Box 4.4), some forms of official statistics can be very
uates of various kinds of post-secondary pro- misleading if they record only those individuals who
grams; therefore, he had to modify the original are processed by the agencies responsible for compil-
data. Similarly, an analyst might want to see ing the statistics, such as the police. Crime rates have
if a known relationship between two variables been a particular focus of attention and concern
holds even when other variables are taken into among critics of the use of official statistics, but other
account, but might find that those other vari- official statistics suffer from the same problem. For
ables are not part of the data set. Considering example, official unemployment statistics may mis-
more than one independent variable at a time represent the “real” level of unemployment. People
is a form of multivariate analysis, a topic we who have given up trying to find work are often
will touch on in Chapter 8. missed in the statistics, while those who work in the
“underground or informal economy” (and thus are
not actually unemployed) may be included in the
The need for survey data as a check unemployment statistics. In addition, the definition
and counterpoint to official statistics of “unemployment” used by those who compile offi-
Official government statistics are important sources cial statistics may not be exactly the same as the ones
of data for secondary analysis. Some of these sources, used by social researchers.
such as the census, are based on interviews and ques- Measurement problems may also arise if policies
tionnaires conducted by state-run organizations. regarding the phenomenon to be counted vary over
Others, such as the police data sometimes used to time. For example, if a local government decides
calculate the crime rate, are better understood as in- to devote more police resources to monitoring ac-
formation gathered by state agencies to record what tivities such as drug use, the sex trade, or driving
they have done in the process of conducting their while intoxicated, official estimates of the incidence
duties. Here we explore potential deficiencies in the of those behaviours will almost certainly increase.
latter and how survey data may sometimes provide a Moreover, during such crackdowns police officers
needed correction to them. may be less likely than usual to let perpetrators off
To be sure, official statistics offer the social re- with a warning. Zero-tolerance programs have the
searcher certain advantages: same result. This illustrates how variations over time
in official crime statistics may not be the result of
• The data are often based on whole populations variations in the frequencies of the activities tracked,
rather than samples, making it possible to but may simply reflect changes in the willingness to
obtain a complete picture of the population in expend resources on surveillance or to proceed with
question. prosecutions.
• There is a greater prospect of analyzing the Problems with official statistics may also stem
data both longitudinally and cross-cultur- from factors related to “race,” ethnicity, and class.
ally. Because the data are compiled over many For example, there may be variations in the like-
years, it’s possible to chart trends over time lihood that members of the public will report a
and perhaps to relate them to broader social crime when the perpetrator is a member of a par-
changes. As well, official statistics from differ- ticular ethnic group (Are Indigenous people more
ent nations can be compared for a specific area likely than others to be noticed and reported?),

bel65796_ch05_100-138 134 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 135

and for a variety of reasons police may be less criminology, which may be used in conjunction
inclined to investigate white-collar crime than with official police statistics. Both types of data have
street crime. their strengths and weaknesses, so a judicious use of
These deficiencies in official statistics may be both may be required to understand the full extent
offset by using other sources of data when perform- and consequences of criminal activity. Boivin and
ing secondary analysis, most notably those derived Leclerc (2016), for instance, point out that survey
from interviews and questionnaires that are based research indicates that only a small fraction of do-
on representative samples. A good example of this mestic violence cases is reported to the police, and
is the utilization of victimization surveys (such as seek to explain why by analyzing incident reports
the aforementioned GSS victimization studies) in compiled by Quebec police officers.

Research in the News

The Gig Economy in Canada
Three Canadian sociologists—Paul Glavin, Alex sample of the Canadian employed population.
Bierman, and Scott Schieman—have done re- Gig workers, who comprise about one-fifth of
search into the “gig economy” in Canada. Their the Canadian workforce, were roughly twice
investigation, done in conjunction with the as likely to show signs of loneliness (such as ex-
2019 Canadian Quality of Work and Economic pressing a lack of companionship, feeling left
Life Study, examines employment in jobs that out, or feeling isolated from other people) com-
are carried out on a per-task basis, such as pared to workers in other sectors. They also re-
those offered by companies like Uber and Skip ported higher levels of anxiety and depression.
the Dishes, and which are found in various free- Similarly, gig workers were 50 per cent more
lance occupations. likely to experience feelings of helplessness and
In an op-ed piece in the newspaper The about 40 per cent more likely to report low levels
Province (Glavin, Bierman, & Schieman, 2019), of control over their lives. The authors conclude
the authors explain that life in the gig economy is that “Canadians will soon need to grapple with
often characterized by feelings of loneliness and the question of whether the ease and flexibil-
powerlessness. They base their findings on online ity provided by the gig economy is worth these
interviews conducted with a representative [psychological] costs.”

Key Points
Survey research is one of the most widely used da- • While open questions undoubtedly have advan-
ta-gathering techniques in the social sciences. tages in survey research, closed ones are typically
Surveys include interviews, in which a researcher en- preferable. They facilitate the asking of the ques-
gages in a verbal discussion with the participant, and tions, the recording of answers, and coding. This
point applies especially to questionnaires.
questionnaires, which respondents read and fill out
• Open questions of the kind used in qualitative in-
on their own. This kind of research permits standard-
terviewing can play a useful role in formulating
ization in the asking of questions and the categoriza-
fixed-choice answers.
tion of answers. • Learning the rules of question asking will help you
avoid some serious problems.

bel65796_ch05_100-138 135 01/13/22 12:19 PM

136 PART II Quantitative Research

• Question order is very important and some gen- • The researcher-driven diary is an alternative to
eral rules should be followed. using questionnaires and interviews, especially
• Response sets are a potential problem in both when the research questions are concerned
structured interviews and questionnaires, but with specific behaviours rather than attitudes or
steps can be taken to lessen their impact. opinions.
• Always put yourself in the position of the respon- • The visual presentation of closed questions and
dent when devising questions. their general layout are important considerations
• Ensure that the survey questions generate data in designing questionnaires.
appropriate to the research questions. • Secondary analyses allow researchers to conduct
• A pilot study can identify problems in question their inquiries without having to collect new data.
formulation. • Very often, secondary analysis involves high-
• Structured interviews can be administered in quality data sets taken from large, representative
person, over the phone, or online. samples.
• It is important for interviewers to keep to the word- • Secondary analysis presents a few disadvantages,
ing and order of questions. Training in the asking such as the absence of theoretically important
of questions, the recording of answers, and how to variables.
establish rapport with respondents is essential. • Official statistics, especially those relating to crime,
• Questionnaires reduce some of the problems en- can be problematic for secondary analysis.
countered in structured interviews, but they also • The problems associated with official statistics
have weaknesses. may be offset by using survey data taken from
• Online surveys take two major forms: Web surveys representative samples.
and email surveys.

Questions for Review (R) and Creative Application (A)

Open or closed questions? a4 Name and define three kinds of response set, pro-
r1 Why are closed questions generally preferred to viding an original example of each one.
open ones in survey research?
a1 Compose a closed question to measure peo- Vignette questions
ple’s attitudes toward recycling, then list possible r5 In what circumstances are vignette questions es-
shortcomings of the question. pecially appropriate?
a5 Make up a set of vignette questions to examine an
Types of question addiction of your choice.
r2 What are the main types of question likely to be
used in a structured interview or questionnaire? Pre-testing questions
a2 Compose a closed question to measure attitudes R6 Why is it important to pre-test questions?
toward gun control, using a seven-point Likert a6 Compose an open question that could be used in
scale. a pilot study to generate closed questions on the
topic of what people find attractive in a romantic
Question order partner.
r3 Explain why question order may significantly
affect answers. Using existing questions
a3 Construct three survey questions to illustrate how r7 Why should researchers consider using questions
question order may significantly affect answers. devised by others?
Place them in two different orders, and explain a7 Assume you are doing a study on the frequency of
how placing them in a different order may make a tobacco smoking among university students, and
difference in how people respond to them. that there are already widely used questions on
this topic that have been administered to the gen-
Response sets eral public. Would it be necessary to devise your
r4 What are response sets and why is it important to own questions, or could you rely on the existing
know about them? ones? Explain.

bel65796_ch05_100-138 136 01/13/22 12:19 PM

 5 Survey Research: Interviews and Questionnaires 137

Interviews a11 If you were asked to participate in an online

r8 Why might a survey researcher prefer a struc- survey concerning cheating on university exams,
tured interview to an unstructured one? would you agree to do it? Explain why or why not,
a8 In an interview study examining attitudes toward then explain how researchers could increase the
transgender people, would it be best to use only response rate for the study.
transgender interviewers, only non-transgender
interviewers, or both? Explain.
The diary as a form of questionnaire
r12 In what circumstances is it preferable to use
Interview contexts a diary approach rather than a conventional
r9 In what circumstances is it preferable to con- questionnaire?
duct structured interviews with more than one a12 Construct five items that would be appropriate for

interviewer? a structured research diary designed to record

a9 Assume you have unlimited time and money to do people’s food intake over a 24-hour period.
an interview study on gender differences in career
aspirations. Would it be better to do in-person in-
Other researchers’ data
terviews or telephone interviews? Explain. r13 What is secondary analysis? What are its advan-
tages and disadvantages?
Questionnaires a13 Assume that you will be conducting a secondary

r10 What advantages do questionnaires have over analysis of survey data on the topic of gender dif-
structured interviews? What disadvantages do ferences in attitudes toward same-sex marriage.
they have compared to structured interviews? List three variables that may not be part of an exist-
a10 Assume your professor just asked you to fill out a ing data set on this topic, and explain why you think
questionnaire in class that will be used in a pub- they may not be included in an existing data set.
lished research paper. What ethical issues are in-
volved in this situation?
Official statistics
r14 What are the drawbacks to using official police
Online social surveys statistics to estimate crime rates?
r11 What advantages do online surveys have over a14 You are planning to do research on the personal

traditional research methods for collecting data? and social factors that contribute to intimate part-
What disadvantages do they have compared to ner violence. What problems would arise if you
traditional research methods? relied entirely on official police statistics for your
data? Explain.

Interactive Classroom Activities

1. Students work in pairs for 15 minutes to design university students’ grades?” Each group is then
an opening statement to be given to prospective paired with another group. A person in Group
participants in a study on shoplifting committed A verbally administers the survey to a person in
by university students. All relevant details are to Group B, and then another person in Group A ad-
be included—e.g., how the potential respondents ministers it to a different person in Group B, and
were selected, how confidentiality will be main- so on, until several people have been interviewed.
tained, and so on. Students are then asked to After each person is interviewed, he or she com-
write their statements on the board/screen, with ments on the experience—e.g., on whether the
the rest of the class commenting on the strengths questions were clear, whether they felt embar-
and weaknesses of the statements. rassed, whether the fixed responses adequately
2. Divide the class into an even number of small covered what came to their mind when they were
groups. Each group is to compose a 10-item asked the questions, and so on. The exercise is
structured interview schedule designed to ad- then repeated, with people in Group B interview-
dress the following research question: “Does the ing people in Group A. Once all class groups are
recreational use of marijuana have an impact on finished, the instructor leads an open discussion

bel65796_ch05_100-138 137 01/13/22 12:19 PM

138 PART II Quantitative Research

that starts with the question “What did you learn designed to measure attitudes toward a contro-
from this exercise?” versial topic (e.g., assisted suicide, the removal
3. The instructor reads each question, one at a time, of statues commemorating historical figures,
from Box 5.5. After each question is read, the class animal rights, etc.) and to gather basic socio-de-
is asked to point out its flaws. All members of the mographic data. Once that has been done, each
class are then given two minutes to come up with group puts its questions up on the board/screen,
a revised version of the question. Individual stu- and the rest of the class provides a critique. Once
dents are called on to produce their revised ver- all groups have had their turn, the groups get
sions, which are written on the board/screen. The together again and revise their questions to ad-
class is then asked to comment on whether the dress the concerns raised by their classmates.
revised versions solve all the problems contained Each group presents its questions to the class a
in the original question, or if further revision would second time, with commentary on how the previ-
be appropriate. ous weaknesses have been corrected. The class
4. The class is divided up into small groups. Each again offers a critique to each group.
group is asked to come up with 10 survey questions

Relevant Websites
This Canadian Election Study site provides access The General Social Survey—Social Identity, 2020,
to structured interviews and questionnaires and the provides the questionnaires used by Statistics
numerical data derived from them for the Canadian Canada to examine how Canadians identify with
elections of 2011, 2015, and 2019. their social and cultural environments.
[Link] [Link]
pl?Function=getSurvey&SDDS=5024
This YouTube video explains how to make an online
survey using Google Docs. The Inter-University Consortium for Political and
Social Research is an international association of
[Link]/watch?v=DEhD0m1fhAQ
about 700 academic institutions and research or-
In the General Social Survey on Time Use, 2015 ganizations that “provides leadership and training
Questionnaire, you can find an example of a re- in data access, curation, and methods of analysis
search diary that forms part of a larger structured for the social science research community.” Click on
interview. “Find Data” to access over 500,000 digital files for ex-
[Link] amples of questions used in a wide variety of social
pl?Function=assembleInstr&lang=en&Item_ research projects.
Id=217656#qb218258
[Link]/icpsrweb/ICPSR
The General Social Survey—Canadians’ Safety, 2019,
The UK Data Service provides a searchable data-
contains the questionnaires used by Statistics Canada
bank of questions used on social surveys in the UK.
to assess perceptions of our criminal justice system and
experiences of criminal victimization in Canada. [Link]
(Websites accessed 4 September 2021.)
[Link]
pl?Function=getSurvey&SDDS=4504

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch05_100-138 138 01/13/22 12:19 PM

 6 Structured Observation

Chapter Overview
Structured observation is a relatively underused method in social research. It entails the direct observation
of behaviour, which is analyzed using categories that are devised before the observation begins. This
chapter explores:
• the limitations of survey research for the study of behaviour;
• different forms of observation in social research;
• the potential of structured observation to contribute to our understanding of social behaviour;
• how to devise an observation schedule;
• different strategies for conducting structured observation;
• issues of reliability and validity in structured observation;
• field studies, in which a researcher intervenes in actual social life and records what happens as a
consequence of the intervention;
• ethical issues in the above; and
• some criticisms of structured observation.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Identify some problems with survey research that structured observation may rectify.
2. Describe the basic principles of structured observation.

▲ TommL/iStockphoto

bel65796_ch06_139-151 139 01/13/22 12:26 PM

140 PART II Quantitative Research

3. List the key points to be considered when devising an observation schedule.

4. State some important strategies for observing behaviour.
5. Explain how the issues of reliability and validity apply to structured observation.
6. Discuss how field experiments may be a form of structured observation.
7. Outline some criticisms of structured observation.

Kate has just arrived at a “meet and greet” recep- Is she the dean? Kate also notices that in the past
tion held by her university’s social sciences division as 10 minutes, the woman who might be the dean has
part of Orientation Week. The invitation said that pro- talked to seven people who appear to be profs,
fessors and even the dean would be in attendance. but only two who look like students. The people
Laid out on a long table in the centre of the room who look like students seem far more likely to be
are fruit and vegetable snacks, cheese and crackers, talking to other students than to profs. Shifting her
wine, beer, and fruit juices, as well as tea and coffee. attention to the table in the centre of the room, Kate
Being naturally curious, Kate decides that instead of notices that the middle-agers who are drinking al-
schmoozing she will take a glass of wine and make cohol have almost all chosen wine rather than beer,
her way to a deserted part of the room. and that the women prefer white to red by a factor
She notices that the people who look mid- of two to one. As a test of the latter observation, she
dle-aged tend to be talking to other people who sets the timer on her watch so she can do a formal
look middle-aged. Are they profs? A group has count for 10 minutes. To her horror, the woman who
gathered around a middle-aged woman in a busi- appears to be the dean walks all the way across the
ness suit who is holding forth about student grades room and asks her brusquely, “Are you doing some
in the social sciences compared to the humanities. kind of study?”

Introduction to some population of interest. Normally it is not

feasible to adopt that sort of methodology in a struc-
As illustrated in Chapter 5, in survey research, people’s tured observation study.
behaviour must be inferred from what they report on
the survey. But as we have noted in previous chap-
ters, such inferences can be problematic because what
Problems with survey
people say and what they do are not always the same research
thing. Structured observation offers a possible solu-
Chapter 5 dealt with several aspects of survey research,
tion to that problem by observing behaviour directly.
including the problems typically associated with it.
Explicitly formulated rules are used to make the ob-
Box 6.1 summarizes some of the main problems. Also
servations and to record the information gathered.
recall that practitioners have developed, with varying
Although structured observation may appear to
degrees of success, ways of dealing with these short-
be a logical alternative to survey research, it has not
comings or at least of reducing their impact.
attracted a large following and tends to be used only
in research settings such as classrooms, courts, and
hospitals. One reason for this is that certain types So why not just observe
of behaviour—such as criminal activity—are in- behaviour directly?
herently difficult to observe. Another reason is that
many social researchers want to generalize their An obvious solution to the problems identified in
findings, which means selecting large, random sam- Box 6.1 is not to rely on research instruments such as
ples to provide information that can be extrapolated surveys, but to observe people’s behaviour directly.

bel65796_ch06_139-151 140 01/13/22 12:26 PM

 6 Structured Observation 141

Box 6.1 Problems with using survey research to investigate behaviour

• Problems of meaning. People may vary in • Threatening or embarrassing questions.
their interpretations of key terms used in a These may lead to replies that are not fully
question or in the answers provided. Does truthful. For example, men who recently un-
“watching television” include having it on derwent surgery for prostate cancer may not
in the background while you are making want to admit to any incontinence.
dinner? • Gap between stated and actual behaviour.
• Problems of memory. Respondents may not How people say they would behave and how
remember certain aspects of their behaviour, they actually behave is often inconsistent (see
or they may have false memories of it. For Box 4.6). Husbands, for example, traditionally
example, people may underestimate the overestimate how much time they devote to
number of alcoholic drinks they consume in a housework. Perrucci et al. (2000) found two quite
typical week. different kinds of racial relations on a predomi-
• Social desirability bias. Perhaps the most nantly White university campus, behaviours they
important shortcoming of survey research is described as “front stage” and “back stage.”
that respondents will often give answers that The “back stage” behaviour, which showed less
they think will reflect well on them; for exam- than full acceptance of racial minorities, did not
ple, overestimating donations to charity or become apparent until questionnaires were
under-reporting traffic tickets. supplemented by direct observation.

Structured observation (sometimes called “system- categories could be used to compare “authoritarian”
atic observation”) is one way to do that. (See Box 6.2 and “shared” leadership styles. For example, is there
for other types of observation research.) Using this less tension and thus less need for tension relief with
method, the researcher formulates explicit rules out- an “authoritarian” leadership style, or with a “shared”
lining what behaviours are to be observed and how one? A scheme like this could also be used to help
the observations are to be recorded. Each person in leadership trainees improve their leadership skills.
the study is observed for a predetermined period of If an observer coded what was happening in the
time following rules set out in a document (similar group as a whole every 15 seconds, the coding for a
to a structured interview schedule) that is usually 12-minute period might look something like Figure
called an observation schedule. The data collected in 6.2. Notice how often a question tends to be followed
this way, like survey data, can then be treated as vari- by an answer and a negative affect by a smoothing
ables for analysis. positive affect (Bales organized his categories so that
One of the classic schedules for the observation of the codes assigned to linked behaviours such as these
small-group behaviour was developed by Bales (1951) add up to 13). In many cases, two types of leaders
(see Figure 6.1). Suppose you wanted to compare lead- emerge: a task specialist and a socio-emotional leader
ers and followers in terms of their behaviour: using who is more responsible for group members’ feelings.
Bales’s observation schedule, it would be possible to
examine the frequency with which each group asks The observation schedule
questions and answers them, talks or keeps silent, and
so forth. Similarly, age and gender differences could Devising a schedule for recording observations is
be explored in relation to Bales’s categories 11 (“Shows clearly a crucial step in a structured observation
tension”) and 2 (“Tension release”). Or the same project. The considerations are very similar to those

bel65796_ch06_139-151 141 01/13/22 12:26 PM

142 PART II Quantitative Research

1 Helps, rewards, affirms others involved in creating a structured interview schedule.

For example:
2 Tension release, jokes, shows satisfaction

3 Agrees, concurs, complies

• A clear focus is necessary, meaning that the
4 Gives direction research question needs to be clearly stated.
5 Gives opinion, evaluation The observer must know exactly who is to be
6 Gives orientation, repeats, clarifies
observed and which specific behaviours are to
be recorded.
7 Asks orientation, for repetition/clarification
• As with the production of a closed question for
8 Asks opinions a structured interview schedule, the categories
9 Asks for direction, possible ways of action of behaviour must be both mutually exclu-
10 Disagrees, withholds help sive (not overlapping) and exhaustive (every-
thing must have a category). What if someone
11 Shows tension, withdraws
knocks on the door to ask about emptying
12 Shows antagonism, deflates, defends self the trash can while the study is underway?
Perhaps the best approach is to have a category
Sum of 13: Balancing task areas (neutral affect) of behaviour coded “Other” or “Interruption.”
6 + 7 Communication Pilot studies help to reveal possible problems
5 + 8 Evaluation associated with a lack of exhaustiveness.
4 + 9 Decision • The classification scheme must be easy to use.
Complex systems listing many types of be-
Sum of 13: Balancing socio-emotional reactions
haviour may be unworkable. Much like inter-
(−) (+)
viewers using a structured interview schedule,
10 + 3 (Dis)agreement
observers need to be trained, but even a well-
11 + 2 Tension (reduction)
trained observer can become flustered or con-
12 + 1 (Dis)integration
fused if faced with too many options.
FIGURE 6.1 Small-group interaction pro- • Problems can arise if the observation schedule
cess: Observation schedule for an imagi- requires too much interpretation on the part
nary study of small-group interaction
of the observer. For example, in Bales’s scheme
Source: Adapted from Bales (1951).
it can be difficult to distinguish between a cat-
egory 10, “Withholds help,” and a category 11,
“Withdraws.” If interpretation is needed, clear
guidelines and considerable training and ex-
perience are required.

Strategies for observing

7 7 6 6 6 7 6 6 8 5 10 2
behaviour
5 8 5 12 1 9 9 4 4 5 4 4
There are different ways to observe and record
10 11 12 12 2 1 1 11 2 8 8 5
behaviours:
5 10 10 2 1 3 2 1 3 6 2 1

FIGURE 6.2 Coding sheet for imaginary study of • Recording incidents means waiting for some-
small group thing to happen and then recording what fol-
Note: Each cell represents a 15-second interval and each row represents lows from it. Essentially, this is what LaPiere
3 minutes.
(1934; see Box 4.6) did when he waited for a

bel65796_ch06_139-151 142 01/13/22 12:26 PM

 6 Structured Observation 143

Box 6.2 Other types of observation research

• Participant observation. This is one of the Canadian separatists is an example of this
best-known methods of social science re- sort of observation. If the participants are
search, especially in sociology and anthro- not aware that they are being observed
pology. It is primarily qualitative and entails and analyzed, such research can be called
the relatively prolonged immersion of the unobtrusive observation. However, the
observer in a particular social setting (for ex- people Bell observed knew that he was
ample, a group, organization, or community). studying them. Non-participant observa-
The main goal is usually to elicit the meanings tion may be either structured or unstruc-
that the people being observed attribute to tured (see below); Bell’s approach was
their environment and behaviour. Participant unstructured.
observers vary considerably in how much • Unstructured observation. As its name im-
they participate in the social settings in which plies, unstructured observation does not
they locate themselves. (For a more detailed follow formal rules for either making or re-
discussion, see Chapter 10.) cording observations. Instead, the observer
• Non-participant observation. This term records people’s behaviour in as much detail
applies to situations in which the observer as possible, and then develops a narrative
does not take part in what is going on in the account of that behaviour. Most participant
social setting. Bell’s (2007) study of western observation is unstructured.

Chinese couple to negotiate entry to a hotel or sample the times at which they make their ob-
restaurant and then noted whether they were servations. The times at which the observations
allowed in or not. are made are chosen in advance, either system-
• A wide variety of behaviours can be observed atically or randomly. An example is a study of
and recorded in either short or long periods of schools known as the ORACLE (Observational
time. In research reported in Chapter 2, chil- Research and Classroom Learning Evaluation)
dren in St Helena were videotaped over a two- project (Galton, Simon, & Croll, 1980). In this
week period during their morning, lunch, and research, eight children (four of each gender)
afternoon breaks. The tapes were then coded and their teacher were observed at regular
using “the Playground Behaviour Observation intervals. A mechanical device made a noise
Schedule, an instrument for recording the oc- every 25 seconds, signalling the observer to
currence of 23 behaviours (e.g., games; fan- record what the teachers or pupils were doing
tasy play; character imitation; anti-social and (in accordance with the observation schedule).
pro-social behaviour) and their behaviour
groupings (i.e., whether the behaviour was
undertaken by an individual, a pair, by 3 to 5 Issues of reliability and
children, or 6 or more). … A separate schedule validity
was completed for each 30-second segment”
(Charlton, Gunter, & Coles, 1998, p. 7). Compared with interviews and questionnaires,
• Time sampling is another approach to the ob- structured observation “provides (a) more reliable
servation of behaviour. Since it is usually not information about events; (b) greater precision re-
possible to observe a situation continuously for garding their timing, duration, and frequency;
extended periods of time, researchers generally (c) greater accuracy in the time ordering of variables;

bel65796_ch06_139-151 143 01/13/22 12:26 PM

144 PART II Quantitative Research

and (d) more accurate and economical reconstruc- people tend to behave differently depending on the
tion of large-scale social episodes” (McCall, 1984, context. Reliability may also be threatened by fac-
p. 277). This is a very strong endorsement, but there tors such as observer fatigue and lapses in attention.
are issues of reliability and validity to consider. Some However, observers can be trained to get highly re-
of them are similar to those faced by any researcher liable results using complex coding schedules. The
seeking to develop measures in social research (es- procedures for assessing intra-observer reliability
pecially survey research), while others are specific to across all possibilities are broadly similar to those
structured observation. used to assess inter-observer consistency.

Reliability Validity
One thing practitioners of structured observation Faulty administration of any measure can obviously
have been concerned with is inter-observer con- affect its validity. However, ensuring measurement
sistency. Essentially, this entails considering how validity—that is, ensuring that an indicator is mea-
closely two or more observers of the same behaviour suring what it is supposed to measure—can be diffi-
agree on how to code it (see Boxes 6.3 and 6.4). cult even when the measure is administered properly.
A second consideration is intra-observer consis- The second point simply means that structured
tency: the degree of consistency in the application of observation demands attention to the same kinds of
the observation schedule by a single observer over issues (assessing face validity, concurrent validity,
time. Consistency is difficult to achieve because and so on) that interviews and questionnaires do.

Box 6.3 Observing the effects of ADHD on childhood friendships

Normand and colleagues (2013) studied how at- The results indicated that children with ADHD
tention deficit hyperactivity disorder (ADHD) may committed a higher proportion of illegal ma-
affect friendships among children aged 7–13. noeuvres in the car race, and that, unlike those
Children from the Ottawa–Gatineau region were without ADHD, increased the proportion of rule
recruited to participate, about two-thirds of whom violations over time. Similarly, in the card-shar-
had a diagnosis of ADHD. The recruited children ing task, ADHD children made a greater propor-
were asked to invite their best friend to participate tion of self-centred and insensitive proposals,
as well. The research was conducted at two dif- and increased their level of insensitive propos-
ferent time periods, six months apart. One aspect als. When choosing games, children with ADHD
of the study involved structured observation of made a higher proportion of insensitive propos-
the friendship pairs engaging in three dyadic ac- als. All in all, it appeared that the ADHD children
tivities: a rule-based car race in which each child were at a marked disadvantage with regard to
controlled a toy vehicle; a sharing activity involving the behaviours associated with creating and sus-
trading cards; and a game-choice task in which taining healthy friendships.
the friendship pair had to decide on which of sev- Given the importance of high-quality
eral games offered they would play at the end of friendships for personal well-being, the re-
the session. During these interactions, observers searchers recommended a number of inter-
counted various types of behaviour that correlate ventions to help children with ADHD, including
with childhood friendship quality and satisfaction, having parents coach their children on the im-
such as the number of rule violations committed in portance of rule following, sensitivity to others,
the first task, and the number of self-centred or in- and compromise when interacting with friends
sensitive negotiation tactics used in the other two. or ­potential friends.

bel65796_ch06_139-151 144 01/13/22 12:26 PM

 6 Structured Observation 145

Box 6.4 Cohen’s kappa

Cohen’s kappa is a measure of the level of agree- The closer it is to 1, the higher the agreement and
ment between two people’s coding decisions the better the inter-observer consistency. As a
that takes into consideration the possibility that rough rule of thumb, a coefficient over .75 is con-
agreement will occur by chance. It can be ap- sidered “very good”; from .6 to .75, “good”; “fair”
plied to any coding. Much like Cronbach’s alpha ranges from .4 to .59.
(see Chapter 4), the coefficient varies from 0 to 1.

The first—error in implementation—relates to LaPiere (1934) did when he arranged for a Chinese
two matters in particular: couple to seek entry to hotels and restaurants in
order to observe the effects (although his study
• Is the observation schedule being adminis-
was an imperfect experiment because there was
tered as directed? This is the equivalent of
no control group of non-Chinese people). In a field
ensuring that structured interviewers follow
experiment—unlike most types of structured ob-
instructions exactly. Variability between ob-
servation—participants do not know that they are
servers or over time makes the measure unre-
being studied.
liable and therefore invalid. Observers should
Another famous field experiment was con-
have a thorough understanding of how the
ducted by Rosenhan (1973). He was one of several
observation schedule is to be implemented.
people in the study who posed as patients and were
• Do people change their behaviour when they
admitted to mental hospitals in the US. Each pseu-
know they are being observed? If they do, that
do-patient was instructed to claim that he or she
would constitute one of several kinds of “reac-
was “hearing voices,” and all except one of them
tive effect” that can occur in various forms of re-
were then diagnosed with schizophrenia. As soon
search (see Box 6.5). If people adjust the way they
as they were admitted, the pseudo-patients were
behave because they know they are being ob-
instructed to cease exhibiting any symptoms. Even
served (perhaps because they want to be seen in a
though they were all “sane,” it took many of them
favourable light by the observer), their behaviour
quite a long time to be released. The length of hos-
would have to be considered atypical or not au-
pitalization reportedly varied from 7 to 52 days,
thentic. However, as McCall (1984) noted, there
is evidence that participants become accustomed with an average of 19 days. Rosenhan noted that in
four of the hospitals, pseudo-patients approached
to being observed and that researchers tend to
psychiatrists and nurses with a request for re-
become less intrusive the longer they are in the
lease and recorded the nature of the response: 71
field. However, these effects are likely to persist
per cent of psychiatrists moved on with their heads
when the behaviour under observation involves
averted and 88 per cent of nurses did likewise. To
sensitive issues such as sexuality or deviance.
Rosenhan this indicated that a mental patient can
become powerless and depersonalized. Generally,
Field experiments as a form in this sort of research, opportunities to employ an
of structured observation observation schedule are limited because excessive
use will blow the observer’s cover, although this
A field experiment is a study in which a researcher did not appear to be a major obstacle in this case.
directly intervenes in a natural setting to observe Rosenhan’s study is highly controversial, in
the consequences of that intervention. That’s what part because of its use of deception, the ethical

bel65796_ch06_139-151 145 01/13/22 12:26 PM

146 PART II Quantitative Research

Box 6.5 Reactive effects in social research

rawpixel/123rf

What sort of challenges might you face as a researcher who is observing this class?
Might the class behave differently because of your presence?

Webb, Campbell, Schwartz, & Sechrest wrote because the workers were aware of being ob-
about “reactive measurement effect” (1966, served and analyzed, not because of the ex-
p. 13). It occurs when someone’s awareness perimental manipulation itself: this is known
that they are participating in research leads to as the “Hawthorne effect.” A well-known gen-
changes in their behaviour that confound the eral problem in experimental studies is that
investigator’s data. Response sets, social de- some individuals will seek out cues about the
sirability bias, and political correctness can all aims of the research, and then adjust what
cause reactive effects. If a respondent reacts they say and do to conform with what they
to some characteristic of the interviewer (e.g., believe to be the purpose of the study. That is
gender or ethnicity), that is another example of a problem, even if such participants are mis-
a reactive effect. Here are a few more: taken about the aims of the research.
• The Hawthorne effect. The classic stud- • Role selection. Webb et al. argued that par-
ies illustrating this effect were done at the ticipants may be tempted to adopt a partic-
Hawthorne Works (located near Chicago) ular kind of role in research. Some people
in the 1920s (see Roethlisberger & Dickson, with no strong opinion on a particular item
1939). A series of quasi-experiments showed may develop one when asked, in order to
that each time working conditions improved appear knowledgeable. Even telling them
(e.g., through improved lighting), productivity how important they are to the research
went up, at least temporarily. But it also went can encourage some display of “expertise.”
up after working conditions deteriorated! It Other, fairly opinionated individuals may
appeared that performance improved simply back off a bit, scared by the audio recorder,

bel65796_ch06_139-151 146 01/13/22 12:26 PM

 6 Structured Observation 147

perhaps, or not wanting to come on too Reactive effects are likely to occur in any re-
strong to a stranger. search in which participants know they are the
• Researcher presence as a change agent. The focus of investigation. Webb et al. called for
simple fact that the researcher is present may greater use of what they called unobtrusive
change the situation that is being observed. measures: non-reactive methods that leave
For example, if an observer is sitting in a participants unaware that they are being stud-
classroom, that means a chair is being occu- ied. Records can be checked at hospitals and
pied that otherwise might have been empty, courts, for instance, removing the reactivity, but
which in turn may cause changes in space and for ethical reasons permission must first be ob-
privacy for those sitting nearby. When Whyte tained, and getting such permission may be dif-
(1955), in his study of “Cornerville,” joined one ficult or impossible.
of the neighbourhood’s bowling teams, he in Alternatively, the usual tools of social re-
effect removed one of his research partici- search can be used. A confidentiality guarantee
pants from the team and turned that person backed up by reminders may help, as may anon-
into a rival on an opposing team. ymous questionnaires and telephone (rather
• Trying to help or to be nice to the researcher. than in-person) interviews, although these have
Although it is often assumed that research their own downsides. It also helps to phrase
participants are always passive, this is not questions so that they are as non-threatening
true, even in experiments. Some participants and non-judgmental as possible; for example:
may try to please the researcher. An old saying “Many people did not have time to vote in the
holds that if you ask friends to do push-ups, referendum or could not make up their minds.
they will ask “what?” or “why?” But people in an Did you vote?” If you want to allow more men to
experiment (especially students, who are often admit they fear being out late at night, don’t ask
asked to be participants) will ask “where?”! In how afraid they are. Use a vignette: “John does
fact, many participants will try to do what they not like to walk alone at night in his neighbour-
think they are supposed to do, although a few hood. There is a lot of crime there, and police
may deliberately do the opposite. Some will officers are rarely if ever seen. Is John almost a
even ask “Was I good?” at the end of the ex- copy of you, a lot like you, a bit like you, not at all
periment. Such behaviour can be problematic like you?”
if it affects the phenomenon being studied. Finally, although it is good to be aware of the
• Different researchers may elicit different shortcomings resulting from reactive effects,
reactions. Some may get the data they want remember that no study is perfect, and that an
by communicating, often unconsciously, their imperfect study can still be an important source
expectations. This is why experiments are of knowledge. Also, maintain the same level
supposed to be double blind, so that no one of skepticism when the results you encounter
(not even the person administering the exper- please you as when they displease you. For ex-
iment) knows certain crucial details, such as ample, research showing that university men are
which participants have received the treat- more violent than university women in hetero-
ment and which have not (single blind means sexual romantic relationships should be held to
that only the people being studied are kept the same methodological standards as research
in the dark). Here, hiring others to collect the showing no gender differences (for example,
data can help. Katz, Kuffel, & Coblentz, 2002).

bel65796_ch06_139-151 147 01/13/22 12:26 PM

148 PART II Quantitative Research

Research in the News

The effect of social trust on Canada’s reaction to the COVID-19 pandemic
collective action,” Wu explained in a CBC inter-
view (Weber, 2020). In a pandemic, trust in fellow
citizens, politicians, the health-care system, and
perceived outsiders could be affected, he main-
tained. Without a high level of social trust, it’s
hard to get people to conform to public health
sockagphoto/Shutterstock

directives such as those regarding physical dis-

tancing and mask wearing.
Both Wu and Kennedy are embarking on
social research programs involving surveys in an
effort to determine what effect the pandemic has
As a researcher choosing between structured
had on social trust in Canada. So far, their results
observation and survey-based measures of be-
suggest that levels of social trust have remained
haviour, which method would help you better
understand the intentions behind participants’ high during the crisis. “By and large, Canadians
social distancing behaviours? are reporting taking a lot of the steps that they’ve
been called upon to take,” Kennedy explained. He
also reported that Canadians are more trusting of
Public health during the COVID-19 pandemic public officials than is the case in other countries.
may depend on things that go beyond discover- “Canadians are more likely to follow social dis-
ies in medical science such as treatment proto- tancing requirements. People perform better in
cols and vaccines. According to Cary Wu, a York terms of complying with policy matters,” he said
University sociologist, and Eric Kennedy, a his- (Weber, 2020). Both Wu and Kennedy are hoping
torian at York, social trust must also be present that their research will help those in charge of
for a nation to cope with the global epidemic. “If promoting public health in Canada to enhance
people do not trust, it’s very difficult to promote levels of social trust as the pandemic wears on.

Methods in Motion Applications to Canadian Society

Systematic social observation of the police
Schulenberg (2014) describes her experiences and the personal characteristics of the people ob-
in using systematic social observation (SSO) to served affect the phenomenon of interest. Since
study decision-making processes and the use of SSO is a method that yields both quantitative
discretion among members of a Canadian re- and qualitative data it has the potential to pro-
gional police service. With this method, structured vide highly meaningful information on a topic, but
observation is employed to gather behavioural it is rarely used because of the heavy burdens it
information on the people studied, while partic- places on the researcher.
ipant observation techniques are used to collect In Schulenberg’s case, she was interested
field notes and develop narratives that provide in how police officers dealt with the anti-social
detailed descriptions of how situational factors and criminal behaviour they encountered while

bel65796_ch06_139-151 148 01/13/22 12:26 PM

 6 Structured Observation 149

on patrol. To research her topic she went on 74 The narratives included information on things
“ride-alongs” with officers, which allowed her to like whether the citizens had weapons or showed
collect information on the 406 police–citizen in- signs of mental illness, and how the police took
teractions that took place in the 637 hours she action to deal with the situation.
spent in the field. A comprehensive observation The “SSO of police work,” Schulenberg writes,
schedule was used to record the behaviours she “is not for the faint of heart. Beyond witness-
witnessed. For example, a number of codes were ing human nature, dead bodies, and citizens in
used to describe the role played by the citizens in crisis, SSO requires a high level of commitment,
the encounter: 1 indicated that a person was a involvement, and personal sacrifice” (p. 308).
complainant or victim; 2, that they were a suspect She found herself immersed in police culture,
or person of interest; 3, that they were a disputant and had to contend with a feeling that she may
whose status as a victim or a suspect was unclear, have become overly sympathetic to the police
and so on. Informal debriefing sessions with offi- and their beliefs and values. At the same time,
cers were used to create detailed descriptions of she found some aspects of what she experienced
how the officers interpreted each citizen encoun- to be highly objectionable. Her status as a single
ter and how they felt about it afterward. In addi- woman made her the target of sexual proposition-
tion, narratives were created that provided a thick ing and rude gestures, which she either ignored or
description of the citizens’ interactions with police. deflected with choice comebacks of her own.

implications of which were discussed in Chapter 3. variables and categories can be identified
There have also been questions about the accu- in advance.
racy of Rosenhan’s descriptions of how the pseu- • Because it concentrates on directly observable
do-patients were treated, and about other critical behaviour, structured observation in itself
details presented in his account of the study (see cannot get at the intentions behind behaviour.
Cahalan, 2019). When intentions are of interest, observers
must impute them. Essentially, the problem
Criticisms of structured is that structured observation cannot tell the
observation observer very much about the meanings that
people attach to their behaviours—informa-
Some problems with structured observation have tion that according to Weber (1947) is crucial
already been mentioned in our discussion of its re- in understanding human thought and action.
liability, validity, and generalizability. Other criti- In a similar vein, contextual factors such as
cisms include the following: events in the larger world may be given insuf-
ficient attention as possible influences on the
• There is a risk of imposing an inappropri- observed behaviour.
ate observation schedule on the setting • There is a tendency for structured observa-
in question; this risk is especially great in tion to generate many small bits of data. As a
settings about which little is known. This result, it can be difficult to find general themes
problem is similar to that of closed-question that illustrate the bigger picture behind the
answer categories that do not fit the ques- scattered bits of data. This problem is com-
tion format. One solution is to precede the pounded by the fact that observers may not
structured observation with a period of un- know the meanings the actors attach to their
structured observation, so that appropriate behaviours.

bel65796_ch06_139-151 149 01/13/22 12:26 PM

150 PART II Quantitative Research

On the other hand . . . asking people to report their behaviour in a survey.

The previous section made it clear that there are Structured observation may work best when accom-
limitations to structured observation. However, panied by another method, especially one that can
it’s also important to remember that when overt probe for the reasons behind people’s behaviours.
behaviour is the focus of analysis and issues of One such method is the systematic social observa-
meaning are less salient, structured observation is tion approach described in the Methods in Motion
almost certainly more accurate and effective than box on pages 148–149.

Key Points
• Structured observation is an alternative to sur- • Reactive effects must be taken into account, but
vey-based measures of behaviour. should not be exaggerated.
• It involves explicit rules for recording behaviour. • The field experiment is a form of structured obser-
• Structured observation has generally been ap- vation, but can involve ethical challenges.
plied to a narrow range of behaviours, such as that • Problems with structured observation include de-
occurring in schools, courts, and hospitals. termining the meaning that people give to their
• It shares with survey research many common behaviour and ensuring that the framework for
problems concerning reliability, validity, and recording the observations is valid.
generalizability.

Questions for Review (R) and Creative Application (A)

Observing behaviour a3 Explain how time sampling could be used to make
r1 Under what circumstances might structured ob- observations of people as they interact while
servation be a better way to gather data than parking their cars in a busy parking lot.
questionnaires or structured interviews? Explain.
a1 You want to observe children in an elementary Issues of reliability and validity
school playground to examine the extent to which r4 How do the considerations of reliability and valid-
they exhibit aggressive behaviours and the con- ity in structured observation mirror those encoun-
ditions that may lead to aggression. Explain the tered in survey research?
pros and cons of using structured versus unstruc- a4 What are reactive effects? How might reactive ef-
tured observation to conduct this research. fects be problematic in structured observation re-
search that is used to examine how parents treat
The observation schedule their children while they shop in a large depart-
r2 What goals and considerations should be kept in ment store?
mind when constructing an observation schedule?
a2 Construct an observation schedule containing five Field experiments as a form of structured
items that could be used to observe fan behaviour observation
at a hockey game, and a second schedule to be r5 What are field experiments and what ethical con-
used for observing people attending a figure cerns do they pose?
skating competition. Then explain how and why a5 What ethical issues should you consider if you do
they are different. a field experiment that involves posing as a con-
fused traveller who does not speak English or
Strategies for observing behaviour French at a busy Canadian airport?
r3 Discuss three different strategies for observ-
ing and recording behaviour in structured
observation.

bel65796_ch06_139-151 150 01/13/22 12:26 PM

 6 Structured Observation 151

Criticisms of structured observation march. Your observation schedule includes cat-

r6 Can structured observation ever reveal the in- egories of behaviour that are peaceful in nature,
tentions that lie behind the observed behaviour? and others that are violent. Explain why in this
Explain. study it may be crucial to consider the intentions
a6 Assume you are using structured observation to that lie behind the behaviours observed.
do a study of people participating in a protest

Interactive Classroom Activities

1. Divide the class into small groups. Each group is 2. The instructor gives a brief synopsis of the field
to come up with a detailed mock proposal for a experiment conducted by Rosenhan (1973). The
structured observation study that involves gain- class is then divided into three groups. Group 1
ing access to a specific location, group, or orga- must argue (a) that the study was ethical and (b)
nization, for example, a mental hospital, school that the methodology used was superior to any
playground, casino, and so on. The proposal must other methodology that could have been used
include a discussion of the following: to conduct the study, citing specific examples of
a. A specific research question that the study other methodologies that would have been less
seeks to answer, a draft of which is to be effective. Group 2 must argue (a) that the study
produced. was not ethical and (b) that other methodologies
b. Whether the research will be overt or covert, would have worked equally well or better, citing
and the ethical implications of that aspect of specific examples of those alternative method-
the study. ologies. Group 3 makes the case that there is
c. How access to the location, group, or organi- no methodology that can be ethically applied
zation will be acquired. to Rosenhan’s topic, and that therefore no one
d. The formal observation schedule that is to be should be allowed to conduct a study on it. After
used, a draft of which is to be produced, along some time for preparation, group 1 presents its
with an explanation of how it is to be adminis- case to the rest of the class, followed by groups
tered and how issues of reliability and validity 2 and 3. Once all groups have presented, each
will be taken into account. group is given some time to meet to assess and
e. What findings are expected from the study. critique the other groups. Each group then pres-
ents its critique to the rest of the class, with a gen-
Each group presents its mock proposal to the rest eral discussion to follow.
of the class for comment and critique.

Relevant Websites
This is a short YouTube video of structured The Nature website offers a review of a book
­observation being done on children. (Cahalan, 2019) that is critical of the Rosenhan
experiment.
(Websites accessed 4 September 2021.)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch06_139-151 151 01/13/22 12:26 PM

 7 Quantitative Sampling

Chapter Overview
Sampling principles are not exclusively the concern of survey and other quantitative researchers. The
qualitative work you will read about in Part III also involves working with samples, but the nature of
qualitative sampling is often rather different from the quantitative kind, so a discussion of the former has
been placed in that part of the book. This chapter explores:
• the role of sampling in relation to the overall process of doing quantitative research;
• why generalizable findings (that is, findings that can be applied to some larger population) require a
representative sample;
• the idea of a probability sample: one using a random selection process;
• the main types of probability sample: simple random, systematic, and stratified random, and the
multi-stage cluster sample;
• the main issues involved in deciding sample size;
• different types of non-probability sampling, including quota sampling, which is widely used in market
research;
• potential sources of sampling error in survey research;
• sampling in structured observation research, where times and contexts are sampled; and
• raising response rates.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Define the key terms used in sampling.

▲ Creative Touch Imaging Ltd./Alamy Stock Photo

bel65796_ch07_152-176 152 01/13/22 12:26 PM

 7 Quantitative Sampling 153

2. Identify three common sources of bias in sampling.

3. Explain how four types of probability sample (simple random, systematic, stratified random, and multi-
stage cluster) are selected.
4. Outline the main characteristics of a probability sample.
5. Summarize the principal criteria used in deciding how large a sample should be.
6. Illustrate how three kinds non-probability sample (convenience, snowball, and quota) are used.
7. Cite some of the limitations involved in generalizing findings from both probability and non-probability
samples to the population of interest.
8. Describe how sampling may be used in content analyses.
9. Enumerate some ways to increase response rates.

Do you like to spend more time and money than is of building a house on a limited budget. Fortunately,
actually necessary to accomplish something? Most this is not necessarily the case with social research.
people would say no, including social researchers, With proper sampling techniques you can vastly
who inevitably face the prospect of doing their work reduce the time and resources needed to complete
with limited time and physical resources. Ideally, the research without seriously compromising the
researchers would study huge populations, such findings. You might think of sampling as “how to
as all women in Canada or all adults living in the do a lot with relatively little.” Under the right con-
Atlantic provinces, but the costs involved make that ditions, you can even calculate the probability that
impossible. In other areas of life, reducing the time sample-based statistical findings are in error. This
and money you devote to a project normally means chapter introduces you to the techniques that make
a reduction in the quality of the final product—think this possible.

Introduction It would be a huge undertaking for you to arrange

for all of them to fill out a questionnaire, and the
The term “sampling” refers to the selection of a subset time and energy required to interview 30,000 people
of a population for research. Since only in very rare would be prohibitive. It’s almost certain that a sample
instances can an entire population be studied, sam- of students would have to suffice. In fact, the need to
pling has an impact on virtually all the research dis- sample is almost universal in quantitative research.
cussed in this book. As we will see below, how the But would any sample do? What if you gave ques-
sampling is conducted determines what information tionnaires to everyone in your classes? Or if you
is acquired and influences the researcher’s interpre- stood in a busy location on campus and then inter-
tation of the results in important ways. viewed anyone who agreed to speak with you?
The key terms used in sampling are explained If your goal is simply to get some experience in
in Box 7.1. We’ll begin with a discussion of sam- data gathering and analysis, or to conduct a pilot
pling in studies involving structured interviews or study, these samples would be fine. However, if the
­questionnaires, and then consider the techniques object is to get a representative sample, one that can
used for other methods of quantitative research. be used to make inferences about all 30,000 students,
Finally, ways to increase response rates are presented. these samples would not do. Why? There are various
Suppose you are wondering about your fellow stu- reasons, but the following stand out:
dents’ attitudes on certain matters, or perhaps about
their behaviours or backgrounds. To examine any of • If you distribute questionnaires in your classes,
these areas, either structured interviews or question- anyone not taking a course with you will be
naires would be appropriate. However, let’s say that excluded. How many music or engineering
there are around 30,000 students at your university. majors take sociology?

bel65796_ch07_152-176 153 01/13/22 12:26 PM

154 PART II Quantitative Research

Box 7.1 Basic terms and concepts in sampling

• Element or unit: A single case in the popu- • Probability sample: A sample selected using
lation. In the social sciences that element or a random process such that each unit in the
unit is usually a person, but many other things population has a known chance of being
can be sampled as well: nations, cities, re- selected. The aim of probability sampling
gions, schools, firms, and so on. Finch and is to keep sampling error (see below) to a
Hayes (1994), for example, based part of their minimum.
research on a random sample of wills of de- • Non-probability sample: A sample selected
ceased people. using a non-random method. Essentially, this
• Population: All the cases about which you implies that some units in the population are
are seeking knowledge, or all the cases to more likely to be selected than others.
which your conclusions are meant to apply. • Sampling error: An error of estimation that
For example, if you are studying voting be- occurs if there is a difference between the
haviour, the population might be all the characteristics of a sample and those of
people in a particular jurisdiction who are the population from which it was selected.
eligible to vote. If the topic is hyperactivity Sampling error can occur even when a
among young children, the population might random method is used. For example, if valid
be all children aged 2 to 11. (Note that “pop- measures indicate that 45 per cent of a na-
ulation” in this context means something tional probability sample is in favour of more
different from “population” in the everyday public funding for the arts, but the actual
sense.) figure in the population is 49 per cent, the dif-
• Sampling frame: The list of elements from ference is caused by sampling error.
which the sample will be selected. • Non-response: A situation that occurs when-
• Sample: The elements selected for investiga- ever some unit selected for the sample re-
tion, a subset of the population. The method fuses to participate in the study, cannot be
of selection may involve probability or contacted, or for some other reason does not
non-probability sampling (see next column). supply the required data.
• Representative sample: A sample that is a • Census: Data collected from all elements in
microcosm of the population, one that “rep- the population rather than from a sample.
resents” its essential characteristics. A sample The phrase “the census” typically refers to the
such as this is most likely to be selected when enumeration of all (or nearly all) members of
a probability sampling process is used (see the population of a nation-state—that is, a
next). national census.

• If you go out looking for people to interview, the factors that could have an impact on your
your sample will exclude anyone who happens selection of prospective respondents.
to be somewhere else at that particular time.
Some students will be at work, others will have The problem with these sampling methods is that
skipped classes that day, and many will have a they make the choice of possible participants depen-
timetable different from yours. Some students dent on non-universal criteria: availability in one
may never need to go where you are. case and personal judgment in the other. In the lan-
• Your decisions about which people to ap- guage of sampling, this means that the sample would
proach may be influenced by your judgments be biased: it would not represent the population
about how friendly or cooperative they appear from which it was selected. It is extremely difficult to
to be. Gender, “race,” and age are only a few of remove bias altogether and get a truly representative

bel65796_ch07_152-176 154 01/13/22 12:26 PM

 7 Quantitative Sampling 155

sample, but steps should be taken to keep it to a find out whether non-respondents reflect the entire
minimum. student body in terms of faculty of registration.
Three sources of bias can be identified: However, it is usually impossible to check charac-
teristics that go beyond basic demographic catego-
• Not using a random method to pick the ries, such as attitudes toward separatism or patterns
sample. Essentially, a random method is one of smoking behaviour, because population data are
in which each element has the same chance generally not available on these topics.
of being selected, like numbers from a lottery
cage (although in some circumstances that Sampling error
principle may be violated for technical rea-
sons; for example, in order to get a sufficiently To appreciate the significance of sampling error
large number of people from a small subgroup for achieving a representative sample, consider
into the sample). As we saw above, if a random Figures 7.1 through 7.4. Imagine a population of 200
method is not used, there is a risk that the se- people and a sample of 50. Imagine as well that the
lection process will be affected by human judg- topic is whether people watch reality TV programs,
ment, so that some members of the population and that the population is divided equally between
are more likely to be selected than others. This those who do (100) and those who do not (100). If the
source of bias is eliminated through the use of sample is representative, the sample of 50 should also
probability sampling, described later in this be equally split in terms of this variable (see Figure
chapter. 7.1). If there is a small amount of sampling error, so
• The sampling frame or list of potential partic- that the sample contains one person too many who
ipants is inadequate. If the sampling frame ex- does not watch reality TV and one too few who does,
cludes some cases or is otherwise inaccurate, the distribution would look like Figure 7.2. Figure
the sample derived from it may not represent 7.3 shows a more serious over-representation in the
the population, even if a random sampling sample of people who do not watch reality programs.
method is employed to select participants This time there are three too many who do not watch
from it. them and three too few who do. Finally, Figure 7.4
• Some people in the sample refuse to par- shows a very serious over-representation of people
ticipate or cannot be contacted—in other
words, there is non-response. The problem Watch reality TV Do not watch reality TV
with non-response is that those who agree to
participate may differ from those who refuse
or cannot be reached in ways relevant to the
subject matter being investigated. If that is the
case, the sample may not be representative. For
example, if you are studying rates of criminal
victimization and people who have been vic-
tims of violent crime decline to participate,
that would skew the results.

If the data are available, a researcher can check to see

how non-respondents differ from the population. It
is often possible to do this in terms of characteristics
such as gender or age, but these are not the only po-
tentially relevant factors; in the case of a sample of
university students, for example, you might want to FIGURE 7.1 A sample with no sampling error

bel65796_ch07_152-176 155 01/13/22 12:26 PM

156 PART II Quantitative Research

Watch reality TV Do not watch reality TV Watch reality TV Do not watch reality TV

FIGURE 7.2 A sample with very little sampling error FIGURE 7.4 A sample with a lot of sampling error

the latter is the most probable single result. Ask your

Watch reality TV Do not watch reality TV
instructor to take a few minutes for an in-class ex-
periment in which a coin is flipped 20 times, sev-
eral times over, and you’ll see the surprisingly large
number of outcomes that are not 10Heads/10Tails.
However, probability sampling stands a better
chance than non-probability sampling of mini-
mizing sampling error, so it’s very unlikely that the
sample will end up looking like Figure 7.4 when it
is used. Moreover, probability sampling allows you
to employ tests of statistical significance that will
permit you to make inferences about the population
from which the sample was selected, with a known
probability of error. This topic will be addressed in
Chapter 8.

FIGURE 7.3 A sample with a fair amount of Types of probability sample

sampling error
Imagine a study on the social variables related to al-
who do not watch reality TV: 35, as opposed to 25 if cohol consumption, to be done at a university that
the sample were perfectly representative. has 9000 full-time students. This hypothetical exam-
It’s important to appreciate that the possibility ple will be used in the illustrations that follow.
of sampling error can never be completely elimi-
nated. Even with a well-crafted probability sample, Simple random sample
a degree of sampling error is likely to creep in, just The simple random sample is the most basic form
as flipping a coin 100 times is likely to yield more of probability sample. With it, each unit of the pop-
49Heads/51Tails and 51Heads/49Tails results com- ulation has an equal probability of inclusion in the
bined than 50Heads/50Tails outcomes, even though sample. Suppose there is enough money to interview

bel65796_ch07_152-176 156 01/13/22 12:26 PM

 7 Quantitative Sampling 157

450 students at the university. This means that the are contacted by someone associated with the
probability of inclusion in the sample is study.

450
, that is, 1 in 20 Step 5 mentions the possible use of a table of random
9000 numbers, found in the back of many statistics books
This is known as the sampling ratio and is ex- or generated by computer software. Such tables are
pressed as usually made up of columns of five-digit numbers;
n for example:
N
09188 08358
90045 28306
where n is the sample size and N is the population 73189 53840
size. 75768 91757
The key steps in devising a simple random sample 54016 87941
are as follows:
The first thing to notice is that, since these are five-
1. Define the population. Here N is all full-time digit numbers and the population size is just 9000
students at the university: 9000. (a four-digit number), the only random numbers
2. Select or devise a comprehensive sampling that might be appropriate are the two that begin with
frame. In this case, let’s assume that (a) the zeros, 09188 and 08358 (and the former is already
office that keeps student records has a list of larger than the largest possible case number for this
all full-time students and (b) you have access study). To deal with this, drop one of the digits in
to it (and all proper ethics protocols have been each number—say the first digit. Using only the last
followed). four digits gives us the following:
3. Decide on the sample size (n) you will use; here
n is 450. 9188 8358
0045 8306
4. List all the students in the population and 3189 3840
assign them consecutive numbers from 1 to N, 5768 1757
here 1 to 9000. 4016 7941
5. Using a table of random numbers, or a com-
puter program that can generate random num- Still, one of the resulting numbers—9188—exceeds
bers, select n (450) different random numbers 9000; no student has that case number. The solu-
from 1 to N (9000). tion is simple: ignore that number and go on to the
6. The students who match the n (450) random next one. This means that the student assigned the
numbers constitute the sample. Two points number 45 will be the first to be included in the
are worth noting here. First, there is almost sample. Moving down the first column, the student
no opportunity for bias, since in this case assigned the number 3189 will be next; the student
the selection process is entirely random: stu- assigned 5768 next; and so on.
dents are not selected on the basis of subjec- To avoid interviewing the same person twice,
tive criteria such as whether or not they look ignore any random number that appears more than
friendly and approachable. Second, selection once and continue down the list. This procedure pro-
does not depend on availability: students do duces what is called a simple random sample without
not have to walk by the interviewer in order replacement: no number is placed back in (replaced
to be included in the sample. The process of after having been selected) for a second chance at
selection takes place without their knowl- inclusion. Since virtually all simple random sam-
edge, and they do not know they will be ples in social research are like this, the qualifier
asked to be part of a social survey until they “without replacement” is usually omitted. Strictly

bel65796_ch07_152-176 157 01/13/22 12:26 PM

158 PART II Quantitative Research

speaking, “without replacement” means that the end in 6, and no even-numbered prefixes will qual-
result is not a true simple random sample, as those ify, such as 26, 86, or 166.
chosen later have a greater chance of being selected
than those selected earlier. (The first person chosen Stratified random sampling
has one chance in 9000 of being selected; removing In the imaginary study of university students dis-
that person means person two has one chance in the cussed earlier, the student’s discipline may be rele-
remaining 8999. Thus, the 450th person chosen will vant to what you will be researching. Generating a
have one chance out of the 8551 people remaining; simple random sample or a systematic sample may
a chance of one in 8551 is greater than a chance of yield one in which, for example, the proportion of
one in 9000.) Researchers generally overlook this kinesiology students in the sample is the same as
problem, since it is a small one compared with other that in the student population, but usually it is not an
sampling problems. exact match. Thus, if there are 1800 students major-
ing in kinesiology, a sampling ratio of 1 in 20 should
Systematic sample produce 90 students in the sample from this faculty.
A systematic sample is selected directly from the But because of sampling error there might be, say,
sampling frame, without using random numbers. 85 or 93.
In the present case, 1 student in 20 is to be se- So long as data are available on each student’s fac-
lected (450/9000). That “1 in 20” figure is called the ulty of registration, it’s possible to ensure that stu-
sampling interval. A random start begins the process, dents are exactly represented in terms of their faculty
which in this case can be achieved by selecting at membership by using stratified random sampling.
random a number from 1 to 20, possibly by using the In the language of sampling, this means stratify-
last two digits in a table of random numbers. With ing the population (dividing it into subgroups) by a
the 10 random numbers above, the first relevant one criterion (in this case faculty) and selecting either a
is 54016, since it is the first one where the last two simple random sample or a systematic sample from
digits yield a number in the desired range (1–20), each of the resulting strata. In the present example,
namely 16. This means that the sixteenth student five faculties would mean five strata, with the sample
in the sampling frame is the first one to be included in each stratum being one-twentieth of the total for
in the sample. Thereafter, take every twentieth stu- each faculty, as in Table 7.1, which also shows a hy-
dent on the list; the sequence of case numbers would pothetical outcome of a simple random or system-
be: 16, 36, 56, 76, 96, 116, and so on. Who is the last atic sample, one that does not exactly mirror the
person in the sample? Case number 8996. population.
In systematic sampling, it is important to ensure The advantage of stratified sampling is clear: it
that there is no inherent ordering or pattern in the ensures that the sample is distributed in the same
sampling frame, a feature called periodicity. For ex- way as the population in terms of the stratifying
ample, imagine that the sampling frame was set up criterion. Using a simple random or systematic sam-
such that case 1 was male, case 2 female, case 3 male, pling approach may result in a distribution exactly
case 4 female, and so on for all cases. In the exam- like that of the stratified sample, but it is unlikely. On
ple above, that would yield a sample of 450 females the other hand, this strategy requires that the rele-
and 0 males, in all likelihood a very unrepresentative vant criteria for stratification be known in advance
sample. If there is any pattern to the list, arrange the of the research, which is not always the case.
cases in random order or choose a different sampling Two more points are relevant here. Stratified sam-
method. pling is feasible only when it is relatively easy to iden-
Note that with systematic sampling, not every tify and allocate units to strata. Five strata based on
possible combination of cases has an equal chance physical activity level, from “competitive athletes”
of being selected. Once that first case number (here at one end to “couch potatoes” at the other, would
16) is chosen, every case number in the sample must be very hard to set up, because it would require an

bel65796_ch07_152-176 158 01/13/22 12:26 PM

 7 Quantitative Sampling 159

TABLE 7.1 The advantages of stratified sampling

Hypothetical simple random
Faculty Population Stratified sample or systematic sample
Humanities 1800 90 85
Social sciences 1200 60 70
Natural sciences 2000 100 120
Business 1800 90 84
Engineering 2200 110 91
TOTAL 9000 450 450

initial study just to get the population data on activ- interview 500 randomly selected students at each
ity level that you would need to create these strata. of the 10 universities selected, to create a sample
Second, more than one stratifying criterion can be of 5000 students. (This would require “probability
used at the same time. In our example, we could proportionate to size” sampling because each uni-
stratify by faculty and gender, or faculty, gender, and versity has a different number of students, but that
whether students are undergraduates or postgradu- technique goes beyond the scope of our discussion
ates, and so on, as long as the criteria are easy to use here.)
and relevant to the research question. But this ap- Suppose the result of sampling 10 universities is
proach is practicable only when data on the relevant the following:
stratifying criteria are available for the population.
Does your school have an accessible list of female un- • Alberta
dergraduate engineering students? • McGill
• Simon Fraser
Multi-stage cluster sampling • Winnipeg
In the previous example, the students to be inter- • Dalhousie
viewed all attend the same university. However, • Guelph
imagine a national sample of students. A major • Brandon
problem with large populations is the general lack • Western
of an adequate sampling frame. Is there a list some- • Nipissing
where of all students registered at Canadian univer- • Toronto
sities? No. Without a sampling frame, it’s impossible
to select a simple random or systematic sample. Note the absence of universities in Saskatchewan,
One way to deal with these problems is to employ Newfoundland and Labrador, New Brunswick, and
cluster sampling. With cluster sampling, the pri- Prince Edward Island. An alternative solution would
mary sampling unit (the first stage of the sampling be to select a stratified cluster sample, which com-
procedure) is not the individuals or units of the bines the notion of stratified sampling discussed
population to be studied, but an aggregate of them, in the previous section with cluster sampling. In
known as a cluster. Imagine you want a nationally our example, we could first group all universities
representative sample of 5000 students. One solu- by region—for example, Atlantic Canada, Quebec,
tion is to sample universities and then students from Ontario, the Prairies, and British Columbia—and
each of the sampled universities, with a probability then randomly sample two universities from each
sample at each stage. For example, you could ran- region. The next step would be to sample 500 stu-
domly sample 10 universities from the entire popu- dents from each of the 10 universities, yielding a total
lation of universities, yielding 10 clusters, and then sample size of 5000. (Again, technical adjustments

bel65796_ch07_152-176 159 01/13/22 12:26 PM

160 PART II Quantitative Research

would be required because the clusters are not of can be used to estimate the population mean (μ) with
equal size.) These are the stages: a known margin of error. Greek letters are used to
denote population characteristics; the Latin alpha-
• Group the universities into five regions. bet (the one used to write in English), for sample
• Sample two universities from each region. characteristics.
• Sample 500 students from each of the 10 To expand on this point, it is necessary to use
universities. some basic statistical ideas. These are presented in
Boxes 7.2 and 7.3, but they can be skipped if all you
Cluster sampling if done properly can be very effec- need is a broad idea of sampling procedures.
tive. Its greatest merit is that it solves the problem
of having no adequate sampling frame. The student Sample size
study described above does not require a list of all
university students in Canada, something that would “How large should a sample be?” “Is the sample big
be very hard to come by. With cluster sampling, no enough?” No issue in sampling raises more questions
such list is ever required. than sample size. Decisions about sample size are not
With large populations, cluster sampling is very straightforward, as we will see below. At the same
economical compared to simple random or system- time, the quest for the perfect sample is usually con-
atic sampling when doing in-person interviews be- strained by the practical considerations of cost and
cause of reduced travel costs—research participants time.
tend to cluster geographically. Using this technique
for research based on telephone interviews has less Absolute and relative sample size
of an economic advantage now than it once did be- People unfamiliar with the field may be surprised
cause long-distance phone charges are so low, and of to learn that it is the absolute size of a sample that
course if the research is to be done on the Internet, is most important, not the proportion of the popu-
travel expenses are not an issue. lation that it makes up. This means that a national
The advantages of multi-stage cluster sampling probability sample of 1000 Canadians has about as
should be clear by now. However, as suggested above, much validity as a national probability sample of
even a very rigorous sampling strategy cannot elim- 1000 Americans, even though the latter involves a
inate all sampling error, as the example in Box 7.2 much larger population.
shows. Increasing the size of a sample increases the pre-
cision of the estimates derived from it; for instance,
a larger sample size means that the 95 per cent con-
The qualities of a probability fidence interval referred to in Box 7.2 is narrowed.
sample However, a large sample cannot guarantee preci-
sion, so it might be better to say that increasing the
Many researchers prefer probability samples because size of a sample probably increases the precision
they allow one to make inferences from the sample of the estimates it can create. In other words, as
to the population from which it was selected. In sample size increases, sampling error tends to de-
other words, with probability samples, the sample crease. Common sample sizes are 100 (minimum)
findings can be generalized to the population. This is and then 400, 900, 1600, and 2500. These sample
not to say that the sample and population character- size increases cut the sampling error by half, then
istics are exactly the same, but the first can be used by one-third, then by one-fourth, and then by one-
to estimate the second with a known probability of fifth respectively. (The reason has to do with the
error. In the example of alcohol consumption in the square roots involved in the denominator of the
sample of 450 students (in the previous section), the calculation; thus the original square root, 10, for a
–
mean number of units consumed by the sample (X ) sample of 100 becomes 20, 30, 40, and 50 [the square

bel65796_ch07_152-176 160 01/13/22 12:26 PM

 7 Quantitative Sampling 161

Box 7.2 Generalizing from a random sample to the population

If an average of 9.7 units of alcohol is consumed in mean, squaring that number, adding up all the re-
the previous seven days by respondents in a prob- sultant scores, dividing that sum by the number of
ability sample, would a similar figure be found in scores, and then square rooting that number. (See
the population? The answer is complex but it is Chapter 8 for more on standard deviation.)
sketched out here and in Box 7.3. Assume for the Thinking about the amount by which a sample
sake of illustration that it would be possible to mean may differ from the population mean is an
take an infinite number of random samples of the important consideration because sampling theory
same size from a population. Not all the resulting tells us that about 95 per cent of all sample means
sample means would be perfect estimates of the lie within (+ or −) 1.96 standard errors of the popu-
population mean. We can never know the popula- lation mean (see Figure 7.5). In other words, we can
tion figure with 100 per cent certainty, but we can be 95 per cent confident that our sample mean lies
imagine another study with a sample mean of 9.6 within 1.96 standard errors of the population mean.
units, another with 9.8, and so on. These outcomes, To illustrate further, if a sample has been se-
as long as there are enough of them and the lected according to probability sampling prin-
sample size is sufficiently high, will take the form of ciples, we can be 95 per cent certain that the
a bell-shaped curve known as a normal distribu- population mean lies between
tion (see Figure 7.5) with the sample means clus-
(a) [the sample mean] − [1.96 times the standard
tering around the population mean. Of the sample error of the mean]; and
means not exactly at the population mean, half will (b) [the sample mean] + [1.96 times the standard
be below the population mean and half above it. error of the mean].
Moving to the left or to the right (away from the
This is known as the 95 per cent confidence in-
population mean), the curve tails off, reflecting
terval. If the mean level of alcohol consumption
the smaller and smaller number of samples gen-
in the previous seven days in our sample of 450
erating means that depart considerably from the
students is 9.7 units and the standard error of the
population mean. The variation of sample means
mean is 1.3, we can be 95 per cent certain that
around the population mean is the sampling error
the population mean lies between
and is measured using a statistic known as the
standard error of the mean. This is an estimate 9.7 – (1.96 × 1.3)
of the amount by which a sample mean is likely to and
differ from the population mean. (Technically, it is
9.7 + (1.96 × 1.3)
the standard deviation of the sampling distribution
of the mean; see Box 7.3). The standard deviation that is, between 7.152 and 12.248 units of alcohol.
of a set of scores is a measure of how dispersed or Under certain conditions, the standard error of
spread out the scores are around the mean, and the mean in a stratified sample is smaller than in
is calculated by taking each score, subtracting the other probability samples. In other words, stratifi-
cation may inject an extra increment of precision
into the probability sampling process (for example,
Number of samples

stratifying by university when levels of alcohol con-

sumption vary widely between universities), since it
eliminates a possible source of sampling error.
By contrast, a sample without stratification
may exhibit a larger standard error of the mean
–1.96 Population +1.96
SE mean SE than a comparable sample with stratification.
Value of the sample mean This occurs because a possible source of vari-
FIGURE 7.5 A distribution of sample means ability between students is disregarded.

bel65796_ch07_152-176 161 01/13/22 12:26 PM

162 PART II Quantitative Research

Box 7.3 Sampling distributions

Assume a population of five cases (with scores = Now count the number of the sample means in
6, 8, 10, 12, 14) from which a random sample of the table; there are 25. There is only one mean of
two cases is to be selected. The five scores sum 14, in the bottom right corner, but five means of
to 50 and therefore the mean of the population 10, on the diagonal going from bottom left to top
is 50/5, or 10. Taking all possible random sam- right. Thus, five of the 25 sample means, or 20 per
ples of two cases with replacement (meaning cent of them, have a value of 10, the same as the
that after a case is chosen it goes back into the population mean, which is what we are trying to
pool, so that two 6s or two 14s are possible), and estimate. Sample means of 9 and 11 are the next
calculating the mean of each sample of size two, most frequent at 32 per cent (a combined total
the following set of means (or “sampling distri- of 8 of 25 means). With 8 and 12, another 24 per
bution of means”) results: cent (6 of 25) are now counted. So sample means
Value of first observation
of 8, 9, 10, 11, and 12 represent 76 per cent of
in sample all possible sample means. The sample means
furthest from the population mean, the 6 and
6 8 10 12 14
14 (in each corner), are the rarest, at 4 per cent
6 6 7 8 9 10
(1 of 25) each.
Value of second 8 7 8 9 10 11
observation in 10
Those percentages are really probabilities.
8 9 10 11 12
sample The probability of getting a sample mean of 7
12 9 10 11 12 13 is 2 chances in 25 or, as it is usually expressed, 8
14 10 11 12 13 14 chances in 100 or 8 per cent.
The logic of a confidence interval should now
Thus, the 7, where 6 and 8 meet, is the mean of be more apparent. A researcher normally takes
those two values, namely: [6 + 8]/2 = 7; the 13 only one sample, but most random sample re-
appears where 12 and 14 meet: [12 + 14]/2 = 13. sults stay close to the population value. In our

root of 2500], yielding the one-half, one-third, one-­ an estimate, the notion of a desired level of precision
quarter, and one-fifth.) Notice that the increases in is not realistic. Instead, to the extent that this notion
precision become less pronounced (going from 100 affects decisions about sample size, it usually does so
to 400 reduces sampling error by half, but 400 to 900 in a general rather than a calculated way. It is usually
reduces it by only one-third) and thus the rate of re- cost that ultimately determines sample size.
duction in the standard error of the mean declines.
Sample size is likely to be profoundly affected by Non-response
considerations of time and cost at such a juncture, Considerations about sampling error do not end
since striving for smaller and smaller increments of with issues of sample size. It’s also important to bear
precision becomes increasingly uneconomic as ev- in mind the problem of non-response.
er-larger samples become less and less cost efficient. The response rate can be defined as the percent-
Fowler (1993) warned against simple acceptance of age of the sample that actually participates in the
this size criterion and argued that it is not normal for study. It is often a good deal less than 100 per cent,
researchers to be able to specify in advance “a desired partly because there are usually some people in the
level of precision” (p. 34). Moreover, since sampling sample who will either refuse or be unable to partici-
error is only one component of any error entailed in pate, who cannot be contacted, or who turn out to be

bel65796_ch07_152-176 162 01/13/22 12:26 PM

 7 Quantitative Sampling 163

example, 76 per cent of the sample means are below. For example, suppose that the population
within (±) 2 units of the population mean. When mean value for testosterone is 10 and athlete
plotted, these results don’t exactly take the X comes in at 14. While that is a possible (non-
shape of the bell curve shown in Figure 7.5, but drugged) outcome, it is rare. So if the doping
you can see the broad outline. Including more committee is willing to be wrong 4 per cent of
and more cases in the sample shapes the lines the time (the frequency with which an honest 14
into a normal curve. occurs), the athlete can be accused. With a rep-
With a real sample, an extension can be cre- utation at stake, however, a 4 per cent chance of
ated on each side of the sample mean to make it error is too high. That is why the tests allow for
more likely to be correct than if the sample mean only much less likely errors.
by itself were used. The extension for a 95 per
cent confidence interval is equal to 1.96 times Sample
the standard error of the mean. (In Box 7.2 the Mean Frequency %
sample mean was 9.7, and the extensions were 14 1 4
± [1.96 × 1.3] or ± 2.548.). The extensions around 13 2 8
the sample mean create what is called an inter- 12 3 12
val. There is only a 5 per cent chance that the 11 4 16
interval does not include the population mean
10 5 20
for a 95 per cent confidence interval.
9 4 16
To illustrate, this is basically how athletes
8 3 12
using banned substances are caught. They
7 2 8
provide one bio-sample (actually two, but one
is a backup), which the authorities then check 6 1 4
against a more sophisticated version of the table 25 100

unsuitable for the study. Others will be excluded if which people who are selected for the sample and
they leave a large number of questions unanswered, participate in the study differ in relevant ways from
or if they do not take the interview or questionnaire those who are selected but do not participate) is in-
seriously. creasingly considered to be more important than the
Unfortunately, survey response rates have been response rate itself (Johnson & Wislar, 2012). Are the
in “steep decline” for decades (Johnson & Wislar, people who have been selected but cannot be con-
2012), with the added problem that there has been tacted richer and on vacation? Are they ill and in hos-
no consensus among researchers regarding how to pital? Such differences may be important (depending
define and calculate response rates. In their research on the research topic) and may have to be taken into
on sustainable consumption in the Edmonton area, account in the analysis. Researchers sometimes take
Kennedy, Krogman, & Krahn (2013) reported a re- steps to determine the level of non-response bias,
sponse rate of 69 per cent. Schieman & Narisada as Schieman & Narisada (2014) did when they con-
(2014), who used a national sample of the Canadian cluded that it was not an issue in their study.
workforce to examine people’s sense of mastery Given that 100 per cent response rates are rare,
over their lives, achieved a response rate of about 40 sample size may have to be adjusted. For example,
per cent. However, non-response bias (the extent to if the goal is a sample of 450 students and earlier

bel65796_ch07_152-176 163 01/13/22 12:26 PM

164 PART II Quantitative Research

surveys suggest that the response rate will be approx- 2 × 2 table there are four cells into which the cases
imately 80 per cent, it may be advisable to sample 575 can fall, let’s say, gender (M/F) × whether the person
individuals, on the assumption that approximately voted in the last election (Y/N). Suppose there were
115 (20 per cent) will not participate, leaving 460 and only 10 males and 10 females in the sample, and that
still room for some additional refusals. 50 per cent of the men and 50 per cent of the women
How far should researchers go to boost response voted, showing no relationship between gender
rates? Smith (1995) showed that response rates are and voting. But suppose the sample by chance in-
affected by things like the subject matter of the re- cluded one more woman who voted and one more
search, the type of respondent, and the effort ex- man who did not. Now the figures show that 40 per
pended on obtaining the cooperation of prospective cent of men voted versus 60 per cent of women—
respondents. Later in the chapter, a number of steps quite a difference. To prevent such a small change
that can improve response rates to questionnaires in the sample from having such a major impact
(which are particularly prone to poor response rates) on the estimate, a larger sample is required. You
are discussed. However, boosting response rates would probably want to have at least 50 males and
can be expensive. Teitler, Reichman, & Sprachman 50 females. Because the population is approximately
(2003) discussed the steps taken to increase the re- half male and half female, this is easy to achieve.
sponse rate of a US sample that was hard to reach, But what if the variable is religion? Fifty Roman
namely both unwed parents of newborn children. Catholics (or 50 Anglicans) are easy to find in any
They found that although increasing the sample random sample of 1000 Canadians, but what about
from an initial 68 per cent to 80 per cent of all cou- 50 Jehovah’s Witnesses or 50 Hindus? One strategy
ples meant that the final sample more closely resem- to achieve statistically meaningful results is to ran-
bled the population from which it had been taken, domly over-sample some of these small groups, as
diminishing returns set in. This does not mean that Dinovitzer, Hagan, & Parker (2003) did to guarantee
steps should not be taken to improve response rates, sufficient social class variation in their study com-
however. For example, following up on people who paring educational attainment among immigrant
do not respond to a questionnaire usually results in and non-immigrant Canadian youth. Similarly, the
an improved response rate, at little additional cost. Canadian Election Study usually over-samples the
less populous provinces (such as Prince Edward
Heterogeneity of the population Island) in order to make interprovincial compari-
Yet another consideration is the heterogeneity of the sons. (A statistical adjustment can be made to com-
population from which the sample is to be taken. pensate for the over-sampling when the data are
When a population is heterogeneous with regard used for purposes other than between-group com-
to characteristics relevant to the topic under study, parisons.) This illustrates how sample size depends
as a whole country or city normally is, the samples in part on the analysis to be conducted.
drawn are likely to be highly varied. When it is rela-
tively homogeneous, such as a population of students
or members of a particular occupation, the amount Types of non-probability
of variation is less. Generally, the greater the hetero- sampling
geneity of a population, the larger a sample should be
in order to maximize the chances that all groups will Non-probability sampling includes all forms of sam-
be adequately represented. pling that are not conducted according to the canons
of probability sampling outlined earlier. It covers a
Kind of analysis wide range of different types of sampling strategy, at
Finally, researchers should bear in mind the kind of least one of which—the quota sample—is claimed by
analysis intended. An example is a contingency table some practitioners to be almost as good as a prob-
showing the relationship between two variables. In a ability sample. This section covers three common

bel65796_ch07_152-176 164 01/13/22 12:26 PM

 7 Quantitative Sampling 165

non-probability samples: the convenience sample; who are dependent on them for a grade to take part
the snowball sample; and the quota sample. in their research.)
This is not to suggest that convenience samples
Convenience sampling should always be avoided. In fact, they are used
A convenience sample is one that is used simply more often than you might think. Mairs & Bullock’s
because the elements are readily available to the re- (2013) research on sexually risky behaviour and
searcher. Imagine a professor in a faculty of educa- HIV testing on the part of Canadian “snowbirds”
tion who is interested in the qualities that teachers in Florida was based on a convenience sample, as
want in their principals, and who has several classes were MacKinnon & Luke’s (2002) study of cultural
made up of teachers working part-time toward change, and Beagan’s (2001) analysis of medical
master’s degrees. Now suppose that she decides to students mentioned in earlier chapters. However,
administer a questionnaire to those students—a studies involving convenience samples should ac-
sample of convenience. Chances are that she will knowledge that they are of limited use in estimating
receive all or almost all of the questionnaires back, population characteristics.
thus achieving a high response rate. However, the Convenience samples are often utilized in pilot
use of the convenience sampling strategy makes it studies. Imagine that part of the education professor’s
impossible to generalize the findings to teachers as research involves developing a battery of questions
a whole: the very fact that her students are taking to measure the leadership preferences of teachers
this degree program marks them off as different that she plans to use later, with a different sample.
from teachers in general. (In addition, of course, it’s Since it is highly desirable to pre-test such a research
ethically questionable for a researcher to ask people instrument before actually using it, administering

Research in the News

The perils of polling
An article in the Winnipeg Free Press (Taniguchi, are becoming aware of racism as a social prob-
2020) cites a 2020 poll conducted by a local survey lem. According to Lori Wilkinson, a sociologist at
research company indicating that a greater the University of Manitoba, “more people are
number of Manitobans believe that racism is a seeing it [racism]. You ask people their opinions
problem in the province than was the case five about anything, and it really depends on what’s
years earlier. A survey done by the company in newsworthy or what’s being discussed online”
2015 suggested that 44 per cent of Manitobans (Taniguchi, 2020). Wilkinson maintains that
thought that racism was a problem in their prov- more people are becoming aware of racism be-
ince, compared to 61 per cent in 2020. cause of recent worldwide anti-racism protests
However plausible these figures may be, and the extensive media attention they have
the samples were not selected randomly, so received.
there is no way to estimate how accurate the Although representative sampling may not be
results were. To their credit, both the author of feasible in a number of research situations, the
the newspaper article and the research firm on results produced by non-probability sampling
its website state that because of the sampling should be interpreted with caution. At a mini-
procedures used, “no error rate can be calcu- mum, the researcher should, as was done here,
lated.” This is not to say that the results have no state explicitly that there is no way to gauge their
value—it’s quite possible that more Manitobans accuracy.

bel65796_ch07_152-176 165 01/13/22 12:26 PM

166 PART II Quantitative Research

Box 7.4 A convenience sample

Suppose you want to study university undergrad- of university, faculty, degree program, and
uates to find out the extent of their part-time em- course are made purposively rather than ran-
ployment while at school. To save time and effort domly. Because of the way the questionnaires
you decide to conduct the study at your own are administered (in class), there is a very high
university. You could use probability sampling response rate. On the other hand, students who
to get participants, but you think the research is are absent from class do not get a chance to
not important enough to justify the cost. Instead, fill out a questionnaire. An important question
you choose five faculties, and within each, one is whether absence from class is connected in
or two of the specific degree programs it offers: some way to part-time work, the latter variable
anthropology, visual arts, chemistry, electrical being the focus of your study. In other words,
engineering, French, kinesiology, media studies, is absence higher among students who work
and psychology. This choice of participants is de- part-time, perhaps because they are working
signed to maximize variety in the types of degree at the time of the class? Are some students too
programs represented while providing similar tired to go to class because of their part-time
numbers of males and females (since in particu- work? If so, you will probably underestimate
lar degree programs one gender frequently pre- the proportion of students who work part-time.
dominates). Questionnaires can then be given Also, because you have selected a convenience
to students in first-, second-, third-, and fourth- sample, you must say so in your research
year courses. report, acknowledging that the results cannot
Together, these procedures represent a be generalized to Canadian students as a
good attempt at generating a varied sample. whole or even to the student population at your
It is a convenience sample, because the choices university.

it to a group that is not part of the main study is a (2003) used this approach to create a sample of im-
legitimate way of exploring issues such as whether migrant women in the Maritimes. Bowen (2015)
respondents will reply honestly to questions on sen- used it to recruit informants for her study of people
sitive topics, and whether the questions are clear and in the Vancouver area who either had experienced
comprehensible. A convenience sample may also be the transition from off-street sex work to “square”
used to test newly created scales for reliability and to (non-sex) work, or were simultaneously employed in
generate ideas for further research. Box 7.4 contains both. And Whalen and Schmidt (2016) chose snow-
another illustration of how a convenience sample ball sampling to recruit participants for their study
may be used. of Newfoundland women whose male partners were
away from home, working in the Alberta oil indus-
Snowball sampling try. Box 7.5 describes the generation of a snowball
Snowball sampling is a form of convenience sam- sample of marijuana users for what is often regarded
pling, but worth distinguishing because it has at- as a classic study of drug use.
tracted quite a lot of attention over the years. With Becker’s comment on creating a snowball sample
this approach, the researcher makes initial con- is interesting: “The sample is, of course, in no sense
tact with a small group of people who are relevant ‘random’; [indeed] it would not be possible to draw
to the research topic and then uses them to estab- a random sample, since no one knows the nature of
lish contact with others. Tastsoglou and Miedema the universe from which it would have to be drawn”

bel65796_ch07_152-176 166 01/13/22 12:26 PM

 7 Quantitative Sampling 167

Box 7.5 A snowball sample

Becker reported on how he generated a sample
of marijuana users:

I had been a professional dance musician . . .

and my first interviews were with people
Source: Irina274/iStockphoto

I had met in the music business. I asked

them to put me in contact with other [mar-
ijuana] users who would be willing to dis-
cuss their experiences with me. … Although
in the end half of the fifty interviews were
conducted with musicians, the other half
Multiple researchers have used snowball sam-
covered a wide range of people, including
pling to study issues around illegal drug use in the
labourers, machinists, and people in the
Downtown Eastside neighbourhood of Vancouver.
Why might snowball sampling be useful in this sce- professions. (1963, pp. 45–46)
nario? What are some other research situations in
which snowball sampling would be useful?

(Becker, 1963, p. 46). What Becker is saying here is immigrant women in the Maritimes, much less to
that there is no accessible sampling frame for the those in the rest of Canada. This is not to suggest that
population from which the sample is to be taken. The snowball sampling is entirely irrelevant to quantita-
difficulty of creating such a sampling frame means tive research: when the researcher needs to focus on
that a non-probability approach is the only feasible relationships between people, tracing connections
one. Moreover, even if it were possible to create a through snowball sampling can be a better approach
sampling frame of marijuana users or British visitors than conventional probability sampling. Statistically
to Disney theme parks, it would almost certainly small groups—gay francophone lawyers in New
become obsolete very soon, because those popula- Brunswick, for example—are researchable with a
tions are constantly changing. People become and snowball approach. Taking a random sample from
cease being marijuana users all the time, while new that population would not be feasible.
theme park visitors arrive every minute.
The problem with snowball sampling is that it is Quota sampling
very unlikely to be representative of the population, Quota sampling is not a predominant form of sample
although, as Becker’s experience suggests, the very selection in academic social research, but it is widely
notion of a population is problematic in some cir- used for commercial purposes in market research
cumstances. However, snowball sampling is gener- and in political opinion polling in some countries.
ally used not within a quantitative research strategy, However, the advent of the online opt-in survey
but within a qualitative one: both Becker’s work on panel has made quota sampling more common in
marijuana users and Bryman’s (1995, 1999) studies academic research in recent years. Opt-in panels are
on theme parks used a qualitative research frame- made up of people who are willing to participate in
work, and concerns about the ability to generalize online surveys, often in exchange for credit points
do not loom large in qualitative research. Indeed, or small amounts of money. Third parties who need
Tastsoglou and Miedema (2003) warned their read- participants for their research contact the people
ers not to generalize from their 40 women to all in charge of the panel and pay a fee to have their

bel65796_ch07_152-176 167 01/13/22 12:26 PM

168 PART II Quantitative Research

Practical Tip | Sample size and probability sampling for students

For most quantitative researchers sample a random sample. Explain why you could not
size does matter: the bigger the sample, the increase your sample size. Above all, though,
more representative it is likely to be (provided don’t make claims about your sample that are
the sample is randomly selected), regardless not warranted. Never claim that your sample
of the size of the population from which it is is representative, or randomly selected, unless
drawn. However, students have to do their that is true. Also, even if your sample is small,
research with very limited resources, and a it may have some good features: the range of
truly random sample may not be feasible for people included, a good response rate, high
them. The crucial point is to be clear about levels of honesty. Make sure to play up these
your sampling decisions and to justify what positive features while being honest about the
you have done. Explain the difficulties that sample’s limitations. This is what professional
you would have encountered in generating researchers do.

surveys administered to people on the panel. The ad- been decided, the researchers or those contracted
vantages of using opt-in panels include fairly quick by them to select the sample try to pick people who
turnaround times and a relatively low cost. fit those categories. The quotas are typically interre-
The aim of quota sampling is to produce a sample lated. As in stratified sampling, the population may
that reflects a population in terms of the relative pro- be divided into strata covering several character-
portions of people in different categories (gender, istics (for example, gender, employment, and age)
ethnicity, age, socioeconomic status, region of resi- all at once (see Table 7.2). For example, fifty 25- to
dence, etc.), or in combinations of these categories. 44-year-old unemployed females may be required
Unlike a stratified sample, however, a quota sample to meet the quota. If necessary, statistical adjust-
is not random, since the final selection of people in ments can be made to the sample data such that the
each category is left up to the researcher or the orga- relative sizes of the categories match those of the
nization in charge of the opt-in panel. For Canadian population.
studies, information about the stratification of the A number of criticisms are frequently levelled at
Canadian population or about certain regions can quota samples:
be obtained from sources such as the census and
from surveys based on probability samples such as • The proponents of probability sampling argue
the General Social Survey, which can then be used that because the choice of respondent is left
to establish the proportion of the sample that is to be to the researcher or the organization collect-
made up of people in the various categories. ing the data, a quota sample is not likely to
Once the categories and the numbers of people be representative of the population. As noted
to be surveyed within each one (the quotas) have earlier, if people are approached in person,

TABLE 7.2 Strata for gender, age, and employment

Older, female, unemployed (5%) Younger, female, unemployed (4%)
Older, female, employed (17%) Younger, female, employed (20%)
Older, female, other (7%) Younger, female, other (1%)
Older, male, unemployed (3%) Younger, male, unemployed (6%)
Older, male, employed (15%) Younger, male, employed (15%)
Older, male, other (6%) Younger, male, other (1%)

bel65796_ch07_152-176 168 01/13/22 12:26 PM

 7 Quantitative Sampling 169

interviewers may be unduly influenced by that have to be considered are not limited to sam-
their perceptions of friendliness or willing- pling people. Several other sampling issues may be
ness to make eye contact (many people will involved:
look at the ground and shuffle past as quickly
as possible, not wanting to be bothered). If • If the structured observation is to take place in
people are recruited online, differences be- public areas, there will be no sampling frames
tween people who have access to the Internet (lists of people who frequent those areas). It
compared to those who do not can result in would be very difficult, for example, to get a
sample selection bias. For example, online complete list of people who walk along a par-
populations tend to be younger, better edu- ticular street, or of men who frequent certain
cated, and have higher incomes than people “tearooms.” One can, however, choose the
without Internet access, which can affect the people by using a table of random numbers:
representativeness of the sample. And among for example, the third, then seventh, then eigh-
people who do go online, there may be rel- teenth person seen in the public area. The same
evant differences between those who join table can be used to establish random samples
opt-in panels—a form of self-selection—and of time, place, or activity.
those who do not. • Time sampling involves an observer recording
• If prospective participants are approached in whatever is happening at a particular time,
person, the characteristics of the people who for example, in a pub every 15 minutes after
are in the vicinity of the investigator when re- a random start time. This can be done on one
cruits are sought may not be typical of the pop- person or on several individuals at once and
ulation of interest. There is a risk, for example, the 15-minute figure can be varied using a
that people not in full-time paid work may be table of random numbers. It may be necessary
over-represented, especially in locations such that individuals watched on more than one day
as malls, making the sample unrepresentative. not be observed at the same time of the day; the
Also, judgments about eligibility may some- observation periods may have to be randomly
times be incorrect, for instance regarding the selected. For example, it would be an error to
respondent’s age, and even more so for social observe a certain pupil in a school classroom
class. For example, someone who is actually always at the end of the day, when most stu-
young enough to be eligible but looks older dents are tired, since tiredness could give a
may not be asked because the quota of the false impression of that pupil’s behaviour.
older group is filled. In such cases an element
of bias is introduced. Time sampling should be combined with the next
• It is not appropriate to calculate a standard strategy:
error term (such as the standard error of the
mean) from a quota sample, so it is not possible • Random sampling of specific places: for exam-
to estimate the amount by which a sample char- ple, in a pub near the entrance, at the bar, by
acteristic is likely to differ from the population. the pool table, outside where the smokers con-
In fact, most statistical estimates are problem- gregate, the washroom area, and the booths.
atic unless random sampling has been used. • “Behaviour sampling,” whereby an entire
group is watched and the observer records
a particular kind of behaviour. Thus, in that
Structured observation and same pub, the researcher can observe the nth
sampling pick-up (second, then fifth, etc.) after the previ-
Structured observation, like survey research, neces- ous one was observed, using a random number
sitates decisions about sampling. However, the issues table to determine the n.

bel65796_ch07_152-176 169 01/13/22 12:26 PM

170 PART II Quantitative Research

Considerations relating to probability sampling difficult; thus settings that are willing to grant access
derive largely from concerns about external ­validity. may not be representative.
For example, if a structured observation study is con-
ducted over a relatively short span of time, questions Limits to generalization
are likely to arise concerning the stability or repre-
sentativeness of the findings over longer periods. If An important point, often not fully appreciated, is
the research is conducted in schools, for instance, that even when a sample is selected using probabil-
observations conducted toward the end of the school ity sampling, any findings can be generalized only
year, when examinations are likely to loom large for to the population from which the sample was taken.
both teachers and students, may produce different This is an obvious point, but it is easy to think that
results than observations made at a different point findings from any study have a broader applicability.
in the academic year. Consequently, consideration If the study is based on a random sample of adult
has to be given to the timing of the observations. Calgarians, for example, its findings cannot be gener-
Furthermore, how are the sites in which structured alized to the whole province of Alberta, much less to
observation is to take place selected? Are they rep- other regions of Canada. The opinions of Calgarians
resentative? Using a random sampling procedure may be very different from those of Edmontonians
to select the schools would assuage any worries in or people who live in Lethbridge, and even more dif-
this connection. However, gaining access to settings ferent from the opinions of Montrealers or residents
such as schools and business organizations is usually of St John’s.

Methods in Motion Applications to Canadian Society

The need for youth mental health services in Canada
Weins et al. (2020) used Statistics Canada’s The researchers conclude that these indica-
Canadian Community Health Survey (2011–2018) tors point to a need for expanded mental health
to examine trends in mental health among youth services for young people in Canada. They are
aged 12–24. The study was based on nation- cautious, however, in interpreting their findings
ally representative samples that were gathered in terms of an actual increase in the prevalence
using stratified cluster sampling in which the of mental illness. The pervasiveness of mental
strata were health regions, geographical areas health problems may have expanded in the years
within health regions, and dwellings within the covered by the study, but they also state that it is
latter. possible that the findings reflect changes in the
The study found that the incidence of per- ways mental and emotional disorders are diag-
ceived poor/fair mental health, formally diag- nosed, greater awareness of mental health mat-
nosed mood and anxiety disorders, and the ters among members of the public, and greater
number of mental health consultations increased acceptance of people who suffer from mental ill-
over the years covered by the study, especially ness. Similarly, the apparent increase in cannabis
among young adult women. Substance abuse dis- use following legalization could reflect real dif-
orders can be related to poor mental health, and ferences in consumption levels, but it may also be
the researchers report that cannabis use among related to less reluctance to admit to marijuana
adolescent males appeared to be decreasing in use following legalization. Wiens and colleagues
the period before the legalization of marijuana conclude that regardless of the reasons for the
(2011–2017), but increased by 5.6 per cent in the higher percentages showing up in their data, the
first year of legalization (2018) compared to the findings reflect a greater need for youth mental
previous year’s levels. health services than was previously recognized.

bel65796_ch07_152-176 170 01/13/22 12:26 PM

 7 Quantitative Sampling 171

One issue rarely discussed in this context, and for example, a systematic sample can be generated
almost impossible to assess, is whether there is a by randomly selecting one date and then selecting
time limit on the findings generated. Quite aside every nth date thereafter. Using a variation of this
from the fact that findings cannot (or at least should method, Collins (2014), in a study of how the gender
not) be generalized beyond the population sampled, of the offender and the victim may affect media por-
is there a point at which one should say, “Well, those trayals of crime, randomly chose two numbers (two
findings applied then, but do they apply now?” To and seven) in order to select years to analyze arti-
take a simple example: no one should assume that cles published in the Vancouver Sun, Saskatoon Star
the findings of a 1995 study of university students’ Phoenix, Winnipeg Free Press, and Toronto Star. This
budgeting and personal finance habits apply to to- produced the years 1982, 1987, 1992, 1997, 2002, and
day’s students. Ever-increasing costs have changed 2007. She then used a pseudo-random number gen-
the ways students finance their education, increasing erator (which uses an algorithm to produce a set of
their reliance on part-time work, parents, and loans. numbers that approximates a sequence of random
But even when there is no definable or recognizable numbers) to select twenty dates that would be used
source of relevant change, there is still the possibility to search for relevant articles in each newspaper,
(even the likelihood) that the findings are time spe- which yielded a total of 1190 articles.
cific. Changes over time are a major reason why it is The time span analyzed in content analysis often
important to conduct replications. depends on the research questions. Dumitrica and
Bakardjieva (2018) wanted to examine how Canadian
Content analysis sampling newspapers viewed the role of social media in grass-
roots political engagement and mobilization, but
There are several phases in the selection of a sample were unsure of what might constitute a good start-
for content analysis. Applying content analysis to the ing date. They speculated that because social media
mass media is explored here but the basic principles usage seemed to reach a critical phase around 2005,
have broader applications. that would be a good year to use as the starting point.
Their search for relevant articles published that year
Sampling media did not yield any that fit their criteria, but they found
Many studies of the mass media specify a research one released in 2006 and twenty-five published in
problem in the form of “examining the representa- 2007, with the latter year’s total being typical of what
tion of X in the mass media.” The X might be trade was available for each of the next seven years. In the
unions, women, food scares, crime, or drunk driving. end they used articles published from 2006 through
But which mass medium to choose—­newspapers, 2014 as the source of their data. Researchers often
television, radio, chat rooms, social media? If news- choose a time span that provides a good chance of
papers, will the study focus on tabloids, broadsheets, producing the data needed to answer their research
or both? Will it include Sunday papers, or free news- questions, and may adjust accordingly depending on
papers? Will it examine feature articles and letters to what they find.
the editor, or news stories only? Buchanan’s (2014)
study of the decline of local news coverage in two Reducing non-response
Ontario metropolitan newspapers analyzed every
article of any kind contained in 19 randomly selected There is little consistent evidence on whether re-
editions of each paper, yielding a total sample of 7129 sponse rates are lower for interviews conducted over
articles. the telephone rather than face-to-face (see Table 7.3).
However, there is a general belief that telephone
Sampling dates interviews achieve slightly lower response rates
With an ongoing general phenomenon such as crime, than personal interviews (Shuy, 2002; Frey, 2004).
probability sampling can be used to sample dates: Developments in telephone communications such

bel65796_ch07_152-176 171 01/13/22 12:26 PM

172 PART II Quantitative Research

TABLE 7.3 Comparing the strengths of different ways of contacting members

of the chosen sample
MODE OF SURVEY ADMINISTRATION
Face-to-face Telephone Posted
interview interview questionnaire Email Web
Resource issues
Is the cost of the mode of ✓ ✓✓ ✓✓✓ ✓✓✓ ✓✓✓
administration relatively low?
Is the speed of the mode of ✓ ✓✓✓ ✓✓✓ ✓✓✓ ✓✓✓
administration relatively fast?
Is the cost of handling a ✓ (✓✓ if ✓✓✓ ✓✓✓ ✓✓✓ ✓✓✓
dispersed sample relatively low? clustered)

Sampling-related issues
Does the mode of ✓✓✓ ✓✓ ✓ ✓ ✓
administration tend to produce
a good response rate?
Is the researcher able to control ✓✓✓ ✓✓✓ ✓✓ ✓✓ ✓✓
who responds (that is, the
person targeted is the one who
answers)?
Is the mode of administration ✓✓✓ ✓✓ ✓✓✓ ✓ (because ✓ (because
accessible to all sample of need for of need for
members? respondents respondents
to be to be
accessible accessible
online) online)
Is the mode of administration ✓✓✓ ✓✓✓ ✓✓ ✓✓ ✓✓
less likely to result in non-
response to some questions?
Notes: Number of ticks indicates the strength of the mode of administration of a survey in relation to each issue. More ticks correspond to more advantages in
relation to each issue. A single tick implies that the mode of administering a questionnaire does not fare well in this area. Two ticks imply that it is acceptable
and three ticks that it does very well. This table was influenced by Czaja and Blair (1996).

as the growing use of call screening and cellphones the day. People living alone may be reluctant to
have certainly had an adverse effect on telephone answer the door, especially after dark.
survey responses. • Reassure people that you are not a salesperson.
Interviewers in both types of study play an im- Because of the unethical tactics of organizations
portant role in maximizing the response rate for a whose representatives falsely say they are doing
survey study. The following points should be borne market or social research, many people have
in mind: become very suspicious of anyone who says
they would “just like to ask a few questions.”
• Interviewers should be prepared to keep call- • When conducting in-person interviews, dress
ing back if interviewees are unavailable. This in a way that is acceptable to a wide spectrum
requires taking into account people’s likely of people.
work and leisure habits: there is no point in • Make it clear that you are happy to find an in-
daytime calling for people who work during terview time to suit the respondent.

bel65796_ch07_152-176 172 01/13/22 12:26 PM

 7 Quantitative Sampling 173

Practical Tip | Response rates

The lower the response rate, the more likely it is to recognize and acknowledge the negative
is that questions will be raised about the rep- implications of a low rate.
resentativeness of the sample. Mailed ques- Also, response rates are usually an issue
tionnaires in particular are often associated only with randomly selected samples. Samples
with low response rates, and as Mangione’s not selected using probability methods are not
classification, shown below, illustrates, some expected to be representative of a population
authorities consider a response rate below 50 even if all the people selected participate. But
per cent unacceptable. Then again, a great here again, it is necessary to acknowledge the
deal of published research achieves only a low limitations of the sample.
response rate, so don’t despair. The key point

Improving response rates to mailed The researcher should try to deal with any po-
questionnaires tential fears in the letter.
Mangione (1995, pp. 60–61) classified response rates • The mailed questionnaires should always be
to mailed questionnaires as follows: accompanied by a stamped return envelope or
(at the very least) return postage.
• Don’t allow the questionnaire to appear unneces-
Over 85% excellent
sarily bulky. Some researchers reduce the size of
70–85% very good
the questionnaire to fit a booklet format. As with
60–69% acceptable structured interviewing, begin with questions
50–59% barely acceptable most likely to be of interest to the respondent.
Below 50% not acceptable • Follow up individuals who do not reply at first,
possibly with two or three further mailings.
Because of the tendency for mailed questionnaires to The importance of reminders cannot be over-
generate lower response rates than structured inter- stated—they do work. One approach is to send
views (and the implications this has for the validity out a reminder letter to non-respondents two
of findings), a great deal of thought and research has weeks after the initial mailing, reasserting the
gone into improving response rates for the former. nature and aims of the survey and suggesting
The following steps are frequently suggested: that the person contact a member of the re-
search team to obtain a replacement copy of
• Write a good covering letter explaining the the questionnaire should the original one have
reasons for the research, why it is important, been lost. Then, two weeks after that, those who
and why the recipient has been selected; pro- still have not responded should be sent another
vide guarantees of confidentiality. letter along with a second copy of the question-
• Make it personal by including the respon- naire. These reminders have a demonstrable
dent’s name and address in the covering effect on the response rate. Some writers argue
letter, and by ensuring that each letter is in- for even more than two mailings of reminder
dividually signed. A cover letter addressed letters. If a response rate is worryingly low, fur-
to “Occupant” will put most people off. The ther mailings are certainly desirable.
downside of the personal approach is that it • Monetary incentives increase the response rate
may raise concerns: How did they know about but can be deemed unethical. They are most
me? Will my responses really be confidential? effective if they arrive with the questionnaire

bel65796_ch07_152-176 173 01/13/22 12:26 PM

174 PART II Quantitative Research

(as opposed to after its return). Apparently, • As mentioned earlier, Internet users are a
most respondents will not take the money and biased sample of the population in that they
discard the questionnaire. The evidence also tend to be better educated, wealthier, younger,
suggests that quite small amounts of money etc. than non-users.
have a positive impact on the response rate, • Few sampling frames exist for the general
but that larger amounts do not necessarily im- online population, and most of these are likely
prove the rate any further. to be either confidential or too expensive to
acquire.

Virtual sampling issues Such issues make it difficult to satisfy all the crite-
A limitation of online surveys is that not everyone is ria for conducting probability sampling. The chief
online and has the technical ability to handle ques- problem with virtual sampling strategies is that the
tionnaires in either email or Web formats, although representativeness of the sample is almost always in
fewer and fewer people lack these characteristics. question. However, online opt-in survey panels are
Other factors that make such surveys problematic making it easier for social researchers to gain access
include the following: to online participants, and one could argue that they
provide better access to large populations and more
• Many people have more than one email ad- diverse samples than other forms of non-probability
dress, and the address known by the researcher sampling. Regardless of the type of sampling chosen,
may not be commonly used by the participant, it is always necessary to discuss the limitations of the
making communication difficult. method used.

Key Points
• Probability sampling is a mechanism for reduc- • Under certain circumstances, quota samples can
ing bias in sample selection. be an alternative to random samples, but they
• Key technical terms in sampling include rep- have some deficiencies.
resentative sample, random sample, non- • Convenience samples can provide useful data,
response, population, and sampling error. but have limited generalizability.
• Randomly selected samples are important be- • Response rates vary by the medium of commu-
cause they permit generalizations to the popula- nication and most can be improved by persistent
tion of interest with a known probability of error. follow-up procedures.
• Sampling error generally decreases as sample
size increases.

Questions for Review (r) and Creative Application (a)

R1 What do each of the following terms mean: Probability sampling
“population”; “probability sampling”; “sampling R2 Why does probability sampling offer a greater
frame”; and “representative sample”? chance of selecting a representative sample
A1 You have been asked to organize your high than non-probability sampling?
school reunion this year, and want to select A2 A researcher positions herself on a street corner
a sample of your classmates to pick a venue and asks every fifth person who walks by for an
for the event. How could you create a sam- interview until she has a sample of 250. Would
pling frame and select a random sample of 50 the results be generalizable to any specific pop-
people? ulation? Explain.

bel65796_ch07_152-176 174 01/13/22 12:26 PM

 7 Quantitative Sampling 175

R3 What are the main types of probability sample? the population in order to get meaningful results
How is each one selected? (e.g., the ethnic breakdown)?
A3 You want to conduct an interview survey of 500
people in Winnipeg that would give you find- Non-probability samples
ings that are generalizable to all adults in the R5 What are the main kinds of non-probability
city. What type of sample should you select? sample, and how are they selected?
Why? A5 Which form of non-probability sample is likely
to be the least useful if your goal is to market a
Sample size new type of cellphone? Which is likely the most
R4 Explain why non-response bias may be a more useful? Explain.
important consideration in sampling than the
response rate. Limits to generalization
A4 You are doing a study on the relationship be- R6 Besides selecting a probability sample of ad-
tween body acceptance and social media use equate size, what factors should be taken into
among young women, and will use an online consideration before generalizing the results of
questionnaire to gather your data. Assume that a study to a particular population?
you have chosen an appropriate method to gen- A6 You just completed a study on racism in Canada
erate a probability sample. What could you do to using appropriate sampling techniques. For how
maximize the response rate? What characteris- long can you consider your results to be applica-
tics of the sample should be the same as those of ble to Canadian society? Explain.

Interactive Classroom Activities

The number of exercises chosen here will depend affect its accuracy? Which sample size has the
on the amount of class time available and the least amount of sampling error? Why?
inclinations of the instructor. It may be advis- 2. Each member of the class then selects a sys-
able for the instructor to determine popula- tematic sample (every nth case, using a random
tion and sample sizes in advance to simplify the start) the size of which matches that of the larg-
calculations. est simple random sample selected in exercise 1.
For example, if the population size was 80 and
1. The instructor gives each student in the class (or the largest simple random sample selected had
a predetermined number of participants) an ID a size of 40, every second case (80/40) would be
number, and asks them how many kilometres selected for the systematic sample. The class
they have to travel to get to campus, which is is asked: How close is the mean of the system-
recorded and then shown on the blackboard/ atic sample to that of the largest simple random
screen, along with the gender of the student. sample? Why are the two sample means not
The data are then treated as population data. always the same?
(For very large classes, a subset of the class can 3. Each person in the class calculates the gender
be randomly selected for this using a “number- breakdown for each sample selected in exercise
ing off” procedure.) The instructor or a volunteer 1, and compares it to the overall gender break-
then adds up the total kilometres travelled for all down of the population. The class is asked: How
participants, which is used to calculate the pop- do these breakdowns differ by sample size?
ulation mean. Each member of the class then What does it mean to “stratify” a population by
uses a random number table to select simple gender? Would stratifying be necessary in order
random samples of increasing size, for example, to have the gender breakdown of the samples
n = 1, n = 10, n = 20, n = 40, and so on, which match that of the population?
are used to calculate sample means. The class 4. The mean number of kilometres travelled by all
is then asked: How does the size of the sample females in the population is calculated, as is the

bel65796_ch07_152-176 175 01/13/22 12:26 PM

176 PART II Quantitative Research

mean for all males. The means for women and Would it be appropriate to stratify by gender for
men are then calculated for each sample se- this variable? Why or why not? How could a gen-
lected in exercise 1. The class is asked: How is the der-stratified sample be selected to calculate the
accuracy of the means affected by sample size? mean number of kilometres travelled to campus?

Relevant Websites
Statistics Canada provides a good general in- Statistics Canada provides the sampling tech-
troduction to sampling. Both probability and niques used for the 2018 General Social Survey—
non-probability sampling techniques are described Giving, Volunteering and Participating here:
here. Navigate the site by using the links on the left [Link]
side of the page. Test your knowledge of sampling pl?Function=getSurvey&SDDS=4430#a2
by doing the exercises provided.
This YouTube video illustrates how to select a
[Link]/edu/power-pouvoir/ch13/5214895- simple random, convenience, systematic, cluster,
[Link]
and stratified sample.
[Link]
(Websites accessed 4 September 2021.)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch07_152-176 176 01/13/22 12:26 PM

 8 Quantitative Data
Analysis
Chapter Overview
This chapter presents some of the most basic and frequently used methods for analyzing quantitative data.
To illustrate them, we’ll use an imaginary data set on attendance at a gym—the kind of data set that could
be generated from a small research project suitable for most undergraduates.
The chapter explores:
• the importance of anticipating questions of analysis early in the research process, before all the data
have been collected;
• the different kinds of variables generated in quantitative research (knowing how to distinguish them
is crucial in deciding which statistics to use);
• methods for analyzing a single variable (univariate analysis);
• methods for analyzing relationships between two variables (bivariate analysis); and
• analysis of relationships among three or more variables (multivariate analysis).

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Explain how to deal with missing data.

2. Differentiate between three types of variable: nominal, ordinal, and interval/ratio.
3. Interpret various forms of univariate analyses, including frequency tables and diagrams, measures of
central tendency, and measures of dispersion.
4. Describe several kinds of bivariate analyses, including those involving contingency tables, Pearson’s r,
scatter diagrams, Kendall’s tau-b, Cramér’s V, eta, and the chi-square test.

▲ kan_chana/Shutterstock

bel65796_ch08_177-204 177 01/13/22 12:26 PM

178 PART II Quantitative Research

5. Outline the purpose of inferential statistics and the meaning of the term “statistical significance.”
6. Discuss the goals of multivariate analyses, and how they can be used to test for spuriousness, i­ ntervening
variables, and statistical interactions.

If you’re like many students in the social sciences, Feel better now? If not, remember that the best way
the idea of learning about statistics doesn’t fill you to learn a new language is to start with the basics.
with joy. How about communing with nature? Yes, That’s what we’ll do in this chapter. In fact, we’re
that would be much better—open fields, wild winds, not going to go beyond a basic level, and we won’t
the bright sky, a riot of colour provided by the plants even be using any formulas—we’ll leave that for
and animals around you. On the other hand, it’s your statistics courses. What we will do is introduce
often said that mathematics is the language that the logic of statistics, which will give you the con-
nature speaks, and statistics is a form of applied ceptual foundations you need to do quantitative
mathematics. Communing with nature via statis- data analysis. We’ll also discuss some important
tics can be just as invigorating as communing any ways of organizing quantitative data. Think of it as
other way, especially if you work at it for a while. a new language class.

Introduction • The size and nature of the sample also impose

limitations on the kinds of techniques that are
Once a sample has been selected and the numerical suitable for the data set.
data have been gathered, the next step is to analyze
the information you have. Here we will discuss some
basic ways to do that. We’ll avoid the formulas that A small research project
underpin those techniques, in order to keep the
discussion focused on interpreting statistics and on The discussion of quantitative data analysis provided
methodological issues. One chapter could not do here is based on an imaginary piece of research car-
justice to the topic of numerical data analysis in any ried out by an undergraduate for her honours thesis
case; readers are advised to consult books that pro- on leisure in modern society. She chose that topic
vide more detailed and advanced treatments (for ex- because of her enthusiasm for gyms and workout fa-
ample, Agresti & Finlay, 2009; Noack, 2018). cilities, and her interest in how and why such venues
Before we begin, a warning is in order. The biggest are used. She had a hunch that they might be indica-
mistake in quantitative research is to think that data tive of a “civilizing process” and used this theory as a
analysis decisions can wait until after the data have framework for her findings (Rojek, 1995, pp. 50–56).
been collected. Data analysis is carried out after that She was also interested in issues relating to gender
stage, but it’s essential to be fully aware of what tech- and body image, and suspected that men and women
niques will be used before data collection begins. differ in their reasons for going to a gym as well as in
Interview schedules, questionnaires, observation their activities there.
schedules, and coding frames should be designed She secured the permission of a gym close to her
with the data analysis in mind, for two main reasons: home to contact a sample of its members by mail.
The gym had 1200 members and she decided to take
• The statistical techniques that can be used a simple random sample of 10 per cent of them (that
depend on how a variable is measured. is, 120 members). She mailed questionnaires to mem-
Inappropriate measurement may make it bers of the sample with a covering letter testifying to
impossible to conduct certain types of data the gym’s support for her research; the letter also in-
analysis. cluded a Web address that allowed participants to fill

bel65796_ch08_177-204 178 01/13/22 12:26 PM

 8 Quantitative Data Analysis 179

out the questionnaire online. One thing she wanted Types of variables
to know was how much time people spent on each of
the three main types of activity in the gym: cardio- Look at the different questions and notice that some
vascular workouts, weightlifting, and other activities of them call for answers in real numbers: questions
(for example, stretching exercises). She defined each 2 (age), and 10, 11, and 12. Question 8 has an either/
of these carefully in the covering letter and asked or answer, which is called a dichotomy (only two
members of the sample to note down how long they possible responses). The rest of the questions take
typically spend on each of the three activities during the form of lists of categories, but there are also dif-
a single visit to the gym. She ended up with 90 com- ferences among them. Some of the answers can be
pleted questionnaires out of the 120—a response rate rank ordered: see questions 4, 5, and 6. Thus, for
of 75 per cent. question 6 the category “every day” implies greater
The entire questionnaire ran to four pages; 12 of frequency than “4–6 days a week,” which in turn is
the questions along with the responses of a fictional a greater frequency than “2 or 3 days a week,” and
respondent are provided in Box 8.1. Questions 1, 3, so on. Compare this to questions 1, 3, 7, and 9 where
4, 5, 6, 7, 8, and 9 were pre-coded; the respondent the categories can’t be rank ordered. In the case of
simply had to circle the appropriate code. With the question 3, for instance, “relaxation” is neither more
others, specific figures were requested, which were nor less of something than “maintain or improve fit-
later transferred to the code column. ness” is.
In short, different types of variable can be gener-
Missing data ated in the course of research. The three main types
The data for all 90 respondents are presented in Box are distinguished by looking at the relationship be-
8.2. Each of the 12 questions is designated initially as tween the categories of the variable as manifested by
a numbered variable (var00001, var00002, etc.); the any one individual. The three types, which pertain to
number corresponds to the question number in Box different levels of measurement, are as follows:
8.1 (var00001 is question 1, var00002 is question 2,
etc.). An important issue in data analysis is how to • Nominal variables. These variables, also known
handle the “missing data” generated when respon- as categorical variables, are composed of cate-
dents do not reply to a question—whether because gories that have no relationship to one another
they accidentally skip it, because they do not want to except that they are different. In the case of
answer it, or because it does not apply to them. For religion, for example, the categories might be
example, respondent 24 failed to answer question 2 Roman Catholic, Protestant, Jewish, Muslim,
on age. This was coded as 999, a number chosen be- Hindu, Other, and No Religion. Any two re-
cause it cannot be mistaken for a person’s age and is search participants will be either in the same
also easy to remember. Also, question 9 has a large category or in different categories: no other
number of 8s, because for this variable an 8 means kind of comparison is possible. This means
“not applicable”—many people did not answer this that the order of the categories is arbitrary: it
question, having been filtered out by the previous could have begun with No Religion and ended
one (that is, they do not have other sources of regular with Roman Catholic. Switching the order
exercise). Everyone answered questions 10, 11, and has no implications for the way the data are
12, so there are no missing data for those variables; interpreted.
if there were, it would be necessary to code the miss- • Ordinal variables. With these variables the
ing data with another number that could not be mis- different categories can be rank ordered. That
taken for an actual number of minutes. The various means that “greater than” (>) and “less than”
missing data codes are stored in the information for (<) statements can be made about the catego-
each variable so they cannot be read by the computer ries and the people in them. For example, in
as anything other than missing data. question 5 “always” indicates greater frequency

bel65796_ch08_177-204 179 01/13/22 12:26 PM

180 PART II Quantitative Research

Box 8.1 Part of a completed questionnaire on use of a gym

Questionnaire Code
1. What is your gender? (please tick)
   Male   ✓   1
   Female      2
Other (please specify)                  3
2. How old are you?
     21 years 21
3. Which of the following best describes your main reason for going to the gym? (please circle one
code only)
Relaxation 1
Maintain or improve fitness   ✓   2
Lose weight 3
Meet others 4
Build strength 5
Other (please specify) 6
4. When you go to the gym, how often do you use the cardiovascular equipment (treadmill, step
machine, bike, elliptical, rower)? (please tick)
Always   ✓   1
Usually 2
Rarely 3
Never 4
5. When you go to the gym, how often do you use the weights? (please tick)
Always   ✓   1
Usually 2
Rarely 3
Never 4
6. Generally, how frequently do you go to the gym? (please tick)
Every day 1
4–6 days a week 2
2 or 3 days a week   ✓   3
Once a week 4
2 or 3 times a month 5
Once a month 6
Less than once a month 7
7. Do you usually go with someone else to the gym or usually on your own? (please circle one code
only)
On my own   ✓   1
With a friend 2
With a partner/spouse 3
8. Do you have sources of regular exercise other than the gym?
Yes No   ✓      1
If you have answered No to this question, please proceed to question 10. 2

bel65796_ch08_177-204 180 01/13/22 12:26 PM

 8 Quantitative Data Analysis 181

Questionnaire Code
9. If you have replied Yes to question 8, please indicate the main source of regular exercise in the last
six months from this list. (please circle one code only)
Casual or organized competitive sports 1
Recreational cycling 2
Jogging 3
Walking 4
Other (please specify) 5
10. During your last visit to the gym, how many minutes did you spend on the cardiovascular
equipment (treadmill, step machine, bike, elliptical, rower)?
  33   minutes 33
11. During your last visit to the gym, how many minutes did you spend on weights?
  17   minutes 17
12. During your last visit to the gym, how many minutes did you spend on other activities (e.g.,
stretching exercises)?
  5   minutes 5

than “usually,” which indicates greater fre- spends 9 minutes on the equipment. At the in-
quency than “rarely,” and so on. Note also that terval/ratio level of measurement, it is possible
it would be illogical to re-order the categories to say that two scores are equal or not equal (as
as, say, “always, never, sometimes, usually.” at the nominal level), for example, 32 ≠ 31; that
   The distance or amount of difference between they have greater or lesser values (as at the or-
the categories, however, may not be equal across dinal level), for example, 32 > 31, 8 < 9; and that
the range. Thus, the difference between the cat- the distance or amount of difference between
egories “always” and “usually” is probably not them can be specified, for example, 32 − 31 = 1.
the same as the difference between “usually” This is the highest level of measurement, the
and “rarely,” and so on. This is because no one that allows the widest range of analysis
unit of measurement is used for this question. techniques, including addition, subtraction,
Ordinal variables are similar to words that have multiplication, and division.
either a comparative “er” or a superlative “est”    Sometimes a distinction is made between
attached to them, as in “taller” and “tallest.” interval and ratio variables. Ratio variables
• Interval/ratio variables. These are variables for are interval variables with a fixed and non-­
which a unit of measurement exists and thus arbitrary zero point (as in a wind speed of 0
the distances or amount of difference between kilometres per hour). Many social science vari-
the categories can be made identical across the ables exhibit this quality, for example, income,
range of categories. The values take the form age, and years of school completed. In the
of actual numbers (such as 1, 2, 7.5, etc.). In discussions that follow we will use the more
the case of variables var00010 to var00012, the inclusive term “interval/ratio,” but to under-
unit is the minute and the difference between stand, take two incomes: $30K and $45K. They
the categories is a one-minute interval. Thus, a are nominally unequal and the first is less than
person who spends 32 minutes on cardiovas- the second, in fact $15K less. But with that real
cular equipment is spending one minute more 0, one can also say that the first is 2/3 of the
than someone who spends 31. That difference second (or the second is 1.5 times the first). The
is the same as the difference between some- first income can also be said to stand in a 2:3
one who spends 8 minutes and another who ratio to the second.

bel65796_ch08_177-204 181 01/13/22 12:26 PM

182 PART II Quantitative Research

Box 8.2 Gym survey data

Case var00001 var00002 var00003 var00004 var00005 var00006 var00007 var00008 var00009 var00010 var00011 var00012

1 1 21 2 1 1 3 1 2 8 33 17 5
2 2 44 1 3 1 4 3 1 2 10 23 10
3 2 19 3 1 2 2 1 1 1 27 18 12
4 2 27 3 2 1 2 1 2 8 30 17 3
5 1 57 2 1 3 2 3 1 4 22 0 15
6 2 27 3 1 1 3 1 1 3 34 17 0
7 1 39 5 2 1 5 1 1 5 17 48 10
8 2 36 3 1 2 2 2 1 1 25 18 7
9 1 37 2 1 1 3 1 2 8 34 15 0
10 2 51 2 2 2 4 3 2 8 16 18 11
11 1 24 5 2 1 3 1 1 1 0 42 16
12 2 29 2 1 2 3 1 2 8 34 22 12
13 1 20 5 1 1 2 1 2 8 22 31 7
14 2 22 2 1 3 4 2 1 3 37 14 12
15 2 46 3 1 1 5 2 2 8 26 9 4
16 2 41 3 1 2 2 3 1 4 22 7 10
17 1 25 5 1 1 3 1 1 1 21 29 4
18 2 46 3 1 2 4 2 1 4 18 8 11
19 1 30 3 1 1 5 1 2 8 23 9 6
20 1 25 5 2 1 3 1 1 1 23 19 0
21 2 24 2 1 1 3 2 1 2 20 7 6
22 2 39 1 2 3 5 1 2 8 17 0 9
23 1 44 3 1 1 3 2 1 2 22 8 5
24 1 999 1 2 2 4 2 1 4 15 10 4
25 2 18 3 1 2 3 1 2 1 18 7 10
26 1 41 3 1 1 3 1 2 8 34 10 4
27 2 38 2 1 2 5 3 1 2 24 14 10
28 1 25 2 1 1 2 1 2 8 48 22 7
29 1 41 5 2 1 3 1 1 2 17 27 0
30 2 30 3 1 1 2 2 2 8 32 13 10
31 2 29 3 1 3 2 1 2 8 31 0 7
32 2 42 1 2 2 4 2 1 4 17 14 6
33 1 31 2 1 1 2 1 2 8 49 21 2
34 2 25 3 1 1 2 3 2 8 30 17 15
35 1 46 3 1 1 3 1 1 3 32 10 5
36 1 24 5 2 1 4 1 1 2 0 36 11
37 2 34 3 1 1 3 2 1 4 27 14 12
38 2 50 2 1 2 2 3 2 8 28 8 6
39 1 28 5 1 1 3 2 1 1 26 22 8
40 2 30 3 1 1 2 1 1 4 21 9 12
41 1 27 2 1 1 2 1 1 3 64 15 8
42 2 27 2 1 2 4 2 1 4 22 10 7
43 1 36 5 1 1 3 2 2 8 21 24 0
44 2 43 3 1 1 4 1 2 8 25 13 8

bel65796_ch08_177-204 182 01/13/22 12:26 PM

 8 Quantitative Data Analysis 183

Case var00001 var00002 var00003 var00004 var00005 var00006 var00007 var00008 var00009 var00010 var00011 var00012

45 1 34 2 1 1 3 2 1 1 45 15 6
46 2 27 3 1 1 2 1 1 4 33 10 9
47 2 38 2 1 3 4 2 2 8 23 0 16
48 1 28 2 1 1 3 3 1 2 38 13 5
49 1 44 5 1 1 2 1 2 8 27 19 7
50 2 31 3 1 2 3 2 2 8 32 11 5
51 2 23 2 1 1 4 2 1 1 33 18 8
52 1 45 3 1 1 3 1 1 2 26 10 7
53 2 34 3 1 2 2 3 2 8 36 8 12
54 1 27 3 1 1 2 3 1 3 42 13 6
55 2 40 3 1 1 2 2 1 4 26 9 10
56 2 24 2 1 1 2 1 1 2 22 10 9
57 1 37 2 1 1 5 2 2 8 21 11 0
58 1 22 5 1 1 4 1 1 1 23 17 6
59 2 31 3 1 2 3 1 1 4 40 16 12
60 1 37 2 1 1 2 3 2 8 54 12 3
61 2 33 1 2 2 4 2 2 8 17 10 5
62 1 23 5 1 1 3 1 1 1 41 27 8
63 1 28 3 1 1 3 3 2 8 27 11 8
64 2 29 2 1 2 5 2 1 2 24 9 9
65 2 43 3 1 1 2 1 2 8 36 17 12
66 1 28 5 1 1 3 1 1 1 22 15 4
67 1 48 2 1 1 5 1 1 4 25 11 7
68 2 32 2 2 2 4 2 2 8 27 13 11
69 1 28 5 1 1 2 2 2 8 15 23 7
70 2 23 2 1 1 5 1 1 4 14 11 5
71 2 43 2 1 2 5 1 2 8 18 7 3
72 1 28 2 1 1 4 3 1 2 34 18 8
73 2 23 3 1 1 2 1 2 8 37 17 17
74 2 36 1 2 2 4 2 1 4 18 12 4
75 1 50 2 1 1 3 1 1 2 28 14 3
76 1 37 3 1 1 2 2 2 8 26 14 9
77 2 41 3 1 1 2 1 1 4 24 11 4
78 1 26 5 2 1 5 1 1 1 23 19 8
79 2 28 3 1 1 4 1 2 8 27 12 4
80 2 35 2 1 1 3 1 1 1 28 14 0
81 1 28 5 1 1 2 1 1 2 20 24 12
82 2 36 2 1 1 3 2 2 8 26 9 14
83 2 29 3 1 1 4 1 1 4 23 13 4
84 1 34 1 2 2 4 2 1 8 24 12 3
85 1 53 2 1 1 3 3 1 1 32 17 6
86 2 30 3 1 1 4 1 2 8 24 10 9
87 1 43 2 1 1 2 1 1 2 24 14 10
88 2 26 5 2 1 4 1 1 1 16 23 7
89 2 44 1 1 1 4 2 2 8 27 18 6
90 1 45 1 2 2 3 3 2 8 20 14 5

bel65796_ch08_177-204 183 01/13/22 12:26 PM

184 PART II Quantitative Research

TABLE 8.1 Types of variable

Type Description Examples in gym study Variable name in SPSS
Nominal Variables whose categories cannot be var00001 gender
rank ordered; also known as categorical var00003 reasons
var00007 accomp
var00008 othsourc
var00009 exercise

Ordinal Variables whose categories can be rank var00004 carduse

ordered but the distances between the var00005 weiuse
categories are not equal across the range var00006 frequent

Interval/ratio Variables where the distances between var00002 age

the categories are identical across the var00010 cardmins
range var00011 weimins
var00012 othmins

The three main types of variable and illustra- Are there more than two categories?

tions from the gym survey are provided in Table 8.1.

Strictly speaking, items that have Likert-style re-
sponse categories (see Box 4.1) produce ordinal vari- Yes No Variable is dichotomous

ables. However, many writers argue that they can be

treated as though they produce interval/ratio vari-
Can the categories be rank ordered?
ables because of the relatively large number of cat-
egories they generate. For a brief discussion of this
issue, see Bryman and Cramer (2001, pp. 58–59). Yes No Variable is nominal/
Figure 8.1 provides guidance about how to identify categorical
variables of each type.
Are the distances between the categories equal?

Univariate analysis
Univariate analysis refers to the examination of one
Yes No Variable is ordinal
variable at a time. This section will outline several
common approaches.
Variable is interval/
ratio
Frequency tables
FIGURE 8.1 Deciding how to categorize a variable
A frequency table provides the number and percent-
age of people (or whatever else constitutes a case in
the study) belonging to each of the categories of the
variable in question and can be created for all three and that they represent 36.7 per cent (percentages
variable types. An example of a frequency table for a are often rounded in frequency tables) of the entire
nominal variable is provided for var00003 (reasons) sample.
in Table 8.2. The table shows, for example, that 33 Frequency tables are typically created using soft-
members of the sample go the gym to lose weight ware. A very commonly used software program for

bel65796_ch08_177-204 184 01/13/22 12:26 PM

 8 Quantitative Data Analysis 185

TABLE 8.2 Frequency table showing reasons for visiting the gym
REASONS FOR VISITING
Frequency Per cent Valid per cent Cumulative per cent
Valid Relaxation 9 10.0 10.0 10.0
Fitness 31 34.4 34.4 44.4
Lose weight 33 36.7 36.7 81.1
Build strength 17 18.9 18.9 100.0
Total 90 100.0 100.0

statistical analysis is IBM SPSS Statistics (SPSS). In to group people in terms of age ranges would mean
the Appendix at the end of this book you will find 40 different categories (one for each year from 18 to
instructions for using SPSS to perform the data anal- 57), which is too many to meaningfully describe.
yses discussed in this chapter. By creating five categories, you make the distri-
When an interval/ratio variable (such as age) is put bution of ages easier to comprehend. Notice that
in frequency table format, some of the categories may the sample totals 89 people, and that the “Valid
be combined in some way: 20–29, 30–39, and so on. per cent” column is based on a total of 89 rather
Be sure that the categories you create don’t overlap (as than 90 because this variable contains one miss-
in 20–30, 30–40, etc.): if they do, you’re likely to end ing value (from respondent 24). Use the “Valid
up with some 30-year-olds in the first group and some per cent” column to cite percentages because the
in the second. This would violate the mutually exclu- “per cent” column includes missing values in the
sive rule, discussed in Chapter 5, that no one should calculations.
be able to fall into more than one category (required
for telling whether two cases are equal or not equal). Diagrams
Also recall from Chapter 5 the exhaustive rule: every Diagrams are sometimes used to display quanti-
case must have a category into which it can be placed, tative data. With nominal or ordinal variables, the
even if it is only “missing data” or “not applicable.” bar chart and the pie chart are two of the easiest to
Table 8.3 shows an example of a frequency table use. A bar chart of the data in Table 8.2 is presented
for an interval/ratio variable: var00002 (age). Not in Figure 8.2. The height of each bar represents the

TABLE 8.3 Frequency table showing ages of gym members

AGES OF GYM VISITORS
Frequency Per cent Valid per cent Cumulative per cent
Valid 20 and below 3 3.3 3.4 3.4
21–30 39 43.3 43.8 47.2
31–40 23 25.6 25.8 73.0
41–50 21 23.3 23.6 96.6
51 and over 3 3.3 3.4 100.0
Total 89 98.9 100.0
Missing User 1 1.1
Total 90 100.0

bel65796_ch08_177-204 185 01/13/22 12:26 PM

186 PART II Quantitative Research

number of people in each category. Figures 8.2, 8.3, 50

and 8.4 were produced with SPSS.
Another way of displaying the same data is in a pie 40
chart (Figure 8.3). This type of diagram also shows
the size of the different categories but more clearly 30

Count
brings out the size of each relative to the total sample.
The percentage of the whole sample that each slice 20

represents is also given in this diagram.

To display an interval/ratio variable such as 10

var00002 (age), a histogram is often employed. Figure

8.4 uses the same data and categories as Table 8.3. 0
20 and 21–30 31–40 41–50 51 and
As in a bar chart, the bars represent the relative size under over
of each of the age bands, but there is no space be- Ages of gym visitors

tween the bars. This is to distinguish histograms, FIGURE 8.4 Histogram showing the ages of gym
which show interval/ratio variables, from bar charts, visitors (SPSS output)
which are used with nominal and ordinal variables.

More univariate analysis

40
Measures of central tendency
Measures of central tendency provide, in one
30
number, a typical or “average” score for a distribu-
tion or group of scores. Three measures of central
tendency are commonly used: the mode, the median,
Count

20
and the mean. Whether a particular one can be used
depends on the level of measurement.
10

• Mode. This is the value that occurs most

0 frequently in a distribution. The mode for
Relaxation Fitness Lose weight Build strength var00002 (age) is 28, meaning that there are
Main reasons for visiting the gym more 28s than any other score. The mode can
FIGURE 8.2 Bar chart showing main reasons for be used with all types of variable but it is most
visiting the gym (SPSS output) applicable to nominal data.
• Median. This is the mid-point in a distribution
of scores, derived by arraying all the scores in
Relaxation
Build strength 10.0 %
order (typically from the lowest to the highest)
18.9% and then finding the middle one. If there is
an even number of values, the median is cal-
culated by taking the mean of the two middle
numbers in the distribution. In the case of
Fitness
34.4% var00002, the median is 31. The median can
be used with both interval/ratio and ordinal
Lose weight variables. It cannot be used with nominal data
36.7%
because those sorts of scores cannot be rank
ordered.
FIGURE 8.3 Pie chart showing main reasons for • Mean. This is the average as it is understood in
visiting the gym (SPSS output) everyday use: that is, the sum of all numbers

bel65796_ch08_177-204 186 01/13/22 12:26 PM

 8 Quantitative Data Analysis 187

in a distribution, divided by the number of the mean. It is calculated by taking the difference
scores. The mean for var00002 is 33.6, mean- between each value in a distribution and the mean,
ing that the average age in our sample of gym squaring it, dividing the total of these squared dif-
visitors is about 34. This is slightly higher than ferences by the number of values, and then taking
the median because a few considerably older the square root. The standard deviation for var00010
members (especially respondents 5 and 10) (cardmins) is 9.9 minutes and for var00011 (weimins)
inflate it. The mean is vulnerable to such out- it is 8.0 minutes. This indicates that not only is the
liers (extreme values at either end of the dis- average amount of time spent on the cardiovascu-
tribution), which exert considerable upward lar equipment (26.5 minutes) higher than for the
or downward pressure on the mean. In such weights (14.9 minutes), there is more deviation from
instances the median is recommended, as it the mean too. The standard deviation is also affected
is not affected in this way. (Some researchers by outliers, which are sometimes excluded.
exclude outliers and then calculate the mean,
noting the removal in the text.) Bivariate analysis
Bivariate analysis examines whether there is a rela-
tionship between two variables. Several techniques
Measures of dispersion are available for examining relationships, but their
The amount of variation in a sample—that is, its use depends on the level of measurement of the two
dispersion—can be just as important as its typical variables being analyzed. Figure 8.5 details the main
value. Two in-class tests can have the same mean types of bivariate analysis to be discussed below,
of 60 per cent, but on one most people scored be- classified according to the type of variable involved.
tween 50 and 70, while on the other the grades are
evenly dispersed from the low 20s to 100 per cent. Contingency tables
(In the language of testing, the first test would be Contingency tables are probably the most flexible of
described as one that does not discriminate: the all methods of analyzing relationships in that they
brightest and least bright students get fairly close can be employed in relation to any pair of variables,
marks.) In the gym study, is there more or less vari- from nominal to interval/ratio. They are not the most
ability in the amount of time spent on cardiovascu- efficient method, however, especially for interval/
lar equipment compared with weights? ratio data, which is the reason why the method is not
The most obvious way of measuring dispersion is recommended in all the cells in Figure 8.5. A con-
by calculating the range. This is simply the differ- tingency table is like a frequency table except that it
ence between the maximum and the minimum value allows two variables to be analyzed simultaneously,
in a distribution of interval/ratio scores. (Range can so that relationships between them can be exam-
be applied to an ordinal variable too, but in that case ined. It is normal for contingency tables to include
it is more definitional than descriptive: for exam- percentages, since they make the tables easier to in-
ple, a range from “Working class” to “Upper class.”) terpret. Table 8.4 examines the relationship between
The range is 64 minutes for the cardiovascular ma- gender and reasons for visiting the gym using our
chines and 48 minutes for the weights. This suggests survey data. “Gender” is the independent variable
that there is more variability in the amount of time (presumed cause) and for that reason becomes the
spent on the former, probably because some people column variable—a common preference among re-
spend almost no time on cardiovascular equipment. searchers. “Reasons” is the dependent variable (pre-
However, like the mean, the range is influenced by sumed effect) and is situated as the row variable (but
outliers, such as respondent 41, who spent 64 min- see Box 8.3). In this case, gender is assumed to influ-
utes on the cardio equipment. ence reasons for going to the gym; reasons for going
Another measure of dispersion is the standard to the gym cannot influence gender. The percentages
deviation, which is a measure of variation around are column percentages, representing the independent

bel65796_ch08_177-204 187 01/13/22 12:26 PM

188 PART II Quantitative Research

Nominal Ordinal Interval/ratio

Nominal Contingency table + chi-square ( χ2) Contingency table + Contingency table + chi-square ( χ2)
+ Cramér’s V chi-square ( χ )
2
+ Cramér’s V. If the interval/ratio
+ Cramér’s V variable can be identified as the
dependent variable, compare
means with eta.
Ordinal Contingency table + chi-square ( χ2) Kendall’s tau-b Kendall’s tau-b
+ Cramér’s V
Interval/ratio Contingency table + chi-square ( χ2) Kendall’s tau-b Pearson’s r
+ Cramér’s V. If the interval/ratio
variable can be identified as the
dependent variable, compare means
with eta.

FIGURE 8.5 Methods of bivariate analysis, level of measurement of one variable on top, that of the other
variable on side

TABLE 8.4 Contingency table Pearson’s r

showing the relationship between Pearson’s r is a statistic used to examine relation-
gender and reasons for visiting the gym ships between two interval/ratio variables. Its chief
GENDER features are as follows:
Male Female
Reasons No. % No. % • The coefficient has values from 0 (which indi-
Relaxation 3 7 6 13 cates that there is no linear relationship what-
Fitness 15 36 16 33 ever between the two variables) to +1 or −1 (the
Lose weight 8 19 25 52 first indicating a perfect positive relationship,
the second a perfect negative).
Build strength 16 38 1 2
• The closer a positive coefficient is to 1, the
Total 42 100 48 100
stronger the relationship; the closer it is to
zero, the weaker the relationship.
variable. The number in each cell is calculated as a
• Similarly, the closer a negative coefficient is to
percentage of the total number in its column. Thus,
−1, the stronger the relationship is; the closer it
in the case of the top left-hand cell, the 3 men—of 42
is to zero, the weaker the relationship.
in all—who go to the gym for relaxation represent
• The sign of the coefficient (positive or nega-
3/42 or 7 per cent of the men in the sample.
tive) indicates the direction of a relationship.
Contingency tables are generated to look for
Negative means that as one variable is going
patterns of association. In this case, there are clear
up, the other is going down; it doesn’t matter
gender differences in reasons for visiting the gym.
which is which. Positive means that the two are
As our student anticipated, females are much more
going in the same direction, either both up or
likely than males to go to the gym to lose weight. They
both down.
are also somewhat more likely to go to the gym for
relaxation. By contrast, men are much more likely to
go to the gym to build strength. There is little gender For Pearson’s r to be used, the relationship between
difference in terms of fitness as a reason. In a real the two variables must be broadly linear: when plot-
research situation, the investigator would take these ted on a scatter diagram, the values of the two vari-
results and interpret them in terms of some social ables will approximate a straight line (even though
theory. they may be scattered as in Figure 8.9), not a curve.

bel65796_ch08_177-204 188 01/13/22 12:26 PM

 8 Quantitative Data Analysis 189

Box 8.3 Relationships not causality

Analyzing associations between variables re-
veals relationships but not necessarily causes
(although examining associations helps in the
search for causes). Similarly, the direction of
causation cannot be determined by merely es-
tablishing that a relationship exists between two
variables. Indeed, in some cases what appears

FOTOGRAFIA INC./iStockphoto
to be a causal influence working in one direc-
tion may actually work the other way round.
An example of the problem of causal direction
was presented in Chapter 4, where Sutton and
Rafaeli (1988) expected to find that a display
of positive emotions (for example, smiling or
The old adage “practice makes perfect” suggests
friendliness) by retail checkout staff caused in-
causality. Although this young woman regularly
creased sales. In fact, the relationship appeared
practises her painting technique, are there vari-
to flow in the opposite direction: levels of retail ables other than practice that could influence her
sales seemed to exert a causal influence on the success as a painter?
display of emotions. The more sales, the busier
the staff, and the less time and inclination they be caused by a third variable. For example,
had to smile! Rippeyoung (2013) examined whether the effects
At times one may feel confident in inferring a of breastfeeding can explain the gaps in cogni-
causal direction: the relationship between age tive ability between “non-poor,” “near poor,” and
and voting behaviour, for example. It is impos- “poor” children in Canada (a purportedly causal
sible for the way people vote to influence their relationship), or whether a third factor, a rich
age; so, if the two variables are related, age is educational home environment, influences both
the independent variable. It is not uncommon breastfeeding and cognitive abilities. She found
for researchers, when analyzing their data, to greater support for the spurious effect position,
draw inferences about causal direction based observing that there were no significant differ-
on assumptions of this kind. However, some re- ences in breastfeeding rates between the three
lationships may be spurious or non-causal: that economic groups. Spurious relationships are dis-
is, the relationship between two variables may cussed in more detail later in this chapter.

A scatter diagram should be created to test for lin- five interval/ratio variables. The scatter diagram
earity before Pearson’s r values are used in a study. for variables 1 and 2 is presented in Figure 8.6 and
shows a perfect positive relationship, which yields a
Scatter diagrams Pearson’s r correlation of +1. This means that as one
Scatter diagrams are shown in Figures 8.6 through variable increases, the other variable increases as
8.9. If one variable can be identified as likely to be well, and the value of one variable perfectly predicts
the independent variable, it is by convention placed the value of the other. If the correlation is below 1, at
on the X-axis, the horizontal axis. least one other variable is affecting the relationship
The scatter diagrams are based on the infor- between them, possibly affecting variable 1 as well
mation in Box 8.4, which gives imaginary data for as variable 2.

bel65796_ch08_177-204 189 01/13/22 12:26 PM

190 PART II Quantitative Research

12
Box 8.4  Imaginary data from
10 five variables to show
different types of
8
relationship
Variable 1

6 Variables
1 2 3 4 5
4
1 10 50 7 9
2 2 12 45 13 23
3 14 40 18 7
0
4 16 35 14 15
0 10 20 30
Variable 2 5 18 30 16 6
FIGURE 8.6 Scatter diagram showing a 6 20 25 23 22
­ erfect positive relationship
p 7 22 20 19 12
8 24 15 24 8

60 9 26 10 22 18
10 28 5 24 10
50

40
If, as in Figure 8.8, there is no apparent pattern in
Variable 3

30 the scatter diagram, there is no or virtually no cor-

relation between the variables. In the case of vari-
20 ables 2 and 5, the correlation is close to zero, at −.041.
This means that the variation in the dependent vari-
10
able is probably associated with variables other than
0 the one used in this analysis.
0 10 20 30 If a relationship is strong, a clear pattern will be
Variable 2
evident in the scatter diagram. This is the case with
FIGURE 8.7 Scatter diagram showing a perfect
negative relationship
30

The scatter diagram for variables 2 and 3

(Figure 8.7) shows a perfect negative relationship,
which yields a Pearson’s r correlation of −1. This 20
Variable 5

means that as one variable increases, the other de-

creases, and that the values of one variable perfectly
predict the values of the other. Here, as in the previ- 10
ous case, the line could be extended beyond the data
to predict what would happen to either variable if
the other moved beyond the range shown, although
0
it is risky to go beyond what is known. Also, because 0 10 20 30
social data usually include so many exceptions that Variable 2
few cases fall exactly on the line, predictions based FIGURE 8.8 Scatter diagram showing two
on them tend to be general rather than specific. variables that are not related

bel65796_ch08_177-204 190 01/13/22 12:26 PM

 8 Quantitative Data Analysis 191

30 Going back to the gym survey, the correlation be-

tween age (var00002) and the time spent on weights,
weimins (var00011), is −.27 (see Table 8.5): a weak
20
negative relationship. This suggests that older people
tend to spend less time on such equipment than
Variable 4

younger ones, but that other variables also influence

the time people spend there.
10
Kendall’s tau-b
Kendall’s tau-b is designed for pairs of ordinal vari-
0 ables, but is also used (as Figure 8.5 suggests) when
0 10 20 30 one variable is ordinal and the other is interval/ratio.
Variable 2
(Notice the general rule that data can be moved
FIGURE 8.9 Scatter diagram showing a strong down a level, interval to ordinal, but not up.) It is
positive relationship
exactly the same as Pearson’s r in terms of possible
outcomes in that the computed value of Kendall’s
variables 2 and 4, whose scatter diagram appears in tau-b can be positive or negative, and varies from 0
Figure 8.9. There is clearly a positive relationship; in to ±1. In the gym study there are three ordinal vari-
fact, the Pearson’s r is .88 (positive correlations are ables: var00004 (carduse), var00005 (weiuse), and
usually presented without the + sign). This means var00006 (frequent; see Table 8.1). Using Kendall’s
that the variation in the two variables is very closely tau-b to calculate the correlation between the first
connected, but that other variables also have some two variables—frequency of use of the cardiovascu-
influence. lar and weights equipment—it turns out to be very

TABLE 8.5 Correlations output for age, weimins, and cardmins (SPSS output)
Correlations of p < 0.05 are “flagged” with asterisks

CORRELATIONS
AGE WEIMINS CARDMINS
  AGE Pearson Correlation 1.000 –.273 ** –.109
Sig. (2-tailed) . .010 .311
N 89 89 89
  WEIMINS Pearson Correlation −.273 ** 1.000 –.161
Sig. (2-tailed) .010 . .130
N 89 90 90
  CARDMINS Pearson Correlation −.109 –.161 1.000
Sig. (2-tailed) .311 .130 .
N 89 90 90
**
Correlation is significant at the 0.01 level (2-tailed).

Shows strength of Shows number of cases, less any Shows level of statistical
relationship as indicated cases for which there are missing significance of computed
by Pearson’s r data for either or both variables value of Pearson’s r

bel65796_ch08_177-204 191 01/13/22 12:26 PM

192 PART II Quantitative Research

low. A slightly stronger relationship is found be- the gym (var00003, reasons). The four means shown
tween var00006 (frequency of going to the gym) and suggest that people who go to the gym for fitness or
var00010 (cardmins, the amount of time spent on to lose weight spend considerably more time on this
the cardiovascular equipment): close to .4. Another equipment than those who go to the gym to relax or
option for relating two ordinal variables is to use to build strength.
Spearman’s rho (though it can only accommodate The statistic eta can be calculated with these vari-
a small number of ties: that is, cases with the same ables; it expresses the level of association between
rank on a variable). It ranges from −1, a perfect neg- them. Since one variable is nominal (meaning no rank
ative relationship, to +1, a perfect positive relation- ordering), its values are always positive; and it has a
ship. For example, look at the participation rank (A) range from 0 to 1. The eta for the data in Table 8.6 is
and popularity rank (B) of a group of seven people. .48, a moderate relationship between the two variables.

Person Mary John Bill Sally Kim Susan Joe Amount of explained variance
A 1 2 3 4 5 6 7 Squaring eta, Kendall’s tau-b, Spearman’s rho, and
B 1 2 3 4 5 6 7 Pearson’s r produces statistics that measure how
much of the variation in one variable can be ex-
Spearman’s rho is 1.0; knowing a person’s partici- plained or predicted by the other variable. Thus, if
pation rank perfectly predicts the popularity rank. It r is −.27 (the correlation between weimins and age),
is called a rank correlation coefficient to remind you r2 is .0729. This can be expressed as a percentage by
of its ordinal character and to distinguish it from multiplying r2 by 100: thus 7.29 per cent of the varia-
Pearson’s r. tion in the time spent on weights can be predicted by
age. This also shows that a strong correlation like .7
Cramér’s V explains only 49 per cent of the variance. For nom-
Cramér’s V is suitable for examining the strength of inal data, squaring Cramér’s V (.503 for gender and
a relationship between two nominal variables (see reasons) provides an approximation of the explained
Figure 8.5). Its coefficient ranges from 0 to 1. It is variance, so about 25 per cent of the variation in rea-
always positive because nominal categories cannot sons for visiting the gym is attributable to gender.
be rank ordered; their values cannot go up or down.
The value of Cramér’s V for the relationship be- Statistical significance and
tween gender and reasons for going to the gym inferential statistics
(var00003, reasons) is .503, a moderate relationship. One difficulty with working on sample data is a lin-
Cramér’s V is usually reported with a contingency gering worry about whether the findings can be gen-
table and a chi-square test (see below). It is not nor- eralized to the population from which the sample
mally presented on its own. was drawn. As we saw in Chapter 7, there is always
the possibility of sampling error, even when prob-
Comparing means and eta ability sampling procedures are followed (as in the
There are many other bivariate statistics but we will gym survey). If this happens, the sample is to some
present just one more. To examine the relationship degree unrepresentative of the wider population.
between an interval/ratio variable and a nominal To make matters worse, there is no practical way of
variable (if the latter can be construed as the inde- finding out for sure how extensive the sampling error
pendent variable), one can compare the means of is, as no one can afford the time or money to study
the interval/ratio variable for each subgroup of the every case in the population—that’s why a sample is
nominal variable. As an example, consider Table 8.6, used in the first place. This is where various tests of
which presents the mean number of minutes spent statistical significance come in.
on cardiovascular equipment (var00010, cardmins) What is a test of statistical significance? It pro-
for each of the four categories of reasons for going to vides an indication of the risk we are taking when we

bel65796_ch08_177-204 192 01/13/22 12:26 PM

 8 Quantitative Data Analysis 193

TABLE 8.6 Comparing subgroup means: Time spent on cardiovascular

­equipment by reasons for going to the gym
REASONS
Time Relaxation Fitness Lose weight Build strength Total
Mean number of minutes 18.33 30.55 28.36 19.65 26.47
spent on cardiovascular
equipment
N 9 31 33 17 90

use a particular sample statistic to estimate a popu- statistical significance are expressed as prob-
lation characteristic. For example, the mean age of ability levels: that is, the probability of reject-
the people in the gym sample is 33.6 years. Using the ing the null hypothesis when it is in fact true.
concept of the standard error of the mean, the 95 per The convention among most social researchers
cent confidence interval for the population mean ex- is that an acceptable level of statistical signif-
tends from 31.72 to 35.47 years of age. The risk that icance is p ≤ .05 (“p” stands for probability),
this range does not contain the population mean is 5 meaning that there are at most 5 chances in
per cent. Chapter 7 (see Box 7.2), in its discussion of 100 that the sample shows a relationship not
the standard error of the mean, revealed some of the also found in the population. However, the
ideas behind statistical significance when estimating maximum risk level the researcher is willing
a population mean. The rest of this section looks at to accept may vary, depending on the circum-
tests of significance for measures of bivariate associ- stances. If more stringent tests of significance
ation. All the tests have a common structure, illus- are desired, p ≤ .01 or p ≤ .001 may be used.
trated in the following steps. • Determine the statistical significance of the
findings. The relevant statistical test depends
• Set up a null hypothesis. A null hypothesis is a on the measure of association involved, and is
hypothesis that is to be disproved; for example, usually calculated with software, although sta-
that there is no association between two vari- tistics textbooks outline how significance tests
ables, or that two populations do not differ on can be calculated by hand with small data sets.
some characteristic. In the context of our gym • Decide whether to reject or not reject the null
survey, a null hypothesis might state that there hypothesis. The most common practice in
is no relationship between gender and reasons the social sciences is to reject the null if the
for visiting the gym. If your data lead to a re- findings are statistically significant at the .05
jection of the null hypothesis, you will have level—that is, if the risk of error (i.e., getting
indirect support for the research hypothesis a relationship as strong as the one that was
that there is a relationship in the population. found when in fact there is no relationship in
But there is always a chance that a rejected null the population), represented by the p-level, is
hypothesis is actually true. The amount of risk no higher than 5 in 100. Such results are un-
associated with rejecting a true null hypothesis likely to have occurred by chance alone.
is called the level of statistical significance.
• Establish an acceptable level of statistical sig- Two types of error are possible when inferring sta-
nificance. This is essentially the level of risk tistical significance (see Figure 8.10). A Type I error
associated with rejecting the null hypothesis occurs when a null hypothesis that is actually true
(implying that there is a relationship in the pop- is rejected. This means that the results are the prod-
ulation) when in fact the null is correct (there uct of chance, and that the researcher was mistaken
is no relationship in the population). Levels of in concluding that there was a relationship in the

bel65796_ch08_177-204 193 01/13/22 12:26 PM

194 PART II Quantitative Research

Error information about the likelihood that the correla-

Type I Type II tion exists in the population. For instance, with a
(risk of rejecting the (risk of failing to reject
null hypothesis when the null hypothesis when
Pearson’s r of −.62 in the sample, what is the likeli-
it should be confirmed) it should be rejected) hood that there is no relationship between the two
variables in the population? How likely is a −.62 to
0.05 Greater Lower occur by chance alone, that is, because of sampling
risk risk
error?
p level
Whether a correlation coefficient is statistically
0.01 Lower Greater
risk risk significant or not is affected by two factors:

FIGURE 8.10 Type I and Type II errors • the size of the computed coefficient
• the size of the sample

population. The level of significance is the proba- This second factor may appear surprising, but it is
bility of making a Type I error, so a p = .05 level of true for all statistics. The larger a sample, the more
significance means a greater likelihood of making a likely it is that a given sample result is statistically
Type I error than a p = .01 level of significance does. significant. Thus, even though the correlation be-
The second type of error occurs when the null hy- tween age and the amount of time spent on weight
pothesis should be rejected but is not. This is called a machines in the gym survey is just −.27—a fairly
Type II error. These errors are more likely when the weak relationship—it is statistically significant at the
significance level is .01 than when it is .05, because p = .01 level. This means that there is only one chance
using .01 makes it less likely that the null will be re- in 100 that there is no relationship between age and
jected. The two types of error cannot be minimized at weights in the population. Because the statistical sig-
the same time; if you decrease the chances of making nificance of a correlation coefficient depends so much
a Type I error, you increase the chances of making on the sample size, it is important always to examine
a Type II error, and vice versa. Researchers usually both the correlation coefficient and the significance
make the conservative choice and seek to minimize level. This is true for any calculated statistic.
Type I error; this makes it less likely that the null will This treatment of correlation and statistical sig-
be rejected in error, and hence less likely that the re- nificance applies to both Pearson’s r and Kendall’s
search hypothesis will be erroneously supported. tau-b. A similar interpretation can also be applied to
If an analysis reveals a statistically significant Cramér’s V and the chi-square test.
finding, this does not mean that the finding is im-
portant. It simply means that the results are prob- The chi-square test
ably not due to chance alone. For example, a study The chi-square (χ2) test is applied to contingency
may find a correlation between time spent jogging tables like Table 8.4. It is a measure of the likelihood
per week and the amount of money spent on run- that a relationship between two variables in a sample
ning shoes. Such a finding might be statistically would also be found in the population. The test works
significant, but is it important? Recall that statisti- by calculating for each cell in the table an expected
cal significance gets easier to achieve as sample size frequency—that is, one that would occur on the basis
goes up. Also, it is important to appreciate that tests of chance alone. Think of the days of the week and
of statistical significance can be conducted only on crime: one might expect that 14.29 per cent (1/7) of
probability samples. all crimes would occur on each day of the week. That
is what would be expected if there were no relation-
Correlation and statistical significance ship between day of the week and crime. The data
Examining the statistical significance of a correla- say otherwise; more crimes are committed on Friday
tion from a randomly selected sample provides and Saturday. The chi-square value, which for the

bel65796_ch08_177-204 194 01/13/22 12:26 PM

 8 Quantitative Data Analysis 195

Research in the News

Big data in Canada
“Big data” is a term that lacks a specific defini- democracies. What makes the matter ethically
tion. As the first word suggests, it refers to large complex is that there can be clear social benefits
data sets, some of which are so huge that con- to the use of big data, such as the contribution it
ventional data processing systems are unable can make to medical research, crime prevention,
to handle them. The expression also pertains to and national security.
large quantities of data gathered in ways that to According to CTV News (Jones, 2020), a
date are not commonly used by social scientists. Canadian company collected big data early
This would include largely electronic methods to in the COVID-19 pandemic that indicated the
acquire information on things like credit card use, extent to which people were abiding by gov-
cellphone use and location, email activity, bodily ernment recommendations to stay home.
functions as recorded by monitors strapped to Ostensibly, the purpose of gathering the in-
the body for prolonged periods, and above all formation was to assist government officials in
Internet use. Big data sources also tend to be un- reaching people who were not abiding by their
structured in that the data gathered must be put recommendations so the officials could help
in some meaningful or systematic configuration them stay safe, although critics have raised pri-
before they can be used, which is often accom- vacy concerns. The data were gathered using
plished with complex algorithms. location apps people downloaded onto their
If you’ve ever Googled a topic, say “cosmet- smartphones, although it’s not clear that those
ics” or “lawnmowers,” and then found advertise- who downloaded the apps knew the informa-
ments for those products on the websites you visit, tion could be provided to third parties. One pat-
that means than someone or some organization tern the data showed was that there appeared
has placed you in a big data set. That may be to be more people “out and about” in lower-in-
useful for you, or it may literally cause you grief. come neighbourhoods and rural areas as op-
A pregnant graduate student in Charlottetown posed to more affluent localities. The company
did an Internet search for topics relating to birth that gathered the information suggested that
and babies, but unfortunately lost her pregnancy this may reflect the fact that people with lower
after 27 weeks. Yet Internet ads for things like incomes often work in occupations such as con-
baby clothes and breast pumps kept coming, struction or in food, retail, or other service oc-
which worsened her agony (Anderssen, 2014). cupations that are not compatible with working
The ethics surrounding big data become even from home. As with other frontline workers, for
murkier when the techniques are used by gov- people in these jobs, obeying a stay-at-home
ernments for surveillance purposes, a prac- recommendation would be considered “a
tice that is surprisingly common even in liberal luxury” (Jones, 2020).

data in Table 8.4 is 22.726, is calculated by taking the here). The chi-square value means nothing by itself
differences between the actual and expected values and can be meaningfully interpreted only in rela-
for each cell in the table and then summing those tion to its associated level of statistical significance,
differences (the operation is slightly more compli- which in this case is p < .0001. This means that there
cated than this, but the details need not concern us is less than one chance in 10,000 of rejecting a true

bel65796_ch08_177-204 195 01/13/22 12:26 PM

196 PART II Quantitative Research

null hypothesis (that is, inferring that there is a rela- pertains only to whether the results may have re-
tionship in the population when none exists). We can sulted from chance alone, and says nothing about
be extremely confident that there is a relationship practical significance or importance. There is a big
between gender and reasons for visiting the gym practical difference between a situation in which
among all gym members, since the chance of obtain- half the respondents at all levels of sports partici-
ing a sample that shows this chi-square value when pation have low self-esteem and one in which half
there is no relationship among all gym members is have high self-esteem, but that makes no difference
less than 1 in 10,000. to the level of statistical significance—it remains
But a chi-square value is also affected by the the same in each situation. The practical implica-
number of cases. You want a large chi-square to tions of a finding go beyond its level of statistical
reject a null hypothesis, and a larger n makes this significance.
easier to achieve. This is why it is necessary to look
at the data and not just at the final statistic. Examine Comparing means and statistical
Table 8.7, which shows the level of self-esteem for significance
three categories of people (those who engage in con- A test of statistical significance called analysis of
tact sports, those who engage in non-contact sports, variance can be applied to the comparison of means
and those who do not engage in any sports). Because in Table 8.6. This procedure entails treating the total
the independent variable is nominal, chi-square is amount of variation in the dependent variable—
more appropriate than Kendall’s tau-b. Note the in- amount of time spent on cardiovascular equip-
formation in the bottom row of the table. ment—as made up of two types: variation within each
The first two parts of Table 8.7, A and B, show of the four subgroups that make up the independent
that just by doubling the size of the sample, the re- variable, and variation between them. The latter is
sults change from being very likely due to chance often called the explained variance (explained by the
in A (thus the null hypothesis of no relationship group one is in) and the former the error variance. A
cannot be rejected) to less than 5 chances in 100 test of statistical significance for the comparison of
that they occurred by chance in B, and thus a re- means entails relating the two types of variance to
jection of the null. This is problematic because it form what is known as an F statistic, which expresses
suggests that one could collect more and more data the amount of explained variance in relation to the
until the null can be rejected. Now look at the last amount of error variance. In the case of the data in
part of the table, C. Why was it included? When Table 8.6, the resulting F statistic is statistically sig-
the data from B for high and low self-esteem are nificant at the p < .001 level. This finding suggests
switched around in C, which in most instances that there is less than one chance in 1000 that there
should mean something, chi-square stays the same. is no relationship between the two variables among
This illustrates the point that statistical significance all gym members.

TABLE 8.7 How chi-square is affected by increasing the size of n and not
affected by changes to column headings
A B C
Self- Contact Non-contact No Contact Non-contact No Contact Non-contact No
esteem sports sports sports sports sports sports sports sports sports
H 7 3 10 14 6 20 24 26 50
M 5 10 15 10 20 30 10 20 30
L 12 13 25 24 26 50 14 6 20
χ = 1.93, not sig.
2
χ = 5.08, p < .05
2
χ = 5.08, p < .05
2

bel65796_ch08_177-204 196 01/13/22 12:26 PM

 8 Quantitative Data Analysis 197

Multivariate analysis Age

Multivariate analysis entails the simultaneous anal-
ysis of three or more variables, and can only be in-
troduced here. It is sometimes called elaboration as Income Voting behaviour
it is more complicated and creates a more valuable
picture than bivariate analysis: one rarely finds an FIGURE 8.11 A spurious relationship
adequate explanation of a particular phenomenon
Is there an intervening variable?
by looking at only one possible cause. (For more
information on the techniques, consult a textbook An intervening variable suggests that the relation-
on quantitative data analysis: for example, Agresti ship between the two original variables is not a
& Finlay, 2009; Noack, 2018.) There are three main direct one. Assume that there is a positive relation-
contexts in which multivariate analysis is used. Each ship between income and self-esteem—the higher
is explained below. one’s income, the more positive one feels about one-
self. But maybe there is more to it than that. Maybe
income affects overall levels of physical vibrancy
Is the relationship spurious? (richer people tend to be more physically fit and to
To establish a relationship between two variables, smoke less than poorer people), which in turn affects
not only must there be logical and temporal evidence self-esteem:
of a relationship but the relationship must be shown
to be non-spurious. A spurious relationship exists income → physical vibrancy → level of self-esteem
when there appears to be a relationship between two
variables, but the relationship is not real: it is being If this is true, what should happen to the relationship
produced because each variable is itself related to a between income and self-esteem if physical vibrancy
third variable. Think about the positive relationship is controlled? (To control means to hold constant.
between the number of fire engines at a fire and the Controlling in this instance might involve taking
fire damage: the more engines, the greater the devas- only people who had high levels of physical vibrancy,
tation. Does this positive association mean that the and seeing whether there is still a positive relation-
number of fire engines present causes the amount ship between income and self-esteem among them.)
of fire damage? Of course not. The size of the fire If physical vibrancy were an intervening variable,
accounts for both the number of engines respond- that would weaken or eliminate the association be-
ing and the damage. Variation in the two variables tween income and self-esteem. Also, notice the “con-
(number of fire engines and amount of damage) is trol” word? Yes, the purpose of controls is to make
caused by a third factor, the size of the fire. cross-sectional research more like an experiment
Consider a social science example. Assume there in which random assignment makes all other things
is a positive relationship between income and voting equal or controlled.
behaviour such that the more income a person has,
the more likely he or she is to vote for a conserva- Is there an interaction?
tive party. Does this mean that having more money If the relationship between two variables holds for
causes people to move to the right? Or could this re- some groups or situations but not for others, an in-
lationship be explained by age (see Figure 8.11)? The teraction exists. The word “interaction” is used in a
older one is, the more likely one is to have a higher statistical sense here, one that is different from its
salary; and the older one is, the more conservative everyday meaning. In statistical terms, if the effect
one tends to be. If the apparent relationship between of one independent variable varies at different levels
income and voting behaviour is in fact a function of of a second independent variable, there is an inter-
age, that relationship is spurious. See Box 8.3 for an- action. In the gym study, for example, is the rela-
other example. tionship between age and other sources of regular

bel65796_ch08_177-204 197 01/13/22 12:26 PM

198 PART II Quantitative Research

exercise (var00008, othsourc) different for men and Other uses for multivariate analysis
women? Table 8.8 shows the relationship between Multivariate analysis can also be used to determine
age and other sources of exercise and includes both how much of the variation in the dependent variable
men and women. (Age has been broken down into is explained or predicted by the independent vari-
just three age bands to make the table easier to read.) ables. In addition, it provides a test to find out which,
The table suggests that the 31-to-40 age group is if any, of the independent variables are significant
less likely to have other sources of regular exercise predictors after controlling for the others. These uses
than the 30-and-under and 41-and-over age groups. will be illustrated by examining multiple linear re-
However, Table 8.9, which breaks the relationship gression. Our brief discussion of this topic will also
down by gender, suggests that the patterns for males acquaint you with some key concepts in multivariate
and females are somewhat different. Among males, analysis.
the pattern shown in Table 8.9 is very pronounced Consider the variable weimins, the number of
(the 31-to-40 age group is less likely to have other minutes spent on weights. What might cause it to
sources of exercise than the other two age groups), vary? One relevant factor might be age. Perhaps the
but for females the likelihood of having other sources older people get, the less likely they are to work out
of exercise declines with age. We can say that the using weights. To test this idea, we could perform a
relationship between age and having other sources bivariate regression with weimins as the dependent
of exercise is moderated by gender. Another way to variable and age as the independent variable.
put this is that there is an interaction between age The first part of Table 8.10 shows an R-square
and gender in predicting whether gym visitors have value of .074, which indicates that age by itself ex-
other sources of regular exercise. plains 7.4 per cent of the variation in weimins—not

TABLE 8.8 Contingency table showing the relationship between age and
whether gym visitors have other sources of regular exercise (percentages)
AGE
Other source of exercise 30 and under 31–40 41 and over
Other source 64 43 58
No other source 36 57 42
N 42 23 24

TABLE 8.9 Contingency table showing the relationship between age

and whether gym visitors have other sources of regular exercise for males and
females (percentages)
GENDER
Male Female
Other source 30 and under 31–40 41 and over 30 and under 31–40 41 and over
of exercise
Other source 70 33 75 59 50 42
No other 30 67 25 41 50 58
source
N 20 9 12 22 14 12

bel65796_ch08_177-204 198 01/13/22 12:26 PM

 8 Quantitative Data Analysis 199

TABLE 8.10 SPSS output showing results of bivariate regression

MODEL SUMMARY
Model R R-Square Adjusted R-Square Std. Error of the Estimate
1 .273 a
.074 .064 7.72123
a
Predictors: (Constant), age

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 417.236 1 417.236 6.999 .010b
Residual 5186.719 87 59.617
Total 5603.955 88
a
Dependent variable: weimins
b
Predictors: (Constant), age

COEFFICIENTSa
Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 23.209 3.218 7.213 .000
age −.245 .093 −.273 −2.645 .010
a
Dependent variable: weimins

very much. The value of R-square tends to be some- effect of independent variables that may be measured
what inflated, especially if the sample size is small (as in different units: the higher the standardized coeffi-
in our example) or the number of independent vari- cient, the more important the independent variable
ables is large (Agresti & Finlay, 2009, p. 366). A more is considered to be. Finally, the significance level for
realistic estimate is the adjusted R-square, which in age is shown to be .010, indicating that the proba-
our case has a value of .064 or 6.4 per cent. bility that the association between age and weimins
The ANOVA section of the output indicates a is due to chance alone is 1 in 100: this suggests it’s
significance or p-value of .010, which is the signif- a pretty safe bet that there is a relationship between
icance level of the model as a whole. Our model at those two variables in the population.
this point consists of only one independent variable As we have noted, a single cause or independent
(age). The final section of the output, “Coefficients,” variable is rarely if ever enough to provide a good ex-
has two subsections: “Unstandardized coefficients” planation for anything. Let’s see what happens if we
and “Standardized coefficients.” The unstandardized put a second independent variable—gender—into
coefficient for age is −.245, which represents the esti- the model. Normally, the variables used in linear
mated change in the dependent variable for each unit regression have to be at the interval/ratio level of
change in the independent variable. In our example, measurement, but a special procedure allows us to
the model predicts that for each one-year increase in use nominal variables like gender as well. Such vari-
age, the time spent on weights decreases by .245 min- ables have to be converted to indicator or “dummy”
utes. The standardized coefficient of −.273 indicates variables, a process that goes beyond the scope of our
that for every standard deviation increase in age, the discussion here. Assume that the dummy variable we
time spent on weights decreases by .273 standard de- have created is called gender2.
viation units. Standardized coefficients (sometimes Table 8.11 shows the output generated after
called beta weights) are useful in comparing the gender2 was added to the model. The “Model

bel65796_ch08_177-204 199 01/13/22 12:26 PM

200 PART II Quantitative Research

TABLE 8.11 Multiple regression output with gender2 (gender dummy variable)
in the model
MODEL SUMMARY
Model R R-square Adjusted R-square Std. Error of the Estimate
1 .478a .228 .210 7.09088
a
Predictors: (Constant), gender2, age

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 1279.830 2 639.915 12.727 .000b
Residual 4324.125 86 50.281
Total 5603.955 88
a
Dependent variable: weimins
b
Predictors: (Constant), gender2, age

COEFFICIENTSa
Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 20.831 3.010 6.920 .000
age −.260 .085 −.289 −3.053 .003
gender2 6.251 1.509 .393 4.142 .000
a
Dependent variable: weimins

Summary” shows that the R-square value has in- for age is .003, which means that age remains signifi-
creased to .228 (or that 22.8 per cent of the variation cant even when gender2 is held constant. Had the sig-
in weimins can be explained or predicted by the new nificance level exceeded .050, we would have had to
model), and that the more conservative adjusted consider dropping age from the model.
R-square has risen to .210 (or 21.0 per cent). This The “Coefficients” section shows an unstandard-
indicates that by adding gender2 we have produced ized coefficient of 6.251 for gender2. Because of the
a much better model, although the majority of the way this dummy variable was coded, the coefficient
variation in weimins is still unaccounted for. The means that, holding age constant, the model esti-
ANOVA significance level is .000, which indicates mates that the average time men spend on weights
that the findings for the new model as a whole (which is 6.251 minutes greater than that spent by women.
includes both age and gender2 as independent vari- The standardized coefficient of .393 tells us that ac-
ables) have a probability of less than 1 in 1000 of cording to the model, the average time men spend on
being due to chance alone. weights is .393 standard deviation units higher than
The “Coefficients” sections show an unstandard- that of women (controlling for age). Also, gender2
ized coefficient of −.260 for age, which means that, has a p-value of .000, indicating that the association
controlling for gender2, for each one-year increase in observed between it and weimins (after controlling
age, the model predicts a decline of .260 minutes spent for age) has a probability of less than 1 in 1000 of
on weights. The standardized coefficient for age is being due to chance alone. Had the significance level
−.289, indicating a decrease of .289 standard deviation been greater than .050, we would have had good
units in weimins for each standard deviation increase reason to exclude gender2 from the model. But with
in age (controlling for gender2). The significance level these results, it stays in.

bel65796_ch08_177-204 200 01/13/22 12:26 PM

 8 Quantitative Data Analysis 201

Methods in Motion Applications to Canadian Society

A multiple regression analysis of Indigenous people’s attitudes toward the police
Indigenous people in Canada have substantially enforced laws, were responsive, were approach-
lower incomes than other Canadians, have higher able, provided information to reduce crime, cared
incarceration rates, experience more racism than about neighbourhood safety, and treated people
other ethnic groups, and are far more likely to be fairly. The results indicated that both Indigenous
victims of homicide than non-Indigenous people and non-Indigenous people were more likely to
(Cao, 2014). All these factors suggest that under- have favourable than unfavourable attitudes
standing the relationship Indigenous people have toward the police. However, Indigenous people
with the police is important in coming to terms had levels of confidence in the police that were
with the larger issue of the plight of Indigenous lower than those for non-Indigenous people on
people in Canada. all six measures, and all the differences were sta-
An important indicator of the status of tistically significant. In the multivariate analyses,
Indigenous–police relations is the level of con- the association between being Indigenous and
fidence Indigenous people have in the police, having lower levels of confidence in the police on
in particular how confident they are that they the composite measure remained significant at
will receive fair treatment by police services p < .05 even after 13 relevant variables were con-
in Canada. Cao (2014) examined this issue by trolled for, including visible minority status, social
performing a multiple regression analysis of trust, contact with police, total criminal victimiza-
Indigenous attitudes toward the police using the tions, and several socio-demographic variables.
2009 Victimization Study, which is part of Statistics Cao suggests that “the differential confidence
Canada’s General Social Survey and is based on between Aboriginal and non-Aboriginal people
a nationally representative sample (see Box 5.10). found here is only the tip of the iceberg in the
To measure confidence in the police, the complicated relationship between the police and
author created an index by combining the re- Aboriginal people” (p. 516), and that “the domi-
sponses to six items that indicated how well, on nant mythology of Canada as a non-racist nation
average, participants believed their local police must be challenged” (p. 517).

Checklist
Checklist for performing and writing up quantitative data analysis

□ Have missing data codes been specified for inferences are included, are their limitations
all variables? outlined?
□ Are the statistical techniques used appro- □ If the data come from a cross-sectional
priate for the level of measurement (that is, design, have unsustainable inferences about
nominal, ordinal, or interval/ratio)? causality been resisted?
□ Are the most appropriate and powerful tech- □ Does the analysis go beyond univariate
niques for answering the research questions to include bivariate and even multivariate
used? analyses?
□ If the sample is not randomly selected, are □ Are the research questions answered, and
inferences about a population avoided? If only the analyses relevant to them presented?

bel65796_ch08_177-204 201 01/13/22 12:26 PM

202 PART II Quantitative Research

Key Points
• Think about data analysis before designing your analysis are applicable to some types of variable
surveys and other research instruments; in par- and not others.
ticular, consider whether they are in a format • Become familiar with computer software such as
that will permit you to answer your research SPSS, which is discussed in the Appendix.
questions in an optimal way.
• Don’t confuse statistical significance with impor-
• Know the difference between nominal, ordinal, tance or practical significance.
and interval/ratio variables. Techniques of data

Questions for Review (r) and Creative Application (a)

R1 What are missing data and how do they arise? Univariate analysis
A1 You are using a variable that is measured with R3 What is an outlier? How does the presence of an
a five-point Likert scale. What codes could be outlier affect the mean and the range?
used to designate missing data? Explain why A3 You have just gathered data that include a
they would be appropriate. variable measuring the participants’ ethnic
identity. Which measure of central tendency
Types of variables would be most appropriate for this variable?
R2 Define each of the three types of variable out- Explain.
lined in this chapter (nominal, ordinal, and inter-
val/ratio), and provide an example of each one. Bivariate analysis
A2 Identify the level of measurement (nominal, or- R4 Can one infer causality from bivariate analysis?
dinal, or interval/ratio) associated with each of Explain.
the following survey questions: A4 You have just found out that people who watch
violent movies are more likely to commit a vio-
1. Do you enjoy talking about politics?
lent crime than people who do not watch those
Yes
movies. Can you conclude from this that watch-
Unsure
ing violent movies causes people to commit vio-
No lent crimes? Explain.
2. How many times have you talked about pol- R5 In what circumstances is each of the following
itics with friends or family in the last seven used: Pearson’s r, Kendall’s tau-b, Cramér’s V,
days? Please write the number here. Spearman’s rho, eta?
A5 Your study indicates that the Pearson’s r coeffi-
3. What is your main source for news about cur- cient for the association between the variables
rent political events? Please tick one only. “years of formal education” and “income” is 0.6.
Social media (Facebook, Twitter, Instagram, How much of the variation in income can be pre-
etc.) dicted by years of formal education?
Television
Radio Statistical significance
Hard copy newspapers or magazines R6 What does “statistical significance” mean and
Other (please specify): how does it differ from “importance” or “practi-
cal significance”?
4. How important is it for you to be informed A6 What does it mean to say that an eta of .42 is
about political issues? statistically significant at p < .05?
Extremely important R7 What does the chi-square test achieve?
Very important A7 Your results indicate that the chi-square statistic
Moderately important for the variables “gender” and “highest level of
Slightly important education achieved” has a significance level of
Not at all important p = .12. What does that mean?

bel65796_ch08_177-204 202 01/13/22 12:26 PM

 8 Quantitative Data Analysis 203

Multivariate analysis A8 The results from your multivariate model that

R8 Define and provide an example of each of the uses "gender" and "father's occupation" to pre-
following: dict annual personal income has an adjusted
R-square value of .44. What does that mean?
• a spurious relationship
• an intervening variable
• a statistical interaction

Interactive Classroom Activities

1. These activities require student access to SPSS, h. Produce the regression output shown in
and may be facilitated by the instructor demon- Table 8.10. The class is asked to interpret the
strating the techniques to the class on the R-square and adjusted R-square values, the
classroom screen, one at a time. Step-by-step significance level shown in the ANOVA table,
instructions for each procedure are provided in the unstandardized and standardized coeffi-
the Appendix. Each student is to do the following: cients for age, and the significance level for
age.
a. Enter the data for the gym study, provided in
i. Introduce a second variable into the regres-
Box 8.2, into an SPSS file. Save the file to a flash
sion analysis by including the variable that
drive or some other reliable storage device.
measures the frequency with which patrons
b. Provide variable names, variable labels, and
go to the gym (frequent). For the purposes of
value labels for each variable, as appropriate.
this exercise, disregard any level of measure-
Then specify missing values for the age variable.
ment issues. The class is asked to interpret
c. Check that the data have been entered cor-
the R-square and adjusted R-square values,
rectly by doing frequency distributions for
then compare them to what was found for
each variable. Each student’s frequency dis-
the bivariate regression done in exercise h
tributions should match those shown on the
above, explaining any similarities or differ-
screen by the instructor.
ences. The class is then asked to interpret the
d. Recode the age variable following the instruc-
significance level shown in the ANOVA table,
tions in the Appendix, then provide a variable
the unstandardized and standardized coeffi-
label and value labels for the new variable.
cients for age and frequent, and the signifi-
The instructor puts the frequency distribution
cance levels for age and frequent.
for the new variable on the screen so students
can check that they have created the variable 2. This activity does not require access to SPSS.
properly. Save the file. After presenting an overview of multivariate
e. Produce the bar chart, histogram, and pie analysis, the instructor divides the class into
chart shown in Figures 8.2 to 8.4. The class is small groups. Each group is to come up with a
asked to interpret each figure. real or hypothetical example of:
f. Create the contingency table shown in Table
a. a spurious relationship;
8.4, and include a chi-square test. The class
b. an intervening variable; and
is asked: What information is conveyed by the
c. a statistical interaction.
contingency table? What do the chi-square
results indicate? The examples must be original, i.e., they cannot
g. Produce the correlation matrix shown in Table be taken from the textbook or lecture materials.
8.5. The class is asked to interpret the correla- Each group must then explain, at least con-
tion coefficient and significance level shown ceptually, how a researcher would go about
in each cell. Then they are asked: Why is there testing for each one. That is, they are to explain
a diagonal of “1s” in the table? Why is some of how one can determine if a particular relation-
the information in the table redundant? ship is spurious, whether a certain variable is in

bel65796_ch08_177-204 203 01/13/22 12:26 PM

204 PART II Quantitative Research

fact an intervening variable, and whether an have completed their work, they present their
interaction exists, and illustrate that with the ex- material to the rest of the class for comment and
amples they just came up with. Once the groups critique.

Relevant Websites
Statistics Canada offers a useful introduction to If you’d like to do further work with SPSS software,
quantitative data gathering and analysis. Navigate the IBM site offers some free trial downloads.
the site using the tabs on the right. [Link]/products/spss-statistics
[Link]
ch2/[Link] To learn more about big data, visit the Surveillance
Studies Centre (Queen’s University) website.
This YouTube video explains levels of measurement.
[Link]/
[Link]/watch?v=B0ABvLa_u88 (Websites accessed 4 September 2021.)

This YouTube video illustrates the notion of

correlation.
[Link]/watch?v=SaSpZdf1oHU

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch08_177-204 204 01/13/22 12:26 PM

 PART III
Qualitative Research
Part III of this book is concerned with qualitative research. Chapter
9 explores its main features, while Chapter 10 deals with ethnog-
raphy and participant observation, two important ways of collect-
ing qualitative data. Chapter 11 discusses qualitative interviewing
and focus groups, which are also valuable ways to acquire qual-
itative information, while Chapter 12 looks at content analysis, a
method of assessing texts (in the broad sense of the word) derived
from a variety of sources. The final chapter in Part III, Chapter 13,
illustrates the various ways in which data analysis is conducted in
qualitative research. The discussion provided in these chapters
gives you some important background knowledge that will help
you understand qualitative social science research, as well as
some practical instructions on how to conduct it.

bel65796_ch09_205-224 205 01/13/22 12:27 PM

 9 The Nature of
Qualitative Research
Chapter Overview
Qualitative analysis uses mainly words and images as data rather than numbers. Qualitative researchers
tend to produce inductive, constructionist, and interpretivist studies, although they do not necessarily
subscribe to all three of those perspectives. This chapter will consider:
• the main steps in qualitative research (although they are not followed as closely as the stages of
quantitative research);
• the relationship between theory and research in qualitative studies;
• the role of concepts in qualitative research;
• criteria for evaluating qualitative research;
• the main goals of qualitative researchers: seeing through the eyes of research participants; rich de-
scription; presentation of context, including the presence of conflict; concepts and theories as the
outcomes of research; and seeing social life and interaction as processes rather than static events;
• some common criticisms of qualitative research; and
• the main contrasts between qualitative and quantitative research.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Describe the variety of methods qualitative researchers use to collect data and the general steps they
follow as part of the research process.
2. Explain how qualitative researchers work towards developing theory and concepts iteratively as they
emerge during data collection through an inductive process.

▲ The Canadian Press/Nathan Denette

bel65796_ch09_205-224 206 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 207

3. Identify some of the different criteria that qualitative researchers use to assess the quality of their re-
search findings, such as credibility, transferability, dependability, and confirmability.
4. Describe the main goals of qualitative research, including: seeing through the eyes of research partic-
ipants; rich description; presentation of context, including the presence of conflict; concepts and theo-
ries as the outcomes of research; and seeing social life and interaction as processes rather than static
events.
5. Discuss some of the critiques of qualitative research and the main ways qualitative and quantitative
research differ.

Imagine that you are a non-Indigenous person and friends? What would be the most appropri-
participating in the National Inquiry into Missing ate way to get the information you need to con-
and Murdered Indigenous Women and Girls. duct the study? If it involves interviewing people
While some estimates maintain that the number who knew the victim, how could you get access to
of victims since 1980 is well into the thousands, them? What sort of interview would you use? What
you have been appointed to provide an in-depth would you talk about? Would it be useful to weave
study of one particular woman who was killed in a historical account of colonialism in Canada into
northern Saskatchewan five years ago. Imagine your analysis? How would you construct your re-
as well that you’re free to gather the “facts” of search so as to not harm those who have already
the crime—time of death, physical cause of death, experienced hardship? Will consultation and com-
the victim’s whereabouts before the murder, and munity involvement improve the research? Finally,
so on—but you feel that such information, although is it even possible for a non-Indigenous person
important, doesn’t address the core issues in- to fully understand what happened? This chap-
volved. How important would it be for you, in order ter discusses how qualitative researchers deal
to reach a full understanding of this event, to find with these sorts of questions as they conduct their
out what the murder meant to the victim’s family analyses.

Introduction • Qualitative writers are often constructionist in

that they understand social life to be an out-
As the opening vignette suggests, with qualitative come of the interactions and negotiations be-
research one moves beyond “the world of brute facts tween individuals, rather than a fixed structure
toward the realm of human meanings” (Bochner & to which individuals must conform and adapt.
Ellis, 2003, p. 509). This is accomplished with a focus • Its approach is naturalistic in that qualitative
on words and images rather than numbers, but there researchers try to understand social life in
are other attributes of qualitative research you should its natural setting as people reflect on and go
know about. The following are also significant: about their everyday lives.

• It usually involves an inductive view of the rela- Although all qualitative research tends to share these
tionship between theory and research: qualita- features, qualitative researchers can also differ signifi-
tive researchers usually start with field research cantly in both approach and subject matter. The fol-
and then develop theories and concepts from it. lowing are different varieties of qualitative research:
• It is normally interpretivist in that it seeks to
understand the social world through other • Ethnography/participant observation. These
people’s interpretations of it. are two very similar approaches to qualitative

bel65796_ch09_205-224 207 01/13/22 12:27 PM

208 PART III Qualitative Research

data collection. In both cases, the researcher is 11. General research question(s)
immersed in a social setting for some time, ob-
serving and listening to people in order to gain
an appreciation of their culture and experiences.
It has been employed in such social research clas- 2
2. Selecting relevant site(s) and participants

sics as Whyte’s (1955) study of street corner life in

a slum community and Gans’s (1962) research on
residents in the throes of urban redevelopment. 3
3. Collection of relevant data
• Qualitative interviewing. This is a very broad
term used to refer to in-depth, semi-structured 5b Collection of further data
5b.
or unstructured interviewing. Qualitative
researchers conducting ethnographic or par- 4 Interpretation of data
4.
ticipant observation research typically also
engage in a substantial amount of qualitative
interviewing.
5
5. Conceptual and theoretical work
• Focus groups. Several people are interviewed
together, often using a semi-structured format. 5a Tighter specification of
5a.
• Language-based approaches such as discourse the research question(s)
and conversation analysis, including critical 6. Write up findings/conclusions
6
discourse analysis.
FIGURE 9.1 An outline of the main steps of quali-
• Qualitative analysis of texts and documents, tative research
which falls under the general heading of con-
tent analysis. • Step 1: General research question(s). Since 2005,
• Participatory action research. As seen in all same-sex couples in Canada have had the
Chapter 1, professional researchers and others right to get married. While some people in
collaborate with the people directly affected by the LGBTQ community view the change in the
a social problem to understand the issue and to law as a victory for marriage equality, others
take action to resolve it. have reservations. For instance, the legal right
to marry may bring with it conventional as-
Each of these approaches will be examined in detail in sumptions of what marriage is supposed to
later chapters. Quite often, qualitative studies will use entail, such as the belief that child-rearing and
a multi-method approach. For example, in addition to monogamy are key features of the institution.
conducting qualitative interviews, an ethnographer Lyon and Frohard-Dourlent (2015) conducted
may analyze texts and documents as well. qualitative interviews to explore how unmar-
ried gay couples who cohabit deal with the
The main steps in qualitative tension that exists between the acceptance
research of marriage as a civil right and how having
The sequence of steps in the qualitative research pro- that right may impact the nature of LGBTQ
cess is outlined in Figure 9.1. These steps illustrate, in relationships.
general terms, how research may progress from data • Step 2: Selecting relevant site(s) and partici-
gathering to conceptualization and theorization, a pants. Participants were recruited through
process that is examined in detail in Chapter 13. The blogs and websites (including the University
work of Lyon and Frohard-Dourlent (2015) on how of Toronto online announcements board),
same-sex common-law partners view conventional email lists, physical posters, and word of
notions of marriage will be used as an example of mouth. There were 22 people, 12 women and
how this is done: 10 men, chosen for the study, all of whom were

bel65796_ch09_205-224 208 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 209

unmarried and living with their same-sex • Step 5a: Tighter specification of the research
partner in the Greater Toronto Area. question(s); and
• Step 3: Collection of relevant data. The authors • Step 5b: Collection of further data. Usually,
conducted in-depth, open-ended interviews in the initial collection and interpretation of
which the participants were encouraged to talk data lead to a sharpening of the research
about their current and former romantic rela- question(s), which is followed by additional
tionships, their views on marriage, how their data gathering to address the more specific
relationships were viewed by friends and rela- concerns contained in the revised research
tions, their levels of social support, and their questions. The authors’ account of the tra-
future relationship goals. The interviews typi- jectory of their research does not state ex-
cally lasted from 90 to 120 minutes. plicitly that they collected additional data
• Step 4: Interpretation of data. One of the key after starting to interpret the information
findings to emerge from the data analysis was that was gathered, but this may have taken
that participants felt that their relationships place spontaneously in the process of in-
were not viewed as having the same level of le- terviewing. When this does occur, there
gitimacy as those of married same-sex couples. can be an interplay between interpretation
One participant stated that “if I were straight and theorizing on the one hand, and data
I might not be that into marriage. But for our collection on the other. For instance, once
queerness, I think for some of our families . . . a particular interpretation has been made,
it’s necessary to take us seriously as a couple” further data may be collected to determine
(Lyon & Frohard-Dourlent, 2015, p. 413). The whether the interpretation is valid. This
researchers also found that several of the people strategy is frequently described as iterative,
interviewed explained their qualms about mar- and is a defining characteristic of grounded
riage in personal rather than ideological terms. theory, discussed next and in Chapter 13.
Instead of discussing the social or political im- • Step 6: Writing up findings/conclusions. In
plications of marriage, opposition to it often any form of research, an audience has to be
took the form of personal preferences. For ex- convinced of both the credibility and the
ample, one interviewee said that “it’s nice that significance of the interpretations offered.
people ask [if we will get married], but I say ‘No, Researchers are more than just conduits for
we’re not into it,’” (p. 417) and left it at that. what they see and hear. Lyon and Frohard-
• Step 5: Conceptual and theoretical work. No Dourlent made clear to their readers how the
new concepts emerged from Lyon and Frohard- institution of marriage can both lend legiti-
Dourlent’s work, but their article provided good macy to same-sex relationships and at the
illustrations of key terms used by researchers in same time invoke social expectations for the
this area, such as “heteronormative,” “hetero- people in those relationships that they may
sexist,” “homonormativity,” and “queer-based not accept. This may have led some research
meaning systems.” In terms of theory, they pro- participants to feel ambivalent about mar-
posed that the hard-fought political battle for riage or even make contradictory statements
marriage equality that was won by mainstream about it as they tried to navigate a relatively
LGBTQ movements may have made people in new legal terrain and the multiple messages
same-sex relationships reluctant to criticize and meanings that surround it. Also, the au-
marriage as an institution, which in the long run thors used heterosexual marriage as a back-
could influence conceptualizations of what con- drop to their analyses of the same-sex variety,
stitutes appropriate attitudes and behaviours for which allowed readers to reflect on the impli-
people in LGBTQ romantic relationships. (See cations of the institution of marriage for soci-
also the Methods in Motion box.) ety as a whole.

bel65796_ch09_205-224 209 01/13/22 12:27 PM

210 PART III Qualitative Research

An important part of qualitative research concerns with empirical data” (Becker, 1998, p. 109). The data
how the theory and concepts of the study relate to inform the utility of the concept. If data indicate
the research data. It is to that issue that our discus- that the concept is not adequately summarizing
sion now turns. the experiences and perspectives of participants, it
is refined or discarded. This process begins during
Theory and concepts in data collection, as researchers reflect on the fit
qualitative research between the concept and the observed reality of
participants’ life-worlds. The revision or rejection
Most qualitative researchers treat theory as some- of one concept gives rise to another that better ap-
thing that emerges out of the collection and anal- proximates the story told by the data.
ysis of data. Practitioners of grounded theory—a Blumer’s (1954) distinction between “definitive”
common approach to the analysis of qualitative and “sensitizing” concepts provides further insight
data—stress the importance of using data to develop into how qualitative researchers develop and employ
theoretical ideas. But some qualitative research- concepts. Blumer argues that fine nuances in the
ers argue that qualitative data can and should play form a concept can assume and alternative ways of
a role in testing theories as well. Silverman (1993) viewing its manifestations may be lost in the pro-
maintains that there is no reason why qualitative re- cess of operationalization. For that reason, Blumer
search cannot be employed to test theories specified recommends that social researchers regard their
in advance of data collection. Besides, much quali- concepts not as “definitive” but as “sensitizing”—
tative research entails the testing of theories in the providing “a general sense of reference and guidance
iterative process described above. Instead of seeking in approaching empirical instances” (1954, p. 7).
to test a theory from the outset, theoretical hunches For Blumer, concepts should provide a very general
or working hypotheses are more often constructed idea of what to look for, allowing the researcher to
and tested in the midst of qualitative data collection. discover the variety of forms that the phenomenon
In Figure 9.1, the loop back from step 5a, “tighter identified by the concept can take.
specification of the research question(s),” to step 5b, But Blumer’s distinction is not without problems.
“collection of further data,” implies that a theoretical It is not clear how useful a very general formulation
position may emerge in the course of research that of a concept can be as a guide to empirical inquiry. If
spurs the collection of further data to test it. This os- it is too general, it will fail to provide a useful starting
cillation between testing emerging theories and col- point because its guidelines are too broad. If it is too
lecting data is a defining feature of grounded theory. narrow, it is likely to repeat some of the difficulties
Blumer (1931) argues that a science without con- he identified with “definitive” concepts. However,
cepts would be analogous to “a carver without tools, his view is important in that it illustrates the value of
a railroad without tracks, a mammal without bones, beginning with a broad definition of a concept and
a love story without love” (p. 515): it would be im- then narrowing it down during the course of data
possible. It’s not surprising, then, that concepts collection (for an example, see Box 9.1).
are significant in both quantitative and qualitative
research. Unlike quantitative researchers, though, Criteria for evaluating
qualitative researchers do not make the measure- qualitative research
ment of concepts a central part of their work. For
qualitative researchers, the emphasis is on concep- In Chapters 2 and 4 we noted that reliability and
tualizing “how human group life is accomplished validity are important criteria for establishing and
in practice” and doing so “from participants’ assessing the quality of quantitative research. But
viewpoints in every instance” (Prus, 1997, p. 4). many qualitative writers argue that these criteria are
Concept development, as a feature of theory con- not directly applicable to their work (Finlay, 2006).
struction, often occurs “in a continuous dialogue For example, since measurement is not a major

bel65796_ch09_205-224 210 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 211

Methods in Motion Applications to Canadian Society

The influence of same-sex marriage on social institutions and lesbian and gay
relationships
What happens to the institution of marriage Extended family members, people at work,
when the law changes to allow same-sex and society in general seemed more accepting
couples to marry? This is a research question of them after they were married. One lesbian
posed by Green (2010), who used qualitative remarked, “It was absolutely incredible, over-
methods to examine three different predic- whelming, just, even my mother. . . . [P]eople we
tions regarding same-sex marriage. Social hadn’t talked to in years would phone and ask if
conservatives maintain that gay marriage will it was okay to come to the wedding. . . .” (Green,
contribute to further decline in the nuclear 2010, p. 413).
family, increase marital infidelity, and lead to But contrary to both the critical feminist/
less stable marriage bonds. Critical feminists queer theory position and the lesbian and
and queer theorists argue that it will produce gay assimilationist views, married same-sex
same-sex marriage institutions with the same couples did not completely buy into the con-
problematic characteristics as those found in ventional heterosexual idea of marriage. For
heterosexual marriage: obligatory monogamy, one thing, there was more support for non-­
gender-specific social roles, and conventional monogamous sexual relationships among
expectations for child-rearing. Lesbian and those interviewed than in the public at large,
gay assimilationists offer a similar predic- although the support was more pronounced
tion, but one that sees conventional marriage among the male interviewees. In fact, some
norms as largely positive: same-sex marriage couples were monogamous before marriage
will strengthen ties between gay couples, en- but not after. Said one gay man, “So, it sounds
courage monogamy, and help to stabilize kind of backwards to the traditional model, but
queer families. the fact that we’re legally married to each other
Green explored the various positions on gay and permanently committed makes us both
marriage by conducting 30 semi-structured in- feel very secure about [having sexual relations
terviews with people from two Ontario cities who outside of marriage]” (Green, 2010, p. 419).
were in same-sex marriages; half the interview- Green also found that the people in same-sex
ees were lesbians and the other half gay men. marriages claimed to have an egalitarian di-
Contrary to the social conservative position, vision of labour regarding household chores
many interviewees reported that their relation- and yard work, and relatively equal power re-
ships with their spouses were strengthened after lationships within the marriage, although he
they exchanged marriage vows, and that they emphasized that further research was needed
came to value stability and permanence in their to substantiate those claims. Green speculated
relationship more after they were married. One about the future of same-sex marriage, in par-
gay man said: “I think it’s just the sense of com- ticular whether it can retain its unique charac-
mitment that you feel. You’ve made a vow and, teristics if queer institutions continue to gain
it’s hard to describe, it definitely feels different acceptance by the larger society and in the
than prior to [marriage]” (Green, 2010, p. 411). process lose their oppositional tenor. Like other
The people studied also mentioned that being predictions for social change, those regarding
married bestowed a sense of legitimacy on their gay marriage can be tested only with the pas-
relationship and on themselves as individuals. sage of time.

bel65796_ch09_205-224 211 01/13/22 12:27 PM

212 PART III Qualitative Research

Box 9.1 The emergence of a concept in qualitative research: The

case of emotional labour
• watched training sessions for flight attendants
and had many conversations with both train-
ees and experienced attendants;
• interviewed airline personnel such as manag-
ers and advertising agents;
• examined Delta advertisements spanning 30
years;
Sean Locke/iStockphoto

• observed the flight attendant recruitment pro-

cess at Pan American Airways (because she
had not been allowed to do this at Delta); and
• conducted “open-ended interviews last-
Is this woman’s smile spontaneous and sincere, or ing three to five hours each with thirty flight
is she just doing her job? attendants in the San Francisco Bay Area”
Hochschild’s (1983) concept of emotional (Hochschild, 1983, p. 15).
labour—labour that “requires one to induce or For a contrasting occupational group also involved
suppress feelings in order to sustain the outward in emotional labour, she interviewed five debt
countenance that produces the proper state of collectors. Her book explored topics such as the
mind in others” (p. 7)—has become very influen- human costs of emotional labour and the issue of
tial in the sociology of work and in the developing gender in relation to it. It’s clear that Hochschild’s
area of the sociology of emotions. She gathered concept of emotional labour began with a some-
data on emotional labour by exploring how air- what imprecise idea that was gradually developed
line workers manage to keep smiling at truly to address its wider significance. The concept has
obnoxious customers. Hochschild’s initial con- been picked up by other qualitative researchers in
ceptualization emerged from a questionnaire the sociology of work. For example, Leidner (1993)
she had created on a related topic. To develop did an ethnographic study of a McDonald’s restau-
the idea of emotional labour she gained access rant and an insurance company to investigate how
to Delta Airlines, and in the course of her inves- organizations seek to “routinize” emotional labour.
tigations she: See Box 9.3 for another adaptation of her work.

concern among qualitative researchers, measure- observing, identifying, or ‘measuring’ what you say
ment validity seems to have little bearing on their you are” (2002, p. 39). Others have used terms simi-
investigations. lar to those employed in quantitative research but in-
One approach is to adopt most of the core ideas vested them with somewhat different meanings (e.g.,
used to assess quantitative research without making Golafshani, 2003).
any special adjustments. This is what Mason (2002) Many writers take a different approach, suggesting
does, using terms such as “validity,” “reliability,” that qualitative studies should be judged or evaluated
and “generalizability” (external validity) in much according to criteria quite different from those used
the same way that quantitative researchers do: for in quantitative research. Lincoln and Guba (1985)
example, “If your research is valid, it means you are and Guba and Lincoln (1994), for instance, argued

bel65796_ch09_205-224 212 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 213

that it is necessary to assess the quality of qualitative represented in the findings, and whether ver-
research in new ways, using new terms. They pro- batim quotes were accurate.
posed two primary criteria for assessing a qualitative • Researchers may show pre-publication versions
study: trustworthiness and authenticity. The latter in of their articles or books to their participants.
this context refers to the degree to which the research They may also provide the study participants
is transformative and emancipatory for the people with an opportunity to ask questions about the
studied and society at large. Here we will focus on methods and findings of the research and offer
trustworthiness, which is made up of four criteria: feedback and criticism. Taken further, Lassiter
(2005, p. 15) makes a case for collaborative eth-
• credibility nography, which entails researchers and par-
• transferability ticipants collaborating in all aspects of a study,
• dependability “from project conceptualization, to fieldwork,
• confirmability and, especially, through the writing process.”

We’ll consider each of these terms individually. In each case, the goal is to confirm that the researcher’s
findings and impressions are congruent with those of
Credibility the people on whom the research was conducted and,
The idea of credibility is tied to the notion that dif- if there are discrepancies, to find out why. However, the
ferent people may interpret the social world in dif- idea is not without practical difficulties:
ferent ways. Since there can be several possible ways
of experiencing or coming to terms with a particular • Respondent validation may lead some research
social situation or event, the researcher has to ensure participants to respond defensively and even
that the interpretations presented in the study ring demand censorship.
true to the people observed. The establishment of • Bloor (1997, p. 45) observed that research par-
credibility entails both following proper research ticipants who develop relationships of “fond-
procedures and submitting the findings to the people ness and mutual regard” with the researcher
studied for confirmation that the account is consis- may be reluctant to criticize the study.
tent with the way they see their world. The researcher • It’s doubtful that research participants can
is, after all, describing their reality. This practice of validate all of a researcher’s analysis, since it
getting feedback on the study from the participants will be written for social scientists and will
involved is referred to as respondent validation, therefore include many specialized references,
member validation, or member reflection. concepts, theories, and contextual issues un-
The aim of respondent or member validation is familiar to them. Skeggs (1994, p. 86), for
to seek corroboration or criticism of the researcher’s example, reported that the “most common
observations and interpretations. It can take several response” among her participants was “Can’t
different forms: understand a bloody word it says.”

• Each research participant may be provided Transferability

with an account of what he or she said to the Because qualitative research typically entails in-depth,
researcher or others. For example, in a study intensive study of a small number of people, qualitative
that explored issues of trust between mothers findings tend to flow out of the context in which the
of children with disabilities and school prin- observations are made. As Lincoln and Guba (1985,
cipals, Shelden. Angell, Stoner, & Roseland p. 316) put it, whether the findings “hold in some other
(2010) gave all respondents a summary of the context, or even in the same context at some other time,
findings by either telephone or email. They is an empirical issue.” Instead of trying to come up with
also asked each participant if their views were findings that can definitely be applied to other times,

bel65796_ch09_205-224 213 01/13/22 12:27 PM

214 PART III Qualitative Research

Research in the News

What’s missing in media portrayals of violent crime
In 2017, a Canadian veteran of the war in Lost from view in these media accounts was
Afghanistan shot and killed his wife, his the fact that there were victims besides the person
mother, and his daughter, and then killed who committed the murder-suicide. The lives and
himself. Many of the media accounts of the circumstances of the other three people who died
tragedy focused on the perpetrator, in par- were largely overlooked and forgotten. Another
ticular on the fact that he had suffered thing that was missing was an acknowledgment
from post-traumatic stress disorder (PTSD) that the murders were part of the larger societal
brought on by his combat experience. A issue of deadly violence against women (Renzetti,
common narrative in the media portrayals 2017). As Mount Allison University sociology pro-
of the event involved the idea that Canadian fessor Ardath Whynacht put it, “There were four
soldiers suffering from PTSD and other victims that day and we’re talking only about the
combat-related ailments typically cannot get services that could have helped him, and not,
the help they need, despite the significant for example, services that might have helped his
sacrifices they have made in Canada’s mili- spouse be safer, in trying to leave that relationship
tary operations. and get space” (Renzetti, 2017).

places, and people, qualitative researchers are encour- research. A rare example is a study of behaviour at an
aged to produce what Geertz (1973) called thick de- American “swap meet,” where second-hand goods are
scription: rich, detailed accounts of a group’s culture bought and sold (Belk, Sherry, & Wallendorf, 1988).
or people’s experiences. Lincoln and Guba argued that A team of three researchers collected data over four
thick description provides others with the database days through observation, interviews, photography,
they need in order to assess the possible transferability and video records. The researchers conducted several
of findings to other milieus. trustworthiness tests, including respondent validation.
In addition, they submitted their draft manuscript and
Dependability data set to three peers whose task it “was to criticize the
Lincoln and Guba proposed that to establish depend- project for lack of sufficient data for drawing its conclu-
ability, researchers should adopt an “auditing” ap- sions, if they saw such a void” (1988, p. 456). The same
proach. That is, they should keep complete records of study highlighted some problems with this approach,
all phases of the research process—problem formu- among them the fact that, because qualitative research
lation, selection of research participants, fieldwork generates extremely large data sets, validation is very
notes, interview transcripts, data analysis decisions, demanding for the auditors. That may be the major
and so on—and ensure that the records are accessi- reason why auditing has not been widely used.
ble. Peers would then act as auditors, possibly during
the course of the research and certainly at the end, to Confirmability
make sure that proper procedures have been followed Objectivity is a difficult if not impossible standard to
and to assess the degree to which the study’s interpre- live up to in social research. For that reason, the con-
tations and theoretical inferences can be justified. firmability criterion is designed to ensure that the
Auditing has not, however, become a popular ap- researcher has acted in good faith. In other words, it
proach to enhancing the dependability of qualitative should be apparent that personal values or theoretical

bel65796_ch09_205-224 214 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 215

inclinations did not blatantly and unduly sway either the rooted in “Verstehende sociology—trying to
conduct of the research or the findings derived from it. think oneself into the situations of the people
Lincoln and Guba suggest that establishing confirma- one is interested in . . . in this case the ‘hooli-
bility should be one of the objectives of auditors. gan.’ This approach involves recognizing social
phenomena as due not to any single or simply
Overview of the criteria issue identifiable cause and attempting to make
There is a general recognition that quantitative re- sense from the [multi-causal] social actors’
searchers’ notions of reliability and validity cannot viewpoint” (Armstrong, 1993, pp. 5–6).
be applied to qualitative research without some mod- • For their research on teenaged girls and vio-
ification, although critics vary in their views on how lence, Burman, Batchelor, & Brown (2001, p. 447)
much modification is necessary. To bolster their ac- “sought to ground the study in young women’s
counts, qualitative researchers may use some of the experiences of violence, hearing their accounts,
strategies advocated by Lincoln and Guba, such as and privileging their subjective views.”
thick description and respondent validation exercises.
The goal of seeing through the eyes of the people
under study is often accompanied by the closely re-
The main goals of qualitative lated goal of probing beneath surface appearances.
researchers After all, by taking the position of those under study,
researchers may be able to see things in a way that
Seeing through the eyes of the an outsider with little direct contact would not. This
people being studied insight is revealed in:
Many qualitative researchers try to view the social
world through the eyes of the people they study. To • Hallgrimsdottir, Phillips, & Benoit (2006)
accomplish this, Blumer (1969) advocates striving for study of sex-trade workers, which revealed that,
a sense of “intimate familiarity” with research partici- contrary to media depictions, these workers
pants’ everyday lives. Achieving a sense of intimate fa- often see their occupation in a mundane light
miliarity means “gaining an in-depth knowledge of the (for example, as simply a way to make a living);
research participants, their setting or settings, and their • Kleinknecht’s (2011) research on computer hack-
situations and actions” (Charmaz, 2008, p. 53). This ers, which found that popular portrayals of hack-
approach is in keeping with qualitative researchers’ at- ers as “computer criminals” were often at odds
tempt to achieve verstehen (empathetic understanding) with hackers’ activities and how they defined
and their focus on studying the meanings people attri- themselves: those who take a passionate, creative
bute to the social world in situ. approach towards problem-solving; and
Qualitative researchers maintain that “(1) . . . face- • McDonough and Polzer’s (2012) examination
to-face interaction is the fullest condition of partic- of the daily struggles of municipal employees
ipating in the mind of another human being, and following the creation of the “mega-city” of
(2) . . . you must participate in the mind of another Toronto, which showed that rather than cre-
human being (in sociological terms, ‘take the role of ating a leaner, more efficient workforce, the
the other’) to acquire social knowledge” (Lofland & restructuring led many manual workers to
Lofland, 1995, p. 16). This tendency reveals itself in question their commitment to excellence on
frequent references to seeking and privileging an in- the job, while white-collar employees felt they
sider perspective. Here are some examples: had to do extra work just to maintain profes-
sional standards of service.
• Armstrong (1993) carried out research on
British soccer hooliganism through partici- On the other hand, sometimes qualitative research
pant observation. He described his work as confirms popular notions of a particular group.

bel65796_ch09_205-224 215 01/13/22 12:27 PM

216 PART III Qualitative Research

For example, Bell (2007) found that media depic- on context in qualitative research goes back to
tions of the western Canadian separatist movement the classic studies in social anthropology, many
as decidedly right wing were essentially correct. of which demonstrated how seemingly illogical
The empathetic effort to see the world through practices (e.g., a magical ritual that accompanies
the eyes of their participants can present research- the sowing of seeds) make sense when they are un-
ers with practical problems. For example: How far derstood as part of the society’s belief system. The
should one go to develop intimate familiarity and provision of descriptive detail is also a manifes-
empathy? What if the research participant’s experi- tation of qualitative researchers’ desire for natu-
ence includes illegal or dangerous activities? There ralism: that is, their belief that researchers should
is also the risk that the researcher will be able to see study the social world as it actually is, rather than
through the eyes of some of the people in a particular relying on contrived settings such as formal inter-
social scene but not those of others, such as people of views or experiments.
another gender or culture. These and other practical
difficulties will be addressed in later chapters. Emphasis on process
Qualitative research tends to view social life in
Description and the emphasis on terms of processes. This tendency reveals itself in
context a number of ways, but the main one is the qualita-
Qualitative researchers are much more inclined than tive researcher’s concern with showing how events
their quantitative colleagues to include descriptive and patterns unfold over time. Qualitative evidence
detail when reporting their research. However, they often conveys a strong sense of change and flux.
are not exclusively concerned with description. They As Pettigrew (1997, p. 338) put it, process is “a se-
also try to explain, and what they endeavour to ex- quence of individual and collective events, actions,
plain may pertain to issues of power and exploitation. and activities unfolding over time in context.”
For example, Skeggs (1997, p. 22) tried to answer this Qualitative research using participant observation
question: “Why do women who are clearly not just is particularly well suited to the study of process.
victims of some ideological conspiracy, consent to a Ethnographers are typically immersed in a social
system of class and gender oppression which appears setting for a long time—in many cases, for years.
to offer few rewards and little benefit?” Consequently, they are able to observe how events
Many qualitative studies do provide a detailed ac- develop over time and how the different elements
count of what goes on in the setting under investiga- of a social system (such as values, beliefs, and be-
tion. On the surface, some of this detail may appear haviour) interconnect. Tracing interdependent
irrelevant; indeed, there is a risk of becoming too streams of actions and events allows the researcher
embroiled in descriptive detail. Lofland and Lofland to present social life as a process (see Box 9.2).
(1995, pp. 164–165), for example, warned against A sense of process can also be developed
what they called “descriptive excess” in qualitative through semi-structured and unstructured in-
research, whereby the amount of detail overwhelms terviewing in which participants are asked to re-
or inhibits the analysis of the data. flect on the activities leading up to or following
But one of the main reasons for providing de- an event. Shaffir and Kleinknecht (2005), for ex-
scriptive detail is that it permits a contextual un- ample, used informal interviews with 45 former
derstanding of social behaviour. This implies that Canadian politicians to explore how they coped
we cannot understand the behaviour of mem- with their involuntary exit from political life. The
bers of a social group without some knowledge enormity of the loss was often described as being
of the specific environment in which they oper- akin to a social death. A former Liberal parliamen-
ate. Behaviour that appears odd or irrational may tarian described the process of coming to terms
make perfect sense when understood in the con- with the defeat as follows: “It’s like the phases of
text within which it takes place. The emphasis death. You have loss, anger, sadness, and then you

bel65796_ch09_205-224 216 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 217

Box 9.2 Process in youth shelters

Karabanow (2002) described his experiences in provided the basis for an analysis of the evo-
two Canadian shelters for homeless and run- lution of their organizational processes. This
away youth. As a participant observer, he was example shows the development of a sense of
able to monitor routine activities there and de- process in at least two ways. First, observation
scribe the shelter culture. In addition, he car- of the shelters over time made it possible to
ried out in-depth interviews with three levels bring out developments and interconnections
of shelter workers and used agency archival between events. Second, connecting these
materials. The dramatic transformations in the events with historical and other data made
shelters’ external environment and internal op- it possible to show how the shelters were af-
erations that took place over a period of years fected by the larger society.

come to accept it” (Shaffir & Kleinknecht, 2005, structured method. For that reason, qualitative re-
p. 714). The life history approach is another search tries to avoid limiting areas of inquiry, and
form of qualitative research that can be used to the research questions it asks tend be fairly general
show process. One of the best-known studies of (see step 1 in Figure 9.1).
this kind is Lewis’s (1961) classic study of a poor Ethnography, with its emphasis on participant
Mexican family, in which he conducted extended observation, is particularly well suited to an unstruc-
taped interviews with family members to recon- tured approach. It allows researchers to immerse
struct their life histories. themselves in a social setting with a general research
idea in mind and then gradually, through observa-
Flexibility and limited structure tion, narrow it down to a more specific topic.
Many qualitative researchers are disdainful of Another advantage of the unstructured nature
approaches that impose a predetermined set of of most qualitative inquiry is that it offers the pros-
assumptions on the social world. This position pect of flexibility. The researcher can change the
is closely related to the preference for seeing the direction of the investigation much more easily
world through the eyes of the people under study. than in quantitative research, which tends to have a
After all, a structured method of data collection built-in momentum once the data collection is un-
must be planned ahead of time on the basis of the derway. (If you send out hundreds of postal ques-
researcher’s prior ruminations and expectations tionnaires and realize only after getting some back
regarding the nature of a social reality that they that you left out an important issue, it’s not easy
may never have encountered before. This struc- to rectify the situation.) Structured interviewing
ture limits the degree to which the researcher can and structured observation can offer some flexi-
genuinely adopt the world view of those being bility, but it will be limited by the requirement that
studied, and can lead to serious misunderstand- interviews be as comparable as possible. Having
ings. Keeping structure to a minimum is supposed conducted ethnographic fieldwork in sites as dis-
to enhance the likelihood that the research will parate as a psychiatric institute in Quebec to a day
reveal the perspectives of the people being ob- shelter for individuals experiencing homelessness
served. This approach allows aspects of people’s in Iceland, Dolson (2018) found that the fieldwork
social world that are particularly important to experience was often unpredictable and uncertain.
them to come to light—aspects that might never To respond to these situations, he advocates foster-
even cross the mind of a researcher using a more ing skills in adversity management: “attempting to

bel65796_ch09_205-224 217 01/13/22 12:27 PM

218 PART III Qualitative Research

be comfortable with the unexpected, and trying to can suggest alternative avenues of inquiry or ways
make the best of ambiguous situations in which you of thinking.
have no idea what will happen next” (Dolson, 2018,
p. 22). Qualitative researchers, especially ethnog- Ultimate goals
raphers who confront some of the most precarious The goals of qualitative research mentioned above—
research situations, must remain open, flexible, and seeing through the eyes of others, bringing out a sense
ready to improvise their research plans. of process, and having a flexible and unstructured
As noted previously, some qualitative research- method of inquiry—are essentially immediate or
ers use sensitizing concepts as a starting point for proximate goals that are used as means of achieving
their research. While this might be seen as remov- a deep understanding of the people or groups being
ing some of the flexibility and unstructured char- studied. But qualitative researchers’ goals often go
acter of qualitative research, Charmaz (2003) states, beyond simply understanding. They may also wish
“We may use sensitizing concepts only as points of to do research in order to pursue social justice and
departure from which to study the data” (p. 259, em- bring about social change, as mentioned in our dis-
phasis in original). When used in this fashion, con- cussion of critical approaches to science in Chapter 1.
cepts offer tentative analytical direction to a project. Box 9.4 illustrates how people doing participatory
Qualitative researchers believe it’s important to action research can use qualitative methods to ad-
remain receptive to new and alternative under- dress a serious social problem. Indeed, qualitative
standings as they emerge from the data. See Box 9.3 studies have shed light on a wide range of societal
for an illustration of how a conceptual angle can injustices in Canada, including racism (Branker,
help give direction to a study, while the unstruc- 2017), homophobia (Abramovich, 2016), and poverty
tured data collection style of qualitative research (Coloma & Pino, 2016).

Box 9.3 Emerging concepts

In an update to Hochschild’s (1983) work (dis- but also protecting them. Their new role as
cussed in Box 9.1), Canadian researchers Santin guardian partially shielded them from workplace
and Kelly (2017) used participant observation repercussions that traditionally followed from
and informal interviews with flight attendants not treating passengers in a passive, courteous
to revisit the concept of emotional labour fol- manner. In line with Hochschild’s findings, institu-
lowing the 9/11 terrorist attacks. Consistent with tional norms continued to structure and constrain
Hochschild’s findings, Santin and Kelly found that flight attendants’ behaviour, but now new safety
flight attendants continued to engage in “surface protocols offered them more control over their
acting:” they would conceal their true emotions emotional labour. By exploring flight attendants’
in order to please their employer or customers. emotional labour post-9/11, Santin and Kelly
However, they also discovered that changes to expanded on Hochschild’s ideas, adding three
airline security protocols post-9/11 provided new concepts to our social science lexicon: the
cabin crew with more discretionary power 9/11 effect, emotional agency, and role shields.
and emotional agency in responding to unruly Their study demonstrates how existing theoret-
passengers. The “9/11 effect” led to a shift in ical ideas can be used as sensitizing concepts
how emotional labour was carried out in the new in subsequent research and then built upon to
pro-security context. It meant that cabin crew develop new ways of understanding changing
were not only providing a service to passengers, social contexts.

bel65796_ch09_205-224 218 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 219

Box 9.4 The use of qualitative methods in participatory action

research
Type 2 diabetes among Indigenous children and past researchers to marginalize Indigenous ex-
youth is a growing problem, one with import- periences and ways of knowing. The goal was
ant implications for the quality of life among to make the members of the community equal
Indigenous people in Canada. Sharma, Reimer- partners in the research, and in so doing allow
Kirkham, & Meyerhoff (2011) sought to do some- them to bring their own issues and perspectives
thing about it, but rather than approach the to bear on it.
issue from the perspective of conventional social The various participants decided collectively
science, they used the methods of participatory that the project would take the form of a short
action research (PAR) to address it. Instead of film to be created by five Indigenous youths aged
devising the methodology themselves and then 14–16. The film featured interviews with Elders
seeking consent from prospective research suffering from diabetes, a description of what di-
participants, they first consulted members of abetes is, and some tips on healthy eating. The
the community where the research would take production of the film involved a constant nego-
place to get their permission to do a study on tiation of the relationship between the profes-
the topic, and then asked for their views on what sional researchers and the youths. The former
an appropriate methodology might be. After tried to stay in the background as much as pos-
gaining permission to go ahead with the project sible, although they did end up editing the first
from the band and chief of a rural Indigenous version of the film. In order to enhance the col-
community in British Columbia, Sharma et al. laborative nature of the project, the professional
held focus groups with several community researchers then showed the edited version to
groups, including Elders with diabetes, the Chief the youths and removed any parts that the young
and Council, a health-care services team, and people thought were inappropriate. This illus-
some Indigenous youths to determine how the trates several aspects of PAR: explicit recognition
study would be conducted. This was done in of the power differentials that often characterize
an attempt to neutralize the power differen- conventional research; an attempt to minimize
tial between the researchers (who were non- those differentials in order to address the subject
Indigenous and had PhDs, secure middle-class matter from the perspective of the people par-
incomes, and Western cultural outlooks) and ticipating in the research; and full participation
the members of the community. It was espe- by the group that is the focus of the inquiry in
cially important to recognize the potential for order to produce knowledge and actions that are
exploitation in this research, given the history beneficial to them and not just to the professional
of colonialism in Canada and the tendency of researchers.

Critiques of qualitative Too subjective

research Quantitative researchers sometimes criticize qual-
itative research for being too impressionistic and
Just as quantitative research has been criticized by subjective. By this they usually mean that qualita-
people in the qualitative tradition, the latter has been tive findings depend too much on the researchers’
taken to task by quantitative social scientists. Some values and opinions about what is significant. A
of the more common criticisms of qualitative re- related contention stems from the close personal
search are discussed below. relationships that many researchers strike up with

bel65796_ch09_205-224 219 01/13/22 12:27 PM

220 PART III Qualitative Research

the people they study, which can lead to bias. Also, criterion is largely irrelevant to their research goals
because qualitative researchers often begin in a rel- because it is not consistent with their epistemolog-
atively open-ended way and then gradually narrow ical and ontological positions. For example, they
down their research questions or problems, they maintain that it is to be expected that different re-
may not explain clearly at the outset why they searchers will view a particular situation or event
came to focus on one area rather than another. In differently, and then take the study in different
response to this critique, qualitative researchers directions.
argue that researchers who attempt to preserve
their objectivity by remaining distant from the Issues of generalization
social world, risk the worst form of subjectivity: It is sometimes suggested that the scope of qualita-
“the objective observer is likely to fill in the pro- tive findings is restricted. Because research based
cess of interpretation with [their] own surmises in on participant observation or unstructured inter-
place of catching the process as it occurs in the ex- views typically involves only a small number of in-
perience of the acting unit which uses it” (Blumer, dividuals in a single organization or locality, it may
1969, p. 86). By not getting close to the social world be difficult to know whether the findings can be
and trying to see things from participants’ points applied to other people or other settings. Can just
of view, we become prone to filling in gaps in our one or two cases be representative of some larger
understanding with our own perspectives instead population? Can Karabanow’s (2002) study of two
of the participants’. Canadian shelters for homeless youth (Box 9.2)
be considered generalizable to all youth shelters
in the country? In the case of research based on
Difficult to replicate qualitative interviews rather than participation, are
Some quantitative researchers argue that the prob- those who were interviewed representative of some
lem of subjectivity is made worse by the difficulty of larger population? Do the interview findings from
replicating a qualitative study, even though replica- sex-trade workers in Victoria (Hallgrimsdottir
tion is uncommon in the social sciences, regardless et al., 2006), for example, apply to sex workers in
of the research methods used. Because it is unstruc- Toronto?
tured and relies on the researcher’s ingenuity, and The answer to these questions in most qualita-
because it may examine processes and situations tive research studies is “probably not.” But a case
that are in flux, qualitative work is almost always study is not a sample of one drawn from a known
impossible to replicate. What is observed and heard population. The people interviewed in qualitative
often depends on the researcher, as does the choice research are not necessarily meant to be repre-
of what to concentrate on; different researchers may sentative of some larger group. In fact, with some
choose to observe different things and as a result social groups, such as workers in the sex trade, it
they may come up with different findings. Also, the may be impossible to enumerate the population in
behaviour of the people being observed or inter- any precise manner, which makes the selection of
viewed is likely to be affected by the characteristics a representative sample impracticable.
of the researcher (personality, age, gender, and so Moreover, in most cases it is not the purpose of
on). The difficulties that qualitative social scientists qualitative research to produce generalizable knowl-
experience when they revisit ground previously edge. Such research normally stands on its own,
covered by another researcher (often referred to as providing in-depth analysis of a person or a small
a “restudy”) do not inspire confidence in the rep- group of people—something that quantitative re-
licability of qualitative research (Bryman, 1994). search rarely does. Also, qualitative research can
Qualitative researchers answer that the replicability contribute to the creation of generalizable knowledge

bel65796_ch09_205-224 220 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 221

insofar as it produces theories or concepts that can be TABLE 9.1 Common contrasts between
assessed using other research methods. For instance, quantitative and qualitative research
Williams (2000) argued that qualitative researchers
are often in a position to produce generalizations re- Quantitative Qualitative
garding the people or events studied (a group of drug Numbers Words
users or soccer hooligans, a strike) that “can be seen Point of view of Points of view of
to be instances of a broader set of recognisable fea- researcher research participants
tures” (2000, p. 215). In addition, Williams argued Researcher distant Researcher close
that a researcher studying the hooligans who follow Theory testing Theory development
a certain soccer club will often draw comparisons Structured Unstructured
with the findings of other researchers who have stud- Generalizable Contextual
ied comparable groups. He or she might also draw knowledge understanding
comparisons with followers of other professional Hard, reliable data Rich, deep data
sports teams, or with violent groups in contexts
Macro Micro
other than sport. Such generalizations will always be
Behaviour Meaning
limited and somewhat more tentative, though no less
Artificial settings Natural settings
useful, than the statistical generalizations produced
by quantitative research.
• “Numbers” versus “Words.” Quantitative re-
Lack of transparency searchers set great store by precise, numerical
Another criticism of qualitative research is that it may measurements of social phenomena. They use
be less than transparent about what the researchers ac- those measurements to calculate statistics to
tually did and how they arrived at their conclusions. For understand social life, whereas qualitative re-
example, qualitative research reports are sometimes searchers use mainly words in their analyses of
unclear about how the participants were chosen for ob- society.
servation or interview. This contrasts with the tendency • “Point of view of researcher” versus “Points of
of quantitative researchers to include detailed accounts view of participants.” In quantitative research,
of sampling procedures in their reports. However, since investigators are in the driver’s seat; their
qualitative research does not claim generalizability the concerns structure the investigation. In qual-
way quantitative research does, sampling procedures itative research, the investigation is struc-
are less relevant to their work. tured by the people being investigated: their
perspective—what they see as ­ i mportant—
provides the point of orientation.
Some contrasts between • “Researcher is distant” versus “Researcher is
quantitative and qualitative close.” In quantitative research, researchers
research tend to be less involved with their subjects and
in some cases, as in research based on online
Several writers have used tables to contrast quan- or mailed questionnaires, they may have no
titative and qualitative research (for example, direct contact at all. Quantitative researchers
Hammersley, 1992). Table 9.1 highlights the main often consider this distance desirable, because
differences. As with any summary of a large body of they want to be as objective as possible and feel
material, the table outlines general tendencies. The their objectivity would be compromised if they
full picture is more nuanced: there are exceptions to were to become too involved with the people
these general points, as we will see in Chapter 14. they study. By contrast, qualitative researchers

bel65796_ch09_205-224 221 01/13/22 12:27 PM

222 PART III Qualitative Research

seek involvement with the people they study in • “Hard, reliable data” versus “Rich, deep data.”
order to see the world through the eyes of their Quantitative researchers often describe their
participants and promote a more equalitarian data as “hard”—robust and unambiguous—
relationship. because they strive for precision when devis-
• “Theory and concepts tested in research” versus ing their measurement techniques. Qualitative
“Theory and concepts developed from data.” researchers, by contrast, claim that their con-
Before data collection begins, quantitative textual approach and often prolonged involve-
researchers typically have a theory in mind ment in a setting mean that the information
that they want to test. In qualitative research, they collect is rich, nuanced, and very detailed.
concepts and theories develop as the data are • “Macro” versus “Micro.” Quantitative researchers
collected. often seek to uncover large-scale social trends
• “Structured” versus “Unstructured.” Quantitative and connections between variables; as we noted
research is typically highly structured to maxi- above, their research is generally designed to be
mize validity and reliability; in qualitative re- applicable to large populations. Qualitative re-
search, the approach is less structured in order to searchers are usually concerned with in-depth
allow the researcher to get a sense of the mean- analysis of small-scale aspects of social reality,
ings people derive from their everyday lives, and such as personal interaction, although these may
to permit the development and elaboration of be linked to larger social and political issues.
concepts and theories as the data are collected. • “Behaviour” versus “Meaning.” It is sometimes
• “Generalizable knowledge” versus “Contextual suggested that quantitative researchers focus
understanding.” Whereas quantitative re- more on people’s behaviour, and qualitative
searchers want their findings to be applicable researchers more on the meaning of that be-
to some larger population, qualitative research- haviour for the actor.
ers focus on understanding the behaviour, • “Artificial settings” versus “Natural settings.”
values, and beliefs of the people in their study. Whereas quantitative research is often conducted
Qualitative findings usually develop out of the in a contrived context, such as an experiment or a
context these people find themselves in; hence formal interview setting, qualitative researchers
the insights gained may or may not apply to investigate people in their natural environments,
other people. such as social gatherings or workplaces.

Key Points
• Qualitative research does not lend itself to a • Lincoln and Guba (1985; 1994) propose two key
clear set of linear steps. As a research strategy criteria to evaluate a qualitative study: trust-
it tends to be more open-ended than is typically worthiness and authenticity. Trustworthiness re-
the case with quantitative research. quires four conditions: credibility, transferability,
• In qualitative research, theories and concepts dependability, and confirmability.
are outcomes of the research process. • Most qualitative researchers try to see the
• There is considerable unease about the direct social world through the eyes of their research
application of reliability and validity criteria to participants.
qualitative research. Many writers prefer alter- • Like all approaches to social research, qualitative
native criteria. methods have both strengths and weaknesses.

bel65796_ch09_205-224 222 01/13/22 12:27 PM

 9 The Nature of Qualitative Research 223

Their major strengths include the development perspectives when making research decisions,
of empathetic understanding, rich description of difficulties in replication, limited generalization,
context, and emphasis on process. Weaknesses and lack of transparency.
include a reliance on the researcher’s own

Questions for Review (R) and Creative Application (A)

R1 Do research questions have the same charac- A4 Why would it be important to provide a rich de-
teristics and significance in qualitative research scription of the context in which new immigrants
as in quantitative research? Explain. to Canada find themselves in order to under-
A1 Imagine that you want to do a qualitative re- stand poverty in immigrant communities?
search project on homelessness in your city. R5 What are some of the main criticisms of qualita-
Come up with a general research question that tive research? Are those criticisms valid? Explain.
could be the starting point for your inquiry. A5 Sometimes qualitative studies are criticized be-
R2 Is the approach to theory in qualitative research cause they are difficult to replicate. Explain why
primarily inductive or deductive? Explain. critics see replication as a problem, then make
A2 Explain how a researcher could use grounded the case that this criticism reflects a misunder-
theory research to investigate how feelings of standing of methods and purposes of qualitative
guilt arise among victims of childhood sexual research.
abuse. R6 “The difference between quantitative and quali-
tative research revolves entirely around the con-
Evaluating qualitative research cern with numbers in the former and with words
R3 Why do qualitative researchers reject the no- in the latter.” To what extent is this statement ac-
tions of validity and reliability used in quantita- curate? Explain.
tive research? A6 The claim is sometimes made that quantitative
A3 Explain how respondent validation might be research focuses on behaviour while qualita-
used in a study of friendship networks among tive work is primarily concerned with meaning.
teenagers working at a fast-food restaurant, How might quantitative researchers investigate
and what difficulties might arise in using this issues of meaning, and qualitative practitioners
technique. study behaviour?
R4 What are the main goals of qualitative research-
ers? Explain.

Interactive Classroom Activities

1. Divide the class into small groups. Each group is g. Make up some qualitative data; for example,
to do the following: imagine how the person or persons stud-
a. Come up with a topic that could be re- ied might interpret the phenomenon under
searched using qualitative methods. study.
b. Produce a research question that could be Each group then presents a summary of their
used to investigate the topic. discussions to the rest of the class for comment
c. Select a hypothetical person or group to be and critique.
studied.
d. Describe some qualitative methods that could 2. Divide the class into six groups. The groups are to
be used to study the topic. marshal arguments making the following points:
e. Explain what verstehen means, and how it a. Group 1: Qualitative research is too subjec-
could be achieved in their research project. tive to be of any scientific value.
f. Explain how the research will be able to ex- b. Group 2: The idea that qualitative research is
amine the topic as a process. too subjective to be of any scientific value is

bel65796_ch09_205-224 223 01/13/22 12:27 PM

224 PART III Qualitative Research

false and is based on a misunderstanding of the results of qualitative studies can rarely be
the purposes of qualitative research. applied to people or groups not studied.
c. Group 3: Qualitative research is difficult or f. Group 6: Qualitative research can produce
impossible to replicate, so any conclusions knowledge of tremendous value regardless
based on it cannot be trusted. of whether the results can be generalized to
d. Group 4: Qualitative research can be used to other people or groups.
produce valid knowledge.
Each group presents its position to the class. General
e. Group 5: Qualitative research produces
class discussions on the positions taken are held
knowledge that is of little of no value because
after Groups 2, 4, and 6 have made their case.

Relevant Websites
The International Institute for Qualitative Examples of qualitative studies can be found in the
Methodology, based at the University of Alberta in journal Qualitative Research and in the Qualitative
Edmonton, sponsors conferences and research train- Sociology Review; you may have to access these
ing programs involving qualitative research meth- journals through your institution’s library server.
ods. It also publishes the International Journal of [Link]
Qualitative Methods, which is available free online.
[Link]/ENG/index_eng.
[Link]/international-institute-for-qualitative- php
methodology/[Link] (Websites accessed 4 September 2021)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch09_205-224 224 01/13/22 12:27 PM

 10 Ethnography and
Participant Observation
Chapter Overview
Ethnography and participant observation require extended involvement in the activities of the people
under study. This chapter explores:
• the problems of gaining access to different settings and ways of overcoming them;
• whether covert research is practicable and acceptable;
• the role of key informants;
• the different roles that ethnographers can assume in the course of their fieldwork;
• the function of field notes and the forms they can take;
• the role of visual materials in ethnography; and,
• bringing an ethnographic study to an end.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Describe the methodological and ethical issues involved in gaining and maintaining access to field sites
and participants, conducting overt and covert research in open and closed research settings, and work-
ing with key informants.
2. Identify the different roles ethnographers can take on while collecting data, including complete partic-
ipant, participant-as-observer, observer-as-participant, and complete observer, as well as how active
or passive they will be in these roles.
3. Discuss strategies for writing different types of field notes and analytic memos while conducting ethno-
graphic research.

▲ Chase Clausen/Shutterstock

bel65796_ch10_225-248 225 01/13/22 12:27 PM

226 PART III Qualitative Research

4. Describe the use of visual ethnography as a way of incorporating and analyzing visual materials such as
photographs in qualitative research.
5. Explain how institutional ethnography can be used to study the relationships between institutions and
people’s everyday lives.
6. Describe how ethnographers sample participants and field sites for inclusion in their studies and make
decisions about how much data is sufficient.
7. Explain how ethnographers decide when to conclude their data collection and why they find it difficult
to leave the field.

Do you like to travel to places you’ve never been to Ethnography and participant observation in-
before? Have you ever observed a group of people volve placing yourself in a social environment that
you don’t know very well and wondered what it would may be foreign to you, and staying there for an
be like to be a member of their group? Have you ever extended period of time. What kinds of groups or
witnessed profound human suffering and asked your- social settings intrigue you? Non-governmental
self how things ended up that way and how the people organizations? Political movements? Sports
suffering managed to endure? Would you like to give teams? Criminal gangs? Hospital emergency
such people a voice or expose the hardships that they rooms? Women’s shelters? All of these can be sub-
face? If so, doing or at least reading about ethnogra- jects of ethnographic and participant observation
phy and participant observation should interest you. research.

Introduction The terms “participant observation” and “ethnog-

raphy” are essentially synonymous, although here
Truly remarkable insights into the human condition “ethnography” will be used in a broader sense, in-
have been achieved by researchers who have im- cluding participant observation; the latter will refer
mersed themselves in a social setting for extended specifically to the observational component of eth-
periods of time (sometimes for years). Different nographic research.
terms are used to describe this process. Discussions This chapter outlines some of the main decisions
about participant observation have been fairly stan- that ethnographers must make, along with some of the
dard in textbooks on social research, but more re- many contingencies they face. However, given the
cently writers on research methods have preferred diversity of the experiences that confront ethnogra-
the word ethnography. In both cases, researchers in- phers and the variety of ways in which they deal with
teract with the people they are studying, listening to those experiences, it’s difficult to generalize about
what they say in conversation and asking them ques- this kind of research or to provide specific recom-
tions. Typically, participant observers and ethnog- mendations about it. The following comment makes
raphers gather further data through documents and this point well:
interviews, especially on matters that are not directly
observable, or about which they are unclear. Santin Every field situation is different and initial luck
and Kelly (2017), for example, used participant ob- in meeting good informants, being in the right
servation as their primary method to study flight at- place at the right time, and striking the right
tendants’ emotional labour, but they also conducted note in relationships may be just as important
interviews with 15 flight attendants, a survey of 40 as skill in technique. Indeed . . . many unsuc-
flight attendants, and a review of flight attendant cessful episodes are due as much to bad luck as
training manuals. to bad judgement. (Sarsby, 1984, p. 96)

bel65796_ch10_225-248 226 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 227

However, this statement should not be taken to mean example of Canadian research is given for each of
that forethought and an awareness of alternative ap- the four types.
proaches are irrelevant. It is with those issues that Three points should be noted about Figure 10.1.
the rest of this chapter is concerned. Issues involving First, the “open versus closed setting” distinction is
the conduct of interviews by ethnographers are re- not hard and fast. Sometimes the process of gain-
served for Chapter 11. ing access in “open” settings can take on an almost
formal quality; for instance, if you are talking to a
Access gang leader in the street, you may have to answer
some pointed questions about your goals. Also,
One of the most important and yet most difficult “closed” organizations and social movements some-
steps in ethnography is gaining access to the social times create contexts that have a public character,
setting one wants to research. How to gain access such as meetings arranged for members or prospec-
depends on several things, one of which is whether tive recruits. Bell (2007) attended many western
the setting is relatively open (public) or closed (pri- Canadian separatist meetings and rallies that were
vate or restricted) (see Lofland & Lofland, 1995). open to the public, and was able to get a lot of in-
Closed settings generally include organizations of formation on the movement and its leaders that way.
various kinds, such as firms, schools, cults, and Second, the “overt versus covert” distinction can
social movements. Open settings are areas where vary from context to context even within the same
anyone can gain access, such as libraries, parks, research project. Although ethnographers may seek
and sidewalks, but even there it can be quite dif- access through an overt route, many of the people
ficult to make observations and talk to people. with whom they come into contact may be unaware
“Open” does not necessarily mean easy access of their status as researchers. Atkinson (1981, p. 135)
to people. noted in connection with his research on the train-
ing of doctors in a medical school that although he
Overt versus covert ethnography was “an ‘open’ observer with regard to doctors and
One way to ease the access problem is to assume students,” he was “a ‘disguised’ observer with regard
a covert role: in other words, to not disclose that to the patients.” Also, some ethnographers move be-
you are a researcher. Getting access can be very tween the two roles (see Box 10.1).
problematic, and the adoption of a covert role Another interesting case is provided by
removes some of the difficulties. These two dis- Glucksmann (1994), who in the 1970s left her ac-
tinctions—open versus closed setting, and overt ademic post for a job on a factory assembly line
versus covert role—suggest a fourfold distinc- in an effort to find out why feminism appeared
tion in forms of ethnography (see Figure 10.1); an not to be relevant to working-class women. In a

Practical Tip | Micro-ethnography

In doing research for an undergraduate proj- focus on how staff members manage to inter-
ect or master’s thesis, it’s generally not feasible act and discuss work problems with one an-
to conduct a full-scale ethnographic study, if other despite continuous incoming calls and
only because of time limitations. Nevertheless, monitoring by supervisors. A shorter period of
it may be possible to carry out a form of mi- time (from a couple of weeks to a few months)
cro-ethnography (Wolcott, 1990), focusing on can be spent in the organization—on either a
one specific aspect of a group. For example, if full- or a part-time basis—to achieve this more
you were interested in call centres, you might limited goal.

bel65796_ch10_225-248 227 01/13/22 12:27 PM

228 PART III Qualitative Research

Open/public setting Closed setting

Overt role Type 1 Type 2
Wilson’s (2002) study of Toronto raves Karabanow’s (2002) study of two
Totten’s (2001) study of youth gang members Canadian youth shelters (Box 9.2)
Restoule et al.’s (2013) research
on Indigenous students

Covert role Type 3 Type 4

Desroches’s (1990) updates on homosexual Lauder’s (2003) discussion of covert
activities in public washrooms research on the Heritage Front

FIGURE 10.1 Four types of ethnography

Box 10.1 Perils of covert observation: Field notes in the lavatory

who shifted positions from covert to overt ob-
server at least in part because of those difficulties:

Right from the start, I found it impossible to

Toronto Star Archives/Toronto Star via Getty Images

keep everything that I wanted to remember

in my head until the end of the working day
. . . and so had to take rough notes as I was
going along. But I was stuck “on the line,” and
had nowhere to retire to privately to jot things
down [except the washroom]. [Eventually] . . .
my frequent requests for “time out” after in-
teresting happenings or conversations in the
bakehouse and the amount of time I was
During her undercover stint at Fiera Foods, Sara spending in the lavatory began to get no-
Mojtehedzadeh worked on the croissant production ticed. I had to pacify some genuinely con-
line. What kinds of difficulties might researchers cerned work-mates . . . and “come out” as an
face when attempting to collect information without observer—albeit in a limited way. I eventually
revealing their true identity and purpose? began to scribble notes more openly, but still
not in front of people when they were talking.
Ditton’s (1977) research in a bakery provides an When questioned about this, as I was occa-
interesting case study of the practical difficulties sionally, I coyly said that I was writing things
involved in taking notes during covert observa- down that occurred to me about “my studies”
tion, as well as an illustration of an ethnographer (Ditton, 1977, p. 5).

sense, she was a covert observer, but her motives which was eventually published under a pseud-
for the research were primarily political, not ac- onym. Was she an overt or a covert observer (or
ademic. At the time she was undertaking the re- neither, or both)? Whichever description applies,
search she had no intention of writing the book, this is an interesting case of what might be termed

bel65796_ch10_225-248 228 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 229

“retrospective ethnography”: using observations Access to closed settings

that were gathered before the decision was made Gaining access to most organizations requires strate-
to conduct a study. gic planning, hard work, and sometimes luck. In se-
A third point with regard to Figure 10.1 is that lecting a case study for an ethnographic investigation,
the preferred choice is an overt role. There are sev- the researcher may employ several criteria, depending
eral reasons for this. As Box 10.2 reveals, these rea- on the research area of interest. One may choose a
sons have to do with practical and especially ethical certain case or setting because of its “fit” with the re-
considerations. The ethics of covert research, which search questions, but there is no guarantee of access.
were addressed in depth in Chapter 3, involve issues Sometimes sheer perseverance pays off. Leidner (1993)
of deception, lack of informed consent, and inva- was determined that McDonald’s would be one of the
sion of privacy, since the researcher has essentially organizations in which she conducted ethnographic
fooled the participants into believing that they are research on the routinization of service work:
not being observed for research purposes. Because
of the ethical problems (and some of the practical The company was a pioneer and exemplar of
difficulties) that beset covert research, the bulk of routinized interaction, and since it was locally
the following discussion of access issues focuses on based, it seemed like the perfect place to start.
overt research. McDonald’s had other ideas, however, and

Box 10.2 The covert role in ethnography

Advantages • Anxiety. Ethnography is frequently a stress-
ful research method, and worries about de-
• Easier access. Adopting a covert role can help
tection only add to the anxiety. Moreover,
with access problems; no special permission
if the ethnographer is found out, the whole
to gain entry to a social setting or organiza-
research project may be jeopardized. In
tion may be needed.
extreme cases, such as covert research on
• Less reactivity. Because participants do
gangs, being found out could even put the
not know they are being observed by a re-
researcher’s life at risk.
searcher, they will speak and act more natu-
rally than they would otherwise. • Ethical problems. Covert observation trans-
gresses two important ethical tenets: deception
Disadvantages of participants and failure to obtain informed
• The problem of taking notes. It is difficult and consent (see Chapter 3). It can also involve a
sometimes impossible to take notes without violation of privacy. Indeed, many writers think
revealing that you are conducting research. covert investigations can harm the practice of
However, notes are very important to an eth- research in general, since innocent researchers
nographer, and it’s too risky to rely on memory may be tarred with the same brush as outed
alone. covert observers, who are considered snoops
• The problem of not being able to use other or voyeurs (or worse). On the other hand,
methods. If the researcher is in a covert role, covert participant observation may be justified
steering conversations in a certain direction in some situations, as Lauder (2003) concluded
can increase the risk of detection, and it is es- regarding his investigation of the neo-Nazi
sentially impossible to conduct interviews. Heritage Front.

bel65796_ch10_225-248 229 01/13/22 12:27 PM

230 PART III Qualitative Research

only after tenacious pestering and persuasion • Be frank about the amount of people’s time
did I overcome corporate employees’ polite you are likely to need. Time is an important
demurrals, couched in terms of protecting issue for commercial organizations, and for
proprietary information and the company’s many non-profit groups too.
image. (Leidner, 1993, pp. 234–235)
Access to open settings
This kind of determination is necessary whenever
Gaining access to people in open settings is similar
the target is a specific organization or group (e.g.,
in many ways. For instance, sometimes ethnogra-
a particular religious sect or social movement).
phers have their paths smoothed by “sponsors” and
Rejection can mean having to find a completely dif-
“gatekeepers.” In seeking access to a group of soccer
ferent research topic.
hooligans, Giulianotti (1995) sought out someone
On the other hand, for many research ques-
who could play both roles for him. Later, he was able
tions you are likely to find that several potential
to call on existing acquaintances to ease his entrée
cases will be suitable. Organizational researchers
into a second group. In seeking access to female in-
have developed a range of tactics, some of which
travenous drug users, Taylor (1993) contacted a drug
may seem rather unsystematic but are still worth
counsellor who introduced her to some local users
considering:
and accompanied her on her first few research visits.
A “research bargain” (Becker, 1970) was struck in
• Use friends, contacts, and colleagues to help which Taylor agreed in return to see any of the coun-
gain access; as long as the organization is rele- sellor’s clients who preferred to discuss issues with a
vant to the research question, the route should woman.
not matter. “Hanging around” is another common access
• If possible, get someone in the organization to strategy. An example of the difficulties that await
vouch for you and the value of your research. the researcher using this approach can be found
Such people are considered “sponsors.” If per- in one of Whyte’s (1955) early field encounters in
mission is granted at a lower level of the hier- Boston’s North End in his classic case study Street
archy, you may still need clearance from top Corner Society. The following incident occurred in
management or senior executives. People who a hotel bar:
must grant their permission for you to gain
I looked around me again and now noticed a
access are sometimes called gatekeepers.
threesome: one man and two women. It oc-
• Offer something in return: for example, a final
curred to me that here was a maldistribution
report (if you provide a working draft, people
of females that I might be able to rectify. I ap-
may ask you to make changes that you would
proached the group and opened with some-
not consider acceptable). However, this strategy
thing like this: “Pardon me. Would you mind
carries a risk of making the researcher a cheap
if I join you?” There was a moment of silence
consultant. Some writers on research method-
while the man stared at me. He then offered to
ology do not recommend offering something
throw me downstairs. I assured him that this
in return, although it is commonplace among
would not be necessary and demonstrated as
researchers on formal organizations.
much by walking right out of there without
• Provide a clear explanation of your aims and
any assistance. (Whyte, 1955, p. 289)
methods. Suggest a meeting to deal with the
participants’ concerns and explain what will Wolf (1991) used a hanging-around strategy to gain
happen in terms that people who are not fa- access to Canadian outlaw bikers. On one occasion
miliar with social research can understand. he met a group of them at a motorcycle shop and
• Be prepared to negotiate; almost no one gets expressed an interest in hanging out with them. But
complete access. he tried to move too quickly in seeking information

bel65796_ch10_225-248 230 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 231

about them and was forced to abandon his plans. Ongoing access
Eventually, Wolf’s hanging-around strategy led to Access problems do not end after initial contact and
his being approached by the leader of a biker group entrée to the group. Maintaining access is an ongo-
(Rebels MC), who acted as his sponsor. To bring this ing activity and is likely to prove a persistent prob-
off Wolf ensured that he was properly attired in biker lem in closed contexts such as organizations.
garb, although this was not difficult for him as he had
been a motorcyclist and had absorbed biker culture • People will be suspicious, perhaps seeing the re-
long before he started his research. Attention to dress searcher as an instrument of top management
and demeanour can be a very important consideration (it’s very common for members of organiza-
when seeking access to either public or closed settings. tions to believe that researchers have been put
Sometimes, researchers may gain access to a set- in place to check up on them). When conduct-
ting before they even think of doing a study there. ing his research on the provision of care in a
Box 10.3 provides an example. Canadian emergency shelter for homeless men,
As these anecdotes suggest, gaining access to Ranasinghe (2017) was suspected by some of the
social settings is a crucial first step in ethnographic shelter’s clients of being a police officer.
research. But it is also difficult and, in some cases, • Group members will worry that what they say
dangerous—for example, when the group in question or do may get back to bosses or colleagues. Van
is engaged in violent or criminal activities. Caution Maanen (1991) noted that when conducting
is essential when choosing an access strategy. ethnographic research among police officers,

Box 10.3 Access to Inuit: From friend to researcher

When Lisa-Jo van den Scott moved to Arviat, As van den Scott became increasingly immersed
Nunavut, as the wife of a music teacher, she in the day-to-day life of Arviat, her sociological cu-
had no intention of researching Inuit culture. riosity about Inuit identity, space, and the built en-
She had recently finished a master’s degree vironment was piqued. In mentioning her research
in ancient Greek and did not envision get- ideas to Inuk friends, she found that although they
ting caught up in conducting an ethnographic did not understand her sociological interests, they
study. Taking to her new life in the remote were eager to assist. Outside researchers do not
northern hamlet, she began developing friend- always receive such a hospitable reception, but
ships, learning Inuktitut, volunteering at a through honest demonstrations of commitment,
drop-in centre, and working as an instructor van den Scott had become accepted as a friend
at Nunavut Arctic College. With a genuine in- and community member. Key to her ongoing access
terest in becoming involved in the community, was an emphasis on learning from her participants
she often moved out of her comfort zone, trying rather than studying them. The difference is subtle
new foods like igunak (raw, aged walrus meat), but important. It involves a shift in the power dy-
and taking part in throat singing. So complete namic, wherein the researcher becomes a student
was van den Scott’s connection to the com- to the experts on a given way of life, the participants.
munity, she was elected to the Elders’ council It also encapsulates van den Scott’s deep respect
as “the first non-Inuk, first non-Elder on the for the people that welcomed her into their lives and
council” (L.K. van den Scott, 2018, p. 22). The shared their stories with her. She writes that she was
relationships she developed were reciprocal, “committed to engaging with them with all the re-
trust-based, and sincere. spect due to research participants as experts in their
(Continued)

bel65796_ch10_225-248 231 01/13/22 12:27 PM

232 PART III Qualitative Research

own lives” (p. 24). Starting her research from the As van den Scott’s experience reveals, entrée
point of friendship also meant reminding herself that into a field site can happen unexpectedly as a
even if her friends did not always remember that she result of chance life circumstances. A positive per-
was a researcher, she could not lose sight of this. She sonal reputation and connection to participants
strove to be reflexive, taking into account her posi- can present opportunities for access, but it is also
tion as a White researcher in a colonial setting, while a balancing act as the ethnographer must take
also being a friend to her participants. She states, into account their roles as both friend and re-
“That meant deciding what information made it searcher. Fostering trust and rapport is often best
into my field notes. Would being a friend affect my achieved through an ongoing attentiveness to
findings? Of course—but in which way? How much? maintaining mutual respect between researcher
What should I do?” (p. 26). and participants.

it’s not unusual to observe activities that reflect Similar considerations apply to research in public
badly on the police and may even be illegal. In settings:
such situations the researcher can establish
credibility by using discretion. • Make sure you have a plan for allaying people’s
• If the people being studied have concerns or suspicions. Ranasinghe (2017) “tried to ensure
suspicions about the researcher, they may that while [he] was conspicuous, [he] was, nev-
appear to go along with the research but in ertheless, not obtrusive or obstructive” (pp.
fact sabotage it, engage in deception, provide 235–236). Over time he found that the emer-
misinformation, or not allow access to “back gency shelter’s clients became less fearful and
regions” (Goffman, 1956). were more receptive to him.
• Be prepared for tests of either competence or
Here are three ways to smooth the path for ongoing credibility. When doing research on gang mem-
access to closed settings: bers in a poor community, Horowitz wrote that
she was frequently told “confidential” stories
• Be prepared to adapt how you present yourself. (which turned out to be fictional) to see if she
Sometimes it’s helpful to play up your creden- could keep a secret (Gerson & Horowitz, 2002).
tials: talk about your past work, your experi- • Be prepared for changes in circumstances.
ence, your knowledge of the group, and your With an interest in how ultra-Orthodox Jews
understanding of its problems. Other times it’s manage social change, Shaffir (2018) worked as
more conducive to take on a student role, pre- a clerical assistant with the Tasher—a Hasidic
senting yourself as someone who is interested Jewish sect—covertly observing their way of life
in learning from the experts as van den Scott in an enclave north of Montreal. When he told
did in her research on Inuit (see Box 10.3). a couple members of the group that he had to
• Don’t give people a reason to dislike you: be reduce his workload so that he could focus on
non-judgmental when hearing about informal his master’s thesis, they suggested he study their
activities or about the organization; make sure culture. By the next day they had changed their
that information given to you doesn’t get back minds and asked him to leave his position due
to others, whether bosses or peers. to a lack of funds. As an office clerk, though,
• Help out occasionally with work, or offer he had made contacts within the community,
advice without being critical. which allowed him to remain in the field.

bel65796_ch10_225-248 232 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 233

Key informants participants develop a sense of the events or encoun-

Ethnographers rely heavily on informants, especially ters that the ethnographer will want to see. However,
those who develop an understanding of the research as Hammersley and Atkinson (1995, pp. 130–131)
and are able to identify situations, events, or people observed, it’s important to recognize that such offers
likely to be helpful to the investigation. Whyte (1955) may on occasion be staged for the ethnographer’s
again provides an extreme example, quoting a key benefit.
informant named Doc: Solicited accounts can be obtained in two ways:
by interviewing (see Chapter 11) or by casual ques-
You tell me what you want to see, and we’ll tioning during conversations (though in ethno-
arrange it. When you want some information, graphic research the boundary between an interview
I’ll ask for it, and you listen. When you want to and a conversation is by no means clear). When
find out their philosophy of life, I’ll start an ar- the ethnographer needs specific information on an
gument and get it for you. If there’s something issue that does not lend itself to direct observation
else you want to get, I’ll stage an act for you. or that is not cropping up during “natural” conver-
(1955, p. 292) sations, solicited accounts are likely to be the only
Doc was also helpful in warning Whyte that he was way forward.
asking too many questions, telling him to “go easy
on that ‘who,’ ‘what,’ ‘why,’ ‘when,’ ‘where’ stuff” Roles for ethnographers
(1955, p. 303). Shaffir (2018) discusses his good
fortune in finding his “Doc” in the Tasher ultra- Related to the issue of ongoing access is the question
Orthodox Jewish community: of the role an ethnographer adopts in relation to the
social setting and its members. Often the various
And when I established a favourable relation-
roles in fieldwork are arrayed along a continuum
ship with a Hasid whose role in the commu-
from complete involvement to complete detachment
nity was central—in effect, he came to serve as
(see Figure 10.2). It should also be noted that the full
my “Doc,” a key informant—my connections
gamut of roles may be employed at different times in
to Tasher increased exponentially. I could and
the course of a single ethnographic project, and for
would visit the enclave at will, meandering in
different purposes.
the synagogues and shopping centre, and en-
For example, the participant observer can be any
gaging Hasidim in friendly conversations. I
of the following:
had gained a measure of trust and was invited
to establish a website featuring the community
• Complete participant. The ethnographer is a
that included several of my publications about
fully functioning member of a social setting
the community. (2018, p. 324)
but one whose true identity is unknown to
Key informants can also provide support that helps members: in other words, a covert observer. In
with the stress of fieldwork. However, undue reliance such cases the ethnographer is engaged in reg-
on them can lead researchers to see social reality ular interaction with people and participates
through their eyes only rather than viewing things in their daily lives but assumes the researcher’s
from the perspectives of several group members or
the group as a whole. Involvement Detachment
In addition, ethnographers encounter many other
people who will act as informants whose accounts Complete Participant-as- Observer-as- Complete
may or may not have been solicited. Some research- participant observer participant observer
ers prefer unsolicited information because of its FIGURE 10.2 Classification of participant
spontaneity and naturalism. Very often, research ­ bserver roles (Gold 1958)
o

bel65796_ch10_225-248 233 01/13/22 12:27 PM

234 PART III Qualitative Research

role in private to write down notes once the sit- inferences. In an interesting variant of this problem,
uation has unfolded. This is what Humphreys Hessler et al. (2003) cautioned observers (in a study
(1970) did when he studied male homosexual of adolescent risk behaviour) against the temptation
encounters in public washrooms. to act as counsellors to respondents.
• Participant-as-observer. The ethnographer Gans (1968) would add that even if it were pos-
participates as described above, except that sible, it is probably undesirable to adhere to a single
members of the social setting are aware that role over the entire course of a project, since this
the ethnographer is studying them, such as in would limit the ethnographer’s flexibility in han-
van den Scott’s research (see Box 10.3). dling situations and people. In addition, the risks
• Observer-as-participant. The researcher is of either excessive involvement or excessive detach-
mainly an interviewer and observer, and partic- ment would loom large.
ipates only marginally in the group’s activities.
This was Bell’s (2007) role in his study of western Active or passive?
Canadian separatists. Ethnographic research on How active or passive should the ethnographer be?
the police is often similar, since the opportunities In most studies—even when the researcher is mainly
for genuine participation are limited by legal and an observer rather than a participant—involvement
safety restrictions. Norris (1993) described how in the group’s activities will be unavoidable from
as an observer-as-participant, he concentrated time to time. For example, Fine’s (1996) research
on gathering two types of data: “naturally occur- on the work of chefs in restaurants was carried out
ring inter-officer talk” and “detailed descriptions largely through semi-structured interviews. In spite
of how officers handled ‘live’ incidents” (1993, of this limited participation, he sometimes found
p. 126). For other examples of situations in which himself in the kitchen washing dishes to help out
minimal participation may be appropriate, see during busy periods. Similarly, in a study of bounc-
Boxes 10.4 and 10.6. ers, the participant observer will not have the luxury
• Complete observer. The final possibility is no of deciding whether to become involved in fights,
interaction with the people observed. Most since involvement comes with the territory (Winlow,
writers do not include this as a form of eth- Hobbs, Lister, & Hadfield, 2001).
nography, since by definition there is little or Sometimes ethnographers feel they have no
no involvement or participation. There may be choice about getting involved because a failure to
less risk of reactivity (participants behaving participate might suggest a lack of commitment
unnaturally because they know they are being to the group. In such situations, standing back
observed) because the researcher is at a dis- could lead to a loss of credibility in the eyes of the
tance; but there is also greatly reduced poten- people studied. This situation can arise especially
tial for understanding because the researcher when the activities are illegal or dangerous. On the
does not ask questions or try in any way to get other hand, many writers counsel against active
into the heads of the people under study. participation in criminal or dangerous activities.
Both Armstrong (1993) and Giulianotti (1995) re-
Each approach has advantages and disadvantages. fused to fight while doing their research into soccer
In the case of covert ethnography, some of these are hooliganism. The latter wrote: “My own rules are
discussed in Box 10.2. The participant-as-observer that I will not get involved in fighting or become a
role carries the risk of overidentification with the go-between for the two gangs in organizing fights”
group being studied (see Boxes 10.5 and 10.6), but at (1995, p. 10). That is an illustration of why it is not
the same time it offers an opportunity to get close to a good idea to do covert research on criminals or
people. The observer-as-participant role carries the those involved in dangerous actions: it is much
risk of not understanding the social setting and its more difficult for someone in such a role to decline
people sufficiently and therefore making incorrect to participate in group activities (see Box 10.5).

bel65796_ch10_225-248 234 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 235

Box 10.4 Virtual ethnography

ethnography is conducted offline; the authors
themselves only ventured into cyberspace after
their attempts to use solely in-person methods
proved to be unworkable.
In researching an organization for undocu-
mented college students in California, one of the
authors began by researching the life history of a
particular student, but found it difficult to keep in
sjenner13/123RF

touch with him. The student didn’t have a phone,

and email messages often went unanswered. The
researcher soon learned that the most depend-
able means of communication was Facebook,
What challenges might an ethnographer face
which the student checked many times a day.
when studying a virtual community as opposed to
an offline community? Would these issues affect Facebook also proved to be an invaluable source
how active or passive the ethnographer must be? of ethnographic information about the organi-
zation, how it established a sense of community,
Hallett and Barber (2014) argue that the study and how its members viewed the plight of undoc-
of online spaces should be included in ethno- umented students. The author states that he “had
graphic research because people’s “natural habi- considered asking participants to journal about
tat” now includes their “online habitat.” Instagram, how they spent time and perceived the educa-
Facebook, Twitter, email, instant messaging, tional process. I soon realized that students were
Skype, YouTube, blogs, and so forth are now already ‘journaling’ on their Facebook pages, and
predominant places for participating in social these posts were more natural than I may have
networks and acquiring information and inter- received if they had used paper and pencil to
pretations of the world. Despite that fact, most write at my request” (316).

Box 10.5 Overidentification with the people one is studying

Sometimes, as a result of prolonged immer- Canadian police officers while doing “ride-
sion in the lives of the people being observed, alongs” with them, she had to contend with
coupled with a commitment to seeing through a feeling that she may have become overly
their eyes, ethnographers become so wrapped accepting of the officers’ beliefs and values.
up in the world of the people they are study- Similarly, Hobbs (1988, p. 6) wrote of his field-
ing that they forget their role as researchers. work that he “often had to remind himself that
Under such circumstances the ethnographer [he] was not in a pub to enjoy [himself] but to
may find it difficult to maintain a social sci- conduct an academic inquiry, and repeatedly
entific perspective on the collection and anal- woke up the following morning with an incred-
ysis of data. For example, when Schulenberg ible hangover, facing the dilemma of whether
(2014) conducted participant observation of to bring it up or write it up.”

(Continued)

bel65796_ch10_225-248 235 01/13/22 12:27 PM

236 PART III Qualitative Research

A related issue for Hobbs was involvement be willingly involved in “normal” business
in illegal activity, which may be tempting if one transactions, legal or otherwise. I was pur-
becomes overly acculturated to a group in which suing an interactive, inductive study of an
criminal actions are known to occur: entrepreneurial culture, and in order to do
so I had to display entrepreneurial skills
A refusal, or worse still an inquiry concern- myself. . . . [My] status as an insider meant
ing the legal status of the “parcel,” would that I was afforded a great deal of trust by
provoke an abrupt conclusion to the rela- my informants, and I was allowed access to
tionship. Consequently, I was willing to skirt settings, detailed conversations, and infor-
the boundaries of criminality on several mation that might not otherwise have been
occasions, and I considered it crucial to available. (1988, pp. 7, 15)

Box 10.6 Annoyance or animosity toward the people one is

observing
Overidentification with the people one is ob- residents. Nonetheless, she felt she “was gath-
serving is not the only risk that participant ering good data, despite [her] feelings of being
observers face in relation to the social situa- an outsider” (2000, p. 120). Similarly, although
tions they study. Lee-Treweek (2000) carried out Bell (2007) spent more than a year observing
research on auxiliary caregivers in two homes Western Canadian separatists, at no point was
for the elderly. In one of them she reacted in he ever tempted to join the group. The lesson
a way that was the opposite of overidentifi- here is that overidentification is not inevitable,
cation. She disliked the home and found the and that researchers must be careful not to
staff unappealing because of their uncaring let either positive or negative feelings for the
approach and lack of sympathy for the elderly people they study distort their findings.

Field notes location, the people involved, what prompted

the exchange or event, and date and time of day.
Because of the frailties of human memory, ethnog- • You may also use an audio recorder to take
raphers must jot down their observations. These notes from later, but you will have to allow
should be fairly detailed summaries of events and plenty of time for transcription.
behaviour, and the researcher’s initial reflections on • Notes must be vivid, clear, and complete. If in
them. The notes need to specify the key dimensions doubt, write it down.
of whatever is observed or heard. Here are some gen-
eral principles: It’s obviously best to take notes as soon as something
significant happens. However, wandering around
• Write down notes, however brief, as soon as with a notebook and continually scribbling in it can
possible after seeing or hearing something make the people you are observing feel self-con-
interesting. scious. Thus, it may be necessary to take small
• Write up full field notes at the end of the day amounts of time away from the group to write down
at the very latest, including details such as your observations. In that case you must take care

bel65796_ch10_225-248 236 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 237

not to make the participants anxious or suspicious findings to the social scientific literature on the topic.
(see Box 10.1). This approach is implied by the sequence suggested
To some extent, strategies for taking field in Figure 9.1.
notes depend on how clearly delineated the re- For most ethnographers, the equipment needed
search questions are. As we noted in Chapter 9, for observation consists mainly of a notepad and
most qualitative research begins with general re- pen (for example, Armstrong, 1993, p. 28). An audio
search questions, but there is considerable varia- recorder may be useful but, as we have noted, tran-
tion in their specificity. Obviously, when there is scribing what you record will take time, and care
a clearly focused research question, the ethnogra- must be taken to ensure the audio recordings are not
pher’s observations should be oriented to it. At the disseminated, as they may unintentionally identify
same time, it’s important to maintain a fairly open research participants. In addition, recording may be
mind, so that the study’s flexibility with respect to more obtrusive than writing notes. Although partic-
the themes it explores—a strength of qualitative ipants usually grow accustomed to the presence of
research—is not impaired. the researcher over time, speaking into a recording
In the context of her research on female drug device can rekindle fears about participating; and
users, Taylor (1993, p. 15) explained that in her early in some gatherings it may be difficult to use one be-
days in the field she tended to listen rather than talk cause of extraneous noise. Photography can be an
because she “did not know what questions [she] additional source of data, and pictures do help to
wanted to ask.” Armstrong (1993, p. 12) wrote in stir the ethnographer’s memory, but in some kinds
connection with his research on soccer hooligan- of research (especially involving crime and deviance)
ism that his research “began without a focus” and photography may not be feasible.
that as a result “he decided to record everything.”
Therefore, a typical Saturday would mean 30 pages Types of field notes
of handwritten notes. Such open-endedness usually Some writers have found it useful to classify the types
cannot last long, because trying to record the details of field notes generated in the process of conducting
of absolutely everything is so tiring. Eventually the ethnography. The following classification is based on
ethnographer begins to narrow down the focus of categories suggested by Lofland and Lofland (1995)
the research, sometimes by relating the emerging and Sanjek (1990):

Research in the News

An ethnographic study of bus-riding behaviour in Vancouver
As Lisa-Jo van den Scott’s (2018) research on in action. In a way, there are two layers—there’s
Inuit culture demonstrates (see Box 10.3), ev- the microcosms of individual behaviour and in-
eryday happenstance can inspire research teractions, as well as the macrocosm of the city
questions. This was the case for Dr. Amy Hanser, as a whole” (Plana, 2019). In the first part of her
an Associate Professor of Sociology at the exploratory study, she rode the bus for three
University of British Columbia (Plana, 2019). As hours at a time and observed the passengers.
a regular bus rider in Vancouver, she became She spent another two hours writing field notes
curious about the unwritten rules of public and recording her observations. The next phase
transit. In an interview with the Daily Hive she of her research will include additional observa-
stated, “The bus struck me as a great place to tion as well as interviews with passengers and
watch and see [mixing and diversity in the city] drivers.

bel65796_ch10_225-248 237 01/13/22 12:27 PM

238 PART III Qualitative Research

• Mental notes. These are particularly useful the methodological problems encountered, and so on.
when it is inappropriate to be seen taking writ- In addition, it’s possible to create analytic memos re-
ten notes; however, they should be put into cording some initial thoughts on what all those obser-
writing at the earliest opportunity. vations might mean. Analytic notes help to bridge the
• Jotted notes (also called scratch or rough notes). gap between the data and the concepts, interpretations,
The researcher makes very brief notes to jog the and theories that researchers develop to make sense of
memory about events that should be written what they are observing. Such memos should be dated
up more fully later. Lofland and Lofland (1995, and reviewed regularly. They should also be kept sepa-
p. 90) described jotted notes as “little phrases, rate from notes on actual observations because they are
quotes, key words, and the like.” Such notes comments on data, not data themselves.
need to be written inconspicuously, preferably
out of sight, since writing in front of research The rise of visual
participants may be off-putting for them. ethnography
• Full field notes. These are the main data source
in ethnographic research. Full notes should be The use of visual materials in social research is by no
written as soon as possible, usually at the end means new. Since the nineteenth century, cultural
of the day. Provide as much detail as possible anthropologists have used photographs in their por-
about events, people, conversations, and so on. trayals of traditional societies. However, books such
A particular comment from a participant or as Pink’s (2001) Visual Ethnography have marked a
an account of an event that you witnessed may transition to a different way of using photos in social
not seem very important at the time, but its sig- research.
nificance may become apparent later on, once In Chapter 12 we will discuss visual materials
you have developed an interpretive structure. (primarily photographs) that were produced some
Without good notes, valuable data may be lost. time before the research began. Here we will con-
Write down, in brackets, initial ideas about your sider visual materials that are produced more or less
interpretations, impressions, and feelings. exclusively for research purposes: research-driven
images, mainly photographs but also video record-
It’s worth adding that in field notes the ethnogra- ings and other visual media. It’s worth noting that
pher’s presence is frequently evident. This can be although the term “visual ethnography” is increas-
seen in Whyte’s account (above) of almost being ingly popular, it does not always refer to the kind of
thrown down the stairs. However, in the finished sustained immersion in a social setting that is the
work the ethnographer is sometimes written out of subject of this chapter.
the picture. Field notes, except for brief passages, Qualitative researchers use visual materials in at
are primarily for the researcher’s own use (Coffey, least three ways:
1999), whereas the written report is for public con-
sumption and has to be presented as a credible ac- • as memory aids in the course of fieldwork,
count of the social setting and culture in question. where they essentially become components of
References to the ethnographer should be limited; the ethnographer’s field notes
otherwise, the report will look more like an account • as sources of data in their own right
of his or her activities than those of the group pur- • as prompts for discussion by research participants
portedly under study.
Pink (2001) drew an important distinction be-
Analytic memos tween two positions on visual materials. The tra-
Most of the previous discussion focused on the me- ditional framework is a realist one in which the
chanics of note taking. But content is even more im- material simply captures an event or setting that
portant. Notes can be taken on the setting, the people, then becomes a “fact” for the ethnographer to

bel65796_ch10_225-248 238 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 239

Methods in Motion Applications to Canadian Society

Contemporary consequences of the historic relationship between Indigenous people
and Euro-Canadian power structures
Browne (2007) used ethnographic methods and interpretation was that the patients’ behaviour
participant observation to examine how the his- stemmed from cultural factors. As one nurse put
toric political, social, and economic inequalities it, “[This patient] wouldn’t look at me. She wouldn’t
experienced by Indigenous women affected their answer anything. She would nod occasionally. . . .
interactions with nurses in a hospital in western I mean, it’s a cultural thing” (p. 2169). One of the
Canada. She explored how socially constructed patients explained things differently: “. . . a lot of
notions of culture, “race,” and ethnicity can influ- native women, especially, don’t want to say any-
ence the quality of the health care that Indigenous thing, don’t talk. . . . I know my mom, for example,
women receive. Browne, who is both a nurse and had a lot of fear in her from residential school”
a researcher, spent nine months conducting the (p. 2170). Browne points out that these misunder-
study, which involved doing in-depth interviews standings can have serious health consequences,
with 14 registered nurses (four of whom were as in the case of an Indigenous woman who did
Indigenous) and 14 Indigenous female patients. not mention that her intravenous needle had not
She also shadowed the nurses as they worked been removed before she left the hospital. Her
with the Indigenous patients. study shows how routine micro-level interac-
One thing Browne noted was that the nurses tions, such as those between patients and nurses,
tended to ignore or downplay historically based may be influenced by larger, historically based
power inequalities when they tried to make relationships such as those between Indigenous
sense of their patients’ demeanour, which was people and the dominant Euro-Canadian society
often passive, quiet, and introverted. A common and state.

interpret along with other data. The image and what it Further, it acknowledges that different people may
represents are presented as unproblematic, as a interpret the images in different ways. In Pink’s
window on reality. This has been the dominant research on Spanish bullfighters, enthusiasts inter-
framework within which visual resources have been preted the images she took of bullfights in terms
produced and analyzed. By contrast, Pink also drew of bullfighter performance. Other viewers might
attention to a position that she described as reflex- see the same images in a different interpretive
ive. This entails an awareness of and sensitivity to framework—for example, one centred on animal
the ways in which researchers themselves determine rights and cruelty.
what the visuals reveal. This sensitivity requires Plate 10.1 presents an image called The Bullfighter’s
a grasp of the ways in which researchers’ personal Braid, from Pink’s research. Showing a female bull-
characteristics—age, gender, social background, fighter, it appealed to many connoisseurs of the sport
academic proclivities, and so on—may influence who saw in it various artistic and other meanings.
what images they choose to record, and how those For Pink, it held additional significance in terms of
images are presented. her interests in gender and the broader discipline of
This approach to the visual is frequently collab- social anthropology (for more information on this
orative in the sense that research participants may photograph see Pink, 2001, p. 101, and [Link]
be involved in decisions about what visuals should [Link]/revealingpictures/[Link] [ac-
be recorded and how they should be interpreted. cessed 4 September 2021]).

bel65796_ch10_225-248 239 01/13/22 12:27 PM

240 PART III Qualitative Research

used the technique to explore how these young

people confronted the “hostile gaze” that is often
directed at people in the LGBTQ community. The
15 people in the study were given training in pho-
tography and then asked to take five to ten pictures
that reflected their experiences as queer or trans
youth. The participants were then invited to discuss
their photos by responding to questions such as,
“What is seen here?” and “How could this image be
used to educate people?” One of the themes to come
out of the discussion was invisibility, whereby par-
ticipants were not recognized as having the gender
or sexual identity that they held. This experience
was particularly painful when it occurred within
the LGBTQ community itself. To illustrate invisi-
Pink, Visual Ethnography, 2001

bility, one participant took a photograph of empty

clothing arranged to make it appear as if a person’s
body were missing.
Visual research methods require researchers to
“read” images while being sensitive to the context
in which they were generated; the potential for
Plate 10.1: The Bullfighter’s Braid
multiple meanings among researchers, study par-
ticipants, and others; and the potential for the re-
Photovoice is an approach to visual ethnog- searcher to influence the image and its presentation.
raphy that takes collaboration with research In addition, researchers usually include non-visual
participants even further. Using this technique, research methods, such as interviews, in their in-
the participants take photographs of their daily vestigations. This raises the question of the relative
experiences and interpret them in terms of their significance of words versus images in the analy-
own perspectives and identities, which trans- sis of data and the presentation of findings. Since
fers a measure of power and control away from words are the traditional medium, it’s easy to slip
the researcher to them. Since photovoice may be into seeing the visual as less important, but this is
used with people from marginalized communi- not always justified. Finally, visual research meth-
ties, those who employ it often do so with the in- ods raise certain ethical issues, such as invasion of
tention of creating social change, which can be privacy and anonymity.
undertaken through an exhibition of the photo-
graphs. When used in this way, photovoice is a Institutional ethnography
form of participatory action research (discussed
in Chapters 1 and 9). With the power to cross Canadian sociologist Dorothy Smith has developed
cultural boundaries and language barriers, pho- an approach called institutional ethnography
tographs can also help make findings accessible (Smith, 2005) that explores how institutional dis-
to a wide range of academic and non-academic courses (typically texts found in the workplace)
audiences (see Box 10.7). relate to people’s everyday experiences with in-
Holtby, Klein, Cook, & Travers (2015) used pho- stitutions, and how examining institutional re-
tovoice in their research on queer and trans youth lationships may reveal larger systems of social
in the Kitchener–Waterloo area. The researchers control and power in a society (Devault, 2006).

bel65796_ch10_225-248 240 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 241

Box 10.7 Photovoice as a knowledge dissemination and social

justice strategy
Ratković and Sethi (2018) make a case for using Gallery, “The research launch was attended by
photovoice as part of an arts-based approach over 150 people, including the local member
to knowledge dissemination. As researchers of parliament, the mayor’s office, the local
committed to social justice, they seek to bring member of provincial parliament, employers,
practitioners, policy-makers, and community health care practitioners, academics, social
members into conversations about important workers, immigrants, refugees, religious lead-
issues such as racialization and citizenship. ers, and community members. The local media
Sethi describes a study she conducted on also wrote about the exhibit” (Ratković & Sethi,
Canadian immigrant and refugee women, 2018, pp. 344–345). Using photographs as part
where participants were given a camera and of her approach to knowledge dissemination,
taught by a photographer how to take profes- Sethi was able to achieve what many research-
sional pictures that captured their perspectives ers find difficult to accomplish with traditional
and experiences. She then worked with par- text-based research outputs such as books
ticipants to select photos that they would be and journal articles. Ratković and Sethi (2018,
comfortable sharing with the public. A graphic p. 345) write, “Reaching practitioners, policy-
artist helped turn the images into art pieces, makers, and community members can lead
which were displayed at art exhibits across to new understandings of citizenship and new
southwestern Ontario. For the opening exhibit alliances between visible minority immigrant
at Wilfrid Laurier University’s Yellow Brick Wall women and the Canadian state.”

Institutional relationships are analyzed in detail, a post-secondary education, and that their parents
with the aim of finding out how they affect the were in favour of it as well, a much smaller propor-
personal experiences of the individuals involved. tion of Indigenous youth than non-Indigenous youth
The approach can be explicitly change oriented actually graduated from those institutions. The re-
in that previously unrecognized opportunities searchers found that some informants declined to
to transform institutions may come to light as identify as Indigenous in their applications, citing a
the research runs its course, and the researcher lack of trust in government institutions, which itself
may even collaborate with the people studied in indicated a problematic relationship between pro-
spective Indigenous students and post-secondary
an effort to create institutional relationships that
bureaucracies. Participants also stated that the pro-
serve them better.
motional information provided by universities did
For example, Restoule and colleagues (2013)
not address concerns felt by Indigenous applicants,
combined Anishinaabe methodology (which hon- although a majority of informants who made it to
ours Indigenous Elders’ wisdom and places a strong college or university said they felt welcome once they
emphasis on representing and serving the people got there. In keeping with the activist orientation of
involved) with institutional ethnography to exam- institutional ethnography, a member of the research
ine how well Indigenous students in Ontario made team took some of the recommendations made by
the transition from high school to post-secondary Indigenous youth to a university assistant registrar
institutions. They first established that although the for discussion. Box 10.8 provides another illustration
vast majority of young Indigenous people hoped for of this type of ethnography.

bel65796_ch10_225-248 241 01/13/22 12:27 PM

242 PART III Qualitative Research

Box 10.8 Institutional ethnography

Campbell (2000) did institutional ethnographic enough—which is such a joke” (Campbell, 2000,
research to investigate how people with disabil- p. 138). The lack of assistance put the woman’s
ities in Victoria, British Columbia, interacted with relationship with her partner in jeopardy because
local health-care providers. She was particularly he had to devote what little time they could spend
interested in how that interaction was influenced together to doing heavy housework for her.
by government regulations and policies. The In another case, a woman confined to a
health-care system in the province was in a state wheelchair suffered because she qualified for
of transition at the time and had been subject to only 12 physiotherapy and massage sessions
funding reductions associated with government a year—the same number as a person without
deficits and debt. As is typical in this kind of re- a disability. The institutional justification given
search, Campbell’s goal was not only to describe was that all citizens had to be treated equally.
how the situation was perceived by the people Campbell makes the point that “generic” rules
experiencing it, but also to give them—in this about health care, which essentially establish
case both people with disabilities and health- “relations of ruling” in the health-care system,
care providers—the means to transform the sit- are created by and for the able-bodied and thus
uation themselves. Both groups were trained as do not serve the needs of people with disabilities
researchers and paid for their involvement. (Campbell, 2000, pp. 141, 143). The texts used to
The study focused on Home Support services implement government health policies (such as
and especially how access to those services af- assessment forms), Campbell argues, serve to
fected the lives of disabled people. One person bolster the government’s legitimacy in the face
mentioned that because of government criteria, of challenges from dissenting parties such as
she was unable to get the support she needed: the disabled, who would benefit from a more in-
“I’m dog tired. And I just get the basics done. clusive system of health-care delivery. Bringing
My partner is away. I’ve been on my own since such matters to light can be a first step in improv-
August and it’s been hard because my home care ing service provision for people with disabilities
was cut off two years ago because I’m not sick or anyone else who is ill served by an institution.

Sampling often face opposition, or at least indifference, to their

research and for that reason are usually willing to
Informants in ethnographic research are usually se- listen to anyone who is prepared to divulge informa-
lected through a version of purposive sampling that tion or views. One way to get access to such people is
involves searching for people who are likely to be a through snowball sampling (see Chapter 7) in which
rich source of information on the group or setting a viable contact is used to identify others who may
under study. Sometimes it is obvious who these people be willing to provide information on the topic of the
are, as in the case of political or religious movements study, who are then used to establish further contacts,
in which the key players and organizational insiders and so on. Snowball sampling was a component of
are well known. In other situations, it may take some Taylor’s (1993) strategy. Her female drug users were
digging to discover who would be a knowledgeable
informant. However, once such people are identified eventually obtained by a mix of “snowballing
it may be difficult or even impossible to get them to techniques” . . . and my almost continuous
open up about the topic of the study. Ethnographers presence in the area. . . . Rather than ask to be

bel65796_ch10_225-248 242 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 243

introduced or given names of others I could is influenced by the emerging theory and—unlike
contact, when I met a woman I would spend quantitative data collection—is an ongoing pro-
as much time with her as she would allow, par- cess, subject to change as the study progresses
ticipating in her daily round, and through this (Breckenridge & Jones, 2009). For Charmaz (2000,
come to meet others in her social circle. My p. 519), theoretical sampling is a “defining property
continued presence in the area also led other of grounded theory” and is concerned with refining
women drug users to approach me when I was ideas rather than boosting sample size.
alone. . . . In addition, the drug worker in the Moreover, theoretical sampling can involve not
area would mention my presence and interest only people but also settings and events. This can
to women with whom he came in contact and be seen in Strauss and Corbin’s (1998, p. 201) defi-
facilitate introductions where possible. (Taylor, nition: “Data gathering driven by concepts derived
1993, p. 16) from the evolving theory and based on the concept
of ‘making comparisons,’ whose purpose is to go to
Ethnographers who take on a role closer to that of
places, people, or events that will maximize oppor-
observer-as-participant rely somewhat more on
tunities to discover variations among concepts and
formal requests for the names of other people who
to identify categories in terms of their properties and
may have relevant information and who can be
dimensions.”
contacted.
Figure 10.3 outlines the main steps in theoretical
Probability sampling is almost never used in eth-
sampling. Data collection (observing, interviewing,
nographic research and is rarely employed even in
and collecting documents) continues until the point
qualitative research based on interviews. In many
of theoretical saturation is reached: “This means,
cases, a probability sample is not feasible because it
until (a) no new or relevant data seem to be emerg-
is difficult, if not impossible, to map “the population”
ing regarding a category, (b) the category is well de-
from which a random sample may be taken. Instead,
veloped in terms of its properties and dimensions
ethnographers have to ensure that they gain access
demonstrating variation, and (c) the relationships
to as wide a range of individuals as possible, so that
among categories are well established and validated”
many different perspectives and ranges of activity
(Strauss & Corbin, 1998, p. 212). In the language of
can be analyzed.
grounded theory, a category operates at a somewhat
higher level of abstraction than a concept in that
Theoretical sampling
Another type of purposive sampling is theoretical
sampling, advocated by Strauss and Corbin (1998). General research question
In their view, because of its reliance on statistical
rather than theoretical criteria, probability sampling
Sample theoretically
is not appropriate to qualitative research. Theoretical
sampling is an alternative strategy, used “in order to
discover categories and their properties and to sug- Collect data
gest the interrelationships into a theory. Statistical
sampling is done to obtain accurate evidence on
Analyze data (concepts, categories)
distributions of people among categories to be used
in descriptions and verifications” (Glaser & Strauss,
1967, p. 62). Theoretical saturation
In theoretical sampling, the researcher simulta-
neously collects and analyzes the data, decides what
data to collect next and where to find them, and Generate hypotheses

develops a theory in the process. Data collection FIGURE 10.3 The process of theoretical sampling

bel65796_ch10_225-248 243 01/13/22 12:27 PM

244 PART III Qualitative Research

it may group together several concepts that have submit a doctoral thesis by a certain date, exhaustion
common features denoted by the category. of funding—necessitate withdrawal from the field:
Saturation means that previous interviews/ Taylor (1993) wrote that one factor contributing to
observations have both formed the basis for the cre- her departure from the field was the lengthy illness
ation of a category and confirmed its importance. of her youngest son.
Thus, there is no longer any need to collect data in Also relevant is the fact that ethnographic research
relation to that category. Instead, the researcher can be highly stressful for many reasons: the nature
should move on and generate hypotheses out of the of the topic, which may place the fieldworker in tense
categories that are forming, and then collect data situations (as in research on crime); the marginal po-
related to those hypotheses. sition of the researcher in the social setting; the con-
stant need to maintain a front; the often prolonged
Not just people absence from one’s normal life. Ethnographers may
As we pointed out in the last section, ethnographic simply feel that they have had enough.
sampling is not just about people. Time and context An important reason for bringing fieldwork to
may also be sampled. Attending to time means that a close is that reasonable answers to the research
the ethnographer must make sure that people or questions have been formulated. The ethnographer
events are observed at different times of the day and may even feel a strong sense of déjà vu toward the
different days of the week, to avoid drawing infer- end of data collection if new data simply reiterate
ences valid only for particular times (mornings, for what has already been discovered. In the language of
example, or weekdays rather than weekends). grounded theory, all the researcher’s categories may
Because behaviour is influenced by contextual become thoroughly saturated, although Glaser and
factors, it is also important to observe in a variety Strauss’s (1967) approach suggests that one should be
of locations. For instance, soccer hooliganism is not certain that there are no new questions to be asked
a full-time occupation. To understand the hooli- and no new comparisons to be made.
gans’ culture and world view, writers like Armstrong In sum, the reasons for bringing ethnographic
(1993) and Giulianotti (1995) had to ensure that they research to a close can range from personal factors
interacted with their participants in a variety of con- to matters of research design. Whatever the reason,
texts other than soccer stadiums, such as the pubs disengagement has to be managed. For one thing,
where they engaged in general socializing. Rosenhan this means that promises must be kept (was a report
(1973) had to do the same in his study of psychiatric promised as a condition of entry?). It also means
hospitals (see Chapter 6). that ethnographers must provide good explanations
for their departure. Participants may know that
The end the researcher is not a permanent presence in their
social setting, but over a long period of time they
Knowing when to stop is not an easy or straightfor- can forget, especially if there has been genuine par-
ward matter in ethnography. Because of its unstruc- ticipation in activities within that setting; therefore,
tured nature and the absence of specific hypotheses farewells may have to be arranged. It’s also essential to
for testing (other than those that may emerge during remember the ethnographer’s ethical commitments,
data collection and analysis), there is a tendency for such as the need to ensure that persons and settings
ethnographic research to lack an obvious end point. are kept anonymous—unless, of course, there has
Only occasionally will there be a natural end to the been an agreement that the social setting can be dis-
research; this was the case with Hier (2002) and closed (as often occurs in studies of religious sects and
Wilson (2002), in that the rave scene was waning cults). Because of how involved ethnographic research
when they studied it. Sometimes the rhythms of the can be, it’s also important to consider the continuing
ethnographer’s occupational career or personal and social and emotional impact fieldwork can have on re-
family life—the end of a sabbatical leave, the need to searchers and participants (see Box 10.9).

bel65796_ch10_225-248 244 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 245

Box 10.9 Leaving the field: There and back again

Ethnographers often wonder whether it’s truly five years researching the role of music in Inuit
possible to leave the field. The question is not culture. He found that “leaving” was not simply a
about one’s physical departure, but about the matter of physical departure. Left lingering were
shared stories, emotions, and experiences that questions about having collected enough data,
remain well after a study has ended. The intimate doing right by his participants, and reintegrating
familiarity ethnographers cultivate often has a re- into his home culture. He concludes, “I continue
sidual effect on both participants and researchers. asking to what degree the field has left me”
In an anthology exploring qualitative research’s (J. van den Scott, 2018, p. 309). Similarly, look-
craft-like features (see Kleinknecht, van den Scott, ing back on over 40 years of researching Hasidic
& Sanders, 2018), some of the chapter authors re- Jewish culture, Shaffir (2018) writes,
flect on the actualities of completing their research
I would make the case that even while dimin-
and leaving the field. For instance, Edwards (2018)
ishing one’s ties with informants, or severing
found that having conducted a qualitative study
one’s connections completely, the researcher
on the relationship between feminism and het-
never leaves the field entirely. To be sure,
erosexuality, ongoing emotional attachments to
there are degrees of leaving. But some parts
her participants were par for the course. Given
of one’s experiences from the field remain—
the type of in-depth, personal research she was
be these recollections or reminiscences—
doing, she found it natural to become invested in
that, in some form or manner, impact upon
her participants’ lives. She writes,
the researcher. (Shaffir, 2018, p. 324)
To say that I have left the field would be
Benbow and Hall (2018), who studied the lives
naive as these women have played, and
of marginalized groups such as street-level sex
continue to play, an important role in my life
workers and mothers experiencing homeless-
as an academic, a feminist researcher, and
ness, reached the point of emotional exhaustion.
a feminist-identified woman who engages
They purposely embody the stories of their par-
in relationships with men. They are also in-
ticipants, carrying their findings and experiences
herently the reason I have a PhD—a simple
forward as part of their ongoing activism. For
fact of reality. So I ask you, how could I not be
those undertaking similarly difficult research,
emotionally attached to the field? How could
they advise creating a self-care plan. Taking time
I truly ever leave? [. . .] If we, as researchers,
for oneself, spending time with friends and family,
are interested in unpacking the complexi-
and reaching out to supportive colleagues can
ties and realities of various populations in
help mitigate some of the enduring stressors that
order to produce the most nuanced under-
accompany emotionally demanding research.
standings of their lived experiences, is it not
Given how involved ethnographic research is,
our duty to ensure that we do not become
it’s hard to envision this form of data collection not
disengaged, detached, and disembodied?
having lasting effects. Writing up findings, revisiting
(Edwards, 2018, p. 330, emphasis in original)
field sites, and reflecting on the stories that were
In the same volume, Jeffrey van den Scott, the shared take field researchers back to their partic-
husband of Lisa-Jo van den Scott (see Box 10.3), ipants’ lives. So, not only is it important to consider
also questions, “Can you leave the field?” He had how to secure and maintain access to a field site, it’s
lived in Arviat, Nunavut, for five years before equally important to prepare for the end of a proj-
departing and then returning to spend another ect and its ongoing social and emotional effects.

bel65796_ch10_225-248 245 01/13/22 12:27 PM

246 PART III Qualitative Research

Key Points
• The ethnographer is typically a participant ob- • Field notes are important memory aids for the
server who also uses interviews and documents. ethnographer.
• The ethnographer may adopt an overt or covert • Visual materials such as photographs and videos
role, but the latter comes with serious ethical have attracted considerable interest among eth-
difficulties. nographers in recent years, not just as adjuncts
• The method of access to a social setting depends to data collection but as objects of interest in
in part on whether it is open (public) or closed their own right.
(private or restricted). • Ethnographers must decide when to end their
• Key informants frequently play an important role data collection and how to manage the ongoing
in ethnography, but care is needed to ensure that social and emotional effects of their research
their impact on the research is not excessive. after they leave the field.

Questions for Review (R) and Creative Application (A)

Access A3 With regard to the ethnography of the tech
r1 “Covert ethnography can make it easier to gain start-up mentioned earlier, outline the benefits
access to certain settings and therefore has and drawbacks to approaching this research
much to recommend it.” Explain how this may as a complete participant, a participant-
be true, then outline the shortcomings of covert as-observer, an observer-as-participant, and a
research. complete observer.
A1 Assume that you have decided to do an ethno-
graphic study of a tech start-up company that Field notes
has just arrived in your community. Explain r4 Why are field notes important for ethnogra-
how you could gain access to the company, phers? What are the main types of field notes
how you would approach potential infor- and why is it useful to distinguish between them?
mants, and how you would interact with mem- A4 Assume you are doing the ethnography of the
bers of the company once ongoing access was start-up mentioned earlier, and have observed
achieved. that no one in the company is over the age of 25.
r2 What is the role of key informants in ethno- Write an analytic memo that explains the signif-
graphic research? Is there anything to be con- icance of that fact.
cerned about when using them?
A2 Assume that you are doing the ethnographic The rise of visual ethnography
study of the start-up mentioned above and have r5 How can visual materials be used in ethnography?
been present in the company for several weeks, A5 The start-up described earlier has just found
but the CEO refuses to speak to you. What could out that it has successfully negotiated a busi-
you do to ensure that your study will produce ness deal with a major corporation. You take
enough information to provide a meaning- a photograph of several members of the
ful analysis of the activities and culture of the start-up at the meeting where the deal was
start-up? announced to all employees. Would the pho-
tograph provide an unproblematic image of
Roles for ethnographers reality? Discuss.
r3 How does it happen that some ethnogra-
phers overidentify with the people they are Sampling
observing? What are some consequences of r6 What is theoretical sampling? How can it be used
­overidentification ? in ethnographic research?

bel65796_ch10_225-248 246 01/13/22 12:27 PM

 10 Ethnography and Participant Observation 247

A6 You’ve decided to do an ethnographic study of A7 Your study on the women’s shelter described
a women’s shelter that will be shut down in six above (A6) is now complete. In what ways
months because of a lack of government fund- might you have become emotionally attached
ing. How could you use snowball sampling to to your participants? What obligations do you
gain access to informants? have to them once the research is complete?
What impact might your study have on future
The end researchers’ attempts to gain access to similar
R7 How do ethnographers decide when it’s time to field sites?
conclude their data collection?

Interactive Classroom Activities

1. Hold a class discussion to decide on a setting or d. Explain what the researchers’ demeanour
group for which ethnographic research would and dress would be as they interact with
be appropriate, for example, biker gang, reli- people at the site.
gious organization, sports team, dating service, e. Explain how they would record the data they
and so on. Divide the class into three groups: get, for example, with field notes, photo-
• Group 1 prepares a mock funding proposal to graphs, video recording, and so on.
conduct a covert study of the site. f. Explain how the study will make a valuable
• Group 2 prepares a mock funding proposal contribution to knowledge.
to conduct a participant-as-observer study g. Upon completion of these tasks, each group
of the site. presents the mock proposal to the rest of the
• Group 3 prepares a mock funding proposal to class, who are to take the role of a funding
conduct an observer-as-participant study of committee that will decide whether the proj-
the site. ect will be funded.
Each group must do the following: 2. As homework, each person in the class is re-
quired to “hang out” at a public, safe location
a. Explain how its methodology would be su- on campus (e.g., residence dining hall, campus
perior to that employed by the other two café, weight room, library space, etc.) for 30 min-
groups—for example, Group 1 would ex- utes as a complete observer. On an appointed
plain how a covert study is superior to a date, each student gives a five-minute mini-
participant-as-observer study and an presentation on his or her experience. The pre-
observer-as-participant study. sentation should:
b. Discuss the ethical implications of the pro-
a. highlight anything of significance that was
posed research.
observed; and
c. Explain how access to the site could be
b. provide suggestions as to how a more in-
gained.
depth ethnographic study of the site could be
conducted.
Relevant Websites
For examples of ethnographic studies, take a look Since 1984, the annual Canadian Qualitative
at the journal Ethnography and the Journal of Analysis Conference has brought ethnographers
Contemporary Ethnography (you may have to log and other qualitative researchers together from
on to your institution’s library server to access them). around the world to discuss their research findings
[Link] and methodological issues in qualitative research.

[Link] [Link]

bel65796_ch10_225-248 247 01/13/22 12:27 PM

248 PART III Qualitative Research

The Ethnography Lab at the University of Toronto PhotoVoice is a charity based in the United
is a collaboration between anthropology faculty Kingdom that assists with ethical participatory
and students. It was established to “promote ethno- photography projects to help promote positive
graphic research methods and practice inside and social change.
outside the university.” The lab runs a speakers series [Link]
and serves as a resource centre for ethnographers. (Websites accessed 9 September 2021.)

[Link]

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch10_225-248 248 01/13/22 12:27 PM

 11 Interviewing in
Qualitative Research
Chapter Overview
This chapter examines the use of interviews in qualitative research. Two types are discussed: unstructured
and semi-structured. This chapter is concerned mainly with interviews with individuals, but the focus group
method is also examined. The highlights of this chapter include:
• the differences between structured and qualitative interviewing;
• the main characteristics of and differences between unstructured and semi-structured interviewing;
• how to devise and use an interview guide for semi-structured interviewing;
• the kinds of questions that can be placed in an interview guide;
• the importance of recording and transcribing qualitative interviews;
• the potential uses and pitfalls of online interviewing;
• how focus groups should be conducted, including matters such as the number of participants, how to
select them, and how direct the questioning should be;
• issues concerning the interactions between participants in focus group discussions;
• some practical difficulties with focus group sessions, such as loss of control over the proceedings and
the potential for unwanted group effects; and
• the advantages and disadvantages of qualitative interviewing relative to ethnography.

▲ kate_sept2004/iStockphoto

bel65796_ch11_249-280 249 01/13/22 12:33 PM

250 PART III Qualitative Research

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Detail the differences between structured and qualitative interviews.

2. Discuss the features of unstructured and semi-structured interviews, including how to prepare an inter-
view guide, the different kinds of questions qualitative researchers use, and considerations for recording
and transcribing interviews.
3. Explain how qualitative researchers use focus groups as a form of data collection, including how
focus groups are conducted, determining an appropriate focus group size, selecting participants,
asking questions and moderating group discussion, the role of group interaction, and the limitations
of focus groups.
4. Describe qualitative researchers’ approach to online interviewing and focus groups, as well as the ad-
vantages and disadvantages of using these two methods online versus face-to-face.
5. Compare the advantages and limitations of using qualitative interviews to those of ethnography.

Practice makes perfect, so why not hone your out how they felt about the experiences, what the
qualitative interviewing skills in casual, every- experiences meant to them, how they affected
day conversations? It’s often said that the art of their lives, and so on. If you’re with a group of
conversation is dying in modern times, but if you people at a party or on a long trip, you can even
approach every discussion you have as a quali- hold an impromptu focus group session, secretly
tative interview in the making, that may change. appointing yourself the moderator. Besides being
Everyday conversation rarely has a narrowly good practice, chance conversations and seem-
restricted purpose, and neither does qualitative ingly inconsequential interactions can spark some
research. A good way to start a discussion is to ask very good research ideas. Just remember that in-
other people where they grew up, or what their stitutional ethics approval will be required if your
most memorable experiences were; then get them casual conversations morph into actual qualita-
to expand on their answers, taking care to draw tive research.

Introduction ethnography make it an attractive option for many

researchers. Although interviewing, transcribing
The interview is probably the most widely used interviews, and analyzing transcripts are all very
method in qualitative research. Since interviews time-consuming, they don’t intrude on researchers’
were not discussed in detail in previous chapters, we personal lives to the degree that spending extended pe-
will give them our full attention here. riods of time with research informants does.
Ethnography usually involves a substantial amount There are two main types of qualitative inter-
of qualitative interviewing, and this undoubtedly con- view: unstructured and semi-structured. A variety
tributes to its wide use. But it is possible to conduct of specialized interview forms fall under those two
qualitative interviews without doing a full-blown eth- headings and will be explored below, including focus
nographic study. In fact, the flexibility of the qualita- groups, oral history interviews, and life history
tive interview and its relatively low cost compared to interviews.

bel65796_ch11_249-280 250 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 251

Differences between direct the discussion as it developed” (p. 170).

By contrast, structured interviews are typically
structured and qualitative inflexible, because of the need to standardize
research interviews the interactions with all interviewees.
• In qualitative interviewing, the researcher
Qualitative interviewing is quite different from in-
wants rich, detailed answers; in structured in-
terviewing in quantitative research:
terviewing the interview is normally designed
to generate specific answers that can be coded
• In quantitative research, the approach is highly
and processed quickly.
structured to maximize reliability and valid-
• In qualitative interviewing, it is not uncommon
ity in measuring key concepts. The researcher
for the interviewee to be interviewed more than
usually has a clearly specified set of research
once. For example, in his research on “clubbers”
questions that the interview is designed to
(people who frequent night clubs), Malbon (1999)
answer. Qualitative interviewing, on the other
conducted an initial interview to put the partici-
hand, is much less structured and tends to be
pant’s mind at ease about the research, help plan
more open-ended. It also allows greater free-
his approach to observation, and begin collect-
dom to modify and add research ideas once the
ing preliminary data. In a follow-up interview, he
investigation has begun.
discussed club visits, the experience of clubbing,
• There is greater interest in the interviewee’s
and occasionally, wider aspects of participants’
perspectives and concerns in qualitative in-
lives. In structured interviewing, unless the re-
terviewing; in quantitative research, the inter-
search is longitudinal in character, the person is
view is driven mainly by the research agenda
usually interviewed only once.
of the person conducting the study.
• In qualitative interviewing, going off on a tan-
gent is often encouraged. The interviewer may Unstructured and
vary both the order and the wording of ques- semi-structured interviewing
tions, and may even ask unplanned questions to
follow up on interviewees’ replies; the additional As we have noted, there are two main types of qual-
replies may provide new insight into what the itative interview:
interviewee sees as relevant and important.
In quantitative research, departing from the • In an unstructured interview the researcher
interview schedule is usually discouraged be- uses at most a memory aid: a small set of self-
cause it diverges from the specific research plan prompts to investigate certain topics. The in-
and may affect reliability and validity. terviewer may ask just a single question; then
• Qualitative interviewing tends to be flexible, the interviewee is allowed to respond freely,
responding to the direction in which inter- with the interviewer pursuing points that seem
viewees take the interview and perhaps adjust- worthy of follow-up. Unstructured interview-
ing the emphases in the research as a result of ing tends to be similar in character to a conver-
the issues that emerge during the interview. sation (see Box 11.1 for an example).
For instance, in describing her approach to • In a semi-structured interview the researcher
conducting qualitative interviews with social has a list of questions or fairly specific topics to
movement actors in Ontario, Maiolino (2019) be covered (an interview guide), but the inter-
writes, “A set of interview questions was used viewee still has a great deal of leeway in deciding
to guide the interview, but I also tried to be how to reply (see Box 11.2 for an illustration).
flexible and create space for interviewees to Questions may not follow the exact order in

bel65796_ch11_249-280 251 01/13/22 12:33 PM

252 PART III Qualitative Research

Box 11.1 Unstructured interviewing

Torelli (2019) adopted a largely unstructured were collaborating jointly in knowledge
interview approach when conducting field re- production. . . . This interchange produced
search on the “dirty work” of Canadian homeless unique conversational avenues which could
shelter caseworkers. His introspective reflections only be achieved naturalistically rather
offer insights into how qualitative interviewers than with strict, albeit aloof, adherence to
must listen attentively, think on their feet, and a structured interview guide. I therefore
carefully follow lines of inquiry presented to them saw the interview process contingently: un-
by participants. Torelli (2019) described his strat- predictable, spontaneous conversational
egy for interviewing caseworkers as follows: instances emerging out of our interaction
that would have not occurred had I stuck
I perceived the interview as a social event, a necessarily to my interview guide and
process whereby, through our interactions, structured the interview in a determinative
me as a researcher and the researched way. (Torelli, 2019, p. 76)

Box 11.2 Semi-structured interviewing

Wright, Alaggia, & Krygsman (2014) used qual- productions, had a number of long-term benefits
itative interviewing to assess the success of an for the participants that went beyond developing
after-school arts program that was offered to their theatre skills. Here are some sample inter-
Canadian youth living in low-income communi- view questions, along with sample probe ques-
ties. The study indicated that the program, which tions that were used to get the respondents to
involved having the youth participate in theatre elaborate on their initial replies:

Interview questions Probe questions

Did you enjoy the art program? What activities did you like?
Did you like the instructors?
Do you have many friends/get along with others Do you sometimes get into arguments with others
your age? your age?
What do you do when you get into arguments?
Have you been getting along with your parents? Do you sometimes get into arguments? About what?

the guide, and some questions not included on interviewee. Box 11.3 illustrates how these fea-
the list may be asked in response to what the tures of qualitative interviewing are used in life
interviewee says. Sometimes lines of thought history research.
identified by earlier interviewees are taken up
and presented to later ones. Nonetheless, all In semi- and unstructured interviews, the process is
the questions on the list are usually asked and designed to bring out how the interviewees themselves
similar wording is used from interviewee to interpret and make sense of issues and events. Bell

bel65796_ch11_249-280 252 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 253

Box 11.3 Life history interviews

One special form of interview associated with conducted narrative interviews with women
qualitative research is the life history interview. tattoo enthusiasts.
It is often combined with personal documents P. Atkinson (2004) observed that the dura-
such as diaries, photographs, and letters. tion of life story interviews varies considerably
Participants are invited to look back in detail from study to study, but that it usually takes two
across their entire life course and to report or three sessions lasting 60 to 90 minutes each.
their experiences and how they understood He provided a catalogue of questions that can
their world. Valuable insights may be gained be asked, and divided them into the following
from a life history regardless of whether other groups (1990, pp. 43–53):
people have had the same experiences as the • birth and family of origin (e.g., “How would
person described in the work. It also has the you describe your parents?”)
advantage of illustrating process: how events • cultural traditions (e.g., “Was your family dif-
unfold and interrelate in people’s lives over ferent from others in town?”)
long periods of time. • social factors (e.g., “What were some of your
An example of the life history interview ap- struggles as a child?”)
proach is provided by Lewis (1961) in his research • education (e.g., “What are your best memories
on the Sánchez family and their experiences in a of school?”)
Mexican slum: • love and work (e.g., “How did you end up in the
work you do or did?”)
I asked hundreds of questions of [the five
• inner and spiritual life (e.g., “What are the
members of the Sánchez family]. . . . While I
stresses of being an adult?”)
used a directive approach to the interviews,
• major life themes (e.g., “What were the crucial
I encouraged free association, and I was a
decisions in your life?”)
good listener. I attempted to cover system-
• vision of the future (e.g., “Is your life fulfilled
atically a wide range of subjects: their ear-
yet?”)
liest memories, their dreams, their hopes,
• closure questions (e.g., “Have you given a fair
fears, joys, and sufferings; their jobs; their
picture of yourself?”)
relationship with friends, relatives, employ-
ers; their sex life; their concepts of justice, A variant of the life history interview is the oral his-
religion, and politics; their knowledge of tory interview, a technique often used by histori-
geography and history; in short, their total ans. It is usually somewhat more specific in tone
view of the world. (Lewis, 1961, p. xxi) than the interviews described above in that the
participants may be asked to reflect on certain
Miller (2000) points out that interest in the life historical events or eras they have lived through.
history method is reflective of the popularity of The emphasis is often on how the individual’s
“narrative interviews” (discussed in Chapter 13) life was affected by those things. The informa-
in which the person studied, instead of respond- tion gathered is sometimes combined with other
ing to an interviewer’s questions, “tells a story.” In sources of data, such as documents.
the past this approach was associated with the The chief problem with the oral history inter-
study of a single life, or perhaps two, but today it view (a problem it shares with life history inter-
is increasingly applied to several lives in a single views in general) is the possibility of bias caused
project. M. Atkinson (2002; 2004), for example, by memory lapses and distortions. Sugiman (2004)

(Continued)

bel65796_ch11_249-280 253 01/13/22 12:33 PM

254 PART III Qualitative Research

pointed out in her work on Japanese-Canadian hand, oral history testimonies can give a voice to
women interned during the Second World War groups that are typically marginalized in histori-
that remembering is a social and even political cal research (this is true of life history interviews in
act, involving recalling, forgetting, transform- general), either because of their lack of power or
ing, and shaping of recollections. On the other because they are regarded as unimportant.

(2007), for example, wanted to understand what west- Preparing an interview guide
ern Canadian separatists thought of Canadian fed- An interview guide is much shorter and less de-
eralism and the federal party system. Tastsoglou and tailed than a structured interview schedule. In fact,
Miedema (2003) sought to understand the meaning of it is often simply a brief list of memory prompts
community from the perspective of immigrant women, for areas to be covered in unstructured interview-
and Smith (2008) examined the meanings of pain for ing and a somewhat more elaborate list of issues
professional wrestlers. Projects like these, unlike quan- to be addressed or questions to be asked in semi-
titative studies, are usually not designed to test hypoth- structured interviewing. What is crucial is that the
eses or theories. In short, qualitative research is not just actual questioning is flexible, allowing interviewers
quantitative research without the numbers. to pursue leads offered by research participants as
There is a growing tendency to refer to semi-struc- they begin to open up and reveal their view of the
tured and unstructured interviews collectively as in- social world. Ricciardelli’s (2019) approach to inter-
depth interviews or qualitative interviews, although viewing Canadian correctional officers about the re-
the semi- and unstructured types can produce very alities of prison work is illustrative of the flexibility
different results. The choice of one type rather than qualitative researchers attempt to build into their
the other is affected by a variety of factors: interviews and interview guides:
A semi-structured interview guide—a list of
• Researchers who feel that using even the most
open-ended questions I constructed—was
rudimentary interview guide hinders genu-
available during interviews, but was often only
ine access to the world views of members of a
used as a sort of checklist to ensure that con-
social setting are likely to favour an unstruc-
versational paths included select topics. . . . The
tured approach. But even here the interview is
conversation tended to be determined by the in-
rarely completely unstructured: the researcher
terviewee because my objective was to hear what
usually has at least a general topic that is to be
was on each participant’s mind, what stood out
discussed.
in his or her experiences, and how he or she in-
• If the researcher is beginning the investigation
terpreted the correctional officer role. Although
with a fairly clear rather than a general focus, it is
I rarely directed conversation, I did at points ask
likely that the interviews will be semi-structured,
interviewees for clarification or for more detail.
so that they can address more specific issues.
(Ricciardelli, 2019, pp. 159–160)
• If more than one person is to carry out the
fieldwork, semi-structured interviewing may In preparing for qualitative interviews, Lofland and
be preferred, to ensure that interviewing styles Lofland (1995, p. 78) suggested that researchers ask
are comparable. themselves “Just what about this thing is puzzling
• Multiple-case study research generally me?” This question can be applied to each of the re-
needs some structure to ensure cross-case search questions generated; it can also serve as a mech-
comparability. anism for generating research questions. Lofland and

bel65796_ch11_249-280 254 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 255

Lofland suggested that puzzlement may be stimulated observation research) nearly always record
in various ways: by recording random thoughts in dif- and then transcribe their interviews. This is
ferent contexts (writing them down as quickly as pos- important for the detailed analysis required in
sible); through discussions with colleagues, friends, qualitative research, and to ensure that the in-
and relatives; and, of course, by reading the existing terviewees’ answers are captured in their own
literature on the topic. The formulation of the research terms. Simply taking notes makes it too easy to
question(s) should not be so specific as to close off the lose specific words and phrases.
alternative avenues of inquiry that may arise during • Make sure as far as possible that the interview
fieldwork. Premature closure of the research focus takes place in a setting that is quiet (so that the
would be inconsistent with qualitative research’s quality of the recording will be as good as pos-
prime purpose: to explore the world view of the people sible) and private (so that the interviewee won’t
being studied, rather than to test the researcher’s own worry about being overheard).
ideas about that world. • Prepare for the interview by cultivating the
The interview guide should: traits of a quality interviewer suggested by
Kvale (1996) (see Box 11.4).
• establish a certain amount of order, so that
questions flow reasonably well, but still allow After the interview, make notes about:
for changes in the order of the questions and
the impromptu asking of different questions; • how the interview went (obstacles, ethical
• include questions or topics that address the re- issues, surprises, the interviewee’s demeanour,
search questions (without being too specific); what worked well and what can be improved
• use language that is comprehensible and fa- upon);
miliar to those being studied; • how you felt during the interview and how this
• not ask leading questions—that is, questions that may have impacted the process (relaxed, ner-
imply a “correct” or socially acceptable answer vous, in control, sad, cheerful);
(for example, “Have you ever done something • where the interview took place and what the
really stupid, like taking crack?”); and setting was like (noisy/quiet, many/few other
• include prompts to remind the researcher to people in the vicinity); and
record basic information about the participant • tentative insights gleaned about your research
(name, age, gender, etc.) as well as more specific question and potential new lines of inquiry.
information that is relevant to the research
questions (position in company, number of The steps required to formulate questions for an in-
years employed, number of years involved in a terview guide in qualitative research are presented
group, etc.); such information is useful for put- in Figure 11.1.
ting people’s comments into context.
Kinds of questions
There are also some practical details to attend to The kinds of questions asked in qualitative inter-
before the interview begins: views are highly variable; Kvale (1996) suggested
nine. Most interviews contain virtually all of them,
• Become familiar with the research setting, although those that rely on lists of topics are likely to
which in many cases will be the everyday sur- follow a somewhat looser format. Kvale’s nine types
roundings of the interviewees. This will help to are as follows:
put their comments in context.
• Get a reliable recording device. Qualitative • Introducing questions. “Please tell me about
researchers (with the possible exception of when your interest in X first began”; “Have you
ethnographers and people doing participant ever . . . ?”; “Why did you go to . . .?”

bel65796_ch11_249-280 255 01/13/22 12:33 PM

256 PART III Qualitative Research

Box 11.4 Kvale’s ten traits of an effective interviewer (plus three

others)
Kvale (1996) proposed that a successful inter- • Interpreting: clarifies and extends the mean-
viewer has ten characteristics: ing of interviewees’ statements but without
• Knowledgeable: is thoroughly familiar with imposing meaning on them
the topic of the interview To Kvale’s list of traits, please add the following
• Structuring: tells the interviewee the purpose three:
of the interview; asks if the interviewee has • Balanced: does not talk too much, which
questions can make interviewees passive, but does
• Clear: asks simple, easy, and short questions; not talk too little, which can cause inter-
no jargon viewees to feel that their responses are not
• Gentle: lets people finish; gives them time to satisfactory
think; tolerates pauses • Ethically sensitive: for example, assures the
• Sensitive: listens attentively both to what is interviewee that all answers will be treated
said and to how it is said; is empathetic confidentially
• Open: responds to what the interviewee con- • Non-judgmental: does not communicate (even
siders important; is flexible subtly) a moral judgment about what an inter-
• Steering: knows what needs to be found out viewee has said or done; negative judgments
• Critical: is prepared to challenge what is said may cause people to withhold information
when, for example, there is an inconsistency in that could be useful to the study, while positive
the interviewee’s replies judgments may encourage them to focus on
• Remembering: relates what is said to what pleasing the interviewer rather than providing
has previously been said authentic commentary

• Follow-up questions. These get the inter- off it?”; “In what ways do you find X disturb-
viewee to elaborate on an answer; for ex- ing?” Box 11.2 provides more examples.
ample: “What do you mean by that?”; even, • Specifying questions. “What did you do then?”;
“Yeesss?” Kvale suggested repeating signifi- “How did (name) react to what you said?”;
cant words in an answer to stimulate further “What effect did (event) have on you?”
explanation, an approach used by Davies • Direct questions. “Do you find it easy to keep
(2000) in a study of female offenders. One smiling when serving customers?”; “Are you
of Davies’s interviewees mentioned the use happy with the way you and your husband
of foil to block detection of security tags decide how money should be spent?” Such
on clothing, so she asked, “You mentioned questions are perhaps best left until later in the
security tags and foil?” which launched the interview, in order not to influence the direc-
interviewee into a detailed description of tion of the interview too much.
shoplifting techniques. • Indirect questions. Before asking an indirect
• Probing questions. These follow up what has question—for example, “What do most people
been said through direct questioning, for ex- around here think of the ways that manage-
ample: “Can you say some more about that?”; ment treats its staff?”—be sure to ask the inter-
“You said earlier that you prefer not to do X. viewee for his or own view (“How do you feel
Can you say what kinds of things have put you about . . .?”).

bel65796_ch11_249-280 256 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 257

• Structuring questions. “Now I would like but also to what the interviewee is doing. This is
to move on to a different topic. Could you important because body language can indicate that
describe . . .?” the interviewee is becoming uneasy with a line of
• Silence. A pause will give the interviewee an questioning. An ethically sensitive interviewer does
opportunity to reflect and amplify an answer; not place undue pressure on an interviewee and is
just don’t pause for so long that the interviewee prepared to cut short any line of questioning that is
feels embarrassed. clearly a source of anxiety.
• Interpreting questions. “Do you mean that your It’s likely that the kinds of questions asked will
leadership role had to change from one of en- vary in the different stages of a qualitative interview.
couraging others to a more directive one?”; “Is Charmaz (2002) distinguished three types of ques-
it fair to say that you don’t mind being friendly tion in this connection. Note that questions about
toward customers most of the time, but you the past and factual matters come first, then ques-
find it more difficult when they are unpleasant tions about feelings, and finally questions of process
or demanding?” and summing up.

As this list suggests, one of the main jobs of the in- • Initial open-ended questions. “What events led
terviewer is listening: being attentive to what the to . . .?”; “What was your life like prior to . . .?”;
interviewee is saying or not saying. The interviewer “Is this organization typical of others you have
must be active without being intrusive—a difficult worked in?”
balance. This means that the interviewer can’t just • Intermediate questions. “How did you feel
sit back and relax, even if the interview is being about . . . when you first learned about it?”;
recorded. In fact, the interviewer must be attuned “What immediate impact did . . . have on your
and responsive not just to what is said and not said, life?”; “What do you like most/least about
working here?”
General More Interview Formulate • Ending questions. “How have your views about
research specific topics interview . . . changed?”; “What advice would you give to
area research questions
questions someone who is undergoing a similar experi-
ence . . .?”; “If you had to do it over, would you
Review/ choose to work for this organization?”
revise
interview
questions Most questions are likely to be of the intermedi-
ate kind, and some categories are likely to overlap.
Nonetheless, these are useful distinctions to bear in
Pilot guide mind.
Remember as well that interviews include differ-
ent kinds of topics, such as the following:
Identify novel
issues
• Values: of the interviewee, of the group, of the
organization
Revise • Beliefs: of the interviewee, of individual others,
interview
questions of the group
• Behaviour: of the interviewee, of others
• Formal and informal roles: of the interviewee,
Finalize guide of others
FIGURE 11.1 Formulating questions for an inter- • Relationships: of the interviewee, of others
view guide for social research • Places and locales

bel65796_ch11_249-280 257 01/13/22 12:33 PM

258 PART III Qualitative Research

• Emotions: particularly of the interviewee, but are expected to be expansive in their answers. This
also of others sequence occurred about halfway through the in-
• Encounters terview, so by then the interviewees were primed to
• Stories recognize that more details were expected. There is
a tinge of embarrassment that the answer is so brief
Try to vary the types of questions you ask (see Kvale’s and not illuminating. The husband’s answer is more
nine types) and, where appropriate, the topic areas expansive but not particularly enlightening.
(see the list above). Vague or overly general questions Then the first of two important prompts by the
are usually best avoided; as Mason (2002) pointed interviewer follows. The husband’s response after the
out, they only force interviewees to puzzle over them initial prompt is more interesting in that he begins to
or to ask for clarification. wonder whether Black people are under-represented at
Vignette questions can be used to ground inter- attractions like the American Adventure. The second
viewees’ ideas and accounts of behaviour in particu- prompt yields further useful reflection, this time that
lar situations (Barter & Renold, 1999). By presenting developing countries are under-represented in World
interviewees with concrete and realistic scenarios, the Showcase. The couple are clearly aware that it was the
researcher can elicit a sense of how certain contexts prompting that led them to these reflections: “I hadn’t
mould behaviour. Hughes (1998) employed the tech- really given this any consideration before you started
nique in a study of perceptions of HIV risk among asking about it.” This is the whole point of prompting:
intravenous drug users. Context is important for this to get the interviewee to think more about the topic
topic because drug users’ willingness to engage in risky and to provide an opportunity for a more detailed re-
behaviour is influenced by situational factors. Scenarios sponse. It isn’t a leading question, since the interview-
depicting risk behaviour were presented and respon- ees were not asked whether they thought the Disney
dents were asked about the kinds of behaviour they felt company failed to recognize the significance of Black
that drug users should engage in (such as using pro- history or ignored developing countries. There is no
phylactics when having sex) and then about how they doubt that the prompts elicited the more informative
felt the hypothetical users would behave (when, for ex- replies, which is precisely their role.
ample, there was an opportunity for unprotected sex).
Hughes argued that a scenario approach is particularly Recording and transcription
valuable with sensitive topics and for eliciting a range of The point has been made several times that in qualita-
responses to different contexts. tive research the interview is usually audio-recorded
and then transcribed (not just listened to) whenever
Using an interview guide: An possible. Qualitative researchers are frequently inter-
example ested not only in what people say but also in the way
Box 11.5 is taken from a study of visitors to Disney they say it. If this aspect is to be fully woven into an
theme parks (Bryman, 1999). The interviewees were analysis, the complete series of exchanges in an in-
a man in his sixties and his wife who was two years terview must be available. Also, because interview-
younger. They had visited Walt Disney World, and ers are supposed to be highly alert to what is being
were very enthusiastic about the visit. said—following up on interesting points, prompting
The sequence begins with the interviewer asking and probing where necessary, drawing attention to
what is considered a “direct question,” in terms of any inconsistencies in the interviewee’s answers—it’s
the nine question types suggested by Kvale (1996). best if they don’t have to concentrate on writing down
The replies are very bland and do little more than what is said as well. In addition, recording interviews
reflect the interviewees’ positive feelings about their allows scrutiny by other researchers, who can evalu-
visit to Disney World. The wife acknowledges this ate the original analysis or even conduct a secondary
when she says, “. . . but I need to say more than that, analysis, which can help to counter accusations that
don’t I?” Interviewees frequently know that they the findings were influenced by the researcher’s values

bel65796_ch11_249-280 258 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 259

Box 11.5 Part of the transcript of a semi-structured interview

Interviewer: OK. What were your views or feel- American Adventure. Now whether we were there
ings about the presentation of different cultures, on an unusual day in that respect I don’t know, but
as shown in, for example, Jungle Cruise or It’s a we saw plenty of Black Americans in the Magic
Small World at the Magic Kingdom or in World Kingdom and other places, but very few if any
Showcase at Epcot? in that World Showcase. And there was certainly
little mention of Black history in the American
Wife: Well, I thought the different countries at
Adventure presentation, so maybe they felt alien-
Epcot were wonderful, but I need to say more
ated by that, I don’t know, but they were notice-
than that, don’t I?
able by their absence.
Husband: They were very good and some were
Interviewer: So did you think there were any spe-
better than others, but that was down to the host
cial emphases?
countries themselves really, as I suppose each of
the countries represented would have been re- Husband: Well, thinking about it now, because I
sponsible for their own part, so that’s nothing to hadn’t really given this any consideration before
do with Disney, I wouldn’t have thought. I mean you started asking about it, but thinking about
some of the landmarks were hard to recognize for it now, it was only really representative of the
what they were supposed to be, but some were developed world, you know, Britain, America,
very well done. Britain was ok, but there was only Japan, world leaders many of them in technol-
a pub and a Welsh shop there really, whereas ogy, and there was nothing of the Third World
some of the other pavilions, as I think they were there. Maybe that’s their own fault, maybe they
called, were good ambassadors for the countries were asked to participate and didn’t, but now
they represented. China, for example, had an ex- that I think about it, that does come to me. What
cellent 360-degree film showing parts of China do you think, love?
and I found that very interesting.
Wife: Well, like you, I hadn’t thought of it like that
Interviewer: Did you think there was anything before, but I agree with you.
lacking about the content?
Source: Bryman, A. (1999). Global Disney. In P. Taylor & D. Slater (Eds.),
Husband: Well I did notice that there weren’t many The American century. Oxford: Blackwell.
Black people at World Showcase, particularly the

or biases. Recording also makes it possible for the data will be preserved in that way. Most people will allow
to be reused in ways other than those intended by the their interviews to be recorded, though it’s not un-
original researcher; for example, to be analyzed using common for a small number to refuse. When that
new theoretical ideas or analytic strategies. happens (or when the recording device malfunc-
As with just about everything in social research, tions), the interview can still proceed with the inter-
there is a cost (apart from the financial cost of buying viewer taking notes. Among those who do agree to
recording equipment) in that the use of a recorder be recorded, some will still be fearful and as a result
can upset respondents, who may become self-con- their interviews may not be as informative as they
scious or alarmed at the prospect that their words otherwise would be.

bel65796_ch11_249-280 259 01/13/22 12:33 PM

260 PART III Qualitative Research

Practical Tip | Transcribing interviews

A student doing research for a thesis may not software, which transcribes spoken words di-
have the resources to pay for professional tran- rectly into a text file, is also available, although
scription, and unless he or she is an accurate it is not yet widely used and at present has cer-
touch typist, transcription can take a lot of time. tain drawbacks, especially for focus groups.
The usual estimate is five to six hours per hour The important thing is to allow sufficient time
of speech, although using transcription soft- for transcription and to be realistic about how
ware along with a foot pedal for stops and starts many interviews can be transcribed in the time
makes the task much easier. Speech-to-text available.

Another consideration, as we have seen, is the time fatigue, or carelessness. For example, one transcript
required to transcribe a recorded interview. There is contained the following passage:
some debate over who should do the transcribing: the
I think unless we want to become like other
interviewer (more familiar with what was actually
countries, where people have, you know, dem-
said) or someone else (less time-consuming) (Rafaeli,
ocratic freedoms . . .
Dutton, Harquail, & Mackie-Lewis, 1997, p. 14).
Furthermore, the vast amounts of text produced must But the actual words on the audiotape were:
then be read. Beardsworth and Keil (1992) reported
I think unless we want to become like other
that their 73 interviews on vegetarianism generated
countries, where people have no democratic
“several hundred thousand words of transcript mate-
freedoms . . . (1995, p. 294)
rial” (p. 262). Clearly, even though transcription has
the advantage of keeping the interviewee’s (and inter- Clearly, steps need to be taken to check on the qual-
viewer’s) words intact, it does so by piling up the text ity of transcription.
to be analyzed. It’s no wonder that writers like Lofland In qualitative research there is often wide variation
and Lofland (1995) advise researchers not to leave the in the time that different interviews take. For example,
analysis of qualitative data until all the interviews in Wilson’s (2000) rave study, the interviews lasted be-
have been completed and transcribed. Insights may be tween 45 minutes and four hours. Shorter interviews
gained by examining the content of early interviews, are not necessarily inferior, with the exception of
and it may be useful to pursue those points in greater those in which the interviewee was non-cooperative
detail in later interviews. Thus, there are good grounds or anxious about being recorded. Indeed, when a
for making analysis an ongoing activity. Also, to wait long interview contains very little of significance, it
until all interviews have been completed may make may not be worth the time and cost of transcription.
the task of transcription seem overwhelming. Thankfully, such occasions are relatively rare. People
Transcription may seem a relatively unproblematic who have agreed to be interviewed are usually coop-
matter of converting the spoken word into written erative, and most loosen up once they get over their
form. However, given the importance of transcripts in initial anxiety about the microphone. As a result, even
qualitative research, the process should not be taken short interviews are often quite revealing.
lightly. Transcribers need to be trained in much the
same way that interviewers are. Moreover, even ex- Flexibility in the interview
perienced transcribers can make mistakes. Poland Flexibility in interviewing means more than being
(1995) provided some fascinating examples of the responsive to what interviewees say and following
errors that can result from factors such as mishearing, up on the interesting points they make. A flexible

bel65796_ch11_249-280 260 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 261

Practical Tip | Transcribing sections of an interview

It’s not always necessary to transcribe every or twice, and then transcribe only the portions
word of an interview. Quite often you will find that seem useful or relevant. The same applies
that large portions of a particular interview to focus group research, which is often more
(even all of it) won’t be useful, perhaps be- difficult and time-consuming to transcribe
cause the interviewee was reticent or because because of the number of speakers involved.
the comments made were not as relevant The downside is that you may miss things, or
to the research topic as you had expected. certain comments may emerge as significant
There seems to be little point in transcribing only later on. If that happens, you will have
such material. Before you begin transcribing, to go back to the recording to get the needed
therefore, listen to the interviews closely once material.

interviewer is able to vary the order of questions and Another common problem for interviewers is know-
clear up inconsistencies in answers. Flexibility is also ing when to switch off their recording equipment.
important when audio recording equipment breaks Often an interviewee will resume talking about the
down, or when an interviewee refuses permission topic of interest as soon as the machine is turned off.
for recording to take place. Sometimes researchers It’s usually not feasible to switch the machine back
find that their questions do not resonate with partic- on again, so try to keep it going as long as possible.
ipants. Flexibility here can mean revising questions If that fails, take some notes either while the person
on the spot and making adjustments to the interview is talking or as soon as possible after the interview.
guide going forward. This is what Jones did in her Such “unsolicited accounts” can often be the source
2018 study, with Hannem, of sexual scripts used by of revealing information. This is certainly what
escort clients: Parker (2000) found in his study of organizations:
many interviewees offered “unrecorded comments
The initial interview questions asked the par- prefixed with a silent or explicit ‘well, if you want to
ticipants to discuss any concerns they had know what I really think.’ . . . Needless to say, a visit
about criminalization or being arrested for to the toilet to write up as much as I could remember
soliciting to purchase sex, experiences of stig- followed almost immediately” (p. 236).
matization connected to being a client, how
much of their identity and self was invested Being reflexive about how qualitative data
in being a client or “hobbyist,” and their im- are co-constructed
pressions of clients in the media. However, Finally, when transcribing and later interpreting
the first three clients arrived at their inter- qualitative interviews, researchers should reflect on
views prepared to tell the story of how and how their own interactions with participants have
why they hire sex workers, and gave succinct affected what was said, how it was said, and what was
and distracted answers to our intended line of left unsaid. Qualitative interviewing is more than
inquiry. The first author decided to adjust the simply digging out pre-existing verbal gems from in-
interview schedule to better suit what the cli- terviewees. The interviewee and the interviewer con-
ents wanted to talk about: what they felt was struct meaning together through their verbal and
misunderstood by the public and researchers non-verbal symbolic interactions. Interpretations
alike, which was, chiefly, the nuanced experi- emerge out of those interactions; they are not simply
ences that they had hiring sex workers. (Jones “out there” in the minds of the study participants,
& Hannem, 2018, p. 494) waiting to be discovered.

bel65796_ch11_249-280 261 01/13/22 12:33 PM

262 PART III Qualitative Research

Ajodhia-Andrews (2016), in her research on how holding a particular view, but a focus group
six Canadian children with disabilities viewed their allows participants to probe one another’s
experiences at school, makes this co-construction of reasons. This can be more informative and re-
understanding explicit and even weaves it into her vealing than the question-followed-by-answer
methodology: method of ordinary interviews. For exam-
ple, an individual who has already answered
After coding and searching for emerging
a certain question may decide, after hearing
themes and patterns, participants and I col-
others’ answers, to qualify or modify that re-
laboratively told, re-told, wrote, and re-wrote
sponse in some way. Some people may want
their narratives, negotiating which pieces of
to voice agreement with a view they might not
stories to include, what they thought of sur-
otherwise have thought of. These possibilities
facing themes and interpretations. . .. Through
permit focus groups to elicit a wide variety of
this dialogue and debate participants and I
perspectives on an issue.
co-constructed narratives together, whereby
• In conventional one-to-one interviewing,
I also became part of the conversations, re-
interviewees are rarely challenged; they may
sponding to their stories and experiences, thus
say things that are inconsistent with earlier
shaping and re-storying their narratives. These
replies or that patently cannot be true, but
narratives become unique to the interaction
interviewers are often reluctant to point out
between myself and participants; if partici-
such deficiencies. In a focus group, partici-
pants shared their stories and understandings
pants may argue and challenge one another’s
with any other person/researcher, the narra-
views. Arguing often produces more realis-
tives would never be conveyed and told in the
tic accounts of what people think, because
same way. . . . (Ajodhia-Andrews, 2016, p. 262)
it forces them to defend and possibly revise
their views.
Focus groups: An • A focus group offers an opportunity to study
introduction how individuals collectively make sense of
a phenomenon and construct meanings of
Most people think of an interview as an exchange it. It is a central tenet of theoretical posi-
between one interviewer and one interviewee. The tions such as symbolic interactionism that
focus group technique, however, involves speaking meanings and understandings are not de-
with more than one (usually at least four) interview- rived by individuals in isolation: rather, they
ees at the same time. Essentially it is a group inter- develop out of interactions and discussions
view in which the interviewees can speak to and with others. In this sense, focus groups re-
interact with one another. flect the processes through which meaning
is constructed in everyday life, which allows
• Most focus group researchers work within them to be more naturalistic than individ-
a qualitative research tradition. They try to ual interviews (Wilkinson, 1998). On the
provide a fairly unstructured setting in which other hand, it is not clear whether these new
the person who runs the focus group, usually meanings persist beyond the focus group
called the moderator or facilitator, guides the session.
session but does not intrude. A focus group
offers several advantages. One of the most im- Box 11.6 provides an example of focus group research
portant is that it allows the researcher to de- conducted by Ricciardelli and Adorjan (2019), who
velop an understanding of why people feel the were interested in understanding Canadian teens’
way they do. In individual interviews, partic- views on sexting (sending sexually suggestive or ex-
ipants are often asked about their reasons for plicit text messages and images to others).

bel65796_ch11_249-280 262 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 263

Box 11.6 Focus groups in action: Canadian teenagers and

sexting
Focus groups were used by Ricciardelli and did not find any significant differences based on
Adorjan (2019) to examine Canadian teenag- location—“gender and age were the primary driv-
ers’ perspectives on sexting. The focus groups ers of difference” (p. 567). The researchers mostly
were concerned with how “the attitudes and had participants of the same age and gender in
perceptions of youth related to cyber-risks in- each group, and “used a sampling stratification
volved with digital sexual expression” (p. 564). strategy designed to help ensure participants in-
Details of the groups in this study are given in teracted with others that they would not perceive
Table 11.1. as threatening and with whom their experience
By using focus groups, Ricciardelli and Adorjan may also resonate” (p. 567). Each session lasted
sought to better understand how teens interpret an average of one hour and was audio-recorded
their experiences with sexting. Focus groups of- to maintain accuracy and so that the moderators
fered participants a platform to engage in group could focus on participants’ discussion. A key find-
discussion with their peers, “illuminating social ing of the study was the gender double-standard
contexts and personal interpretations which are that exists with online sexual expression: males
largely left out of survey-based methods” (p. disproportionately send females non-consensual
567). Male and female participants were sampled sexts, while females are expected to regulate their
from urban Western and rural Atlantic schools. sexual expression to a much greater extent and
This offered the researchers an opportunity to are more likely to suffer serious stigmatization for
compare the different regional contexts, but they not doing so.

Conducting focus groups is similar to theoretical saturation in grounded

theory, discussed in Chapters 10 and 13.
A number of practical aspects of conducting a focus One factor that can affect the number of groups
group are considered below. required is whether the researcher thinks that the
range of views is likely to be affected by socio-demo-
How many groups? graphic factors such as age, gender, or social class.
How many focus groups are needed? Table 11.1 Many focus group researchers like to ensure that
provides some data on this question, and on some different demographic groups are included, which
other aspects of focus groups (cf. Deacon, Pickering, usually requires a large number of groups. In con-
Golding, & Murdock, 1999). As it suggests, there is nection with Kitzinger’s (1994) research on audi-
a good deal of variation in the number of groups re- ence responses to media messages about AIDS (see
quired for a particular study, but it generally ranges Table 11.1), she wrote that a large number of groups
from 10 to 15. is preferred in order to capture as many different
A single group is unlikely to be sufficient, since perspectives as possible. On the other hand, con-
the responses may not be typical of other groups. ducting more groups increases the complexity of the
Nonetheless, there are good reasons for limiting analysis. For example, Schlesinger, Dobash, Dobash,
the number of groups (besides saving time and re- & Weaver (1992, p. 29; see Table 11.1 and Box 11.7)
sources). Once the moderator is able to anticipate reported that their 14 tape-recorded hour-long ses-
fairly accurately what the next group is going to sions produced more than 1400 pages of transcrip-
say, enough groups have participated. This notion tion for analysis.

bel65796_ch11_249-280 263 01/13/22 12:33 PM

264 PART III Qualitative Research

Size of groups and selecting others were made up of people likely to have a special
interest in AIDS, such as male sex workers and intra-
participants venous drug users. Increasingly, focus group practi-
How large should a single focus group be? Morgan tioners try to discern patterns of variation by putting
(1998) suggested six to ten members. To control for together groups with particular attributes.
the problem of “no-shows,” researchers often find A further issue in designing a focus group is
themselves over-recruiting (for example, Wilkinson, whether to select people who are unknown to each
1999, p. 188). Morgan recommended small groups other, or to use natural groups (for example, friends,
when the participants are likely to have a lot to say, co-workers, or students in the same class). Some
as often occurs when they are emotionally involved researchers prefer to exclude people who know
with the topic, when the topic is controversial or each other, fearing that known status differences
complex, or when exploring personal interpreta- or pre-existing patterns of interaction would con-
tions. A larger group is appropriate when the re- taminate the session. Others prefer to select natural
searcher wants “to hear numerous brief suggestions” groups whenever possible. Holbrook and Jackson
(Morgan, 1998, p. 75). (1996) reported that, for their research on shopping
For most topics there is no need to select a partic- centres, they initially tried to secure participants
ular kind of participant; however, the topic should be who were unknown to each other, but this strategy
relevant to those taking part. A wide range of people attracted no takers. They then sought out partic-
is often required, but participants may be put into ipants in various social clubs. They defended this
separate groups on the basis of age, gender, educa- new approach on the grounds that, because their
tion, ethnicity, whether they have (or have not) had research questions concerned shopping in relation
a certain experience, and so on. The aim is to look to the construction of identity and how it relates to
for any variation in how the different groups discuss people’s sense of place, recruiting people who knew
the matter at hand. For example, to examine wom- each other would be highly appropriate. Recruiting
en’s responses to scenes of violence against women, people from natural groups is not always feasible,
Schlesinger et al. (1992) showed 14 groups (see Table however, because of the difficulty of getting every-
11.1 and Box 11.7) four levels of mass-media vio- one in the group to participate. Morgan (1998) re-
lence: incidental violence, moderate violence, mar- vealed another downside: because people in natural
ital violence, and an extremely vivid sexual assault groups know one another, they may share certain
scene. The authors concluded: assumptions that they feel no need to explain or jus-
tify. He suggested that in cases where it is important
Having a particular experience or a particu- that such assumptions are made explicit, a group of
lar background does significantly affect the strangers is a better choice.
interpretation of a given text. The four pro-
grams screened are obviously open to various
readings. However, on the evidence, how they
Asking questions and level of
are read is fundamentally affected by various moderator involvement
socio-cultural factors and by lived experience.
There are different questioning strategies and ap-
(Schlesinger et al., 1992, p. 168; emphasis in
proaches to moderating focus group sessions. Most
original)
lie somewhere between a rather open-ended style and
A slight variation on this approach is Kitzinger’s a more structured one. In Ricciardelli and Adorjan’s
(1994) study of reactions to media representations of (2019) research on sexting among teenagers (see Box
AIDS (see Table 11.1). Her groups were made up of 11.6 and Table 11.1), they began each session by asking
people in a variety of situations. Some were what she a general, open-ended question about participants’
called “general population groups” (for example, a experiences with technology. Follow-up questions
team of civil engineers working on the same site) but pertained to concerns that were raised by participants,

bel65796_ch11_249-280 264 01/13/22 12:33 PM

 TABLE 11.1 Composition of groups in focus groups research

FROHLICH,
RICCIARDELLI EADY, POTVIN,
AND ADORJAN DOBINSON, & CHABOT, & KITZINGER LETT, HIER, & MIRAFTAB SCHLESINGER

bel65796_ch11_249-280
AUTHORS (2019) ROSS (2011) CORIN (2002) (1993, 1994) WALBY (2010) (2000) ET AL. (1992)

265
Area of research Teenagers’ Bisexual people’s A contextual Audience Extent of public Housing Responses
perspectives experience with understanding of responses to consultation on experiences of women
on sexting mental health pre-adolescent media messages the introduction of Kurdish and to watching
services smoking about AIDS of street Somali refugees violence against
behaviour surveillance in Vancouver women
cameras, and
perceptions
of their
effectiveness in
preventing crime
Number of 35 8 Not specified 52 4 Not specified 14
groups
Size range of 2–5 3–9 Probably 12 Not specified but 5–10 10–15 5–9
groups appears to be 3,
9, or 10
Average (mean) 3.3 Not specified n.a. 6.75 7 n.a. 6.6
size of groups
Criteria (if any) Age, gender, Age, sexual 4 groups out No, but groups People identified None, except Experience
for inclusion rural/urban orientation of 32 Quebec made up of in promotional being recent of violence,
location communities specific groups campaigns and immigrant Scottish/English,
(e.g., retirement justificatory ethnicity, class
club members, arguments
male sex relating to street
workers) surveillance
cameras
Natural groups No No No Yes Some No Some

01/13/22 12:33 PM
11 Interviewing in Qualitative Research
265
266 PART III Qualitative Research

Box 11.7 Asking about violence

• Initially, participants were given the opportu-
nity to discuss the film in terms of things such
as its purpose, its realism, and its storyline.
Stígur Már Karlsson/Heimsmyndir/iStockphoto

• The questioning then moved on to reactions

to the characters: the woman who was raped,
the three rapists, the female lawyer, the male
lawyers.
• Participants were then asked about their re-
actions to particular scenes, including the
rape; the female lawyer’s decision to support
the case after initially not supporting it; and
the victory in court.
Semi-structured interviewing offers researchers
and interviewees a degree of flexibility in shap- • Participants were asked for their reactions to
ing the course and content of an interview. What the inclusion of the rape scene.
are some difficulties that researchers might face • Finally, they were asked to give an assessment
when conducting a semi-structured interview that of the film’s value, in particular whether the
they might not face when conducting a structured fact that it was American made a difference
interview? to their reactions. (The study was conducted
in the UK.)
Researchers usually want to inject some struc-
While the research by Schlesinger and col-
ture into their focus group sessions. An exam-
leagues examined a lot of specific topics, initial
ple can be seen in the research conducted
questions were designed to generate relatively
by Schlesinger et al. (1992; see Table 11.1) on
open-ended reactions. Such a general approach
women viewing mass media presentations of
to questioning is fairly common in focus group
varying levels of violence against women. For a
research. It allows the investigator to address
rape scene depicted in a movie, reactions were
the research questions, ensures that there is
gleaned through “guiding questions” under five
some comparability between sessions, and per-
main headings, the first three of which had sev-
mits participants to raise issues that they see as
eral more specific elements:
significant.

Practical Tip | Number of focus groups

Even one focus group session takes a long time to professional researcher would. In your report,
arrange. It also takes a long time to transcribe the make sure to justify the number of groups used
recordings made during the session. Thus, most in your research, explaining why the data are still
students will not be able to conduct as many focus significant even though you had to make do with
group sessions for a project or dissertation as a a relatively small number of groups.

bel65796_ch11_249-280 266 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 267

which often centred around privacy and sexting. additional direction may be required from the mod-
Their middle-of-the-road approach to question struc- erator. Levels of interest and knowledge among the
turing can be seen in the following question asked by participants can make a difference as well. Limited
one of the focus group moderators: “Do you think interest or knowledge on the part of participants
it’s the same though for boys and girls, cuz you said may require a somewhat more structured approach.
boys send the pictures and they don’t care, but girls If in doubt, the best advice is to err on the side of
obviously send them too (Rebecca and Carolyn reply minimal intervention.
affirmatively). But do you think it’s a different context,
like they think something different is going to come Recording and transcription
of it?” (Ricciardelli & Adorjan, 2019, p. 570). Here, the Recording is even more important with focus groups
moderator focuses on the topic to be addressed but is than it is in other forms of qualitative research. Writing
also able to pick up on what the group says. down not only exactly what is said but who says it is too
How involved should the moderator or facilitator difficult. In an individual interview you may be able to
be? As with question structuring, above, the most ask a respondent to “hold on” while you write down a
common approach is middle-of-the-road. There is a response, but this is not feasible in an interview where
tendency to use a fairly small number of very general several people are speaking rapidly, and would almost
questions to guide a focus group session. Obviously, certainly break the flow of the discussion.
if the discussion goes completely off topic it may be Transcribing focus group sessions is also more
necessary to refocus the participants’ attention, but complicated and hence more time-consuming than
even then the moderator must be careful, because it is with other interview forms. Sometimes voices
apparent digressions can often reveal something of are hard to distinguish, making it difficult to deter-
significance. More direction is probably needed if the mine who is speaking. Also, people sometimes talk
participants are not addressing the research ques- over each other, which can make transcription even
tions, or if a particularly meaningful point made by more problematic. Therefore, a very high-quality
one participant is not followed up by the others. recording device, capable of picking up even faint
Both intervention and non-intervention carry voices from many directions, is a necessity. Focus
risks. The style of questioning and moderating de- group transcripts always seem to have more missing
pends on the nature of the research topic; if it is bits than transcripts from other sorts of interview,
embarrassing for some participants, for example, mainly because of audibility problems.

Research in the News

What does it mean to be a working-class, first-generation student in Canada?
When Proudfoot (2019) first interviewed Wolfgang referring to examined what it means to come from
Lehmann, a Professor of Sociology at Western a working-class home and be the first person in
University in London, Ontario, she recalls his re- the family to attend university (a “first-generation
search findings bringing tears to her eyes. She student”). The idea of “working class” (and “social
writes, “I never forgot his research because he was class” more generally) is a concept that social
talking about me, explaining my life in ways that si- scientists and politicians alike have a hard time
multaneously made a deep sort of sense and had pinning down. Lehmann did not specify “working
never occurred to me before.” The research she is class” in his study’s recruitment materials; rather,
(Continued)

bel65796_ch11_249-280 267 01/13/22 12:33 PM

268 PART III Qualitative Research

he “described those who were the first among background except to highlight its advantages,
their parents or grandparents to attend university, such as fostering resilience, a strong work ethic,
and whose parents worked in fields like construc- and a sense of not having things handed to them.
tion, factories or hairdressing.” From this sample One of Lehmann’s participants stated, “A lot of
of students, he conducted in-depth interviews the people here have a good amount of money
when they were in first, second, and fourth year, behind them, or their parents are paying . . . and
and then again five years after they had grad- at times it’s like ‘Oh man, I wish it was that easy.’
uated. When asking students about their social . . . But honestly, I like the struggle of it. I think it
class, they often initially felt that they were middle makes me a better person.” Striving to become
class. After Lehmann described some of the qual- middle class also presents a conundrum: the
ities of what it means to be working class, though, conditions that working-class students credit
most identified as such and were somewhat sur- as advantages are also what they are trying to
prised by the realization. escape. Climbing the ladder of social mobility
Lehmann found that working-class students leads to being caught between two worlds, no
encounter a variety of hurdles, such as money longer fully belonging to the working-class cul-
concerns and having to balance school and work ture of their up-bringing, while also at times feel-
commitments. He also discovered that his par- ing like an outsider in the middle-class world they
ticipants tended to ignore their working-class aspired to.

Group interaction in focus No. 1: But I think maybe what we’re saying here
is that there’s no one cause of heart attacks,
group sessions there’s no one type of person, there’s probably
Kitzinger (1994) observed that reports of focus umpteen different types of heart attacks and
group research frequently fail to take group in- causes coming from maybe smoking, maybe
teraction into account. This is surprising, given obesity, maybe stress, maybe design fault, he-
that such interaction is a distinguishing feature reditary, overwork, change in life style. Any of
of focus groups. Wilkinson (1998) reviewed more these things in themselves could be . . .
than 200 focus group studies, published between No. 2: And when you start putting them in
1946 and 1996, and concluded: “Focus group data combination [unclear].
is most commonly presented as if it were one-to-
No. 3: Yeah, you may be really magnifying
one interview data, with interactions between
each one of these particular things.
group participants rarely reported, let alone ana-
lyzed” (p. 112). No. 2: Yeah, and depending on how, and in
In the context of her research on the coverage of each person that magnification is different.
AIDS in the mass media, Kitzinger (1994) drew at- Some people can take a little stress without
tention to two types of interaction in focus groups: doing any damage, some people can take a
complementary and argumentative. Complementary little smoking, a little drinking, a little obe-
interaction allows collective interpretations and sity, without doing any damage. But you take
understandings to develop, with each participant a little of each of these and put them together
building on the preceding remark, as in the follow- and you’re starting to increase the chances
ing passage taken from Morgan and Spanish’s (1985, of damage. And any one of these that takes a
p. 414) research on heart-attack victims: magnitude leap increases the chances.

bel65796_ch11_249-280 268 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 269

This sequence brings out the emerging consen- raped, would you like to have [to] sit through
sus around the question of who has heart attacks that again?
and why. However, as Kitzinger (1994) suggested,
The debate then moved on to consider the signifi-
argumentative interaction in focus groups can be
cance of the scene for men:
equally revealing. In such cases moderators can
play an important role in identifying differences Speaker 1: Men would sit down and think,
of opinion and exploring with participants the fac- “Well, she asked for it. She was enjoying it and
tors that lie behind them. Disagreement can give look, the men around enjoyed it.” (Schlesinger
participants a chance to revise their opinions or to et al., 1992, pp. 51–52)
think about why they hold them. A passage from
It seems, then, that behind the unease some of the
Schlesinger et al. (1992; see Table 11.1) illustrates.
women expressed about the rape scene was the idea
The group is made up of women with no experience
that some men might enjoy it and identify with the
of violence, and the debate concerns a rape scene
onlookers depicted in the movie. This interpretation
in a film:
came out because of an earlier disagreement within
Speaker 1: I think . . . that they could’ve ex- the group, which allowed the women to give a fuller
plained it. They could easily leave [out] that account of their reactions to the scene.
rape scene. In sum, as Kitzinger (1994) argued, drawing at-
tention to patterns of interaction in focus groups
Speaker 2: But it’s like that other film we
allows a researcher to determine how group par-
watched. You don’t realize the full impact, like,
ticipants view the issues at hand. The posing of
the one we were watching, the first one, until
questions by focus group members as well as the
you’ve got the reconstruction.
agreement and disagreement among them help to
Speaker 3: Yeah, but I think with that sort bring out this knowledge. The resolution of disagree-
of film, it would cause more damage than it ments also forces participants to explain why they
would good, I mean, if someone had been hold the views they do.

Methods in Motion Applications to Canadian Society

Social capital and Indigenous education
Many scholars have expressed concern over the fact “cultural capital,” which maintains that certain kinds
that in Canada, Indigenous people are less likely to of knowledge, skills, and competence give people
graduate from high school than non–Indigenous in- an advantage in negotiating how they are treated in
dividuals, and likewise that Indigenous youth have formal institutions, served as the theoretical back-
lower levels of participation in post-secondary drop to the study.
educational institutions (e.g., Milne, 2017). To look The interviews revealed that the legacy of colo-
into this issue, Milne (2016) conducted semi-struc- nialism, especially the intergenerational, traumatic
tured interviews with 26 educators and 24 parents/ effects of attending residential schools, created a
guardians of students from off-reserve schools in high level of mistrust of educational institutions on
southern Ontario. Twenty of the teachers identi- the part of Indigenous people. This was evident in
fied as Indigenous, as did 20 of the parents/guard- a remark made by one of the Indigenous parents:
ians. Lareau and Weininger’s (2003) concept of “I have the worst fear of teachers, to be honest with

(Continued)

bel65796_ch11_249-280 269 01/13/22 12:33 PM

270 PART III Qualitative Research

you. Even if the teacher’s the same age as me, I’m away from the schools. When asked about what role
nervous around them” (Milne 2016, p. 276). she played in her children’s education, one low-in-
Another key finding was that class differences come parent said that she “doesn’t know how to go
among Indigenous participants were linked to how about that kind of stuff” (Milne 2016, p. 281). Middle-
well they were able to pursue the educational inter- class Indigenous parents (who had a post-second-
ests of their children with school authorities. Lower- ary education and incomes between $30,000 and
class individuals (defined as having at most a high $112,335) were much better able to navigate the
school education and a household income of less educational system on their children’s behalf. A mid-
than $30,000) encountered a great deal of diffi- dle-class mother, after having a confrontation with
culty in getting their children’s educational needs her daughter’s teacher, told him “You’re fired,” and
met. They were hesitant to approach school officials insisted that the principal move the child to a differ-
about their children’s progress and tended to stay ent class immediately. She got her way.

Limitations of focus groups include inaudible comments from participants

sitting too far from the microphone, and these
Focus groups are particularly useful in illustrating also affect transcription.
how meaning is jointly constructed. What, then, are • The data may be difficult to analyze. Developing
their chief limitations? an analysis strategy that incorporates both the
themes of the discussion and the patterns of
• The researcher probably has less control interaction is not easy.
over the proceedings than in individual in- • Focus groups are difficult to arrange. It may
terviews. As we have seen, not all writers be hard to get people to agree to participate,
on focus groups perceive this as a problem; and harder still to get them to show up. Very
indeed, researchers concerned with reflexiv- small payments, such as gift cards, are some-
ity often see it as an advantage. However, the times offered to induce participation, but it is
question of control raises questions about the still common to find that some people will not
extent to which a researcher should allow turn up.
the participants to “take over” the discussion. • Group effects may be a problem. This cate-
There is clearly a delicate balance to be struck gory includes the obvious challenge of dealing
in deciding how involved moderators should with participants who either are too reticent
be and how much a set of prompts or ques- to speak up or hog the stage. Krueger (1998)
tions should be allowed to influence the con- suggested that the people running the focus
versation. What is not clear is the degree to group should make it clear to participants that
which it is appropriate to surrender control, other people’s views are definitely required; for
especially when there is a set of research ques- example, he suggested saying something like
tions to be answered. “That’s one point of view. Does anyone have
• An unwieldy amount of data is sometimes another point of view?” (1998, p. 59). As for
produced. For example, Bloor, Frankland, those who are reluctant to speak, Krueger rec-
Thomas, & Robson (2001) suggested that ommended that they be actively encouraged to
one focus group session can take up to eight say something.
hours to transcribe—somewhat longer than an
equivalent personal interview, because of vari- It would be interesting to know whether agree-
ations in voice pitch and the need to identify ment among focus group participants is more
who says what. Also, group recordings often common than disagreement; if that is the case,

bel65796_ch11_249-280 270 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 271

one reason might be group pressure to conform. has great potential for facilitating access to groups
It’s also possible that participants are more likely to of participants that were often difficult to reach in
express socially acceptable views in a group setting the past. Even if the technology is not used specif-
than in individual interviews. Morgan (2002) cited ically for data collection, carefully selected web-
a study in which group interviews with boys on the sites, online forums, and social media sites can be
topic of relationships with girls were compared with excellent sources for recruitment.
individual interviews with boys on the same sub-
ject. Alone, the boys expressed a degree of sensitiv- Online qualitative interviews
ity that was not evident in the group context, where
Price and Puddephatt (2018) conducted qualita-
they tended to express more macho views. This sug-
tive research on the perspectives and experiences
gests that in the group setting the boys were trying
of Canadian open access journal editors. Because
to impress each other or avoid embarrassment, and
their participants were geographically dispersed,
were influenced by peer pressure. But this does not
they decided to use a combination of Skype, tele-
mean that such data are tainted: in fact, it may be
phone, email, and face-to-face interviews to collect
the gulf between privately and publicly held views
their data. Of the two online communication medi-
that is of interest.
ums, they discovered that Skype worked better than
email for establishing rapport and eliciting in-depth
• Madriz (2000) proposed that there are cir-
answers. Email participants’ responses tended to be
cumstances when individual interviews are
short and glib. Note the following brief sequence
more appropriate. One is when participants
(Price is Taylor):
are likely to disagree profoundly with each
other; another is when participants are not Taylor: What was the hardest day or week you
comfortable in one another’s presence (for ex- had as an open access journal editor?
ample, people in hierarchical relationships).
Email respondent: Whenever I learn a new
Finally, because of the potential for focus
HTML program. (Price & Puddephatt, 2018,
groups to cause discomfort among partici-
p. 194)
pants (for example, when intimate details of
private lives need to be revealed), an individ- Not being in a shared interactional space made it dif-
ual interview or even a questionnaire may be ficult for the interviewer to know how to respond.
better. Skype is a rich, synchronous communication
medium, meaning that it allows for visual and au-
Online interviews and focus ditory communication cues and instant interactive
groups exchange between the interviewer and interviewee.
The immediate back-and-forth flow of information
Conducting qualitative interviews and focus creates greater opportunity for developing mutual
groups online has become increasingly common understanding and working out misunderstand-
as more and more people incorporate Internet ings in real-time. Hanna (2012) also reports that
communication technologies into their every- Internet interviews using webcam technologies such
day lives. Being aware of what topics and types of as Skype can be as effective as in-person qualitative
participants are best suited for online interviews interviews, and that they have advantages similar to
and focus groups is an important consideration. telephone interviews. No travelling is needed with
While some people are comfortable with online Skype interviews, they can be postponed and re-
communication, there are still many who are not. scheduled without difficulty, and the interviewee is
Selecting participants can also be difficult since allowed to stay home and thus feel safe and secure.
everyone taking part must have access to the nec- Moreover, both audio and visual aspects of the inter-
essary hardware and software. Still, the Internet view are easily recorded.

bel65796_ch11_249-280 271 01/13/22 12:33 PM

272 PART III Qualitative Research

of the ground rules for the discussion. There is ev-

idence that participants respond more positively if
the researchers reveal something about themselves
(Curasi, 2001). This can be done in advance of the
session, as was suggested by Price and Puddephatt
(2018) in the case of online interviews, through the
moderator’s opening remarks or by creating links to
a personal website.
In online focus groups, participants respond
more or less immediately to whatever has just been
“said,” whether by the moderator or by other par-
Skype can be a valuable interview tool, allowing for
ticipants. As all are simultaneously online, text-
rich visual and auditory communication without the based contributions can be responded to as soon
need to travel. What might be some of the drawbacks as they appear on the screen (and with some forms
of Skype interviews? of software, the contributions can be seen as they
insta_photos/Shutterstock are being typed). As Mann and Stewart (2000)
observed, because several participants can type
Price and Puddephatt (2018) suggest that while in a response to something at the same time, the
real-time communication technologies generally convention of taking turns in regular conversa-
offer greater advantages for long distance data col- tion is largely sidelined. Even with the benefits of
lection, the fit between the research question and visual and verbal cues in video-based focus groups,
communication medium should also be consid- turn-taking behaviour is somewhat more difficult
ered. For instance, asynchronous (non-real-time) to read and facilitate online than it is when moder-
communication mediums like email can be useful ating a face-to-face focus group.
for exploring topics where more deliberate, careful Online focus groups are unlikely to replace their
reflection is called for. However, if there is no in- face-to-face counterparts, but they are likely to be used
teractional history between the participant and re- for certain kinds of research topics and/or samples. As
searcher, Price and Puddephatt recommend using a regards the latter, people who are dispersed or inacces-
rich, synchronous medium prior to an email inter- sible are especially suitable for online focus group re-
view to help establish rapport and buy-in from the search; slow typists, for instance, are not well-suited if
participant. the focus group is run in a text-based format. As Sweet
(2001) pointed out, the most appropriate topics for
Online focus groups online research are likely to be those involving sensi-
Mann and Stewart (2000) suggested that an online tive issues or Internet use itself (as in Box 11.8).
focus group should have between six and eight Much research on the Internet treats the technol-
people. Larger groups make it difficult for some ogy as a given, but some studies examine how it is
people to participate. This is especially true when interpreted by users. Hine (2000, p. 9), for instance,
solely using text-based communication technol- presented the Internet “as a product of culture . . .
ogies as it curtails input from those with limited shaped by the ways in which it is marketed, taught,
keyboard skills. Also, moderating an online ses- and used.” Box 11.9 summarizes the main advan-
sion can be more difficult with larger numbers, tages and disadvantages of conducting focus groups
and as Adriaenssens and Cadman (1999) sug- and individual qualitative interviews online in a
gested, large groups can present administrative text-based format compared to administering them
problems. in person. The two types of qualitative interviewing
Before starting an online focus group session, are combined because the tally of advantages and
moderators are advised to send out a welcome mes- disadvantages applies more or less equally well to
sage introducing the research and laying out some both of them.

bel65796_ch11_249-280 272 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 273

Box 11.8 An online focus group study

O’Connor and Madge (2001; 2003; see also Kerry: I feel the same. Like the HV [health vis-
Madge & O’Connor, 2002) employed text-based itor] is judging even though she says she isn’t
conferencing software in a virtual focus group
Kerry: Although my HV has been a lifeline as
study that examined the use of an online in-
I suffer from PND [post-natal depression]
formation website for parents. Initially, the re-
searchers set up a Web survey on the use of the Amy: Also, there are some things that are so
site. When respondents sent in their question- little that you don’t want to feel like you’re
naires, they were thanked for their participation wasting anyone’s time. Asking the HV or GP
and asked via email whether they were willing might get in the way of something mroe
to take part in an in-depth interview. Of the 155 important, whereas sending an email, the
respondents who returned the questionnaires, person can answer it when convenient.
16 agreed to be interviewed and were sent the
Amy: My HV is very good, but her voice does
software to install on their computers. The re-
sound patronising. I'm sure she doesn’t
searchers tried to ensure that each group was
mean it, but it does get to me . . .
asked more or less the same questions. For each
session, the researchers introduced themselves Kerry: Being anon means that you don’t get
and asked participants to do likewise. In addi- embarrassed asking about a little point or
tion, they placed descriptions and photographs something personal. (O’Connor & Madge,
of themselves on a website to which participants 2001, p. 10.4)
were directed. An important part of the process
This extract reveals a good flow without interven-
of building rapport was the fact that both of the
tion by the researchers. It contains several mistakes
researchers were mothers. They found that
(“mroe,” “I’msure”), but they are retained to preserve
the relative anonymity afforded by the Internet
the authenticity of the interaction. Another plus is that
gave participants greater confidence in asking
there is no need to transcribe the material because
embarrassing questions, a finding that has im-
it is already in textual form. Also, the fact that partic-
plications for online focus groups in general. This
ipants appear to relish the anonymity of the Internet
can be seen in the following excerpt:
has implications for online focus groups, since par-
Amy: I feel better asking BW [Babyworld] ticipants may find it easier to ask naive questions or
than my health visitor as they’re not going make potentially embarrassing comments online
to see how bad I am at housekeeping!!! than in face-to-face focus groups.

Qualitative ­interviewing They are probably the two most prominent methods
of data collection in qualitative research, so there is
(without immersion in some value in assessing their relative strengths.
a social setting) versus
Advantages of ethnography
ethnography ­compared to qualitative interviewing
The aim of this section is to compare the merits and Seeing through others’ eyes
limitations of qualitative interviewing that does As we noted in Chapters 1 and 9, seeing through
not involve spending extensive periods of time in a others’ eyes is one of the main purposes of qualita-
particular social setting with those of ethnography. tive research. On the face of it, ethnographers would

bel65796_ch11_249-280 273 01/13/22 12:33 PM

274 PART III Qualitative Research

Box 11.9 Advantages and disadvantages of conducting focus

groups and individual qualitative interviews using an
online, text-based format rather than face-to-face
Advantages the participants’ characteristics; in this way,
• Online interviews and focus groups are much reactivity is reduced.
less expensive to conduct than their face-to- • Online focus groups and interviews allow
face equivalents, especially when conducting people to participate in an environment that
long distance research. is “anonymous, safe and non-threatening”
• People who would otherwise be inaccessible (O’Connor & Madge, 2001, p. 11.2); this may
(for example, in another country) or reluc- be especially helpful for vulnerable groups.
tant to participate (for example, busy senior • Similarly, researchers do not need to enter
executives) are more easily brought into the potentially unsafe environments to reach the
study. people they are studying.
• Interviewees and focus group participants are
able to read and reread their statements (and Disadvantages
in focus groups, those of the other participants • Online focus groups and interviews take
as well), thus allowing for greater reflection. longer than their face-to-face counterparts.
• Audio recording is not required; this elimi- • Only people with access to online facilities and
nates both interviewee apprehension and who are comfortable using them are likely to
the need for time-consuming and expensive participate.
transcription. • It is more difficult for the interviewer to estab-
• Records of online interviews are more accu- lish rapport with interviewees, though this is
rate because nothing has to be transcribed less of a problem when the topic is of particu-
and problems of mishearing don’t arise. This lar interest to the participants.
is a particular advantage with focus group • Probing is more difficult, though not impos-
discussions. sible. Curasi (2001) reported some success
• Participants can use pseudonyms to conceal in eliciting further information from respon-
their identity, making it easier for them to dis- dents, but working online makes it easier for
cuss embarrassing issues or to divulge poten- them to ignore or forget about such requests.
tially unpopular views. • There is less spontaneity in responding since in-
• In focus groups, shy or quiet people may find it terviewees are able to reflect on their answers
easier to participate and overbearing partic- more carefully than is possible in a face-to-face
ipants are less likely to predominate, though situation. However, this can be construed as an
variations in keyboard skills may still prevent advantage in some respects since interviewees
equal participation. are likely to give more considered replies.
• Focus group participants are less likely to be in- • There is a tendency for non-response rates to
fluenced by factors such as the age, ethnicity, be higher in online personal interviews.
appearance, and (if pseudonyms are used) pos- • The researcher cannot be certain that the
sibly even the gender of the other participants. participants are who they say they are (though
• Similarly, interviewees and focus group par- this may sometimes be the case in face-to-
ticipants are less likely to be affected by the face interviews as well).
characteristics of interviewers or moderators, • Online interviews and focus groups require
and the latter are less likely to be affected by considerable commitment from interviewees

bel65796_ch11_249-280 274 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 275

and participants if they have to install soft- • Online connections may be lost and long
ware and remain online for extended periods breakdowns may ensue, interrupting the flow
of time. This is especially problematic if they of the discussion.
have to pay for the software (though reim- • Interviewers cannot rely on body language
bursement may be possible). to tell them when someone is puzzled or (in
• The interviewer/moderator may not be aware the case of focus groups) losing interest in the
that the interviewee/participant is distracted discussion.
by something and in such circumstances will
Sources: Adapted from Adriaenssens & Cadman (1999); Bampton &
continue to ask questions as if they have the Cowton (2002); Clapper & Massey (1996); Curasi (2001); Mann & Stewart
person’s full attention. (2000); O’Connor & Madge (2001); Sweet (2001); Tse (1999).

seem to be especially well placed to gain insight police surveillance on a regular basis, and that it is
into social reality in this way, because of their pro- an important part of their daily lives.
longed immersion in the social settings they study.
Ethnographers are not only in much closer contact Deviant and hidden activities
with people for a longer period of time than inter- There are certain activities that most people are
viewers are, they also participate in many of the same reluctant to talk about in one-on-one interviews:
kinds of activity as the people they are studying. drug-taking, violence, shoplifting, illegal commerce,
Research that relies on interviewing alone is likely hooliganism, and so on. For that reason, much of what
to entail more fleeting contacts, although qualitative is known about criminal and “deviant” subcultures
research interviews can last many hours and re-in- has been gleaned from ethnography. Ethnographers
terviewing is not unusual. have also uncovered information about workplace
resistance practices (such as working-to-rule and
Learning the local language industrial sabotage) and have observed groups that
Becker and Geer (1957) argue that ethnographers are support socially unacceptable ideologies, such as
like social anthropologists visiting a distant land: to white supremacists. Ethnographers are better po-
understand its culture, they must have a good grasp sitioned than interviewers to infiltrate the social
of its language. To penetrate the culture of a group, worlds of people who are wary of talking to outsiders.
it’s particularly important to be familiar with its
“argot”: its informal slang and special uses of words. Sensitivity to context and flexibility
Learning that informal language takes prolonged Extensive contact with a social setting allows the eth-
observation and interaction with others. See Chapter nographer to map the context of people’s behaviour.
12 for more on language analysis. Interacting with people in a variety of different sit-
uations makes it possible to connect behaviour and
Things taken for granted context.
Because interviews rely primarily on the interview-
ees’ accounts of their world, elements of that world Naturalistic emphasis
that the interviewees take for granted may not be Ethnography can be more naturalistic because
mentioned. For example, a street sex worker may the researcher confronts members of a social set-
never mention her relationship with the police, and ting in their natural environments. Interviewing,
an interviewer may never think to ask about it. By which tends to interrupt the normal flow of events
contrast, an ethnographer immersed in the street even when it is informal, is often less naturalistic.
scene may learn that sex workers have to deal with It’s no surprise that when discussing naturalism as

bel65796_ch11_249-280 275 01/13/22 12:33 PM

276 PART III Qualitative Research

a tradition in qualitative research, Gubrium and would be the case with interviews alone. Interviews
Holstein (1997) referred mostly to ethnography. in qualitative research can sometimes be very long,
and as we have noted, re-interviewing is not un-
Advantages of qualitative interviewing common. Nevertheless, the impact of undergoing
in comparison to ethnography an interview is probably still less than the impact
Issues resistant to observation of having to deal with an ethnographer on a regu-
Many issues are simply not open to observation, so lar basis. Ethnography can be especially intrusive in
asking people about them may be the only way to get terms of time when it is conducted in organizational
information about them. For example, consider domes- settings, because it disrupts the rhythm of the work
tic violence. Would it be feasible for a researcher to hang day and other routine activities.
around in other people’s homes, waiting for the violence
Longitudinal research easier
to unfold? Reactivity would impinge in a big way, since
perpetrators would do their best to hide their violence. Interviewing can be carried out within a longitudi-
nal research design somewhat more easily than eth-
Reconstruction of events and future plans nography because repeat interviews may be easier
In-depth reconstruction of past events and plans for to organize than repeat visits to the ethnographer’s
future behaviour is not possible through observa- research settings.
tion alone. Qualitative research entails reconstruc- Ethnography can be used for longitudinal stud-
tion of events when interviewees are asked to think ies, but the time that ethnographers can spend away
about how a series of previous activities might have from their normal lives is usually not more than two
created a current situation. Beardsworth and Keil or three years, and this limits the extent to which an
(1992) used the symbolic interactionist notion of ethnographic study can be longitudinal. When re-
career to understand how people become vegetari- search is conducted on an episodic basis, however,
ans. In fact, most qualitative studies ask about events a longer time period may be feasible. Armstrong’s
that occurred before the study began; for example, (1993) research on soccer hooligans, mentioned in
questions may be asked about the early family expe- Chapters 9 and 10, entailed six years of participant
riences of gang members, recent immigrants, or sex observation, but since hooliganism was not a full-
workers. Some call this “retrospective interviewing.” time occupation for his informants, the research
did not require any sustained absence from work or
Reactive effects other personal commitments.
Reactive effects are by no means straightforward. As
with structured observation (see Chapter 6), it can Greater breadth of coverage
be anticipated that the presence of an ethnographer Interviewing may allow access to a wider variety of
will lead people to behave unnaturally (although eth- people and situations. In ethnographic work, the
nographers, like researchers using structured obser- researcher is invariably limited to a fairly restricted
vation, typically find that their informants become range of people, places, and incidents. An ethnogra-
accustomed to their presence and over time begin pher observing a large organization, for example, is
to behave more naturally). Ethnography also suffers unlikely to have extensive knowledge of operations
from the related problem that an observer disturbs outside the area under study.
the very situation under study, sparking conversa-
tions and interactions, both with and about the ob- Specific focus
server, that otherwise would not occur. As we noted in Chapter 9, qualitative research
sometimes begins with a specific focus; Silverman
Less intrusive in people’s lives (1993) was even critical of the notion that it should
Ethnography can be costly in that the researcher is be regarded as an open-ended form of research.
likely to take up more of the participants’ time than Qualitative interviewing seems to be better suited

bel65796_ch11_249-280 276 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 277

for dealing with a specific issue, since an interview which the participant observer gathers it,” Trow
can address a particular matter in detail. Research (1957, p. 33) reprimanded them for making such
by Bryman, Stephens, & Campo (1996) had a very a universal claim. He argued that “the problem
specific focus: namely, conceptions of leadership under investigation properly dictates the methods
among police officers. Because of its clear focus, it of investigation.” The latter view is very much the
was more appropriate to conduct the research using one taken in this book. Specific research methods
semi-structured interviewing than ethnography, are appropriate for some issues but not for others.
since issues involving leadership are unlikely to Our discussion of the merits and limitations
emerge naturally on a regular basis. of ethnography versus qualitative interviews is
meant to draw attention to some of the factors that
Overview researchers should take into account if given the
When Becker and Geer (1957, p. 28) proclaimed opportunity to choose one method or the other.
over half a century ago that the “most complete The points raised can also be used to evaluate ex-
form of the sociological datum . . . is the form in isting research.

Checklist
Checklist of issues to consider in qualitative interviewing

□ Have you thought about how you will present □ Do the questions offer a real prospect of
yourself in the interview, such as how you will seeing the world from the interviewees’ point
dress? of view rather than imposing a frame of ref-
□ Do you have a clear and comprehensive way erence on them?
of introducing the research to your inter- □ Has the setting in which the interviews
viewees and acquiring their informed con- will take place been checked out? Has the
sent to participate? recording equipment been put through
□ Does the interview guide clearly relate to the a dry run? Have all aids to be used (e.g.,
research questions? visual aids, film clips, case studies) been
pre-tested?
□ Has a pilot test been done with some appro-
priate respondents? □ Is there a plan in place if the interviewee
does not turn up for the interview?
□ Have the interviewers been fully trained?
For a focus group:
□ Does the interview guide contain a good
mixture of question types (e.g., probing, □ Have you planned what you will do if not all
specifying, and direct questions)? participants turn up for the session?

□ Do the interviews allow novel or unexpected □ Have the questions been designed to en-
themes and issues to arise? courage group interaction and discussion?

□ Is the language in the questions free of □ Is there a strategy for dealing with silences and
jargon? for participants who are reluctant to speak?

□ Are the questions relevant to the people □ Is there a strategy for dealing with partic-
being interviewed? ipants who speak too much and “hog” the
discussion?
□ Have the questions been designed to elicit
in-depth responses so that interviewees are □ Is there a strategy to follow if the discussion
not tempted to answer simply “yes” or “no”? goes off on a tangent?

bel65796_ch11_249-280 277 01/13/22 12:33 PM

278 PART III Qualitative Research

Key Points
• Interviewing in qualitative research is typically • The focus group moderator generally tries to
unstructured or semi-structured. give free rein to the discussion. However, there
• Qualitative interviewing is meant to be flexible may be contexts in which it is necessary to ask
in order to explore the world views of research specific questions.
participants. • Group interaction is an important aspect of focus
• If an interview guide is used, it should allow some groups.
flexibility in the way questions are asked. • The Internet offers significant opportunities
• Qualitative interviews should be recorded and for gaining access to potential focus group
then transcribed. participants.

• Qualitative interviews can be conducted online. • Whether to use ethnography or qualitative in-
terviews alone depends mainly on the research
• The focus group allows the researcher to explore questions to be addressed. Still, ethnographers
the joint production of meaning. usually conduct some interviews in the course of
• The selection of focus group participants re- their investigations.
quires several decisions, including whether to
use natural groups.

Questions for Review (R) and Creative Application (A)

Differences between structured inter- R4 Why is it important to record and transcribe
views and qualitative interviews qualitative interviews?
R1 How does qualitative interviewing differ from A4 Imagine that you are reviewing a 50-page tran-
structured interviewing? script of an interview you just did, and you real-
A1 You’ve decided to use qualitative interviewing ize that the interview ended up covering many
to examine the way varsity athletes view their issues that you had no intention of examining
coaches. Explain how using this form of inter- before the interview began. Explain how this
viewing would be more advantageous than could strengthen your research, then explain
structured interviewing for this topic. how it could be a weakness.

Unstructured and semi-structured Introduction to focus groups

interviewing R5 What advantages can a focus group offer com-
R2 What are the differences between unstructured pared to an individual qualitative interview?
and semi-structured interviewing? A5 Come up with three general topics that would be
A2 Devise a semi-structured interview guide that appropriate for focus group research, and explain
could be used to conduct research on what how the data generated from focus groups on
Canadians think about the Black Lives Matter those topics might differ from what would be pro-
movement. The guide must cover the topic of duced through individual qualitative interviews.
racism in Canadian society.
R3 Qualitative data may be “co-constructed” by the Conducting focus groups
researcher and the interviewee. What does that R6 Are there any circumstances in which it is an ad-
mean? vantage to select participants who know each
A3 Invent a hypothetical scenario in which the inter- other? Explain.
pretation of an event (e.g., the death of a loved A6 You’ve decided to conduct a focus group using
one) emerges out of the interaction between a university students to explore the topic of cheat-
researcher and an interviewee. The scenario ing on exams. Would it be better to use students
must include at least three exchanges between who already know each other, or students who
interviewer and interviewee. don’t? Explain.

bel65796_ch11_249-280 278 01/13/22 12:33 PM

 11 Interviewing in Qualitative Research 279

Group interaction in focus group sessions A9 Online focus groups are appropriate for re-
R7 Why is it important to examine group interaction search involving sensitive issues. Identify
when analyzing focus group data? three issues that, because of their sensitive
A7 “The interaction that takes place in a focus nature, would be better researched with online
group is not more ‘natural’ than what tran- rather than in-person focus groups, then ex-
spires in a one-on-one qualitative interview.” plain why the online technique would be more
Discuss. appropriate.

Qualitative research using online per- Qualitative interviewing alone versus

sonal interviews ethnography
R8 What are the pros and cons of doing online per- R10 Outline the advantages and disadvantages of
sonal interviews? qualitative interviewing (without immersion in a
A8 Can online, text-based personal interviews really social setting) compared to ethnography.
be personal interviews? To what extent does the A10 
Is one method more in tune with the research
absence of direct contact mean that the online needs of qualitative researchers than the other?
interview cannot be a true interview? Explain. Explain, using the topic of intimate partner vio-
lence to illustrate your answer.
Qualitative research using online
focus groups
R9 Is the role of the moderator in online focus
groups different from that in the face-to-face
variety? Explain.

Interactive Classroom Activities

1. The instructor divides the class into groups of qualitative interviews for researching the
6–10 people. Each group is to conduct inter- topic chosen;
views using the focus group method. The group d. to explain the disadvantages of the focus
first decides on a general topic (e.g., legalization group method compared to one-on-one
of marijuana, domestic violence, racism, preva- qualitative interviews for researching the
lence of rape culture, climate change, etc.), and topic chosen; and
then produces a list of five general questions that e. to explain how the focus group method may
will be posed by the moderator, who is chosen be better than structured interviews for ex-
by group members from within the group. The ploring the world views of the participants on
moderator conducts the focus group interviews the topic chosen.
with the other members of the group for about
20–30 minutes. When the interviews are finished 2. Each member of the class is given five minutes
the class as a whole then reconvenes, with the in- to think of a general topic that would be appro-
structor asking each group: priate for a semi-structured, one-on-one quali-
a. whether shared meanings and conclusions tative interview (e.g., views on gay marriage, how
emerged from their focus group discussions, the Internet impacts one’s life, life goals and how
and if so, to explain what they were and how they might be achieved, etc.). Each person then
they developed; constructs a five-point interview guide that could
b. whether the moderator’s control of the discus- be used in a semi-structured interview. The in-
sion was excessive, about right, or too weak, structor then uses a random method to pair stu-
and what the consequences of that were; dents up so they can take turns interviewing each
c. to explain the advantages of the focus other on their selected topics. Each interview is
group method compared to one-on-one to last for a minimum of 15 minutes. When the

bel65796_ch11_249-280 279 01/13/22 12:33 PM

280 PART III Qualitative Research

interviews are completed, the class as a whole interview anyway, and how that helped or
reconvenes for a general discussion of: hindered the investigation of the topic; and
a. difficulties in getting the interview to flow c. the sorts of topics that could be usefully re-
smoothly, and how those difficulties may be searched using semi-structured interviews,
resolved; and which topics would be better pursued
b. illustrations of how topics that were not on using ethnography or structured interviews.
the interview guide made their way into the

Relevant Websites
A YouTube video of a qualitative interview uses the A YouTube video offers tips for moderating online
life history approach to examine the experiences of focus groups, with comparisons to in-person focus
a Holocaust survivor. groups.
[Link]/watch?v=-jJ6s7ob_Dg [Link]

A YouTube video of a focus group discusses univer- A YouTube video provides basic tips on qualitative
sity student involvement in extracurricular activities. interviewing.
[Link] [Link]/watch?v=LPwO-vOVxD4

A YouTube video demonstrates how to moderate a A YouTube video demonstrates the use of
focus group. semi-structured interviewing as a Participatory
[Link]/watch?v=xjHZsEcSqwo Action Research method.
[Link]
(Websites accessed 4 September 2021)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch11_249-280 280 01/13/22 12:33 PM

 12 Content Analysis

Chapter Overview
Content analysis involves the examination of various documents and texts, which may be printed, visual,
aural, or virtual. It can be quantitative, focused on coding data into predetermined categories in a systematic
and easily replicable manner, or qualitative, seeking to uncover deeper meanings in the materials. Although
both approaches are discussed, this chapter deals primarily with the qualitative variety. It considers:
• the kinds of research question that content analysis can answer;
• the features of documents or texts that are commonly analyzed;
• how to code (a key part of content analysis);
• semiotics and hermeneutics;
• active and passive audiences;
• conversation analysis and its roots in ethnomethodology;
• the assumptions and analytic strategies of discourse analysis, including critical discourse analysis;
and
• the advantages and disadvantages of content analysis.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Describe how content analyses can be conducted quantitatively and qualitatively and list the criteria
social researchers use to assess the quality of documents to be used as data.
2. Discuss the different types of documents researchers use for conducting content analyses and compare
the strengths and limitations of these data sources.

▲ selimaksan/iStockphoto

bel65796_ch12_281-313 281 01/13/22 12:33 PM

282 PART III Qualitative Research

3. Describe how words, subjects and themes, and value positions are analyzed within documents.
4. Explain how content analysts code their data with or without pre-existing coding schemes.
5. Discuss the different positions in the debate about audiences actively or passively consuming messages
they see and hear.
6. Compare conversation analysis and discourse analysis as distinctive approaches for studying language.
7. Identify the advantages and disadvantages of content analysis.

Rebecca and her boyfriend, Daniel, were sitting at stupid things. But some guys look cool, and they’re
home watching TV on a Saturday night. Neither was the ones who get the girls. They’re better-looking
in a particularly good mood; both thought there and taller than the dorks.” “We could probably do
must be a more exciting way to spend the week- a study comparing men and women in beer com-
end. “Have you ever noticed that there are way mercials,” Rebecca said, her mood beginning to
more food commercials—burgers, pizza, chicken— lighten. “We could even throw in a comparison with
at night than in the daytime?” Rebecca asked. detergent ads. Ever see a man selling detergent?”
“Are there?” Daniel replied. “I know hockey games “You could do that,” Daniel said. “You’d just have to
always seem to come with a lot of beer ads.” find a way to categorize the portrayals—­flattering
“You ever notice how the women look in those or unflattering, active or passive, the age of the
ads?” Rebecca fumed. “They’re always in bathing people, how attractive they are, overtly sexual or
suits or shorts, even if they’re standing on a ski hill non-sexual behaviour—that kind of thing.” “That
in winter. And by some miracle they all have perfect wouldn’t be too hard to do,” Rebecca answered.
bodies.” “Yeah,” said Daniel. “It’s a little different “Sounds like a lot more fun than watching this
with the guys. A lot of them look like dorks, doing stupid stuff.”

Introduction • Do columnists as well as reporters write about

crime?
There can be similarities between quantitative and • Are some crimes given more attention than
qualitative approaches to social research, and even others?
where there are differences the two orientations can • Do more crime stories appear during the week
sometimes be combined or at least used in a comple- or on weekends?
mentary fashion. Content analysis in some ways is • What sorts of crime predominate in newspa-
a blend of the two. It examines forms of communi- per articles (crimes against the person, prop-
cation to see what they reveal about a society, a cul- erty crime, crimes in which society is the
ture, or even the relationships between individuals. victim)?
Meanings and interpretations are important aspects
of content analysis, but so is formal categorization Most content analysis of the media is likely to entail
of the communications, which often includes some several research questions, generally revolving
quantification. For example, suppose you are inter- around the same five W’s that are the basis of any
ested in how newspapers cover crime. You come up news report: who (does the reporting); what (gets
with the following questions: reported); where (does the issue get reported); why
(does the issue get reported); and when (does it get
• Do certain newspapers report more crimes reported). But researchers are also interested in what
than others? media coverage omits. For example, interviews with
• How much crime is reported? Where in the the family of the accused are rare; such inattention is
paper do crime stories appear—on the front itself notable, revealing what is and is not important
page, or inside? to writers and publishers.

bel65796_ch12_281-313 282 01/13/22 12:33 PM

 12 Content Analysis 283

Another common theme in content analysis is • was not produced specifically for the purpose
change in the coverage of an issue over time. For of social research.
example, Buchanan (2014) found that in the period
1894–2005 there was a significant decrease in the Documents are important because they are unobtru-
amount of local news featured in the Toronto Star sive and non-reactive. The fact that the subjects don’t
and the Ottawa Citizen, and that both papers had in- know they are being studied removes a common
creased their national news coverage over this period. threat to the validity of the data.
Content analyses of this kind yield a predomi- In discussing the different kinds of documents
nantly quantitative description of the characteristics used in social science, Scott (1990) distinguished be-
of a communication. Those who perform content tween personal and official documents and, within
analysis in this way claim to be objective and system- the latter group, between private and state docu-
atic, clearly specifying the rules for the categorization ments. These distinctions are used in much of the
of the material in advance. Researchers try to create discussion that follows. Scott also enumerated four
transparency in the coding procedures so that per- criteria for assessing the quality of documents (p. 6):
sonal biases intrude as little as possible. The coding
rules in question may, of course, reflect the research- • Authenticity. Is the evidence genuine and of
er’s interests and concerns and therefore be a product unquestionable origin?
of subjective bias to some degree, but once the rules • Credibility. Is the evidence free from error and
have been formulated, it should be possible to apply distortion?
them without bias. • Representativeness. Is the evidence typical of
Content analysis can be performed on unstruc- what it is supposed to represent (for example,
tured information, such as transcripts of semi- and social life at a particular time and place)? If
unstructured interviews, and even qualitative case not, is the extent of its uniqueness known?
studies of organizations (for example, Hodson, 1996). • Meaning. Is the evidence clear and
Until recently it was used mainly to examine printed comprehensible?
texts and documents, but other sorts of materials may
also be analyzed. If you were interested in gender roles Note that this use of “authenticity” is different from
or representations, for example, you might look at: how the term was used in Chapter 9 to assess quali-
tative research.
• differences in the degree to which men and
women are sexualized on the covers of popular
magazines (Hatton & Trautner, 2013); Personal documents
• pictures on social media (e.g., how parents
portray their children’s gender on Instagram) Diaries, letters, and autobiographies
(Choi & Lewallen, 2018); Diaries and letters are often used by historians, but
• profiles on dating sites and apps (Wada, Hurd they have been given less attention by other social re-
Clarke, & Mortenson, 2019); searchers (see Box 12.1). Whereas letters are written
• animated cartoons (Thompson & Zerbinos, to communicate with other people, diarists presum-
1995); and ably write for themselves. When written for wider
• lyrics of popular songs (e.g., to look for changes consumption, diaries are difficult to distinguish
in the representation of women) (Marcic, 2002). from another kind of personal document: the auto-
biography. Used with a life history or ­biographical
Here we will call all such materials “documents,” method, diaries, letters, and autobiographies
and define them as any data source that: (whether solicited or unsolicited) can be either pri-
mary sources of data or adjuncts to other sources
• can be “read” (including visual materials such such as life story interviews. However, this chapter
as photographs); and is primarily concerned with unsolicited documents.

bel65796_ch12_281-313 283 01/13/22 12:33 PM

284 PART III Qualitative Research

The distinction between biographies and autobi- of problem can arise with other documents. For exam-
ographies can sometimes break down. Walt Disney ple, in the case of Augustus Lamb (Box 12.1), Dickinson
provides a case in point. The first biography of (1993,) noted that there are “only three letters existing
Disney, written by his daughter, Diane Disney Miller from Augustus himself (which one cannot be certain
(1956), almost certainly included information from were written in Augustus’s own hand, since the use of
Mr Disney himself. Subsequent biographies were amanuenses was not uncommon)” (pp. 126–127). This
very similar, because of the tight control exercised by raises the question of whether Augustus was in fact the
the Walt Disney Corporation over the primary mate- author of the letters, especially in the light of his appar-
rials in the Disney archive (letters, notes of meetings, ent learning difficulties.
and so on) out of which the biographies were fash- Turning to the issue of credibility, Scott (1990) ob-
ioned. As a consequence, even though Walt Disney served that there are at least two major concerns with
never wrote an autobiography in the conventional respect to personal documents: their factual accu-
meaning of the term, he (and, later, his company) racy and whether they express the true thoughts and
clearly controlled what was written about him. feelings of the writer. The case of Augustus Lamb
In evaluating personal documents, the authentic- suggests that a definitive, factually accurate account
ity criterion is obviously important. Is the purported is at the very least problematic. Scott recommended
author of the letter or diary the real author? In the case a healthy skepticism regarding the sincerity with
of autobiographies, the increasing use of “ghost” writers which the writer reports his or her feelings. Famous
has made this a standard question. But the same kind people who are aware that their letters or diaries will

Box 12.1 U
 sing historical personal documents: The case of
Augustus Lamb
Dickinson (1993) provided an interesting account
of the use of historical personal documents in the
case of Augustus Lamb (1807–36), the only child of
the novelist Lady Caroline Lamb and her husband,

West7megan/[Link]/GetStock
the second Viscount Melbourne, who was prime
minister of the UK from 1834 to 1841. It is possible
that Augustus suffered from epilepsy, though he
seems to have had other medical problems as well
throughout his short life. Dickinson was drawn to
him because of her interest in nineteenth-­century
reactions to non-institutionalized people with
mental disabilities. In fact, Dickinson doubted that What sorts of dilemmas do you face if you use his-
Augustus was mentally disabled, and suggested torical personal documents as primary sources for
the somewhat milder diagnosis of “learning dif- research?
ficulties.” At the same time, she found that the
people around him had difficulty coming to terms
with his conditions, in large part because of the record of the post-mortem examination conducted
challenge of finding words to describe him that on his remains and extracts from the diary of his
were consistent with his high social status. resident tutor and physician for the years 1817–21.
The chief sources of data were “letters from Despite the many sources she consulted, Dickinson
family and friends; letters to, about and (rarely) still could not conclude with certainty that she un-
from Augustus” (p. 122). Other sources included the derstood definitively what Augustus was like.

bel65796_ch12_281-313 284 01/13/22 12:33 PM

 12 Content Analysis 285

be of interest to others may be very careful about A further problem is the selective survival of doc-
how much of themselves they reveal in their writ- uments like letters. What proportion are damaged,
ings. Similarly, adolescent diary-keepers are often lost, or thrown away? How representative are suicide
aware that their parents could “accidentally” read notes of the thoughts and feelings of all suicide vic-
what they write. In short, what is not said can be of tims? It seems that only a relatively small percentage of
great importance. A particularly poignant illustra- suicide victims leave notes, and some of the notes that
tion is Sugiman’s (2004) suggestion that the reason are written may be destroyed by the victims’ relatives,
many Japanese-Canadian women interned during especially if they contain accusations against family
the Second World War did not write down their members.
experiences was to shield their children from that Finally, understanding the meaning of the docu-
painful episode in their family histories. ments we do have is often problematic. Some pages in
Representativeness is clearly a major concern in a letter or diary may be missing or damaged, or the
assessing these materials. Since literacy rates were writer may have used abbreviations or codes that are
very low in the past, letters, diaries, and autobiog- difficult to decipher. Also, as Scott (1990) observed,
raphies were largely the preserve of a small class letter writers tend to leave much unsaid when they
of wealthy, literate people. Moreover, because boys take it for granted that the people they are writing to
were more likely to receive an education than girls, have the same values, assumptions, and information
the voices of women tended to be under-represented as they do.
in documents of that kind. Some researchers also Researchers often have to search long and hard for
argue that in the past it was less socially acceptable new or surprising materials in diaries. However, this
for women than men to write diaries and autobiog- was not the case with the diaries of William Lyon
raphies. Therefore, such historical documents are Mackenzie King. When the King diaries were first
likely to be biased in terms of their applicability to released to the public in the 1970s, they caused a sen-
the society as a whole. sation (see Box 12.2).

Box 12.2 The Mackenzie King diaries

William Lyon Mackenzie King was prime minister in their proper historical context, the practice of
of Canada for 21 of the 27 years between 1921 spiritualism was quite common when King was a
and 1948. He was an avid diarist for most of his young man in the late nineteenth and early twen-
life, recording his thoughts and feelings about tieth centuries.) Sensationalism aside, the King
both personal issues and national and interna- diaries are a valuable primary source of infor-
tional events. The contents of the diaries suggest mation for a significant portion of Canadian and
that he did not expect them to be made public. world history.
They reveal his political views in considerable Nevertheless, the publication of the King diaries
detail, but also a side of the man that shocked does raise ethical questions. Is it a violation of privacy
many people when the transcripts were released. rights to publish or quote from someone’s diary, es-
King believed strongly in psychic phenomena and pecially if the person is deceased and cannot raise
spiritualism—he attended séances, consulted objections? King’s diaries are now available online
mediums, and believed that he was in communi- at Library and Archives Canada: [Link]
cation with various deceased persons, including [Link]/eng/discover/politics-government/
Franklin Delano Roosevelt and his own mother. prime-ministers/william-lyon-mackenzie-king/
He also believed that he could commune with his Pages/[Link]
deceased dog, Pat. (To place these revelations (accessed 4 September 2021).

bel65796_ch12_281-313 285 01/13/22 12:33 PM

286 PART III Qualitative Research

Visual objects As Scott (1990) observed, many family photos are

taken as records of ceremonial occasions such as wed-
There is a growing interest in the use of visual mate- dings and graduations, or of recurring events such
rials, especially photographs, in social research (see as reunions and holidays. Scott distinguished three
Box 12.3). One of the main reasons why photographs types of home photograph: idealization (a formal
are of interest to social researchers is their potential portrait of a wedding party or the family in its finery);
to reveal important information about families. natural portrayal (an informal snapshot capturing

Box 12.3 Photographs in social research

Photographs can play a variety of roles in social
research. They can be used in qualitative research,
as well as in surveys and experiments. Three prom-
inent roles photographs can play are the following:
• As illustrations. Photographs may be used
to illustrate points and thus to enliven what
might otherwise be rather dry discussions of
findings. This was certainly the case with some
classic reports by anthropologists.
• As data. Photographs may be data in their own

JD Moodie/Library and Archives Canada/C-001823

right. When produced for research purposes,
they essentially become part of the research-
er’s field notes. When based on extant photo-
graphs, they may become the main source of
data about the field in which the researcher is
interested, as in the work of Blaikie (2001) and
Sutton (1992, see Box 12.4). Mannay (2018)
points out that photographs can be particu-
larly useful in providing data about everyday
life in spaces that are often off limits for sus-
This photo of Niviaqsarjuk (left), Suzie (centre),
tained observation, such as people’s homes.
and Jennie (right) was taken by Geraldine Moody
• As prompts. Photographs may be used as
at Fullerton (in what is now Nunavut) in 1904,
prompts to entice people to talk about what
­although the names of the women depicted were
they see in them. Both researcher-created unknown to Library and Archives Canada until
photographs and previously existing ones may 2002. How might photos like this one be used
be used in this way. Sometimes research par- as data? What issues around representation or
ticipants may volunteer their own photographs. ethics might arise when Indigenous cultures are
For example, Riches and Dawson (1998) found represented by people who are not community
in their interviews with bereaved parents that ­members? How might community consultation help
many parents showed them pictures of their mitigate these issues?
deceased children, likely the same pictures
they had shown to neighbours and friends
when they were first grieving their loss.

bel65796_ch12_281-313 286 01/13/22 12:33 PM

 12 Content Analysis 287

action as it happens, though there may be a contrived behind the camera. It’s clear that the question of
component); and demystification ­(depicting the sub- representativeness is much more fundamental than
ject in an atypical—often embarrassing—situation). the issue of what survives, because it points to how
Scott suggested that researchers need to be aware of the selective survival of photographs may be part of
these different types in order to avoid being deceived a reality that family members (or others) deliberately
by the superficial appearance of images. One must seek to fashion. As in Sutton’s example, that man-
probe beneath the surface: ufactured reality itself may then become a focus of
interest for the social researcher.
There is a great deal that photographs do not
tell us about their world. Hirsch [1981, p. 42]
argued, for example, that “The prim poses Government documents
and solemn faces which we associate with
The state is a source of much information of poten-
Victorian photography conceal the reality of
tial significance for social researchers. It produces
child labour, women factory workers, whose
a great deal of quantitative statistical information.
long hours often brought about the neglect of
For example, Bell, Jansen, & Young (2007) used
their infants, nannies sedating their charges
government voting records as well as census infor-
with rum, and mistresses diverting middle
mation to analyze the issue of one-party dominance
class fathers.” (Scott, 1990, p. 195)
in Alberta. The state is also the source of a great
As Scott argued, this means not only that the photo- deal of textual material of potential interest, such
graph must not be taken at face value when used as a as official reports. For instance, in his study of the
research source, but that the viewer must have con- issues surrounding synthetic bovine growth hor-
siderable knowledge of the social context in order to mone, Jones (2000) used transcripts of a Canadian
get its full meaning. Sutton (1992) makes a similar senate inquiry on the topic. Briefly, Monsanto, the
point in Box 12.4. In fact, we may wonder whether manufacturer, lost the battle to have the hormone
photographs in such situations can be of any use accepted for use; health groups claimed that it was
to a researcher. At the same time, a researcher’s in- unnecessary and, worse, that it posed a health risk
terpretations should not be accepted uncritically. both to cows and to the humans who drank their
For example, the “prim poses and solemn faces” of milk. Abraham (1994) used similar materials in his
Victorian portraits might well “conceal” a bleak and research on the role of self-interest and values in sci-
miserable reality; but is that all? Could they not also entists’ evaluations of the safety of medicines, spe-
conceal moments of human warmth or compassion? cifically the drug Opren. The author described his
Or fun? sources as “publicly available transcripts of the tes-
A particular problem for the analyst of photo- timonies of scientists, including many employed in
graphs, according to Scott, is judging representa- the manufacture of Opren, Parliamentary debates,
tiveness. Photos that survive the passage of time—in questions and answers in Hansard, and leaflets,
archives, for example—are very unlikely to be rep- letters, consultation papers and other documenta-
resentative for the simple reason that somebody at tion disposed by the [drug regulatory authority]”
some time decided they should be preserved. The (Abraham, 1994, p. 720). His research showed in-
discussion in Box 12.4 suggests that decisions about consistencies in the scientists’ testimonies, suggest-
which photos to keep or post online may reflect the ing that self-interest can play an important role in
needs and biases of the people who make those deci- such situations. He also used his findings to argue
sions. Another problem relates to the issue of what is that the scientific ethos of “objectivity” has limited
not photographed, as suggested by Sutton’s idea that applicability in areas where self-interest may be a
unhappy events at Disney theme parks may not be factor.
photographed at all. Awareness of what is not photo- In terms of Scott’s (1990) four criteria, such ma-
graphed can reveal the “mentality” of the person(s) terials can certainly be seen as authentic and as

bel65796_ch12_281-313 287 01/13/22 12:33 PM

288 PART III Qualitative Research

Box 12.4 Photographs of the Magic Kingdom

Sutton (1992) noted a paradox about visitors to
Disney theme parks. The Magic Kingdom is sup-
posed to be “the happiest place on Earth,” and
employees (“cast members”) are trained to en-
hance visitors’ experiences. Yet it is clear that

PamelaJoeMcFarlane/iStockphoto
some people do not enjoy themselves while
visiting. Time spent waiting in lines is a particu-
larly common gripe for “guests” (Bryman, 1995).
Nonetheless, because people expect their visit
to be momentous, they take photographs that
support their assumption that Disney theme
parks are happy places. When they return home, This family seems to be enjoying their vacation,
they “discard photographs that remind them but should we conclude that their holiday was fun
of unpleasant experiences and keep those that in every respect? How can the knowledge that a
photo may not provide a complete record of an
remind them of pleasant experiences” (Sutton,
event or experience affect how we approach the
1992, p. 283). In other words, positive feelings
use of photographs as data?
about an experience may be a post hoc recon-
struction, substantially aided by the photos that
are not thrown away or deleted. Thus, photo- holidays? With social media, the purpose of
graphs may provide an incomplete, somewhat ­photo-assisted reconstruction seems to be ex-
distorted record. panding: Now photos may be used not only to
Similar observations can be made with create happy memories, but also to present a
regard to pictures posted on social media. How public image of oneself as a fun-loving, inter-
common is it to see Instagram photographs of esting person. Clearly, holiday photos represent
people having unpleasant experiences on their more than an objective account of a vacation.

having meaning (in that they are clear and compre- produce many documents, some of which are in the
hensible to the researcher). However, the credibility public domain; the latter include annual reports,
criterion requires us to consider whether the docu- press releases, advertisements, and public relations
mentary source is biased. This is exactly the point of material both in printed form and on the Web. Other
Abraham’s (1994) research: such documents can be documents, such as company newsletters, organiza-
interesting precisely because of the bias they reveal. tional charts, minutes of meetings, memos, corre-
Equally, this point suggests that caution is necessary spondence (internal and external), and manuals for
in attempting to treat them as depictions of reality. new recruits, may not be accessible to the public.
This kind of material is often used by organizational
Official documents from ethnographers in their investigations, but the diffi-
culty of gaining access to it means that many other
private sources researchers have to rely on public domain docu-
Official documents available from “private sources” ments. Even if the researcher is an insider with the
vary widely, but a common category is company doc- organization, certain private documents may remain
uments. Companies (and organizations generally) inaccessible.

bel65796_ch12_281-313 288 01/13/22 12:33 PM

 12 Content Analysis 289

Private documents also need to be evaluated Issues of representativeness are also important.
using Scott’s four criteria. As with the materials Did Forster have access to a complete set of docu-
considered in the previous section, documents de- ments? It could be that some had been destroyed,
riving from private sources such as companies are or that he was not allowed access to certain sen-
likely to be authentic and meaningful (in the sense sitive ones. This is not to say that such documents
of being clear and comprehensible to the researcher). necessarily exist, but a healthy skepticism is often
However, issues of credibility and representativeness warranted.
are still likely to require scrutiny.
People who write documents generally want to Mass media outputs
convey a particular point of view. An interesting
illustration is provided by Forster’s (1994) study of Newspapers, magazines, television programs, films,
career development issues in a major retail com- and other mass media are potential sources for social
pany. Forster analyzed company documentation as scientific analysis. For example, Parnaby’s (2003)
well as interviews and a questionnaire. Because he study of how Toronto tried to deal with its squeegee
was able to interview many of the authors of the kids examined 200 newspaper articles that appeared
documents, “both the accuracy of the documents between 1995 and 2000. Hallgrimsdottir, Phillips, &
and their authorship could be validated by the Benoit (2006) looked at how newspapers in Victoria,
individuals who had produced them” (p. 155). In BC, characterized sex-trade workers in the province,
other words, the authenticity of the documents was while Harding (2006; 2010) analyzed press cov-
confirmed and apparently the credibility as well. erage of Indigenous people and their institutions.
However, Forster also said that the documents re- Dumitrica and Bakardjieva (2018) examined how
vealed divergent interpretations of key events and Canadian newspapers construct the role played by
processes: social media in civic engagement. Bhuyan, Jeyapal,
Ku, Sakamoto, & Chou (2017) used documents
One of the clearest themes to emerge was the
published by Citizenship and Immigration Canada
apparently incompatible interpretations of the
as well as articles from the Toronto Star, the Globe
same events and processes among the three
and Mail, and the National Post to investigate how
sub-groups within the company—senior exec-
the use of language in those sources is related to the
utives, HQ personnel staff, and regional per-
development of social ideologies and the implemen-
sonnel managers. . . . These documents were
tation of government policies on immigration. All
not produced deliberately to distort or obscure
of these studies are examples of critical discourse
events or processes being described, but their
analysis (discussed later in this chapter) in that they
effect was to do precisely this. (p. 160)
examined the social and political implications of the
The perspectives that members of the different materials they examined, an approach that can be
groupings expressed in the documents reflected taken with virtually any medium of communication.
their positions in the organization. Consequently, Films, television shows, and magazines have sim-
although the authors of the documents were able to ilar potential for research. For example, when Coté
confirm their authenticity, their contents could not and Allahar (1994) examined magazines aimed at
be regarded as “free from error and distortion,” as adolescents, they concluded that these “teenzines”
Scott put it. In other words, the accounts that docu- turned adolescents into uncritical consumers and
ments provide may not be completely objective. They diverted them from protesting against their lack of
have to be analyzed critically and compared with adult privileges.
other sources of data. As Forster’s study suggests, the Authenticity is sometimes difficult to ascertain in
different stances taken by the authors of documents the case of mass media outputs. The authors are not
can be used to develop insights into the processes always identified, as in the case of a TV news report,
and factors that lie behind their creation. so it is sometimes difficult to know whether a given

bel65796_ch12_281-313 289 01/13/22 12:33 PM

290 PART III Qualitative Research

Research in the News

Tracking the gender gap in Canadian online news sources
Shari Graydon’s non-profit organization, gender gap by news outlet. The aggregate data
Informed Opinions, aims to promote women’s from October 2018 onward shows that women’s
voices to enhance democracy. Collaborating voices accounted for 26 per cent of those quoted
with a group of researchers from Simon Fraser by major Canadian news outlets. Graydon
University, Graydon and her team are examin- (2019) points out, “For context, that's a mere
ing the gender gap in Canadian online news four-point increase over data we collected
coverage (Graydon, 2019). To do so, they almost three decades ago.” She hopes that
are harnessing the power of big data (see making the tracker publicly accessible will help
Chapter 8) to track, in real time, the ratio of promote change, but acknowledges that the
males to females quoted online as sources by tool is just one piece of the puzzle: “To make a
Canadian news outlets. Their digital monitoring difference, journalists need to actively seek to
system, the Gender Gap Tracker, “analyzes in improve the data—and news consumers need
microseconds what it used to take researchers to give them reasons to do so.” The Gender Gap
hours to assess” (Graydon, 2019). The tool, in Tracker can be accessed online at https://
effect, offers an ongoing content analysis of the [Link].

account was prepared by someone in a position to including theme parks, technologies, paintings, and
know all the facts. Credibility is frequently an issue, buildings. Contemporary writers and researchers
and as the examples used in this section show, it is regard such “texts” as materials that can be inter-
often the uncovering of error or distortion that is preted to produce “readings.”
the objective of the analysis. Representativeness may Websites and webpages are further examples
not be an issue with newspaper or magazine articles, of “texts” that can be “read.” The vastness of the
since many publications make a point of maintain- Internet and its accessibility make it a valuable
ing a consistent tone or ideological bent. Finally, al- source of documents for both quantitative and
though the literal meaning of mass media outputs is qualitative data analysis. Hier (2000), for example,
often clear, it usually takes some reflection and the- examined a Toronto-based racial supremacy web-
oretical analysis to appreciate the broader societal site and found that it allowed people access to racist
impact these forms of communication can have. ideas in a relatively anonymous way (e.g., without
having to subscribe to a hard-copy newsletter). For
that reason, he feared that it might be more success-
Virtual outputs and the ful in reaching ordinary citizens than non-­Internet
Internet as objects of sources would be. Johnston, Sanscartier, and
Johnston conducted qualitative content analyses ex-
analysis ploring the online ratings of Canadian psychiatrists
“Text” is a word that has up to this point been used (M.S. Johnston, 2020), retail employers (Johnston,
sparingly in this chapter. It originally referred to a Sanscartier, & Johnston, 2019), and food service em-
written document, but in recent years it has been ployers (Johnston, Sanscartier, & Johnston, 2018).
applied to an increasingly wide range of phenomena Drawing on patient and employee reviews from

bel65796_ch12_281-313 290 01/13/22 12:33 PM

 12 Content Analysis 291

the websites [Link] and RateMyEmployer. a sense of community with citizens, it affords de-
ca, they make a case for how virtual spaces provide partments greater control in managing their public
opportunities for resistance and discourse about image.
healthcare and workplace issues that might other- Websites and other online data sources clearly
wise be suppressed. have huge potential, but it’s important to keep Scott’s
The use of images on websites can also be quite re- criteria in mind. First, authenticity: anyone can set
vealing. Crook and Light (2002) analyzed the photo- up a website, so you have no guarantee that a person
graphs in 10 university prospectuses and found that offering information (such as financial advice) is an
although there were many photos of students appar- authority. Second, credibility: is the information on
ently studying, the setting was rarely a library carrel the site credible, or might it be distorted for some
or a dorm room. Instead, students were usually reason? For example, a site that encourages you to
shown in “social” learning situations that allowed buy stock held by its author might exaggerate its
them to appear active and engaged, frequently out value. Third, given the constant flux of the Internet,
of doors. The authors argued that those less typical it’s doubtful that one can ever know how represen-
learning contexts were chosen because they are more tative websites on a certain topic are. New websites
seductive. In a study of Web-based family photog- are continually appearing, others disappearing,
raphy, Pauwels (2008) concluded that digital tech- still others being modified. Online searching is like
nology has made it easier for families to “construct trying to hit a target that is not only moving but
fictions and fantasies” and to project their values and also in a constant state of metamorphosis. In a re-
norms than was the case when paper-based family lated vein, any one search engine can provide access
albums were more common. to only a portion of the Web; there is evidence that
Other forms of Internet-based communication, even the combined use of several search engines
such as email lists, discussion groups, and chat gives access to just under half of the sites that exist,
rooms, have also been used as objects of analysis. For and there is no way of knowing if they are a biased
their study of online social support groups, Nettleton, sample. Finally, websites are notorious for a kind of
Pleace, Burrows, Muncer, & Loader (2002) exam- “webspeak” that makes it difficult to comprehend
ined the interactions between people who use those what is being said without some insider knowledge.
sorts of virtual communications. Similarly, social
media such as Facebook, Instagram, Twitter, and
YouTube have been analyzed by social researchers.
For instance, Lalancette and Raynauld (2019) stud-
ied Justin Trudeau’s use of Instagram during his first
Practical Tip |
year as prime minister. They found that his embrace Referring to Websites
of social media helped Trudeau present himself as a It has become common practice in academic
happily married and loving husband and father, and work, when referring to websites, to include
as a leader comfortable in both informal and formal the date they were consulted. This convention
settings. Such framing conveyed a personal image is closely associated with the fact that web-
in keeping with his “sunny ways” campaign to re- sites often disappear and frequently change,
introduce a positive approach to politics. O’Connor so that if subsequent researchers want to
(2017) studied how Canadian police departments check or follow up a Web reference, they may
use Twitter to share information and interact with find that the site has changed or even disap-
the public. He points out that social networking sites peared. Citing the date(s) the site was visited
allow police departments to bypass the news media can help to relieve anxieties about citing Web
in getting their message directly to the public. This sources.
not only helps increase police visibility and build

bel65796_ch12_281-313 291 01/13/22 12:33 PM

292 PART III Qualitative Research

Researchers basing their investigations on web- are addressed? Who benefits and how? Who inter-
sites need to recognize these limitations as well as prets for whom and who represents whom?” (p. 243).
the opportunities available. Scott’s suggestions invite Many sources of information, including much of the
consideration of why a website was constructed in the research in the social sciences, serves some larger if
first place: for commercial reasons? Political reasons? unstated political or cultural purpose. Kempf (2006),
In other words, you should be no less skeptical about for example, in an analysis of history textbooks and
websites than about any other kind of document. Ontario Ministry of Education guidelines from 1860
Employing both traditional printed documents and to 2006, concludes that a lot of this material showed
website materials will allow you to cross-validate a disrespect for Indigenous perspectives and ways of
sources. knowing, and in effect provided a rationalization for
the mistreatment of Indigenous Canadians.
Beyond Scott’s criteria
So far, most of the biases discussed in the context
of Scott’s criteria have been of the personal type—
What things need to be
someone may write a flattering if somewhat unreal- analyzed?
istic autobiography, pictures for public display may Obviously, what is to be analyzed depends on the
be chosen in order to create a favourable impression, specific research questions. In mainly quantitative
and so on. Keep in mind that there may be larger cul- studies, these elements are usually specified in
tural and political biases involved in the creation and ­advance in order to guide both the selection of the
use of any type of text or document. Hales (2006), media to be assessed and the construction of
for example, points out that a colonialist mentality the coding schedule. They usually include words,
often pervades research conducted on subject or sub- subjects and themes, and value positions, as dis-
ordinated peoples. Where there is a power differen- cussed next.
tial between the researcher and those who are being
studied, Hales notes that one would do well to ask, Words
“Why is the research done? How is it done? Who de- Determining the frequency with which certain
fines, initiates and conducts the research? On/with/ words are used is often the first step in content anal-
for whom is the research carried out? What topics ysis. Jagger (1998), for example, searched dating

Checklist
Checklist for evaluating documents

Have the following questions been answered? □ Is the document typical of its kind? If not, is it
possible to establish how atypical it is and in
□ Who produced the document?
what ways?
□ Why was the document produced?
□ Is the meaning of the document clear?
□ Was the person or group who produced the
□ Can the events or accounts presented in the
document in a position to write authorita-
document be corroborated?
tively about the subject?
□ Are there different interpretations of the
□ Is the material genuine?
document from the one you offer? If so, what
□ Did the author have an axe to grind or a par- are they? Have you discounted them? If so,
ticular slant to promote? why?

bel65796_ch12_281-313 292 01/13/22 12:33 PM

 12 Content Analysis 293

advertisements and counted words such as “slim” Value positions

and “non-smoker” to compare women and men with A further level of interpretation is likely when the re-
respect to what each deemed desirable in a date (see searcher is seeking to demonstrate that the writer of
Box 12.5). Simple counting of particular words can a particular text has taken a certain value position.
reveal emphasis, style of writing or presentation, and For example, is a journalist who writes about crime
even the overplaying of certain events. For exam- sympathetic or hostile to the accused? Is all the
ple, Dunning, Murphy, & Williams (1988) noted a blame put on the accused, with the implication that
tendency for the British press to sensationalize dis- punishment is appropriate? Or is the focus on social
turbances at soccer matches by using emotive words conditions, with the assumption that less blame
such as “hooligans” and “louts” to refer to audience should be placed on the criminal? If there is no man-
members, and terms such as “battle” to refer to a ifest indication of the writer’s value positions, can
game; less dramatic terms such as “fans” or “con- inferences be made from the latent content?
test” would have encouraged more neutral responses Another way in which content analysis can reveal
among readers. value positions is through the coding of ideologies,
A variation on the search for individual keywords beliefs, and principles. For instance, Sturgeon and
is the search for pairings of keywords. The grow- Morrissette (2010) examined suicide ideation among
ing availability of the written news media online Manitoba farmers who called a rural crisis line.
greatly facilitates this kind of search. Hier (2002) The data were gathered from forms that were filled
found “rave” and “drug use” to be frequently linked, out by the crisis line counsellors (which are similar
which may have encouraged readers to believe that in design to the coding manual depicted in Figure
raves must be controlled. Parnaby (2003), on the 12.2). One of the themes that emerged as codes were
other hand, noted that “squeegee kids” and “home- developed was “family salvation,” which involved
lessness” were not paired, even though many of the the belief that life is worth living if one has meaning-
“kids” were in fact homeless; as a result of this omis- ful relationships with family members. “I just keep
sion, he argued, the “kids” were constructed “as a going because I think someday my son will need
social problem requiring a law and order resolution” me,” said one farmer (p. 199). Surprisingly, only a
(p. 281). The search for pairings of keywords can be a small proportion of farmers mentioned their spouses
starting point for a more in-depth analysis. or life partners as the people who made their lives
bearable; others such as siblings, children, and ex-
Subjects and themes
tended family members were more likely to be cited.
A more interpretative approach is required to code
text in terms of subjects and themes. At this point,
the analyst is searching not just for the obvious or Coding
manifest content, but also for some of the underlying As the foregoing has implied, coding is a crucial part
or latent content as well. This occurs if the researcher of content analysis. There are two main elements to a
wants to probe beneath the surface to ask deeper content analysis coding scheme: designing a coding
questions about what is happening. In reports on schedule and creating a coding manual. To illus-
crime, for example, is the victim blamed along with trate, imagine that you are interested in crime re-
the accused? Is the occupation of the accused or the ports in a local newspaper. To simplify, assume that
victim stated? If not, is it implied in an address or a your study is limited to crimes where the victim is a
picture? Why is it that news stories about men who person rather than an organization, and that it con-
have been mugged sometimes mention their marital siders these variables:
status? In seeking to answer these sorts of questions,
the quantitative aspects of content analysis can serve 1. nature of the offence(s)
as a starting point for a study in which qualitative 2. gender of suspect
research becomes the central focus of the endeavour. 3. occupation of suspect

bel65796_ch12_281-313 293 01/13/22 12:33 PM

294 PART III Qualitative Research

Box 12.5 Finding love: Then and now

Jagger (1998) reported a content analysis of 1094 There was also a somewhat greater propensity
dating advertisements in four newspapers with for women to market themselves in terms of
a general readership. The sample was chosen physical appearance. As an aside, men were just
over two four-week periods in 1996. Three re- as likely to market themselves as “slim,” suggest-
search questions drove the study: ing that certain preferences in body shape were
not exclusive to one gender. More generally, her
• What is “the relative significance of monetary
results pointed to the significance of the body in
resources and lifestyle choices as identity
identity construction for both men and women.
markers and desirable attributes for men and
In a later publication, Jagger (2001) reported the
women”?
findings from a qualitative content analysis of a
• How do men and women vary in the ways
sub-sample of the 1094 advertisements.
they choose to market themselves and de-
Much has changed in the years since Jagger
scribe their preferred (or ideal) partners in
did her research. Internet dating sites and
terms of the body?
apps such as Match and Tinder have eclipsed
• To what extent are “traditional stereotypes of
the classified newspaper advertisement as
masculinity and femininity . . . still in opera-
the mediums of choice for people whose in-­
tion”? (p. 799)
person activities do not provide adequate op-
Jagger noted the tendency for considerable portunities to meet a romantic partner. Since
percentages of both men and women to market space is not at a premium, digital platforms
themselves in terms of their lifestyle choices (film allow participants to give far more informa-
preferences, clubs frequented, sports activities, tion than was the case with newspaper ads,
etc.). She also found that women were far more and of course the provision of photographs
likely than men to stress the importance of eco- means that verbal descriptions of physical ap-
nomic and other resources in a preferred partner. pearance have become far less important than

4. age of suspect Figure 12.1 is a dimension (indicated by the column

5. gender of victim heading) to be coded. The blank cells on the coding
6. occupation of victim form are the places in which codes are to be writ-
7. age of victim ten. One row is used for each media item coded. The
8. depiction of victim codes can then be transferred to a computer data file
9. position of news item in the paper for analysis with a software package such as IBM
SPSS Statistics (SPSS).
Content analysts are normally interested in a much
larger number of variables, but this simple illustra- Coding manual
tion can help to get across how a coding schedule and On the face of it, the coding schedule in Figure 12.1
coding manual operate. seems bare, providing little information about what is
to be done or where. This is where the coding manual
Coding schedule comes in. It is a set of instructions to coders that in-
The coding schedule is a form onto which the data cludes all possible categories for each dimension to
are entered (see Figure 12.1). Each of the columns in be coded. It provides a list of all the dimensions; the

bel65796_ch12_281-313 294 01/13/22 12:33 PM

 12 Content Analysis 295

they once were. However, many sites still fea- meeting before getting involved in a roman-
ture “dating profiles” in which people have a tic relationship (Finkel, Eastwick, Karney, Reis,
few paragraphs to describe themselves and to & Sprecher, 2012), and not without reason.
outline what they would like to find in a part- Hitsch, Hortaçscu, & Ariely (2010) compared
ner, and the content of these profiles is some- the information provided in online dating
what similar to the ­older-style ad descriptions. sites with American national averages, and
For example, Wada et al.’s (2019) research on found that both men and women on the sites
older (60+) Canadians’ use of the dating site claimed higher than average heights, and
Plenty of Fish focused on analyzing the “About that women but not men indicated lower than
Me” and “Interests” sections of users’ profiles. average weights. Using self-reports of prob-
They found that users emphasized an idealized able misrepresentation on dating sites, Hall
image of aging in portraying themselves and et al. (2010) found that men were more likely
what they were looking for in a partner. Being than women to be deceptive about their per-
active, healthy, productive, positive, and per- sonal assets (e.g., income), but that there was
sonable were some of the standout attributes no gender difference in misrepresentation of
people used to characterize themselves. For in- relationship goals (e.g., interest in pursuing a
stance, words such as “active,” “busy,” and “on serious relationship). Women were found to be
the go” were chosen to describe how engaged more likely to misrepresent their weight, while
they were with their communities. men, contrary to expectation, were more likely
Not surprisingly, a major concern with to mislead about their age. Older adults often
people who engage in online dating is the use vague language about age in their profiles
misrepresentation of personal characteris- (e.g., “youthful mindset and optimal health”) to
tics by potential mates (Hall, Park, Song, & portray themselves as being younger than they
Cody, 2010). Most users want an in-person are (Wada et al., 2019, p. 954).

different categories subsumed under each dimen- a simple social-class scheme. The offences are classi-
sion; the numbers (that is, codes) that correspond fied according to the categories used by the police in
to each category; and guidance to coders on what recording crimes. (Since police statistics have been
should be taken into account in coding a particular criticized for weak reliability and validity—recall
dimension. Figure 12.2 shows a coding manual that Chapters 4 and 5—a comparison between police data
corresponds to the coding schedule in Figure 12.1. and crime reports in local newspapers is a possible
Our coding manual includes the occupation of research topic.) Finer distinctions can be used, but
both the suspect and the victim of the crime, using unless you are planning to examine a large sample of

Case Day Month Nature Gender Occupation Age of Gender Occupation Age of Depiction of Nature Position
number of offence I of suspect of suspect of of victim victim victim of of news
suspect victim offence II item

FIGURE 12.1 Coding schedule

bel65796_ch12_281-313 295 01/13/22 12:33 PM

296 PART III Qualitative Research

Nature of offence I 12. Other

01. Violence against the person 13. Unemployed
02. Sexual offences 14. Homemaker
03. Robbery 15. Student
04. Burglary in a dwelling 16. Retired
05. Burglary other than in a dwelling 17. Unknown
06. Theft from a person
07. Theft of bicycle Age of suspect
08. Theft from shops Record age (–1 if unknown)
09. Theft from vehicle
10. Theft of motor vehicle
Gender of victim
11. Other theft and handling stolen goods
1. Male
12. Fraud and forgery
2. Female
13. Drug offences
3. Unknown
14. Other offences

Occupation of victim
Gender of suspect
Same as for occupation of suspect
1. Male
If not applicable, code as 99
2. Female
3. Unknown
Age of victim
Record age (–1 if unknown)
Occupation of suspect
01. Professionals, administrators, officials and managers in
large establishments; large proprietors Depiction of victim
02. Lower-grade professionals, administrators, and of- 1. Victim responsible for crime
ficials; higher-grade technicians; managers in small 2. Victim partly responsible for crime
business and industrial establishments; supervisors of
non-­manual employees 3. Victim not at all responsible for crime
03. Routine non-manual employees in administration and 4. Not applicable
commerce
04. Personal service workers
Nature of offence II (code if second offence mentioned in relation
05. Small proprietors, artisans, etc., with employees to the same incident; code 0 if no second offence)
06. Small proprietors, artisans, etc., without employees Same as for Nature of offence I
07. Farmers and smallholders; self-employed fishermen
08. Lower-grade technicians, supervisors of manual workers Position of news item
09. Skilled manual workers 1. Front page
10. Semi-skilled and unskilled manual workers (not in 2. Inside
agriculture)
3. Back page
11. Agricultural workers

FIGURE 12.2 Coding manual

news items, broader categories are preferable. Note dimensions. It is on the basis of these lists that a coding
also that the coding schedule and manual permit the schedule of the kind presented in Figure 12.1 is created.
recording of two offences for any incident. If there Detailed rules about how to code should be formulated,
are more than two, you will have to make a judgment because reliability, both inter-coder and intra-coder, is
concerning the two most significant offences. always a concern (see further on in this section).
The coding manual is crucial because it provides a Here is a news report of a fictional road-rage inci-
complete listing of all categories for each dimension to dent. Two male motorists, one a retired schoolteacher
be coded, as well as guidance on how to interpret the aged 68, the other a 26-year-old assembly-line

bel65796_ch12_281-313 296 01/13/22 12:33 PM

 12 Content Analysis 297

worker, got into an argument and the worker al- • Clear instructions. Coders should be clear
legedly punched the retired man, causing him to about what factors to take into account when
fall, hit his head, and suffer a concussion. There was assigning codes. Sometimes these have to be
no second offence. The coding of the incident would very elaborate. In the sort of content analysis
then appear as in Figure 12.3 and the data would be described in the previous section, coders gen-
entered into a computer program such as SPSS. erally have little or no discretion in deciding
Suppose a second article, appearing the next how to code the units of analysis.
day, described how an unemployed 34-year-old • A clear unit of analysis. For example, in the
female reportedly took an 86-year-old woman’s imaginary study of the reporting of crime in
purse and then knocked her down. The code is pro- the local press, there is both a media item (for
vided in Figure 12.4 but with a few errors. Can you example, one newspaper article) and a topic to
spot them? Forms like these would be completed be coded (one of two offences). In practice, a
for each news item within the chosen period(s) of researcher is interested in both but needs to
study. keep the distinction in mind.

Potential pitfalls in devising coding To enhance the quality of a coding scheme, it is

schemes advisable to conduct a pilot study to identify diffi-
The potential dangers in devising a content analy- culties, such as finding there is no code to cover a
sis coding scheme are similar to those involved in particular case (not exhaustive). A pilot test will also
designing structured interview and observation help to reveal if one category of a dimension includes
schedules. Points to recall include the following: an extremely large percentage of items. When this
occurs, it may be necessary to break that category
• Mutually exclusive categories. There should down to allow for greater specificity.
be no overlap in the categories supplied for As we have mentioned, the reliability of coding is
each dimension. If the categories are not mu- a further concern. An important part of pre-testing
tually exclusive, coders will not know how the coding scheme is examining consistency between
to code an item that fits into more than one coders (inter-coder reliability) and, if time permits,
category. intra-coder reliability. The process of gauging re-
• Exhaustive categories. Every possible dimen- liability is more or less the same as in structured
sion should have a category. ­observation, discussed in Chapter 6.

Case Day Month Nature of Gender of Occupation Age of Gender Occupation Age of Depiction Nature of Position
number offence I suspect of suspect suspect of victim of victim victim of victim offence II of news
item
001 24 11 01 1 10 26 1 16 68 2 00 2

FIGURE 12.3 Completed coding schedule

Case Day Month Nature of Gender of Occupation Age of Gender of Occupation Age of Depiction of Nature of Position
number offence I suspect of suspect suspect victim of victim victim victim offence II of news
item
002 25 12 06 1 13 32 2 16 86 3 01 2

FIGURE 12.4 Completed coding schedule with errors

bel65796_ch12_281-313 297 01/13/22 12:33 PM

298 PART III Qualitative Research

Content analysis without a Altheide and Schneider (2013) outlined an ap-

proach they called ethnographic content analysis
pre-existing coding scheme (ECA). They described this approach as differing
Content analysis may be undertaken before any from quantitative content analysis in that the re-
specific decisions have been made regarding how searcher is constantly revising the themes or catego-
the information available for analysis will be ries distilled from the examination of documents. As
coded. In such cases the data are searched in order they put it:
to find underlying themes, making the study pri-
ECA follows a recursive and reflexive ­movement
marily qualitative in nature. Like other qualita-
between concept development-­ sampling-data
tive approaches, qualitative content analysis tends
collection-data coding-data analysis-­interpretation.
to be inductive, with themes emerging through
The aim is to be systematic and analytic but not
an iterative process as the researcher interprets
rigid. Categories and variables initially guide
the data. Kennedy (2006), for example, in his
the study, but others are allowed and expected
study of the Canadian fathers’ rights movement
to emerge during the study, including an orien-
called “Fathers 4 Justice,” used content analysis
tation toward constant ­discovery and constant
in conjunction with participant observation and
comparison of relevant situations, settings, styles,
semi-structured interviews to examine how the
images, meanings, and nuances (p. 26; emphases in
movement developed a collective identity. The
original).
content analysis was conducted on documents
such as position papers submitted to government While more quantitatively oriented analysis typically
agencies, newsletters, pamphlets, and minutes entails applying predefined categories to the sources,
from meetings, and was used mainly to make ECA allows for greater refinement of those catego-
sense of the data gathered through participant ries as well as generation of new ones. For example,
observation and interviews. Similarly, Bell (2007) Canadian researchers Schneider and Hannem (2019)
used pamphlets, speeches, newsletters, and other examined how the sexual misconduct allegations
texts to arrive at an overall assessment of western against Donald Trump in the 2016 US election were
Canadian separatism as a neo-liberal movement. framed as a political issue, which had the effect of
The processes by which the themes are extracted minimizing victims’ experiences. Instead of begin-
in this kind of content analysis are often left im- ning with a pre-existing coding scheme, they con-
plicit, although they are usually illustrated with ducted a preliminary analysis of a few potential
quotations from the text in question. documents to be used in their study (articles from
Lynch and Bogen (1997) examined core sociolog- the Washington Post and New York Times). This early
ical textbooks and found that they contained recur- reading of possible data informed their initial cate-
ring themes that presented an upbeat and scientific gories and codes to be used as part of a data collec-
view of the discipline—one that, in the ­researchers’ tion protocol. The subsequent analysis then followed
view, was biased and value laden. Seale (2002) exam- an iterative process to construct qualitative themes
ined newspaper reports about people with cancer. in the data. In addition, ECA emphasizes the con-
One of the phases of his analysis entailed an “NVivo text within which documents are generated. For in-
coding exercise, in which sections of text concern- stance, a study of newspaper reports of crime would
ing themes of interest were identified and retrieved” require some understanding of news organizations
(p. 109). He was especially interested in gender and the work that journalists do (Altheide, 2004).
differences in the representation of sufferers, and
demonstrated that stories about men were much Semiotics
more likely than stories about women to discuss the Another form of predominantly qualitative content
role that the individual’s character played in dealing analysis is semiotics, the science of signs. In social
with the disease. research, semiotics involves analysis of the signs and

bel65796_ch12_281-313 298 01/13/22 12:33 PM

 12 Content Analysis 299

symbols encountered in everyday life. It can be ap- see beyond and beneath the apparent ordinariness
plied to documentary sources as well as many other of everyday life and its manifestations. The main
kinds of data. The main terms employed in semiotics difficulty with semiotic analysis is that, although it
are as follows: may offer a compelling exposition of some aspect of
everyday life, the interpretation may be somewhat
• The sign is something that stands for some- arbitrary. For example, Gottdiener’s association
thing else, such as a yellow traffic light; it has of “Adventureland” with colonialism/imperialism
two components, a signifier and a signified. might not be shared by other analysts. However, the
• The signifier is the thing (here the yellow light) results of a semiotic analysis may be no more arbi-
that points to an underlying meaning. trary than those of any other interpretive approach.
• The signified is the meaning to which the signi- Indeed, it would be surprising not to be struck by a
fier points (“caution: stop if possible”). sense of arbitrariness in semiotics, given its empha-
• The denotative meaning is the manifest or ob- sis on the principle of polysemy.
vious meaning of a signifier and as such indi-
cates its function (here to regulate traffic). Hermeneutics
• A connotative meaning is one that can arise The hermeneutic approach has been used in under-
in conjunction with the denotative meaning standing and interpreting the Bible, but it has also
(for example, “speed up to beat the coming red been employed in the analysis of other texts. It was
light”). influential in the formation of interpretivism as
• Polysemy refers to the notion that signs can be an epistemology (see Chapter 1) and has much in
interpreted in many different ways. common with Weber’s notion of Verstehen. The cen-
tral idea behind hermeneutics is that the analyst of a
Semiotics seeks to uncover the hidden meanings that text must seek its meanings from the perspective of
reside in texts, broadly defined. Consider, by way of its author, considering the social and historical con-
illustration, the curriculum vitae (CV) in academic text within which the text was produced.
life. The typical academic CV contains information on Phillips and Brown’s (1993) hermeneutic study of
matters such as degrees earned; previous and current the corporate image advertisements of a Canadian
posts; administrative responsibilities and experience; synthetic crude oil company was a “formal analy-
teaching experience; research experience; research sis of the structural and conventional aspects of the
grants acquired; and publications. It can be seen as a text” (p. 1563). For them this meant examining the
system of interlocking signifiers denotatively provid- texts of the advertisements in terms of their constitu-
ing a summary of the individual’s experience (its sign ent parts and the writing conventions they reflected.
function). At the connotative level, it is an indication of They also relied on a large database of magazine and
an individual’s value, particularly in connection with newspaper articles relating to the company for ad-
employability. Every CV is capable of being interpreted ditional documentary materials. They showed how
in different ways and is therefore polysemic, but there corporate image advertisements were used to mobi-
is a shared code among academics whereby certain at- lize support for company activities from government
tributes of CVs (such as lists of publications) are seen as and the public, and to ward off the environmental
especially desirable and are therefore less contentious in legislation that threatened them.
terms of their meaning. Applicants for posts know this
and design their CVs to highlight the desired qualities
and downplay others, so that the CV becomes a presen-
Readers and audiences—
tation of self, as Miller and Morgan (1993) suggested. active or passive?
Box 12.6 outlines a semiotic study (Gottdiener, Audience reception is a prominent area of inquiry
1997) in which the “text” was Disneyland. The in media and cultural studies. The key question is
chief strength of semiotics lies in its invitation to whether audiences/readers are active interpreters of

bel65796_ch12_281-313 299 01/13/22 12:33 PM

300 PART III Qualitative Research

Box 12.6 A semiotic Disneyland

and the life they experience at the theme park.
Joshuaraineyphotography/[Link]/GetStock

He identified, through this principle, several sign

systems contrasting the park with the surround-
ing environment: transportation, food, clothing,
shelter, entertainment, social control, econom-
ics, politics, and family. The first of these sign
systems—transportation—revealed a difference
between the Disneyland visitor as pedestrian
(walking with others in a group) and the Los
Angeles resident as passenger (a car is neces-
sary; danger on the congested freeways). A fur-
Visitors see this plaque as they enter Disneyland in ther component of his research entailed analysis
California. How does it support Gottdiener's notion of the connotations of the different “lands” that
of a semiotic Disneyland? make up the park. He suggested that each land
was associated with one or more signifiers of
Gottdiener (1997, pp. 108–115) proposed that capitalism; for example:
Disneyland can be fruitfully explored through • Frontierland: predatory capital, conquering
semiotic analysis. He concluded that its mean- • Adventureland: colonialism/imperialism
ing is based on the contrast between the alien- • Tomorrowland: state capital for space
ated daily lives of nearby Los Angeles residents exploration

what they see or hear. Do they passively accept the arrive at the same interpretations? Do their inter-
meanings that authors or designers infuse into their pretations match those of the original readers or
texts, as in the oil company advertisements just de- audiences? The social researcher is always putting a
scribed, or do they resist those meanings and strive personal “spin” on the texts analyzed. The same is
for independent readings? Do they arrive at a middle true of all social science data: the conclusions derived
point that incorporates both passive and active ele- from questionnaire or ethnographic data reflect the
ments? Much of the research on this issue suggests author’s interpretations, not the complete set of pos-
that audiences frequently come up with readings dif- sible interpretations. The point is that close scrutiny
ferent from those intended by authors (see Fenton, and critical thinking are always required.
Bryman, & Deacon, 1998, for a summary of some of
this research).
Although the idea of the “active audience” has
Two approaches to the study
not gone unchallenged (see, for example, McGuigan, of language
1992), it is so well supported that many of the inter- In this section two approaches to the study of lan-
pretations of texts offered by social scientists have guage are examined: conversation analysis (CA) and
been questioned. This suggests caution in reading discourse analysis (DA). While CA and DA do not
Giulianotti’s (1997) study of “fanzines” or Hier’s exhaust the range of possibilities for studying lan-
(2002) examination of city council minutes to un- guage, they do represent two of the most prominent
derstand the rave issue. Would other social scientists approaches, and each includes both quantitative and

bel65796_ch12_281-313 300 01/13/22 12:33 PM

 12 Content Analysis 301

qualitative aspects. Both have developed a technical situations suggests that CA fits well with another
vocabulary and a set of techniques. This section out- preoccupation among qualitative researchers: a
lines some of their basic elements and draws atten- commitment to naturalism (see Chapter 9).
tion to their contrasting features. Conversation analysts have developed a variety
of ways to study talk. Psathas (1995, p. 1) described
Conversation analysis them as “rigorous, systematic procedures” that can
The roots of CA lie in ethnomethodology, a socio- “provide reproducible results.” Such a framework
logical approach to communication focusing on the brings to mind the procedures used to generate
“practical, common-sense reasoning” that people valid, reliable, and replicable findings in quantita-
use in their everyday lives. It includes notions of tive research. It is not surprising, therefore, that CA
cause and effect (“if I do this, then that will happen”) is sometimes described as having a positivist orien-
and the generalizations that allow people to perform tation. Thus, some features that are broadly in tune
everyday tasks. This process of reasoning and com- with qualitative research (contextual and naturalis-
munication is presented as a way in which social tic analysis without prior theoretical commitments)
order is created. Social order is not seen as a pre-­ are married to traits associated with quantitative
existing force constraining individual action, but as research.
something that is worked at and achieved by people However, the emphasis on context in CA is some-
through interaction. Garfinkel, one of the pioneers what at variance with contextual understanding
of conversation analysis, claimed that the role of as it is normally thought of in qualitative research.
sociology is not to uncover “objective” social facts For CA practitioners, context refers to the specific
as Durkheim suggested, but to see them as an ac- here-and-now context of immediately preceding
complishment, as the eventually taken-for-granted talk, whereas for most qualitative researchers it has
­patterns established through the activities of ordi- a much wider focus, encompassing phenomena such
nary people going about their daily lives (Garfinkel, as the broader culture of the group within which the
1967, p. vii). social action occurs, including their values, beliefs,
Two ideas are particularly central to ethno- and typical modes of behaviour. This is precisely
methodology and find clear expression in CA: the kind of attribution that CA practitioners seek to
indexicality and reflexivity. The former suggests avoid. To import elements that are not specifically
that the meanings of words or utterances, includ- grounded in the here and now of what has just been
ing pauses and sounds, depend on the context in said during a conversation is to risk imposing an un-
which they are used. Reflexivity means that talk derstanding that is not grounded in the participants’
is not a “mere” representation of the social world, immediate concerns. It is no wonder, therefore, that
standing for something else, but is itself a reality. writers like Gubrium and Holstein (1997) treated CA
In these ways, ethnomethodology fits squarely as a separate tradition within qualitative research,
with two aspects of qualitative research: a prefer- whereas Silverman (1993) found it difficult to fit CA
ence for a contextual understanding of action (see into broad descriptions of qualitative research.
Chapter 9) and an ­ontological position associated
with ­constructionism (see Chapter 1). Assumptions of conversation analysis
In the years following its introduction, ethno- Analysts often begin practising conversational anal-
methodological research sought to conduct fine- ysis when they notice something significant about
grained analyses of the sequences of interaction the way a speaker says something, and this recogni-
revealed in conversations recorded in naturally oc- tion generates interest in the function that the turn
curring contexts. As such, CA is a multifaceted ap- of phrase serves. To illustrate, Clayman and Gill
proach, part theory, part method of data acquisition, (2004) referred to the way a child will often begin
part method of analysis. The predilection for the a conversation with an adult by saying “You know
analysis of talk gleaned from naturally occurring what, Daddy [or whoever]?” The question generally

bel65796_ch12_281-313 301 01/13/22 12:33 PM

302 PART III Qualitative Research

produces a “What?” reply that allows the child to means breathing out. The more h’s, the longer
find a slot in a sequence of communications, or to the breath.
inaugurate such a sequence. The use of a question,
(0.8)  A figure in parentheses indicates a
rather than a declarative statement, may reflect the
period of silence, usually measured in tenths
child’s lesser power.
of a second. Thus, (0.8) indicates eight-tenths
Once such a focus has been identified, conversa-
of a second of silence.
tion analysts typically follow certain basic assump-
tions. Heritage (1984; 1987) proposed three: you and knowing   An underline indicates an
emphasis in the speaker’s talk.
• Talk is structured. Talk comprises patterns,
you- you   A hyphen means the speaker was
and participants are implicitly aware of the
interrupted or stopped speaking.
rules that underpin these patterns. As a result,
conversation analysts do not attempt to infer (.)   Indicates a very slight pause.
the motivations of speakers from what they
say, or to ascribe their talk to purely personal The attention to detail in the sequence of talk in
characteristics. Such information is unneces- Box 12.7 is striking, but what is significant in it?
sary, since the conversation analyst is oriented Silverman (1994) drew two main inferences. First,
to the underlying structures that are revealed P initially tries to deflect any suggestion that there
in pauses, emphases, questions followed by an- may be a special reason why she needs an HIV test.
swers, and so on. As a result, the disclosure that she has been engaging
• Talk is forged contextually. Talk must be ana- in potentially risky behaviour is delayed. Second, P’s
lyzed in terms of its context and understood in use of “you” depersonalizes her behaviour. Silverman
terms of the talk that has preceded it. argued that sequences like these show how “people
• Analysis is grounded in data. Conversation receiving HIV counselling skilfully manage their
analysts shun prior theoretical schemes and talk about delicate topics” (p. 75). The hesitations
argue that the characteristics of talk and social are designed by patients to establish that issues like
order must be derived from the data. these are not the subject of normal conversation. The
rather general replies to questions indicate that the
In doing a project based on CA, it is important not speaker is not the kind of person who immediately
to collect too much data. The real work of CA is the launches into a discussion about difficult sexual
painstaking analysis that its underlying theoretical matters with a stranger. As an aside, Silverman sug-
stance requires. It may be that just one or two por- gested that P’s hesitancy and depersonalization are
tions of transcribed text will be sufficient to answer minimal. Others stall more, lie, or totally refuse to
the research questions. answer (p. 76).

Transcription and attention to detail Some basic characteristics of

The transcript in Box 12.7 includes some basic sym- conversations
bols used by conversation analysts: There are recurring features in the way talk is orga-
nized that can be discerned in sequences of conver-
sation. Among them are the following.
We:ll   A colon indicates that the sound that
occurs directly before the colon is prolonged. Turn-taking
More than one colon means further prolonga-
One of the most basic features of everyday conversa-
tion (e.g., Ye: : : :s).
tion is the fact that the participants take turns speak-
.hh   A row of h’s preceded by a dot indicate ing. Only one speaker tends to talk at a time, the
an intake of breath. If no dot is present, it other listening, and turns are taken with minimal

bel65796_ch12_281-313 302 01/13/22 12:33 PM

 12 Content Analysis 303

Box 12.7 Conversation analysis showing a question-and-answer

adjacency pair
Silverman (1984, p. 72) provided the following ex- 4. I mean that you have to think about these
tract from a conversation between an HIV coun- da:ys, and
sellor (C) and a patient (P): 5. I just uh:m felt (0.8) you- you have had sex
with
1. C: Can I just briefly ask why: you thought 6. several people and you just don’t want to go
about having on (.)
2. an HIV test done: 7. not knowing.
3. P:.hh We:ll I mean it’s something that you
have these

gaps between them (Hutchby & Wooffitt, 1998, analyst not just because it invites a response in its
p. 47). Turn-taking is particularly important to con- own right but because compliance with the model
versation analysts because it points to the existence indicates recognition of the way one is supposed to
of shared codes indicating the ends of utterances. respond to the initial phase. In this way “intersubjec-
Without such codes, conversation could not proceed tive understandings” are continuously reinforced.
in an organized manner, with smooth transitions be- This does not mean that the second phase always fol-
tween speakers. lows the first. Failure to respond a­ ppropriately—for
Of course, things sometimes go wrong, as when example, when one answers a question with another
people speak at the same time. Silverman (1993, question—has itself been the focus of attention for
p. 132) described some repair mechanisms for in- conversation analysts.
stances in which turn-taking conventions are not
followed. For example, when a turn transfer does not Preference organization
occur at an appropriate point (such as when some- Some responses are clearly preferred to others. An
one does not respond to a question), the first speaker example is when an invitation or a request is prof-
may speak again, perhaps reinforcing the need for fered: acceptance is the preferred response and re-
the other person to speak (possibly by rephrasing the fusal the non-preferred response. To illustrate, Potter
question). (1996, p. 59) contrasted a sequence in which an offer
The crucial point to note about such repair mecha- is met with a straightforward acceptance—“thank
nisms is that they allow the rules of turn-taking to be you”—with a sequence (reproduced in Box 12.8) in
maintained even after the rules have been breached. which an invitation is declined.
According to Potter, A’s response is fairly typical
Adjacency pairs of rejections, which have their own standard fea-
One feature of turn-taking is the tendency for an ex- tures. For example, A delays the start of his or her
change to have two linked phases known as an response and fills it with “hehh.” In addition, the
­adjacency pair; for instance, a question followed by rejection is “softened” by A’s saying “I don’t think I
an answer, as in Box 12.7; an invitation followed by a can make it” (which suggests that acceptance is still
response (accept/decline); or a greeting followed by possible) and then offering an explanation for his or
a returned greeting. The first phase assumes that the her failure to provide the preferred response. The key
other part of the adjacency pair will be ­forthcoming— point is that the participants recognize the prefer-
for example, that an invitation will get a response. ence structure of this kind of adjacency pairing and
The second phase is of interest to the conversation that this recognition affects the response (hesitancy,

bel65796_ch12_281-313 303 01/13/22 12:33 PM

304 PART III Qualitative Research

social interaction, and has contributed to our under-

Box 12.8 Conversation analysis standing of how social order is created—one of the
in action: A non- classic concerns of social theory.
preferred response
Discourse analysis
1. B: Uh if you’d care to come over and visit a
little while Discourse analysis incorporates aspects of CA, but
2. this morning I’ll give you a cup of coffee. is more flexible and can be applied to forms of com-
3. A: hehh munication other than talk. It puts less emphasis on
4. Well naturally occurring talk, so the language used in
5. that’s awfully sweet of you, research interviews can also be a legitimate subject
6. I don’t think I can make it this morning. hh of analysis. According to Potter (1997), DA “empha-
uhm sizes the way versions of the world, of society, events
7. I’m running an ad in the paper and-and uh and inner psychological worlds are produced in dis-
I have to course” (p. 146). For continental philosophers such
8. stay near the phone. (Atkinson & Drew,
as Foucault (1926–84), “discourse” refers to the way
1979, p. 58; quoted in Potter, 1996, p. 59)
a particular set of linguistic categories relating to an
object frames people’s understanding of that object.
acknowledgment of the invitation, explanation of For example, a certain discourse concerning
failure). Acceptance of an invitation does not have to mental illness can produce particular conceptions of
be justified, whereas refusal generally does, in order what mentally ill persons are like, the nature of their
to protect the relationship between the two par- illness, how they should be treated, and who is legit-
ties from the harm that the non-preferred response imately entitled to treat them. In this way discourse
might otherwise do. serves to justify the power of those who provide such
Conversational patterns can be a good subject for treatment as well as the treatment regimens they pre-
student research to see if different groups, such as scribe. This discourse is much more than language: it
men and women, or students and professors, exhibit is part of the social world of mental illness.
varying patterns. (Keep in mind that status hierar- Several different approaches fall under the DA
chies can sometimes determine who the interrupters heading (Potter, 1997). The version discussed here
are, who defers to the interrupters, etc.) Your class is one associated with writers such as Potter (1997),
could go to an eating area on campus to collect data, Potter and Wetherell (1994), and Billig (1991). This
each student listening to different conversations. To version of DA exhibits two distinctive features at the
catch body movements, video recordings can be used level of epistemology and ontology (Potter, 1997):
(Heath, 1997), so long as ethical protocols have been
followed. • It is generally anti-realist in that it denies the
existence of an external reality awaiting a de-
A final note on conversation analysis finitive portrayal by the researcher. It therefore
CA considers it illegitimate to invoke cultural fac- disavows the notion that researchers can arrive
tors when analyzing conversations. However, some at accurate and objective accounts of the social
researchers see that principle as unnecessarily re- world.
strictive. In many exchanges, much of the talk is • It is constructionist in that it gives priority to
informed by the participants’ shared knowledge of the version of reality propounded by mem-
contexts, including cultural contexts, yet the pro- bers of the social setting being investigated.
hibition against cultural arguments limits CA to More specifically, it recognizes that discourse
research questions that pertain to talk itself. On analysis entails a selection from many possi-
the other hand, CA reduces the risk of making un- ble interpretations of a given situation. For
warranted speculations about what is happening in example, is a person who speaks to herself

bel65796_ch12_281-313 304 01/13/22 12:33 PM

 12 Content Analysis 305

dangerous or harmless? Is she more in need of suggested, what is said is always a way of not saying
treatment or social support? In the process of something else. Either way, discourse can be seen
answering such questions, a particular reality as providing a solution to a problem (Widdicombe,
is constructed. 1993).
DA shares with CA a preference for locating con-
Discourse is not simply a neutral device for im- textual understanding in terms of the situational
parting meaning. After all, there are many things specifics of talk. As Potter (1997, p. 158) put it, dis-
that people seek to accomplish when they talk or course analysts prefer “to see things as things that
write. DA is concerned with the strategies they are worked up, attended to, and made relevant in in-
employ in order to create different kinds of effect. teraction rather than being external determinants.”
This is illustrated in three basic questions that DA However, DA practitioners are less wedded to this
asks: principle than are conversation analysts.
Discourse analysts resist the idea of codifying
• What is this discourse doing? their procedures and argue that such codification
• How is it constructed to make this happen? is probably impossible. Potter (1997) described his
• What resources are available to perform this work as “a craft skill, more like bike riding or chicken
activity? (Potter, 2004, p. 609) sexing than following [a] recipe” (pp. 147–148). Gill
(2000) suggested adopting a posture of “skeptical
This action orientation is revealed in a study of the reading.” This means searching for a purpose lurk-
first few moments of telephone calls to the helpline ing behind the way things are said or presented. Gill
of the UK’s National Society for the Prevention of proposed that DA can be usefully thought of in terms
Cruelty to Children. Through analysis of these call of four main themes, outlined in Box 12.9.
openings, Potter and Hepburn (2004) showed that a
variety of actions are performed during the first few Producing facts
moments: In this section, the emphasis is on the strategies
employed to convey allegedly factual knowledge.
• The callers spell out the details of their Among them is quantification rhetoric: the various
concerns. ways in which statements involving numbers or
• The callers seek to establish, in collabora- quantities can be made to either support or refute
tion with the “experts” on the line (the child arguments. The interest in this issue lies partly in
protection officers) that the incidents they the importance of quantification in everyday life, but
are reporting do in fact require some kind of also in the tendency of many social scientists to use
intervention. this strategy themselves (John, 1992). In their anal-
• The callers convey the idea that the activity ysis of data such as the portions of transcript cited
they are describing is serious and that they are in Box 12.10, Potter, Wetherell, & Chitty (1991) and
concerned about it, but not so concerned that Potter and Wetherell (1994) show how these sorts of
the police should be contacted immediately. devices are used.
• The child protection officers show a willing-
ness to treat the reports as serious, without Using variation as a lever
making judgments as to their actual truth or The authors drew attention to the phrase “1 per cent
seriousness. of a quarter of a million” because it incorporates two
different types of quantitative expression: a relative
Analysis of these brief moments of conversation expression (a percentage) and an absolute frequency
shows that the flow of discourse achieves a number (quarter of a million). The difference is important,
of objectives for both parties and in that sense con- because it allowed the program makers to empha-
stitutes action. On the other hand, as Gill (1996) size the very low cure rate (just 1 per cent) compared

bel65796_ch12_281-313 305 01/13/22 12:33 PM

306 PART III Qualitative Research

Box 12.9 Four themes in discourse analysis

Gill (2000) drew attention to four prominent 3. Discourse is a form of action. As Gill put it,
themes in DA: language is considered “a practice in its own
right” p. 175). Language is a way of accom-
1. Discourse is a topic. In other words, discourse
plishing acts, such as attributing blame, pre-
is a focus of inquiry itself and not just a means
senting oneself in a particular way, or getting
of gaining access to the aspects of social re-
an argument across. Moreover, a person’s
ality that lie behind it. This idea contrasts with
discourse is affected by the situation he or she
the view traditionally taken by research in-
is in. For instance, the reasons given for want-
terviewers, in which language simply reveals
ing a job may vary depending on whether
what people think, or serves as a mechanism
the job seeker is talking to a prospective em-
for learning about their behaviour and the
ployer, a family member, or a friend.
reasons for it.
4. Discourse is rhetorically organized. DA practi-
2. Language is constructive. Discourse is a way
tioners recognize that discourse is concerned
of constituting a particular view of social re-
with “establishing one version of the world
ality. Moreover, in order to create that view,
in the face of competing versions” (p. 176).
choices have to be made regarding the most
In other words, the version of events that
appropriate way of presenting it, and these
a person presents is intended to persuade
choices reflect the disposition of the person
others.
concerned.

with the very large number of cases diagnosed each they were looking for ways to present a convincing
year. Instead, they could have pointed to the abso- argument to show that cancer remains largely in-
lute number of people who are cured (2500), but the curable in spite of the money spent on it. The pro-
impact would have been less dramatic. It’s also worth gram makers very consciously devised the strategy
noting that the phrase “a quarter of a million” not outlined in the discussion of “using variation as a
only exaggerates the number of new cases (adding lever” of playing down the numerical significance of
7000 to the actual 243,000) but also sounds signifi- the cancers that are amenable to treatment. Another
cantly larger. element of their strategy was to employ a tactic
that Potter et al. (1991) called a “preformulation,”
Reading the detail the discounting of a possible counterargument, as
Discourse analysts, like conversation analysts, attend when the commentary says: “When faced with this
to the details of discourse. For example, Potter and challenge the first thing the charities point to is the
Wetherell (1994) suggested that the description of small number of cancers which are now effectively
the three “curable cancers” as “amongst the rarest curable.”
cancers” was deployed to make the point that these
are atypical cancers, and that it would be unwise to Critical discourse analysis
generalize from them to all cancers. There may be important political conflicts and social
justice issues underlying the texts examined by dis-
Looking for rhetorical detail course analysts. For example, the makers of the tele-
Attention to rhetorical detail entails sensitivity to the vision program analyzed by Potter and colleagues in
way arguments are constructed. During the editing Box 12.10 infer that despite lavish funding, cancer
of the film, the program makers’ discourse suggested researchers have had limited success in producing

bel65796_ch12_281-313 306 01/13/22 12:33 PM

 12 Content Analysis 307

Box 12.10 Discourse analysis in action: Producing facts through

quantification rhetoric
The study of the television program Cancer: Your who had cast doubt on whether their research,
Money or Your Life (Potter et al., 1991; Potter much of it funded by charities, resulted in suc-
& Wetherell, 1994) used a variety of sources, cessful treatment:
including
Commentary: The message from these sci-
• a video recording of the program; entists is clear—exactly like the public—they
• the observations of one of the people making hope their basic research will lead to cures
the program (who acted as a participant ob- in the future—although at the moment
server while it was being made);
they can’t say how this will happen. In the
• drafts of the script;
meantime, their aim is to increase scientific
• recordings of editing sessions;
knowledge on a broad front and they’re
• entire interviews (of people such as cancer
research specialists and heads of charities) certainly achieving this. But do their re-
conducted for the program; and sults justify them getting so much of the
• research interviews with some of the latter money that has been given to help fight
people as well as some of the people involved cancer? When faced with this challenge the
in making the program. first thing the charities point to is the small
number of cancers which are now effec-
With regard to the coding process, the authors
tively curable.
reported that they
[on screen: Dr. Nigel Kemp Cancer Research
made a list of about a dozen keywords
Campaign]
and phrases that related to the sequence—
percentage, cure rates, death rates, 1 per Kemp: The outlook for individuals suffering
cent, etc.—and then ran through each from a number of types of cancer has been
of the interview and interaction files, totally revolutionized. I mean for example—
looking for them with a standard word-­ children suffering from acute leukemia—in
processor. . . . Whenever we got a “hit” we the old days if they lived six months they
would read the surrounding text to see if were lucky—now more than half the
it had relevance to our target sequence. children with it are cured. And the same
When it did we would copy it across to an applies to a number of other cancers—
already opened coding file . . . noting the Hodgkin’s Disease in young people and
transcript page numbers at the same time. testicular tumours in young men. (Potter &
If we were not sure if the sequence was Wetherell, 1994, pp. 52–53)
relevant we copied it anyway, for, unlike
At this point a table showing the annual incidence
the sorts of coding that take place in tra-
of 34 types of cancer begins to scroll on the
ditional content analysis, the coding is not
screen. The total incidence is 243,000, and the
the analysis itself but a preliminary to make
individual incidences range from placenta (20)
the task of analysis manageable. (Potter &
to lung (41,400). The three forms of cancer men-
Wetherell, 1994, p. 52)
tioned by Kemp and their levels of incidence are
Following is a sequence used in the research. highlighted in yellow: childhood leukemia (350),
It occurred roughly halfway through the pro- testis (1000), and Hodgkin’s disease (1400). The
gram, following interviews with cancer scientists program continues while the table is scrolling.

(Continued)

bel65796_ch12_281-313 307 01/13/22 12:33 PM

308 PART III Qualitative Research

Commentary: But those three curable sometimes—the outlook is very bad and ob-
types are amongst the rarest cancers—they viously one is frustrated by the s[low], rela-
represent around 1 per cent of a quarter of tively slow rate of progress on the one hand
a million cases of cancers diagnosed each but equally I think there are a lot of real op-
year. Most deaths are caused by a small portunities and positive signs that advances
number of very common cancers. can be made—even in the more intractable
cancers. (Potter & Wetherell, 1994, p. 53)
Kemp: We are well aware of the fact that
[pause] once people develop lung cancer Source: Excerpts from Analyzing Qualitative Data © 1994 Routledge.
or stomach cancer or cancer of the bowel Reproduced by permission of Taylor & Francis Books UK.

cures, which one could take to mean that scarce coverage was “reason” versus “emotion.” Press cover-
societal resources are being misallocated. While age in the more recent era depicted a non-­Indigenous
misallocation of funds is certainly a political issue, judge presiding over an Indigenous land claims case
notions of power and exploitation are not confronted as someone who would “rule” and offer “insight” by
directly or in any depth in their research. making his “decision” using “common sense” and
Those who conduct critical discourse analysis, a “firm” interpretation of “the rule of law” (p. 217).
by contrast, are much more explicit in exposing the Indigenous individuals, however, on hearing of
political nature of the texts they examine. The effect his ruling, are described as dealing with the death
of power hierarchies, structural inequalities, and his- of their “dream,” and as people who “argue,” cry
torical political struggles are among the issues raised “tears,” “vow to fight,” and use tactics of “confron-
by these researchers (Blommaert & Bulcaen, 2000). tation” (p. 217).
They also bring to their work a commitment to social Similarly, G. Johnston (2020) has shown how
change and the empowerment of the oppressed. Canadian mainstream media accounts of opioid
Harding, for instance, has examined how dis- overdose deaths often portray White youth as “in-
course has been used to suppress the aspirations of nocent, normal, and victimized” while Indigenous
Indigenous peoples in Canada since the 1860s. In people are characterized as “addicted and hopeless”
one study (2006), he compared Canadian newspa- (p. 128). She also found that mainstream reporting
per coverage of Indigenous peoples’ activities in the on White overdose deaths was nearly 30 times more
mid-nineteenth century with coverage from the late common than media mentions of Indigenous deaths,
twentieth century. He concluded that although in- even though the actual rates of Indigenous fatal and
stances of overt racism in Canadian papers declined non-fatal opioid overdoses are much higher. She
in this period, the press continued to serve domi- argues that differences in media coverage are likely
nant interests by portraying Indigenous people as a to have a significant impact on public perception and
threat to White power and dominance. One way in understanding: “the choice not to depict Indigenous
which this occurred was through the framing of the overdose deaths continues to allow many Canadians
text with a binary opposition of “us” versus “them.” to remain ignorant of the issues facing these com-
An article published in the 1860s stated, “To allow munities” (p. 141).
these [Indigenous] people all the privileges of others
[with regard to homesteading] would be to throw the Overview
whole Colony [of British Columbia] into confusion. DA draws on insights from CA, particularly when
Just imagine our 80,000 Indians . . . being allowed analyzing strings of talk. The CA injunction to focus
to locate land wherever they please” (Harding, 2006, on the talk itself and the ways in which intersubjec-
p. 213). Another binary opposition evident in press tive meaning is arrived at in sequences of talk is also

bel65796_ch12_281-313 308 01/13/22 12:33 PM

 12 Content Analysis 309

Methods in Motion Applications to Canadian Society

Interpreting the Highway of Tears
Morton (2016) conducted a critical discourse needed to travel independently by car or public
analysis of two billboards placed on the “Highway transport. Also missing from the signs, she points
of Tears,” a stretch of road (Highway 16) in cen- out, is any indication of why Indigenous women
tral British Columbia that is infamous for the large need to travel the highway in the first place,
number of Indigenous women who have disap- which typically includes gaining access to social
peared while hitchhiking on it. The first billboard services such as women’s shelters, health care,
was co-sponsored by the Province of British and education. Morton asserts that what is
Columbia, the regional district of Kitimat-Stikine, and is not stated on the billboards are equally
and the Gitxkan First Nation. The second was significant.
co-sponsored by the province and the Aboriginal A key issue Morton addresses in her analysis
Women’s Action Network. is that the billboards in no way try to portray the
Morton analyzed the images, text, and other social and political position Indigenous women
symbols on the billboards. The first road sign in- find themselves in, and there is no attempt to set
cludes the words “HITCHHIKING/Is It Worth the the tragedy of the Highway of Tears in its appro-
Risk?” in large letters, with the statement, “ain’t priate socio-historical context. Morton sees the
worth the risk sister” in smaller, less distinct print. “intersection of race, gender, mobility and vio-
The second billboard reads in part, “GIRLS DON’T lence” (2016, p. 300), all in the larger setting of
HITCHHIKE/on the/Highway of Tears.” Morton European conquest and colonialism, as a more
observes that one thing the billboards do is iden- realistic interpretive frame to use when trying to
tify hitchhiking as a form of “contentious mobil- remedy what happens on the Highway of Tears.
ity,” one which any prudent, responsible person But rather than taking a sociologically informed
would avoid. But nowhere on the signs are there approach to the issue, the billboards imply that
suggestions for alternative, legitimate forms of the disappearance of Indigenous women along
transportation, nor is there any acknowledgment this route is merely the result of reckless be-
that Indigenous women often lack the resources haviour or misadventure.

incorporated into DA. However, DA practitioners or sizeable slice of social life but usually a tiny frag-
break with their CA counterparts when they specu- ment” (Wetherell, 1998, p. 402).
late on motives, which is quite evident in the critical Writing from a critical realist position, Reed
discourse analysis discussed above. It is precisely this (2000) argued that discourses should be examined
practice of speculating on things that are not directly in relation to social structures such as the power re-
discernible in the sequences of talk that conversation lationships that create the discourses. This approach,
analysts reject, as when Schegloff (1997, p. 183) wrote which is exemplified in the work of critical discourse
of DA: “Discourse is too often made subservient to analysts, shows how discourses work through ex-
contexts not of its participants’ making, but of its isting structures, and expands discourse from a
analysts’ insistence.” For their part, discourse ana- self-referential concept in which nothing of signifi-
lysts object to the restrictions that CA imposes, be- cance exists outside itself into a “generative mecha-
cause those restrictions cause conversation analysts nism.” That position is perhaps closer to the classic
to “rarely raise their eyes from the next turn in the concerns of the social sciences than the anti-realist
conversation, [which] is not an entire conversation stance taken by some CA and DA researchers.

bel65796_ch12_281-313 309 01/13/22 12:33 PM

310 PART III Qualitative Research

Advantages of content recommended assessing documents in terms

of criteria such as authenticity (the document
analysis is what it purports to be); credibility (there are
Content analysis has several advantages: no grounds for thinking that its contents have
been distorted in any way); and representative-
• In its most quantitative form, it is a very trans- ness (the specific documents examined are rep-
parent research method, making replication resentative of all possible relevant documents).
relatively easy. It is because of this transpar- • Even in quantitative content analysis, it’s
ency that content analysis is often said to be an almost impossible to devise coding manuals
“objective” method of analysis. that do not require some interpretation by the
• It allows for longitudinal analysis. For exam- coder. It seems unlikely that there is ever a per-
ple, a crime study can be expanded to examine fect correspondence of interpretation between
changes in newspaper crime reporting over different coders.
two different time periods. • Problems of interpretation are especially likely
• Content analysis is an unobtrusive, non-­ to arise when the aim is to identify latent
reactive method. Newspaper articles and tele- meanings (as opposed to the more readily ap-
vision scripts are generally not written with parent manifest meanings). For example, in
the expectation that a content analysis might searching for traditional markers of masculin-
one day be carried out on them. On the other ity and femininity (see Box 12.5), the potential
hand, if the object of analysis is an interview for conflicting interpretations is magnified. A
transcript or ethnography, its content may related distinction is sometimes made between
have been at least partly influenced by antici- mechanical analysis (in particular, counting
pation of such scrutiny and thus may contain certain words) and analysis that emphasizes
some reactive error. themes in the text, which entails a higher level
• Content analysis is a highly flexible method, of abstraction and a correspondingly greater
applicable to several different kinds of un- risk of measurement invalidity.
structured information. Although it is pri- • Content analysis may not be helpful in answer-
marily applied to mass media outputs, it has a ing “why?” questions. For example, Jagger’s
much broader applicability. (1998) analysis of dating advertisements
• Content analysis can permit the study of social clearly showed that both men and women put
groups that are difficult to access. For exam- less emphasis on the “attractiveness, shape and
ple, much of what social scientists know about size” they desired in a partner than they did
the social backgrounds of elite groups, such as on their own appearance (p. 807; see Box 12.5).
company directors and top military personnel, But further research would be needed to deter-
comes from content analyses of publications mine why this was the case. Similarly, Fenton
such as Who’s Who and the business pages of et al. (1998) found that sociology was only the
newspapers. fourth-most common discipline explicitly re-
ferred to (after psychology, economics, and
Disadvantages of content social policy) when social science research
was reported in the mass media, but it was
analysis by far the most frequently inferred discipline
Like all research techniques, content analysis suffers (not directly mentioned, but apparently the
from certain limitations: discipline under discussion). Again, however,
the reasons behind this phenomenon could
• A content analysis can only be as good as the only be a matter for speculation (Fenton et al.,
documents it explores. Recall that Scott (1990) 1998).

bel65796_ch12_281-313 310 01/13/22 12:33 PM

 12 Content Analysis 311

• Some content analysis studies are accused of (1998), for instance, placed her findings on
being atheoretical, and it’s easy to see why. The dating advertisements in the context of cur-
emphasis on measurement in content analysis rent ideas about consumerism and the body.
can easily lead researchers to focus on what is And Harding’s (2006; 2010) content analysis
measurable rather than what is theoretically of media portrayals of Canadian Indigenous
significant or important. However, content people was underpinned by theoretical ideas
analysis is not necessarily atheoretical. Jagger pertaining to racism and ethnic conflict.

Key Points
• Depending on how it is done, content analy- • Semiotics and hermeneutics are qualitative ap-
sis can be mainly quantitative or qualitative in proaches to content analysis.
nature. • There is disagreement over whether readers of
• It’s important to have clear research questions documents are active or passive consumers of
and to be explicit about what is to be analyzed. the messages they receive.
• Documents (as the term is used here) provide • Both conversation analysis (CA) and discourse
many different kinds of information and can take analysis (DA) approaches make language itself
the form of personal documents, official docu- a focus of research.
ments from state or private sources, and output • CA is a systematic approach based on the princi-
from the mass media. ple that talk is structured by certain rules.
• Such materials can be used for both quantitative • Practitioners of CA avoid making inferences
and qualitative research. about talk that are not grounded in its immediate
• Documents may be in printed, visual, digital, or context.
any other retrievable format. • DA shares many features with CA but comes in
• Many researchers believe that almost anything several different versions and can be applied to
can be “read” as a “text.” a wider variety of phenomena.
• Criteria for evaluating the quality of documents in- • DA sees discourse as a means of conveying
clude authenticity, credibility, representativeness, meaning and generally relates meaning in talk
and meaning; some may be more relevant than to contextual factors.
others, depending on the nature of the document. • Critical discourse analysis situates texts in the
• The design of a coding schedule and preparation context of larger social structures such as power
of a coding manual are crucial steps in content hierarchies and systems of inequality.
analysis. In quantitative content analysis, this is • As with all research methods, there are advan-
done before the materials are examined. tages and disadvantages to content analysis.
• Content analysis can be particularly challenging
when it is used to search for latent meanings and
themes.

Questions for Review (R) and Creative Application (A)

R1 What types of material may serve as “docu- Personal documents
ments” in social research? R2 List the different kinds of personal documents
A1 Should there be a fifth criterion for assessing and rank them in terms of Scott’s four criteria.
documents? If so, what should it be?

bel65796_ch12_281-313 311 01/13/22 12:33 PM

312 PART III Qualitative Research

A2 Go online and take a look at the Mackenzie King Coding

diaries. What themes would you want to ana- R7 What potential pitfalls need to be guarded
lyze if you were to do a content analysis of this against when devising coding schedules and
information? manuals?
A7 You are planning to do a content analysis of
Official government documents daytime TV commercials. What themes would
R3 How might official government documents be you examine? Explain why your themes are
biased? How do these documents fare in terms important.
of Scott’s criteria?
A3 You’ve decided to do a content analysis of Semiotics
Question Period in the House of Commons, R8 What is the difference between denotative and
using Hansard as a source. Pick an issue to be connotative meaning?
­analyzed, and explain what words or expres- A8 Why are “used” cars sometimes referred to as
sions you would search for to provide informa- “pre-owned”?
tion on it.
Conversation analysis
Official documents deriving from pri- R9 Explain what each of the following means:
vate sources “turn-taking,” “adjacency pairs,” “preference
R4 What kinds of documents may be considered organization,” “repair mechanism.” How do they
private official documents? How do these docu- relate to the production of social order?
ments fare in terms of Scott’s criteria? A9 Evaluate the argument that CA should examine
A4 You’ve chosen to analyze the private docu- a participant’s motives.
ments of a major oil company, but the company
will only provide you with materials that have Discourse analysis
been cleared by its public relations department. R10 In what ways may DA be anti-realist and
Prepare a summary of what you do not expect to constructionist?
find in the documents. A10 You’ve decided to pursue a career in journalism
because you believe that discourse is a form of
Mass media outputs action. Explain.
R5 How might mass media outputs be biased? How
do they fare in terms of Scott’s criteria? Critical discourse analysis
A5 You will be comparing the content provided by R11 Why do critical discourse analysts make refer-
the print version of a major national newspaper ence to power structures, systems of inequality,
in Canada with what it offers in its online version. and other political and sociological factors in
How might the content differ? How might it be their work?
similar? A11 Should all researchers doing CA and DA make
reference to macro political and social factors
Manifest and latent content in their studies? Explain.
R6 What is the difference between manifest and
latent content? What are the implications of that Advantages and disadvantages of
distinction for content analysis? content analysis
A6 Record the first two minutes of any TV news pro- R12 Explain how content analysis can be used as a
gram. Present a summary of its manifest con- form of longitudinal analysis.
tent, and then an account of its latent content. A12 How might the need for coders to interpret
meaning undermine latent content analysis?

Interactive Classroom Activities

1. The class is divided up into small groups. Each reported in newspapers—for example, a school
group selects a topic of interest that has been shooting, an act of terrorism. Each group then:

bel65796_ch12_281-313 312 01/13/22 12:33 PM

 12 Content Analysis 313

a. searches the Internet for newspaper accounts of a movie or television drama that it considers to
of the event, and selects three for analysis; be worthy of analysis. Each group then:
b. decides on at least three themes in the ac- a. decides on at least three themes in the clip
counts that are noteworthy, as well as other that are noteworthy, as well as other aspects
aspects of the stories that seem relevant—for of it that seem relevant—for example, under-
example, number of victims, type of weapons lying racist assumptions, gender stereotypes,
used; jingoism;
c. produces both a coding schedule and a b. produces both a coding schedule and a
coding manual to do a content analysis of the coding manual to do a content analysis of the
three accounts; clip; the former should be created in such a
d. creates a numerical summary of the three way that the number of instances of a partic-
articles based on the data gathered in the ular theme can be recorded;
coding schedule; and c. creates a numerical summary of the clip
e. discusses salient interpretations that arose based on the data gathered in the coding
out of the content analysis, and if possi- schedule; and
ble, relates them to hypotheses or theories d. discusses salient interpretations that arose
(either original or already established in the out of the content analysis, and if possi-
literature) that may follow from the content ble, relates them to hypotheses or theories
analysis. (either original or already established in the
The class as a whole then reconvenes. Each literature) that may follow from the content
group presents the results of its analysis to the analysis.
class, which is encouraged to ask questions and The class as a whole then reconvenes. Each
provide critical commentary. group presents the results of its analysis to the
2. The class is divided up into small groups. Each class, which is encouraged to ask questions and
group searches the Internet for a five-minute clip provide critical commentary.

Relevant Websites
Information on ethnomethodology and conversa- An overview of the software available to do content
tion analysis can be found at the EMCA wiki page. analysis can be found at the Audience Dialogue
[Link] website.
[Link]/[Link]
The journal Discourse & Society is an international
journal that features articles on discourse analysis The Discourse and Rhetoric Group at
(you may have to go through your institution’s li- Loughborough University provides a tutorial on
brary sever to access the material). conversation analysis.
[Link] [Link]

A YouTube video describes how to do qualitative (Websites accessed on 4 September 2021)

content analysis.
[Link]/watch?v=BhQX-zKultw

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch12_281-313 313 01/13/22 12:33 PM

 13 Qualitative Data
Analysis
Chapter Overview
Analyzing qualitative data from interviews, participant observation, content analysis, or any other
qualitative source is never straightforward. For one thing, such data typically take the form of a large body
of unstructured textual material. Moreover, whereas quantitative research has clear-cut rules for data
analysis, this is not the case in qualitative research. This chapter will examine some general approaches
to qualitative data management. Computer software is often used by qualitative researchers to assist with
coding: the main feature of qualitative data analysis. This software, often referred to as computer-assisted
qualitative data analysis software (CAQDAS), eliminates many, if not most, of the clerical tasks associated
with the manual coding and retrieving of data. Although there is no industry leader among the different
programs available for qualitative data analysis, this chapter introduces one of the most popular programs:
QSR NVivo (discussed in detail in the Appendix).
This chapter will explore:
• grounded theory as a general strategy of qualitative data analysis; its main features, processes, and
outcomes are presented along with some criticisms;
• coding as a key process in grounded theory and in qualitative data analysis more generally; there is
an extended discussion of what it entails and some of its limitations;
• some of the debates about the desirability of using CAQDAS;
• the criticism that coding tends to fragment data; and
• the idea of narrative analysis, an approach that reduces that fragmentation.

▲ PrathanChorruangsak/iStockphoto

bel65796_ch13_314-332 314 01/13/22 12:33 PM

 13 Qualitative Data Analysis 315

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Describe the general strategies of qualitative analysis, including analytic induction and, in particular,
grounded theory.
2. Discuss the basic operations in qualitative data analysis.
3. Detail how coding fragments qualitative data and how computer software aids in the coding process.
4. Identify the problems inherent in coding qualitative data.
5. Explain how narrative analysis is used by qualitative researchers.

Einstein once famously said that the most incom- the “concepts” and “categories” that qualitative re-
prehensible thing about the universe is that it is searchers look for as they navigate through vast
comprehensible. What is it that allows us to under- seas of data are simply patterns. This suggests
stand how the universe works? The simple fact that that the social world of meanings, perceptions, and
it shows certain regularities or patterns. Science, feelings is not chaotic, but has a coherence to it that
including social science, is all about finding those we can know and describe if we use the right meth-
patterns and then explaining them. odology. This chapter explores how qualitative
Qualitative social research may seem a long researchers make sense of the data they gather, in-
way away from Einsteinian physics, but in essence cluding how some of them use it to generate theory.

Introduction but at least there are unambiguous rules for handling

such data. Analytic procedures for qualitative data
Making sense of the words contained in field notes, analysis have not reached the same degree of codifica-
interview transcripts, or other documents presents tion, and many feel that would be undesirable anyway
some challenges not seen in numerical analysis, but (cf. Bryman & Burgess, 1994). Nevertheless, there are
the qualitative research process provides oppor- some broad guidelines (Okely, 1994) which we will
tunities to pursue emergent lines of thought and explore. This chapter has two main sections:
interpretation that would be largely impossible in
quantitative studies. One challenge of qualitative • General strategies of qualitative data analysis:
research is that it rapidly generates a large, cumber- analytic induction and grounded theory
some database. Miles (1979) described qualitative • Basic operations in qualitative data analysis:
data as an “attractive nuisance”—attractive because coding and narrative analysis, the latter dif-
of their richness and the insight that may be gained fering in style from both grounded theory and
from them, but also a nuisance because that very the secondary analysis of qualitative data
richness can create analytical difficulties that are not
always acknowledged.
Finding a path through a thicket of prose is not General strategies of
easy, especially for researchers confronting such data qualitative data analysis
for the first time. “What do I do with it now?” is a
common refrain. In large part this is because there are Analytic induction
few well-established and widely accepted rules for the One difference between qualitative and quantitative
analysis of qualitative data. Learning the techniques data analysis is that, with the latter, analysis occurs
of quantitative data analysis can be challenging too, after all the data have been collected. Qualitative

bel65796_ch13_314-332 315 01/13/22 12:33 PM

316 PART III Qualitative Research

analysis, on the other hand, is often iterative (as or recursive approach in which data collection and
noted in Chapter 9), meaning that analysis takes analysis proceed in tandem, repeatedly referring
place after some of the data have been collected and back to each other.
then the implications of that analysis shape further Grounded theory is by far the most widely used
data collection. framework for analyzing qualitative data. Yet there
Analytic induction is an example of a type of is considerable controversy about what it entails
qualitative research that uses an iterative process. It (Charmaz, 2000). For example, it is vague on the dif-
begins with a rough definition of a research ques- ference between concepts and categories. As we will
tion, proceeds to a hypothetical answer, and then see later in this chapter, it seems that the former term
moves to data collection. What makes it unique is is increasingly being replaced by the latter, and this
that analytic induction seeks universal explanations inconsistent use of key terms is not helpful either to
of phenomena that permit no exception. If a case in- practitioners of the craft or to people trying to un-
consistent with the hypothesis is encountered, the derstand the overall process.
analyst either redefines the hypothesis to exclude Against such a background, defining grounded
the deviant or negative case or reformulates the hy- theory is not easy. As well, grounded theory cannot
pothesis and proceeds with further data collection. be described here in all its facets; we will simply out-
With each new deviant case they find, analysts must line its main features.
choose again between reformulation and redefini-
tion. Data collection continues until no new incon- The basic features of grounded theory
sistent piece of evidence is found. It is, in effect, a Some of the basic features of grounded theory have
special case of grounded theory, one that holds 100 been introduced in earlier chapters:
per cent of the time. But it is also rare, as the re-
quirement to be able to explain all cases means that • Coding: the key process in grounded theory,
the explanation may become too broad to be useful. whereby data are broken down into compo-
While analytic induction is an extremely rigor- nent parts and given names. It begins soon
ous method of analysis, it is not favoured by many after the initial collection of data. According
current qualitative researchers. In Canada it has to Charmaz (2000): “Unlike quantitative re-
been used by Whitehead and Carpenter (1999) in a search that requires data to fit into precon-
qualitative study on unsafe sexual behaviour in the ceived standardized codes, the researcher’s
military. They found that the greater the social and interpretations of data shape his or her emer-
cultural distance between sex partners, the greater gent codes in grounded theory” (p. 515, em-
the likelihood of condom use. phasis in original).
One further problem with analytic induction • Constant comparison: continuous comparison
is that (unlike grounded theory) it does not pro- of new and existing data within a particular
vide useful guidelines on the number of cases re- concept or category. Glaser and Strauss (1967)
quired before the absence of negative cases can advised writing a memo (see page 320) on each
be assumed and the validity of the hypothetical concept/category after a few phenomena have
explanation (whether reformulated or not) can be been coded. Comparison also entails sensitiv-
confirmed. ity to differences between emerging concepts/
categories.
Grounded theory • Theoretical saturation: the point at which
Grounded theory is defined as “theory that was there is nothing to be gained by further re-
derived from data, systematically gathered and an- viewing of old data or collection of new in-
alyzed through the research process” (Strauss & formation to see how it fits with emerging
Corbin, 1998, p. 12). Its two central features are its concepts or categories; new data are no longer
development of theory out of data, and an iterative illuminating.

bel65796_ch13_314-332 316 01/13/22 12:33 PM

 13 Qualitative Data Analysis 317

Coding in grounded theory to other categories, validating those relation-

Coding in grounded theory entails reviewing tran- ships, and filling in categories that need fur-
scripts and/or field notes and giving labels (names) ther refinement and development” (p. 116).
to items that seem to share a similar theme, to be of A core category is the focus around which
potential theoretical significance, and/or to be par- other categories are integrated—what Strauss
ticularly salient within the social worlds under inves- and Corbin called the storyline that frames the
tigation. As Charmaz (1983) put it, “Codes . . . serve account. In the case of hardship and loss, the
as shorthand devices to label, separate, compile, and core category might be “adaptation.”
organize data” (p. 186; emphases in original). This
coding process differs from the coding used in the The three types of coding are really different levels
analysis of quantitative data (e.g., social survey data) of the same process, each relating to a different
in two ways. First, whereas quantitative analysis point in the process of elaborating concepts and
often involves testing a pre-existing theory, grounded categories. Not all grounded theory practitioners
theory uses coding as an important first step in the operate with this threefold distinction; the notion
generation of theory. Second, whereas in quantitative of axial coding has been especially controversial
analysis coding is more or less a way of managing because it is sometimes perceived as closing off the
data that have already been at least broadly catego- coding process too quickly. Charmaz (2004) pre-
rized, in grounded theory (and other approaches to ferred to distinguish between open or initial coding
qualitative data analysis), coding is more tentative and selective or focused coding. The former tends
and fluid. The data are treated as potential indicators to be very detailed and may produce as much as
of concepts/categories, and the indicators are repeat- one code per line of text. It is crucial at this stage of
edly compared to see which concepts/categories they the research to be open-minded and to generate as
fit with best. Ad hoc compromises may have to be many new ideas, and hence codes, as necessary to
made when more than one researcher is doing the organize the data. Selective or focused coding em-
coding (Tastsoglou & Miedema, 2003). phasizes the codes that (a) appear most frequently
Strauss and Corbin (1990) distinguished three and (b) seem most revealing. Combining the initial
types of coding: codes generates new codes. The data are then re-ex-
plored and re-evaluated in terms of the selected
• Open coding: “the process of breaking down, codes. Pidgeon and Henwood (2004) provided a
examining, comparing, conceptualizing, and useful example in their study of 60 mother–daugh-
categorizing data” (p. 61); this process stays ter relationships:
very close to the data and yields the concepts
that are later grouped together to form catego- The initial coding led to the development of a
ries. Thus, concepts such as anger, jealousy, or long and varied, but highly unwieldy, list of in-
affection could form the category “emotion.” stances under the label “Relational Closeness.”
• Axial coding: “a set of procedures whereby data … [A] closer reading and comparison of the
are put back together in new ways after open individual instances indicated a much more
coding, by making connections between cat- mixed view of the emotional intensity of the
egories” (p. 96). This is done by linking codes relationships, ranging from a welcome but
to contexts, consequences, patterns of interac- painful sense of gratitude and debt to a stance
tion, and apparent causes. Thus, the category of hypersensitivity and a desire to flee from a
of emotion, above, could be linked to the con- relationship which involved “confinement”
texts in which it is expressed: for example, con- or “smothering.” … [T]his subdivision was
texts of hardship or loss. retained and coded through their respec-
• Selective coding: “the procedure of selecting tive labels “Closeness” and “Overcloseness.”
the core category, systematically relating it (Pidgeon & Henwood, 2004, p. 638)

bel65796_ch13_314-332 317 01/13/22 12:33 PM

318 PART III Qualitative Research

Research in the News

Natural disasters and mental health
In the spring of 2019, Woodhall-Melnik con-
ducted focus groups with over two dozen resi-
dents and interviewed 10 people involved in the
official response to the 2018 flood. Her research
considered the effect of informal supports like
neighbours and family, and formal supports
Jason Jeandron/Shutterstock

such as first responders and the provincial gov-

ernment. She found that people had a strong
bond to their property and community. They
wanted to stay after recovering from the flood,
but they were worried about courting disas-
Flooded streets in downtown Fredericton. ter. Some were optimistic that the flood was a
once-in-a-lifetime, while others were concerned
Julia Woodhall-Melnik, an assistant professor of about more flooding in 2019. People’s stress and
sociology at the University of New Brunswick, St anxiety varied, with the uncertainty about an-
John, has been studying how the extensive flood- other potential disaster leading some to despair.
ing in southern New Brunswick in 2018 and 2019 Working alongside neighbours and banding to-
impacted residents’ mental health (Bissett, 2019; gether in responding to the flood through such
CBC News, 2019; Fraser 2019). Woodhall-Melnik things as sandbagging was found to help resi-
points out, “Housing loss in general can lead to dents cope with the situation. When the area
fatigue, depression, anxiety, grief, sleep distur- experienced more flooding in 2019, it provided
bances and the development of post-traumatic Woodhall-Melnik with an opportunity to study
stress disorder” (Bissett, 2019). She hopes that the impact of a multi-year disaster. In her fol-
her research will shed light on how to reduce the low-up research she plans to also examine how
impact of these types of events and determine the experience has influenced residents’ views
what is needed to help people recover. on climate change.

Overall, then, the process of coding begins with • Categories: at a higher level of abstraction, sub-
the data themselves and then gradually moves to sume two or more concepts. An especially cru-
more selective and abstract ways of conceptualizing cial category may become a core category (see
the phenomenon of interest. Box 13.1).
• Properties: attributes or aspects of a category
Outcomes of grounded theory • Hypotheses: initial hunches about relationships
The following are the products of different phases of between concepts
grounded theory. • Theory: “a set of well-developed categories . . .
that are systematically interrelated through
• Concepts: the “building blocks of theory” statements of relationship to form a theoretical
(Strauss & Corbin, 1998, p. 101), the discrete framework that explains some relevant social
phenomena produced through open coding . . . or other phenomenon” (Strauss & Corbin,

bel65796_ch13_314-332 318 01/13/22 12:33 PM

 13 Qualitative Data Analysis 319

Box 13.1 Categories in grounded theory

Whiting, Parker, & Houghtaling (2014) used behaviour, one of the categories that emerged
grounded theory to explore how male perpetra- was “remorse.” One interviewee stated, “I have
tors of intimate partner violence (IPV) perceived always been upset by the abusive behaviours,”
their actions. The researchers suggest that ex- while another said, “I was in the wrong forever
amining the perpetrator’s point of view is import- [sic] putting my hands on her” (Whiting et al.,
ant in order to fully understand IPV and develop 2014, p. 282). The authors mention that a find-
effective therapies and interventions. ing of perpetrator remorse is rare in studies of
Qualitative interviews with 13 men were con- IPV, possibly because researchers do not want
ducted, with three research questions guiding to appear to be softening or excusing violence.
the analysis: How do men who have been vio- They take the position that violence is inexcus-
lent describe their relationship in terms of abuse? able and make their views on that issue quite
What do men believe contributes to the violence? clear in the article, and also point out that the
How do they feel about the violence in their re- offenders’ remorse may simply be an attempt
lationship? Prior to open coding, the data were at manipulation. Nonetheless, the researchers
examined to identify “significant statements” make the case that counsellors and other pro-
or “meaning units.” Open coding was then per- fessionals should take the remorse expressed by
formed by labelling the selected statements with perpetrators as a starting point to help them de-
descriptors. Following that, axial coding was done velop empathy for their victims and to reflect on
to group the data into “identifiable categories how IPV goes against their values and self-im-
that tied together” (Whiting et al., 2014, p. 279). age. Doing so may reduce the tendency to ra-
Although it was not uncommon for perpe- tionalize IPV and could ultimately result in less
trators to try to justify or rationalize their violent physical abuse.

1998, p. 22). In grounded theory there are two collection and analysis (indicated by arrows pointing
levels of theory: substantive and formal. A sub- in both directions). The figure implies the following:
stantive theory pertains to a certain empir-
ical instance of a phenomenon such as racial • The researcher begins with a general research
prejudice in a hospital setting (see Box 14.1). A question (step 1).
formal theory is at a higher level of abstraction • Relevant people and/or incidents are theoreti-
and has applicability to several substantive cally sampled (step 2).
areas, such as prejudice generally or in a vari- • Relevant data are collected (step 3).
ety of spheres. The generation of formal theory • Data are coded (step 4), which may, at the level
requires data collection in different settings. of open coding, generate concepts (step 4a).
• There is a constant movement backward and
The various elements are portrayed in Figure 13.1. forward among the first four steps, so that
Any diagram is only an approximation, but this is early coding suggests a need for new data,
particularly true in this instance, partly because which leads to theoretical sampling, and so on.
there are different approaches to grounded theory • Through constant comparison of indicators
and partly because it is difficult to convey diagram- and concepts (step 5) categories are generated
matically the iterative nature of grounded theory— (step 5a). It’s crucial to ensure a fit between in-
in particular, the recursive relationship between data dicators and concepts.

bel65796_ch13_314-332 319 01/13/22 12:33 PM

320 PART III Qualitative Research

Processes Outcomes • The substantive theory may eventually be ex-

1 Research question plored using grounded theory processes in a
different setting from the one in which it was
generated (step 12). In this way a formal theory
2 Theoretical sampling
(step 12a)—relating to more abstract categories
not specifically examined in the research—can
3 Collect data be generated.

4 Coding 4a Concepts
Note that Step 12 is relatively uncommon, because
researchers using the grounded theory approach
typically concentrate on one particular setting.
5 Constant comparison 5a Categories Sometimes researchers who use grounded theory
procedures produce surprising findings, as Box
6 Saturate categories 13.3 shows. When you read grounded theory stud-
ies, keep in mind that some authors will claim to
have used the “grounded theory” approach when
7 Explore relationships 7a Hypotheses
between categories in fact they have not used it at all (Charmaz, 2000)
or have used just one or two features of it (Locke,
1996). Concepts and categories are at the heart of the
8 Theoretical sampling
grounded theory approach, but coding, theoretical
sampling, and theoretical saturation are also key fea-
9 Collect data tures. Again, Box 13.2 provides an illustration.

Memos
10 Saturate categories
An important aid to grounded theory research is the
memo: an analytical note that researchers write for
11 Test hypotheses 11a Substantive theory themselves or colleagues about a code, category, or
other ideas they might have about their data. Memo-
12 Collection and analysis of 12a Formal theory writing occurs throughout the research process and
data in other settings assists researchers in developing their tentative an-
FIGURE 13.1 Processes and outcomes in grounded alytical hunches. Charmaz (2006) advises research-
theory ers to write memos in whatever way advances their
thinking and research. Memos can be very useful
• Categories become saturated in the course of in helping researchers to crystallize their ideas and
the coding process (step 6). keep track of their thoughts on various topics; they
• Relationships between categories are explored also help to shape the researchers’ reflections on
(step 7) in such a way that hypotheses about con- broader issues.
nections between categories emerge (step 7a). Reflecting on their separate research projects on
• Further data are collected via theoretical sam- Canadian first responders’ use of emergency tech-
pling (steps 8 and 9). nologies and the meanings Inuit attach to their
• The collection of data is likely to be governed living spaces, Sanders and van den Scott (2018) de-
by the theoretical saturation principle (step 10) scribe memo-writing as an integral part of how they
and the testing of the emerging hypotheses made sense of their data. They kept notebooks with
(step 11), which leads to specification of sub- them throughout the research process so they could
stantive theory (step 11a). See Box 13.2 for an record “any thoughts, questions, insights, and con-
illustration. cerns that might arise” (p. 280). During their coding,

bel65796_ch13_314-332 320 01/13/22 12:33 PM

 13 Qualitative Data Analysis 321

Box 13.2 Grounded theory in action

Charmaz (1997) used grounded theory to exam- “identity dilemmas”: the ways in which men
ine the identity dilemmas of men with chronic but approach and possibly resolve the assault
not terminal illnesses. She outlined clearly the on their traditionally masculine self-images.
chief steps in her analysis, which were: She showed that men often use strategies to
• interviews with men and a small number of re-establish earlier selves in order to preserve
women; their identities, at least in their own eyes;
• exploring the interview transcripts for gender • further interviews designed to refine the
differences; categories;
• searching for themes in the men’s interviews • rereading personal accounts of chronic illness
and published personal accounts such as au- with a particular focus on gender;
tobiographies. An example was the theme of • reading a new group of personal accounts;
“accommodation to uncertainty,” as the men and
found ways of dealing with the unpredictable • making “comparisons with women on selected
paths of their illnesses; key points” (Charmaz, 1997, p. 39).
• building “analytic categories from men’s defi- Charmaz proposed a substantive theory that
nitions of and taken-for-granted assumptions helped to explain the importance of notions of
about their situations” (p. 39). Of particular masculinity in maintaining an identity for chron-
significance in Charmaz’s work is the idea of ically ill men.

memo-writing helped them to contemplate connec- is conditioned by what is already known about
tions between their codes and make links between the social world under study (in terms of both
theories and ideas from the literature. It also assisted social scientific concepts and general, everyday
them in moving their abstract coding to a more the- ones). Many writers consider it an advantage to
oretical level, and helped them identify areas where build on the work of others.
clarification and more data collection were neces- • In practical terms, the time required to tran-
sary. For a further illustration of the use of memos, scribe recordings of interviews, for example,
see Box 13.4. can make it difficult for a researcher with a tight
deadline to carry out a genuine grounded theory
Criticisms of grounded theory analysis, given the constant interplay of data col-
Grounded theory has limitations: lection and conceptualization that it requires.
• It isn’t clear that the grounded theory method
• Some question whether researchers can fully necessarily results in theory. It offers a rigor-
suspend their awareness of existing theories or ous approach to the generation of concepts
concepts until the late stages of analysis when and categories, but it is often difficult to see
their theories are supposed to emerge. Most what theory, in the sense of an explanation of
social researchers are sensitive to the concep- something, is put forward. Moreover, although
tual armoury of their disciplines and it seems lip-service is frequently paid to the idea of gen-
unlikely that this awareness can be put aside. erating formal theory, most grounded theories
In fact, few people today believe that theo- pertain only to the specific social phenomenon
ry-neutral observation is possible. It is gener- under investigation—not to a broader range of
ally agreed that what is “seen,” even in research, phenomena.

bel65796_ch13_314-332 321 01/13/22 12:33 PM

322 PART III Qualitative Research

Box 13.3 Some counterintuitive results arising from a grounded

theory approach
One of the key insights of qualitative research was not conducted in person, Whitehead was
is that people often interpret their situations in not able to get information on the demographic
ways that an outsider might not expect. Take characteristics of the participants and she makes
the example of women with eating disorders. no mention of their nationality. Like many other
You might think that they would definitely want virtual interactions, the ones she observed were
to change their eating habits, but Whitehead probably not limited to people from a single
(2010) found that that is not always the case. country, although much of what she describes is
Instead, she found that some women suffer- applicable to Canadian society.
ing from anorexia nervosa and bulimia do not Taking a grounded theory approach to the
want to change that aspect of their lives, and issue, Whitehead identified five practices that
in fact may seek to develop a collective iden- contributed to the development of a collective
tity with women in similar situations by interact- identity: encouraging surreptitious eating be-
ing with them on pro–eating disorder (Pro-ED) haviours by sharing stories about how they had
websites. These sites are controversial in that hidden their bulimia from their partners or family
they do not present eating disorders in a neg- members; focusing on domestic tasks such as
ative light or recommend that sufferers seek cooking (for example, sharing recipes for dishes
treatment, but instead foster a sense of soli- that are delicious but not calorific and that are
darity and companionship for people who al- easily purged); encouraging the idea that
ready have the condition and who do not want beauty, in particular thinness, is essential to self-
to change. Some openly revel in their situation, worth; promoting friendships among people with
such as one woman who wrote that “Ana [an- eating disorders and even casting the disorder
orexia] loves me, She may use me, But she won’t as a friend; and drawing inspiration from celeb-
forget me. I am not her victim, I am one of her rity women who have battled eating disorders
lovers. Every day I pray to be loved above the or who are simply thin and beautiful. One of the
rest” (Whitehead, 2010, p. 617). author’s main conclusions is that these practices
The research involved the author regularly are “gendered” in the sense that they reinforce
visiting pro-ED websites and closely monitor- and reproduce dominant norms and stereotypes
ing the postings made there. Since the research about what it means to be a woman.

• Grounded theory tends to invite researchers to of allowing theory to emerge from data, have also
code their data into discrete chunks. However, been very influential. It is striking that the use
some commentators believe that this kind of of computer software has implicitly promoted
fragmentation results in a loss of context and grounded theory, because the software programs
narrative flow (Coffey & Atkinson, 1996); we have in many cases been written with its processes
will return to this point later in this chapter. in mind (Lonkila, 1995).

That said, and even though many studies do not Basic operations in qualitative data
use it consistently, grounded theory is probably analysis
the most influential overall strategy in use today Coding (or indexing) is the starting point for most
for qualitative data analysis. Many of its core el- forms of qualitative data analysis. The princi-
ements, from coding and memos to the very idea ples involved have been well developed by writers

bel65796_ch13_314-332 322 01/13/22 12:33 PM

 13 Qualitative Data Analysis 323

Box 13.4 A memo

In their research into the bus industry, Bryman, • the survival of attitudes and behaviour pat-
Gillingwater, and McGuinness (1996) noticed that terns, particularly among bus drivers, that
managers frequently mentioned that their com- are seen as inappropriate to the new envi-
panies still followed officially discontinued rules ronment (for example, lack of concern for
and practices. Managers often suggested that customer service) and that hinder service
“inherited” characteristics were holding them innovation; and
back in their efforts to adapt to a new competitive • high wages from the earlier era, which make it
environment. “Inheritance” is what Strauss (1987) possible for competitors who pay their drivers
called an in vivo code: one derived from the lower wages to enter the market and charge
language of people in the social context under less for their services.
study, as opposed to a sociologically constructed Sample comments:
code, which is a label created by the researcher.
The following memo outlines the concept of in- “I suppose another major weakness is
heritance, provides some illustrative quotations, that we are very tied by conditions and
and suggests some properties of the concept. practices we’ve inherited.” (Commercial
Director, Company G)
Memo for inheritance “We have what we’ve inherited and we now
Inheritance: Many of our interviewees have a massive surplus of double decks [i.e.,
suggest that they have inherited certain too many buses of the wrong size]. … We
company traits and traditions that they have to go on operating those.” (Managing
themselves would not have chosen. The key Director, Company B)
point about this inheritance is that our in-
terviewees see it as hindering their ability Managing Director of Company E said
to respond to a changing environment. the company had inherited staff steeped
in old attitudes: “We don’t have a staff
Inherited features might include: where the message is ‘the customer is
• expensive and often inappropriate fleets of number one.’”
vehicles and depots;

concerned with grounded theory and other perspec- • What are people doing?
tives. Among the issues they consider in developing • What do people say they are doing?
codes (cf. Lofland & Lofland, 1995) are the following: • What kind of event is going on?

• Of what general category is this datum an

instance? Steps and considerations in coding
• What does this datum represent? Following are some guidelines for coding:
• What is this datum about?
• What question does this datum suggest? • Code as soon as possible (that is, as data are
• What sort of answer to that question does this being collected), as grounded theory suggests.
datum imply? This may sharpen an understanding of the
• What is happening here? data and help with theoretical sampling.

bel65796_ch13_314-332 323 01/13/22 12:33 PM

324 PART III Qualitative Research

It can also help to alleviate the feeling of being to use the existing terms? Are there connec-
swamped by the data. At the very least, begin tions between the codes? Is there evidence that
transcribing recorded interviews at a relatively respondents believe one thing tends to be asso-
early stage. ciated with or caused by something else?
• Read through the initial set of transcripts, field • Consider more general theoretical ideas in re-
notes, documents, and so on, without taking lation to codes and data. Now you can start
any notes or considering any interpretations. generating some general theoretical ideas
Only when you have finished should you con- about the data. Try to outline connections be-
sider jotting down a few general notes about tween concepts and any developing categories.
what seems especially interesting, important, Consider in more detail how they relate to the
or significant. existing literature. Develop hypotheses about
• Do it again. Read through the data again, but the linkages you perceive and go back to the
this time make marginal notes about signifi- data to see if they can be confirmed.
cant remarks or observations—as many as • Finally, keep coding in perspective. It’s only one
possible. Initially, they will be very basic: per- part (albeit an important part) of the analysis:
haps keywords used by respondents or themes a mechanism for thinking about the meaning
in the data. Now you are coding: generating of the data and for reducing the data to a man-
terms that will help in interpreting the data. ageable size. The larger task of interpretation
• Don’t worry about generating too many codes— awaits, including forging interconnections be-
at least in the early stages of analysis. Some will tween codes, reflecting on the overall impor-
be fruitful, others not. The important thing is tance of the findings for the research literature,
to be as inventive and imaginative as possible; and pondering the significance of the coded
tidying up can be done later. Remember that material for the lives of the people who are the
any one datum can, and often should, be coded subjects of the research.
in more than one way. An outburst of anger,
for example, can be seen as an emotion, a Turning data into fragments
cause of stress, or the beginning of a new level The coding of materials such as interview transcripts
of integration. As a first stage, Charmaz (2004) typically entails writing marginal notes on them and
recommended “line by line coding,” in which gradually refining those notes into codes and then
virtually every line in the transcript (or other cutting and pasting (sometimes in the literal sense,
data source) has a code attached to it: this way, with scissors and glue). It entails cutting transcripts
the qualitative researcher does not lose contact into chunks of data (and of course carefully identi-
with the data and the perspectives and inter- fying the origins of each chunk with, for example,
pretations of those being studied. Although name, position in organization, date) for later data
the line-by-line approach almost always results retrieval. Although word-processing programs can
in a proliferation of codes, this should not be be used for this task, CAQDAS has become popular.
alarming. What qualitative researchers need to CAQDAS has been a growth area in terms of
do is ask what these codes have in common so both the proliferation of programs and the numbers
that they can be combined into higher-order of people using them. Most of the best-known pro-
and more abstract codes. grams allow analysts to code text and later retrieve
• Review the codes, possibly in relation to the it, tasks that were once done manually. For example,
transcripts. Are two or more words or phrases the software can search for all chunks of text relat-
being used to describe the same phenomenon? ing to a code, and then cut and paste them together.
If so, remove one of them. Do some of the Human input is still crucial, however. CAQDAS
codes relate to concepts and categories in the cannot help with decisions about the codes, or the
existing literature? If so, would it make sense coding of textual materials, or the interpretation of

bel65796_ch13_314-332 324 01/13/22 12:33 PM

 13 Qualitative Data Analysis 325

Methods in Motion Applications to Canadian Society

Using grounded theory to make sense of mothers’ experiences with food insecurity in
Nova Scotia
Lesley Frank (2020), an associate professor of so- insecurity. The impact of public policy on moth-
ciology at Acadia University, conducted a qualita- ers’ lives was also studied and became a sub-
tive study on infant food insecurity in Nova Scotia. theme of each of the three overarching themes
The questions that guided her research included: identified by Frank.
“What is the relationship between household food Frank learned that relying exclusively on
insecurity and infant feeding practice?” and “What breastfeeding is largely unsustainable for low-in-
role, if any, does Canadian public policy play in come mothers. Some of the mothers in her study
this relationship?” (Frank, 2020, p. 121). By seeking described wanting to breastfeed because it was
answers to these questions, her book, Out of Milk: seen as healthier and less expensive than formula,
Infant Food Insecurity in a Rich Nation (2020), ad- necessary for parent-child bonding, or because it
dresses the breastfeeding paradox: why mothers was in keeping with cultural traditions; however,
who can least afford to feed their infants formula many mothers found breastfeeding challenging,
are the ones who use it the most. The experience and some found it impossible given the stress
of food insecurity and poverty was at the heart of of living in poverty. Mary, one of the mothers in
the issue. Frank’s study, put it this way:
In the first phase of Frank’s research, she
I don’t know if what I was producing was
conducted 20 in-depth interviews with low-in-
really enough, nutrient wise, if it was—How
come mothers of infant children. To assist with
would you put it?—healthy enough for him,
her inductive analysis, Frank used NVivo as part
giving him what he needed. Plus, in order to
of the coding process. She started by conducting
be able to produce milk, the mother needs
line-by-line coding of participants’ responses.
to be food secure and you don’t have that
The codes were not predetermined; instead, she
all the time, and the situation could change
allowed them to emerge from the data as she
for you on the drop of a dime.
worked toward developing grounded theory.
After completing her open coding, she sought Frank discovered that infant feeding practices
to reduce the number of broad categories by have to do with more than what happens at the
making connections between the codes. She individual level. If the problem was to be un-
then re-coded her data through the lens of derstood, she realized that her analysis needed
this new set of categories. During the last step to also take into account how public policy and
of her analysis, she grouped the themes into community supports intertwine with people’s
three broad categories: “breastfeeding as a everyday lives. She writes, “Building infant food
food security measure, breastfeeding as an in- security means enhancing [ . . . ] practical in-
secure food system, and infant food insecurity stitutional and community supports to form an
related to formula feeding” (Frank, 2020, p. 127). integral component of health and social services
Together, the categories that emerged out of for families. Doing so will help us to fulfil our col-
Frank’s analysis offer a typology of infant feed- lective responsibility for the welfare of children
ing practices for mothers’ experiencing food and the social reproduction of society” (p. 119).

bel65796_ch13_314-332 325 01/13/22 12:33 PM

326 PART III Qualitative Research

findings. As with quantitative data analysis software themes. The responses in the Box 11.5 tran-
such as IBM SPSS Statistics (SPSS), someone must script were coded in terms of characteristics
still choose the variables to be analyzed and the an- such as enthusiasm (“uncritical enthusiasm”),
alytic techniques to be used, and then make sense of attitudes toward the Disney Corporation (“not
the results. Each form of software requires creativity. critical of Disney”); comments made about
CAQDAS differs from SPSS largely in terms of the typical visitors (“visitors’ ethnicity”); and the
type of data that can be used with it. nature of critical comments (“aesthetic cri-
There is no CAQDAS industry leader, but NVivo is tique,” “ethnicity critique,” “nationality cri-
a package that most researchers would at least know tique”). Interestingly, the passage also reveals
by name. Earlier versions of the software became the potential for a code employed by Coffey
very popular in the 1990s. The Appendix outlines and Atkinson (1996, pp. 43–45): the use of a
how to use NVivo for qualitative data analysis. “contrastive rhetoric” (making a point about
To use or not to use CAQDAS? With a very one thing by comparing it to something else).
small data set, it is probably not worth the time and This feature occurred when the husband made
trouble required to become familiar with new soft- a point about the representation of British cul-
ware. Catterall and Maclaran (1997) have argued ture, which he regarded as poor compared to
that CAQDAS is not very suitable for focus group that of China.
data because the code and retrieve function tends
to hide the communication process typical of focus As Coffey and Atkinson (1996) observed, following
groups. On the other hand, learning new software Strauss and Corbin’s account (1990) of grounded
will give you skills that you may need in the future. theory, codes should not be thought of purely as
CAQDAS is likely to be too expensive for personal mechanisms for the fragmentation and retrieval of
purchase, though there are student and educational text. They can do more than simply manage the data
discounts. Demonstration copies of some of the gathered. For example, examination of the intercon-
main packages can be downloaded from the dis- nections between codes may reveal that some are
tributors’ websites. dimensions of a broader phenomenon. For example,
There is no one correct approach to coding data. “ethnicity critique” came to be seen as a dimension
As we suggested earlier, grounded theory conceives of “ideology critique,” along with “class critique”
of different types of code. Coffey and Atkinson and “gender critique.” In this way, a map of the more
(1996) pointed to three different levels of coding, ap- general or formal properties of the concepts under
plied below to a passage from an interview that first development can be started.
appeared in Box 11.5 (p. 259).
Problems with coding
• First there is a very basic coding, which, in the One of the most common criticisms of coding qual-
case of the Box 11.5 material could be liking itative data is the risk of losing the context of what is
or disliking the visit to a Disney theme park. said. (For a general critique of qualitative data anal-
However, such a coding scheme is unlikely ysis packages see Box 13.5.) By plucking chunks of
to provide anything more than a superficial text out of the context in which they appear, such
analysis. as a particular interview transcript, the social set-
• A second level involves a deeper awareness of ting can be lost. A second criticism of coding is
the content of what is said and is organized that it results in fragmentation of the data, causing
around the focus of the research. An exam- the narrative flow to be lost (Coffey & Atkinson,
ple might be countries “well-represented” and 1996). Sensitivity to this issue has been heightened
“missing” at Disney World. since the late 1980s by a growing interest in nar-
• A third level moves slightly away from what rative analysis (see later in this chapter). Riessman
the respondent says to explore broader analytic (1993) became concerned about fragmentation when

bel65796_ch13_314-332 326 01/13/22 12:33 PM

 13 Qualitative Data Analysis 327

Box 13.5 Lack of agreement about the utility of CAQDAS

Whereas the use of computer software is almost as a new standard is inconsistent with a key
universal in quantitative data analysis, among strength of qualitative research: its flexibility.
qualitative data analysts it is sometimes avoided, On the other hand, several writers have extolled
for several reasons. the virtues of such packages:
• Some writers worry that because qualitative • CAQDAS, like NVivo, invites thought about
data analysis packages make it so easy to “trees” of interrelated ideas—a useful feature
quantify coded text, qualitative research will in that it encourages the analyst to consider
become subject to the reliability and validity possible connections between concepts.
criteria of quantitative research (Hesse-Biber, • Quantitative researchers often criticize the
1995). tendency toward “anecdotalism” found in
• Others feel that CAQDAS reinforces and even much qualitative research—that is, the ten-
exaggerates the tendency toward a frag- dency to use quotations from interview tran-
mentation of the textual materials (Weaver scripts or field notes with little sense of how
& Atkinson, 1995), destroying the natural representative they might be. CAQDAS makes
flow of interview transcripts and field notes. it possible to count the frequency with which a
Awareness of context is crucial to many qual- particular form of behaviour occurs or a par-
itative researchers, and they are concerned ticular viewpoint is expressed in interviews.
by the prospect that this awareness may be • CAQDAS enhances the transparency of qual-
pushed to the sidelines. Stanley and Temple itative data analysis. How qualitative data
(1995) suggested that most of the coding were analyzed is often unclear in published
and retrieval features needed for qualitative reports (Bryman & Burgess, 1994). CAQDAS
data analysis, such as search, cut, and paste, may force researchers to be more explicit and
already exist in word-processing software. reflective about the process of analysis, and
Using Microsoft Word, for example, would may indirectly encourage replication, a fea-
save not only money but also the time re- ture often lacking in qualitative analysis.
quired to learn new software. • When working with large amounts of qualita-
• Coffey, Holbrook, & Atkinson (1996) argued tive data (“big qual”) on multi-person interna-
that the style of qualitative data analysis in tional research teams, technological advances
most CAQDAS presumes a methodology— in CAQDAS, like cloud-based storage, can
based on coding and retrieving text—that assist with data management and collab-
owes a great deal to grounded theory. In oration (Brower, Bertrand Jones, Osborne-
their view, the emergence of grounded theory Lampkin, Hu, & Park-Gaghan, 2019).

coding themes in her structured interview study on I searched the texts for common thematic ele-
divorce and gender: ments. But some individuals knotted together
several themes into long accounts that had co-
Some [interviewees] developed long accounts
herence and sequence, defying easy categori-
of what had happened in their marriages to
zation. I found myself not wanting to fragment
justify their divorces. I did not realize these
the long accounts into distinct thematic cate-
were narratives until I struggled to code them.
gories. (p. vi)
Applying traditional qualitative methods,

bel65796_ch13_314-332 327 01/13/22 12:33 PM

328 PART III Qualitative Research

Riessman’s account is interesting because it suggests situations; to the stories they generate about them;
several possibilities: that the coding method can and to the significance of context for the unfolding
fragment the data; that some forms of data may be of events and people’s sense of their roles in them.
unsuitable for coding; and that researchers can pro- The way people forge connections between events
duce narrative analysis, since what she provided in and the sense they make of those connections are the
this passage is precisely a narrative. Interest in nar- raw material of narrative analysis. Riessman (2004b)
rative analysis has grown and in large part this trend helpfully distinguished four models of narrative
parallels the rebirth of interest in the life history analysis:
approach (see Box 11.3). Nonetheless, coding is un-
likely to become less prominent, for several reasons: • Thematic analysis examines what is said rather
it is widely accepted in the research community; not than how it is said.
all analysts are interested in research questions that • Structural analysis emphasizes the way a story
lend themselves to narratives; grounded theory and is told. Issues of content do not disappear, but
the techniques associated with it are very influential; attention is focused on the use of narrative
and the use of computer software for qualitative data mechanisms to increase the persuasiveness of
analysis frequently invites a coding approach. a story.
Regardless of analytical strategy, it is unaccept- • Interactional analysis looks at the dialogue be-
able to simply say: “This is what my participants tween the teller of a story and the listener. The
said and did; isn’t that incredibly interesting?” co-construction of meaning by the two parties
Interpretation and theorizing are necessary. Many is especially prominent, although content and
researchers worry that focusing on those tasks may form are also taken into account.
prevent them from doing justice to what they have • Performative analysis analyzes narrative as
seen and heard, or lead them to contaminate their a performance and explores the use of words
subjects’ words and behaviour. Those are real risks, and gestures to get a story across. This model
but they must be balanced against the fact that find- of narrative analysis also examines audience
ings acquire significance in an intellectual commu- responses to the narrative.
nity only when they have been subject to reflection.
The researcher has to do more than simply take notes With narrative analysis, attention shifts from “What
or record answers. happened?” to “How do people make sense of what
happened?” Proponents of narrative analysis argue
Narrative analysis that most approaches to the collection and analysis
The term “narrative analysis” covers a wide variety of data ignore the fact that people perceive their lives
of approaches based on the search for and analysis of in terms of continuity and process, and that efforts
stories that people tell to understand their lives and to understand social life that do not take that fact
the world around them. It has become particularly into account may misunderstand the perspective of
prominent in connection with the life history or bi- those being studied. Narrative analysis is obviously
ographical approach. However, as Box 13.6 shows, suitable for life history research, but its applications
narrative analysis is not exclusively concerned with can be much broader. For example, it can also be
telling life histories. As Roberts (2002) observed, the applied to accounts of shorter episodes and their
term is often used to refer both to the research ap- interconnections.
proach itself and to the accounts that it examines— Some researchers apply narrative analysis to inter-
the stories that people tell about their lives and other views. In her account of her “click moment” as a nar-
events. In general, narrative analysis entails sensitiv- rative researcher, Riessman (1993) described how she
ity to the connections between people’s accounts of applied narrative analysis to conventional interview
past, present, and future events and states of affairs; transcript material, and then began to uncover the
to people’s sense of their place in those events and stories her interviewees were telling her. In this case,

bel65796_ch13_314-332 328 01/13/22 12:33 PM

 13 Qualitative Data Analysis 329

Box 13.6 Tattoo narratives

As part of a three-year participant observation
study of tattooing, Atkinson (2002, 2004) elicited
tattoo narratives from the women he interviewed.
The great increase in the numbers of women
getting tattoos has clearly challenged the stan-
dard “masculinity” explanations of male tattoo-
ing. Drawing on feminist theories about bodies,
grandriver/iStockphoto

he found that these women are not misfits, and

that they use tattoos to signify their “established”
or “outsider” constructions of femininity. Ideas
about femininity, including conformity to and
How can narratives be used to analyze a social resistance against cultural norms, are crucial in
trend such as the increasing popularity of tattoos explaining women’s tattooing.
for women?

Riessman was applying a narrative approach to ma- between events and contexts. Miller (2000) proposed
terials gathered in a conventional way for conven- that narrative interviews in life story or biographical
tional purposes. Other researchers start out with the research are far less concerned with eliciting facts
intention of conducting a narrative analysis and de- than with drawing out interviewees’ reflections and
liberately ask people to recount stories (for example, interpretations, as revealed in their accounts of their
Miller, 2000). While stories can emerge in response lives or families.
to questions not designed to elicit a narrative, cer- Narrative analysis, then, is an approach to the
tain kinds of question are especially likely to produce analysis of qualitative data that examines the stories
them. Riessman (2004a) suggested that prompts that people use to explain events. It can be applied
such as “Tell me what happened” followed by “And to data that have been acquired through a variety
then what happened?” are much more likely to elicit of research methods (notably semi-structured and
a narrative account than “When did X happen?” unstructured interviewing and participant ob-
While some narrative researchers prefer to start servation), but it has also become an interviewing
people off by asking them to tell their story of an approach in its own right. Also, there is growing
event, Riessman argued that it is usually necessary recognition of ethnography as a narrative designed
to keep asking follow-up questions to stimulate the to tell a story about a way of life (Falconi & Graber,
flow of details and impressions. For example, in her 2019).
study of divorce, she often asked: “Can you remem- Narrative analysis has been criticized. Bury
ber a time when . . . ?” and then followed it up with (2001), while noting the increased interest in illness
“What happened that makes you remember that par- narratives (accounts of illnesses people have expe-
ticular moment in your marriage?” rienced), argued that there has been a tendency for
Coffey and Atkinson (1996) argued that a narra- narrative researchers to treat such stories uncriti-
tive should be examined in terms of the functions cally. For example, he suggested that the frequent
that it serves for the teller. The aim of narrative in- references in illness narratives to coping with and
terviews is to elicit interviewees’ reconstructed ac- “normalizing” illness may often represent attempts to
counts of connections between events and links convince the audience (whether the interviewer or the

bel65796_ch13_314-332 329 01/13/22 12:33 PM

330 PART III Qualitative Research

reader of the narrative) of the subject’s competence; conditions that prompt such narratives and the
in other words, such accounts may be motivated forms the narratives take are themselves revealing.
more by the desire to be seen as a fully functioning In drawing attention to the motives behind illness
member of society than by the desire to give a re- narratives, he was not seeking to undermine nar-
alistic account of coming to terms with a medical rative analysis but to show how the range of issues
condition. However, as Bury recognized, the social addressed in this form of research can be expanded.

Key Points
• The collection of qualitative data frequently re- • Coding is a key process in most qualitative data
sults in the accumulation of a large volume of analysis strategies, but it may have the effect of
information. fragmenting and decontextualizing information.
• Qualitative data analysis is not governed by cod- • Narrative analysis emphasizes the stories that
ified rules to the same extent that quantitative people tell in the course of interactions with a
data analysis is. qualitative researcher; it has become a strategy
• There are different approaches to qualitative for producing and analyzing qualitative data in
data analysis; grounded theory is probably the its own right.
most prominent.

Questions for Review (R) and Creative Application (A)

General strategies of qualitative data a4 How realistic is the assumption in grounded
analysis theory that a researcher can begin to collect
r1 What are the main components of analytic data without being influenced by existing theo-
induction? ries and concepts? Explain.
a1 Why is analytic induction a rarely used
approach? Basic operations in qualitative data
r2 What are the main features of grounded theory? analysis
What are the different types of coding used in r5 What are the main steps in coding?
grounded theory? a5 Is narrative analysis the answer to the problem
a2 You are conducting grounded theory research of data fragmentation that results from coding?
on divorced couples, using data derived from Explain.
qualitative interviews. What concepts might
emerge when you begin open coding? Narrative analysis
r3 What is theoretical sampling? R6 Explain how the emphasis on stories in narrative
a3 Charmaz (2000) wrote that theoretical sampling analysis represents a distinctive approach to the
“represents a defining property of grounded production and analysis of qualitative data.
theory” (p. 519). Why do you think she feels this A6 Come up with a research question pertaining to
way? gender roles that can be answered using narra-
r4 What are some of the main criticisms of grounded tive analysis.
theory?

Interactive Classroom Activities

1. Each student in the class is paired up with an- hopes to be doing with his or her life in 10 years’
other student. For 20 minutes, one student in time, why they have chosen those goals, how
the pair conducts a qualitative interview with they hope to achieve their goals, and so on. The
the other on the topic of what the interviewee interviewer takes field notes the whole time.

bel65796_ch13_314-332 330 01/13/22 12:33 PM

 13 Qualitative Data Analysis 331

When the 20 minutes are up the roles are re- of the strengths and weaknesses of the grounded
versed: the interviewer becomes the inter- theory approach to qualitative research.
viewee. Once the second interview is completed, 2. Each student in the class is paired up with an-
each person in the class is asked to: other student. For 20 minutes, one student con-
a. conduct “open coding” of their field notes, ducts a qualitative interview with the other using
identifying “concepts” and “categories” the “narrative analysis” technique. The topic to be
where possible; addressed is the interviewee’s most memorable
b. write at least one “analytic memo” that childhood or adolescent experience, and how that
describes noteworthy themes that have experienced shaped their lives over time. Students
emerged in their data; are reminded that their goal as interviewers is to
c. put the concepts and categories together to get the interviewees to “tell their stories”—i.e., to
form a tentative hypothesis or interpretative provide narratives that illustrate how their lives
framework; have been a process rather than a series of dis-
d. identify topics that would benefit from further connected events and experiences. When the 20
questioning or investigation. minutes are up the students reverse roles, with the
interviewer becoming the interviewee. Once the
Each student then re-interviews their interviewee second set of interviews has been completed, the
for 20 minutes. When both interviews have been instructor asks students to share any difficulties
completed, students conduct open, axial, or se- they encountered in conducting the interviews, in
lective coding, as appropriate, on all the data particular any issues they had in establishing and
they have gathered. They then revisit their orig- sustaining the flow of the narrative. The class is
inal hypotheses or interpretive framework, and also asked for suggestions as to how such difficul-
revise as needed. When the process is complete ties could be overcome. The instructor then leads
students are asked to share with the rest of the a general class discussion on the strengths and
class the concepts, constructs, hypotheses, and weaknesses of the narrative analysis approach to
interpretive frameworks they have developed. qualitative research compared to the grounded
Then the instructor facilitates an open discussion theory method.

Relevant Websites
A wealth of information is available at the website [Link]/watch?v=4SZDTp3_New

of the Grounded Theory Institute. [Link]/watch?NR=1&v=gn7Pr8M_Gu8

[Link] [Link]/watch?v=vi5B7Zo0_OE&NR=1

In this YouTube video, Kathy Charmaz provides If you’d like to experiment with NVivo software, trial
an overview of the grounded theory method, de- downloads are available at the NVivo website.
scribes its historical emergence, and defines con-
[Link]
structivist grounded theory. nvivo-qualitative-data-analysis-software/try-nvivo

[Link] (Websites accessed 4 September 2021)

These YouTube videos discuss grounded theory,

including coding and constant comparison.

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch13_314-332 331 01/13/22 12:33 PM

bel65796_ch13_314-332 332 01/13/22 12:33 PM
 PART IV
Bringing It All Together:
Revisioning Quantitative and
Qualitative Research, and Some
Practical Advice
This section of the book expands on the comparison between
quantitative and qualitative approaches to research that were
touched on in earlier chapters. Chapter 14 invites readers to con-
sider the distinction between those two approaches in greater
depth, and makes the point that some of the differences between
them alluded to previously may not be as fundamental as one
might think. The aim here is not to deny the distinction, but to show
that in addition to differences, there are some similarities between
the two. The final chapter of the book, Chapter 15, offers practi-
cal advice on completing a small-scale research project, including
some tips on writing up social research. These two chapters return
to issues raised in earlier sections of the book, probing them more
extensively and making connections between them.

bel65796_ch14_333-358 333 01/13/22 12:32 PM

 14 Bridging the Quantitative/
Qualitative Divide
Chapter Overview
This chapter is concerned with how far the quantitative/qualitative distinction should be taken. Although
there are many differences between the two research strategies, there is also much that unites them, as we
will see when we discuss how they can be combined. This chapter explores:
• aspects of qualitative research that can contain elements of the natural science model;
• aspects of quantitative research that can contain elements of interpretivism;
• the idea that research methods are more independent of epistemological and ontological assump-
tions than is sometimes supposed;
• ways in which the quantitative/qualitative contrast may break down;
• studies in which qualitative research is used to analyze quantitative research and vice versa;
• the use of quantification in qualitative research;
• arguments against the combination of quantitative and qualitative research;
• two versions of the debate on combining quantitative and qualitative research, one concentrating on
methods of research and the other on epistemological issues;
• different ways in which multi-strategy research has been carried out; and
• the claim that multi-strategy research is not inherently superior to research employing just one re-
search strategy.

▲ SeanZeroThree/iStockphoto

bel65796_ch14_333-358 334 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 335

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:

1. Describe how qualitative research exhibits some of the features of the natural science model.
2. Discuss how quantitative research can be used to study meaning.
3. Explain how quantitative research can contribute to constructionist research and how the tenets of
constructionism can help inform our understanding of quantitative data.
4. Identify how research methods are not always guided by epistemological and ontological
considerations.
5. Summarize the ways in which quantitative and qualitative approaches are not as divergent as they
are often portrayed.
6. Describe how a qualitative approach can be applied to quantitative research and vice versa.
7. Detail how quantification is used in qualitative research.
8. Discuss different approaches to multi-strategy research, criticisms against combining methods, and
reasons why a multi-strategy approach is not necessarily a better approach than using a single
method.

Philosophers of science sometimes use the term common and can complement each other. It makes
“consilience” to refer to the notion that there is a the case that both are required to gain knowledge
unity or compatibility between all branches of of the social world.
knowledge. The word came into wide usage after As you progress through your academic career,
biologist E.O. Wilson published Consilience: The you may be tempted to take a dismissive attitude
Unity of Knowledge (1998), in which he argued toward approaches and methodologies that are
that the natural sciences, the social sciences, and different from your own. Try to resist that tempta-
even the humanities may one day have a unity of tion: giving in to it can lead to needless tensions and
form and purpose that will allow for a more thor- rivalries between researchers and can ultimately
ough and profound understanding of natural and be quite destructive. Drawing on the insights of
social phenomena. This chapter is written in a sim- multiple perspectives is much more conducive
ilar spirit, seeking to show that quantitative and to good research and produces a deeper level of
qualitative research strategies have a great deal in understanding.

Introduction • There are differences between quantitative and

qualitative research strategies.
Given the emphasis we have placed on the distinc- • Contrasting the two is a useful way of organizing
tion between quantitative and qualitative research, research methods and approaches to data analysis.
it might seem strange at this stage to suggest that
the differences between them are sometimes exag- As we have seen in earlier chapters, certain episte-
gerated. The dichotomy has been maintained to this mological and ontological positions may be associ-
point for two simple reasons: ated with particular research methods: for example,

bel65796_ch14_333-358 335 01/13/22 12:32 PM

336 PART IV Bringing It All Together

a natural science epistemology is often paired with approaches (sometimes referred to as mixed meth-
social survey research, and an interpretivist episte- ods research; Timans, Wouters, & Heilbron, 2019).
mology with qualitative interviewing. However, the This term does not refer to the use of several
connections posited in Chapter 1 between episte- quantitative methods in a single study, or the in-
mology and ontology, on the one hand, and research tegration of two or more qualitative methods—for
method, on the other, are best thought of as ten- example, combining structured interviewing with
dencies rather than definitive connections. The use structured observation, or ethnography with
of a structured interview or self-completion ques- semi-structured interviewing. In the current dis-
tionnaire does not necessarily imply an exclusive cussion, “multi-strategy research” refers only to
commitment to a natural science model; nor does studies that combine qualitative and quantitative
ethnographic research necessarily mean an inter- methods in the same project.
pretivist epistemology.
Research methods are much more free-floating The natural science model
than is sometimes supposed. A method of data col- and qualitative research
lection like participant observation can be employed
in such a way that it is in tune with constructionism, One of the chief difficulties in linking issues of epis-
but it can also be used in a way that reveals an ob- temology and ontology with research methods is the
jectivist orientation. Also, it’s easy to underestimate frequent characterization of the natural sciences as
the importance of practical considerations in deter- inherently positivist in orientation. But the term
mining how social research is carried out. It’s not im- “positivism” is often used in a polemical way, usu-
possible to conduct a study of drug dealers by using ally as an unhelpful criticism of another’s work.
questionnaires, but for practical reasons such a strat- Moreover, qualitative research itself frequently ex-
egy is unlikely to succeed. The rest of this chapter will hibits features associated with a natural science
demonstrate why the distinction between quantitative model. This tendency is revealed in several ways:
and qualitative research should not be overstated.
It is of course possible to combine the two ap- • Empiricist overtones. Although empiricism
proaches so that their various strengths can be cap- (see Chapter 1) is typically associated with
italized on and their weaknesses compensated for. quantitative research, many qualitative re-
Indeed, the number of studies that combine different searchers place a similar emphasis on the im-
strategies has been increasing since the early 1980s. portance of direct, sensory contact with social
However, not all research methodologists would reality. They frequently stress the importance
agree that such integration is either feasible or desir- of direct experience with social settings, and
able. In discussing the combination of quantitative advocate understanding social reality via that
and qualitative strategies, this chapter looks at three contact. The very idea that theory should be
main issues: grounded in data is central to empiricism. It’s
not surprising, therefore, that some writers
• the arguments against integrating quantitative claim to detect “covert positivism” in qualita-
and qualitative methods tive research.
• the different ways in which quantitative and Another example of empiricist overtones
qualitative strategies have been combined can be seen in the idea that social reality
• whether a combined research strategy is nec- must be studied from the vantage point of
essarily superior to one relying on a single the research participants. The fact that eth-
approach nographers rely on extended contact to gain
access to the perspectives of the people they
This chapter uses the term multi-strategy research observe implies acceptance of the empiri-
for work that integrates quantitative and qualitative cal notion that meaning is accessible to the

bel65796_ch14_333-358 336 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 337

senses. The empiricism of qualitative re- Festinger, Riecken, & Schachter (1956) on a
search is perhaps most notable in conversa- doomsday cult used the qualitative technique
tion analysis (examined in Chapter 12), an of participant observation to test a theory
approach that starts with precise transcrip- about how people respond when a belief they
tions of speech and applies rules of analysis zealously endorse—in this case, the immi-
to them. The analyst is actively discouraged nent end of the world—is disconfirmed. The
from speculating about intention or context; authors argued that it is possible to imagine a
in other words, the empiricist notion that number of conditions that, if met, would lead
one should not stray too far from the data is group members to embrace their belief even
strictly upheld. more fervently after its disconfirmation. They
• A specific problem focus. Qualitative research thought that the cult would provide an ideal
can investigate quite specific, tightly defined opportunity for testing their theory. The re-
research questions of the kind normally asso- searchers (and hired observers) pretended to
ciated with the natural science model. be converts and gained entry to the group in
• Hypothesis and theory testing. Following from order to make first-hand observations. The
the last point, qualitative researchers typi- theory was substantiated when cult members
cally test hypotheses or theories generated in decided that their faith had saved the world
the course of conducting their research, as in and began proselytizing—an activity they had
analytic induction or grounded theory. This not engaged in before.
is well within the bounds of a natural science • Realism. Critical realist ethnography (Porter,
model, even if the natural sciences tend to be 1993; Barron, 2013; see Box 14.1) illustrates
deductive. Also, there is no reason why hy- how ethnographic studies can be based on
pothesis or theory testing in qualitative social epistemological and ontological positions
research cannot be done with previously spec- that have considerable affinity with those
ified hypotheses or theories. In fact, this has found in the natural sciences. It also relates to
been done on many occasions. For example, the previous point in providing a further il-
Wilson’s (2002) qualitative study of raves used lustration of hypothesis testing in qualitative
theory as a departure point. The research by research.

Box 14.1 Critical realist ethnography

Critical realism (Bhaskar, 1975; 1979; 1986) pres- and fallible, and as such should be subject to em-
ents a view of reality that in many ways is in keep- pirical scrutiny, and on the notion that research-
ing with the ontological approaches taken in the ers should seek to uncover causal tendencies to
natural sciences, despite important differences explain their subject matter.
(Gorski, 2013). Critical realist ethnography, for These sorts of positivist-friendly assumptions
example, maintains that the social world has a are evident in the work of Porter (1993; 2002),
form and structure that exists independently of who took a critical realist stance in his ethno-
our perception of it and which affects human graphic study of racism in a large Irish hospital
consciousness and behaviours in complex ways. where he spent three months as a staff nurse. He
This sort of ethnography is also based on the idea found that racism was expressed in the form of
that the descriptions and explanations presented remarks made behind the backs of members of
by social researchers are inherently provisional racial minorities. However, it did not intrude into
(Continued)

bel65796_ch14_333-358 337 01/13/22 12:32 PM

338 PART IV Bringing It All Together

work relationships, because more weight was Similarly, Barron (2013) used the critical real-
given to achievements and performance (such ist paradigm in his ethnographic study of ethnic
as qualifications and medical skills) than to “race” identity among white British and Pakistani-British
when judging members of professions. In terms children. He concluded that his previous eth-
of critical realism, one possible structural mech- nographic work on the topic, although insight-
anism (racism) was countered by the operation ful, could have been improved by acquiring a
of another structural mechanism (professional broader range of empirical data, a practice that
ideology). Only on rare occasions did racial ten- has important parallels in the natural sciences.
sion surface.

In addition, some qualitative researchers include It may be objected that survey questions do not
explicitly quantitative elements in their research. really tap issues of meaning because they are based on
Miller (2000), for example, engaging in what he categories devised by the designers of the interview
called neo-positivism, made theoretically based pre- schedule or questionnaire and not by the participants
dictions about people’s lives using the life history themselves. Two points are relevant here. First, the
method. Another illustration is Charmaz’s (2000) notion that qualitative research is better at gaining
suggestion that in spite of the differences that de- access to the point of view of those being studied is gen-
veloped between Glaser (1992) and Strauss (for ex- erally assumed rather than demonstrated. Qualitative
ample, Strauss & Corbin, 1998)—who were major researchers frequently claim to have tapped into par-
proponents of grounded theory—both maintained ticipants’ world views because of their extensive par-
that an objective, external reality exists (the position ticipation in the daily life of the people who are the
taken in the natural sciences). Each posited a social focus of their study, the length of time they spend in
world beyond the researcher and maintained that the setting under investigation, or the lengthy and in-
it was the job of the social investigator to reveal its tensive interviews they conduct. However, the explicit
nature and functioning. demonstration that interpretative understanding has
been accomplished—for example, through respon-
Quantitative research and dent validation (see Chapter 9)—is rarely undertaken.
interpretivism Second, if the design of attitude questions is based on
prior questioning that seeks to bring out the range of
Qualitative research would seem to have a near-mo- possible attitudinal positions on an issue, attitudinal
nopoly on the ability to study meaning. But is it really questions can certainly provide access to meaning.
the case that only qualitative research makes it pos- Also, the practice in much survey research
sible to see the social world through the eyes of the of asking respondents the reasons for their ac-
people studied? That contention seems at odds with tions implies that quantitative researchers are
the widespread study of attitudes using structured frequently concerned with uncovering meaning.
interviews and questionnaires. For example, there For example, in the research on delinquency in
is a huge literature on political attitudes, and many Box 1.2, the boys were asked to give the reasons
indexes and scales have been designed to measure for their actions in their own words and then to
them (see Bell, Kandler, & Riemann, 2018; Robinson, choose the sociological theories that came closest
Shaver, & Wrightsman, 1999). The widespread inclu- to explaining them. Examples such as these point
sion of questions about attitudes in social surveys to the possibility that the gulf between quantita-
suggests that quantitative researchers are interested tive and qualitative research is not as wide as is
in matters of meaning too. sometimes supposed.

bel65796_ch14_333-358 338 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 339

Quantitative research and under age 50, and that 85 per cent of the anecdotes
and case stories related to women in this age group.
constructionism This emphasis on younger women helped to create
We noted in Chapter 1 that a keynote in qualita- the impression that they were the most at risk. But in
tive analysis is constructionism: how people build reality, fewer than 20 per cent of new cases of breast
up images or representations of the social world. cancer were in women under 50, and the mean age at
Qualitative content analysis plays an important diagnosis was 65. This inconsistency allowed Lantz
role in developing an understanding of how people and Booth to uncover how the media drew a con-
construct their visions of reality, in the same way nection between youthful lifestyles (working outside
that discourse analysis (Chapter 12) does using the home, postponing parenthood, greater sexuality)
materials such as newspaper reports and television and breast cancer—a connection consistent with the
programs. articles’ “blame the victim” theme. The quantitative
Lantz and Booth’s (1998) research on the social content analysis of articles on breast cancer played
construction of breast cancer provides an example. an important part in revealing the social construc-
As Box 1.5 makes clear, much of their understand- tion of beliefs about the disease. More generally, this
ing of the representation of breast cancer came example shows how quantitative research can con-
from a qualitative content analysis, but they con- tribute to constructionist research.
ducted a quantitative content analysis as well. The It is also important to remember that quantitative
latter showed that 80 per cent of the women in the data are not formulated in a social vacuum: statistics
photographs attached to breast cancer articles were themselves are social constructions (see Box 14.2).

Box 14.2 The Social Construction of Numbers and COVID-19

From a constructionist standpoint, interpreting something a fact always occurs within a par-
statistics can benefit from understanding how ticular social context” (p. 3). Best is not argu-
claims about numbers are socially constructed. ing that all claims are equally true. Instead,
Best (2020) writes, “Teaching numeracy cannot he is pointing out that one group might treat
be understood only as a mathematical project something as true, while another might not—
each has different standards for evaluating
that can ignore the way numbers are brought
claims and determining which are factual.
into being” (p. 13). He suggests that those inter-
Just because one group deems something
ested in using numbers to make sense of social
to be a fact, and brings numbers to bear to
problems should take into account the following support their contention, does not mean that
constructionist observations: (1) facts are social another group will agree and the debate will
constructions; (2) measuring involves making disappear.
decisions; (3) counting is not straightforward; 2. Measuring involves making decisions: “A long
(4) all comparisons involve choices; and, (5) series of choices shapes whatever is being
social patterns shape numbers. To illustrate counted and how the counting occurs. When
his five propositions, Best (2020) considers how we talk about the resulting numbers as facts
numbers are used to make sense of the COVID- (e.g., equating the crime rate with the amount
19 pandemic: of crime), we ignore the very real processes by
which those numbers are socially constructed,
1. Facts are social constructions: “Facts are prod- and the inevitable flaws and limitations in
ucts of particular social groupings at particu- the results” (p. 4). For instance, COVID-19
lar times and in particular places. . . . Labelling infection and fatality rate estimates have
(Continued)

bel65796_ch14_333-358 339 01/13/22 12:32 PM

340 PART IV Bringing It All Together

significant variations between countries, par- comparisons to data from other epidemics,
tially owing to different decisions made about other geographical areas, and other causes
the measurement process and the social con- of death.
texts from which the numbers are drawn. 5. Social patterns shape numbers: Social factors
3. Counting is not straightforward: “Any time influence the data in terms of the direction a
we count anything, we are inevitably forced problem takes and how it is understood. For
to choose—to decide what to count and how example, Best indicates that about a month
to go about counting” (p. 6). In the case of after COVID-19 was declared a pandemic,
COVID-19, this leads to both an under- and commentators began drawing attention to
over-counting of cases. As an example, an how the disease was disproportionately im-
American who might be sick with COVID-19, pacting ethnic minority groups. It was not a
but who does not have health insurance cov- matter of inter-group biological differences,
erage, is typically less likely to seek treatment though: ethnicity was being used as a proxy
and be counted compared to someone who for social class. Best suggests that for differ-
does have coverage. ent reasons, such as the messiness of measur-
4. All comparisons involve choices: “It is often ing social class, advocates drawing attention
hard to know what to make of a number. to ethnic inequalities, and Americans’ greater
Is it large or small? Ordinarily, placing a receptivity toward making racial over class
figure in some interpretive context requires distinctions, ethnicity became the initial gov-
making some sort of comparison” (p. 8). With erning narrative.
COVID-19, this has involved making statistical

Research methods and rather than … carefully chosen fundamen-

tal assumptions. Frequently methodological
­epistemological and choices are steered by considerations of a prac-
­ontological considerations tical nature … and are just slogans and aspira-
tions. … (p. 275)
That each research method does not carry with it a
full cluster of epistemological and ontological com- Again, even when there are discernible links be-
mitments can be seen in studies on social research tween research methods and assumptions about
itself. For example, Platt’s (1986) historical research on knowledge, the connections are not absolute.
American sociology suggested that the alleged connec- Further evidence of the autonomy of research
tion between conservative functionalist theory (which methods is the fact that both quantitative and quali-
is often associated with positivism) and survey research tative approaches may be employed in a single piece
is greatly exaggerated. Her research indicated that “the of research, a point that will be illustrated later in
two originated independently, and that leading func- this chapter.
tionalists had no special propensity to use surveys and
leading surveyors no special propensity for function-
alism” (p. 527). Moreover, Platt’s (1996) general con-
Problems with the quantitative/
clusion on the use of research methods in American qualitative contrast
sociology between 1920 and 1960 is very revealing:
The contrasts between quantitative and qualitative
Research methods may on the level of theory, research drawn in Chapter 9 suggest a sharp distinc-
when theory is consciously involved at all, tion between the two (see, in particular, Table 9.1).
reflect intellectual … post hoc justifications Here is a more nuanced view.

bel65796_ch14_333-358 340 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 341

Behaviour versus meaning taken place. The analysis of quantitative survey data
A distinction is sometimes drawn between a quan- is often more exploratory than is generally appreci-
titative focus on behaviour and a qualitative focus ated and consequently offers many opportunities for
on meaning. Yet, as noted, quantitative research generating theories.
frequently seeks to study meanings in the form of The common depiction of quantitative research
attitudes. In recent decades the social sciences in as an exercise in theory testing also fails to take
general have increasingly recognized the importance into account how often its findings can result in
of examining not only what people do, but also how new theoretical departures and contributions.
they think and make sense of the world. The “cog- Reflecting on his career in survey research, Glock
nitive revolution” in psychology has made people’s (1988, pp. 45–46) provided the following example
thoughts and feelings the topic of extensive research. from his study of American church involvement.
The use of rational choice theory and game theory in Having found that women, older persons, the poor,
economics, political science, and sociology reflects and those without families were more religiously
a similar emphasis on examining how people think active than people not in those categories, he sug-
and make choices. Many different perspectives in gested that such people might have more time
the social sciences now acknowledge the importance than others to become involved. Alternatively (or
of meanings and interpretations in understanding in addition), these religiously active people might
social phenomena. be compensating for some kind of deprivation rel-
Looking at the other side of the divide, qualita- ative to their counterparts—a conflict view that
tive researchers frequently, if not invariably, exam- Marx would have appreciated. Since the existing
ine behaviour, although they also pay close attention data did not allow him to rule out either explana-
to the context in which it occurs. Qualitative re- tion, he collected new data to test them: in this way
searchers often seek to interpret people’s behaviour his original quantitative data led to new research
in terms of the norms, values, and culture of the ideas. Quantitative research commonly follows
group or community in question. In other words, this path.
quantitative and qualitative researchers alike are
Numbers versus words
typically interested both in what people do and
in what they think, even if they go about their in- Even the most basic distinction between quantitative
vestigations in different ways. The degree to which and qualitative research—numbers versus words—is
quantitative and qualitative research reflect the problematic. Qualitative researchers will sometimes
“behaviour versus meaning” distinction should not undertake to quantify a limited amount of their
be overstated. data in an effort to determine the generality of the
phenomena they are describing. After observing
Theories and concepts tested in doctor–patient interactions in public and private on-
research versus those emerging cology clinics, for example, Silverman (1985) quanti-
from the data fied some of his data to show that patients in private
A related point concerns the characterization of clinics have more influence over what goes on in
quantitative research as driven by theory testing. the consultations. Still, Silverman warned that such
Experimental investigations probably fit this model quantification should reflect the research partici-
well, but survey-based studies are often more ex- pants’ own ways of understanding their social world.
ploratory in nature. Although concepts have to be It has often been noted that qualitative research-
measured, specific hypotheses are frequently not ers engage in “quasi-quantification” when they use
identified in advance. Surveys ask so many questions, terms such as “many,” “often,” and “some” (see page
and there are so many possible correlations and ways 344). The only difference in studies like Silverman’s
of organizing the findings, that hypotheses are often is that the researcher makes such estimates of fre-
not formulated until after the data collection has quency more precise.

bel65796_ch14_333-358 341 01/13/22 12:32 PM

342 PART IV Bringing It All Together

Artificial versus natural As we noted in Chapter 11, focus group research

The distinction between artificial and natural is sim- is often described as more natural than qualitative
ilarly open to criticism. It is often assumed that be- interviewing because it resembles the way people
cause much quantitative research employs research discuss issues in real life. Natural groups (people al-
instruments (such as questionnaires) that offer only ready known to each other) are often used to empha-
limited and indirect indicators of people’s lives, its size this element. Whether focus group participants
account of the social world is artificial. Qualitative see their interaction as “natural” is unclear, however.
research tends to be considered more naturalistic in In particular, since the participants are often paid
that it focuses on observing people in their natural strangers who have travelled some distance to dis-
settings, behaving as they normally do. This would cuss topics they rarely if ever talk about, it’s clear that
seem to be particularly true of ethnographic re- this sort of research may not be as natural as some
search, since the participant observer studies people have assumed.
in their normal social worlds and contexts, as they In participant observation, the researcher can be
go about their everyday activities. However, when a source of interference, making the research situ-
qualitative research is based on interviews, the “nat- ation less natural than it might appear. It’s difficult
ural” label is less applicable. Interviews still have to to estimate how much impact reactivity has on the
be arranged and interviewees have to be taken away research findings in such cases, but once again the
from activities they would otherwise be engaged in, naturalism of such studies is questionable. When
even when the interviewing style is of the more con- the ethnographer also engages in interviewing (as
versational kind. Furthermore, little is known about opposed to casual conversation), the naturalistic
interviewees’ reactions to and feelings about being quality is definitely compromised, although the re-
interviewed. search will probably be somewhat less artificial than
Phoenix (1994) reported on the responses of it would be if quantitative methods were used.
interviewees to in-depth interviews in connection These observations cast doubt on the rigidity of
with two studies: one focusing on mothers under the quantitative/qualitative distinction. Once again,
the age of 20 and the other on the social identities this is not to suggest that the usual contrasts are un-
of young people. Although many of the partici- helpful: only that they should not be exaggerated.
pants seemed to enjoy being interviewed, they were Quantitative and qualitative research strategies are
clearly conscious of the fact that they had been not absolutely divergent.
engaged in interviews rather than conversations.
This was revealed in some of their replies. For ex- Mutual analysis
ample, Phoenix reported one young Black woman
as saying that she had liked the interview, adding: The divisions between quantitative and qualitative
“I had the chance to explain how I feel about cer- research are also undermined when one approach is
tain things and I don’t really get the opportunity to used to analyze research conducted using the other.
do that much” (p. 61). Another interviewee said it
was a “good interview” and added: “I have never A qualitative approach to
talked so much about myself for a long time, too ­quantitative research
busy talking about kids and their problems” (p. 61). There has been growing interest in using some of
While such qualitative interviews are clearly valu- the methods associated with qualitative research to
able in allowing the perspectives of people whose examine quantitative research. In part, this trend
voices are normally silent to surface, they don’t fit reflects qualitative researchers’ growing interest in
the definition of “naturalistic” used by critics of ethnography. The qualitative study of quantitative
quantitative social science. Thus, it would be incor- research is part of this trend because it reveals that
rect to assume that artificiality is a problem only in the written account of research constitutes not only
quantitative research. the formal presentation of findings but an attempt to

bel65796_ch14_333-358 342 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 343

persuade readers of the credibility of those findings. criteria resulted in the exclusion of 279 out of the 365
This is true of the natural sciences as well; research books he had considered. A second crucial issue was
by Gilbert and Mulkay (1984) showed that scien- the coding of the studies. Hodson stressed the im-
tists, when writing up their findings, took pains to portance of knowing the subject area, adopting clear
demonstrate that proper procedures were followed. coding rules, and pilot testing the coding schedule.
They also learned from interviewing scientists that In addition, he recommended checking the reliabil-
the research process was influenced by the scientists’ ity of coding by having 10 per cent of the documents
personal biographies. coded by two people. The process was time-consum-
One way in which a qualitative approach has ing; each book took at least 40 hours to code.
been applied to quantitative studies is through what Among the many attractions of this approach is
Gephart (1988) called ethnostatistics: “the study of the fact that it allowed a single researcher to investi-
the construction, interpretation, and display of sta- gate a varied set of organizations and provided more
tistics in quantitative social research” (p. 9). For ex- in-depth data than is typically gathered by quanti-
ample, Gephart showed that the use of statistics in tative researchers. It also allowed for the testing of
itself can be regarded as a rhetorical device, because hypotheses derived from established theories: for
quantification makes social research look more like example, the “technological implications” approach,
natural science—an association that tends to give it which sees technology as having an impact on the
greater legitimacy and credibility. This is just one ex- experience of work (Hodson, 1996). However, the
ample of the way qualitative research can shed light loss of social context is likely to make Hodson’s ap-
on the nature of quantitative research. proach unattractive to many qualitative researchers.
Of particular significance is Hodson’s remark
A quantitative approach to that “the fundamental contribution of the systematic
­qualitative research analysis of documentary accounts is that it creates
When Hodson (1996) conducted a content analysis of an analytic link between the in-depth accounts of
workplace ethnographies (discussed below), he essen- professional observers and the statistical methods of
tially applied a quantitative content analysis to qualita- quantitative researchers” (Hodson, 1999, p. 68). In
tive research. This type of research, sometimes called other words, the application of quantitative meth-
meta-ethnography, may have potential in other areas ods to qualitative research may provide a meeting
of social research where ethnography has been a pop- ground for the two research strategies.
ular method, such as the study of social movements,
religious sects, and cults. Hodson’s approach was one Quantification in qualitative
solution to the problem of comparing ethnographic
studies in a given area. However, he largely ignored research
contextual factors in order to explore relationships be- As we noted in Chapter 9, perhaps the most basic
tween variables abstracted from the studies. difference between quantitative and qualitative re-
Among the key issues in a study of this kind is search is the “numbers versus words” distinction. Yet
the selection of works suitable for analysis. Hodson most qualitative research does contain some quanti-
chose to analyze just books, excluding articles be- fication. Three observations are worth making about
cause they usually contain only a limited amount of quantification in the analysis and writing up of qual-
information. To qualify for selection, a book had to itative data.
meet three criteria: “(a) the book had to be based on
ethnographic methods of observation over a period Thematic analysis
of at least 6 months, (b) the observations had to be Chapter 13 noted that qualitative data analysis
in a single organization, and (c) the book had to often takes the form of a search for themes in tran-
focus on at least one clearly identified group of work- scripts or field notes. The choice of themes to look
ers” (Hodson, 1999, p. 22). The application of these for is often determined by how frequently certain

bel65796_ch14_333-358 343 01/13/22 12:32 PM

344 PART IV Bringing It All Together

incidents, words, phrases, and so on recur. This pro- a further example, Aurini, Missaghian, & Pizarro
cess may also account for the prominence given to Milian (2020) used quantification to a limited
some themes over others. In other words, a kind of degree in their qualitative study of how Canadian
implicit quantification probably influences both the upper-middle-class parents structure and ra-
identification of themes and the elevation of some tionalize their children’s after-school activities.
themes over others. Tables of frequencies and percentages were used
to summarize their data on children’s extracur-
Quasi-quantification in qualitative ricular activities and their parents’ rationales for
research selecting those activities. This approach allowed
As we have noted, qualitative researchers engage in the researchers to offer a big-picture overview of
“quasi-quantification” when they use terms such as parental choices, which was then followed up by
“many,” “frequently,” “rarely,” and “some,” which by a deeper look at the data by presenting excerpts
definition are based on the relative frequency of the from interviews conducted with the participants.
phenomenon of interest. As expressions of quantity, Using actual numbers can counter the criticism
however, these terms are imprecise, and it’s often that qualitative data are too anecdotal and do noth-
difficult to tell why they are being used at all. When ing to indicate the extent to which certain beliefs
quantification is needed to support a point, it may be or behaviours occur. The precision of numbers also
better to use actual numbers. makes them superior to the estimates of frequency
that must be inferred from quasi-quantification
Combatting anecdotalism through terms such as “some” or “many.”
limited quantification The use of “layering” has been presented as an-
Qualitative research is sometimes criticized for other way of addressing concerns about anecdotalism
the anecdotal nature of its data, which leaves (McAleese & Kilty, 2019). Layering involves bringing
the reader with no way of assessing their gener- together statements and stories from participants in
alizability. Brief sequences of conversation, snip- multiple research projects and considering which
pets from interview transcripts, and accounts of narratives are supported. Given that qualitative re-
encounters between people say little or nothing searchers often work with a small number of cases,
about the prevalence of the phenomena that such layering can lend credibility to researchers’ findings
evidence is supposed to indicate. There is also by demonstrating that a pattern holds across a series
the related risk that the researcher may attribute of studies. Layering is similar to replication in quan-
undue significance to a particularly striking state- titative research, but with layering, qualitative re-
ment or unexpected activity. searchers seek to validate the narrative even though
At least partly in response to these problems, the research methods may differ from study to study.
qualitative researchers sometimes undertake McAleese and Kilty (2019) argue that “By layering
a limited quantification of their data. Gabriel stories from different people in different contexts
(1998) collected 377 stories about organizational over time, qualitative research can solidify a partic-
culture in the course of 126 interviews in five or- ular narrative, which may then be considered a more
ganizations. He identified different types of sto- valid and reliable one” (p. 836). Although findings
ries and noted the frequency of each type. For from small qualitative studies can be credible and
instance, there were 108 comic stories (usually a trustworthy in their own right, layering helps to ad-
mechanism for the disparagement of others); 82 dress the critique that qualitative findings are “just
epic stories (survival against the odds); 53 tragic anecdotal.”
stories (undeserved misfortune); and 40 gripes We may see greater use of quantification in
(personal injustices). The stories could have been qualitative research in the future as a result of the
presented as anecdotes, but simple counting con- growing use of software programs for qualitative
veyed a clear sense of their relative prevalence. As data analysis. Most of the major software programs

bel65796_ch14_333-358 344 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 345

make it possible to produce simple counts for the • Particular research methods are associated
use of a particular word or the incidence of a coded with particular epistemological and ontologi-
theme. In many cases, they can also produce simple cal positions.
cross-tabulations—for example, showing which sex • Quantitative and qualitative research strate-
uses the passive voice more often. Software programs gies are inherently incompatible on epistemo-
also assist with managing large amounts of qualita- logical and ontological grounds.
tive data, which makes layering as an analytical ap-
proach more feasible. These two positions are briefly reviewed below.

The embedded methods argument

Multi-strategy research
Advocates of the first position, outlined earlier in
The ways in which quantitative and qualitative re- this chapter, argue that every research method is
search can complement each other can be seen most firmly rooted in a particular set of epistemological
clearly in multi-strategy research, which by defini- and ontological commitments. According to this
tion involves combining the two modes of investiga- view, a decision to employ (for example) partici-
tion. However, there are two basic arguments against pant observation is not simply a choice about data
this sort of research. collection but a commitment to an epistemological
position that is consistent with interpretivism and
Arguments against multi-strategy incompatible with positivism or a natural science
research model in general.
Arguments against multi-strategy research tend to This view has led some writers to argue that
be based on the following assumptions: multi-strategy research is not feasible. An ethnographer

Research in the News

COVID-19 and intimate partner violence
Early in the COVID-19 pandemic, women’s exposing themselves and their children to the
shelters in Manitoba reported an increased coronavirus. Also, women who have lost em-
demand for their services (Hoye, 2020). ployment during the pandemic may not have
Researchers suspect that more time spent at the financial means to leave an abusive part-
home during the pandemic with a violent part- ner. Nixon and Henriquez’s research will in-
ner is creating greater risks of victimization clude participants from both rural and urban
among abuse survivors. To better understand areas and different ethnic groups, as well as
this issue, Kendra Nixon, an associate professor refugees and immigrants. Partnering with the
in the Faculty of Social Work at the University Manitoba government’s family violence pre-
of Manitoba, and Nadine Henriquez, an assis- vention program, the researchers hope their
tant professor in the Faculty of Health Studies findings will improve services and help inform
at Brandon University, will be interviewing policy decisions. Nixon points out, “It’s import-
abuse survivors and service providers. Nixon ant, now that we have this opportunity, that we
believes that there are several pandemic-re- look at the research, that we look at the risk so
lated factors putting survivors at greater risk. we can develop better policies and practices
For example, women may be concerned about should a future pandemic or a future surge of
leaving their home for a shelter and potentially COVID-19 surface” (Hoye, 2020).

bel65796_ch14_333-358 345 01/13/22 12:32 PM

346 PART IV Bringing It All Together

may collect questionnaire data about a slice of social life it is by no means clear that quantitative research and
not amenable to participant observation, but this does qualitative research are fully separate paradigms. As
not represent an integration of quantitative and quali- we have suggested throughout this chapter, there are
tative research because the epistemological positions of areas of overlap and commonality between them.
the two methods are irreconcilable. The chief difficulty
with this kind of argument is that the idea of research Two positions in the debate
methods carrying fixed epistemological and ontologi-
cal implications is very difficult to sustain. over quantitative and
­qualitative research
The paradigm argument
There are two opposing arguments regarding the
The paradigm argument is closely related to the pre-
question of whether quantitative and qualitative re-
vious one. It conceives of quantitative and qualita-
search can be combined.
tive research as paradigms. A paradigm is a set of
beliefs and assumptions about how the world works
• An epistemological argument, as in the em-
and how knowledge of it is to be gained. Kuhn (1970)
bedded methods and paradigm positions dis-
popularized the term in his portrayal of natural sci-
cussed above, sees quantitative and qualitative
ence as going through periods of revolution, whereby
research as grounded in incompatible episte-
“normal” science (science carried out in terms of the
mological principles (and ontological ones too,
prevailing paradigm) is increasingly challenged by
though these tend to be given less attention).
new findings inconsistent with the assumptions and
According to this version, multi-strategy re-
established findings of the discipline. The increas-
search is not possible in principle.
ing frequency of such anomalies eventually sparks a
• A technical argument perceives research meth-
crisis, which in turn sparks a revolution. The period
ods as independent of any specific epistemolog-
of revolution comes to an end when a new paradigm
ical position. According to this view, a research
gains acceptance and a period of the new “normal”
method from one research strategy can be
science sets in. An important feature of paradigms
pressed into the service of another: quantitative
is that they are incommensurable: inconsistent with
and qualitative methods can be fused. This is
each other because of their divergent assumptions
the position taken by most researchers whose
and methods. Disciplines in which no paradigm has
work is mentioned in the next section.
emerged as pre-eminent, such as sociology and other
social sciences, are deemed “pre-paradigmatic.”
The technical argument views the two research strat-
The paradigm argument maintains that quanti-
egies as compatible, and multi-strategy research as
tative and qualitative research approaches are based
both feasible and desirable. It is in that spirit that we
on incompatible paradigms in which epistemologi-
will now consider the ways in which quantitative and
cal assumptions, values, and methods are inextri-
qualitative research can be combined.
cably intertwined (Morgan, 1998). Therefore, when
researchers combine participant observation with a
questionnaire, they are not really combining quanti- Approaches to multi-strategy
tative and qualitative research, since the paradigms research
are incompatible: the integration is only superficial
and within a single paradigm. Hammersley (1996) proposed three approaches to
The problem with the paradigm argument, as with multi-strategy research:
the embedded methods claim, is that it rests on a con-
nection between method and epistemology that has • Triangulation: the use of quantitative research
not been demonstrated. Moreover, although Kuhn to corroborate qualitative research findings or
(1970) argued that paradigms are incommensurable, vice versa

bel65796_ch14_333-358 346 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 347

• Facilitation: the use of one research strategy to 824 12- to 17-year-olds. The first part was an
aid research using another interview and the second was a questionnaire
• Complementarity: the use of two different re- designed to elicit more sensitive information.
search strategies so that diverse aspects of an
investigation can be combined The results achieved with the two research strategies
were mutually reinforcing. The qualitative findings
The logic of triangulation showed age differences in attitudes toward designer
Triangulation entails using more than one method drinks and other types of alcohol: the youngest
in the study of social phenomena. The term has been people (12- and 13-year-olds) tended to adopt a gen-
used somewhat more broadly by Denzin (1970, p. erally experimental approach; the 14-and 15-year-
310) to refer to an approach that uses “multiple ob- olds thought of drinking as a way of having fun and
servers, theoretical perspectives, sources of data, losing inhibitions, and felt that designer drinks met
and methodologies,” but the emphasis has tended to their needs well; the oldest group (16- and 17-year-
be on methods of investigation and sources of data. olds) were mainly concerned with appearing mature
One of the reasons Webb, Campbell, Schwartz, & and establishing relationships with the other sex,
Sechrest (1966) called for greater use of unobtrusive and tended to think of designer drinks as targeted
methods is their potential for triangulation (see Box at immature younger drinkers. These connections
6.5). Triangulation can operate within and across with age were confirmed by the quantitative evi-
research strategies. It was originally conceptualized dence, which also corroborated the suggestion from
by Webb et al. (1966) as a way of developing addi- the qualitative evidence that the consumption of de-
tional measures, resulting in greater confidence in signer drinks was largely associated with a desire to
findings, and was strongly associated with a quan- get drunk.
titative research strategy. However, triangulation In this study the use of triangulation appears
can also take place within qualitative research. In to have been planned, but this need not always be
fact, ethnographers often check their observations the case. Researchers may start out using different
with interview questions to look for possible misun- research strategies for different purposes but in the
derstandings. Bloor (1997) reported that he tackled process discover that they have generated quantita-
the process of death certification in two ways: inter- tive and qualitative findings on related issues, and
viewing clinicians responsible for certifying causes that they can treat their overlapping results as a tri-
of deaths, and asking the same people to complete angulation exercise.
dummy death certificates based on case summaries Whether triangulation is planned or unplanned,
he had prepared. Increasingly, triangulation refers there is always the possibility that the two sets of
to the cross-checking of findings from both quanti- findings will not corroborate one another. In the
tative and qualitative research (Deacon, Bryman, & event that the results are inconsistent, the solution is
Fenton, 1998). not to arbitrarily decide which set of findings is cor-
An illustration of a triangulation approach is rect. Usually, further research is required to resolve
Hughes et al.’s (1997) study of the consumption of the matter.
“designer drinks” (fortified wines and strong white Deacon and colleagues (1998) used several quan-
ciders) by young people. The authors used two main titative and qualitative research methods to gather
research methods: information, without intending to conduct a triangu-
lation exercise. But analysis of their data revealed an
• Qualitative method: eight focus groups with 56 inconsistency: the quantitative data suggested a broad
children and young adults, with each discus- consensus between journalists and social scientists on
sion lasting around two hours the reporting of social scientific research in the media,
• Quantitative method: a survey administered but the qualitative findings suggested some conflict
in two parts to a multi-stage cluster sample of of approaches and values between the two groups.

bel65796_ch14_333-358 347 01/13/22 12:32 PM

348 PART IV Bringing It All Together

Instead of favouring one set of findings over the other, methods of data collection, qualitative research
the researchers re-examined the data and found that is often a source of hypotheses that can be
social scientists tended to answer questions about tested using a quantitative strategy. An exam-
media coverage of their work differently, depending ple is Phelan’s (1987) research in which she con-
on the research instrument. In response to survey ducted qualitative interviews and conversations
questions, they typically expressed relief that the cov- with people attending a treatment program
erage was not as bad as they had expected, whereas for families dealing with incest. After a con-
in interviews they tended to emphasize “war stories” siderable amount of qualitative data had been
of wounding encounters with the media. In general, collected, Phelan became aware of differences
then, the questionnaires showed social scientists to be between biological fathers and stepfathers in
relatively pleased with the reporting of their research, how they interpreted their incestuous acts, and
but their replies were more negative when they were in the frequency with which certain behaviours
encouraged to reflect on specific problems in the past. occurred. Quantifiable data were collected
Another example of triangulation can be found through interviews with family counsellors
in Fenton, Bryman, & Deacon’s (1998) study of the that supported her hypothesis that “the pro-
same topic. It employed both quantitative and qual- cess of incest in structurally different families
itative methods: may vary” (Phelan, 1987, p. 39). Similarly, Bell’s
(2007) study of western Canadian separatism
1. content analysis of news and current affairs started out as a qualitative investigation involv-
coverage (local and national newspapers, TV, ing attendance at separatist events and hanging
and radio) out with separatists. Analysis of these data led
2. mailed questionnaires soliciting social scien- to the hypothesis that support for separatism
tists’ views about media coverage and their was associated with a neo-liberal ideology and
own practices (674 replies) a partisan dislike for the ruling federal Liberal
3. mailed questionnaires soliciting the views of Party, a view that was later substantiated in a
social scientists identified in the content analysis multiple regression analysis using survey data.
as having received media coverage (123 replies) • By aiding measurement. The in-depth knowl-
4. semi-structured interviews with 20 of the edge of social contexts acquired through
latter qualitative research can be used to inform
5. semi-structured interviews with 34 journalists the design of survey questions for structured
identified in the content analysis interviewing and self-completion question-
6. semi-structured interviews with 27 represen- naires. Johnson and colleagues (2003) used
tatives of funding bodies and government data from unstructured interviews to develop
7. observing journalists’ movements and activi- highly structured questions for later interviews
ties at three conferences with adolescent smokers. Walklate (2000) ex-
8. focus group analysis of audience reception of plained that for her research on fear of crime
media coverage (13 focus groups) and safety issues in high-crime areas, a survey
using traditional questions about victimiza-
tion was used. However, the questions were
Qualitative research facilitates amended to reflect the local context following
quantitative research six months of interviews, ethnography, and ex-
Here are two ways in which qualitative research can amination of local newspapers. Similarly, the
serve as a guide for quantitative research: survey questions that Bell (2007) used in his
study of western Canadian separatism were
• By suggesting hypotheses. Because of its ten- prompted in part by his experiences in gather-
dency toward unstructured, open-ended ing qualitative data.

bel65796_ch14_333-358 348 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 349

Quantitative research facilitates Priority

qualitative research Quantitative Qualitative

One way in which quantitative research can prepare Preliminary M1 M2

the ground for qualitative work is through the se- Sequence
Follow-up M3 M4
lection of people to be interviewed. This can occur
in several ways. In the case of Fenton et al.’s (1998) FIGURE 14.1 Morgan’s classification of
research on the reporting of social science research approaches to multi-strategy research
in the mass media, discussed above, a media con-
tent analysis (method 1) was used as a source of One difficulty with using Morgan’s typology is
data. However, it also served as a means of identify- that it may be difficult to know whether quantitative
ing journalists who had reported relevant research or qualitative research had priority, and which one
(method 5). In addition, replies to questions in the came first in the sequence. Sometimes research uses
general survey of social scientists (method 2) were both strategies and several methods, but no single
used to help identify two groups of social scientists, one is dominant.
those with high and those with low levels of media
Static and process features
coverage of their research, who were then inter-
viewed with a semi-structured approach (method Together, the characteristics listed in Table 9.1 sug-
4). Similarly, Jamieson (2000) administered a ques- gest that quantitative research presents a static pic-
tionnaire to a sample of young men on their crimi- ture of social life, while qualitative research presents
nal activity. On the basis of their replies, qualitative a more dynamic picture, emphasizing process. In
interviews were conducted with equal numbers of some contexts “static” may be considered a negative
young men in each of three categories: those who term, but in this case it’s positive. In fact, it is the reg-
had never committed a criminal offence; those ularities revealed by quantitative research that allow
who had offended but not recently; and persistent for the analysis of process. A multi-strategy research
offenders. approach makes it possible to combine the two. An
illustration is provided by MacKinnon and Luke’s
Filling in the gaps (2002) study of cultural change from 1981 to 1995.
When the researcher cannot rely on either a quanti- Their main method of data collection was survey re-
tative or a qualitative method alone, a multi-strategy search, but to complete the picture they also examined
approach can be used to fill in the gaps. For example, census and public opinion data, as well as newspaper
ethnographers may employ structured interviewing articles on historical events. They found that levels
or self-completion questionnaires if not everything of homophobia and anti-Semitism decreased over
they need to know is accessible through direct ob- that period, but that sympathy for Indigenous people
servation, or if they have difficulty gaining access to also declined. They cautioned that their small sample
certain groups of people. (70) and what were essentially “good guesses” meant
Morgan (1998) suggested a two-step approach to that their conclusions, “like the interpretations of an
planning multi-strategy research: ethnographer [were] subject to alternative interpreta-
tions by others” (MacKinnon & Luke, 2002, p. 332).
• The priority decision. Will the principal da-
ta-gathering tool be qualitative or quantitative? Researcher and participant
• The sequence decision. Should the “complemen- perspectives
tary” method be used first, as a preliminary to Sometimes researchers want to gather two kinds of
the principal method, or second, as a follow-up? data: qualitative data that show the general perspec-
tives of the people they are studying, and quantita-
These decisions yield four possible types of research tive data on specific issues. For example, Milkman
design (see Figure 14.1). (1997) was interested in the meaning of industrial

bel65796_ch14_333-358 349 01/13/22 12:32 PM

350 PART IV Bringing It All Together

work: in particular, whether factory conditions differ in their levels of education), which in turn
had changed since the 1950s, when they were por- has implications for the jobs that people in different
trayed very negatively. To find out, she employed ethnic groups attain. Qualitative work could examine
semi-structured interviews and focus groups with how different ethnic groups perceive education, which
General Motors production workers. She was also could lead to a fuller understanding of education as
interested in a “buyout” plan that the company’s an intervening variable. It could also provide leads to
management introduced in the mid-1980s after it a different theoretical perspective on the general topic
had initiated a variety of changes to work practices. of ethnicity, education, and occupation.
The plan offered workers the opportunity to give up Another illustration is provided by Barnard and
their jobs for a substantial cash payment. In 1988 Frischer’s (1995) research on HIV-related risk be-
and again in 1991, Milkman carried out a question- haviour among drug injectors. Structured interviews
naire survey of workers who had accepted the com- with 503 injectors revealed that “females report[ed]
pany’s buyout offer. The surveys inquired about the significantly higher levels of needle sharing, sexual
reasons for taking the buyout, how those taking it activity, and AIDS awareness than their male coun-
had fared since leaving General Motors, how they terparts,” and that “women . . . co-habiting with
felt about their current employment, and differences sexual partners who [were] themselves injectors,
among social groups (in particular ethnic groups) in [were] particularly likely to report high levels of
current earnings relative to those at General Motors. risk behaviour and also AIDS awareness” (p. 357).
What produced this relationship between gender,
The problem of generality risk behaviour, and cohabitation? Semi-structured
As we noted above, there is a tendency in qualita- interviews with 73 injectors suggested that the rela-
tive research to present findings without evidence tionship between these variables could be explained
indicating how typical they are, or how the group “by the tendency for women to be in sexual rela-
under study compares with some larger popula- tionships with men who themselves inject and with
tion. Partly to counter this problem, Tastsoglou and whom they are unlikely to use condoms” (p. 360).
Miedema (2003) used quantitative national data on Here qualitative data were used to shed light on rela-
women’s volunteer work to compare their admittedly tionships among variables derived from quantitative
non-random sample of immigrant women with the research.
Canadian population; they found that the immi-
grant women actually volunteered more than the Studying different aspects of a
national average. phenomenon
There is a tendency to think of quantitative research
Qualitative research may help to as best suited to the investigation of “macro” phe-
interpret the relationship between nomena (such as social mobility and social stratifi-
variables cation) and qualitative research as more appropriate
A frequent problem for quantitative researchers is for “micro” issues (such as interactions between wait
how to explain relationships between variables. One staff and their customers). The macro/micro dis-
strategy is to look for an intervening variable: one tinction can also be discerned in Table 9.1. In the
that is influenced by the independent variable but example outlined in Box 14.3, Wajcman and Martin
in turn has an effect on the dependent variable. For (2002) used a quantitative method in the form of a
example, in a relationship between ethnicity and oc- questionnaire survey to explore the career patterns
cupation, education may be an intervening variable: of male and female managers. However, they also
carried out qualitative, semi-structured interviews
ethnicity → education → occupation
to explore how managers made sense of their career
This sequence implies that the variable ethnicity has patterns in terms of their identity. Thus, the choice of
an impact on education (for example, ethnic groups methods was determined by the research question.

bel65796_ch14_333-358 350 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 351

Box 14.3 Combining survey research and qualitative

interviewing in a study of managers
in terms of career orientations and attitudes:
contrary to what many had anticipated, wom-
en’s career experiences and orientations were
not found to be distinctive. They then exam-
ined the qualitative interviews for narratives of
Weekend Images Inc/iStockphoto

identity and found that both male and female

managers depicted their careers in “market”
terms (they responded to the requirements of
the managerial labour market by developing
their skills and experience). But whereas for
men narratives of career meshed seamlessly
Why do the findings from structured inter- with narratives of domestic life, for women
views sometimes differ from those derived from there was a disjuncture. Female managers
semi-structured or unstructured interviews? found it harder to reconcile managerial iden-
Wajcman and Martin (2002) used a question- tities with domestic ones and more often had
naire to survey male and female managers to opt for one or the other. In this way, choices
(470 in total) in six Australian companies and about career and family were still gendered.
conducted semi-structured interviews with 136 This example illustrates how a multi-strategy
managers in each company. The survey evi- approach can work to reveal much more than
dence showed that male and female manag- could have been gleaned through the use of
ers were generally more similar than different one method alone.

The use of multi-strategy research to study differ- anticipated that these would tell us about dif-
ent aspects of a phenomenon can also be seen in the ferent aspects of family obligations. … [O]ur
family obligations studies conducted by Finch (1985) view was that an understanding of kin obli-
and Mason (1994). Focusing on the distribution gations in practice would require an analy-
within families of the obligation to care for relatives, sis of the relationship between the two data
this research had two main data collection elements: sets and the social processes they expressed.
a survey of a sample of nearly 1000 people by struc- (Mason, 1994, pp. 90–91)
tured interview, and semi-structured interviewing
What were the two sets of research questions to which
with 88 people. A major component of the survey
Mason referred? The survey was designed to elicit in-
interviewing was the use of the vignette technique
formation about “‘the proper thing to do’ for rela-
described in Box 5.4. Mason described the purpose
tives in a variety of circumstances” (Mason, 1994, p.
of integrating quantitative and qualitative research
90). Through the semi-structured interviews, Mason
as follows:
tried “to discover what people actually did in prac-
From the beginning … we were using the tice for their own relatives, and also the processes by
two parts of our study to ask different sets which they came to do it and make sense of it: did a
of questions about family obligations. Not sense of obligation or responsibility have a role in the
only were we employing different methods to process? How did people in practice work out what
generate different types of data, but we also to do for their kin, or ask of their kin?” (p. 90).

bel65796_ch14_333-358 351 01/13/22 12:32 PM

352 PART IV Bringing It All Together

Box 14.4 Qualitative and quantitative methods combined to

study children’s knowledge of learning difficulties
Nowicki, Brown, & Stepien (2014) combined qual- understanding that often cannot be verbalized
itative and quantitative approaches in a study but may manifest itself in recognition tasks. To
that examined how children aged 9–11 viewed do this the students were given cards on which
the causes of classroom learning difficulties. were written each of the 42 statements. They
The study was conducted at five schools in a were asked to place the cards in groups based
medium-sized Ontario city where students with on similarity, and to produce labels for each
learning issues were fully integrated with other group they had created. The cluster analysis
students in the classroom. The researchers em- showed that five distinct clusters of statements
ployed the technique of concept mapping, which could be identified, and that the clusters were
combines qualitative interviewing with cluster meaningful, coherent categories that indicated
analysis, a statistical procedure. The qualitative a surprising level of understanding of learning
interviewing of individual students formed the difficulties. The clusters had the themes fate,
first phase of the study, with the focal question family stress, neurological problems, informa-
being “Why do some children have learning diffi- tion processing difficulties, and issues related
culties?” After editing for redundancy and clarity, to motivation and instruction. It turned out
this produced 42 unique statements, or a mean that the students were more astute in identi-
of 1.2 responses per student. This result was in fying a diverse range of possible causes of
keeping with previous qualitative research that learning difficulties when given a list of state-
indicated that children at this age have a very ments produced by their peers than they were
limited store of “explicit” knowledge of learning in verbalizing the causes on their own. This
difficulties, which refers to knowledge that can study illustrates how combining qualitative and
be verbalized on one’s own. quantitative methods may produce more infor-
But the researchers also tested the children’s mative results than employing just one of those
“implicit” knowledge of the topic, a type of approaches.

In multi-strategy research, then, the different to write an item on social science research?
methods may be geared to addressing different (methods 5 and 7).
kinds of research questions. The study on the re- • Questions about social scientists’ attitudes to media
porting of social science research in the British mass reports of research. This includes reporting of re-
media by Fenton et al. (1998), discussed earlier, is a search in general (method 2) and of their own re-
further example of a project that used quantitative search (methods 3 and 4). Method 4 was designed
and qualitative research to answer different types of to allow the findings reached through method 3
questions: to be elaborated and more fully understood.
• Questions about reception. For example: How
• Questions about coverage. For example: How do readers/viewers interpret media reporting
much coverage is there of social science re- of social science research? (method 8).
search? What gets covered? Where? (method • Questions about the communication environ-
1). ment. For example: What are the policies of
• Questions about the production of media cover- universities, government departments, and
age. For example: What kinds of attributes do funding bodies concerning the media report-
journalists look for when deciding whether ing of research? (method 6).

bel65796_ch14_333-358 352 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 353

This form of multi-strategy research entails for that led them to study a different set of archival
making decisions about which kinds of research materials. This, in turn, produced further research
question are best answered using a quantitative questions that were well-suited for study through
method and which by a qualitative method. (Box interviews.
14.4 provides another example.) It also requires that Childress et al. (2020) make a case for the value
researchers think about how best to interweave the of propulsive facilitation in moving researchers
different elements, especially since, as was suggested beyond triangulation. They state that “It is worth
in the context of the discussion about triangulation, acknowledging that when methodological trian-
the outcomes of mixed method research are not gulation is evoked, findings almost always seem
always predictable. to triangulate” (p. 370). While triangulation seeks
Like unplanned triangulation, this category of to confirm findings using different methods, pro-
multi-strategy research is more or less impossible pulsive facilitation is characterized by ongoing
to plan in advance. Essentially, it offers quantitative methodological adaptation and flexibility: re-
researchers an alternative to either revising their hy- search discoveries raise new questions which may
potheses or filing the inconsistent results away (and “present an occasion to apply a new or alternative
probably never looking at them again). method which is better equipped to making sense
of the unexpected datum or phenomena one is
Moving beyond Triangulation presented with along the way” (p. 380). Instead of
Building on Hammersley’s (1996) idea of facilita- directing one’s research energies to confirming
tion, University of Toronto sociologists Childress, what is already known through application of the
Calonga, & Schneiderhan (2020) propose using first method, Childress et al. (2020) propose re-
what they term propulsive facilitation for conducting directing our research energies toward thinking
multi-strategy research. They point out that while iteratively about the application of our research
this approach is not new, their goal is to reinforce methods.
the value of facilitation in fostering new insights.
The idea of “propulsion” captures both (a) the move- Reflections on multi-strategy
ment across different findings and methods as the research
research progresses, and (b) the difficulty of getting
the research started, but once underway keeping it There can be little doubt that multi-strategy research
in motion becomes easier. is becoming more common (Timans, Wouters, &
Propulsive facilitation entails beginning with Heilbron, 2019), although research that incorporates
the research method best suited for answering one’s only one approach remains the norm, especially in
research question. As the data begin to shed light disciplines such as sociology, psychology, and politi-
on the question, new questions often arise that are cal science. One factor that is relevant to the growth
best answered through a different form of data col- in the popularity of multi-strategy methods is a grow-
lection. This, in turn, might lead to other questions ing willingness to see research methods simply as
arising that are studied using additional methods. techniques of data collection, unencumbered by epis-
With each successive step in the process the re- temological and ontological baggage. For example,
searcher is propelled in new directions, producing research on audience reception of media and cul-
iterative questions that require different research tural texts had traditionally relied mainly on qual-
techniques to answer them. For example, Childress itative methods (in particular, focus groups), but
et al.’s (2020) research on Nelson Mandela’s presi- lingering unease regarding the reliability and gen-
dential memoirs began with archival research. In eralizability of their findings led some researchers
answering their initial question, a new one arose to recommend using quantitative research methods
which was best answered through an ethnogra- in tandem with qualitative methods (for example,
phy of the archive itself. The data they collected Schrøder, 1999).

bel65796_ch14_333-358 353 01/13/22 12:32 PM

354 PART IV Bringing It All Together

However, multi-strategy research is not necessar- challenges presented by diverse and fluid research
ily superior to one-method or one-strategy research. contexts. To navigate the messiness of multi-strat-
Four points must be borne in mind: egy research, Sanscartier recommends developing a
“craft attitude,” which he describes as follows:
• Multi-strategy research must be competently
a disposition (not a paradigm, method, or
designed and conducted: otherwise its findings
design type) towards the mixed methods re-
will be suspect, no matter how many methods
search process that (a) is comfortable with
are employed.
uncertainty, (b) favors non-linearity and re-
• Multi-strategy research must be appropriate
cursiveness in research design, and (c) treats
for the research questions asked. There is no
research as an exercise in storytelling, about
point in collecting additional kinds of data on
both the research object and our engagement
the assumption that “more is better.”
with that object. (p. 53)
• Multiple methods are likely to take consider-
ably more time and financial resources than Adjusting to the messiness of multi-strategy re-
research using just one approach, and spread- search requires researchers to be open to the mess,
ing resources too thinly can dilute the effec- to be prepared to improvise and remain flexible
tiveness of the research. with their methods, to be reflexive about their ap-
• Not all researchers have the skills and train- proach, and to be transparent when sharing accounts
ing required to carry out both quantitative and of the decisions made in conducting their research
qualitative research, and their “trained inca- (Sanscartier, 2020).
pacities” may prevent them from integrating Multi-strategy research is not a panacea. It may
the different forms of research. provide a better understanding of a phenomenon
than if just one method is used. It may enhance
Compared to studies involving a single research confidence in the findings, and it may even improve
method, conducting multi-strategy research is often the chances of access to research settings. Milkman
more complex and “messy” (Sanscartier, 2020). (1997, p. 192), for example, suggested in her research
Sanscartier (2020) explains that mixed methods on a General Motors factory that her promise of
researchers encounter two broad types of “messi- “hard,” quantitative data facilitated her entry to the
ness:” “empirical mess” and “design-related mess.” plant even though she had no experience with this
Empirical mess occurs when quantitative and quali- method. But the general point remains: multi-strat-
tative methods produce conflicting findings. Design- egy research, while offering great potential in many
related mess is the result of researchers needing to instances, is subject to constraints similar to those in
make ad hoc design adjustments to respond to the research relying on a single strategy.

Methods in Motion Applications to Canadian Society

Combining quantitative and qualitative research to understand poverty and mental
health among bisexuals
Although bisexuals form the largest sexual mi- relationships between bisexuality, poverty, and
nority group, there is relatively little research on mental health among a sample of 302 adult bi-
the experiences of bisexual people. Ross and col- sexuals living in Ontario.
leagues (2016) sought to rectify this by using a The research grew out of an awareness
multi-strategy approach to examine the empirical that previous studies had shown that low

bel65796_ch14_333-358 354 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 355

socioeconomic status was associated with higher In the quantitative part of the study, Ross
risk of negative mental health outcomes, and that and her associates found that a sizable propor-
people with a minority sexual orientation (gays, tion (25.7 per cent) of all study participants lived
lesbians, and bisexuals) tended to have lower in- below the LICO poverty line, and that poverty was
comes than heterosexuals, although some data associated with depression and PTSD. Bisexuals
indicated that non-heterosexual women earned living in poverty also experienced more discrimi-
more on average than heterosexual women. nation, and after demographic controls were in-
Previous research also found that bisexuals troduced they were more likely to be open about
were more likely to live in poverty than lesbians their sexual orientation.
or gays, and had more health problems than The qualitative data analysis gave rise to
people with other sexual orientations. Earlier four models of the pathways by which pov-
studies also indicated that bisexuals experienced erty and bisexuality may intersect with mental
discrimination in the form of homophobia, het- health. One male participant described being
erosexism, biphobia, and monosexism, with the gang raped at age 15, which led to a “lost
latter two forms of prejudice coming not only decade” of substance abuse and risky sexual
from heterosexual individuals but also from les- behaviours at a stage in his life that might oth-
bian and gay people. erwise have been devoted to advancing his ed-
Ross et al. gathered quantitative data with ucation and developing his job credentials. This
an online survey that measured whether partic- and other accounts of early life experiences
ipants had incomes below the Low Income Cut- were used by the researchers to produce a
Off (LICO), a widely used indicator of poverty. The model in which bisexuality and poverty impact
survey also included measures of mental health adolescent development, which in turn affects
issues such as depression, anxiety, post-trau- both mental health and subsequent poverty.
matic stress disorder (PTSD), and suicidality. Other participants discussed how biphobic or
Variables measuring “outness” (how open a homophobic discrimination resulted in job loss,
person with a minority sexual orientation is about promotion denials, or lack of hiring, which led
their sexual identity) and perceived discrimination the authors to develop a theoretical model in
were included on the survey as well. In addition, which bisexuality affects employment experi-
information was collected on the participants’ ences and earning potential, which then influ-
demographic characteristics such as their age, ences mental health in a reciprocal way. Other
sex assigned at birth, gender identity, and racial/ models explored how bisexuality and pov-
ethnic/cultural identity. erty may impede access to social support and
After the quantitative data were collected, mental health services, which can contribute to
Ross and her co-researchers acquired a quali- poor mental health.
tative perspective on the matter by conducting This study provides a good example of how
semi-structured interviews with a subset of the quantitative and qualitative approaches can
survey participants. The interviews were de- complement each other, and how a combination
signed to provide a social context for the quanti- of the two may be used to generate empirically
tative results, and to explore possible theoretical based theories that can be the subject of further
connections between poverty, bisexuality, and research.
mental health.

bel65796_ch14_333-358 355 01/13/22 12:32 PM

356 PART IV Bringing It All Together

Key Points
• It’s important not to exaggerate the differences research methods can be used to analyze the
between quantitative and qualitative research. rhetoric of quantitative researchers.
• Connections between epistemology and ontol- • Some qualitative researchers employ quantifica-
ogy on the one hand, and research methods on tion in their work.
the other, are not fixed or absolute. • Although the practice of multi-strategy research
• Qualitative research can exhibit features nor- has increased, not all writers support it.
mally associated with a natural science model. • The view that there are epistemological and
• Quantitative research can incorporate an inter- ontological impediments to the combination of
pretivist stance. quantitative and qualitative research is a barrier
• The artificial/natural contrast used to distinguish to multi-strategy research.
quantitative and qualitative research is often • There are several different ways of combining
exaggerated. quantitative and qualitative research; some can
• A quantitative research approach can be used be planned in advance, others cannot.
to analyze qualitative data, and qualitative

Questions for Review (R) and Creative Application (A)

The natural science model and quali- A3 You do a quantitative content analysis that ex-
tative research amines the portrayal of women in detergent
R1 Under what circumstances can some qualitative commercials. How might your study be con-
research use a natural science model? structivist in nature?
A1 A qualitative researcher finds that many of the
homeless people she encounters in her fieldwork Research methods and epistemologi-
have addiction issues. How might she use that cal and ontological considerations
finding to launch a quantitative study? R4 How closely tied are research methods to
epistemological and ontological positions?
Quantitative research and Explain.
interpretivism A4 You decide to do a secondary analysis of quan-
R2 Under what circumstances can some quan- titative data taken from the General Social
titative research exhibit characteristics of Survey, Victimization Study. You have no pre-
interpretivism? conceived theoretical position when you begin,
A2 A quantitative researcher finds that 25 per cent but decide to see if there is any association be-
of the people aged 18–25 in a national sample tween gender and fear of crime. How does your
have no intention of voting in the next federal approach deviate from a strictly positivist orien-
election, while the figure for people aged 65 and tation to research?
over is only 10 per cent. Explain this difference,
making reference to how people at different Problems with the quantitative/quali-
ages may perceive the political process differ- tative contrast
ently. If you were to write up your answer in a R5 Outline some of the ways in which the quantita-
research report, would it be appropriate to de- tive/qualitative contrast is not as hard and fast
scribe it as qualitative in nature? Explain. as is often supposed.
A5 Explain how grounded theory methods could be
Quantitative research and used to develop a theory of the relationship be-
constructionism tween body shaming and the use of social media,
R3 Under what circumstances can some quantita- then describe how this sort of research can be
tive research be constructivist? thought of as a form of theory testing.

bel65796_ch14_333-358 356 01/13/22 12:32 PM

 14 Bridging the Quantitative/Qualitative Divide 357

The mutual analysis of quantitative against multi-strategy research that combines

and qualitative research quantitative survey approaches with qualitative
R6 How might a researcher take a qualitative ap- interviewing?
proach to quantitative research?
A6 What are some implications of Gilbert and Opposing arguments about combining
Mulkay’s (1984) work (on how scientists write quantitative and qualitative research
up their findings) for the qualitative analysis of R11 What are the epistemological and technical
quantitative research? arguments regarding the prospect of com-
R7 What is ethnostatistics and why is it used? bining quantitative and qualitative research?
A7 How can ethnostatistics inform the debate about What are their implications for multi-strategy
whether quantitative and qualitative research research?
are based on incompatible paradigms? A11 Explain how multi-strategy research could be
R8 What is a meta-ethnography? carried out on the topic of white-collar crime.
A8 Assess the significance of Hodson’s (1996; 1999)
research on workplace ethnographies. How can Approaches to multi-strategy research
it be used to evaluate the embedded methods R12 What are the chief ways in which quantitative
argument? and qualitative research have been combined?
A12 Traditionally, qualitative research is depicted as
Quantification in qualitative research having a preparatory role in relation to quan-
R9 What is meant by quantification in qualitative titative research. To what extent do the differ-
research? ent forms of multi-strategy research reflect this
A9 Should there be more quantification in qualita- view? How can quantitative studies of conser-
tive studies? What are some drawbacks to in- vative political attitudes be a precursor to qual-
creased quantification in qualitative research? itative research?
Explain your answer to these questions as they
would pertain to a qualitative content analysis of Reflections on multi-strategy research
a local gay pride parade. R13 Why has multi-strategy research become more
prominent?
The argument against multi-strategy A13 Is multi-strategy research necessarily superior
research to single-strategy research? Describe a spe-
R10 What is multi-strategy research? cific, hypothetical instance in which it would be
A10 What are the strengths and weaknesses of the superior, and one in which it would not.
embedded methods and paradigm arguments

Interactive Classroom Activities

1. The instructor divides the class into small groups. political parties and any other organization
Each group: that has a pronounced power hierarchy;
a. selects a general topic that could be re- c. produces an outline of how the research
searched using quantitative methods—e.g., could be conducted, including sample selec-
variables that predict people’s participation tion, wording for key survey questions (if ap-
in political activities; plicable), description of statistical techniques
b. develops a theory or hypothesis pertaining that could be employed (e.g., correlation,
to their general topic—for example, young cross-tabulation, regression), and so on;
people are less likely to participate (vote, d. discusses how qualitative methods could be
run as a candidate, etc.) in national elections incorporated into the research at each step
than older people, and this occurs because in the process—for example, by holding focus
of a sense of alienation from conventional groups with young people to explore how

bel65796_ch14_333-358 357 01/13/22 12:32 PM

358 PART IV Bringing It All Together

they feel about politics and their participation c. discusses how quantitative methods could be
in it; acquiring and analyzing qualitative data incorporated into the research at each step in
to develop testable hypotheses on the topic; the process—for example, by coding qualita-
using the qualitative data to create survey tive data; addressing the issue of how prevalent
questions; or doing qualitative research to certain eating disorders are among particular
make sense of quantitative findings; and social groups or in the larger society; examining
e. considers the following question: “Does quanti- whether people with eating disorders share cer-
tative research differ from qualitative research tain psychological traits or demographic char-
in fundamental, irreconcilable ways? Explain.” acteristics; or generating hypotheses that could
be tested using quantitative methods; and
The class as a whole then reconvenes. Each group
d. considers the following question: “Does qualita-
presents a summary of its discussions to the class,
tive research differ from quantitative research
which is invited to ask questions and provide criti-
in fundamental, irreconcilable ways? Explain.”
cal commentary on each group’s presentation.
The class as a whole then reconvenes. Each group
2. The instructor divides the class into small groups. presents a summary of its discussions to the class,
Each group: which is invited to ask questions and provide criti-
a. selects a general topic that could be re- cal commentary on each group’s presentation.
searched using qualitative methods—for ex-
ample, how people with eating disorders view 3. A class debate is held to evaluate the following
themselves and their relationships with food; statement: “Quantitative research differs from
b. produces an outline of how the research could qualitative research in fundamental, irreconcil-
be conducted, including the recruitment of able ways.” The instructor divides the class into
informants; questions that could be explored three groups. The first group must present argu-
in qualitative interviews (if applicable); com- ments supporting the statement, the second must
mentary on whether ethnographic research oppose it, and the third can chose to either sup-
would be appropriate; a consideration of how port or oppose the statement. At the conclusion of
narrative analysis might be used; and so on; the debate a class vote is held to determine which
group presented the most convincing case.

Relevant Websites
In this YouTube video, one of this book’s authors (the qualitative and quantitative methods in telling a
late Professor Alan Bryman) discusses multi-strat- more complete story.
egy research in the social sciences.
[Link]
[Link]/watch?v=L8Usq_
TPfko&list=PLoem9zOwhTaD3dpk1aCp4e For examples of multi-strategy research, see the
OBzFgmOQ6DC Journal of Mixed Methods Research (to access the
journal you may have to go through your institu-
Using examples about holocaust survivors and
tion’s library server).
baseball, in this YouTube video Professor John W.
Creswell speaks about the value of combining [Link]
(Websites accessed on 4 September 2021)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch14_333-358 358 01/13/22 12:32 PM

 15 Conducting a Research
Project
Chapter Overview
The goal of this chapter is to guide you through a small-scale research project from beginning to end.
The guidance offered goes beyond the discussions presented in previous chapters in that it focuses on the
practical aspects of doing research. A wide variety of issues is explored, including:
• generating research questions;
• time management when doing research;
• dealing with the existing literature on your subject; and
• writing up social research.

Key Learning Outcomes

Here is a list of goals you should try to achieve as you read and think about the material in this chapter. After
completing this chapter, you should be able to:
1. Identify several ways of generating research questions.
2. Produce a plan for managing your time effectively while completing a research project.
3. Search the existing literature on your topic, and integrate relevant material into your project in various
ways and at different stages of your research.
4. Develop a strategy for writing up your project in an informative and engaging way.

When one of the authors of this book was a gradu- somewhat acerbically, “Just do it!” (This was years
ate student and had his first meeting with his disser- before that phrase became a marketing slogan.)
tation advisor, the advisor listened to his research Those three words contain a lot of wisdom. It’s
proposal, offered a few suggestions, and then said, easy to overthink things and waste time worrying

▲ Hemis/Alamy Stock Photo

bel65796_ch15_359-373 359 01/13/22 12:32 PM

360 PART IV Bringing It All Together

that your plans aren’t worthwhile or that you’re not The same goes for the writing. Try to write
up to the task. Just assume that with proper guid- something related to the project—anything—
ance and a lot of hard work, you’ll be just fine. But every day. And for your first draft, allow yourself
you must follow through with concrete actions, to be sloppy. Don’t worry about finding the per-
every day. Dreaming is not enough. fect word or crafting the perfect sentence or en-
Don’t wait for the perfect time to start. There is suring that the logic flows flawlessly. Those issues
no such thing. Imagine you’re at work and your boss can be dealt with later, at the editing stage. When
has given you an unbreakable deadline. You’ll find Michael Jackson was asked about his method for
that once you get absorbed in the task, your inspi- writing songs, he replied, “Don’t get in the way of
ration and motivation will pick up even if they were the music.” That advice applies to completing a
minimal when you began. It also helps if you think of research project too: don’t get in the way of the
your research project as a series of small jobs rather ideas. Just let them flow. Write first and evaluate
than one monumental undertaking. later. Just do it.

Introduction of an abstract, the structure of the presentation,

­plagiarism, and deadlines.
This chapter provides advice for those readers car-
rying out their own small-scale research projects. Identifying research
Previous chapters have provided helpful information
about the methodological and conceptual choices questions
available and how to implement them. Here we offer Most students want to conduct research in areas
some hands-on advice on how to generate research of personal interest. This is not a bad thing; as
ideas and take them through the various stages of de- we noted in Chapter 1, many professional social
velopment to produce a successful project of your own, researchers want to do the same thing (see also
such as an honours thesis or a course assignment. The Lofland & Lofland, 1995, pp. 11–14). However, even
advice is tailored for students engaged in empirical open-ended qualitative research must begin with
research, that is, studies in which they either collect the formulation of research questions. Although
new data or analyze existing data. Although the dis- Chapter 10 referred to some studies that were not
cussion is mainly geared toward undergraduate proj- driven by specific research questions, open-ended
ects, students in graduate programs will also find the research is risky. It can lead to the collection of
observations helpful. data that will never be used, and to serious prob-
lems in data organization and analysis. The lack of
Know what is expected by focus can also cause the project to take too long or
your institution even to be abandoned. So, unless your supervisor
or instructor advises you to the contrary, you need
Quite often a course instructor will have specific to formulate some research questions that are not
guidelines and requirements for class projects. completely open-ended. What is it about your area
Look them over carefully and follow them. People of interest that you want to know? Figure 15.1 il-
doing undergraduate theses and students in grad- lustrates some steps that can be taken to select re-
uate school should be aware that their institution search questions. If you need ideas for research, talk
or department may have specific requirements things over with your advisor or course instructor,
concerning things like binding and page margins, or take a look at the list of sources for research ques-
as well as the format for referencing, the inclusion tions provided in Box 15.1.

bel65796_ch15_359-373 360 01/13/22 12:32 PM

 15 Conducting a Research Project 361

Research area
Example:
People’s willingness or unwillingness to abide by pandemic health protocols
(mask wearing, social distancing, avoiding large gatherings, etc.)
The meanings and interpretations people attach to abiding by or rejecting pandemic health protocols

Select aspect of research area

Examples:
Social causes of the variations in people’s willingness to abide by pandemic protocols
How meanings and interpretations pertaining to pandemic protocols develop and become
integrated into more comprehensive views of the self and society

Possible research questions

Examples:
Do political orientation, age, gender identification, socio-economic status, education, or ethnic
background affect people’s willingness to abide by pandemic health protocols? If so, which of these factors has the
greatest impact, and why? What are the theoretical and public policy implications of these issues?
How do people feel about those who do not follow pandemic health protocols? Do people’s views
on this matter emerge and develop through discourse with others who are similarly socially situated?
If so, how does that process unfold? How do people’s perceptions of the protocols relate to their larger views
of themselves and society? What are the theoretical and public policy implications of these issues?

Modification of selected research questions

Research questions may be revised and refined as the research proceeds.

FIGURE 15.1 Steps in selecting research questions

Interacting with work is formally examined. Supervisors have to go

through the same process themselves whenever they
a supervisor apply for a research grant or submit an article or book
Most institutions will assign a student writing a thesis for peer review.
or dissertation to a supervisor or advisor. However, the Students who get stuck or fall behind in their
kind and amount of assistance you can expect varies work will sometimes avoid their supervisors. But
greatly. Of course, students also vary in what they ask the longer they wait, the harder it is to ask for help.
of their advisors. Our advice here is simple: use su- Then, because the work has been put off, it has to be
pervisors or advisors to the fullest extent allowable rushed at the end, and in some cases is never com-
and give due consideration to their suggestions. Such pleted. Try to avoid this situation by confronting any
people are usually well versed in the research process difficulties head-on and getting advice on how to
and can provide help and feedback at all stages of the deal with them as soon as possible.
project, subject to your institution’s strictures in this Undergraduate students doing course projects
regard. If they criticize your research questions, inter- should follow the same basic advice. Consult your
view schedule, early drafts, or whatever, try to respond course instructor if you run into difficulties, and the
positively. Criticism isn’t a personal attack. Such com- sooner the better. Even if there are no apparent prob-
ments are usually accompanied by the r­easons for lems, it’s always worthwhile to discuss your research
them and some suggestions for revision. Be thankful with your instructor, who may pick up on issues that
for the opportunity to address problems before your you have missed.

bel65796_ch15_359-373 361 01/13/22 12:32 PM

362 PART IV Bringing It All Together

Box 15.1 Marx’s (not Karl) sources of research questions

• “gaps between official versions of reality and
the facts on the ground” (1997, p. 113). Examine
the standard explanations offered by govern-
ments, the media, or social researchers for
social phenomena such as poverty, substance
abuse, international conflict, or anything else
of interest. Are those explanations consistent
5./15 WEST/iStockphoto

with what you know or have experienced? If

not, develop some research questions per-
taining to the phenomenon in question, in-
cluding questions about the validity of the
What types of research questions could emerge conventional wisdom on the topic.
from a close investigation of new media technolo- • the counterintuitive; for example, cases in which
gies and social networking websites? social scientific truths seem to fly in the face of
common sense, as they did when a doomsday
cult began to proselytize rather than slink away
Marx (1997) suggested the following possible in embarrassment when “doomsday” passed
sources of research questions: and life carried on as usual (Festinger, Riecken,
• intellectual puzzles and contradictions & Schachter, 1956)
• the existing literature • “empirical examples that trigger amazement”
• replication (1997, p. 114); for example, atypical events,
• a feeling that a certain theoretical perspective cases that contradict widely accepted theo-
or notable piece of work is misguided (in this ries, and so on
case you would explore the reasons for the • new methods and theories—in what settings
shortcomings) and with what people might they be applied?
• a social problem in need of a social scientific • “new social and technical developments and
(for example, sociological) explanation social trends” (1997, p. 114)

Managing time and • Work out a timetable, preferably with your

­supervisor or instructor, detailing the differ-
­resources: Start thinking early ent stages of the research (including the litera-
about the research area ture review and the writing up). The timetable
should specify the different stages and the spe-
Students are usually asked to start thinking cific dates for starting and finishing them (see
about a potential research topic well before they Figure 15.2 for an example of a quantitative
are expected to start work on the project. That research outline). Some stages are likely to be
thinking is worth doing, because all research ongoing—for example, searching the literature
is constrained by time. Two pieces of advice are for references (see below)—but that should
­relevant here: not delay the development of a timetable.

bel65796_ch15_359-373 362 01/13/22 12:32 PM

 15 Conducting a Research Project 363

Issue(s) to be researched

Review literature/theories relating to topic/area

Formulate research question(s)

Consider whether a social survey is appropriate (if not, consider an alternative research design)

Consider what kind of population is appropriate

Consider what kind of sample design is appropriate

Determine whether there is a sampling frame available

Decide sample size

Decide on mode of administration (face-to-face, telephone, postal, email, web)

Develop questions (and devise answer alternatives for closed questions)

Review questions and assess face validity

Complete ethics protocols

Pilot study to test the questions

Revise questions

Finalize questionnaire/schedule

Sample from population

Administer questionnaire/schedule to sample

Follow up non-respondents at least once

Transform completed questionnaires/schedules into computer readable data (coding)

Enter data into statistical analysis program such as SPSS

Analyze data

Interpret findings

Consider implications of findings for research questions, existing theories, future research,
and the discipline as a whole.
FIGURE 15.2 Steps in conducting a social survey study

bel65796_ch15_359-373 363 01/13/22 12:32 PM

364 PART IV Bringing It All Together

• Find out what resources, if any, are available for and sexism or prejudice” is better because it expands
carrying out your research. For example, can the first option above to cover university prejudice in
your institution help with the cost of printing, addition to university sexism. Finally, “not” restricts
postage, stationery, and so on? Can it lend you the search. “Sexism not universities” will find arti-
hardware to record and transcribe interviews? cles about sexism in places other than universities.
Does it have the necessary software, such as IBM You can experiment with the use of keywords and
SPSS Statistics (SPSS) or NVivo? This kind of Boolean options; if you don’t find what you need, ask
information will help you decide whether your a librarian for help.
research design and methods are financially Following these procedures will produce a list
feasible and practical. The fictional “gym study” of journal articles in which the keyword(s) appears,
in Chapter 8 is an example of a project feasible with full citation details and an abstract, and some-
within the kind of time frame usually allocated times a link to a full-text download. Other data-
to undergraduate projects. However, it would bases are more specialized: examples include the
require resources for printing covering letters Canadian Periodicals Index Quarterly (CPI.Q) (for-
and questionnaires for participants who are merly the Canadian Periodical Index), the LGBT Life
unable to respond electronically; postage both with Full Text, Family & Society Studies Worldwide,
for sending the questionnaires out to them and and Contemporary Women’s Issues. Mark the records
for follow-up letters; and return postage for the you want and email them to yourself for a permanent
questionnaires. In addition, quantitative data record. The bibliographies in the articles you find
analysis would require software such as SPSS. can provide additional sources to examine.
Two particularly interesting tools are the Social
Sciences Citation Index (accessible through the Web of
Searching the existing Science) and Google Scholar, which look forward from
literature the original article to later ones that cite it, revealing
what others have thought of it. Perhaps some who
Online bibliographical databases (accessible through cited the article were critical; perhaps others accepted
most university libraries) are an invaluable source or expanded on it. In this way a history of the research
of references. The best one for sociology students is in the area of interest can be built up. To many pro-
probably Sociological Abstracts; databases for other fessors, these indexes provide measures of how often
disciplines are available as well. They normally allow they have been cited. To those granting tenure and
searches by keyword (topic), author, title, journal, raises, it can be a crude indicator of merit, following
and other descriptors. The focus is easily narrowed by the logic of “more citations, better researcher.”
specifying language, year, and type of presentation. The library catalogue is an obvious tool for finding
For example, “English, post-2015, scholarly journal books, even those held at other universities. Again,
articles only” can be searched and listed from latest ask a librarian if you run into difficulty. However,
to earliest. The “scholarly journal articles only” re- since books are often dated, checked out, or too spe-
striction removes papers presented at meetings but cialized for student papers, the above sources may be
not published in journals, and also excludes disserta- preferable. The rule of thumb is similar to the idea
tions, which are too hard for most students to access of theoretical saturation in qualitative research: stop
in any event. Searches can be conducted using key- searching when the same items keep appearing.
words in different combinations with the Boolean Why review the existing literature? The most obvi-
operators “and,” “or,” and “not.” Thus, a search for ous reason is to find out what is already known about
material on “sexism and universities” will produce your area of interest, in order to avoid “reinventing
articles that deal with both topics in the same piece. the wheel.” In addition, the process of reviewing the
Choosing “sexism or universities” will produce those literature will help you revise and refine your re-
articles plus articles on each one alone. That is prob- search questions. Beyond this, the existing literature
ably not a good choice—it’s too broad. “Universities on your topic will help you develop an argument

bel65796_ch15_359-373 364 01/13/22 12:32 PM

 15 Conducting a Research Project 365

regarding the significance of your research and where timetable (see above). This is not to say that your lit-
it will lead. The metaphor of a story is sometimes used erature search will cease: only that you need to force
in this context (see below). A competent review of yourself to move on. Ask your supervisor or instruc-
the literature will help to establish your credibility as tor for advice on whether you need to extend your
someone knowledgeable in your chosen area. search of the literature.
As you explore the existing literature, look for an-
swers to the following questions: Preparing for research
• What is already known about this area? In choosing a methodology you need to consider its
• What concepts and theories are relevant? ethical implications and what you will do to prevent
• What research methods and strategies have any potential ethical violations. Keep in mind the
been employed in studying this area? various ethical issues discussed in Chapter 3 and
• Are there any significant controversies? flagged throughout this book.
• Are there any inconsistencies in the findings Don’t begin collecting data until you have clearly
relating to this area? identified your research questions, and design your
• Are there any unanswered research questions? data collection instruments with those questions
in mind. If at all possible, conduct a small pilot study
While reading, take good notes, and be sure to to see how well your research instruments work. You
record full citations for all the materials you consult. will also need to think about access to research sites.
It’s infuriating to find that you’ve forgotten some If the research requires entry to a closed setting such
detail, such as the volume number of an article, when as an organization, you need to get permission at the
you come to assemble your bibliography. Also keep earliest opportunity. This usually takes so long that
in mind that your written review of the literature many advisors do not recommend using closed set-
should be critical as well as descriptive. How does a tings for student research. You also need to consider
particular article relate to others you have read? Does how you will go about gaining access to people. This
it contradict them? Are there any apparent strengths issue leads you into sampling considerations, such as
or deficiencies—for example, in the methods used or the following:
the conclusions drawn? What theoretical ideas influ-
enced the author(s)? • Whom do you need to study in order to answer
In some areas of research you will find huge num- the research questions?
bers of references. Try to identify the major ones and • Is an adequate sampling frame available?
work outward from there. Move on to the next stage • What kind of sampling strategy is feasible—
of the research at the point you identified in your probability, quota, theoretical, convenience?

Practical Tip | Reasons for writing a literature review

The following are reasons for writing a literature 4. to get help in developing an analytic framework
review: 5. to find additional variables to include in the
research
1. to find out what is already known in a research 6. to search for further research questions
area, and what remains to be researched 7. to view examples of how findings can be
2. to learn from other researchers’ mistakes interpreted
and omissions 8. to compare and contrast your findings with
3. to learn about different theoretical and meth- those of other researchers
odological approaches to the research area 9. it’s expected

bel65796_ch15_359-373 365 01/13/22 12:32 PM

366 PART IV Bringing It All Together

Writing up research Start early

There are good reasons for beginning the writ-
Once the data gathering and analysis have been com- ing process early on. It forces you to think about
pleted, the findings must be conveyed to an audience. issues such as how best to present and justify the
This is an important matter because regardless of research questions, and how to structure the dis-
how well your research is conducted, readers have to cussion of the literature cited. A second reason is
be convinced that your findings are valid and worth entirely practical: many people find it difficult to get
knowing about, and that requires good writing. started and tend (often unwittingly) to use procras-
As with other skills, the best way to become profi- tination strategies. But putting off the inevitable
cient at writing up social research is to do it as often almost always results in rushed, last-minute writ-
as you can. And be patient with yourself—even the ing. Writing under that kind of pressure is not ideal,
best writers edit and revise their work many times to say the least. How you present your findings and
over before they consider it finished. It’s also helpful conclusions is a crucial aspect of the research pro-
to get a knowledgeable person to read what you’ve cess, and if you fail to produce a convincing account
written and offer advice. of your findings because of time pressures, you will
There are structural and technical requirements not do your work justice.
that vary depending on the type of research you’ve
done, but it’s equally important to pay attention
to the quality of your writing as writing. Writing, Get feedback
even the formal variety, shouldn’t be thought of
Try to get as much feedback on your writing as pos-
as drudgery: consider it an art to be cultivated.
sible, and respond positively to whatever comments
Among other things, try to bring your enthusi-
you receive. Your supervisor or instructor is likely to
asm for your topic to the page. If your exuberance
be the main source of feedback; show as many drafts
has waned by the time you start writing, think
of your work as regulations allow, leaving plenty of
about how your project relates to some larger so-
time for a response. You will not be the only student
cietal or theoretical issue, and keep that in mind
seeking advice, and if comments are rushed, they
as you write.
may not be particularly helpful. You can also ask
Here are few more suggestions:
others in your program or class to read your drafts
and comment on them, and they may ask you to do
the same. Their comments may be very useful, but
those from your supervisor or instructor are likely to
be the most valuable.

Avoid sexist, racist, and other

­prejudicial language
inbj/123RF, vm/iStockphoto

Remember that writing should be free of sexist,

racist, and other prejudicial language.

Even formal writing should showcase your enthusi-

Structure your writing
asm for your topic. What strategies could you use to A 10,000- to 15,000-word research project entailing
maintain your interest in your research project? data collection is typically structured as follows.

bel65796_ch15_359-373 366 01/13/22 12:32 PM

 15 Conducting a Research Project 367

Box 15.2 Safety in research

In 2002, a 19-year-old sociology student was language), calmly withdraw from the situation.
supposed to interview a homeless person in Further guidelines on these issues can be found
Manchester, UK. Because of safety concerns, in Craig (2004).
her advisor recommended that she take a Sometimes researchers may be unaware
friend with her and conduct the interview in a of possible hazards. Lankshear (2000), for in-
public place. When she did not return, she was stance, found out only after her study had
reported missing (Barkham & Jenkins, 2002). begun that her research in a hospital labora-
Four days later she showed up in Dublin, where tory could expose her to dangerous pathogens.
she claimed to have fled after the man robbed Be sure to do a thorough risk assessment of the
her; however, it was later suggested that her research environment before you undertake
disappearance might have been sparked by a your project.
family argument. Finally, once you’ve assessed the risk and
Whether or not her story was true, there is are convinced that you will not be placing
an important lesson in it: social research can yourself in harm’s way, it may be necessary to
place you in potentially dangerous circum- persuade your instructor, advisor, or research
stances. Avoid situations where personal harm ethics board (REB) that your safety will not be
is a real possibility. Just as you should ensure compromised. Small, Maher, & Kerr (2014),
that no harm comes to research participants, who did ethnographic research with people in
individuals conducting research should not Vancouver who inject themselves with illegal
place themselves in unsafe situations. The drugs, were told by their REB that they had to
advice given by the student’s advisor—to take consult with local police about safety issues and
someone with you and conduct the interview in provide police with the names of the ethnogra-
a public place—is sensible. Keep your cellphone phers who would be observing the drug users.
nearby and switched on. Establish a routine The researchers had to go to great lengths
whereby you keep in regular contact with to convince the REB that proper precautions
friends or family members. Even in seemingly would be taken and that they would not be in
safe conditions, a researcher can be faced danger, and that involving the police in their
with a sudden outburst of abuse or threaten- safety would make it very difficult to recruit
ing behaviour; you can’t always predict how participants and ensure the confidentiality of
people will react to a particular trigger, such the results. Eventually the board relented and
as an interview question. If there are signs that the study went ahead as planned, but only after
trouble is imminent (for example, through body a prolonged review process.

Title page Acknowledgments

Your institution or class may have specific require- You may want to acknowledge the help of people such
ments for the title page. Typically, it should include as gatekeepers who gave you access to an organiza-
your name, your instructor’s name, the course, and tion and those who read your drafts and provided you
the date, as well as the title of the paper. with feedback or advice, including your supervisor.

bel65796_ch15_359-373 367 01/13/22 12:32 PM

368 PART IV Bringing It All Together

Table of contents (if applicable) with the problem of careers.” This kind of sentence
Your institution or class may have recommendations “is evasive, pointing to something without saying
or prescriptions about the form the Table of Contents anything, or anything much, about it. What about
should take. careers?” (Becker, 1986, p. 51). To those concerned
that too much detail might give away the plot,
An abstract he countered that it is much better to give readers
An abstract is a brief (less than one page) summary a quick and clear indication of what is going to be
of your work. Not all institutions or instructors ask presented.
for this component, so find out if it is required. Most
Literature review
journal articles include abstracts; draw on them for
ideas and models. As we have seen, the Literature Review section pro-
vides an overview of the main ideas and research
Introduction done in your area of interest. However, your review
The purpose of the Introduction is to explain what should do more than simply summarize the relevant
you are writing about and why it is important. It literature.
should also:
• Remember that, whenever appropriate, your
• describe in general terms the theoretical approach should be critical.
perspective you have chosen to take (where • You should use your review of the literature
applicable); to show why your research questions are im-
• explain your choice; and portant. For example, suppose that the basis
• outline your research questions (remember for your research questions is the idea that
that in qualitative studies, research questions although a lot of research has been done on
are often more open-ended than is the case X (a general topic or area, such as food inse-
with quantitative work). curity), little research has been done on Y (an
aspect of X, like food insecurity in rural areas).
The opening sentence is often the most difficult of In that case the literature review is where you
all. Becker (1986) strongly advised against “vacuous” should justify this assertion. Alternatively, it
opening sentences on the model of “This study deals may be that there are two competing positions

Practical Tip | Non-sexist writing

One of the biggest challenges (though by no This sentence, although grammatically cor-
means the only one) when trying to write in rect, can be rephrased as:
an inclusive way is how to avoid those awk-
I wanted to give all respondents the op-
ward “she/he” and “his/her” formulations. One
portunity to complete their questionnaires
­solution is to write in the plural where possible.
at times and locations that were conve-
For example:
nient for them.

I wanted to give each respondent the “They” and “them” are also acceptable as sin-
­opportunity to complete the questionnaire gular pronouns, and some people prefer to be
at a time and location that was convenient addressed that way. In such cases their wishes
for him or her. should be respected.

bel65796_ch15_359-373 368 01/13/22 12:32 PM

 15 Conducting a Research Project 369

with regard to Y and you are going to investi- and data analysis. When discussing each of these
gate to see which one provides a better under- matters, you should describe and defend the choices
standing. In the literature review, you should you made: for example, why you decided to use an
outline the differences between the competing online questionnaire rather than a structured inter-
positions. The literature review, then, locates view, or why you chose a particular population for
your own research within a tradition of related sampling.
research.
• Bear in mind that you will return to points Results
made in the review when you present your In the Results section you present the gist of
findings and write up the conclusions. your findings. If you will be including a separate
• Don’t try to get everything you’ve read into Discussion section, the results should generally be
the literature review. Forcing it all in (be- presented with little or no commentary on how they
cause of the effort you’ve put into finding relate to the claims made in the literature. If there is
and reading the material) will not help. The no Discussion section to follow, you need to include
literature review is there to assist you in some reflections on the significance of your findings
developing an argument, and if you stuff it for your research questions and for the literature.
with material of passing relevance you will Bear the following points in mind:
undermine your ability to get your argument
across. • If you are presenting tables, graphs, and so on,
• Recall that reading the relevant literature you need to comment on each one; otherwise,
should continue more or less throughout the readers will be left wondering why they were
research process. This means that a litera- included. Don’t just summarize what a table
ture review written before the data collection shows: direct the reader to the parts of it that
begins is provisional. If you find that you want are especially striking from the point of view
to revise your initial review, go ahead. of your research questions. Ask yourself what
• Additional reflections on producing a liter- story you want the table to convey; then relay
ature review are presented in Box 15.3. They that story to your readers.
were derived from a review of qualitative stud- • Whichever approach you choose, remember
ies of organizations, but they also apply to not to include all your results. You should
quantitative research. present and discuss only those findings that
relate directly to your research questions.
Research methods Omitting some of your findings can be pain-
The term “research methods” is a kind of catch-all ful, but it’s necessary to ensure that the thread
for several things that need to be outlined: research of your argument is not lost. This is especially
design; sampling approach; how access was achieved true in the case of qualitative research. As one
(if relevant); specific procedures used (for example, experienced qualitative researcher put it: “The
in the case of an online questionnaire, whether and major problem we face in qualitative inquiry
how non-respondents were contacted for follow-up); is not to get data, but to get rid of it!” (Wolcott,
and, where relevant, the nature of the question- 1990, p. 18). If you don’t leave out some of
naire, interview schedule, participant observation, your results, you will not only obscure your
observation schedule, or coding frame. Although argument but run the risk that your account
copies of these usually appear in an appendix, you of your findings will appear too descriptive
should comment on matters such as your style of and lack an analytical edge. This is why it’s
questioning or observation and why you asked the essential to use your research questions as a
questions you did. Other issues to be discussed in focus and to orient your presentation of the
this section may include procedures for note taking findings to them.

bel65796_ch15_359-373 369 01/13/22 12:32 PM

370 PART IV Bringing It All Together

Box 15.3 Presenting qualitative research literature

In their examination of journal articles ­ reporting Each of these strategies is designed to leave
qualitative research on organizations, G ­olden- room for the writer to make their own contribu-
Biddle & Locke (1993) argued that good articles in tion to the field.
this area develop a story; that is, a clear and com- • Problematizing the situation. The literature
pelling framework around which the writing is struc- is then criticized by identifying a problem, for
tured (see also Golden-Biddle, 1997). This idea is example:
very much in tune with Wolcott’s (1990, p. 18) advice » Incomplete. There is a gap in the existing
on writing up qualitative research: “determine the literature.
basic story you are going to tell.” Golden-Biddle & » Inadequate. The existing literature has ig-
Locke suggest that establishing how your research nored useful ways of looking at the phe-
relates to the existing literature on the topic is an nomenon; alternative perspectives or
­important component of storytelling. They identi- frameworks are then introduced.
fied two things that are done to effectively review
the l­iterature and convey the relevance and signifi- According to Golden-Biddle & Locke, the authors
cance of your research: they studied used their accounts of the literature
in various ways:
• Constructing inter-textual coherence. The author
shows how existing contributions to the literature » They demonstrated their knowledge and
relate both to each other and to the research competence by referring to prominent
that will be reported. Golden-Biddle & Locke dis- writings in the field.
tinguished three approaches to the coherence » They developed their review of the literature
question: in such a way as to highlight the contribution
» Synthesized coherence. The author pieces they would make in the article.
together theory and research previously » They explained how the gaps or problems
regarded as unconnected. they identified in the literature corresponded
» Progressive coherence. The author traces to their research questions.
the building of consensus in a particular
The idea of writing up research as storytelling
area of knowledge.
serves as a useful reminder that reviewing the
» Non-coherence. The author emphasizes
literature, which is part of the story, should link
disagreements in the contributions to a
seamlessly with the rest of the article.
certain research program.

Discussion Conclusion
If you are including a Discussion section, it should The main guidelines for the Conclusion are as
address the implications of the findings for your re- follows:
search questions. If you have specified hypotheses,
the discussion should revolve around whether or • A conclusion is not the same as a summary.
not they were supported. If they were not, you might However, it is frequently useful to briefly reca-
speculate about the reasons. Was the sample too pitulate your main arguments at the beginning
small? Did you forget a key variable? In this section of the Conclusion section. Once that is done,
you can also bring out the main theoretical contribu- hammer home to readers the significance of
tions of your research and explore their implications. your research.

bel65796_ch15_359-373 370 01/13/22 12:32 PM

 15 Conducting a Research Project 371

Checklist
Checklist of issues to consider for writing up research

□ Is there a good correspondence between the □ Does the discussion of the findings relate to
title of the project and its contents? the research questions?
□ Have you clearly specified your research □ Does that discussion show how the findings
questions? shed light on the literature presented?
□ Have you clearly linked the literature cited to □ Do the conclusions clearly establish what
your research questions? your research contributes to the literature?
□ Is your discussion of the literature critical and □ Have you explained the limitations of your study?
not just a summary of what you have read? □ Do your conclusions consist solely of a sum-
□ Have you clearly outlined your research mary of findings? If so, rewrite them, ex-
design and research methods? Make sure plaining their significance.
you have explained: □ Do the conclusions provide clear answers to
• why you chose a particular research your research questions?
design or method; □ Have you broken up the text in each chapter
• why you implemented your research or section with appropriate subheadings?
design in the way you did (for example, □ Have you provided signposts so that read-
how the interview questions relate to the ers know what to expect next, and why it is
research questions, or why you observed there?
people in particular situations); □ Does the writing avoid sexist, racist, and
• how you selected your research participants; other prejudicial language?
• whether you experienced any problems □ Have you checked to ensure that the text
with cooperation (for example, response does not make excessive use of jargon?
rates);
□ Have you included all the necessary appen-
• if your research required access to an or- dices (for example, interview schedule, let-
ganization, how and on what basis the ters requesting access, communications with
agreement was achieved; research participants)?
• any difficulties encountered in implement- □ Does the list of references include all the
ing your research; items referred to in your text?
• the steps taken to ensure that the research
□ Does the format of the references follow pre-
was ethical; and cisely the style required by your institution or
• how the data were analyzed. instructor?
□ Have you presented your data in such a □ Have you met all requirements regarding
way that they clearly relate to your research matters such as word count (so that the
questions? paper is neither too long nor too short)
□ Are your interpretations of the data fully sup- and inclusion of an abstract and table of
ported with tables, figures, or segments from contents?
transcripts? □ Have you ensured that you do not quote ex-
□ Is every table and/or figure properly labelled cessively when presenting the literature?
with a title and number? □ Have you fully acknowledged the work of
□ Is every table and/or figure commented on in others, so that you cannot be accused of
the text? plagiarism?

bel65796_ch15_359-373 371 01/13/22 12:32 PM

372 PART IV Bringing It All Together

□ Have you acknowledged, preferably in an □ Finally, have you checked to make sure
Acknowledgments section, the help of others that your tenses, margins, pagination, and
where appropriate (for example, your su- capitalization are consistent? Many forms
pervisor, people who helped with interviews, may be acceptable, but all must be used
people who read drafts)? consistently.

• This is the place to draw attention to any lim- and from gatekeepers, and so on—can be included
itations of your work that cannot be rectified in appendices.
without unduly expanding the scope of the
research. References
• You may also suggest avenues for further research. Be sure to include all the sources cited in the text.
Follow the format suggested by your department
Two things to avoid are (a) engaging in speculations or instructor. The format is usually an author-date
that either take you too far from your data or cannot system such as the one used in this book.
be substantiated by the data, and (b) discussing issues
or ideas that have not already been introduced. Finally
Remember to fulfill any obligations you have
Appendices ­incurred, such as supplying a copy of your report
Researchers often have some material that would to those who have been promised one. Maintain the
help readers assess their work, but is too detailed confidentiality of the information given and the ano-
or technical to include in the main text. Such ma- nymity of informants and other research participants
terial—questionnaires, coding frames, observation by securing—and later, as appropriate, ­destroying—
schedules, letters sent to those sampled, letters to the primary data.

Interactive Classroom Activities

1. The instructor reviews the various aspects of a 2. The instructor, or better yet a reference librar-
good research question. The class is then divided ian, provides a detailed description of the pro-
into small groups. Each group is to select a topic of cedures for using the databases available via
interest—for example, sexual assault on Canadian the institution’s library system. The steps taken
university campuses. The group then carefully can be shown on the classroom screen.
constructs a research question or set of research 3. In situations where students have ready access
questions based on the topic. Once completed, to the Internet, each student uses library data-
each group takes turns putting its questions on bases to find five articles that address the re-
the classroom screen or board for class discussion search question(s) his or her group came up with
and critical comment. The class is asked a series in exercise 1. Once completed, the groups then
of evaluative questions such as: Is the question(s) reconvene so students can compare the arti-
clear? Could anything be done to improve the cles they found with those selected by the other
question(s)? Should additional questions be asked? members of their group.

bel65796_ch15_359-373 372 01/13/22 12:32 PM

 15 Conducting a Research Project 373

Relevant Websites
Databases such as Sociological Abstracts, In this YouTube video, Jared Wright walks you
PsycINFO, Worldwide Political Science Abstracts, through the steps that need to be taken to write a
Social Work Abstracts, the Canadian Periodicals sociological research paper.
Index Quarterly (CPI.Q), LGBT Life with Full Text, [Link]/watch?v=n-5xIA71aRQ
Family & Society Studies Worldwide, Contemporary
Women’s Issues, and the Social Sciences Citation In this YouTube video, David Taylor discusses how
Index (through the Web of Knowledge) are all ac- to choose and narrow down a research topic.
cessible online. To use them, log on to your institu- [Link]/watch?v=jSHXb83Xtsk
tion’s library server and call them up from there.
This Social Sciences and Humanities Research
Google Scholar can be a ­ ccessed directly.
Council of Canada site discusses best practices to
This YouTube video from the University of Maryland promote equity, diversity, and inclusion in research.
discusses the fundamentals of doing a literature review. [Link]
nfrf-fnfr/[Link]
[Link]/watch?v=2IUZWZX4OGI&
feature=related (Websites accessed 4 September 2021)

More resources are available on Oxford Learning Link.

Visit [Link] for:

• Student self-quiz • Audio clips • Activities

• Flash cards • Videos

bel65796_ch15_359-373 373 01/13/22 12:32 PM

APPENDIX
Using IBM SPSS Statistics and
NVivo Software
Chapters 8 and 13 described a number of methods variables are in bold italics (e.g., gender or reasons).
for analyzing quantitative and qualitative data. This Labels (longer versions of names; see pp. 377–78)
appendix illustrates how to perform those anal- given to variables or values are in bold but are not
yses using IBM SPSS Statistics (SPSS) and NVivo italicized (e.g., reasons for visiting or male). A
software.* right-pointing arrow (→) means “left-click once with
the mouse” to make selections. Box A.1 presents a
list of basic operations in SPSS.
Quantitative data analysis If you have SPSS software loaded onto the hard
using SPSS disk of the computer you are using, click on the SPSS
icon on the Desktop screen or find the program
The gym survey described in Chapter 8 will be used using the Start button in the bottom left-hand corner
to illustrate how SPSS software is used. Learning of the screen. If you don’t have SPSS software on your
new software requires some perseverance, but is well computer, it may be available from a special server
worth the effort. It would take far longer to perform provided by your university.
the calculations by hand (even on the small gym After SPSS loads, an opening dialogue box with a
survey sample of 90) than to learn the software. With number of options may appear. If so, for the purposes
more advanced techniques and larger samples, the of this exercise disable it: → Close. This will give you
time saved is even more substantial. Imagine calcu- the SPSS Data Editor, made up of two components:
lating the mean age of 2500 people by hand! Data View and Variable View. Move between these
two by selecting the appropriate button at the bottom
Getting started in SPSS left of the screen. The Data View provides a spread-
sheet grid into which data are entered (see Plate A.1).
Introduction The columns represent variables—in our example,
SPSS is the software most widely used for this kind information about the characteristics of each person
of analysis, probably because it is relatively straight- in the gym study sample, such as gender and age.
forward. It has been in existence since the mid-1960s Until data are entered, each column simply has var
and over the years has undergone many revisions, as its heading. The rows represent cases, which can be
particularly since the arrival of personal computers. people (as in this example) or other units of analysis.
The version used in preparing this section is IBM Each block in the grid is referred to as a cell. Note that
SPSS Statistics 25. when the data are entered in the SPSS spreadsheet,
SPSS operations and other on-screen terms are they will look different: for example, 1 becomes 1.00
presented in bold (e.g., Analyze). The names given to (this can be adjusted if necessary).

* All screen captures in this appendix are reprinted courtesy of International Business Machines Corporation, © International
Business Machines Corporation.

bel65796_app_374-403 374 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 375

Box A.1 Basic operations in SPSS

• IBM SPSS Statistics Data Editor. This is the mode • Value Label. This is a label attached to the
in which data may be entered and subsequently code used when entering the data. Thus, for
edited and defined. It is made up of two screens: var00001, the label male can be attached to
Data View and Variable View. Move between 1 and the label female to 2. In output, such as
these two views by selecting the appropriate a frequency table or chart, the labels for each
button at the bottom left of the screen. value appear, making the interpretation of
• Data View. This is the spreadsheet into which output easier than if the code were used. It is
the data are entered. It appears when SPSS is generated from the Variable View.
started. • Missing Values. If data for a particular
• Variable View. Another spreadsheet, this one variable are missing, you must specify how
displays information about each of the vari- missing values are coded so that the com-
ables (such as the variable name, the variable puter can omit them from the calculations.
labels, and value labels) and allows that infor- Missing values are generated from the
mation to be changed (see below). Variable View.
• IBM SPSS Statistics Viewer. After a statistical • Analyze. This is the button on the menu bar
analysis or the production of a graph (some- along the top of the Data Editor from which
times called a “chart” in SPSS), the output is (via a drop-down menu) the method of anal-
deposited here. The output window superim- ysis is selected. Note that whenever a menu
poses itself over the Data Editor after an anal- item appears with a right-pointing arrowhead
ysis has been performed or a graph has been after it, a further sub-menu is available.
generated. • Graphs. This is the button (at the top of the
• Variable Name. This is the name given to a Data Editor) used to get access to the Chart
variable (for example, gender). Until they are Builder (via a drop-down menu), which can be
given names, variables are referred to as used to create various graphs and charts.
var00001, var00002, and so on. Once a vari- • Chart Builder. This feature is used to make
able is given a name, the name appears in charts and graphs. A very useful tutorial is
the column for that variable in the Data View available from the Data Editor by clicking
window. It is generated from the Variable View. Help → Topics. Use the search engine to find
• Variable Label. This is a label attached to a Tutorial → Creating and editing charts.
particular variable, and is optional. It is usu- • Chart Editor. This is used to make various
ally longer than the variable name and there- changes to graphs. To activate it, double-click
fore more explanatory. Spaces can be used anywhere in the graph. A small Chart Editor
(for example, reasons for visiting). The label window appears along with a version of the
appears in any output generated and comes main graph and stays until the Chart Editor is
from the Variable View. exited.

Entering data in the Data View cell—that is, 1. This number goes directly into that
To input the data given in Box 8.2 into the Data View, cell and into the box beneath the toolbar. Although
make sure that the top left-hand cell in the grid is you can use the mouse, many people find it easier to
highlighted (Plate A.1). If it is not, simply click once use the arrow keys on their keyboard to move from
in that cell. Then type the appropriate number in that cell to cell. If you make a mistake, simply click once

bel65796_app_374-403 375 12/16/21 06:12 PM

376 Appendix Using IBM SPSS Statistics and NVivo Software

in the cell in question, type in the correct value, and data from the gym survey entered (though only part
hit the Enter key. The last piece of data goes into the of the data set is visible). The first row of data contains
bottom right-hand cell of what will be a perfect rect- the coded answers (provided by the first respondent)
angle of data. Plate A.2 shows the Data View with the from the completed questionnaire in Box 8.1.

Each row represents a case. Each column represents a variable.

PLATE A.1 The SPSS

Data Editor Data View

For the Variable View to examine each

variable in detail, click on this tab.

This row shows the data for the first person

who answered the questionnaire in Box 8.1.

PLATE A.2 The Data

View with the “gym
study” data entered

bel65796_app_374-403 376 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 377

To proceed further, SPSS works in the following Defining variables: Variable names,
typical sequence for selecting variables and analyz- missing values, variable labels,
ing data. and value labels
Once you have finished entering the data, prepare the
1. Make a selection from the menu bar at the top variables. The following steps explain how this is done:
of the screen, for example, → Analyze.
2. From the menu that will appear, make a selec- 1. → Variable View tab at the bottom left of the Data
tion, for example, → Descriptive Statistics. Editor [opens the window shown in Plate A.3]
3. This will bring up another menu where you 2. To name a variable, click on the current variable
will specify which statistics you want, for ex- name (for example, var00003) and type a name
ample, Frequencies. for it (for example, reasons). This name, which
4. This will bring up a dialogue box in which you cannot include spaces, is the identification
provide further information, such as the vari- needed for instructing the computer to perform
ables to be analyzed. an analysis with this variable. It can be anything
5. Very often, further information is needed; → a you like, but you should choose a name that in-
button that brings up a sub-dialogue box. dicates what the variable is measuring.
6. Provide the information in the sub-dialogue box 3. To give the variable a more detailed designa-
and then go back to the dialogue box. Sometimes, tion, known in SPSS as a variable label, → the
a further sub-dialogue box is required before you cell in the Label column for this variable. Then
can return to the dialogue box. type in the label (for example, reasons for vis-
iting). This label will appear on your output,
When you have finished going through the entire just as a designer label appears on a piece of
procedure, → OK. The toolbar beneath the menu bar clothing. If no variable label is specified, the
allows shortcut access to certain SPSS operations. variable name will be used for the output.

To insert Value Labels for var00003, click here.

PLATE A.3 The Data Editor Variable View

bel65796_app_374-403 377 12/16/21 06:12 PM

378 Appendix Using IBM SPSS Statistics and NVivo Software

4. Provide “value labels” for the values of the vari-

ables, where appropriate. The procedure gener-
ally applies to variables that are not interval/ratio
(the interval/ratio level of measurement is called
“scale” in SPSS terminology). Scale variables do
not need value labels unless you are grouping
them in some way. To assign value labels, → in
the cell in the Values column for the variable in
question. A small button with three dots on it ap-
pears. → the button. The Value Labels dialogue
box appears (Plate A.4). → the box to the right of
Value and begin to define the value labels. To do
this, enter the value (for example, 1) in the area to
the right of Value and then type in the value label PLATE A.5 The Missing Values Dialogue Box
(for example, relaxation) in the area to the right
of Label. Then → Add. Do this for each value. In the Missing Values dialogue box, enter
When finished, → OK. A variable name for ques- the missing value (0) below Discrete missing
tion one on the questionnaire could be gender; values and then → OK.
the variable label “gender of respondent;” and 6. Designate the level of measurement of the vari-
the value labels, 1, “male” and 2, “female.” The able. Reasons is a nominal variable. To enter
computer reads only numbers (male is 1; female, this information, → the cell for this variable in
2) so the value labels are optional, but if you omit the Measure column. A drop-down menu will
them the output will tell you about 1s and 2s in- appear; → Nominal.
stead of males and females.
5. Designate missing values. In the case of rea- To simplify the following presentation, reasons is the
sons, missing data are given a value of 0. To only variable for which a variable label, value labels,
assign the missing value, → the cell for this missing values, and the level of measurement are
variable in the Missing column. Again, → the defined.
button that has three dots on it. This will gener-
ate the Missing Values dialogue box (Plate A.5). Saving the data
To save the data for future use, make sure that the
Data Editor is the active window. Then, → File
→ Save As. The Save Data As dialogue box will
appear. It needs a name for the data, which is placed
after File name and a place to save the data—for ex-
ample, onto a USB drive or the hard drive of your
computer. To select the destination drive, → the
downward pointing arrow to the right of the box
by Look in. Then choose the drive on which to save
the data and → Save.
This procedure saves the data and any work
done on it—for example, value labels and missing
values specifications. If you do more work on the
Remember to click here after entering data, such as creating a new variable, the data must
each Value and Value Label.
be saved again or the new work will be lost. SPSS
PLATE A.4 The Value Labels Dialogue Box gives you a choice of renaming the data—in which

bel65796_app_374-403 378 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 379

case there will be two data files (one with the origi- and other activities). Adding them up gives the total
nal data and one with the revised data)—or keeping number of minutes spent on various activities in the
the same name. In the latter case the original file gym, and a new variable, ­totalmin, can be created for
is lost but its name is retained and applied to the that purpose. To do this, take the following steps:
modified file.
1. → Transform → Compute Variable . . . [opens
Retrieving data the Compute Variable dialogue box shown in
To retrieve a data file, → File → Open → Data . . . . Plate A.6]
The Open Data dialogue box will appear. Go to the 2. Under Target Variable, type totalmin
location in which the data are deposited to retrieve 3. Under Function group, → Statistical; under
the file containing the data, and then → Open. Functions and Special Variables, → Sum; then →
the button with an upward-pointing arrowhead
Computing a new variable to send it into the box under Numeric Expression
The total amount of time a respondent spent in the gym 4. From the list of variables at the left, → cardmins
is made up of three variables: cardmins, weimins, and → adjacent button [puts cardmins in box after
othmins (questions 10, 11, and 12 ­respectively in Box SUM]; → weimins → button [puts weimins in
8.1, referring to cardiovascular equipment, weights, box after cardmins]; → othmins → button [puts

PLATE A.6 The Compute Variable Dialogue Box

bel65796_app_374-403 379 12/16/21 06:12 PM

380 Appendix Using IBM SPSS Statistics and NVivo Software

othmins in box after weimins]; be sure to put 1. → Transform → Recode into Different
commas (but not spaces) between the variable Variables [opens the Recode into Different
names, and to delete any “?” that remain in the Variables dialogue box shown in Plate A.7]
Numeric Expression box 2. → age → adjacent button [puts age in the
5. → OK Numeric Variable → Output Variable box];
→ box beneath Output Variable Name
The new variable totalmin is created and appears in and type agegp → Change [puts agegp in
the Data Editor. the Numeric Variable → Output Variable
box]; → Old and New Values . . . [opens the
Recoding variables Recode into Different Variables: Old and
Sometimes you will need to recode variables—for ex- New Values sub-dialogue box shown in
ample, to group scores together, as was done to produce Plate A.8]
Table 8.3 for an interval/ratio variable (var00002, now 3. If there are missing values for a variable
given the variable name age). SPSS offers two choices: (as in this case), in the Old Value box, →
change age, or keep age as it is and create a new vari- the circle by System- or user-missing, and
able. The latter option is appropriate here, because you in the New Value box, → System-missing,
may need the data for age in its original form for some then → Add
other analysis. We will preserve the original variable 4. In the Old Value box, → circle by Range:
and create a new one, agegp, for age groups, with five LOWEST through value and type 20 in the
age categories, as in Table 8.3. box. In the New Value box, → circle by Value

Original name Recoded name

of variable. of variable.

Click here to specify how

values are to be recoded.

PLATE A.7 The Recode into Different Variables Dialogue Box

bel65796_app_374-403 380 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 381

and type 1; → Add [the new value appears in The new variable agegp is created and appears in the
the Old → New box] Data View. To generate value labels for the five age
5. In the Old Value box, → first circle by Range bands and a variable label, repeat the approach de-
and type 21, and in the box after through type scribed above for those procedures.
30. In the New Value box, → circle by Value
and type 2; → Add
6. In the Old Value box, → first circle by Range Using SPSS for data analysis
and type 31 and in the box after through type
40. In the New Value box, → circle by Value Generating a frequency table
and type 3; → Add To produce a frequency table like the one in Table 8.2:
7. In the Old Value box, → first circle by Range
and type 41, and in the box after through type 1. → Analyze → Descriptive Statistics →
50. In the New Value box, → circle by Value Frequencies . . . [opens the Frequencies dia-
and type 4; → Add logue box shown in Plate A.9]
8. In the Old Value box, → circle by Range: value 2. → reasons → adjacent button [puts reasons into
through HIGHEST and type 51 in the box. Variable(s) box]
In the New Value box, → circle by Value and 3. → OK
type 5; → Add → Continue [closes the Recode 4. The table then appears in the IBM SPSS
into Different Variables: Old and New Values Statistics Viewer (see Plate A.10)
sub-dialogue box shown in Plate A.8 and re-
turns to the Recode into Different Variables Note that in the Frequencies dialogue box, the vari-
dialogue box shown in Plate A.7] able name appears. If you would prefer to work with
9. → OK variable labels (which provide more information on

PLATE A.8 The Recode into Different Variables: Old and New Values Sub-Dialogue Box

bel65796_app_374-403 381 12/16/21 06:12 PM

382 Appendix Using IBM SPSS Statistics and NVivo Software

what the variable measures), go into the Data Editor, Generating a bar chart
→ Edit → Options → General, and in the top left- To produce a bar chart like the one in Figure 8.2:
hand corner you can select variable labels to be used.
Similarly, if you prefer that the variables be arranged 1. → Graphs → Chart Builder → Define Variable
in alphabetical order, you can specify that there. Properties [opens the Define Variable
Properties dialogue box shown in Plate A.11]
2. → reasons → adjacent button [puts reasons in
Click here to send selected the Variables to Scan box]
variables into Variable(s) list.
3. → Continue [opens the Define Variable
Properties sub-dialogue box shown in Plate
A.12]. This allows you to edit properties such as
variable labels, value labels, and so on before cre-
ating a graph, or to enter that information if you
have not already done so. We’ve already entered
all the information needed, so exit this box.
4. → Graphs → Chart Builder → OK [opens the
Chart Builder dialogue box shown in Plate A.13]
5. → Gallery → Bar, and drag the type of bar
chart you want into the “canvas,” the large area
above the Gallery
Select variable(s) to be 6. Drag reasons into the X-Axis? box at the
analyzed from here.
bottom of the canvas
PLATE A.9 The Frequencies dialogue box 7. → OK

Number of cases in each category.

Percentage of cases in each category.

Percentage of cases in each category
taking missing values into account.

PLATE A.10 The IBM SPSS Statistics Viewer

bel65796_app_374-403 382 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 383

PLATE A.11 The Define

Variable Properties
Dialogue Box

As mentioned earlier, for some useful details on 3. Drag reasons into the Slice by? drop zone at
this procedure, do the tutorial available in the Data the bottom of the canvas
Editor at Help → Topics. Use the search engine to 4. → OK
find Tutorial → Creating and editing charts. 5. Double-click anywhere in the chart to bring up
the Chart Editor (Plate A.14)
Generating a pie chart 6. → Elements → Show Data Labels. This will
To produce a pie chart like the one in Figure 8.3: place the percentage in each category in the
appropriate slice of the pie (see Plate A.15).
1. → Graphs → Chart Builder → OK [opens the
Chart Builder dialogue box] This pie chart is in colour, but if you only have access
2. → Gallery → Pie/Polar, and drag the pie chart to a monochrome printer, you can change the co-
into the “canvas,” the large area above the Gallery lours into patterns. Get into the Chart Editor, which

bel65796_app_374-403 383 12/16/21 06:12 PM

384 Appendix Using IBM SPSS Statistics and NVivo Software

PLATE A.12 The Define Variable Properties Sub-Dialogue Box

PLATE A.13 The Chart Builder Dialogue Box

bel65796_app_374-403 384 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 385

PLATE A.14 The Chart Editor

PLATE A.15 The Chart Editor Showing Pie

Chart Percentages

enables editing of all figures: → Edit → Properties → ages to be grouped together. The general procedure
Variables. By reasons for visiting (the variable label) for doing this is → Graphs → Chart Builder → OK
→ Style: Colour, and → Style: Pattern → Apply. Use → Gallery → Histogram, and drag the histogram
the same procedure for all charts. into the “canvas.” Then, → Element Properties →
Statistic: Histogram → Set Parameters. However,
Generating a histogram since we have already recoded age into the categories
Producing the histogram in Figure 8.4 is just as we want, you can use the newly created agegp vari-
simple, except that it requires you to define the able. Although technically it is an ordinal variable, to

bel65796_app_374-403 385 12/16/21 06:12 PM

386 Appendix Using IBM SPSS Statistics and NVivo Software

create a histogram it is necessary to define agegp as dialogue box. When the Print dialogue box ap-
a scale (interval/ratio) variable by right-clicking on pears, make sure Selected output under Print
it in the Variables list while in the Chart Builder. Range is selected.
Drag agegp into the X-Axis? drop zone at the bottom
of the canvas; → OK. Other ways of presenting inter- Generating the arithmetic mean,
val/ratio data, such as measures of central tendency median, standard deviation, and range
and dispersion, will be discussed below. To produce the mean, median, standard deviation,
and the range for an interval/ratio variable like age,
Printing output follow these steps:
To print all the output in the IBM SPSS Statistics
Viewer, make sure that the Viewer is the active 1. → Analyze → Descriptive Statistics → Explore
window and then → File → Print . . . . When the . . . [opens the Explore dialogue box]
Print dialogue box appears, → OK. To print just 2. → age → the button to the left of Dependent
some of the output, hold down the Ctrl button List [puts age in the Dependent List box]; →
on the keyboard and click once on the parts of Statistics under Display; → OK
the output you want to print. The easiest way to 3. The output also includes the 95 per cent confi-
do this is to select the elements from the output dence interval for the mean, which is based on
summary in the left-hand segment of the Viewer the standard error of the mean. The output is
shown in Plate A.10. Then bring up the Print in Table A.1.

TABLE A.1 Explore output for age (SPSS output)

EXPLORE
CASE PROCESSING SUMMARY
Cases
Valid Missing Total
N Per cent N Per cent N Per cent
AGE 89 98.9% 1 1.1% 90 100.0%
DESCRIPTIVES
Statistic Std. Error
AGE Mean 33.5955 .9420
95% Confidence Lower bound 31.7235
Interval for mean Upper bound 35.4675
5% Trimmed mean 33.3159
Median 31.0000
Variance 78.971
Std. deviation 8.8866
Minimum 18.00
Maximum 57.00
Range 39.00
Interquartile range 14.0000
Skewness .446 .255
Kurtosis 2.645 .506

bel65796_app_374-403 386 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 387

Generating a contingency table simply → at the relevant point. → Continue

To generate a contingency table like Table 8.4, follow [closes the Crosstabs: Cell Display sub-
this procedure: dialogue box and returns to the Crosstabs
­dialogue box shown in Plate A.16]
1. → Analyze → Descriptive Statistics → 4. → Statistics . . . [opens the Crosstabs: Statistics
Crosstabs . . . [opens the Crosstabs dialogue sub-dialogue box shown in Plate A.18]. For
box shown in Plate A.16] example, suppose Cramér’s V is needed (be-
2. → reasons → button by Row(s) [reasons cause of having two nominal variables). → Chi-
appears in the Row(s) box]; → gender → square → Phi and Cramér’s V → Continue
button by Column(s) [gender appears in [closes the Crosstabs: Statistics sub-dialogue
the Column(s) box]; → Cells [opens the box and returns to the Crosstabs dialogue box
Crosstabs: Cell Display sub-dialogue box shown in Plate A.16]
shown in Plate A.17] 5. → OK
3. Make sure Observed in the Counts box is se-
lected. Make sure Column under Percentages The resulting output is in Table A.2. Cramér’s V is
is selected. If either of these was not selected, just one of the available statistics.

If a variable can be identified as a likely

independent variable, place it here.

PLATE A.16 The Crosstabs Dialogue Box

bel65796_app_374-403 387 12/16/21 06:12 PM

388 Appendix Using IBM SPSS Statistics and NVivo Software

TABLE A.2 Contingency table for reasons for visiting by gender (SPSS output)
CROSSTABS
CASE PROCESSING SUMMARY
Cases
Valid Missing Total
Case Processing
N Per cent Summary Per cent N Per cent
reasons for visiting * 90 100.0% 0 .0% 90 100.0%
GENDER

REASONS FOR VISITING * GENDER CROSSTABULATION

GENDER
male female Total
reasons
for
visiting relaxation Count 3 6 9
% within GENDER 7.1% 12.5% 10.0%
fitness Count 15 16 31
% within GENDER 35.7% 33.3% 34.4%
lose weight Count 8 25 33
% within GENDER 19.0% 52.1% 36.7%
build Count 16 1 17
strength
% within GENDER 38.1% 2.1% 18.9%
Total Count 42 48 90
% within GENDER 100.0% 100.0% 100.0%

CHI-SQUARE TESTS
This is the χ2 value Asymp. Sig.
referred to in the
Value df [(2-sided)]
text
Pearson Chi-Square 22.726a 3 .000
Likelihood Ratio 25.805 3 .000
Linear-by-Linear Association 9.716 1 .002
N of Valid Cases 90
2 cells (25.0%) have expected count less than 5. The minimum expected count is 4.20.
a

Since this is not a 2 × 2 table, interpret Cramér’s V

SYMMETRIC MEASURES
Approx. Shows level of statistical
Value Sig. significance of computed value
Nominal by Phi .503 .000 of Cramér’s V

Nominal Cramér’s V .503 .000

N of Valid Cases 90 Shows strength of relationship

bel65796_app_374-403 388 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 389

Must be selected for number of cases identified as likely to be the independent variable, it
in each cell to be included in table. is by convention placed on the X-axis (the horizontal
axis). Since age is bound to be the independent vari-
able, follow these steps:

1. → Graphs → Chart Builder → OK → Gallery →

Scatter/Dot
2. Drag the Simple Scatter diagram into the canvas
3. Drag age into the X-Axis? box at the bottom of
the canvas
4. Drag cardmins into the Y-Axis? box at the left
side of the canvas
4. → OK

Generating Pearson’s r
Select to give the percentage To produce Pearson’s r, in particular the correlation
of cases of each category of
between age, weimins, and cardmins shown in Table
the column variable.
8.5, follow these steps:
PLATE A.17 The Crosstabs: Cell
Display Sub-Dialogue Box 1. → Analyze → Correlate → Bivariate . . . [opens
the Bivariate Correlations dialogue box
Select to provide the 2 statistics
shown in Plate A.19]
for the contingency table.
2. → age → button → weimins → button → cardmins
→ button [age, weimins, and cardmins should
now be in the Variables box]; → Pearson (if not
already selected) → OK

The resulting output is shown in Table A.3.

To produce Kendall’s tau-b, follow the same
procedures, but instead of selecting Pearson, →
Kendall’s tau-b. Spearman’s rho is another option.

Comparing means
To produce a table like Table 8.6, follow these steps:

1. → Analyze → Compare Means → Means . . . [opens

the Means dialogue box shown in Plate A.20]
Select to generate Phi/Cramér’s V.
2. → cardmins → button to the left of Dependent
PLATE A.18 The Crosstabs: List → reasons → button to the left of
Statistics Sub-Dialogue Box Independent List → Options [opens the
Means: Options sub-dialogue box]
3. → Anova table and eta under Statistics for
Generating scatter diagrams First Layer → Continue [closes the Means:
The production of scatter diagrams, known as scat- Options sub-dialogue box and returns to
terplots in SPSS, is illustrated in the relationship the Means dialogue box shown in Plate
between age and cardmins. If one variable can be A.20]; → OK

bel65796_app_374-403 389 12/16/21 06:12 PM

390 Appendix Using IBM SPSS Statistics and NVivo Software

All variable(s) to be
correlated go here.

Select Pearson , Kendall's tau-b, or Spearman

depending on the kinds of
variables being analyzed.

PLATE A.19 Bivariate Correlations Dialogue Box

TABLE A.3 Correlations output for age, weimins, and cardmins (SPSS output)
CORRELATIONS Correlations of p < 0.05 are “flagged” with asterisks
AGE WEIMINS CARDMINS
AGE Pearson Correlation 1.000 −.273 ** −.109
Sig. (2-tailed) . .010 .311
N 89 89 89
WEIMINS Pearson Correlation −.273 ** 1.000 −.161
Sig. (2-tailed) .010 . .130
N 89 90 90
CARDMINS Pearson Correlation −.109 −.161 1.000
Sig. (2-tailed) .311 .130 .
N 89 90 90
**Correlation is significant at the 0.01 level (2-tailed).

Shows strength of Shows number of cases, less Shows level of

relationship as any cases for which there statistical significance
indicated by are missing data for either of computed value of
Pearson’s r or both variables Pearson’s r

bel65796_app_374-403 390 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 391

Dependent variable(s) go here.

Independent variable(s) go here.

PLATE A.21 The Linear Regression Dialogue Box
PLATE A.20 The Means Dialogue Box

Generating a contingency table with Regression

three variables To perform a bivariate regression in SPSS:
To create a table like Table 8.9, do as follows:
1. → Analyze → Regression → Linear [opens the
1. → Analyze → Descriptive Statistics → Crosstabs Linear Regression dialogue box shown in
. . . [opens the Crosstabs dialogue box shown in Plate A.21]
Plate A.16] 2. → weimins → button to the left of the Dependent
2. → othsourc → button by Row(s) [othsourc ap- box
pears in the Row(s) box] 3. → age → button to the left of the Independent(s)
3. → age3 (assume this is the name given to a box
newly created variable with age recoded into 4. → OK. The output appears in Table 8.10.
the three categories listed in Table 8.9); → button
by Column(s) [age3 appears in the Column(s) To perform multiple regression, simply add addi-
box]; → gender → button in Layer 1 of 1 box tional independent variables to the model. Once
[gender appears in the box under Layer 1 of you are in the Linear Regression dialogue box (de-
1]; → Cells [opens the Crosstabs: Cell Display scribed above and shown in Plate A.21), just click
sub-dialogue box shown in Plate A.17] on whatever variable you would like to add, then
4. Make sure Observed in the Counts box has click the arrow button to put the variable in the
been selected. Make sure Column under Independent(s) box along with age.
Percentages has been selected. If either has
not, simply click at the relevant point. → Questions for Review
Continue [closes the Crosstabs: Cell Display
sub-dialogue box and returns to the Crosstabs Getting started in SPSS
dialogue box shown in Plate A.16] • Outline the differences between variable
5. → OK names, variable labels, and value labels.
• In what circumstances is it appropriate to
The resulting table looks somewhat different from recode a variable?
Table 8.9 in that gender appears as a row rather than • In what circumstances is it appropriate to
as a column variable. create a new variable?

bel65796_app_374-403 391 12/16/21 06:12 PM

392 Appendix Using IBM SPSS Statistics and NVivo Software

Exercises: Data analysis with Qualitative data analysis

SPSS (see Table 8.1 to identify
the variables)
using NVivo
Introduction
Using the gym survey data, create:
To illustrate how to use NVivo software, version 10, we
will use the interview transcripts from a study of vis-
• a frequency table for the variable exercise;
itors to Disney theme parks, discussed in Chapter 13.
• a bar chart and pie chart for exercise and com-
This illustration of NVivo and its functions addresses
pare their usefulness;
just its most basic features. There is a very good help
• a histogram for cardmins;
facility included with this program as well as tutorials
• measures of central tendency and dispersion
that are recommended for more in-depth instruction.
for cardmins;
As in the discussion of SPSS, in the following account
• a contingency table and Cramér’s V for gender
→ signifies “left-click once with the mouse.”
and exercise;
• a scatter diagram for age and cardmins;
• Pearson’s r for age and cardmins; Getting started with NVivo
• Kendall’s tau-b for carduse and weiuse; and On opening NVivo, you will be presented with a
• a difference of means analysis for reasons and welcome screen (Plate A.22). This screen shows any
totalmin. existing NVivo projects and is the springboard for

PLATE A.22 The Opening Screen

bel65796_app_374-403 392 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 393

either opening one of the existing projects or starting notes. NVivo 10 can accept documents in both
a new one. If you are starting a new project, as in the rich text (.rtf) and Word (.doc, .docx) formats. To
example that follows, → File → New Project. The New import the documents (see Plate A.23), → Internals
Project dialogue box appears and you are asked to (below Sources at the top of the Navigation view)
provide a Title for your project. For this exercise, the → External Data on the ribbon (see Plate A.24 for a
title “Disney Project” was chosen. You are also asked description of how to find the ribbon and other key
to give a Description of the project, although this is areas of the screen) → Documents button on the find
an optional feature. When you have done this, → OK. bar [opens the Import Internals dialogue box] →
You then need to import the documents you want Browse to locate the documents that are to be im-
to code. In this case, they will be interview transcripts ported. → the documents to be imported (you can
from the project on visitors to Disney theme parks, hold down the Ctrl key to select several documents;
referred to in Box 11.5 and Chapter 13. Other kinds if you want to select all of them, hold down the Ctrl
of documents can be imported such as fieldwork key and tap the A key); → Open. The documents

2. Select External Data and then 3. In the Import Internals dialogue

1. Select Documents. This brings up the box, click on Browse… to locate
Internals. Import Internals dialogue box. the documents to be imported.

4. Select the documents to be 5. Click on Open. The

imported from the location documents will then be
identified in step 3. imported into NVivo.

PLATE A.23 Stages in Importing Documents into NVivo

bel65796_app_374-403 393 12/16/21 06:12 PM

394 Appendix Using IBM SPSS Statistics and NVivo Software

Ribbon—contains the main

NVivo commands. The Find
bar changes when you select
a different command.

Find bar—to search for items

in your NVivo project.

List view—displays the

contents of your folders.

Detail view—here you can

examine contents of your
documents, nodes, etc.

Quick coding bar.

Navigation view—provides
access to documents,
nodes, etc.

PLATE A.24 The Document Viewer and Its Components

will then be visible in the document viewer (this is There are several ways of going about the coding
the term used to describe the general screen shown process in NVivo. The approach taken in relation to
in Plate A.24). Once the documents have been im- the coding of the Disney Project was as follows:
ported, they can be read and edited. All you need to
do is to double-click on the yellow icon to the left of • The interviews, both in printed form and in
each interview in the list view. the document viewer (Plate A.24), were read
through. The viewer is treated as having a
Coding number of different components or sections
Coding your data is obviously one of the key phases and these are highlighted in Plate A.24.
in the whole process of qualitative data analysis. • Some codes that seemed relevant to the docu-
For NVivo, coding is accomplished through nodes. ments were created.
NVivo’s help system in earlier releases defined coding • The documents were then coded using NVivo.
as “the process of marking passages of text in a proj-
ect’s documents with nodes” (emphasis added). Thus, An alternative strategy is to code while browsing the
nodes are the route by which coding proceeds. In documents.
turn, the latest release defines a node as “a collection
of references about a specific theme, place, person or Creating nodes
other area of interest.” When a document has been The nodes that were relevant to the passage in Box
coded, the node will incorporate references to those 11.5 are presented in Figure A.1. Prior to NVivo 9,
portions of documents in which the code appears. when creating a node, the researcher had to choose
Once established, nodes can be changed or deleted. either a free node or a tree node. The latter is a node

bel65796_app_374-403 394 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 395

that is organized in a hierarchy of connected nodes, 1. In the document viewer → Create in the ribbon
whereas free nodes were not organized in this way. 2. → Node in the find bar [opens the New Node
This distinction has been dropped in versions after dialogue box shown in Plate A.25]
NVivo 9, and the software assumes that a tree node is 3. Enter the node Name (Not critical of Disney)
being created. Two points are crucial to note here for and a Description (the latter is optional)
users of earlier releases of the software. First, the ten- 4. → OK
dency now is not to refer to “tree nodes” but to treat
them as hierarchically organized nodes. Second, Creating hierarchically organized nodes
free nodes (that is, nodes that are not hierarchi- To create a hierarchically organized node, the ini-
cally organized) can still be created; they are simply tial process is exactly the same as with a non-hier-
nodes without “children” (to use the latest NVivo archically organized node. The following example
terminology). will show how to create the hierarchically organized
Notice that there are three groups of hierarchi- node Class critique, which is a child of the hierar-
cally organized nodes and two non-hierarchically chically organized node Ideological critique, which
organized nodes in Figure A.1. The nodes can be cre- is itself a child of the hierarchically organized node
ated in the following way. Critique (see Figure A.1). The following steps will
generate this node.
Creating non-hierarchically organized
nodes 1. In the document viewer → Create in the ribbon
This sequence of steps demonstrates how to create 2. →Node in the find bar [opens the New Node
the non-hierarchically organized node Not critical dialogue box shown in Plate A.26]
of Disney. 3. Enter the node Name (Critique) and a
Description (the latter is optional)
Hierarchical nodes Critique
4. →OK
5. →Critique in the list of nodes in the list viewer
Aesthetic Ideological Content Other 6. →Node in the find bar [opens the New Node
critique critique critique critique dialogue box shown in Plate A.26]
7 Enter the node Name (Ideological critique)
and a Description (the latter is optional). This
Class Ethnicity Gender Nationality node will form a child of the hierarchically or-
critique critique critique critique
ganized node (make sure that in Hierarchical
name it reads Nodes\\Critique, as this will
Visitors
mean it is a child of Critique). See Plate A.26.
8. → Ideological critique in the list of nodes in
Visitors’ Visitors’ Visitors‘ Visitors‘ the list viewer
age gender class ethnicity
9. → Node in the find bar [opens the New Node
dialogue box shown in Plate A.26]
Significance
10. Enter the node Name (Class critique) and a
Description (the latter is optional). This node
Educational Recreational Ideological Other
will form a child of the hierarchically orga-
significance significance significance significance nized node (make sure that in Hierarchical
name it reads Nodes\\Critique\Ideological
Non-hierarchical nodes critique, as this will mean it is a child of
Uncritical enthusiasm Ideological critique, which is itself a child of
Not critical of Disney Critique). See Plate A.26.
FIGURE A.1 Nodes used in the Disney Project 11. → OK

bel65796_app_374-403 395 12/16/21 06:12 PM

396 Appendix Using IBM SPSS Statistics and NVivo Software

1. Select Create. 2. Select Node.

3. In the New Node dialogue box, 4. Click on OK.

enter the node Name and a
Description (latter is optional).

PLATE A.25 Stages in Creating a Non-Hierarchically Organized Node

Applying nodes in the coding process want to use has been created, NVivo 10 allows you
Coding is carried out by applying nodes to segments to drag and drop text into it (see Plate A.27). To do
of text. Once you have set up some nodes (remember this, highlight the text to be coded and then, holding
that you can add and alter them at any time), you down the left-hand button, drag the text over to the
can look at a document in the viewer and highlight appropriate node in the list view.
the area of the document that you want to code; then Alternatively, you can highlight the text you want
right-click on the mouse while holding the cursor to code and right-click over it. Then → Code Selection
over the highlighted text. If the node that you want → Code Selection at Existing Nodes, which opens
to use has not been created yet, highlight the text the Select Project Items dialogue box (see Plate
you want to code, right-click on the highlighted text, A.28). → the node(s) you want to use. Thus, in the
and then → Code Selection → Code Selection at example in Plate A.28, the tick by Uncritical enthu-
New Node . . . . This opens the New Node dialogue siasm will code the highlighted text at that node. If
box. You can then create a new node in the manner you also wanted to use a hierarchically organized
outlined in the previous sections. If the code you node, you would need to find the appropriate parent

bel65796_app_374-403 396 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 397

3. If the node is a child of an existing

1. Click on 2. Click on node, make sure that the appropriate
Create Node node has been selected

4. In the New Node dialogue box, 5. Click on OK

enter the node Name and a
Description (latter is optional) Hint: Ensure that the
correct sequence of
children has been
selected here
Ensure that Nodes
has been selected
here

PLATE A.26 Stages in Creating a Hierarchically Organized Node

in the list of nodes within the list view and then → Selecting this feature allows you to see multi-
the plus sign to the left of it. To uncode at any point, coloured stripes that represent portions of coded
simply highlight the passage to be uncoded, and → text and the nodes that have been used. Overlapping
the button with a red cross in it in the quick coding codes do not represent any problem at all.
bar (see Plate A.24). Alternatively, you can right-click To activate this feature, → View in the ribbon
on the highlighted text and → Uncode. and then → Coding Stripes in the find bar → Nodes
Recently Coding. Plate A.29 shows these stripes.
Coding stripes We can see that some segments have been coded at
It’s very helpful to be able to see the areas of text that two or more nodes (for example, Visitors’ ethnicity
have been coded and the nodes applied to them. To and Ethnicity critique). All the nodes that have been
this end NVivo has a useful aid called coding stripes. used are clearly displayed.

bel65796_app_374-403 397 12/16/21 06:12 PM

398 Appendix Using IBM SPSS Statistics and NVivo Software

Hint: To uncode at
any time, highlight the
text to be uncoded and
click on this button.
This will clear the
coding at that point.

Highlight text to be coded …drop into the appropriate

and, holding down the left- node, in this case
hand button of the mouse… uncritical enthusiasm. PLATE A.27 Using Drag and
Drop to Code

3. Click on OK.

1. Highlight the text to be coded, 2. Select the appropriate node

right click, and, from the menu, select by clicking on the box to the
Code Selection and then select left of the node.
Code Selection at Existing Nodes.
This brings up the Select Project
Items dialogue box.
PLATE A.28 Coding in NVivo

bel65796_app_374-403 398 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 399

PLATE A.29 Coding Stripes

Searching text 3. → on the + to the left of Ideological critique

Once you have coded your data, however preliminary [this brings up a list of all branches of the node
that coding may be, at some point you will want to Ideological critique]
conduct a search of your data. For instance, suppose 4. Double-click on Ethnicity critique
that you want to retrieve all occurrences of a particular 5. All instances of coded text at the node
node. NVivo allows you to trawl rapidly through all Ethnicity critique will appear at the bottom of
your documents and find all the text that was coded at the screen, as in Plate A.30.
a particular node. This is very easy to do in NVivo 10.

To search for occurrences of a single node To search for the intersection of two nodes

These steps describe how to conduct a search for se- This section is concerned with searching for se-
quences of text that have been coded in terms of the quences of text that have been coded at two nodes:
node Ethnicity critique. The stages are outlined in Aesthetic critique and Not critical of Disney. This
Plate A.30. type of search is known as a Boolean search. It will
locate text coded in terms of the two nodes to-
1. In the document viewer → Nodes in the nav- gether (that is, where they intersect), not text coded
igation view. This will bring up your list of in terms of each of the two nodes. The steps are as
nodes in the list view. follows:
2, If you cannot find the parents of Ethnicity cri-
tique, → on the little box with a plus sign (+) to 1. In the document viewer, → Queries in the nav-
the left of Critique [this brings up a list of all igation view
branches of the node Critique] 2. → Query tab in the ribbon

bel65796_app_374-403 399 12/16/21 06:12 PM

400 Appendix Using IBM SPSS Statistics and NVivo Software

1. Click on Nodes. 2. Locate the node to be

analyzed and double click on it.

PLATE A.30 Stages in Retrieving Text from a Node

3. → Coding . . . from the menu of options [opens 1. → Home on the ribbon

the Coding Query dialogue box shown in 2. → Find . . . [opens the Find Content dialogue
Plate A.31] box shown in Plate A.32]
4. → Coding Criteria tab 3. Type magic kingdom to the right of Text.
5. → Advanced tab 4. To the right of Look in, make sure Text has
6. In the Define more criteria panel, → Coded at been selected.
from the drop-down menu 5. → Find Next
7. → Select. You then need to choose the two
nodes to be analyzed from the Select Project Text searching can be useful for the identification
Items dialogue box. of possible in vivo codes (introduced in Chapter 13).
8. Once the nodes have been selected, → OK You would then need to go back to the documents to
which returns you to the Coding Query dia- create nodes to allow you to code in terms of any in
logue box → Add to List vivo codes.
9. Make sure AND has been selected immedi-
ately below Define more criteria Output
10. → Run To find the results of coding at a particular node, →
the Nodes button in the bottom left. This will bring
up your node structure. Find the node that you are
To search for specific text interested in and simply double-click on it. This will
NVivo can also perform searches for specific words or bring up all text coded at that node along with in-
phrases, often referred to as “strings” in computer jargon. formation about which interview(s) the text comes
For example, to search for magic kingdom, you would: from.

bel65796_app_374-403 400 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 401

3. Select Coding. This brings up

the Coding Query dialogue box.
2. Select Query.
4. Select the Coding
Criteria tab.

5. Select the
Advanced tab.

8. Choose the nodes to

be analyzed from the
Select Project Items
dialogue box and
click on Add to List.
They will appear here.

7. Click on Select (opens

the Select Project Items
dialogue box).

6. Select Coded at.

1. Click on 9. Ensure AND has 10. Click on Run. PLATE A.31 The Coding
Queries. been selected. Query Dialogue Box (search-
ing for the intersection of two
nodes).

1. Select Home. 3. Insert text to be searched for 2. Select Find.

PLATE A.32 The Find Content

4. Select Find Next.
Dialogue Box

bel65796_app_374-403 401 12/16/21 06:12 PM

402 Appendix Using IBM SPSS Statistics and NVivo Software

Memos 5. To the right of Name, type in a name for the

In Chapter 13 we noted that one feature of the memo (e.g., Gender critique). You can also
grounded theory approach to qualitative data anal- provide a brief description of the document in
ysis is the use of memos in which ideas and illustra- the window to the right of Description, as in
tions might be stored. Memos can be easily created Plate A.33.
in NVivo. The steps, which are outlined in Plate 6. → OK
A.33, are as follows:

1. In the navigation view, → Sources Saving an NVivo project

2. Under Sources → Memos When you have finished working on your data, you
3. → Create tab on the find bar will need to save it for future use. To do this, on the
4. → Memo [opens the New Memo dialogue box menu bar at the top, → File → Save. This will save
shown in Plate A.33] all the work you have done. You will then be given

4. Select Memo. This brings up

2. Select Memos. 3. Select Create. the New Memo dialogue box.

1. Select Sources. 5. Enter the memo Name 6. Click on OK.

and Description here.

PLATE A.33 Stages in Creating a Memo

bel65796_app_374-403 402 12/16/21 06:12 PM

 Appendix Using IBM SPSS Statistics and NVivo Software 403

the opportunity to exit NVivo or to create or open Final thoughts

a project without worrying about losing all your As with SPSS, space limitations allow us to introduce
hard work. You might also consider backing up the only the most basic features of the NVivo software.
project. In so doing, we hope to have given readers unfamil-
iar with CAQDAS a sample of how the system works.
Opening an existing NVivo project No doubt some will decide that they prefer the tried-
To retrieve a project you have created, at the Welcome and-tested scissors-and-paste approach. On the
screen, → File → Open. This opens the Open Project other hand, the software warrants serious consider-
dialogue box. Search for and then select the project ation because of its power and flexibility.
you want to work on. Then → Open. Alternatively, For useful online help with NVivo, see the Online
simply click on the project you want to retrieve on QDA website and the CAQDAS Networking Project
the opening screen. website at:
You can also open a NUD*IST project, or one de- [Link]
signed using an earlier release of NVivo, by selecting w w [Link]/computer-assisted-
the appropriate project type from the drop-down qualitative-data-analysis
menu to the right of File name. (Websites accessed 4 September 2021)

Questions for review

Learning NVivo • In NVivo, what is the difference between a docu-
ment and a memo?
• What is a node?
• How do you go about searching for a single node
• What is the difference between a hierarchically or-
and the intersection of two nodes?
ganized node and a node that has not been hier-
archically organized? • Why might it be useful to display coding stripes?
• What is in vivo coding? • How do you search for specific text?
• Do nodes have to be set up in advance?

bel65796_app_374-403 403 12/16/21 06:12 PM

Glossary
Terms in italic type are defined elsewhere in the Glossary.
action research Same as participatory action research. code, coding In quantitative research, codes are the
tags used to assign the data on each variable to a cat-
adjacency pair Two kinds of talk activity that are
egory of the variable in question. Numbers are usually
linked together, such as an invitation and a response.
assigned to each category to allow easier computer
analytic induction An approach to the analysis of processing. In qualitative research, coding is the pro-
qualitative data in which the collection of data con- cess in which data are broken down into component
tinues and the hypothesis is modified until no cases parts, which are then assigned names.
inconsistent with it are found.
coding frame or coding manual A list of the codes to
arithmetic mean What everyday language refers to be used in the analysis of a particular set of data. For
as the “average”: the sum of all the scores divided by answers to a structured interview schedule or ques-
the number of scores. Also known simply as the mean. tionnaire, the coding frame delineates the categories
biographical method See life history method. used for each open question. With closed questions,
the coding frame is essentially incorporated into the
bivariate analysis Examination of the relationship fixed answers from which respondents must choose;
between two variables, as in contingency tables; hence the term “pre-coded question.”
correlation.
concept A general or abstract idea; a category
CAQDAS An abbreviation of “computer-assisted (or that serves to organize observations and ideas about
computer-aided) qualitative data analysis software.” some aspect of the social world.
case study A research design that entails detailed concurrent validity A type of validity that is tested by
and intensive analysis of either a single case or (for relating a measure to an existing criterion or a differ-
comparative purposes) a small number of cases. ent indicator of the concept to see if one predicts the
causality A connection between variables in which other; one of the main forms of measurement validity.
one variable changes as a result of a change in connotation A term used in semiotics to refer to the
­another, as opposed to a mere correlation between meanings of a sign associated with the social context
them. within which it operates: a sign’s connotations are
cells The areas in a table where the rows and col- supplementary to its denotation and less immediately
umns intersect and data are inserted. apparent.
census A count of an entire population; by contrast, a constant An attribute on which cases do not differ;
sample counts only some units of a population. compare with variable.
chi-square test Chi-square (χ2) is a test of statistical constructionism, constructionist An ontological
significance used to establish confidence that a find- position (the antithesis of objectivism) according to
ing displayed in a contingency table can be general- which social phenomena and their meanings are con-
ized from a probability sample to the population from tinually being created by social actors; also known as
which it is drawn. constructivism.
closed, closed-ended question A question in an in- construct validity (1) Same as measurement validity;
terview schedule or questionnaire that presents the (2) a type of measurement validity that is established
respondent with a fixed set of possible answers to by determining whether the concepts being measured
choose from; also called a fixed-choice question. relate empirically in a manner that would be predicted
by relevant theories.
cluster sampling A procedure in which the re-
searcher first samples sets of cases (“clusters”) and content analysis An approach to the analysis of
then samples units within them, usually using a proba- documents and texts that seeks to quantify content
bility sampling method. in terms of predetermined categories in a systematic

bel65796_glos_404-411 404 12/16/21 05:53 PM

 Glossary 405

and replicable manner. The term is sometimes used in deduction, deductive An approach to inquiry that
connection with qualitative research as well; see qual- begins with the statement of a theory from which hy-
itative content analysis. potheses may be derived and tested; compare with
inductive.
contingency table A table made up of rows and
columns that shows the relationship between two denotation A term used in semiotics to refer to
variables. Usually, at least one of the variables is a the principal and most manifest meaning of a sign;
nominal variable or ordinal variable. Each cell in the ­compare with connotation.
table shows the number or (more often) the percent-
dependent variable A variable that is caused (or as-
age of cases for that specific combination of the two
sumed to be caused) by an independent variable.
variables.
diary A written memoir. Three types of diary are
control group See experiment.
discussed in this book: personal diaries that are
convenience sample A sample that is selected be- ­spontaneously produced and can be analyzed as per-
cause of its availability to the researcher; a form of sonal documents; diaries written at the behest of a
non-probability sample. researcher; and diaries written by researchers to log
their activities and reflections.
conversation analysis The fine-grained anal-
ysis of talk (recorded in naturally occurring sit- dichotomous variable A variable with just two
uations and then transcribed) to uncover the categories.
underlying structures in interaction that make social
dimension An aspect of a concept.
order possible. Conversation analysis is grounded in
ethnomethodology. discourse analysis An approach to the analysis of
talk and other forms of communication that empha-
correlation An approach to the analysis of rela-
sizes the way language can create versions of reality.
tionships between interval/ratio variables and/or or-
dinal variables that seeks to assess the strength and dispersion The degree of variation in a set of scores.
direction of the relationship between the variables Commonly used measures of dispersion include the
concerned. Pearson’s r and Spearman’s rho are both range and the standard deviation.
correlational measures. The corresponding term
ecological fallacy The error of assuming that
measure of association is often used with nominal
­inferences about individuals can be made from data
variables.
on the characteristics of groups.
covert research A term frequently used in con-
empiricism An approach to the study of reality
nection with ethnographic research in which the re-
­according to which only knowledge gained by obser-
searcher does not reveal their true identity and/or
vation through the senses is acceptable.
intentions. Such research may violate the ethical prin-
ciple of informed consent. epistemology, epistemological A branch of philosophy
concerned with what constitutes knowledge and how
Cramér’s V A statistical measure used to assess the
knowledge is to be acquired; see positivism, realism, and
strength of the relationship between two nominal
interpretivism.
variables.
eta A test of the strength of the relationship between
critical discourse analysis A type of content analy-
two variables. The independent variable is usually a
sis that brings issues such as power hierarchies, struc-
nominal variable while the dependent variable must
tural inequalities, and historical political struggles to
be an interval variable or ratio variable.
bear on the analyses of texts.
ethnography, ethnographer, ethnographic Like
critical realism A realist epistemology accord-
participant observation, a research method in which
ing to which the study of the social world should be
the researcher is immersed in a social setting for an
concerned with the identification of social structures
extended period of time, observing behaviour, asking
in order to change them and thereby counteract in-
questions, and analyzing what is said in conversations
equalities and injustices. Unlike positivism, which is
both between participants and with the fieldworker.
an empiricist epistemology, critical realism maintains
As a term, “ethnography” is more inclusive than “par-
that the structures may not be directly perceivable.
ticipant observation,” which emphasizes the observa-
cross-sectional design A research design in which tional component. Written accounts of ethnographic
data are collected at a single point in time. research are often referred to as ethnographies.

bel65796_glos_404-411 405 12/16/21 05:53 PM

406 Glossary

ethnomethodology A sociological perspective con- grounded theory An approach to the analysis of

cerned with the way social order is established and qualitative data in which the goal is to use the data
maintained through talk and interaction; the intellec- to generate theory; the data collection and analysis
tual foundation of conversation analysis. proceed in an iterative (recursive) fashion.
evaluation research Research that concentrates on hermeneutics An approach to the interpretation of
the evaluation of real-life interventions, such as policy texts that emphasizes the need to understand them
changes. from the perspective of their authors.

experiment A research design that rules out alter- hypothesis An informed speculation, which is set up
native explanations of findings deriving from it (in to be tested, about the possible relationship between
other words, that possesses internal validity) because two or more variables.
it involves (a) both an experimental group, which is IBM SPSS Statistics software A computer program
exposed to a treatment, and a control group, which that facilitates the management and analysis of quan-
is not, and (b) random assignment to the two groups. titative data.
experimental group See experiment. idiographic An approach to understanding that
external validity A type of validity that is achieved seeks specific, unique knowledge about a person or
if the results of a study can be generalized beyond group, typically concerning interpretations or mean-
the specific research context in which they were ings held by the persons studied.
generated. independent variable A variable that has (or is a
­ ssumed
face validity A type of validity that is achieved if, on to have) a causal impact on a dependent variable.
inspection, an indicator appears to measure the con- index See scale.
cept in question.
indicator Something employed to measure a con-
facilitator See moderator. cept; it may refer to any measure, but sometimes it
means an indirect measure used when no direct mea-
factor analysis A statistical technique used for large
sure is available.
numbers of indicators to establish whether there is a
tendency for groups of them to be interrelated. It is induction, inductive, inductivist An approach to in-
often used with multiple-indicator measures to see if quiry that begins with the collection of data, which are
they cluster into one or more groups (factors). then used to develop theories, hypotheses, and con-
cepts; compare with deductive.
field experiment A study in which the researcher di-
rectly intervenes in and/or manipulates a natural set- informed consent The principle that prospective
ting to observe what happens as a consequence. participants in social research should be given as
much information as they need to make a sound de-
field notes A detailed ethnographic account of cision about whether to participate in a study; a key
events, conversations, and behaviour, and the re- principle in social research ethics.
searcher’s initial reflections on them.
institutional ethnography A type of ethnography
focus group A form of group interview in which there that explores how institutional discourses (typically
are several participants (in addition to the modera- workplace texts) relate to people’s everyday experi-
tor/facilitator), there is an emphasis in the questioning ences with institutions, and how institutional relation-
on a particular topic or related topics, and interaction ships intersect with larger systems of social control
within the group and the joint construction of meaning and power in a society.
is observed.
interaction See statistical interaction.
frequency table A table that displays the number
and/or percentage of units (for example, people) in inter-coder reliability The degree to which two
different categories of a variable. or more individuals agree on the coding of an item;
a frequent concern in the coding of answers to open
gatekeeper A non-researcher who controls re- questions in research based on questionnaires or
searchers’ access to a research setting. structured interviews.
generalization, generalizability To generalize a re- internal reliability or internal consistency The
search finding is to apply it to people or groups who degree to which the items that make up a scale or
were not in the study. index are consistent or correlated.

bel65796_glos_404-411 406 12/16/21 05:53 PM

 Glossary 407

internal validity A type of validity that is achieved if of agreement with a series of attitude statements that
there is sufficient evidence that a causal relationship together form a multiple-indicator measure. The scale
exists between two or more variables. is designed to measure the intensity of respondents’
feelings about an issue.
interpretivism, interpretivist An epistemological
position that requires the social scientist to grasp the longitudinal research A research design in which
subjective meanings that people attach to their ac- data are collected on at least two separate occasions.
tions and behaviours.
mean See arithmetic mean.
intersubjectivity A condition in which two or more measure of central tendency A measure that summa-
observers of the same phenomenon are in agreement rizes the magnitude of a set of scores; examples include
as to what they have observed. Empiricists assume the arithmetic mean, median, and mode.
that intersubjectivity is possible insofar as knowledge
is based on data acquired through the senses. measure of dispersion A measure (such as the range
or standard deviation) that summarizes the amount of
interval variable A variable for which the intervals variation in a set of scores.
between the categories are identical and quantifiable.
measurement validity The degree to which a mea-
intervening variable A variable that is affected by an- sure of a concept actually measures what it is supposed
other variable and in turn has a causal impact on a third. to measure; see also face validity and concurrent
Taking an intervening variable into account often facili- validity.
tates the understanding of the relationship between two
median The mid-point in a set of scores that is
variables. An intervening variable is the “Y” in “X→Y→Z.”
­arranged in order.
interview guide A brief list of memory prompts re-
member validation See respondent validation.
garding areas to be covered in unstructured and
semi-structured interviewing. missing data Data that are not available, for exam-
ple, when a respondent in social survey research does
interview schedule A collection of questions de-
not answer a question. These are referred to as “missing
signed to be asked by an interviewer; always used in a
values” in IBM SPSS Statistics software.
structured interview.
mode The score that occurs most frequently in a set
intra-coder reliability The degree to which an indi- of scores.
vidual coder is consistent over time in the coding of an
item; likely to be an issue in the coding of answers to moderated relationship A relationship between two
open questions in research based on questionnaires variables is said to be “moderated” when the effect of
or structured interviews. the independent variable varies at different levels of a
second independent variable (also known as a statis-
Kendall’s tau-b A test of the strength of the relation- tical interaction).
ship between two ordinal variables.
moderator The person who guides the questioning
key informant Someone who offers the researcher, of a focus group, also called a facilitator.
usually an ethnographer, particularly perceptive in-
multiple-indicator measure A measure that uses
formation about the social setting, important events,
more than one indicator.
and individuals.
multi-strategy research Research that combines
life history interview An unstructured interview that
quantitative and qualitative approaches.
is similar to an oral history interview, but designed
to gather information on the entire biography of a multivariate analysis The examination of relation-
respondent. ships among three or more variables.

life history method A method (often referred to as narrative analysis An approach focused on the
the biographical method) that emphasizes the inner search for and analysis of stories that people use to
experience of individuals and its connections with understand their lives and the world around them.
larger societal events throughout the life course. It naturalism, naturalistic A style of research d
­ esigned
usually entails life history interviews and the use of to minimize disturbance to the natural or everyday
personal documents as data. social world.
Likert item, Likert scale A widely used format in negative relationship A relationship between two
which respondents are typically asked their degree variables in which one decreases as the other increases.

bel65796_glos_404-411 407 12/16/21 05:53 PM

408 Glossary

nominal variable A variable comprised of catego- oral history interview A largely unstructured inter-
ries that cannot be ranked; also known as a categor- view in which respondents are asked to recall and re-
ical variable. flect on events they have experienced.
nomothetic An approach to explanation that utilizes ordinal variable A variable whose categories can be
general laws and principles, which are said to apply to rank-ordered, but the distances between the catego-
some population beyond the people studied. ries are not equal or known across the range.
non-probability sample A sample selected using a outlier An extreme value (either very high or very low)
non-random sampling method. Essentially, this means in a distribution of scores. If a variable has an outlier,
that some units in the population are more likely than it will distort both the arithmetic mean and the range.
others to be selected.
paradigm A cluster of beliefs and assumptions,
non-response Occurs if someone in a sample refuses often unstated, that influence views on what should be
to participate in the study, cannot be contacted, or for studied, how research should be done, and how re-
some other reason does not supply the required data. sults should be interpreted.
null hypothesis A hypothesis of no relationship participant observation Research in which field-
between two variables; the hypothesis that the re- workers are immersed in a social setting for an ex-
searcher hopes to disprove. tended period of time, observing behaviour, asking
NVivo See QSR NVivo. questions, and analyzing what is said in conversations
both between the people under study and with the
objectivism, objectivist An ontological position ac- researcher. It usually includes interviewing key infor-
cording to which social phenomena have an existence mants and studying documents. In this book, “partic-
independent of social actors or their perceptions; ipant observation” refers to the observational aspect
compare with constructionism. of ethnography.
observation schedule A device used in structured participatory action research Research in which
observation that specifies the categories of behaviour local people affected by a particular social problem
that are to be observed and gives instructions on how collaborate as equals with professional researchers
behaviour should be allocated to those categories. and government officials to generate knowledge per-
official statistics Data compiled by (or on behalf tinent to the problem and to take action to ameliorate it.
of) state agencies in the course of conducting their
Pearson’s r A measure of the strength and direc-
business.
tion of the relationship between two interval/ratio
ontology, ontological A branch of philosophy con- variables.
cerned with the nature of reality: for example, whether
personal documents Documents not written for an
social entities can and should be considered objective
official purpose that provide first-person accounts of
entities with a reality external to specific social actors,
the writer’s life and events within it. Examples include
or as social constructions built up through the percep-
diaries, letters, and autobiographies.
tions and actions of these actors. See objectivism and
constructionism. population All cases or people covered by a theory
open, open-ended question In an interview sched- or explanation; the universe of units from which a
ule or questionnaire, a format that does not pres- sample is selected.
ent the respondent with a set of possible answers to positive relationship A relationship between two
choose from; compare with closed question. variables whereby both move in the same direction,
operationalization The way in which a concept is either increasing or decreasing simultaneously.
measured empirically; for example, one could opera- positivism, positivist An epistemological position
tionalize the concept “level of education” with a survey that advocates using the methods of the natural sci-
item that asks people to indicate their highest level of ences in the study of social reality.
education by choosing from categories such as “some
elementary,” “high school,” “university graduate,” etc.; postal questionnaire A questionnaire that is sent out
see also operational definition. and returned by regular (non-electronic) mail.
operational definition A definition that spells out postmodernism A position that questions the notion
the operations that are to be performed to measure of a pre-existing external reality that dispassionate
a concept. social scientists can simply uncover, and argues that

bel65796_glos_404-411 408 12/16/21 05:53 PM

 Glossary 409

no single account of social reality can be authoritative. random sampling The form of sampling in which all
Postmodernists prefer qualitative methods. units of a population have the same chance of being
selected.
pre-coded question Another name for a closed
question; used because a numerical code has been range The difference between the maximum and
pre-assigned to each of the predetermined answers the minimum score in a set of scores associated with
(hence there is no need for a coding frame). an interval or ratio variable.
probability sample A sample, selected at random, in ratio variable An interval variable with a true zero
which each unit in the population has a known chance point.
of being selected.
reactivity, reactive effect The effect on research
purposive sampling A form of non-probability sam- participants of knowing that they are being studied,
pling in which cases are selected on the basis of their which may result in atypical or inauthentic behaviour.
ability to provide information relevant to the topic of
realism, realist An epistemological position ac-
the study.
cording to which reality is independent of the senses
QSR NVivo A CAQDAS package that derives from but but is to some degree accessible to the researcher’s
goes beyond NUD*IST (Non-numerical Unstructured tools and theoretical speculations. See also critical
Data Indexing Searching and Theorizing). realism.
qualitative content analysis An approach to con- reflexive, reflexivity Terms used to refer to social
structing the meaning of documents and text that researchers’ awareness of the implications that their
allows categories to emerge out of data analysis and methods, values, biases, decisions, and mere pres-
recognizes the significance of the context in which ence in the situations they investigate have for the
items appear. knowledge they generate.
qualitative research Inquiry that uses mainly words, relationship An association between two variables
images, and other non-numerical symbols as data whereby the variation in one variable coincides with
and involves little or no quantification. As a research variation in another variable.
strategy it tends to be inductivist, constructivist, and in-
terpretivist, but qualitative researchers do not always reliability The degree to which a measure of a concept
subscribe to all three features; compare with quanti- is stable or consistent.
tative research. replication, replicability The degree to which a
quantitative research Inquiry using quantitative study can be repeated using the same methods.
data-gathering techniques and statistical analysis. As representative sample A sample that is similar to
a research strategy it tends to be deductivist and ob- the population in all important respects.
jectivist and to incorporate a natural science model of
the research process (in particular, one influenced by research design The term used in this book to refer
positivism), but quantitative researchers do not always to a framework for the collection and analysis of data.
subscribe to all three features; compare with qualita- The choice of research design reflects the goals that
tive research. the researcher hopes to achieve (for example, gener-
alization, establishing causality, or producing empa-
quasi-experiment A research design that resembles thetic understanding).
an experiment but does not meet all the requirements
and therefore does not exhibit complete internal respondent validation A process whereby research-
validity. ers provide the people on whom they conducted
research with an account of their findings and re-
questionnaire A collection of written questions or quest their feedback on it; sometimes called member
response items that the respondent completes without validation.
the aid of an interviewer.
response set A term for the tendency for some
quota sample A type of non-probability sample that people, when responding to multiple-indicator mea-
matches the proportions of people in different cate- sures, to respond to every item in the same way, sug-
gories in the population. gesting that their answers are motivated by something
random assignment The random allocation of re- other than their actual feelings. Three of the most
search participants to either the experimental or the common response-set effects are “acquiescence,”
control group(s) in an experiment. “social desirability,” and “laziness or boredom.”

bel65796_glos_404-411 409 12/16/21 05:53 PM

410 Glossary

sample The segment or subset of the population se- Spearman’s rho [ρ] A measure of the strength and
lected for research. The selection may be based on direction of the relationship between two ordinal
either probability or non-probability sampling. variables.
sampling error Differences between the charac- spurious relationship A relationship in which two
teristics of a random sample and the population from variables are correlated but not causally related. It is
which it is selected. produced by the impact of a common third variable
on each of the two variables. When the third variable
sampling frame A listing of units in the population
is controlled, the relationship disappears.
from which a sample is to be selected.
standard deviation A measure of how widely the
scale A term that can be used interchangeably with
scores that make up a set are dispersed around the
index when referring to a multiple-indicator measure
arithmetic mean.
of a variable in which the scores a research partici-
pant provides for each component indicator are com- standard error of the mean An estimate of the
bined to produce a composite score for that person. degree to which a sample mean is likely to differ from
A scale can also refer to a single-item measure of a the population mean.
variable.
statistical interaction Occurs if the effect of one
secondary analysis The analysis of data (quanti- independent variable varies at different levels of a
tative or qualitative) by researchers other than those second independent variable.
responsible for their collection, often for purposes that
statistical significance (test of) A test (such as the
the latter may not have envisaged.
chi-square test) that allows analysts to estimate the
semiotics An approach to the analysis of documents probability that the results of a study of a randomly
and other materials that emphasizes the importance selected sample are generalizable to the population
of signs and symbols, seeking out their deeper mean- from which the sample was drawn. Note that statis-
ing and exploring their intended effects. tical significance has nothing to do with substantive
significance or importance. Using a test of statistical
semi-structured interview A type of interview in
significance to generalize from a sample to a popula-
which the interviewer has a series of questions in the
tion is known as statistical inference.
general form of an interview guide and is able to vary
the sequence of questions. The questions are typically stratified random sample A sample in which units
more general than those specified by a structured are randomly sampled from a population that has
interview schedule, and the interviewer usually has been previously divided into sub-groups (strata).
some latitude to ask further questions in response to
structured interview One in which all respondents
replies that appear significant.
are asked exactly the same questions in the same
sensitizing concept A concept that is not fixed in ad- order with the aid of a formal interview schedule.
vance (as with an operational definition) but rather is
structured observation Often called systematic
treated as a guide that suggests what may be relevant
observation, this is a research method in which the
or important in an investigation.
researchers follow explicitly formulated rules regard-
sign A term used in semiotics. Each sign has two ing not only what they should look for, but when and
parts: the signifier (manifestation of a sign) and the where, and how they should record what they have
signified (the deeper meaning to which the signifier observed.
refers).
survey research A research design in which data are
simple random sample A sample in which each unit collected from respondents, mainly by questionnaire
selected from the population, and each combination or structured interview; typically, the data are then ex-
of units, has an equal probability of being included. amined to detect relationships among variables.
snowball sample A non-probability sample in which symbolic interactionism A theoretical perspective in
the researcher makes initial contact with a small which social interaction is understood to be based on
group of people connected to the research topic and the meanings that actors attach to their actions and
then uses them to establish contact with others. the contexts in which they occur.
social desirability bias Distortion of data caused by systematic sample A probability sampling method
respondents’ efforts to construct accounts that con- in which units are selected from a sampling frame at
form to a socially acceptable model of behaviour. fixed intervals (for example, every fifth unit).

bel65796_glos_404-411 410 12/16/21 05:53 PM

 Glossary 411

text A written document or, in more recent years, turn-taking In conversation analysis, the idea that
anything (from paintings to buildings to theme parks) taking turns to speak is a rule that helps to maintain
that may be “read” and analyzed for its symbolic order in everyday conversation.
value.
univariate analysis The analysis of one variable at
theoretical sampling A term used mainly in the a time.
context of grounded theory to refer to sampling
unobtrusive measures, unobtrusive methods, unob-
carried out in such a way that emerging theoretical
trusive observations Measures that do not cause
considerations guide the collection of data and/or
reactivity because they do not make research partici-
the selection of cases (usually research participants).
pants aware that they are being studied.
This process is supposed to continue until a point of
theoretical saturation is reached. unstructured interview An interview in which the in-
terviewer is free to explore any topic, although an in-
theoretical saturation In grounded theory, the point
terview guide is often used. The questioning is usually
where emerging concepts have been fully explored
informal and the content, phrasing, and sequencing of
and no new insights are being generated. See also
questions may vary from one interview to the next.
theoretical sampling.
validity A research criterion concerned with the in-
thick description A term coined by Geertz (1973; see
tegrity of the conclusions generated by a particular
Chapter 9) to refer to detailed accounts of a social set-
study. There are several types of validity: see, in par-
ting or people’s experiences that can form the basis
ticular, measurement validity, internal validity, and ex-
for general statements about a culture and its signifi-
ternal validity. When used on its own, validity is usually
cance in peoples’ lives.
taken to refer to measurement validity.
transcription, transcript The verbatim written
variable An attribute or characteristic that may vary
record of a taped interview or focus group session.
over time or from case to case. See also dependent vari-
triangulation The use of more than one method or able and independent variable; compare with constant.
source of data so that findings may be cross-checked.
vignette technique A method in which people are
trustworthiness A general criterion (composed of presented with “vignettes” (hypothetical scenarios)
four more specific criteria) used by some writers in as- and then asked how they would respond if they faced
sessing the quality of qualitative research. the circumstances depicted in the scenarios.

bel65796_glos_404-411 411 12/16/21 05:53 PM

References
Preface immigrants’ experience with the police. Race and Justice, 10
(3), 341–62. [Link]
World Bank. (2018). Nearly half the world lives on less
Foster, J. (1995). Informal social control and community crime
than $5.50 a day. [Link]
prevention. British Journal of Criminology, 35, 563–83.
press-release/2018/10/17/nearly-ha lf-t he-world-lives-
Frisby, W., Crawford, S., & Dorer, T. (1997). Reflections on par-
on-less-than-550-a-day
ticipatory action research: The case of low-income women
accessing local physical activity services. Journal of Sport
Chapter 1 Management, 11, 8–28.
Beagan, B. (2001). Micro inequities and everyday inequali- Giddens, A. (1984). The constitution of society. Cambridge, UK:
ties: “Race,” gender, sexuality and class in medical school. Polity.
Canadian Journal of Sociology, 26, 583–610. Goffman, E. (1963). Stigma: Notes on the management of spoiled
Bell, E. (2007). Separatism and quasi-separatism in Alberta. identity. Harmondsworth, UK: Penguin.
Prairie Forum, 32, 335–55. Hallgrimsdottir, H., Phillips, R., & Benoit, C. (2006). Fallen
Bikos, L. (2017, 17 May). It’s not just the RCMP: Police culture women and rescued girls: Social stigma and media narra-
is toxic. Globe and Mail. [Link] tives of the sex industry in Victoria, B.C., from 1980 to 2005.
com/opinion/its-not-just-the-rcmp-police-culture-is-toxic/ Canadian Review of Sociology and Anthropology, 43, 265–80.
article35014971/ Hay, D. (2014). An investigation into the swiftness and inten-
Blumer, H. (1962). Society as symbolic interaction. In A. Rose sity of recent secularization in Canada: Was Berger right?
(Ed.), Human Behavior and Social Processes. London: Sociology of Religion, 75, 136–62.
Routledge & Kegan Paul. Hier, S. (2002). Raves, risks and the ecstasy panic: A case study in
Charpentier, M., & Quéniart, A. (2017). Aging experiences of the subversive nature of moral regulation. Canadian Journal
older immigrant women in Québec (Canada): From deskill- of Sociology, 27, 33–57.
ing to liberation. Journal of Women & Aging, 29, 437–47. Hochschild, A. (1983). The managed heart. Berkeley and Los
Collins, R. (1994). Four sociological traditions (Rev. ed.). New Angeles: University of California Press.
York: Oxford University Press. Hughes, G. (2000). Understanding the politics of criminologi-
Creswell, J. (2013). Qualitative inquiry and research design: cal research. In V. Jupp, P. Davies, & P. Francis (Eds.), Doing
Choosing among five approaches (3rd ed.). Los Angeles: SAGE. criminological research. London: SAGE.
Debbink, G., & Ornelas, A. (1997). Cows for Campesinos. In Kandler, C., Bell, E., & Riemann, R. (2016). The structure
S.E. Smith & D.G. Willms (Eds.), Nurtured by knowledge: and sources of Right-Wing Authoritarianism and Social
Learning to do participatory action research. New York and Dominance Orientation. European Journal of Personality, 30,
Ottawa: Apex Press and International Development Research 406–20.
Centre. Lantz, P., & Booth, K. (1998). The social construction of the
Demaiter, E.I., & Adams, T.L. (2009). “I really didn’t have any breast cancer epidemic. Social Sci. and Medicine, 46, 907–18.
problems with the male–female thing until . . . ”: Successful Layder, D. (1993). New strategies in social research. Cambridge,
women’s experiences in IT organizations. Canadian Journal UK: Polity.
of Sociology, 43, 31–53. Lewin, K. (1946). Action research and minority problems.
Dinovitzer, R., Hagan, J., & Parker, P. (2003). Choice and circum- Journal of Social Issues, 2, 34–46.
stance: Social capital and planful competence in the attain- Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide
ments of immigrant youth. Canadian Journal of Sociology, 28, to qualitative observation and analysis (3rd ed.). Belmont,
463–88. CA: Wadsworth.
Duncan, L. (2010). Money and membership: Effects of neigh- Lyon, K., & Frohard-Dourlent, H. (2015). Let’s talk about the in-
bourhood poverty, income inequality and individual income stitution: Same-sex common-law partners negotiating mar-
on voluntary association membership in Canada. Canadian riage equality and relationship legitimacy. Canadian Review
Journal of Sociology, 35, 573–93. of Sociology, 52, 402–28.
Durkheim, E. (1938). The rules of sociological method. S. Solavay Maiolino, E. (2015). Political pugilists: Recuperative gender
& J. Mueller (Trans.). New York: Free Press. strategies in Canadian electoral politics. Canadian Review of
Durkheim, E. (1952). Suicide: A study in sociology. J. Spaulding Sociology, 52, 115–33.
& G. Simpson (Trans.). London: Routledge & Kegan Paul. Marx, K. (1998). The German ideology including theses on
(Original work published 1897) Feuerbach and Introduction to the critique of political econ-
Ellis, B.H., Lincoln, A.K., Abdi, S.M., Nimmons, E.A., Issa, O., omy. Amherst, NY: Prometheus Books. (Original work pub-
& Decker, S.H. (2020). “We all have stories”: Black Muslim lished 1845)

bel65796_ref_412-429 412 12/11/21 05:21 PM

 References 413

McKeganey, N., & Barnard, M. (1996). Sex work on the streets. households for both sexes, total, presence and age of children,
Buckingham, UK: Open University Press. 2016 counts, Canada, provinces and territories, 2016 Census—
Merton, R. (1938). Social structure and anomie. American 100% data. Families, Households and Marital Status Highlight
Sociological Review, 3, 672–82. Tables. [Link]
Merton, R. (1967). On theoretical sociology. New York: Free Press. dp-pd/hlt-fst/fam/[Link]?Lang=E&T=11&Geo=00&SP=
Mies, M. (1993). Towards a methodology for feminist research. 1&view=1&sex=1&presence=1
In M. Hammersley (Ed.), Social research: Philosophy, politics Strauss, A., Schatzman, L., Ehrich, D., Bucher, R., & Sabshin, M.
and practice. London: SAGE. (1973). The hospital and its negotiated order. In G. Salaman
Milgram, S. (1963). A behavioural study of obedience. Journal of & K. Thompson (Eds.), People and organizations. London:
Abnormal and Social Psychology, 67, 371–78. Longman.
Morgan, R. (2000). The politics of criminological research. In R. Sugiman, P. (2004). Memories of the internment: Narrating
King & E. Wincup (Eds.), Doing research on crime and justice. Japanese-Canadian women’s life stories. Canadian Journal of
Oxford: Oxford University Press. Sociology, 29, 359–88.
Nagra, B., & Maurutto, P. (2016). Crossing borders and manag- Tastsoglou, E., & Miedema, B. (2003). Immigrant women and
ing racialized identities: Experiences of security and surveil- community development in the Canadian Maritimes:
lance among young Canadian Muslims. Canadian Journal of Outsiders within? Canadian Journal of Sociology, 28, 203–34.
Sociology, 41, 165–94. Teevan, J., & Dryburgh, H. (2000). First person accounts and so-
Neves, B.B., Waycott, J., & Malta, S. (2018). Old and afraid of new ciological explanations of delinquency. Canadian Review of
communication technologies? Reconceptualizing and con- Sociology and Anthropology, 37, 77–93.
testing the “age-based digital divide.” Journal of Sociology, 54, Turnbull, P. (1973). The mountain people. London: Cape.
236–48. Weber, M. (1947). The theory of social and economic organization.
Nietzsche, F. (1968). The will to power. W. Kaufmann & R. J. A.M. Henderson & T. Parsons (trans.). New York: Free Press.
Hollingdale (Trans.). New York: Vintage Books. (Original Wilson, B. (2002). The Canadian rave scene and five theses on
work published 1883) youth resistance. Canadian Journal of Sociology, 27, 373–412.
Neuman, W.L. (2003). Social research methods. Toronto: Pearson Wolcott, H. (2008). Ethnography: A way of seeing (2nd ed.).
Canada. Toronto: AltaMira Press.
Orwell, G. (1933). Down and out in Paris and London. Toronto:
Harcourt. Chapter 2
Peltier, C. (2018). An application of two-eyed seeing: Indigenous
research methods with participatory action research. Alford, J.R., Funk, C.L., & Hibbing, J.R. (2005). Are political ori-
International Journal of Qualitative Methods, 17, 1–12. entations genetically transmitted? American Political Science
Pew Research Center. (2019). 5 facts about religion in Canada. Review, 99, 153–67.
[Link] Baer, D., Curtis, J., & Grabb, E. (2001). Has voluntary association
about-religion-in-canada/ membership declined? Cross-national analyses for fifteen
Potter, J. (1996). Representing reality: Discourse, rhetoric and countries. Canadian Review of Sociology, 38, 249–74.
social construction. London: SAGE. Bahr, H., Caplow, T., & Chadwick, B. (1983). Middletown III:
Pratt, A., & Valverde, M. (2002). From deserving victims to Problems of replication, longitudinal measurement, and tri-
“masters of confusion”: Redefining refugees in the 1900s. angulation. Annual Review of Sociology, 9, 243–64.
Canadian Journal of Sociology, 27, 135–62. Beardsworth, A., & Keil, T. (1992). The vegetarian option:
Ring, K. (1967). Experimental social psychology: Some Varieties, conversions, motives and careers. Sociological
sober questions about some frivolous values. Journal of Review, 40, 253–93.
Experimental Social Psychology, 3, 113–23. Bell, E., Aitken Schermer, J., & Vernon, P.A. (2009). The origins
Schutz, A. (1962). Collected papers I: The problem of social reality. of political attitudes and behaviours: An analysis using twins.
The Hague: Martinus Nijhof. Canadian Journal of Political Science, 42, 855–79.
Smith, D. (2004). Women’s perspective as a radical critique Bell, E., Jansen, H., & Young, L. (2007). Sustaining a dynasty in
of sociology. In S. Harding (Ed.), The feminist standpoint Alberta: The 2004 provincial election. Canadian Political
theory reader: Intellectual and political controversies. New Science Review, 1, 27–49.
York: Routledge. Bell, E., & Kandler, C. (2017). The genetic and the sociological:
Smith, D. (2005). Institutional ethnography: A sociology for Exploring the possibility of consilience. Sociology, 51, 880–96.
people. Walnut Creek, CA: AltaMira Press. Bell, E., Kandler, C., & Riemann, R. (2018). Genetic and envi-
Smith, S.E. (1997). Deepening participatory-action research. In ronmental influences on socio-political attitudes: addressing
S.E. Smith & D.G. Willms (Eds.), Nurtured by knowledge: some gaps in the new paradigm. Politics and the Life Sciences,
Learning to do participatory-action research. New York and 37, 236–49.
Ottawa: Apex Press and International Development Research Berthoud, R. (2000). Introduction: The dynamics of social
Centre. change. In R. Berthoud & J. Gershuny (Eds.), Seven years in
Statistics Canada. (2013). Cycle 27: Giving, volunteering and the lives of British families: Evidence on the dynamics of social
­participating. General Social Survey. change from the British Household Panel Survey. Bristol:
Statistics Canada. (2017). Marital status and opposite- and same- Policy Press.
sex status by sex for persons aged 15 and over living in private Blaxter, M. (1990). Health and lifestyles. London: Routledge.

bel65796_ref_412-429 413 12/11/21 05:21 PM

414 References

Bowen, R. (2015). Squaring up: Experiences of transition from Huey, L. (2003). Explaining Odlin Road: Insecurity and exclusiv-
off-street sex work to square work and duality—concurrent ity. Canadian Journal of Sociology, 28, 367–86.
involvement in both—in Vancouver, BC. Canadian Review of Hughes, E. (1943). French Canada in transition. Chicago:
Sociology, 52, 429–49. University of Chicago Press.
Brannigan, A. (2004). The rise and fall of social psychology: The Kerr, D. (2004). Family transformations and the well-being of
use and misuse of the experimental method. Hawthorne, NY: children: Recent evidence from Canadian longitudinal data.
Aldine de Gruyter. Journal of Comparative Family Studies, 35, 73–90.
Canadian Government News. (2020, June 8). Will the pandemic Kerr, D., & Michalski, J. (2007). Family structure and children’s
change the world of work? U of T researcher Scott Schieman hyperactivity problems: A longitudinal analysis. Canadian
aims to find out. Journal of Sociology, 32, 85–112.
Carroll, L. (2009). Alice’s adventures in Wonderland. Toronto: Knopik, V.S., Neiderhiser, J.M., DeFries, J.C., & Plomin, R. (2017).
Penguin Group. (Original work published 1865) Behavioural genetics (7th ed.). New York: Worth.
Charlton, T., Coles, D., Panting, C., & Hannan, A. (1999). Knott, C. (2016). Contentious mobilities and cheap(er) labour:
Behaviour of nursery class children before and after the avail- Temporary foreign workers in a New Brunswick seafood pro-
ability of broadcast television: A naturalistic study of two cessing community. Canadian Journal of Sociology, 41, 375–97.
­cohorts in a remote community. Journal of Social Behavior Laplante, B. (2006). The rise of cohabitation in Quebec: Power
and Personality, 14, 315–24. of religion and power over religion. Canadian Journal of
Charlton, T., Gunter, B., & Coles, D. (1998). Broadcast television Sociology, 31, 1–24.
as a cause of aggression? Recent findings from a naturalistic Lewis, O. (1961). The children of Sánchez. New York: Vintage.
study. Emotional and Behavioural Difficulties, 3, 5–13. Lipset, S. (1990). Continental divide: The values and institutions of
Chowdhury, F.D. (2018). Immigration through marriage: A case the United States and Canada. New York: Routledge.
study of Bangladeshis in Toronto. Journal of Muslim Minority Lynd, R., & Lynd, H. (1929). Middletown: A study in contempo-
Affairs, 38, 280–300. rary American culture. New York: Harcourt Brace.
Clairborn, W. (1969). Expectancy effects in the classroom: A Lynd, R., & Lynd, H. (1937). Middletown in transition: A study in
failure to replicate. Journal of Educational Psychology, 60, cultural conflicts. New York: Harcourt Brace.
377–83. Mead, M. (1928). Coming of age in Samoa. New York: Morrow.
Cook, T., & Campbell, D. (1979). Quasi-experimentation: Design Menard, S. (1991). Longitudinal research. Newbury Park, CA:
and analysis for field settings. Boston: Houghton Mifflin. SAGE.
Demers, A. (1996). Effect of support groups on family caregivers Michaud, S. (2001). The National Longitudinal Study of Children
to the frail elderly. Canadian Journal on Aging, 15, 129–44. and Youth—Overview and changes after three cycles.
Dinovitzer, R., Hagan, J., & Parker, P. (2003). Choice and circum- Canadian Studies in Population, 28, 391–405.
stance: Social capital and planful competence in the attain- Nemni, M., & Nemni, M. (2006). Young Trudeau: Son of Quebec,
ments of immigrant youth. Canadian Journal of Sociology, 28, father of Canada, 1919–1944. W. Johnson (Trans.). Toronto:
463–88. Douglas Gibson Books.
Dobash, R.E., & Dobash, R.P. (2000). Evaluating criminal justice Pawson, R., & Tilley, N. (1997). Realistic evaluation. London:
interventions for domestic violence. Crime & Delinquency, SAGE.
46, 252–70. Phinney, A., Moody, E., & Small, J. (2014). The effect of a
Dyer, W., & Wilkins, A. (1991). Better stories, not better con- ­community-engaged arts program on older adults’ well-
structs, to generate better theory: A rejoinder to Eisenhardt. being. Canadian Journal on Aging, 33, 336–45.
Academy of Management Review, 16, 613–19. Pratt, A., & Valverde, M. (2002). From deserving victims to
Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy “masters of confusion”: Redefining refugees in the 1900s.
fails. New York: Harper & Row. Canadian Journal of Sociology, 27, 135–62.
Fisher, R., & Ma, Y. (2014). The price of being beautiful: Negative Radley, A., & Chamberlain, K. (2001). Health psychology and the
effects of attractiveness on empathy for children in need. study of the case: From method to analytic concern. Social
Journal of Consumer Research, 41. doi:10.1086/676967 Science and Medicine, 53, 321–32.
Foran, C. (2010). Mordecai: The life and times. Toronto: Knopf Rinehart, J. (1996). The tyranny of work (3rd ed.). Toronto:
Canada. Harcourt Brace.
Goyder, J., Guppy, N., & Thompson, M. (2003). The allocation of Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom:
male and female occupational prestige in an Ontario urban Teacher expectation and pupils’ intellectual development.
area: A quarter-century replication. Canadian Review of New York: Holt, Rinehart & Winston.
Sociology and Anthropology, 40, 417–39. Rosnow, R.L., & Rosenthal, R. (1997). People studying people:
Greene, J. (2000). Understanding social programs through Artefacts and ethics in behavioral research. New York: W.H.
evaluation. In N. Denzin & Y. Lincoln (Eds.), Handbook of Freeman.
Qualitative Research (2nd ed.). Thousand Oaks, CA: SAGE. Russell, R., & Tyler, M. (2002). Thank heaven for little girls: “Girl
Hanson, R.K., Broom, I., & Stephenson, M. (2004). Evaluating Heaven” and the commercial context of feminine childhood.
community sex offender treatment programs: A 12-year Sociology, 36, 619–37.
­follow-up of 724 offenders. Canadian Journal of Sociology, 36, Sanderson, S.K. (2001). The evolution of human sociality: A
87–96. Darwinian conflict perspective. Lanham, MD: Rowman &
Houghton, E. (1998). Sex is good for you. Guardian, 92, 14–15. Littlefield.

bel65796_ref_412-429 414 12/11/21 05:21 PM

 References 415

Scriven, M. (1991). Evaluation thesaurus (4th ed.). Newbury Park, Canadian Institutes of Health Research, Natural Sciences and
CA: SAGE. Engineering Research Council of Canada, and Social Sciences
Seabrook, J., & Avison, W. (2015). Family structure and children’s and Humanities Research Council (2018). Tri-council policy
socioeconomic attainment: A Canadian sample. Canadian statement: Ethical conduct for research involving humans. Her
Review of Sociology, 52, 66–88. Majesty the Queen in Right of Canada.
Shalla, V. (2002). Jettisoned by design? The truncated employ- Canadian Institutes of Health Research, Natural Sciences
ment relationship of customer sales and service agents under and Engineering Research Council of Canada, and Social
airline restructuring. Canadian Journal of Sociology, 27, 1–32. Sciences and Humanities Research Council of Canada (1998,
Shepherd, R.P., & Graham, K.A.H. (2020). Evaluation in with 2000, 2002, 2005 amendments). Tri-council policy state-
Indigenous contexts: An introduction to practice. Canadian ment: Ethical conduct for research involving humans. Public
Journal of Program Evaluation, 34.3, 391–99. Works and Government Services Canada.
Smith, N., Lister, R., & Middleton, S. (2004). Longitudinal Castonguay, P. (2017, June 7). Quebec ruling reaffirms ­research
qualitative research. In S. Becker & A. Bryman (Eds.), data confidentiality. University Affairs. [Link]
Understanding research for social policy and practice: Themes, [Link]/news/news-article/quebec-ruling-
methods, and approaches. Bristol: Policy Press. reaffirms-research-data-confidentiality/
Stake, R.E. (1995). The art of case study research. Thousand Oaks, Charbonneau, L. (2005, June 6). Ethics boards harming survey
CA: SAGE. research, says York professor. University Affairs. [Link]
Statistics Canada. (2010). Detailed information for 2008–2009 university [Link]/ethics-boards-harming-survey-research-
(Cycle 8). National Longitudinal Survey of Children and [Link]
Youth (NLSCY). [Link] Desroches, F. (1990). Tearoom trade: A research update.
pl?Function=getSurvey&SDDS=4450 Qualitative Sociology, 13, 39–61.
Tilley, N. (2000). Doing realistic evaluation of criminal justice. Le Devoir. (2016, November 3). L’affaire Maillé, ou l’avenir de la
In V. Jupp, P. Davies, & P. Francis (Eds.), Doing criminological confidentialité dans la recherche scientifique. [Link]
research. London: SAGE. [Link]/opinion/libre-opinion/483756/l-affaire-maille-
Truth and Reconciliation Commission of Canada. (2015). ou-l-avenir-de-la-confidentialite-dans-la-recherche-scientifique
Honouring the truth, reconciling for the future: Summary of Dinovitzer, R., Hagan, J., & Parker, P. (2003). Choice and circum-
the final report of the Truth and Reconciliation Commission of stance: Social capital and planful competence in the attain-
Canada. [Link] ments of immigrant youth. Canadian Journal of Sociology, 28,
Walsh, M., Hickey, C., & Duffy, J. (1999). Influence of item con- 463–88.
tent and stereotype situation on gender differences in mathe- Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy
matical problem solving. Sex Roles, 41, 219–40. fails. New York: Harper & Row.
Walters, D. (2004). A comparison of the labour market outcomes Gans, H.J. (1962). The urban villagers. New York: Free Press.
of postsecondary graduates of various levels and fields over a Goode, E. (1996). The ethics of deception in social research:
four-cohort period. Canadian Journal of Sociology, 29, 1–27. A case study. Qualitative Sociology, 19, 11–33.
White, J. (1990). Hospital strike. Toronto: Thompson. Haney, C., Banks, C., & Zimbardo, P. (1973, September). A study
Wilson, B. (2002). The Canadian rave scene and five theses on of prisoners and guards in a simulated prison. Naval Research
youth resistance. Canadian Journal of Sociology, 27, 373–4412. Reviews, 1–17.
Yin, R. (1984). Case study research: Design and methods. Beverly Homan, R. (1991). The ethics of social research. London: Longman.
Hills, CA: SAGE. Homan, R., & Bulmer, M. (1982). On the merits of covert meth-
Young, N., & Dugas, E. (2012). Comparing climate change cover- ods: A dialogue. In M. Bulmer (Ed.), Social research ethics.
age in Canadian English- and French-language print media: London: Macmillan.
Environmental values, media cultures, and the narration of Humphreys, L. (1970). Tearoom trade: Impersonal sex in public
global warming. Canadian Journal of Sociology, 37, 25–54. places. Chicago: Aldine.
Kimmel, A. (1988). Ethics and values in applied social research.
Newbury Park, CA: SAGE.
Chapter 3
Kondro, W. (2016, November 22). Canadian researcher in legal
Aronson, E. (1992). The social animal (6th ed.). New York: W.H. battle to keep her interviews confidential. [Link]. http://
Freeman and Company. [Link]/news/2016/11/canadian-researcher-
Bailey, K.D. (1994). Methods of social research (4th ed.). Toronto: legal-battle-keep-her-interviews-confidential
Maxwell Macmillan Canada. Le Texier, T. (2019). Debunking the Standford prison ­experiment.
Beagan, B. (2001). Micro inequities and everyday inequali- American Psychologist, 74, 823–39.
ties: “Race,” gender, sexuality and class in medical school. Milgram, S. (1963). A behavioural study of obedience. Journal of
Canadian Journal of Sociology, 26, 583–610. Abnormal and Social Psychology, 67, 371–78.
Blair, G., Imai, K., & Zhou, Y.Y. (2015). Design and analysis of Mosby, I. (2013). Administering colonial science: Nutrition re-
the randomized response technique. Journal of the American search and human biomedical experimentation in Aboriginal
Statistical Association, 110, 1304–19. communities and residential schools, 1942–1952. Histoire
Blum, B. (2018, June 7). The lifespan of a lie: The most famous ­sociale/Social History, 46, 145–72.
psychology study of all time was a sham. Why can’t we escape O’Connell Davidson, J., & Layder, D. (1994). Methods, sex, and
the Stanford Prison Experiment? Medium GEN. madness. London: Routledge.

bel65796_ref_412-429 415 12/11/21 05:21 PM

416 References

Perry, G. (2012). Behind the shock machine: The untold story of Brannigan, A. (2004). The rise and fall of social psychology: The
the notorious Milgram psychology experiments. New York: use and misuse of the experimental method. Hawthorne, NY:
New Press. Aldine de Gruyter.
Punch, M. (1994). Politics and ethics in qualitative research. Bridgman, P. (1927). The logic of modern physics. New York:
In N. Denzin & Y. Lincoln (Eds.), Handbook of Qualitative Macmillan.
Research. Thousand Oaks, CA: SAGE. Bryman, A., & Cramer, D. (2001). Quantitative data analysis
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom: with SPSS release 10 for Windows: A guide for social scientists.
Teacher expectation and pupils’ intellectual development. London: Routledge.
New York: Holt, Rinehart & Winston. Bulmer, M. (1984). Facts, concepts, theories and problems. In
Small, W., Maher, L., & Kerr, T. (2014). Institutional ethical review M. Bulmer (Ed.), Social research methods. London: Macmillan.
and ethnographic research involving injection drug users: A Cicourel, A. (1964). Method and measurement in sociology.
case study. Social Science & Medicine, 104, 157–62. New York: Free Press.
Totten, M. (2001). Legal, ethical, and clinical implications of Cicourel, A. (1982). Interviews, surveys, and the problem of eco-
doing fieldwork with youth gang members who engage in se- logical validity. American Sociologist, 17, 11–20.
rious violence. Journal of Gang Research, 8, 35–49. Cramer, D. (1998). Fundamental statistics for social research.
van den Hoonaard, W.C. (2001). Is research-ethics review a moral London: Routledge.
panic? Canadian Review of Sociology and Anthropology, 38, 19–36. Deng, J., Walker, G., & Swinnerton, G. (2006). A comparison
van den Hoonaard, W.C., & Connolly, A. (2006). Anthropological of environmental values and attitudes between Chinese in
research in light of research-ethics review: Canadian master’s Canada and Anglo-Canadians. Environment and Behavior,
theses, 1995–2004. Journal of Empirical Research on Human 38, 22–47.
Research Ethics, 1, 59–69. Edin, K., & Kefalas, M. (2005). Promises I can keep: Why poor
van den Hoonaard, W.C., & Tolich, M. (2014). The New Brunswick women put motherhood before marriage. Berkeley: University
Declaration of Research Ethics: A simple and radical perspec- of California Press.
tive. Canadian Journal of Sociology, 39, 87–97. Foddy, W. (1993). Constructing questions for interviews and ques-
Van Maanen, J. (1991). The smile factory: Work at Disneyland. tionnaires: Theory and practice in social research. Cambridge:
In P. Frost, L. Moore, et al. (Eds.), Reframing organizational Cambridge University Press.
culture. Newbury Park, CA: Sage. Frazer, R., & Wiersma, U. (2001). Prejudice vs. discrimination
Vidich, A., & Bensman, J. (1968). Small town in mass society. in the employment interview: We may hire equally, but our
Princeton, NJ: Princeton University Press. memories harbour prejudice. Human Relations, 54, 173–91.
Von Hoffman, N. (1970, January 30). The sociological snoopers. Gabor, T., Hung, K., Mihorean, S., & St-Onge, C. (2002).
Washington Post. Canadian homicide rates: A comparison of two data sets.
Zimbardo, P. (1971). The psychological power and pathology of Canadian Journal of Criminology, 44, 351–63.
imprisonment. Statement prepared for the U.S. House of Gazso-Windlej, A., & McMullin, J. (2003). Doing domestic
Representatives Committee on the Judiciary, Subcommittee labour: Strategizing in a gendered domain. Canadian Journal
No. 3: Hearings on Prison Reform, San Francisco. of Sociology, 28, 341–66.
George, U., & Chaze, F. (2014). Discrimination at work: Comparing
the experiences of foreign-trained and locally-trained engi-
Appendix to Part I neers in Canada. Canadian Ethic Studies, 46, 1–21.
Orwell, G. (2004). Why I write. London: Penguin. (Original work Goyder, J., Guppy, N., & Thompson, M. (2003). The allocation of
published 1946) male and female occupational prestige in an Ontario urban
area: A quarter-century replication. Canadian Review of
Sociology and Anthropology, 40, 417–39.
Chapter 4
Guppy, L., & Siltanen, J. (1977). A comparison of the allocation of
Beardsworth, A., Bryman, A., Ford, J., & Keil, T. (n.d.). “The dark male and female occupational prestige. Canadian Review of
figure” in statistics of unemployment and vacancies: Some Sociology and Anthropology, 14, 320–30.
sociological implications. Discussion paper, Department of Hay, D. (2014). An investigation into the swiftness and inten-
Social Sciences, Loughborough University. sity of recent secularization in Canada: Was Berger right?
Bell, E., Aitken Schermer, J., & Vernon, P.A. (2009). The origins Sociology of Religion, 75, 136–62.
of political attitudes and behaviours: An analysis using twins. Kennedy, E., Krogman, N., & Krahn, H. (2013). Sustainable
Canadian Journal of Political Science, 42, 855–79. consumption and the importance of neighbourhood:
­
Berthoud, R. (2000). A measure of changing health. In R. A ­ central city/suburb comparison. Canadian Journal of
Berthoud & J. Gershuny (Eds.), Seven years in the lives of Sociology, 38, 359–83.
British families: Evidence on the dynamics of social change LaPiere, R.T. (1934). Attitudes vs. actions. Social Forces, 13,
from the British Household Panel Survey. Bristol: Policy Press. 230–37.
Blumer, H. (1956). Sociological analysis and the “variable.” Logie, C.H, Lys, C.L., Okumu, M. & Fujioka, J. (2019). Exploring
American Sociological Review, 21, 683–90. factors associated with condom use self-efficacy and condom
Bourdieu, P. (1987). The historical genesis of a pure aesthetic. use among Northern and Indigenous adolescent peer leaders
Journal of Aesthetics and Art Criticism, 46 (Analytic in Northern Canada. Vulnerable Children and Youth Studies,
Aesthetics), 201–10. 14, 50–62.

bel65796_ref_412-429 416 12/11/21 05:21 PM

 References 417

Murphy, L., & Fedoroff, J. (2013). Sexual offenders’ views of Bampton, R., & Cowton, C.J. (2002). The e-interview. Forum:
Canadian sex offender registries: A survey of a clinical Qualitative Social Research, 3. [Link]
sample. Canadian Journal of Behavioural Science, 45, 238–49. [Link]/[Link]/fqs/article/view/848
Reiner, R. (2000). Crime and control in Britain. Sociology, 34, Beagan, B. (2001). Micro inequities and everyday inequalities:
71–94. “Race,” gender, sexuality and class in medical school. Canadian
Ricciardelli, R., & Clow, K. (2009). Men, appearance, and cos- Journal of Sociology, 26, 583–610.
metic surgery: The role of self-esteem and comfort with the Beardsworth, A., & Keil, T. (1997). Sociology on the menu: An in-
body. Canadian Journal of Sociology, 34, 105–34. vitation to the study of food and society. London: Routledge.
Ritchie, S. (2020). Science fictions: How fraud, bias, negligence, Beardsworth, A., Keil, T., Goode, J., Haslam, C., & Lancashire, E.
and hype undermine the search for truth. New York: Henry (2002). Women, men and food: The significance of gender for
Holt and Company. nutritional attitudes and choices. British Food Journal, 104,
Saba, R. (2020, July 22). New study confirms COVID-19 rates and 470–91.
deaths far more prevalent at for-profit long-term care homes Bell, E., Aitken Schermer, J., & Vernon, P.A. (2009). The origins
than non-profits. Toronto Star. [Link] of political attitudes and behaviours: An analysis using twins.
business/2020/07/22/new-study-confirms-covid-19-cases- Canadian Journal of Political Science, 42, 855–79.
and-deaths-far-more-prevalent-at-for-profit-long-term-care- Bell, E., & Kandler, C. (2017). The genetic and the sociological:
[Link] Exploring the possibility of consilience. Sociology, 51, 880–96.
Schieman, S., & Narisada, A. (2014). In control or fatalistically Beran, T., Hughes, G., & Lupart, J. (2008). A model of achieve-
ruled? The sense of mastery among working Canadians. ment and bullying: Analyses of the Canadian National
Canadian Review of Sociology, 51, 343–74. Longitudinal Survey of Children and Youth data. Educational
Schuman, H., & Presser, S. (1981). Questions and answers in at- Research, 50, 25–39.
titude surveys: Experiments on question form, wording, and Blauner, R. (1964). Alienation and freedom: The factory worker
context. San Diego, CA: Academic Press. and his industry. Chicago: University of Chicago Press.
Silverman, R., Sacco, V., & Teevan, J. (2000). Crime in Canadian Boivin, R., & Leclerc, C. (2016). Domestic violence reported to
society. Toronto: Harcourt Canada. the police: Correlates of victims’ reporting behavior and sup-
Sobel, N. (2015). A typology of the changing narratives of port to legal proceedings. Violence and Victims, 31, 402–15.
Canadian citizens through time. Canadian Ethnic Studies, Bottomore, T.B., & Rubel, M. (1963). Karl Marx: Selected writ-
47, 11–39. ings in sociology and social philosophy. Harmondsworth, UK:
Stall, N.M., Jones, A., Brown, K.A., Rochon, P.A., & Costa, A.P. Penguin.
(2020). For-profit long-term care homes and the risk of Brannigan, A., Gemmell, W., Pevalin, D., & Wade, T. (2002).
COVID-19 outbreaks and resident deaths. Canadian Medical Self-control and social control in childhood misconduct and
Association Journal, early-released July 22. [Link] aggression: The role of family structure and hyperactivity.
org/10.1503/cmaj.201197 Canadian Journal of Criminology, 44, 119–42.
Sutton, R.I., & Rafaeli, A. (1992). How we untangled the relation- Cobanoglu, C., Ward, B., & Moreo, P.J. (2001). A comparison
ship between displayed emotion and organizational sales: of mail, fax and web-based survey methods. International
A tale of bickering and optimism. In P. Frost & R. Stablein Journal of Market Research, 43, 441–52.
(Eds.), Doing exemplary research. Newbury Park, CA: SAGE. Coleman, C., & Moynihan, J. (1996). Understanding crime data:
Teevan, J., & Dryburgh, H. (2000). First person accounts and so- Haunted by the dark figure. Buckingham: Open University
ciological explanations of delinquency. Canadian Review of Press.
Sociology and Anthropology, 37, 77–93. Comeau, J., Duncan, L., Georgiades, K., Wang, L., & Boyle,
M.H. (2020). Social assistance and trajectories of child
mental health problems in Canada: evidence from the
Chapter 5
National Longitudinal Survey of Children and Youth.
Ames, M., Rawana, J., Gentile, P., & Morgan, A. (2015). The pro- Canadian Journal of Public Health. [Link]
tective role of optimism and self-esteem on depressive symp- s41997-020-00299-1
tom pathways among Canadian Aboriginal youth. Journal of Corti, L. (1993). Using diaries in social research. Social Research
Youth and Adolescence, 44, 142–54. Update, 2.
Anderson, C., & McGregor, M. (2020). National identity, atten- Crook, C., & Light, P. (2002). Virtual society and the cul-
tiveness and the two sides of partisanship: the case of the 2015 tural practice of study. In S. Woolgar (Ed.), Virtual society?
Canadian federal election. National Identities, 22, 347–66 Technology, cyberbole, reality. Oxford: Oxford University
.doi:10.1080/14608944.2019.1634036 Press.
Anderson, K., Sebaldt, R., Lohfeld, L., Burgess, K., Donald, F., & Curasi, C. (2001) A critical exploration of face-to-face inter-
Kaczorowski, J. (2006). Views of family physicians in south- viewing vs. computer-mediated interviewing. International
western Ontario on preventive care services and performance Journal of Market Research, 43, 361–75.
incentives. Family Practice, 23, 469–71. Dale, A., Arber, S., & Proctor, M. (1988). Doing secondary analy-
Andrews, K., Smith, L., Henzi, D., & Demps, E. (2007). Faculty sis. London: Unwin Hyman.
and student perceptions of academic integrity at U.S. and Durkheim, E. (1952). Suicide: A study in sociology. J. Spaulding
Canadian dental schools. Journal of Dental Education, 71, & G. Simpson (Trans.). London: Routledge & Kegan Paul.
1027–39. (Original work published 1897)

bel65796_ref_412-429 417 12/11/21 05:21 PM

418 References

Elliott, H. (1997). The use of diaries in sociological research on Kevins, A., & Soroka, S. (2018). Growing apart? Partisan sorting
health experience. Sociological Research Online, 2. http:// in Canada, 1992–2015. Canadian Journal of Political Science,
[Link]/socresonline/2/2/[Link] 51, 103–33.
Finch, J. (1987). The vignette technique in survey research. Kingsbury, M., & Coplan, R. (2012). Mothers’ gender-role atti-
Sociology, 21, 105–14. tudes and their responses to young children’s hypotheti-
Fournier, P., Cutler, F., Soroka, S., & Stolle, D. (2015). The cal display of shy and aggressive behaviours. Sex Roles, 66,
2015 Canadian Election Study. [Link] 506–17.
english-section/surveys Krosnick, J., Holbrook, A., Berent, M., Carson, R.T., Hanemann,
Fowler, F. (1993). Survey research methods (2nd ed.). Newbury W.M., Kopp, R.J . . . Conaway (2002). The impact of “no opin-
Park, CA: SAGE. ion” response options on data quality: Non-attitude reduc-
Frey, J. (2004). Telephone surveys. In M. Lewis-Beck, A. Bryman, tion or an invitation to satisfice? Public Opinion Quarterly,
& T. Liao (Eds.), The SAGE. Encyclopedia of social science re- 66, 371–403.
search methods. Thousand Oaks, CA: SAGE. Laurenceau, J.-P., & Bolger, N. (2005). Using diary methods
Friedenreich, C.M, Barberio, A.M., Pader, J., Poirier, A.E., Ruan, to study marital and family processes. Journal of Family
Y., Grevers, X . . . Brenner, D.R., on behalf of the ComPARe Psychology, 19, 86–97.
Study Team. (2019). Estimates of the current and future Légaré, J., Décarie, Y., & Bélanger, A. (2014). Using microsimu-
burden of cancer attributable to lack of physical activity in lation to reassess aging trends in Canada. Canadian Journal
Canada. Preventative Medicine, 122, 65–72. on Aging, 33, 208–19.
Gidengil, E., Everitt, J., Blais, A., Fournier, P., & Nevitte, N. Li, P. (2003). Initial earnings and catch-up capacity of immi-
(2006). Gender and vote choice in the 2006 Canadian elec- grants. Canadian Public Policy, 29, 319–37.
tion. Paper prepared for the Annual Meeting of the American Mann, C., & Stewart, F. (2000). Internet communication and
Political Science Association, Philadelphia. qualitative research: A handbook for researching online.
Glavin, P., Bierman, A., & Schieman, S. (2019, November 23). London: SAGE.
Gig workers report feeling lonely and powerless. The Province. Marin, A., & Hayes, S. (2017). The occupational context of mis-
[Link] match: Measuring the link between occupations and areas
report-feeling-lonely-and-powerless of study to better understand job-worker match. Canadian
Goodyear-Grant, E., & Croskill, J. (2011). Gender affinity effects Journal of Sociology, 42, 1–22.
in vote choice in Westminster systems: Assessing “flexible” Mata, F., & Pendakur, R. (2017). Of intake and outcomes: Wage
voters in Canada. Politics and Gender, 7, 223–50. trajectories of immigrant classes in Canada. Journal of
Goyder, J., Guppy, N., & Thompson, M. (2003). The allocation of International Migration and Integration, 18, 829–44.
male and female occupational prestige in an Ontario urban Mayhew, P. (2000). Researching the state of crime. In R. King
area: A quarter-century replication. Canadian Review of & E. Wincup (Eds.), Doing research on crime and justice.
Sociology and Anthropology, 40, 417–39. Oxford: Oxford University Press.
Grabb, E., & Curtis, J. (2004). Regions apart: The four societies of O’Connor, H., & Madge, C. (2001). Cyber-mothers: Online
Canada and the U.S. Don Mills, ON: Oxford University Press. synchronous interviewing using conferencing software.
Hay, D. (2014). An investigation into the swiftness and inten- Sociological Research Online, 5. [Link]/5/4/
sity of recent secularization in Canada: Was Berger right? o’[Link].
Sociology of Religion, 75, 136–62. O’Connor, H., & Madge, C. (2003). “Focus groups in cyberspace”:
Hessler, R., Downing, J., Beltz, C., Pelliccio, A., Powell, M., & Vale, Using the Internet for qualitative research. Qualitative
W. (2003). Qualitative research on adolescent risk using e-mail: Market Research, 6, 133–43.
A methodological assessment. Qualitative Sociology, 26, 111–24. O’Sullivan, L., Udell, W., & Patel, V. (2006). Young urban adults’
Historica Canada. (2020, June 26). While half of Canadians know heterosexual risk encounters and perceived risk and safety: A
their country contributed to eradicating polio, most don’t structured diary study. Journal of Sex Research, 43, 343–51.
know the face on their $10 bill. [Link] Pahl, J. (1990). Household spending, personal spending and the
ca/content/polls/while-half-canadians-know-their-coun- control of money in marriage. Sociology, 24, 119–38.
try-contributed-eradicating-polio-most-dont-know Robinson, J., & Lee, C. (2014). Society’s (virtually) time-free
Holbrook, A., Green, M., & Krosnick, J. (2003). Telephone versus transition into the digital age. Social Indicators Research, 117,
face-to-face interviewing of national probability samples 939–65.
with long questionnaires: Comparisons of respondent satis- Robinson, J., Shaver, P., & Wrightsman, L. (1999). Measures of po-
ficing and social desirability response bias. Public Opinion litical attitudes. Toronto: Academic Press.
Quarterly, 67, 79–125. Schaeffer, D., & Dillman, D. (1998). Development of a standard
Kelley, J., & De Graaf, N. (1997). National context, parental so- e-mail methodology. Public Opinion Quarterly, 62, 378–97.
cialization, and religious belief: Results from 15 nations. Schuman, H., & Presser, S. (1981). Questions and answers in at-
American Sociological Review, 62, 639–59. titude surveys: Experiments on question form, wording, and
Kent, R., & Lee, M. (1999). Using the Internet for market re- context. San Diego, CA: Academic Press.
search: A study of private trading on the Internet. Journal of Shafer, K., & Renick, A.J. (2020). Depressive symptoms and father
the Market Research Society, 41, 377–85. involvement in Canada: Evidence from a national study.
Kerr, D. (2004). Family transformations and the well-being of Canadian Review of Sociology, 57, 197–222.
children: Recent evidence from Canadian longitudinal data. Sheehan, K. and Hoy, M. (1999). Using e-mail to survey Internet
Journal of Comparative Family Studies, 35, 73–90. users in the United States: Methodology and assessment.

bel65796_ref_412-429 418 12/11/21 05:21 PM

 References 419

Journal of Computer-Mediated Communication, 4. [Link] Normand, S., Schneider, B., Lee, M., Maisonneuve, M.,
.org/jcmc/vol4/issue3/[Link] Chupetlovska-Anastasova, A., Kuehn, S., & Robaey, P. (2013).
Shuy, R.W. (2002). In-person versus telephone interviewing. In Continuities and changes in the friendships of children with
J.F. Gubrium & J.A. Holstein (Eds.), Handbook of interview and without ADHD: A longitudinal, observational study.
research: Context and method. Thousand Oaks, CA: SAGE. Journal of Abnormal Child Psychology, 41, 1161–75.
Sinclair, M., O’Toole, J., Malawaraarachchi, M., & Leder, K. Perrucci, R., Belshaw, R., DeMerritt, A., Frazier, B., Jones, J.,
(2012). Comparison of response rates and cost-effectiveness Kimbrough, J., . . . Williams, B. (2000). The two faces of racial-
for a community-based survey: postal, Internet and tele- ized space at a predominantly white university. International
phone modes with generic or personalized recruitment ap- Journal of Contemporary Sociology, 37, 230–44.
proaches. BMC Medical Research Methodology, 12, 1–8. Roethlisberger, F., & Dickson, W. (1939). Management and the
Smith, D.J., & McVie, S. (2003). Theory and method in the worker. Cambridge, MA: Harvard University Press.
Edinburgh Study of Youth Transitions and Crime. British Rosenhan, D.L. (1973). On being sane in insane places. Science,
Journal of Criminology, 43, 169–95. 179, 350–58.
Statistics Canada. (2019). The General Social Survey: An over- Schulenberg, J.L. (2014). Systematic social observation of police
view, 2019. [Link] decision-making: The process, logistics, and challenges in a
/[Link] Canadian context. Quality and Quantity, 48, 297–315.
Sudman, S., & Bradburn, N. (1982). Asking questions: A practical Webb, E.J., Campbell, D.T., Schwartz, R.D., & Sechrest, L. (1966).
guide to questionnaire design. San Francisco: Jossey-Bass. Unobtrusive measures: Non-reactive measures in the social
Sullivan, O. (1996). Time co-ordination, the domestic division of sciences. Chicago: Rand McNally.
labour and affective relations: Time use and the enjoyment of Weber, B. (2020, March 31). Experts study how coronavirus
activities within couples. Sociology, 30, 79–100. pandemic affects trust in officials, ourselves. CBC News.
Tourangeau, R., & Smith, T.W. (1996). Asking sensitive ques- [Link]
tions: The impact of data collection mode, question format, coronavirus-public-trust-research-1.5515929
and question context. Public Opinion Quarterly, 60, 275–304. Weber, M. (1947). The theory of social and economic organization.
Tse, A. (1998). Comparing the response rate, response speed and A.M. Henderson & T. Parsons (Trans.). New York: Free Press.
response quality of two methods of sending questionnaires: Whyte, W.F. (1955). Street corner society (2nd ed.). Chicago:
E-mail vs. mail. Journal of the Market Research Society, 40, University of Chicago Press.
353–61.
Walklate, S. (2000). Researching victims. In R. King & E. Wincup Chapter 7
(Eds.), Doing research on crime and justice. Oxford: Oxford
University Press. Beagan, B. (2001). Micro inequities and everyday inequali-
Walters, D. (2004). A comparison of the labour market outcomes ties: “Race,” gender, sexuality and class in medical school.
of postsecondary graduates of various levels and fields over a Canadian Journal of Sociology, 26, 583–610.
four-cohort period. Canadian Journal of Sociology, 29, 1–27. Becker, H. (1963). Outsiders: Studies in the sociology of deviance.
Wright, K.B. (2005). Researching Internet-based populations: New York: Free Press.
Advantages and disadvantages of online survey research, online Bowen, R. (2015). Squaring up: Experiences of transition from
questionnaire authoring software packages, and web survey off-street sex work to square work and duality—concurrent
services. Journal of Computer-Mediated Communication, 10. involvement in both—in Vancouver, BC. Canadian Review of
[Link] Sociology, 52, 429–49.
Bryman, A. (1995). Disney and his worlds. London: Routledge.
Bryman, A. (1999). Global Disney. In P. Taylor & D. Slater (Eds.),
Chapter 6 The American century. Oxford: Blackwell.
Bales, R. (1951). Interaction process analysis. Cambridge, MA: Buchanan, C. (2014). A more national representation of place
Addison-Wesley. in Canadian daily newspapers. Canadian Geographer, 58,
Bell, E. (2007). Separatism and quasi-separatism in Alberta. 517–30.
Prairie Forum, 32, 335–55. Collins, R.E. (2014). “Beauty and bullets”: A content analysis of
Cahalan, S. (2019). The Great Pretender. New York and Boston: female offenders and victims in four Canadian newspapers.
Grand Central. Journal of Sociology, 52, 296–310.
Charlton, T., Gunter, B., & Coles, D. (1998). Broadcast television Czaja, R., & Blair, J. (1996). Designing surveys: A guide to decisions
as a cause of aggression? Recent findings from a naturalistic and procedures. Thousand Oaks, CA: SAGE.
study. Emotional and Behavioural Difficulties, 3, 5–13. Dinovitzer, R., Hagan, J., & Parker, P. (2003). Choice and circum-
Galton, M., Simon, B., & Croll, P. (1980). Inside the primary class- stance: Social capital and planful competence in the attain-
room. London: Routledge & Kegan Paul. ments of immigrant youth. Canadian Journal of Sociology, 28,
Katz, J., Kuffel, S., & Coblentz, A. (2002). Are there gender dif- 463–88.
ferences in sustaining dating violence? Journal of Family Dumitrica, D., & Bakardjieva, M. (2018). The personalization of
Violence, 17, 247–71. engagement: The symbolic construction of social media and
LaPiere, R.T. (1934). Attitudes vs. actions. Social Forces, 13, grassroots mobilization in Canadian newspapers. Media,
230–37. Culture & Society, 40, 817–37.
McCall, M.J. (1984). Structured field observation. Annual Review Finch, J., & Hayes, L. (1994). Inheritance, death and the concept
of Sociology, 10, 263–82. of the home. Sociology, 28, 417–33.

bel65796_ref_412-429 419 12/11/21 05:21 PM

420 References

Fowler, F. (1993). Survey research methods (2nd ed.). Newbury Cao, L. (2014). Aboriginal people and confidence in the police.
Park, CA: SAGE. Canadian Journal of Criminology and Criminal Justice, 56,
Frey, J. (2004). Telephone surveys. In M. Lewis-Beck, A. Bryman, 499–525.
& T. Liao (Eds.), The SAGE encyclopedia of social science re- Jones, A.M. (2020, April 11). Phone data reveals who is staying
search methods. Thousand Oaks, CA: SAGE. home during COVID-19. CTV News. [Link]
Johnson, T., & Wislar, J. (2012). Response rates and nonre- ca/health/coronavirus/phone-data-reveals-who-is-staying-
sponse errors in surveys. Journal of the American Medical home-during-covid-19-1.4892194
Association, 307, 1805–06. Noack, A.M. (2018). Social statistics in action: A Canadian intro-
Kennedy, E., Krogman, N, & Krahn, H. (2013). Sustainable con- duction. Don Mills: Oxford University Press.
sumption and the importance of neighbourhood: A central Rippeyoung, P. (2013). Can breastfeeding solve inequality? The
city/suburb comparison. Canadian Journal of Sociology, 38, relative mediating impact of breastfeeding and home envi-
359–83. ronment on poverty gaps in Canadian child cognitive skills.
MacKinnon, N., & Luke, A. (2002). Changes in identity attitudes Canadian Journal of Sociology, 38, 65–85.
as reflections of social and cultural change. Canadian Journal Rojek, C. (1995). Decentring leisure: Rethinking leisure theory.
of Sociology, 27, 299–338. London: SAGE.
Mairs, K., & Bullock, S. (2013). Sexual-risk behaviour and HIV Sutton, R.I., & Rafaeli, A. (1988). Untangling the relationship be-
testing among Canadian snowbirds who winter in Florida. tween displayed emotions and organizational sales: The case of
Canadian Journal on Aging, 32, 145–58. convenience stores. Academy of Management Journal, 31, 461–87.
Mangione, T.W. (1995). Mail surveys: Improving the quality.
Thousand Oaks, CA: SAGE. Chapter 9
Schieman, S., & Narisada, A. (2014). In control or fatalistically
ruled? The sense of mastery among working Canadians. Can. Abramovich, A. (2016). Preventing, reducing and ending
Review of Sociology, 51, 343–74. LGBTQ2S youth homelessness: The need for targeted strate-
Shuy, R.W. (2002). In-person versus telephone interviewing. In gies. Social Inclusion, 4, 86–96.
J.F. Gubrium & J.A. Holstein (Eds.), Handbook of interview Armstrong, G. (1993). Like that Desmond Morris? In D. Hobbs &
research: Context and method. Thousand Oaks, CA: SAGE. T. May (Eds.), Interpreting the field: Accounts of ethnography.
Smith, T.W. (1995). Trends in non-response rates. International Oxford: Clarendon Press.
Journal of Public Opinion Research, 7, 157–71. Becker, H.S. (1998). Tricks of the trade: How to think about your
Taniguchi, K. (2020, July 10). Majority view racism as pro- research while doing it. Chicago: University of Chicago Press.
vincial problem: poll. Winnipeg Free Press. [Link] Belk, R., Sherry, J., & Wallendorf, M. (1988). A naturalistic in-
[Link]/local/majority-of-manitobans-support- quiry into buyer and seller behaviour at a swap meet. Journal
[Link] of Consumer Research, 14, 449–70.
Tastsoglou, E., & Miedema, B. (2003). Immigrant women and Bell, E. (2007). Separatism and quasi-separatism in Alberta.
community development in the Canadian Maritimes: Prairie Forum, 32, 335–55.
Outsiders within? Canadian Journal of Sociology, 28, 203–34. Bloor, M. (1997). Addressing social problems through qualitative
Teitler, J., Reichman, N., & Sprachman, S. (2003). Costs and ben- research. In D. Silverman (Ed.), Qualitative research: Theory,
efits of improving response rates for a hard-to-reach popula- method and practice. London: SAGE.
tion. Public Opinion Quarterly, 67, 126–38. Blumer, H. (1931). Science without concepts. American Journal of
Whalen, H., & Schmidt, G. (2016). The women who remain Sociology, 36, 515–33.
behind: challenges in the LDC lifestyle. Rural Society, 25, Blumer, H. (1954). What is wrong with social theory? American
1–14. Sociological Review, 19, 3–10.
Wiens, K., Bhattarai, A., Pedram, P., Dores, A., Williams, J., Blumer, H. (1969). Symbolic interactionism: Perspective and
Bulloch, A., & Patten, S. (2020). A growing need for youth method. Berkeley: University of California Press.
mental health services in Canada: Examining trends in Bochner, A., & Ellis, C. (2003). An introduction to the arts and
youth mental health from 2011 to 2018. Epidemiology and narrative research: Art as inquiry. Qualitative Inquiry, 9,
Psychiatric Sciences 29, e115, 1–9. [Link] 506–14.
S204579602000028 Branker, R. (2017). Labour market discrimination: The lived
experiences of English-speaking Caribbean immigrants in
Toronto. Journal of International Migration & Integration, 18,
Chapter 8 203–22.
Anderssen, E. (2014, October 2). Big data is watching you. Has Bryman, A. (1994). The Mead/Freeman controversy: Some impli-
online spying gone too far? The Globe and Mail. https:// cations for qualitative researchers. In R. Burgess (Ed.), Studies
w w w.t he g lo b e a nd m a i l .c om / l i fe /re l at ion s h ip s / bi g- in Qualitative Methodology (Vol. 4). Greenwich, CT: JAI Press.
data-is-watching-you-has-online-spying-gone-too-far/ Burman, M., Batchelor, S., & Brown, J. (2001). Researching girls
article20894498/?page=all and violence: Facing the dilemmas of fieldwork. British
Agresti, A., & Finlay, B. (2009). Statistical methods for the social Journal of Criminology, 41, 443–59.
sciences. Upper Saddle River, NJ: Prentice Hall. Charmaz, K. (2003). Grounded theory: Objectivist and con-
Bryman, A., & Cramer, D. (2001). Quantitative data analysis structivist methods. In N.K. Denzin & Y.S. Lincoln (Eds.),
with SPSS release 10 for Windows: A guide for social scientists. Strategies for qualitative inquiry (2nd ed.). Thousand Oaks,
London: Routledge. CA: SAGE.

bel65796_ref_412-429 420 12/11/21 05:21 PM

 References 421

Charmaz, K. (2008). A future for symbolic interactionism. Pettigrew, A. (1997). What is a processual analysis? Scandinavian
Studies in Symbolic Interaction, 32, 51–59. Journal of Management, 13, 337–48.
Coloma, R., & Pino, F. (2016). “There’s hardly anything left”: Prus, R. (1997). Subcultural mosaics and intersubjective realities:
Poverty and the economic insecurity of elderly Filipinos in An ethnographic agenda for pragmatizing the social sciences.
Toronto. Canadian Ethnic Studies, 48, 71–97. New York: SUNY Press.
Dolson, M.S. (2018). The role of unpredictability in ethnographic Renzetti, E. (2017, January 6). Women killed by their spouses are
fieldwork. In S.W. Kleinknecht, L.K. van den Scott, & C.B. not casualties in someone else’s story. The Globe and Mail.
Sanders (Eds.), The craft of qualitative research: A handbook. [Link] [Link]/opinion/women-killed-
Toronto: Canadian Scholars. by-their-spouses-are-not-casualties-in-someone-elses-story/
Finlay, L. (2006). “Rigour,” “ethical integrity” or “artistry”? article33535098
Ref lexively reviewing criteria for evaluating qualitative Santin, M., & Kelly, B. (2017). The managed heart revisited:
research. British Journal of Occupational Therapy, 69, Exploring the effect of institutional norms on the emotional
319–26. labor of flight attendants post 9/11. Journal of Contemporary
Gans, H.J. (1962). The urban villagers. New York: Free Press. Ethnography, 46, 519–43.
Geertz, C. (1973). Thick description: Toward an interpretive Shaffir, W., & Kleinknecht, S. (2005). Death at the polls:
theory of culture. In The Interpretation of Cultures. New Experiencing and coping with political defeat. Journal of
York: Basic Books. Contemporary Ethnography, 34, 707–38.
Golafshani, N. (2003). Understanding reliability and validity in Sharma, S., Reimer-Kirkham, S., & Meyerhoff, H. (2011).
qualitative research. The Qualitative Report, 8, 597–606. Filmmaking with Aboriginal youth for type 2 diabetes pre-
Green, A.I. (2010). Queer unions: Same-sex spouses marrying vention. Pimatisiwin: A Journal of Aboriginal and Indigenous
tradition and innovation. Canadian Journal of Sociology, 35, Community Health, 9, 423–40.
399–436. Shelden, D., Angell, M., Stoner, J., & Roseland, B. (2010). School
Guba, E., & Lincoln, Y. (1994). Competing paradigms in qualita- principals’ influence on trust: Perspectives of mothers of chil-
tive research. In N. Denzin & Y. Lincoln (Eds.), Handbook of dren with disabilities. Journal of Educational Research, 103,
qualitative research. Thousand Oaks, CA: SAGE. 159–70.
Hallgrimsdottir, H., Phillips, R., & Benoit, C. (2006). Fallen Silverman, D. (1993). Interpreting qualitative data: Methods for
women and rescued girls: Social stigma and media ­narratives analysing qualitative data. London: SAGE.
of the sex industry in Victoria, B.C., from 1980 to 2005. Skeggs, B. (1994). Situating the production of feminist ethnogra-
Canadian Review of Sociology and Anthropology, 43, 265–80. phy. In M. Maynard & J. Purvis (Eds.), Researching women’s
Hammersley, M. (1992). Deconstructing the qualitative– lives from a feminist perspective. London: Taylor & Francis.
quantitative divide. In What’s wrong with ethnography? Skeggs, B. (1997). Formations of class and gender. London: SAGE.
London: Routledge. Whyte, W.F. (1955). Street corner society (2nd ed.). Chicago:
Hochschild, A. (1983). The managed heart. Berkeley and Los University of Chicago Press.
Angeles: University of California Press. Williams, M. (2000). Interpretivism and generalization.
Karabanow, J. (2002). Open for business: Exploring the life stages Sociology, 34, 209–24.
of two Canadian street youth shelters. Journal of Sociology
and Social Welfare, 29, 99–116. Chapter 10
Kleinknecht, S. (2011). The hacker spirit: An interactionist analysis
of the hacker ideology. In E. Ksenych (Ed.), Exploring deviance Armstrong, G. (1993). Like that Desmond Morris? In D. Hobbs &
in Canada. Toronto: Oxford University Press. T. May (Eds.), Interpreting the field: accounts of ethnography.
Lassiter, L.E. (2005). The Chicago guide to collaborative ethnography. Oxford: Clarendon Press.
Chicago: University of Chicago Press. Atkinson, P. (1981). The clinical experience. Farnborough, UK:
Leidner, R. (1993). Fast food, fast talk: Service work and the routin- Gower.
ization of everyday life. Berkeley and Los Angeles: University Becker, H. (1970). Practitioners of vice and crime. In R.
of California Press. Habenstein (Ed.), Pathways to data. Chicago: Aldine.
Lewis, O. (1961). The children of Sánchez. New York: Vintage. Bell, E. (2007). Separatism and quasi-separatism in Alberta.
Lincoln, Y., & Guba, E. (1985). Naturalistic inquiry. Beverly Hills, Prairie Forum, 32, 335–55.
CA: SAGE. Benbow, S., & Hall, J. (2018). Negotiating tensions in exiting the
Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide field of critical qualitative research. In S.W. Kleinknecht, L.K.
to qualitative observation and analysis. (3rd ed.). Belmont, van den Scott, & C.B. Sanders (Eds.), The craft of qualitative
CA: Wadsworth. research: A handbook. Toronto: Canadian Scholars.
Lyon, K., & Frohard-Dourlent, H. (2015). Let’s talk about the Breckenridge, J., & Jones, D. (2009). Demystifying theoretical
­institution: Same-sex common-law partners negotiating mar- sampling in grounded theory research. Grounded Theory
riage equality and relationship legitimacy. Canadian Review of Review, 8, 113–26.
Sociology, 52, 402–28. Browne, A.J. (2007). Clinical encounters between nurses and
Mason, J. (2002). Qualitative researching (2nd ed.). London: First Nations women in a western Canadian hospital. Social
SAGE. Science & Medicine, 64, 2165–76.
McDonough, P., & Polzer, J. (2012). Habitus, hysteresis, and or- Campbell, M. (2000). Research on health care experiences of
ganizational change in the public sector. Canadian Journal of people with disabilities: Exploring the social organization
Sociology, 37, 357–79. of service delivery. Canadian Journal of Sociology, 1, 131–54.

bel65796_ref_412-429 421 12/11/21 05:21 PM

422 References

Charmaz, K. (2000). Grounded theory: Objectivist and construc- Kleinknecht, S.W., van den Scott, L.K., & Sanders, C.B. (Eds.).
tivist methods. In N. Denzin & Y. Lincoln (Eds.), Handbook (2018). The craft of qualitative research: A handbook. Toronto:
of qualitative research (2nd ed.). Thousand Oaks, CA: SAGE. Canadian Scholars.
Coffey, A. (1999). The ethnographic self: Fieldwork and the repre- Lauder, M. (2003). Covert participant observation of a devi-
sentation of reality. London: SAGE. ant community: Justifying the use of deception. Journal of
Devault, M. (2006). Introduction: What is institutional ethnog- Contemporary Religion, 18, 185–96.
raphy? Social Problems, 53, 294–98. Lee-Treweek, G. (2000). The insight of emotional danger:
Ditton, J. (1977). Part-time crime: An ethnography of fiddling and Research experiences in a home for the elderly. In G. Lee-
pilferage. London: Macmillan. Treweek & S. Linkogle (Eds.), Danger in the field: Risk and
Edwards, N. (2018). On (still) being emotionally attached to the ethics in social research. London: Routledge.
field. In S.W. Kleinknecht, L.K. van den Scott, & C.B. Sanders Leidner, R. (1993). Fast food, fast talk: Service work and the routin-
(Eds.), The craft of qualitative research: A handbook. Toronto: ization of everyday life. Berkeley and Los Angeles: University
Canadian Scholars. of California Press.
Fine, G. (1996). Justifying work: Occupational rhetorics as Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide
resources in kitchen restaurants. Administrative Science to qualitative observation and analysis (3rd ed.). Belmont, CA:
Quarterly, 41, 90–115. Wadsworth.
Gans, H.J. (1968). The participant-observer as human being: Norris, C. (1993). Some ethical considerations on fieldwork with
Observations on the personal aspects of field work. In H. the police. In D. Hobbs & T. May (Eds.), Interpreting the field:
Becker (Ed.), Institutions and the person: Papers presented to Accounts of ethnography. Oxford: Clarendon Press.
Everett C. Hughes. Chicago: Aldine. Pink, S. (2001). Visual ethnography. London: SAGE.
Gerson, K., & Horowitz, R. (2002). Observation and interview- Plana, V. (2019, December 4). UBC professor to
ing: Options and choices. In T. May (Ed.), Qualitative re- debunk unwritten rules, behaviour of public tran-
search in action. London: SAGE. sit riders. Daily Hive. [Link]
Giulianotti, R. (1995). Participant observation and research into ubc-amy-hanser-rules-behaviour-bus-riders
football hooliganism: Reflections on the problems of entrée Ranasinghe, P. (2017). Helter shelter: Security, legality and an
and everyday risks. Sociology of Sport Journal, 12, 1–20. ethic of care in an emergency shelter. Toronto: University of
Glaser, B., & Strauss, A. (1967). The discovery of grounded theory: Toronto Press.
Strategies for qualitative research. Chicago: Aldine. Ratković, S., & Sethi, B. (2018). Dissemination and social jus-
Glucksmann, M. (1994). The work of knowledge and the knowl- tice. In S.W. Kleinknecht, L.K. van den Scott, & C.B. Sanders
edge of women’s work. In M. Maynard & J. Purvis (Eds.), (Eds.), The craft of qualitative research: A handbook. Toronto:
Researching women’s lives from a feminist perspective. Canadian Scholars.
London: Taylor & Francis. Restoule, J., Mashford-Pringle, A., Chacaby, M, Smillie, C.,
Goffman, E. (1956). The presentation of self in everyday life. New Brunette, C., & Russel, G. (2013). Supporting successful
York: Doubleday. t ransitions to post-secondary education for Indigenous
­
Gold, R. (1958). Roles in sociological fieldwork. Social Forces, 36, Students: Lessons from an institutional ethnography in
217–23. Ontario, Canada. International Indigenous Policy Journal,
Hallett, R., & Barber, K. (2014). Ethnographic research in a cyber 4, 1–10.
era. Journal of Contemporary Ethnography, 43, 306–30. Rosenhan, D.L. (1973). On being sane in insane places. Science,
Hammersley, M., & Atkinson, P. (1995). Ethnography: Principles 179, 350–58.
in practice (2nd ed.). London: Routledge. Sanjek, R. (1990). A vocabulary for fieldnotes. In R. Sanjek (Ed.),
Hessler, R., Downing, J., Beltz, C., Pelliccio, A., Powell, M., & Fieldnotes: The making of anthropology. Ithaca, NY: Cornell
Vale, W. (2003). Qualitative research on adolescent risk using University Press.
e-mail: A methodological assessment. Qualitative Sociology, Santin, M., & Kelly, B. (2017). The managed heart revisited:
26, 111–24. Exploring the effect of institutional norms on the emotional
Hier, S. (2002). Raves, risks and the ecstasy panic: A case study in labor of flight attendants post 9/11. Journal of Contemporary
the subversive nature of moral regulation. Canadian Journal Ethnography, 46, 519–43.
of Sociology, 27, 33–57. Sarsby, J. (1984). The fieldwork experience. In R. Ellen (Ed.),
Hobbs, D. (1988). Doing the business: Entrepreneurship, the work- Ethnographic research: A guide to general conduct. London:
ing class and detectives in the east end of London. Oxford: Academic Press.
Oxford University Press. Schulenberg, J.L. (2014). Systematic social observation of police
Holtby, A., Klein, K., Cook, K., & Travers, R. (2015). To be seen decision-making: The process, logistics, and challenges in a
or not to be seen: Photovoice, queer and trans youth, and the Canadian context. Quality and Quantity, 48, 297–315.
dilemma of representation. Action Research, 13, 317–35. Shaffir, W. (2018). Leaving the field trajectories: Researching
Humphreys, L. (1970). Tearoom trade: Impersonal sex in public Hasidic Jews. In S.W. Kleinknecht, L.K. van den Scott, & C.B.
places. Chicago: Aldine. Sanders (Eds.), The craft of qualitative research: A handbook.
Karabanow, J. (2002). Open for business: Exploring the life stages Toronto: Canadian Scholars.
of two Canadian street youth shelters. Journal of Sociology Smith, D. (2005). Institutional ethnography: A sociology for
and Social Welfare, 29, 99–116. people. Walnut Creek, CA: AltaMira Press.

bel65796_ref_412-429 422 12/11/21 05:21 PM

 References 423

Strauss, A., & Corbin, J. (1990). Basics of qualitative research: Bampton, R., & Cowton, C.J. (2002). The e-interview. Forum:
Grounded theory procedures and techniques. Newbury Park, Qualitative Social Research, 3. [Link]
CA: SAGE. [Link]/[Link]/fqs/article/view/848
Strauss, A., & Corbin, J. (1998). Basics of qualitative research: Barter, C., & Renold, E. (1999). The use of vignettes in qualitative
Techniques and procedures for developing grounded theory. research. Social Research Update, 25.
Thousand Oaks, CA: SAGE. Beardsworth, A., & Keil, T. (1992). The vegetarian option:
Taylor, A. (1993). Women drug users: An ethnography of an inject- Varieties, conversions, motives and careers. Sociological
ing community. Oxford: Clarendon Press. Review, 40, 253–93.
Totten, M. (2001). Legal, ethical, and clinical implications of Becker, H., & Geer, B. (1957). Participant observation and inter-
doing fieldwork with youth gang members who engage in se- viewing: A comparison. Human Organization, 16, 28–32.
rious violence. Journal of Gang Research, 8, 35–49. Bell, E. (2007). Separatism and quasi-separatism in Alberta.
van den Scott, J. (2018). Leaving the field/Can you leave the field? Prairie Forum, 32, 335–55.
In S.W. Kleinknecht, L.K. van den Scott, & C.B. Sanders Bloor, M., Frankland, S., Thomas, M., & Robson, K. (2001). Focus
(Eds.), The craft of qualitative research: A handbook. Toronto: groups in social research. London: SAGE.
Canadian Scholars. Bryman, A. (1999). Global Disney. In P. Taylor & D. Slater (Eds.),
van den Scott, L.K. (2018). Role transitions in the field and reflex- The American century. Oxford: Blackwell.
ivity: From friend to researcher. (In special issue: Emotion Bryman, A., Stephens, M., & Campo, C. (1996). The importance
and the researcher: Sites, subjectivities and relationships). of context: Qualitative research and the study of leadership.
Studies in Qualitative Methodology, 16, 19–32. Leadership Quarterly, 7, 353–70.
Van Maanen, J. (1991). Playing back the tape: Early days in the Charmaz, K. (2002). Qualitative interviewing and grounded
field. In W. Shaffir & R. Stebbins (Eds.), Experiencing field- theory analysis. In J. Gubrium & J. Holstein (Eds.), Handbook
work: An inside view of qualitative research. Newbury Park, of interview research: Context and method. Thousand Oaks,
CA: SAGE. CA: SAGE.
Whyte, W.F. (1955). Street corner society (2nd ed.). Chicago: Clapper, D., & Massey, A. (1996). Electronic focus groups: A
University of Chicago Press. framework for exploration. Information and Management,
Wilson, B. (2002). The Canadian rave scene and five theses on 30, 43–50.
youth resistance. Canadian Journal of Sociology, 27, 373–412. Curasi, C. (2001). A critical exploration of face-to-face interviewing
Winlow, S., Hobbs, D., Lister, S., & Hadfield, P. (2001). Get ready vs. computer-mediated interviewing. International Journal of
to duck: Bouncers and the realities of ethnographic research Market Research, 43, 361–75.
on violent groups. British Journal of Criminology, 41, 536–48. Davies, P. (2000). Doing interviews with female offenders. In V.
Wolcott, H. (1990). Writing up qualitative research. Newbury Jupp, P. Davies, & P. Francis (Eds.), Doing criminological re-
Park, CA: SAGE. search. London: SAGE.
Wolf, D. (1991). High risk methodology: Reflections on leav- Deacon, D., Pickering, M., Golding, P., & Murdock, G. (1999).
ing an outlaw society. In W. Shaffir & R. Stebbins (Eds.), Researching communications: A practical guide to methods in
Experiencing fieldwork: An inside view of qualitative research. media and cultural analysis. London: Arnold.
Newbury Park, CA: SAGE. Eady, A., Dobinson, C., & Ross, L. (2011). Bisexual people’s experi-
ences with mental health services: A qualitative investigation.
Community Mental Health Journal, 47, 378–89.
Chapter 11
Frohlich, K., Potvin, L., Chabot, P., & Corin, E. (2002). A theo-
Adriaenssens, C., & Cadman, L. (1999). An adaptation of mod- retical and empirical analysis of context: Neighbourhoods,
erated e-mail focus groups to assess the potential of a new smoking, and youth. Social Science and Medicine, 54,
online (Internet) financial services offer in the UK. Journal of 1401–17.
the Market Research Society, 41, 417–24. Gubrium, J., & Holstein, J. (1997). The new language of qualitative
Ajodhia-Andrews, A. (2016). Reflexively conducting research method. New York: Oxford University Press.
with ethnically diverse children with disabilities. The Hanna, Paul. (2012). Using Internet technologies (such as Skype)
Qualitative Report, 21, 252–87. as a research medium: A research note. Qualitative Research,
Armstrong, G. (1993). Like that Desmond Morris? In D. Hobbs & 12, 239–42.
T. May (Eds.), Interpreting the field: Accounts of ethnography. Hine, V. (2000). Virtual ethnography. London: SAGE.
Oxford: Clarendon Press. Holbrook, B., & Jackson, B. (1996). Shopping around: Focus
Atkinson, M. (2002). Pretty in ink: Conformity, resistance and group research in North London. Area, 28, 136–42.
negotiation in women’s tattooing. Sex Roles, 47, 219–35. Hughes, R. (1998). Considering the vignette technique and its
Atkinson, M. (2004). Tattooing and civilizing processes. ­application to a study of drug injecting and HIV risk and
Canadian Review of Sociology and Anthropology, 41, 125–46. safer behaviour. Sociology of Health and Illness, 20, 381–400.
Atkinson, P. (1990). The ethnographic imagination: Textual con- Jones, Z., & Hannem, S. (2018). Escort clients’ sexual scripts and
structions of society. London: Routledge. constructions of intimacy in commodified sexual relationships.
Atkinson, P. (2004). Life story interview. In M. Lewis-Beck, A. Symbolic Interaction, 41, 488–512.
Bryman, & T. Liao (Eds.), The SAGE. encyclopedia of social Kitzinger, J. (1993). Understanding AIDS: Researching audi-
science research methods. Thousand Oaks, CA: SAGE. ence perceptions of acquired immune deficiency syndrome.

bel65796_ref_412-429 423 12/11/21 05:21 PM

424 References

In J. Eldridge (Ed.), Getting the message: News, truth and O’Connor, H., & Madge, C. (2001). Cyber-mothers: Online synchro-
power. London: Routledge. nous interviewing using conferencing software. Sociological
Kitzinger, J. (1994). The methodology of focus groups: The impor- Research Online, 5. [Link]/5/4/o’[Link].
tance of interaction between research participants. Sociology O’Connor, H., & Madge, C. (2003). “Focus groups in cyberspace”:
of Health and Illness, 16, 103–21. Using the Internet for qualitative research. Qualitative
Krueger, R. (1998). Moderating focus groups. Thousand Oaks, Market Research, 6, 133–43.
CA: SAGE. Parker, M. (2000). Organizational culture and identity. London:
Kvale, S. (1996). Interviews: An introduction to qualitative re- SAGE.
search interviewing. Thousand Oaks, CA: SAGE. Poland, B.D. (1995). Transcription quality as an aspect of rigor in
Lareau, A., & Weininger, E. (2003). Cultural capital in educa- qualitative research. Qualitative Inquiry, 1, 290–310.
tional research: A critical assessment. Theory and Society, 32, Price, T., & Puddephatt, A. (2018). “Opening access” to open
185–96. access editors: Communication technologies in long distance
Lett, D., Hier, S.P., & Walby, K. (2010). CCTV surveillance and interviewing. In S.W. Kleinknecht, L.K. van den Scott, & C.B.
the civic conversation: A study in public sociology. Canadian Sanders (Eds.), The craft of qualitative research: A handbook.
Journal of Sociology, 35, 437–62. Toronto: Canadian Scholars.
Lewis, O. (1961). The children of Sánchez. New York: Vintage. Proudfoot, S. (2019, July 16). What does it mean to be work-
Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide ing class in Canada? Maclean’s. [Link]
to qualitative observation and analysis (3rd ed.). Belmont, CA: society/what-does-it-mean-to-be-working-class-in-canada/
Wadsworth. Rafaeli, A., Dutton, J., Harquail, C.V., & Mackie-Lewis, S. (1997).
Madge, C., & O’Connor, H. (2002). On-line with e-mums: Navigating by attire: The use of dress by female administra-
Exploring the Internet as a medium of research. Area, 34, tive employees. Academy of Management Journal, 40, 9–45.
92–102. Ricciardelli, R. (2019). Also serving time: Canada’s provincial
Madriz, M. (2000). Focus groups in feminist research. In N. and territorial correctional officers. Toronto: University of
Denzin & Y. Lincoln (Eds.), Handbook of qualitative research Toronto Press.
(2nd ed.). Thousand Oaks, CA: SAGE. Ricciardelli, R., & Adorjan, M. (2019). “If a girl’s photo gets sent
Maiolino, E. (2019). Words and deeds: Social movement actors’ around, that’s a way bigger deal than if a guy’s photo gets sent
assessments of allied politicians. In J.M. Everitt & A. Wagner around”: Gender, sexting, and the teenage years. Journal of
(Eds.), Gendered mediation: Identity and image making in Gender Studies, 28, 563–77.
Canadian politics. Vancouver: UBC Press. Schlesinger, P., Dobash, R.E., Dobash, R.P., & Weaver, C. (1992).
Malbon, B. (1999). Clubbing: Dancing, ecstasy and vitality. Women viewing violence. London: British Film Institute.
London: Routledge. Silverman, D. (1993). Interpreting qualitative data: Methods for
Mann, C., & Stewart, F. (2000). Internet communication and analysing qualitative data. London: SAGE.
qualitative research: A handbook for researching online. Smith, R. (2008). Pain in the act: The meanings of pain among
London: SAGE. professional wrestlers. Qualitative Sociology, 31, 129–48.
Mason, J. (2002). Qualitative interviewing: Asking, listening, in- Sugiman, P. (2004). Memories of the internment: Narrating
terpreting. In T. May (Ed.), Qualitative Research in Action. Japanese-Canadian women’s life stories. Canadian Journal of
London: SAGE. Sociology, 29, 359–88.
Miller, R.L. (2000). Researching life stories and family histories. Sweet, C. (2001). Designing and conducting virtual focus groups.
London: SAGE. Qualitative Market Research, 4, 130–35.
Milne, E. (2016). “I have the worst fear of teachers”: Moments of Tastsoglou, E., & Miedema, B. (2003). Immigrant women and com-
inclusion and exclusion in family/school relationships among munity development in the Canadian Maritimes: Outsiders
Indigenous families in southern Ontario. Canadian Review of within? Canadian Journal of Sociology, 28, 203–34.
Sociology, 53, 270–89. Torelli, J. (2019). On entering the field: Notes from a neophyte
Milne, E. (2017). Implementing Indigenous education policy di- researcher. Qualitative Sociology Review, 15, 64–92.
rectives in Ontario public schools: Experiences, challenges Trow, M. (1957). Comment on “Participant Observation and
and successful practices. International Indigenous Policy Interviewing: A Comparison.” Human Organization, 16,
Journal, 8, 1–20. 33–35.
Miraftab, F. (2000). Sheltering refugees: The housing experience Tse, A. (1999). Conducting electronic focus group discussions
of refugees in Metropolitan Vancouver, Canada. Canadian among Chinese respondents. Journal of the Market Research
Journal of Urban Research, 9, 42–63. Society, 41, 407–15.
Morgan, D. (1998). Planning focus groups. Thousand Oaks, CA: Wilkinson, S. (1998). Focus groups in feminist research: Power,
SAGE. interaction, and the co-production of meaning. Women’s
Morgan, D. (2002). Focus group interviewing. In J. Gubrium & J. Studies International Forum, 21, 111–25.
Holstein (Eds.), Handbook of interview research: Context and Wilkinson, S. (1999). Focus group methodology: A review.
method. Thousand Oaks, CA: SAGE. International Journal of Social Research Methodology, 1,
Morgan, D., & Spanish, M. (1985). Social interaction and the cog- 181–203.
nitive organization of health-relevant behaviour. Sociology of Wilson, B. (2002). The Canadian rave scene and five theses on
Health and Illness, 7, 401–22. youth resistance. Canadian Journal of Sociology, 27, 373–412.

bel65796_ref_412-429 424 12/11/21 05:21 PM

 References 425

Wright, R., Alaggia, R., & Krygsman, A. (2014). Five-year follow- grassroots mobilization in Canadian newspapers. Media,
up study of the qualitative experiences of youth in an after- Culture & Society, 40, 817–37.
school arts program in low-income communities. Journal of Dunning, E., Murphy, P., & Williams, J. (1988). The roots of
Social Science Research, 40, 137–46. football hooliganism: An historical and sociological study.
London: Routledge.
Fenton, N., Bryman, A., & Deacon, D. (1998). Mediating social
Chapter 12
science. London: SAGE.
Abraham, J. (1994). Bias in science and medical knowledge: The Finkel, E., Eastwick, P., Karney, B., Reis, H., & Sprecher, S. (2012).
Opren controversy. Sociology, 28, 717–36. Online dating: A critical analysis from the perspective of
Altheide, D.L. (2004). Ethnographic content analysis. In M. ­psychological science. Psychological Science in the Public
Lewis-Beck, A. Bryman, & T. Liao (Eds.), The SAGE. ency- Interest, 13, 3–66.
clopedia of social science research methods. Thousand Oaks, Forster, N. (1994). The analysis of company documentation. In C.
CA: SAGE. Cassell & G. Symon (Eds.), Qualitative methods in organizational
Altheide, D.L., & Schneider, C.J. (2013). Qualitative media analy- research. London: SAGE.
sis (2nd ed.). Thousand Oaks, CA: SAGE. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood
Atkinson, J.M., & Drew, P. (1979). Order in court: The orga- Cliffs, NJ: Prentice Hall.
nization of verbal interaction in judicial settings. London: Gill, R. (1996). Discourse analysis: Practical implementation. In
Macmillan. J. Richardson (Ed.), Handbook of qualitative research methods
Bell, E. (2007). Separatism and quasi-separatism in Alberta. for psychology and the social sciences. Leicester: BPS Books.
Prairie Forum, 32, 335–55. Gill, R. (2000). Discourse analysis. In M. Bauer & G. Gaskell
Bell, E., Jansen, H., & Young, L. (2007). Sustaining a dynasty in (Eds.), Qualitative researching with text, image and sound.
Alberta: The 2004 provincial election. Canadian Political London: SAGE.
Science Review, 1, 27–49. Giulianotti, R. (1997). Enlightening the North: Aberdeen fan-
Bhuyan, R., Jeyapal, D., Ku, J., Sakamoto, I., & Chou, E. (2017). zines and local football identity. In G. Armstrong & R.
Branding “Canadian experience” in immigration policy: Giulianotti (Eds.), Entering the field: New perspectives on
Nation building in a neoliberal era. Journal of International world football. Oxford: Berg.
Migration & Integration, 18, 47–62. Gottdiener, M. (1997). The theming of America: Dreams, visions
Billig, M. (1991). Ideology and opinions: Studies in rhetorical psy- and commercial spaces. Boulder, CO: Westview Press.
chology. Cambridge: Cambridge University Press. Graydon, S. (2019, February 8). Graydon: In numbers there is
Blaikie, A. (2001). Photographs in the cultural account: Contested strength—how big data can help close the gender gap in Canadian
narratives and collective memory in the Scottish islands. journalism. Ottawa Citizen. [Link]
Sociological Review, 49, 345–67. columnists/graydon-in-numbers-there-is-strength-how-big-
Blommaert, J., & Bulcaen, C. (2000). Critical Discourse Analysis. data-can-help-close-the-gender-gap-in-canadian-journalism
Annual Review of Anthropology, 29, 447–66. Gubrium, J., & Holstein, J. (1997). The new language of qualitative
Bryman, A. (1995). Disney and his worlds. London: Routledge. method. New York: Oxford University Press.
Buchanan, C. (2014). A more national representation of place Hales, J. (2006). An anti-colonial critique of research methodol-
in Canadian daily newspapers. Canadian Geographer, 58, ogy. In G. Dei & A. Kempf (Eds.), Anti-colonialism and edu-
517–30. cation: The politics of resistance. Rotterdam: Sense.
Choi, G.Y. & Lewallen, J. (2018). “Say Instagram, kids!”: Hall, J., Park, N., Song, H., & Cody, M. (2010). Strategic mis-
Examining sharenting and children’s digital representa- representation in online dating: The effects of gender, self-
tions on Instagram. Howard Journal of Communications, 29, monitoring, and personality traits. Journal of Social and
144–64. Personal Relationships, 27, 117–35.
Clayman, S., & Gill, V.T. (2004). Conversation analysis. In Hallgrimsdottir, H., Phillips, R., & Benoit, C. (2006). Fallen
M. Hardy & A. Bryman (Eds.), Handbook of data analysis. women and rescued girls: Social stigma and media narra-
London: SAGE. tives of the sex industry in Victoria, B.C., from 1980 to 2005.
Coté, J., & Allahar, A. (1994). Generation on hold: Coming of age Canadian Review of Sociology and Anthropology, 43, 265–80.
in the late twentieth century. Toronto: Stoddart. Harding, R. (2006). Historical representations of Aboriginal
Crook, C., & Light, P. (2002). Virtual society and the cul- people in the Canadian news media. Discourse & Society, 17,
tural practice of study. In S. Woolgar (Ed.), Virtual society? 205–35.
Technology, cyberbole, reality. Oxford: Oxford University Harding, R. (2010). The demonization of Aboriginal child welfare
Press. authorities in the news. Canadian Journal of Communication,
Dickinson, H. (1993). Accounting for Augustus Lamb: Theoretical 35, 85–108.
and methodological issues in biography and historical sociology. Hatton, E., & Trautner, M. (2013). Images of powerful women in
Sociology, 27, 121–32. the age of “choice feminism.” Journal of Gender Studies, 22,
Disney Miller, D. (1956). The story of Walt Disney. New York: 65–78.
Dell. Heath, C. (1997). The analysis of activities in face to face interac-
Dumitrica, D., & Bakardjieva, M. (2018). The personalization of tion using video. In D. Silverman (Ed.), Qualitative research:
engagement: The symbolic construction of social media and Theory, method and practice. London: SAGE.

bel65796_ref_412-429 425 12/11/21 05:21 PM

426 References

Heritage, J. (1984). Garfinkel and ethnomethodology. Cambridge, Marcic, D. (2002). Respect: Women and popular music. New York:
UK: Polity. Texere.
Heritage, J. (1987). Ethnomethodology. In A. Giddens & J. Turner McGuigan, J. (1992). Cultural populism. London: Routledge.
(Eds.), Social theory today. Cambridge, UK: Polity. Miller, N., & Morgan, D. (1993). Called to account: The CV as an
Hier, S. (2000). The contemporary structure of Canadian racial autobiographical practice. Sociology, 27, 133–43.
supremacism; Networks, strategies, and new technologies. Morton, K. (2016). Hitchhiking and missing and murdered
Canadian Journal of Sociology, 25, 471–94. Indigenous women: A critical discourse analysis of billboards
Hier, S. (2002). Raves, risks and the ecstasy panic: A case study in on the Highway of Tears. Canadian Journal of Sociology, 41,
the subversive nature of moral regulation. Canadian Journal 299–326.
of Sociology, 27, 33–57. Nettleton, S., Pleace, N., Burrows, R., Muncer, S., & Loader, B.
Hirsch, J. (1981). Family photographs. New York: Oxford (2002). The reality of virtual social support. In S. Woolgar
University Press. (Ed.), Virtual society? Technology, cyberbole, reality. Oxford:
Hitsch, G., Hortaçscu, A., & Ariely, D. (2010). Matching and Oxford University Press.
scoring in online dating. American Economic Review, 100, O’Connor, C.D. (2017). The police on Twitter: image manage-
130–63. ment, community building, and implications for policing in
Hodson, R. (1996). Dignity in the workplace under participative Canada. Policing and Society, 27, 899–912.
management. American Sociological Review, 61, 719–38. Parnaby, P. (2003). Disaster through dirty windshields: Law,
Hutchby, I., & Wooffitt, R. (1998). Conversation analysis. order and Toronto’s squeegee kids. Canadian Journal of
Cambridge, UK: Polity. Sociology, 28, 281–307.
Jagger, E. (1998). Marketing the self, buying an other: Dating in a Pauwels, L. (2008). A private visual practice going public? Social
post modern, consumer society. Sociology, 32, 795–814. functions and sociological research opportunities of web-
Jagger, E. (2001). Marketing Molly and Melville: Dating in a post- based family photography. Visual Studies, 23, 34–49.
modern, consumer society. Sociology, 35, 39–57. Phillips, N., & Brown, J. (1993). Analyzing communications in
John, I.D. (1992). Statistics as rhetoric in psychology. Australian and around organizations: A critical hermeneutic approach.
Psychologist, 27, 144–49. Academy of Management Journal, 36, 1547–76.
Johnston, G. (2020). The kids are all white: Examining race and Potter, J. (1996). Representing reality: Discourse, rhetoric and
representation in news media coverage of opioid overdose social construction. London: SAGE.
deaths in Canada. Sociological Inquiry, 90, 123–46. Potter, J. (1997). Discourse analysis as a way of analysing nat-
Johnston, G., Sanscartier, M.D., Johnston, M.S. (2019). Retail urally occurring talk. In D. Silverman (Ed.), Qualitative
therapy: Making meaning out of menial labour. Journal of ­research: Theory, method and practice. London: SAGE.
Sociology, 55, 446–62. Potter, J. (2004). Discourse analysis. In M. Hardy & A. Bryman
Johnston, M.S. (2020). “He sees patients as lesser people”: (Eds.), Handbook of data analysis. London: SAGE.
Exploring mental health service users’ critiques and apprais- Potter, J., & Hepburn, A. (2004). The analysis of NSPCC call
als of psychiatrists in Canada. Disability & Society, 35, 258–79. openings. In S. Becker & A. Bryman (Eds.), Understanding
Johnston, M.S., Sanscartier, M.D., & Johnston, G. (2018). Dirty research for social policy and practice: Themes, methods, and
work, dirty resistance: Digital warfare in the era of precarious approaches. Bristol: Policy Press.
labor. Canadian Review of Sociology, 55, 278–97. Potter, J., & Wetherell, M. (1994). Analyzing discourse. In A.
Jones, K. (2000). Constructing rBST in Canada: Biotechnology, Bryman & R. Burgess (Eds.), Analyzing qualitative data.
instability, and the management of nature. Canadian Journal London: Routledge.
of Sociology, 25, 311–41. Potter, J., Wetherell, M., & Chitty, A. (1991). Quantification rhetoric—
Kempf, A. (2006). Anti-colonial historiography: Interrogating Cancer on television. Discourse and Society, 2, 333–65.
colonial education. In G. Dei & A. Kempf (Eds.), Anti- Psathas, G. (1995). Conversation analysis: The study of talk-in-
colonialism and education: The politics of resistance. interaction. Thousand Oaks, CA: SAGE.
Rotterdam: Sense. Reed, M. (2000). The limits of discourse analysis in organiza-
Kennedy, R. (2006). Researching the intersection between collec- tional analysis. Organization, 7, 524–30.
tive identity and conceptions of post-separation and divorced Riches, G., & Dawson, P. (1998). Lost children, living memories:
fatherhood: A case study fathers for justice, fathers for just The role of photographs in processes of grief and adjustment
us, or fathers are us. Qualitative Sociology Review, 2, 75–97. among bereaved parents. Death Studies, 22, 121–40.
Lalancette, M., & Raynauld, V. (2019). The power of political Schegloff, E. (1997). Whose text? Whose context? Discourse and
image: Justin Trudeau, Instagram, and celebrity politics. Society, 8, 165–87.
American Behavioral Scientist, 63, 888–924. Schneider, C., & Hannem, S. (2019). Politicization of sexual miscon-
Lynch, M., & Bogen, D. (1997). Sociology’s asociological “core”: duct as symbolic annihilation: An analysis of news media cover-
An examination of textbook sociology in the light of the age of the 2016 “rape election.” Sexuality & Culture, 23, 737–758.
sociology of scientific knowledge. American Sociological Scott, J. (1990). A matter of record. Cambridge, UK: Polity.
Review, 62, 481–93. Seale, C. (2002). Cancer heroics: A study of news reports with
Mannay, D. (2018). Ethnography in inaccessible fields: Drawing particular reference to gender. Sociology, 36, 107–26.
on visual approaches to understand the private sphere of Silverman, D. (1984). Going private: Ceremonial forms in a pri-
the home. In S.W. Kleinknecht, L.K. van den Scott, & C.B. vate oncology clinic. Sociology, 18, 191–204.
Sanders (Eds.), The craft of qualitative research: A handbook. Silverman, D. (1993). Interpreting qualitative data: Methods for
Toronto: Canadian Scholars. analysing qualitative data. London: SAGE.

bel65796_ref_412-429 426 12/11/21 05:21 PM

 References 427

Silverman, D. (1994). Analysing naturally occurring data Charmaz, K. (1983). The grounded theory method: An explica-
on AIDS counselling: Some methodological and practi- tion and interpretation. In R. Emerson (Ed.), Contemporary
cal issues. In M. Boulton (Ed.), Challenge and innovation: field research: A collection of readings. Boston: Little, Brown.
Methodological advances in social research on HIV/AIDS. Charmaz, K. (1997). Identity dilemmas of chronically ill men.
London: Taylor & Francis. In A. Strauss & J. Corbin (Eds.), Grounded theory in practice.
Sturgeon, R., & Morrissette, P. (2010). A qualitative analysis of Thousand Oaks, CA: SAGE.
suicide ideation among Manitoba farmers. Canadian Journal Charmaz, K. (2000). Grounded theory: Objectivist and construc-
of Counselling, 44, 191–207. tivist methods. In N. Denzin & Y. Lincoln (Eds.), Handbook
Sugiman, P. (2004). Memories of the internment: Narrating of qualitative research (2nd ed.). Thousand Oaks, CA: SAGE.
Japanese-Canadian women’s life stories. Canadian Journal of Charmaz, K. (2004). Grounded theory. In M. Lewis-Beck, A.
Sociology, 29, 359–88. Bryman, & T. Liao (Eds.), The SAGE. encyclopedia of social
Sutton, R.I. (1992). Feelings about a Disneyland visit: science research methods. Thousand Oaks, CA: SAGE.
Photography and the reconstruction of bygone emotions. Charmaz, K. (2006). Constructing grounded theory: A practical
Journal of Management Inquiry, 1, 278–87. guide through qualitative analysis. Thousand Oaks, CA: SAGE.
Thompson, T., & Zerbinos, E. (1995). Gender roles in animated car- Coffey, A., & Atkinson, P. (1996). Making sense of qualitative
toons: Has the picture changed in 20 years? Sex Roles, 32, 651–73. data: Complementary research strategies. Thousand Oaks,
Wada, M., Hurd Clarke, L., & Mortenson, W.B. (2019). “I am busy CA: SAGE.
independent woman who has sense of humor, caring about Coffey, A., Holbrook, B., & Atkinson, P. (1996). Qualitative data anal-
others”: Older adults’ self-representations in online dating ysis: Technologies and representations. Sociological Research
profiles. Ageing & Society, 39, 951–76. Online, 2. [Link] [Link]/socresonline/1/1/[Link]
Wetherell, M. (1998). Positioning and interpretative repertoires: Falconi, E.A., & Graber, K.E. (Eds.). (2019). Storytelling as nar-
Conversation analysis and post-structuralism in dialogue. rative practice: Ethnographic approaches to the tales we tell.
Discourse and Society, 9, 387–412. Boston: Brill.
Widdicombe, S. (1993). Autobiography and change: Rhetoric and Frank, L. (2020). Out of milk: Infant food insecurity in a rich
authenticity of “gothic” style. In E. Burman & I. Parker (Eds.), nation. Vancouver: UBC Press.
Discourse analytic research: Readings and repertoires of text. Fraser, E. (2019, May 2). Stress of floodwaters can take toll
London: Routledge. on mental health. CBC News. [Link]
canada/new-brunswick/mental-illness-f looding-stress-
st-john-river-1.5119636
Chapter 13
Glaser, B., & Strauss, A. (1967). The discovery of grounded theory:
Atkinson, M. (2002). Pretty in ink: Conformity, resistance and Strategies for qualitative research. Chicago: Aldine.
negotiation in women’s tattooing. Sex Roles, 47, 219–35. Hesse-Biber, S. (1995). Unleashing Frankenstein’s monster?
Atkinson, M. (2004). Tattooing and civilizing processes. The use of computers in qualitative research. Studies in
Canadian Review of Sociology and Anthropology, 41, 125–46. Qualitative Methodology, 5, 25–41.
Bissett, K. (2019, April 29). Study of 2018 New Brunswick flooding Locke, K. (1996). Rewriting The discovery of grounded theory
looks at ways of easing mental health toll. National Post. https:// after 25 years? Journal of Management Inquiry, 5, 239–45.
[Link]/pmn/news-pmn/canada-news-pmn/study- Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide
of-2018-new-brunswick-flooding-looks-at-ways-of-easing- to qualitative observation and analysis (3rd ed.). Belmont, CA:
mental-health-toll Wadsworth.
Brower, R.L., Bertrand Jones, T., Osborne-Lampkin, L., Hu, S., Lonkila, M. (1995). Grounded theory as an emergent paradigm for
& Park-Gaghan, T.J. (2019). Big qual: Defining and debating computer-assisted qualitative data analysis. In U. Kelle (Ed.),
qualitative inquiry for large data sets. International Journal of Computer-aided qualitative data analysis. London: SAGE.
Qualitative Methods, 18, 1–10. Miles, M.B. (1979). Qualitative data as an attractive nuisance.
Bryman, A., & Burgess, R. (1994). Reflections on qualitative data Administrative Science Quarterly, 24, 590–601.
analysis. In A. Bryman & R. Burgess (Eds.), Analyzing quali- Miller, R.L. (2000). Researching life stories and family histories.
tative data. London: Routledge. London: SAGE.
Bryman, A., Gillingwater, D., & McGuinness, I. (1996). Industry Okely, J. (1994). Thinking through fieldwork. In A. Bryman
culture and strategic response: The case of the British bus & R. Burgess (Eds.), Analyzing qualitative data. London:
industry. Studies in Cultures, Organizations and Societies, 2, Routledge.
191–208. Pidgeon, N., & Henwood, K. (2004). Grounded theory. In M.
Bury, M. (2001). Illness narratives: Fact or fiction? Sociology of Hardy & A. Bryman (Eds.), Handbook of data analysis.
Health and Illness, 23, 263–85. London: SAGE.
Catterall, M., & Maclaran, P. (1997). Focus group data and qual- Riessman, C.K. (1993). Narrative analysis. Newbury Park, CA:
itative analysis programs: Coding the moving picture as well SAGE.
as snapshots. Sociological Research Online, 2. [Link] Riessman, C.K. (2004a). Narrative interviewing. In M. Lewis-
[Link]/socresonline/2/1/[Link] Beck, A. Bryman, & T. Liao (Eds.), The SAGE. encyclopedia of
CBC News. (2019, June 10). Have back-to-back floods changed social science research methods. Thousand Oaks, CA: SAGE.
views on climate change? UNB prof launches study. https:// Riessman, C.K. (2004b). Narrative analysis. In M. Lewis-Beck,
[Link]/news/canada/new-brunswick/new-brunswick- A. Bryman, & T. Liao (Eds.), The SAGE. encyclopedia of social
flood-climate-change-research-woodhall-melnik-1.5168912 science research methods. Thousand Oaks, CA: SAGE.

bel65796_ref_412-429 427 12/11/21 05:21 PM

428 References

Roberts, B. (2002). Biographical research. Buckingham, UK: Bloor, M. (1997). Addressing social problems through qualitative
Open University Press. research. In D. Silverman (Ed.), Qualitative research: Theory,
Sanders, C.B., & van den Scott, L.K. (2018). Analyzing material- method and practice. London: SAGE.
ity. In S.W. Kleinknecht, L.K. van den Scott, & C.B. Sanders Charmaz, K. (2000). Grounded theory: Objectivist and construc-
(Eds.), The craft of qualitative research: A handbook. Toronto: tivist methods. In N. Denzin & Y. Lincoln (Eds.), Handbook
Canadian Scholars. of Qualitative Research (2nd ed.). Thousand Oaks, CA: SAGE.
Stanley, L., & Temple, B. (1995). Doing the business? Evaluating Childress, C., Calonga, A., & Schneiderhan, E. (2020). Beyond
software packages to aid the analysis of qualitative data sets. triangulation: Reconstructing Mandela’s writing life through
Studies in Qualitative Methodology, 5, 169–97. propulsive facilitation at the archive. Qualitative Sociology,
Strauss, A. (1987). Qualitative analysis for social scientists. New 43, 367–84.
York: Cambridge University Press. Deacon, D., Bryman, A., & Fenton, N. (1998). Collision or col-
Strauss, A., & Corbin, J. (1990). Basics of qualitative research: lusion? A discussion of the unplanned triangulation of
Grounded theory procedures and techniques. Newbury Park, quantitative and qualitative research methods. International
CA: SAGE. Journal of Social Research Methodology, 1, 47–63.
Strauss, A., & Corbin, J. (1998). Basics of qualitative research: Denzin, N. (1970). The research act in sociology. Chicago: Aldine.
Techniques and procedures for developing grounded theory. Fenton, N., Bryman, A., & Deacon, D. (1998). Mediating social
Thousand Oaks, CA: SAGE. science. London: SAGE.
Tastsoglou, E., & Miedema, B. (2003). Immigrant women and Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy
community development in the Canadian Maritimes: fails. New York: Harper & Row.
Outsiders within? Canadian Journal of Sociology, 28, 203–34. Finch, J. (1985). “It’s great to have someone to talk to”: The ethics
Weaver, A., & Atkinson, P. (1995). Microcomputing and qualita- and politics of interviewing women. In C. Bell & H. Roberts
tive data analysis. Aldershot, UK: Avebury. (Eds.), Social researching: Politics, problems, practice. London:
Whitehead, K. (2010). “Hunger hurts but starving works”: A case Routledge & Kegan Paul.
study of gendered practices in the online pro-eating-disorder Gabriel, Y. (1998). The use of stories. In G. Symon & C. Cassell
community. Canadian Journal of Sociology, 35, 595–626. (Eds.), Qualitative methods and analysis in organizational
Whitehead, P., & Carpenter, D. (1999). Explaining unsafe sexual ­research. London: SAGE.
behaviour: Cultural definitions and health in the military. Gephart, R. (1988). Ethnostatistics: Qualitative foundations for
Culture, Health, and Sexuality, 1, 303–15. quantitative research. Newbury Park, CA: SAGE.
Whiting, J., Parker, T., & Houghtaling, A. (2014). Explanations Gilbert, G., & Mulkay, M. (1984). Opening Pandora’s box: A socio-
of a violent relationship: The male perpetrator’s perspective. logical analysis of scientists’ discourse. Cambridge: Cambridge
Journal of Family Violence, 29, 277–86. University Press.
Glaser, B. (1992). Basics of grounded theory analysis. Mill Valley,
CA: Sociology Press.
Chapter 14
Glock, C. (1988). Reflections on doing survey research. In H.
Aurini, J., Missaghian, R., & Pizarro Milian, R. (2020). O’Gorman (Ed.), Surveying social life. Middletown, CT:
Educational status hierarchies, after-school activities, Wesleyan University Press.
and parenting logics: Lessons from Canada. Sociology of Gorski, P. (2013). What is critical realism? And why should you
Education, 93, 173–89. care? Contemporary Sociology, 42, 658–70.
Barnard, M., & Frischer, M. (1995). Combining quantitative Hammersley, M. (1996). The relationship between qualitative and
and qualitative approaches: Researching HIV-related risk quantitative research: Paradigm loyalty versus methodologi-
behaviours among drug injectors. Addiction Research, 2, cal eclecticism. In J. Richardson (Ed.), Handbook of research
351–62. methods for psychology and the social sciences. Leicester, UK:
Barron, I. (2013). The potential and challenges of critical realist BPS Books.
ethnography. International Journal of Research and Method Hodson, R. (1996). Dignity in the workplace under participative
in Education, 36, 117–30. management. American Sociological Review, 61, 719–38.
Bell, E. (2007). Separatism and quasi-separatism in Alberta. Hodson, R. (1999). Analyzing documentary accounts. Thousand
Prairie Forum, 32, 335–55. Oaks, CA: SAGE.
Bell, E., Kandler, C., & Riemann, R. (2018). Genetic and envi- Hoye, B. (2020, September 23). Domestic violence research
ronmental influences on socio-political attitudes: Addressing to shed light on heightened risks facing survivors amid
some gaps in the new paradigm. Politics and the Life Sciences, pandemic. CBC News. [Link]
37, 236–49. manitoba/intimate-partner-violence-pandemic-manito-
Best, J. (2020). COVID-19 and numeracy: How about them num- ba-research-1.5735289
bers?. Numeracy, 13, Article 4. Hughes, K., MacKintosh, A.M., Hastings, G., Wheeler, C.,
Bhaskar, R. (1975). A realist theory of science. London: Verso. Watson, J., & Inglis, J. (1997). Young people, alcohol, and
Bhaskar, R. (1979). The possibility of naturalism: A philosoph- designer drinks: A quantitative and qualitative study. British
ical critique of the contemporary human science. Atlantic Medical Journal, 314, 414–18.
Highlands, NJ: Humanities Press. Jamieson, J. (2000). Negotiating danger in fieldwork on crime: A re-
Bhaskar, R. (1986). Scientific realism and human emancipation. searcher’s tale. In G. Lee-Treweek & S. Linkogle (Eds.), Danger
Abingdon, UK: Routledge. in the field: Risk and ethics in social research. London: Routledge.

bel65796_ref_412-429 428 12/11/21 05:21 PM

 References 429

Johnson, J., Bottorff, J., Moffat, B., Ratner, P., Shoveller, J., & Silverman, D. (1985). Qualitative methodology and sociology:
Lovato, C. (2003). Tobacco dependence: Adolescents’ per- Describing the social world. Aldershot, UK: Gower.
spectives on the need to smoke. Social Science and Medicine, Strauss, A., & Corbin, J. (1998). Basics of qualitative research:
56, 1481–92. Techniques and procedures for developing grounded theory.
Kuhn, T. (1970). The structure of scientific revolutions (2nd ed.). Thousand Oaks, CA: SAGE.
Chicago: University of Chicago Press. Tastsoglou, E., & Miedema, B. (2003). Immigrant women and
Lantz, P., & Booth, K. (1998). The social construction of the breast community development in the Canadian Maritimes:
cancer epidemic. Social Science and Medicine, 46, 907–18. Outsiders within? Canadian Journal of Sociology, 28, 203–34.
MacKinnon, N., & Luke, A. (2002). Changes in identity attitudes Timans, R., Wouters, P., & Heilbron, J. (2019). Mixed methods
as reflections of social and cultural change. Canadian Journal research: What it is and what it could be. Theory and Society,
of Sociology, 27, 299–338. 48, 193–216.
Mason, J. (1994). Linking qualitative and quantitative data anal- Wajcman, J., & Martin, B. (2002). Narratives of identity in
ysis. In A. Bryman & R. Burgess (Eds.), Analyzing qualitative modern management. Sociology, 36, 985–1002.
data. London: Routledge. Walklate, S. (2000). Researching victims. In R. King & E. Wincup
McAleese, S., & Kilty, J.M. (2019). Stories matter: Reaffirming (Eds.), Doing research on crime and justice. Oxford: Oxford
the value of qualitative research. The Qualitative Report, 24, University Press.
822–45. Webb, E.J., Campbell, D.T., Schwartz, R.D., & Sechrest, L. (1966).
Milkman, R. (1997). Farewell to the factory: Auto workers in the Unobtrusive measures: Non-reactive measures in the social
late twentieth century. Los Angeles: University of California sciences. Chicago: Rand McNally.
Press. Wilson, B. (2002). The Canadian rave scene and five theses on
Miller, R.L. (2000). Researching life stories and family histories. youth resistance. Canadian Journal of Sociology, 27, 373–412.
London: SAGE. Wilson, E.O. (1998). Consilience: The unity of knowledge. New
Morgan, D. (1998). Practical strategies for combining qualitative York: Knopf.
and quantitative methods: applications for health research.
Qualitative Health Research, 8, 362–76. Chapter 15
Nowicki, E., Brown, J., & Stepien, M. (2014). Children’s struc-
tured conceptualizations of their beliefs on the causes of Barkham, P., & Jenkins, R. (2002, December 13). Fears for fresher
learning difficulties. Journal of Mixed Methods Research, 8, who vanished on mission to talk to the homeless. The Times
69–82. [London].
Phelan, P. (1987). Comparability of qualitative and quantitative Becker, H. (1986). Writing for social scientists: How to start and
methods: Studying child sexual abuse in America. Education finish your thesis, book, or article. Chicago: University of
and Urban Society, 20, 35–41. Chicago Press.
Phoenix, A. (1994). Practising feminist research: The intersection Craig, G. (2004). Managing safety in policy research. In S. Becker
of gender and “race” in the research process. In M. Maynard & A. Bryman (Eds.), Understanding research for social policy
& J. Purvis (Eds.), Researching women’s lives from a feminist and practice: Themes, methods, and approaches. Bristol:
perspective. London: Taylor & Francis. Policy Press.
Platt, J. (1986). Functionalism and the survey: The relation of Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy
theory and method. Sociological Review, 34, 501–36. fails. New York: Harper & Row.
Platt, J. (1996). A history of sociological research methods in Golden-Biddle, K. (1997). Composing qualitative research.
America 1920–1960. Cambridge: Cambridge University Press. Thousand Oaks, CA: SAGE.
Porter, S. (1993). Critical realist ethnography: The case of racism Golden-Biddle, K., & Locke, K. (1993). Appealing work: An in-
and professionalism in a medical setting. Sociology, 27, vestigation of how ethnographic texts convince. Organization
591–609. Science, 4, 595–616.
Porter, S. (2002). Critical realist ethnography. In T. May (Ed.), Lankshear, G. (2000). Bacteria and babies: A personal reflection
Qualitative research in action. London: SAGE. on researcher’s risk in a hospital. In G. Lee-Treweek & S.
Robinson, J., Shaver, P., & Wrightsman, L. (1999). Measures of po- Linkogle (Eds.), Danger in the field: Risk and ethics in social
litical attitudes. Toronto: Academic Press. research. London: Routledge.
Ross, L., O’Gorman, L., MacLeod, M., Bauer, G., MacKay, J., & Lofland, J., & Lofland, L. (1995). Analyzing social settings: A guide
Robinson, M. (2016). Bisexuality, poverty and mental health: to qualitative observation and analysis (3rd ed.). Belmont, CA:
A mixed methods analysis. Social Science & Medicine, 156, Wadsworth.
64–72. Marx, G.T. (1997). Of methods and manners for aspiring sociolo-
Sanscartier, M.D. (2020). The craft attitude: Navigating mess in gists: 37 moral imperatives. American Sociologist, 28, 102–25.
mixed methods research. Journal of Mixed Methods Research, Small, W., Maher, L., & Kerr, T. (2014). Institutional ethical
14, 47–62. review and ethnographic research involving injection drug
Schrøder, K.C. (1999). The best of both worlds? Media audience users: A case study. Social Science & Medicine, 104, 157–62.
research between rival paradigms. In P. Alasuutari (Ed.), Wolcott, H. (1990). Writing up qualitative research. Newbury
Rethinking the media audience. London: SAGE. Park, CA: SAGE.

bel65796_ref_412-429 429 12/11/21 05:21 PM

Index
absolute sample size, 160, 162 bio-medical model, 66
abstract, 368 bivariate data analysis: comparing means and eta in, 192; con-
access: to closed settings, 229–30, 232; ethnography and, tingency table in, 187, 188t; Cramer’s V and, 192; explained
227–33; ongoing, 231–32, 231b; to open settings, 230–31; variance in, 192; Kendall’s tau-b and, 191–92; methods of,
to research, 22 188f; Pearson’s r and, 188–89; quantitative data analysis and,
acknowledgments, 367 187–92; scatter diagrams in, 189–91, 190f, 191f
acquisition of knowledge, 54 Blum, Ben, 70–71
active audience, 299–300 Blumer, H., 210, 215
Adams, T. L., 20, 21 Booth, K., 16b
ADHD. See attention deficit hyperactivity disorder Browne, A. J., 239
adjacency pairs, 303, 303b Bryman, A., 84
advisors, 361 Bryman et al., 323b
Ajodhia-Andrews, A., 262 Bullfighter’s Braid (Pink), 239, 240
ambiguous terms, 106–7 Bulmer, M, 81
analysis. See specific types bus-riding behaviour study, 237
analysis of variance test, 196
analytic induction, 315–16, 337 CA. See conversation analysis
anecdotalism, 344–45 Cambridge–Summerville Youth Study, 68
animosity, 236b Campbell, M., 34–35, 242b
annoyance, 236b Canadian Sociological Association, 55
anomie theory, 5 Cancer television program study, 307b
anonymity, 61 Cao, L., 201
appendices, 372 CAPI. See computer-assisted personal interviews
Atkinson, P., 326, 329 CAQDAS. See computer-assisted qualitative data analysis
attention deficit hyperactivity disorder (ADHD), 144b Carroll, Lewis, 30
attitudes, 104, 201, 338 case study design, 31, 45b, 47–49, 50t
audience, 299–300 causal inferences, 42, 47–48
audio recordings, 237, 258–60 causality, 90, 189
auditing, 214–15 causation, 30–31, 36–37
authenticity, 213, 283, 284, 287–90 census data, 154b
autobiographies, 283–84 central tendency measures, 186–87
axial coding, 317 change agents, 145b
Charmaz, K., 317, 320, 321
balanced answers, 109 Charter of Rights and Freedoms, 55
Banks, C., 68–69, 70–71 children, 61, 144b, 352b
Barber, K., 235b chi-square test, 194–96, 196t
bar chart, 382–83, 384f Cicourel, A., 94
Becker, H., 166–67, 167b Clairborn, W., 36
behaviour, 11–12, 237, 244; gap, 95b, 141b; “helping” or “being classic experimental design, 33–36, 33f
nice” to researchers and, 146; intentions behind, 149; mean- closed questions, 102–3, 104t, 105b, 109–10
ing compared to, 341; structured observation of, 142–43 Clow, K., 92–93
behavioural genetic research, 38 cluster analysis, 83
beliefs, 104, 106 cluster sampling, 159–60
Bell, E., 4, 236b Code of Ethics (Canadian Sociological Association), 55
Bell et al., 43, 116 codes, 84–85
Benoit, C., 16b coding, 80; axial, 317; basic principles of, 85–86; Cohen’s kappa
Bensman, J., 63 and, 145b; content analysis and, 293–99; errors and, 297,
bias, 19–20, 141b, 155 297f; in grounded theory, 316, 317–18; manual, 293, 294–97,
Bierman, Alex, 135 296f; nodes and, 395–97, 395f, 396f, 397f; with NVivo,
big data, 195 394–99; open, 317; open questions, 85b, 102; pitfalls in,
Bikos, Lesley, 10 297; post-coding, 85; pre-coding, 179; without pre-­existing
biographical method, 283–84 scheme, 298–99; problems with, 326–28; qualitative data

bel65796_idx_430-446 430 12/24/21 08:23 AM

 Index 431

analysis and, 316–18, 322–28; reliability, 297; schedule, 294, Cook, T., 34–35
295f, 297f; schemes, 297; selective, 317; steps and consider- “Cornerville” study, 145b
ations in, 323–24; stripes, 397–98, 398f, 399f; of structured correlation, 82; causation and, 30, 31; statistical significance
observation, 142f; turning data into fragments, 324–26; un- and, 194
structured data, 84–85; variables, 293–94, 380–81 covert ethnography, 227–29, 228b, 228f, 229b
coefficients, 199–200, 200t covert research, 63–64
Coffey, A., 326, 329 COVID-19, 46, 87, 96, 345; social construction of numbers and,
Cohen’s kappa, 145b 339b; social trust and, 148
cohort study, 44–45 Cramer, D., 84, 87
common-sense thinking, 11–12 Cramer’s V, 192
comparative design, 49, 49b credibility, 213, 283, 284–85, 287–90
comparison, 40–41, 45, 49, 172t, 326, 339b crime statistics, 88–89, 88b, 89f, 214
complementarity, 347 critical case, 48
complete observer, 234 critical discourse analysis, 208, 306–8
complete participants, 233–34 critical realism, 309, 337b
computer-assisted personal interviews (CAPI), 118 critical realist ethnography, 337b
computer-assisted qualitative data analysis (CAQDAS), 324, critical social science, 11–15
326, 327b, 403 Cronbach’s alpha, 86–87, 145b
concepts: definition of, 81; definitive, 210; dimensions of, 82–83; cross-cultural research, 49b; secondary analysis and, 131–32
emerging, 218b; indicators and, 82–84; multiple-indicator cross-sectional design, 41, 42–44, 44f, 50t
measures of, 82–83; operational definition of, 82; in quan- Curasi, C., 119
titative research, 81–86; self-concept, 12; sensitizing, 210; curriculum vitae (CV), 299
variables and, 81
concurrent validity, 87 DA. See discourse analysis
confidentiality, 61, 63–66, 125b Darwinian theory, 30
confirmability, 213, 214–15 data, 3, 8; big, 195; census, 154b; complexity of, 133, 149–50;
conformity, 270–71 in cross-sectional design, 44, 44f; existing data sets, 129b;
Connolly, A., 66 familiarity with, 133; on internet, 62b, 290–91; missing, 179;
connotative meaning, 299 official statistics and, 134–35; photographs as, 286b; in qual-
conscious partiality, 20 itative research, 208–9; quality of, 129, 133; in quantitative
consent, 56–57, 58f, 60–61, 63–64 research, 42; in questionnaires, 123; in research process, 76;
consilience, 43 secondary analysis of survey, 128–35; turning data into frag-
consistency, 87, 144 ments, 324–26; unstructured, 84–85
constant comparison, 326 data analysis: bivariate, 187–92; multivariate, 134, 197–201;
constructionism, 15, 15b, 17, 336, 339, 339b qualitative, 314–31; quantitative, 177–204; statistics and,
construct validity, 88 192–96; univariate, 184–87. See also qualitative data analy-
content analysis, 208; advantages of, 310; audience and readers of, sis; quantitative data analysis
299–300; checklist for, 292; coding and, 293–99; conversation databases, 364
analysis and, 300–304; disadvantages of, 310–11; discourse dates, 171
analysis and, 300–301, 304–9; ethnographic, 298; government Deacon et al., 347–48
documents and, 287–88, 288b; hermeneutics and, 299; intro- debriefing, 59
duction to, 282–83; language and, 300–309; mass media out- deception, 36; as controversial issue, 60, 147, 149
puts and, 289–90; official documents and, 288–89; overview deduction process, 8, 8f
of, 281–82; personal documents and, 283–87, 284b, 285b, 286b; deductive approaches, 2, 7–10, 8f, 79, 337
qualitative, 339; review of, 311–13; sampling, 171; semiotics definition, 4
and, 298–99, 300b; of subjects and themes, 293; value positions definitive concepts, 210
and, 293; virtual outputs and internet as object of, 290–92; Demaiter, E. I., 20, 21
what things need analyzed?, 292–93; of words, 292–93 demand characteristics, 70–71
contingency table, 164; in bivariate data analysis, 187, 188t; in Demers, A., 38–39
multivariate data analysis, 198, 198t; in SPSS software, 387, demystification, 286–87
387f, 388t, 389f, 391 denotative meaning, 299
control group, 32, 33–34 dependability, 213, 214
convenience sampling, 165–66, 166b dependent variable, 32, 81, 187–88, 350
convenience store study, 90b deprivation-compensation theory, 9b
convergent validity, 88–89, 88b description, 213–14, 216; theory and, 4
conversation analysis (CA): adjacency pairs and, 303, 303b; Desroches, F., 63–64
assumptions of, 301–2; basic characteristics of conversation deterministic statements, 5
and, 302–4; final note on, 304; language and, 208, 300–304, deviant activities, 275
337; preference organization and, 303–4, 304b; transcription diagrams: scatter, 189–91, 190f, 191f, 389; univariate data analy-
and, 302; turn-taking in, 302–3 sis, 185–86, 186f

bel65796_idx_430-446 431 12/24/21 08:23 AM

432 Index

diaries, 37–38, 127–28, 253b, 283–85, 284b, 285b evaluation research, 38–41, 40b
dichotomy, 179 exhaustive categories, 297
direct questions, 256 exhaustive rule, 185
discourse analysis (DA): critical, 208, 306–8; facts and, 305–6, existing data sets, 129b
307b; four themes in, 306b; language and, 300–301, 304–9; existing questions, 116
overview of, 304–5; reading detail in, 306; rhetorical detail experimental group, 33
in, 306; variation in, 305–6 experimental research, 50t, 80, 341; classic, 33–36, 33f; designs,
Disney, Walt, 284, 287 32–41, 50t, 54–55; double blind, 145b; ethics in, 32, 36,
Disneyland, 300b 54–55, 59–60; evaluation research and, 38–41, 40b; labora-
dispersion measures, 187 tory experiments and, 36–37, 38b; manipulation and, 32–33;
Ditton, J., 228b quasi-experiments and, 37–40, 39b; reactive effects in, 59,
division of labour, 15 145b; significance of, 40–41
double-barrelled questions, 107 experimental treatment, 34
double blind experiments, 145b explained variance, 192
Dryburgh, H., 7 external validity, 35, 42, 47–48, 80, 91, 170
Durkheim, E., 19 extreme case, 48

ECA. See ethnographic content analysis face validity, 87

ecological fallacy, 133 facilitation, 347, 353
Edin, K., 96 factor analysis, 83
education, 269–70, 352b facts, 19, 305–6, 307b, 339b
Edwards, N., 245b factual questions, 104
elaboration, 197. See also multivariate data analysis fairness, 70
element, 154b Father parent involvement study, 120
Ellis et al., 18 Fedoroff, J., 85
email interviews, 118–19. See also online interviews feedback, 366
email surveys, 124. See also online social surveys field experiments, 33; as structured observation, 147–49
embarrassing questions, 141b field notes, 236–38, 315
embedded methods argument, 345–46 findings, in research process, 76
emerging concepts, 218b “First do no harm” principle, 54
emotional labour, 6, 212b, 218b Fisher, R., 38, 40
emotions, 90b flexibility, 217–18, 251, 275; in qualitative interviews, 260–61
empathetic understanding, 12 focus groups, 116, 208; composition of groups in, 265t; conduct-
empiricism, 10–11 ing, 263, 263b; group interaction in, 268–69, 270; introduc-
empiricist overtones, 336–37 tion to, 262; limitations of, 270–71; moderator involvement
ending questions, 257 in, 264, 267; number of, 264, 266; online, 271–72, 273b, 274b;
Enquête ouvrière, 107 qualitative interviews and, 262–73, 274b; questions for, 264,
epistemological argument, 346 266b, 267; recording and transcription, 267, 270; size and
epistemological assumptions, 79 participant selection for, 264
epistemology, 299, 335–36; epistemological considerations, Foddy, W., 85b
10–15; interpretivism and, 11–12; positivism and, 10–11; follow-up questions, 256
quantitative/qualitative divide and, 340; social research and, food insecurity study, 325
12–15 Forster, N., 289
errors: coding and, 297, 297f; error variance, 196; sampling, Foster, J., 13b
154b, 155–56, 155f, 156f, 161b, 162, 192–94; standard error of Frank, Lesley, 325
the mean, 161b; statistical significance, 193–94; Type I and frequency tables, 184–85, 185t, 381–82, 382f
Type II, 193–94, 194f
eta, 192 Gans, H. J., 60
ethnographic content analysis (ECA), 298 gatekeepers, 22, 230
ethnography, 13b, 24, 31, 207–8, 217, 336; access and, 227–33; gender gap, 290
active compared to passive, 234, 235b; advantages of, 273–76; generality, 350
consent in, 57; covert, 227–29, 228b, 228f, 229b; critical re- generalizations, 8, 10; of case study design generalizability,
alist, 337b; end point in, 244, 245b; field notes in, 236–38; 48–49; limits to, 170–71; in qualitative research, 220–21; in
institutional, 240–41, 242b; introduction to, 226–27; key quantitative research, 91–92; from random sample, 161b
informants in, 233; meta-ethnography and, 343; micro- General Social Survey (GSS), 88, 168; features of, 130b; second-
ethnography, 227; overt, 227–29, 228f; overview of, 225–26; ary analysis and, 129–30, 130b
qualitative interviews compared to, 273–77; research ethics genetic research, 38
in, 55, 244; review of, 246–48; roles for ethnographers, 233– Gidden, A., 5b
36, 233f; sampling in, 242–44; virtual, 235b; visual, 238–40 gig economy study, 135
ethnostatistics, 343 Glavin, Paul, 135

bel65796_idx_430-446 432 12/24/21 08:23 AM

 Index 433

Goffman, E., 6 inclusiveness, 67

Golden-Biddle, K., 370b incomplete questionnaires, 123
government documents, 287–88, 288b independent variable, 32, 80, 187–88, 350
government research, 39 indexicality, 301
Goyder et al., 44, 93 indicators, 82–84
grand theories, 5 Indigenous people, 14–15, 39–40, 80, 201, 219b, 231b, 239; social
Green, A. I., 211 capital and, 269–70
grounded theory, 8, 209–10, 337; basic features of, 316; cat- indirect questions, 256
egories in, 319b; coding in, 316, 317–18; counterintuitive induction process, 8
results from, 322b; criticisms of, 321–22; defining, 316; inductive approaches, 2, 7–10, 8f, 207
memos and, 316, 320–21, 323b; outcomes of, 318–20, inferential statistics, 192–96
320f; phases of, 318–19; qualitative data analysis and, informants, 233
316–22, 325 information sheet, 56, 57f
GSS. See General Social Survey informed consent, 57, 60–61
“inheritance, ” 323b,
Hallett, R., 235b initial open-ended questions, 257
Hallgrimsdottir et al., 16b institutional ethnography, 240–41, 242b
Haney, C., 68–69, 70–71 institution requirements, 360
“hanging around, ” 230–31 instrumentation, 34
Hanser, Amy, 237 interactional analysis, 328
Hanson et al., 39 interactionism, symbolic, 12, 95
harm, 59, 64; defining, 54–55, 67; “do no harm principle” and, inter-interviewer variability, 102
54; justice and, 67–68, 69 intermediate questions, 257
Hawthorne effect, 145b internal reliability, 86–87
Hay, D., 9b, 84, 84b internal validity, 90
“helping” researcher, 145b International Social Survey Programme (ISSP), 132
Henriquez, Nadine, 345 internet: communication, 291; data on, 62b, 290–91; as object of
hermeneutics, 299 content analysis, 290–92; online focus groups, 271–72, 273b,
heterogeneity of population, 164 274b; online interviews, 118–19, 271–72, 274b; online opt-in
Hier, S., 18 survey panel, 168–69; online social surveys, 124–26, 125b,
“Highway of Tears,” 309 178–79, 273b
histogram, 186, 186f, 385–86 inter-observer consistency, 87, 144
history, 34–35 interpreting questions, 257
Hobbs, D., 235b interpretivism, 11–12, 207, 338
Hochschild, A., 6, 212b, 218b intersubjectivity, 11
Homan, R., 57–59 inter-textual coherence, 370b
horizontal questionnaire arrangement, 126, 126b interval/ratio variables, 181, 184t
Hughes et al., 347 intervening variables, 197
Humphreys, L., 63–64 interview guide, 254–55, 258
hypotheses, 11, 80, 337, 348; null, 193–96 interviews, 10; alternatives to in-person, 116–18; assessing inter-
viewers for, 122; computer-assisted, 118; conducting, 119–22;
IBM SPSS Statistics (SPSS) software, 184–85, 186f, 191t, 198–99, formats for, 116–19; inter-interviewer variability in, 102;
199t, 294, 326; basic operations in, 375b; comparing means interviewer effects, 123; intra-interviewer variability in, 102;
in, 389, 391f; computing new variables in, 379–80, 379f; for introducing research in, 119; introductory statement and,
data analysis, 381–91; entering data in Data View, 375–77, 120–21; multiple interviewers or interviewees, 116; online
376f, 377f; generating arithmetic mean, median, standard or email, 118–19, 271–72, 274b; oral and life history, 250,
deviation, and range with, 386; generating bar chart in, 253b; probing in, 121; prompting in, 122, 258; in qualitative
382–83, 384f; generating contingency table in, 387, 387f, research, 208, 249–80; questionnaires compared to, 122–24;
388t, 389f, 391; generating frequency table in, 381–82, 382f; rapport in, 119–20, 272; response rates for, 172–73; schedule
generating histogram in, 385–86; generating Pearson’s r in, for, 31, 80, 119, 141, 251; semi-structured, 5b, 251–54, 252b,
389; generating pie chart in, 383–85, 385f; generating scatter 259b, 336; show cards in, 118b, 121–22; in simple random
diagrams in, 389; introduction to, 374; printing output in, sample, 156–58, 178–79; structured, 41, 153, 217, 251, 297,
386, 386t; quantitative data analysis with, 374–92; questions 336; in survey research, 102, 116–22; telephone, 116–17; tran-
for review, 391–92; recoding variables in, 380–81, 380f, 381f; scribed, 43–44; unstructured, 43–44, 208, 251–62. See also
regression in, 391, 391f; retrieving data in, 379; saving data qualitative interviews
in, 378–79; variables in, 377–80, 378f intra-interviewer variability, 102
idealization, 286–87 intra-observer consistency, 144
illness narratives, 329–30 introducing questions, 255
illustrations, 286b introductory statement, 120–21
immigration, 4, 18, 20 ISSP. See International Social Survey Programme

bel65796_idx_430-446 433 12/24/21 08:23 AM

434 Index

Jacobson study, 33–36 Martin, B., 351b

Jagger, E., 294b Marx, G. T., 362b
justice: fairness and, 70; harm and, 67–68, 69; research ethics Marx, Karl, 13, 107
and, 67–70; social, 13–15, 61–66, 240, 241b Marxism, 13
mass media outputs, 289–90
Kefalas, M, 96 maturation, 34–35
Kelly, L., 218b Mead, George Herbert, 12
Kendall’s tau-b, 191–92 Mead, Margaret, 48
Kennedy et al., 83 mean, 186–87; comparing subgroup, 192–94, 193t; eta compared
key informants, 233 to, 192; in SPSS software, 386, 389, 391f; standard error of
Kimmel, A., 68 the mean, 161b; statistical significance compared with, 196
Klasse an sich, 13 meaning, 283, 285; of actions, 11–12; behaviour compared to,
knowledge: acquisition of, 54; questions about, 106; requisite, of 341; denotative and connotative, 299; problems of, 141b, 149
respondents, 109 measured variation, 103, 103f
Kvale’s ten traits of effective interviewer, 255, 256b measurement: aiding, 348; central tendency measures, 186–87;
dispersion, 187; reactive measurement effect and, 145b;
laboratory experiments, 36–37, 38b reasons for, 81–82; testing and, 210; unobtrusive measures,
labour, 351b; division of, 15; emotional, 6, 212b, 218b; gig econ- 145b, 347
omy study and, 135; women in male-dominated occupa- measurement validity, 35, 47–48, 339b; causality and, 90; con-
tions, 21 current validity, 87; consistency and, 86; construct validity,
Lamb, Augustus, 284b 88; convergent validity, 88–89, 88b; face validity, 87; of of-
language: content analysis and, 300–309; conversation analysis ficial statistics, 134–35; in quantitative research, 86–90, 94;
and, 208, 300–304, 337; discourse analysis and, 300–301, reflections on, 89; reliability and, 82; of structured observa-
304–9; learning, 275; prejudicial, 366, 368; words and, 292– tion, 144, 146
93, 341; in writing up research, 366, 368 media: on gender gap, 290; mass media outputs, 289–90; sam-
Lantz, P., 16b pling, 171; social, 288b; study on crime in, 214; study on
LaPiere, R. T., 95b violence and, 266b
latent content, 293 median, 186, 386
law, 30, 65 medical research, 66, 67, 69
Layder, D., 64 memory: problems of, 141b, 236; questions stretching, 109–10
“layering,” 344 memos, 238, 316, 320–21, 323b; in NVivo, 402, 402f
leadership styles, 141 mental health studies, 120, 144b, 214, 322b, 354–55; on mental
leading questions, 108 health services, 170; natural disasters and, 318
learning difficulties, 352b Merton, R., 5
Lee-Treweek, G., 236b meta-ethnography, 343
Lehmann, W., 267–68 methodology, 69
letter of information, 56, 57f micro-ethnography, 227
letters, 283–84 Milgram, S., 4, 59–60
Lewin, K., 13 missing data, non-response, in sampling, 154b, 162–64,
life history approach, 216–17, 283–84, 328, 338 171–74, 179
life history interviews, 250, 253b mode, 186
Likert, Rensis, 83b morality, 11. See also research ethics
Likert items, 82, 83b, 84b mortality, 34
Likert scale, 83b, 87, 102 Morton, K., 309
literature review: in research process, 75; in research project, motivation, 4, 73
363–64, 370b; writing up research and, 368–69, 370b multiple-indicator measures, 82–83
Locke, K., 370b multiple linear regression, 198–201, 200t
logic of comparison, 40–41 multiple regression analysis, 201
longitudinal research, 276; case study design and, 45b; designs, multi-stage cluster sampling, 159–60
44–47, 45b, 46b, 50t; forms and types of, 44–45; qualitative multi-strategy research: approaches to, 346–53, 349f; arguments
approaches in, 46; secondary analysis and, 129–30 against, 345–46; facilitation in, 353; filling in gaps with,
long questions, 107 349; generality in, 350; quantitative/qualitative divide and,
love study, 294b 336, 345–55; questions in, 352–53; reflections on, 353–54;
researcher and participant perspectives in, 349–50; static
Ma, Y., 38, 40 and process features of, 349; for studying different aspects of
Mackenzie King, William Lyon, 285b phenomenon, 350–53
Maillé, Marie-Ève, 65 multivariate data analysis, 134; contingency table in, 198, 198t;
manifest content, 293 intervening variables in, 197; other uses for, 198–200; quan-
manipulation: in cross-sectional designs, 41, 42–44; experimen- titative data analysis and, 197–201; relationships in, 197–98
tal designs and, 32–33; of variables, 32, 42–44 Murphy, L., 85

bel65796_idx_430-446 434 12/24/21 08:23 AM

 Index 435

mutual analysis, 342–43 outliers, 187

mutually exclusive rule, 185 Out of Milk (Frank), 325
overlapping answers, 109
narrative analysis, 328–30, 329b overt ethnography, 227–29, 228f
National Longitudinal Survey of Children and Youth (NLSCY),
45, 46b panel conditioning, 46
natural disasters, 318 panel study, 44–45
“natural” experiments, 37–38 PAR. See participatory action research
naturalistic approach, 207, 275–76, 301, 336–38, 342 paradigm argument, 346
natural portrayal, 286–87 paradigms, 346
natural science, 12, 336–38 participant-as-observer, 234
negative relationships, 42 participant observation, 13b, 31, 143b, 207–8, 226–27, 336. See
negatives, in questions, 108–9 also ethnography
neo-positivism, 338 participatory action research (PAR), 13–15, 208, 219b, 240
Nixon, Kendra, 345 passive audience, 299–300
NLSCY. See National Longitudinal Survey of Children and Youth passive role, in ethnography, 234, 235b
nodes, 395–97, 395f, 396f, 397f Pearson’s r, 188–89, 389
nominal variables, 179, 184t Peltier, C., 14
nomothetic approach, 30 performative analysis, 328
nomothetic causation, 36–37 personal documents: authenticity of, 284, 287–88; checklist for
non-participant observation, 143b evaluating, 292; content analysis and, 283–87, 284b, 285b,
non-probability sampling types, 164–70 286b; credibility of, 284–85, 287–88; diaries, 37–38, 127–28,
non-response, in sampling, 154b, 162–64, 171–74, 179 253b, 283–85, 284b, 285b; visual objects and photographs,
non-spuriousness, 31 286, 286b
Nowicki et al., 352b personal questions, 104
null hypothesis, 193–96 Phillips, N., 16b
numbers, 339b, 341 photography, 237, 239, 240; personal, 286–87, 286b; representa-
NVivo software, 326, 326b; coding with, 394–99; getting started tiveness and, 287, 288b; in social research, 286b
with, 392–94, 392f, 393f, 394f; introduction to, 392; memos photovoice, 240, 241b
in, 402, 402f; opening existing projects in, 403; output in, physical appearance, 92–93, 294b; attractiveness and,
400; qualitative data analysis with, 392–403; saving projects 38b, 40–41
in, 402–3; searching text in, 399–400, 400f, 401f pie chart, 383–85, 385f
pilot study, 114–16
objectivists, 15 Pink, S., 238–39, 240
objectivity, 20, 79, 96 point of view, 11–12, 349–50
observation: of behaviour, 142–43; classic experimental design police studies, 10, 148–49, 201
and, 34; “helping” or “being nice” to researchers and, 146; politics, 19, 20–22, 43
inter-observer consistency and, 87, 144; intra-observer con- polling, 165
sistency and, 144; issues resistant to, 276; non-participant, polysemy, 299
143b; participant, 13b, 31, 143b, 207–8, 226–27, 336; re- population, 154b; generalizing from random sample to, 161b;
searcher as change agent in, 145b; schedule, 141–42, 142f, heterogeneity of, 164; strata, 168, 168t
297; SSO, 148–49, 150; structured, 41, 80, 139–51, 217, 336; positivism, 79, 81; epistemology and, 10–11; neo-positivism, 338
unobtrusive, 143b; unstructured, 143b postal questionnaires, 217
observer-as-participant, 234 post-coding, 85
O’Connell Davidson, J., 64 post-traumatic stress disorder (PTSD), 214, 355
official documents, from private sources, 288–89 Potter, J., 304–7
official statistics, 13b, 134–35 Potter et al., 307b
online focus groups, 271–72, 273b, 274b praxis, 13
online interviews, 118–19, 271–72, 274b pre-coding, 179
online opt-in survey panel, 168–69 preference organization, 303–4, 304b
online social surveys, 124–26, 125b, 178–79, 273b prejudicial language, 366, 368
ontological assumptions, 79 Presser, S., 85
ontological issues, 2, 15–17, 340 pre-testing, 35–36, 37
open coding, 317 principles, 30
open questions, 85b, 102, 104t, 105b priority decision, 349
operationalization, 80 privacy. See confidentiality
ORACLE study, 143 probabilistic statements, 5
oral history interviews, 250, 253b probability sampling, 91–92, 154, 156–60, 192, 243
ordinal variables, 179, 181, 184t probing, 123; in interviews, 121; questions, 256
Orwell, George, 4 prompting: in interviews, 122, 258; with photographs, 286b

bel65796_idx_430-446 435 12/24/21 08:23 AM

436 Index

propulsive facilitation, 353 quantitative research, 75; approaches, 2, 17–19, 18t, 30–31, 42,
Proudfoot, 267–68 101–2; concepts in, 81–86; critiques of, 94–96; data in, 42;
PTSD. See post-traumatic stress disorder generalization in, 91–92; interpretivism and, 338; introduc-
publication restrictions, 22 tion to, 79; laws and principles in, 30; main goals of, 89–93;
public settings, 232 main steps in, 79–81, 79f; measurement validity in, 86–90,
purposive sampling, 242 94; objectivity and, 96; overview of, 78–79; qualitative
approach to, 342–43; qualitative research combined with,
qualitative content analysis, 339 346–55; qualitative research compared to, 17–19, 18t, 221–22,
qualitative data analysis: analytic induction and, 315–16, 337; 221t, 251; qualitative research facilitating, 348; reliability in,
basic operations in, 322–24; with CAQDAS, 324, 326, 86–89; replicability in, 92–93; research designs and, 30–31;
327b, 403; coding and, 316–18, 322–28; general strategies research ethics in, 55, 80; review of, 97–99
of, 315–30; grounded theory and, 316–22, 325; narrative quantitative sampling. See sampling
analysis and, 328–30, 329b; with NVivo, 392–403; overview quasi-experiments, 37–40, 39b
of, 314–15; quantitative data analysis compared to, 315–16; quasi-quantification, 344
review of, 330–31; turning data into fragments, 324–26 questionnaires, 31, 80; advantages of, 122–23; bad questionnaire
qualitative interviews, 208; advantages of, 276–77; checklist example, 115b; clear instructions in, 126; clear presentation
for, 277; ethnography compared to, 273–77; flexibility in, of, 126; data in, 123; design rules for, 126–28; disadvantages
260–61; focus groups and, 262–73, 274b; interview guide in, of, 123–24; incomplete, 123; interviews compared to, 122–24;
254–55, 258; Kvale’s ten traits of effective, 255, 256b; online, online social surveys and, 124–26, 125b, 178–79; overview
271–72, 274b; overview of, 249–50; questions in, 255–58, of, 122; postal, 217; researcher-driven diaries and, 127;
257f; recording and transcription, 258–61, 259b; reflexivity response rates to, 173–74; sampling and, 153–54, 166b; in
in, 261–62; review of, 278–80; structured interviews com- survey research, 122–28; vertical or horizontal arrangement
pared to, 251; unstructured interviews and, 251–62 of, 126, 126b
qualitative questions, 23–24, 216 questions: ambiguous terms in, 106–7; asking for help with
qualitative research, 75; approaches, 2, 7b, 17–19, 18t, 31, 46; designing questions, 114; about attitudes, 104; balanced
confidentiality in, 63; confirmability in, 213, 214–15; credi- answers to, 109; about beliefs, 104, 106; closed, 102–3, 104t,
bility and, 213; critiques of, 219–21; data in, 208–9; depend- 105b, 109–10; common mistakes in, 112; considerations
ability in, 213, 214; emerging concepts in, 218b; emphasis on when developing, 25b; design rules for, 106–16; direct and
process in, 216–17; evaluation criteria, 210–15; flexibility and indirect, 256; “don’t know” answer use in, 110; double-
limited structure of, 217–18; generalizations in, 220–21; in- barrelled, 107; ending, 257; existing question use, 116;
terviews in, 208, 249–80; introduction to, 207–10; literature, factual, 104; for focus groups, 264, 266b, 267; formulating,
370b; in longitudinal research, 46; main goals of, 215–18; 23–24, 25b, 360, 361f; general rules of thumb for, 106; in-
main steps in, 208–10, 208f; naturalistic approach and, 336– termediate, 257; interpreting, 257; introducing, 255; issues
38; overview of, 206–7; participant familiarity and, 215–16; related to, 22–25; about knowledge, 106; leading, 108; long,
participatory action research and, 219b; quantification in, 107; minimizing technical terms in, 109; in multi-strategy
343–45; quantitative approach to, 343; quantitative research research, 352–53; negatives in, 108–9; open, 85b, 102, 104t,
combined with, 346–55; quantitative research compared to, 105b; ordering of, 110–11, 111b; about others, 106; overlap-
17–19, 18t, 221–22, 221t, 251; quantitative research facilitat- ping answers to, 109; overstretching memories and, 109–10;
ing, 349; quasi-quantification in, 344; replicability of, 220; personal, 104; for pilot study, 114–16; probing, 256; quali-
research designs and, 31; research ethics in, 55, 66; review of, tative, 23–24, 216; in qualitative interviews, 255–58, 257f;
222–24; subjectivity in, 219–20; theory and concepts in, 210; quantitative, 23; requisite knowledge and, 109; research
transferability and, 213–14; transparency in, 221; ultimate designs and, 75; research process, 73–75; response sets and,
goals of, 218; validity of, 212–13; variables and, 350 112; secondary analysis and new, 132–33, 132b; social re-
quantification rhetoric, 305–6, 307b, 343–45; numbers and, search, 22–25; sources of, 362b; specifying, 256; standardiza-
339b, 341 tion of, 103; threatening or embarrassing, 141b; too general,
quantitative data analysis: bivariate data analysis and, 187–92; 107; two or more, 108–9; types of, 104, 106, 256–58; vignette
checklist for, 201; data collection in, 178–79; with IBM SPSS questions and, 113–14, 113b, 258
Statistics, 374–92; multivariate data analysis and, 197–201; quota sampling, 120, 167–69
overview of, 177–78; qualitative data analysis compared to, quota strata, 168, 168t
315–16; review of, 202–4; statistics and, 192–96; univariate
data analysis and, 184–87; variables in, 179–84 race, 4
quantitative/qualitative divide: constructionism and, 339, 339b; Rafaeli, A., 90b, 189b
epistemology and, 340; interpretivism and, 338; introduc- random assignment, 33, 34, 35
tion to, 335–36; multi-strategy research and, 336, 345–55; random-digit dialing, 117
mutual analysis and, 342–43; naturalistic approach and, randomized response technique, 61–63
336–38; ontological issues and, 340; overview of, 334–35; random samples, 47, 161b; simple, 156–58, 178–79; stratified,
problems with, 340–42; quantification, in qualitative re- 158–59, 159t
search, 343–45; research methods and, 340; review of, range, 187, 386
356–58; two positions in debate of, 346 rapport, 119–20, 272
quantitative questions, 23 reactive effects, 36, 59, 145b, 276

bel65796_idx_430-446 436 12/24/21 08:23 AM

 Index 437

reactive measurement effect, 145b safety in, 367b; supervisors and, 361; time management and
reactivity, 229b, 234, 342 resources for, 362–64; writing up, 366–72
reading, 306 respect, 56–61
realism, 309, 337, 337b respondent validation, 213
REB. See research ethics board response rates, 163–64, 171–74
reconstruction of events, 276 response sets, 84, 112, 145b
recording: audio, 237, 258–60; focus groups, 267, 270; incidents, revelatory case, 48
142–43; interviews, 258–61, 259b rhetorical detail, 306
references, 365, 372 Ricciardelli, R., 92–93, 254
reflexivity, 19, 261–62, 301 Riessman, C. K., 326–28
regression, 198–201, 200t, 391, 391f rights, 55–56, 68–69
relational statements, 5 role selection, 145b
relationships: causality and, 189; in multivariate data analysis, Rosenhan, D. L., 147, 149
197–98; negative, 42; spuriousness of, 31, 197, 197f; variables Rosenthal study, 33–36
and, 188–92 Ross et al., 354–55
relative sample size, 160, 162
reliability, 82, 86–89, 143–44, 297 safety, 367b
religiosity, 9b same-sex marriage study, 211
Renick, A.J., 120 sampling, 47; bias in, 155; comparison of different contact meth-
Renzetti, E., 214 ods in, 172t; content analysis, 171; convenience, 165–66,
replicability, 36, 42; of case study design, 47–48; of qualitative 166b; dates, 171; distributions, 162b; error, 154b, 155–56,
research, 220; in quantitative research, 92–93 155f, 156f, 161b, 162, 192–94; in ethnography, 242–44; frame,
representativeness, 35, 283, 285–89, 288b 154b; generalization limits and, 170–71; generalizations
representative sample, 91, 154b, 165 from random sample, 161b; heterogeneity of population in,
research: funding, 20, 22, 60–61, 65–66; information sheet, 56, 164; introduction to, 153–55; media, 171; multi-stage cluster,
57f; orientation, 17–19, 49–50; overview of research process, 159–60; non-probability types of, 164–70; non-response in,
73–76; sponsors, 60–61, 65–66; theory and, 4–7. See also 154b, 162–64, 171–74, 179; overview of, 152–53; polling and,
specific topics 165; probability, 91–92, 154, 156–60, 192, 243; purposive,
research designs: case study design, 31, 45b, 47–49, 50t; choos- 242; quality of samples, 160; quantitative, 152–76; question-
ing, 75; comparative design, 49; cross-sectional design, naires and, 153–54, 166b; quota, 120, 167–69; representative
41–44, 50t; ethics and, 54–55; experimental designs, 32–41, samples, 91, 154b, 165; response rates, 163–64, 171–74;
50t, 54–55; introduction to, 30–31; longitudinal designs, review of, 174–76; simple random sample, 156–58, 178–79;
44–47, 46b, 50t; overview of, 29–30; qualitative approaches sizes, 160, 162–64, 168; snowball, 166–67, 167b, 242–43;
and, 31; quantitative approaches and, 30–31; questions and, standard deviation in, 161b; standard error of the mean in,
75; research orientation and, 49–50; research strategy and, 161b; stratified random sampling, 158–59, 159t; structured
50t; review of, 51–52 observation and, 169–70; systematic sample, 158; terms and
researcher-driven diaries, 127–28 concepts in, 154b; theoretical, 243–44, 243f; time, 143, 169;
research ethics, 3; big data and, 195; conclusions and review types of, 156–60, 164–70; virtual, 174
on, 70–72, 76; consent and, 56–57, 60–61; in covert ethnog- Santin, M., 218b
raphy, 229, 229b; in covert research, 63–64; deception and, scatter diagrams, 189–91, 190f, 191f, 389
36, 60, 147, 149; in ethnography, 55, 244; in experimental Schieman, Scott, 46, 135
research, 32, 36, 54–55, 59–60; general ethical principles, Schlesinger et al., 266
56–70; internet and, 62b; introduction to, 54–56; justice and, Schulenberg, J. L., 235b
67–70; overview of, 53–54; in qualitative research, 55, 66; in Schuman, H., 85
quantitative research, 55, 80; research designs and, 54–55; in science: natural, 12, 336–38; positivism and, 11; of signs, 298–
research process, 76; respect for persons and, 56–61; rights 99; social, 11–15; as “value-free,” 11
and, 55–56, 68–69; welfare and, 61–66 scientific statements, 11
research ethics board (REB), 55–56, 59, 69–70, 367b Scott, J., 283–89, 292
research participants: annoyance or animosity toward, 236b; fa- secondary analysis: advantages of, 129–33; cross-cultural re-
miliarity and seeing through eyes of, 215–16, 273, 275; focus search and, 131–32; ecological fallacy and, 133; existing
groups selection of, 264; observation of, 13b, 143b, 207–8, data sets and, 129b; GSS and, 129–30, 130b; limitations of,
226–27, 336; overidentification with, 235b; representative- 133–34; longitudinal research and, 129–30; new research
ness of, 35, 283, 285–89; underage, 61; volunteers as, 37 questions and, 132–33, 132b; official statistics and, 134–35;
research process: choosing quantitative or qualitative, 75; data subgroup analysis and, 130–31; of survey data, 128–35
in, 76; findings in, 76; literature review in, 75; motivation secularization, 9b, 84
in, 73; overview of, 73–76; questions, 73–75; research ethics selection, 35
in, 76 selective coding, 317
research project: conducting, 359–73; institution requirements self-concept, 12
for, 360; literature review in, 363–64, 370b; overview of, semiotics, 298–99, 300b
359–60; preparation for, 365; question selection in, 360, 361f; semi-structured interviews, 5b, 251–54, 252b, 259b, 336

bel65796_idx_430-446 437 12/24/21 08:23 AM

438 Index

sensitivity, 275 Street Corner Society case study, 230

sensitizing concepts, 210 stress, 214, 229b, 244, 355
sequence decision, 349 structural analysis, 328
settings, effects of, 35 structuration theory, 5, 5b
Shafer, K., 120 structured interviewing, 41, 153, 217, 251, 297, 336
Shaffir, W., 232–33, 245b structured observation, 41, 80, 217, 336; of behaviour, 142–43;
Sharma et al., 219b coding of, 142f; criticisms of, 149–50; field experiments as,
show cards, 118b, 121–22 147–49; introduction to, 140; measurement validity of, 144,
signifiers, 299 146; observation schedule and, 141–42, 142f; overview of,
signs, 298–99 139–40; problems with, 140–41; reliability of, 143–44; review
silence, 257 of, 150–51; sampling and, 169–70; SSO and, 148–49, 150;
simple random sample, 156–58, 178–79 strategies, 142–43; survey research and, 140–41
single blind experiments, 145b subgroup analysis, 130–31
Smith, Dorothy, 5 subgroup means, 192–94, 193t
snowball sampling, 166–67, 167b, 242–43 subjectivity, 11, 219–20
social capital, 269–70 substance abuse, 170
social class, 267–68 suicide, 5
social desirability, 112, 141b supervisors, 361
social facts, 19 survey research: closed questions in, 102–3, 104t; GSS, 88,
social justice, 13–15, 61–66, 240, 241b 129–30, 130b, 168; interviews in, 102, 116–22; introduction
social media, 288b to, 101–2; from online social surveys, 124–26, 125b, 178–79,
social ontology, 15–17 273b; open questions in, 102, 104t; overview of, 100–101;
social organization, 16–17 problems with, 140–41, 141b; question design rules in,
social patterns, 339b 106–16; questionnaires in, 122–28; question types in, 104–6;
social reality, 15–17 review, 135–38; secondary analysis of survey data, 128–35;
social research: deductive approaches to, 2, 7–10; epistemology steps in, 363f; structured observation and, 140–41
and, 12–15; general orientations, 17–19; inductive ap- Sutton, R. I., 90b, 189b
proaches to, 2, 7–10, 8f, 207; influences on, 19–25, 19f; symbolic interactionism, 12, 95
interactive activities on, 26–27; introduction to, 3–4; motives symbols, 298–99
for, 4, 73; ontological considerations in, 15–17; photography systematic sample, 158
in, 286b; politics in, 19, 20–22; questions, 22–25; review of, systematic social observation (SSO), 148–49, 150
25–26; theory and, 4–7; values and, 19–20
social science, 11–15 table of contents, 368
social trust, 148 talk. See conversation analysis
Sociological Abstracts database, 364 tattoo narratives, 329b
software, 184–85, 324, 326, 327b, 364. See also IBM SPSS TCPS. See Tri-Council Policy Statement on Ethics in Human
Statistics; NVivo software Research
Spearman’s rho, 192 Tearoom Trade (Humphreys), 63
specifying questions, 256 technical argument, 346
split-half method, 87 technical terms, minimizing, 109
sponsored research, 60–61, 65–66 Teevan, J., 7
SPSS. See IBM SPSS Statistics telephone interviews, 116–17
spuriousness, 31, 197, 197f testing, 34–37, 86, 210
SSO. See systematic social observation thematic analysis, 328, 343–44
stability, 86 theoretical sampling, 243–44, 243f
Stall, Nathan, 96 theoretical saturation, 243–44, 316
standard deviation, 161b, 187, 386 theory: common components in, 4–5; definition and, 4; degree
standard error of the mean, 161b of abstraction and, 4–7; description and, 4; generation of,
standpoint theory, 5 317; grounded, 8, 209–10, 316–22, 325, 337; need for, 7b;
Stanford mock prison study, 68–69, 70–71 qualitative research, 210; research and, 4–7; structuration,
statistical significance: chi-square test and, 194–96, 196t; cor- 5, 5b; tested theories versus those emerging from data, 341;
relation and, 194; errors, 193–94; establishing, 193; inferen- theories of middle range, 5; types of, 5. See also specific
tial statistics and, 192–96; mean compared with, 196 theories
statistics: data analysis and, 192–96; ethnostatistics, 343; thick description, 213–14
inferential, 192–96; official, 13b, 134–35. See also IBM SPSS threatening questions, 141b
Statistics software time: management, 362–64; order, 30–31, 34, 42; sampling, 143,
stigma, 6 169
stratified random sampling, 158–59, 159t title page, 367
Strauss, A., 323b Torelli, J., 252b
Strauss et al., 16–17

bel65796_idx_430-446 438 12/24/21 08:23 AM

 Index 439

transcription, 102; conversation analysis and, 302; field notes 188–92; in SPSS software, 377–80, 378f; types of, 181, 184,
and, 237–38; for focus groups, 267, 270; of qualitative inter- 184t
views, 258–61, 259b; transcribed interviews, 43–44 Verstehen, 12
transferability, 213–14 Verstehende sociology, 215
transparency, 221 vertical questionnaire arrangement, 126, 126b
trauma, 59 Vidich, A., 63
triangulation, 346–48, 353 vignette questions, 113–14, 113b, 258
Tri-Council Policy Statement on Ethics in Human Research violence, 214, 266b, 309, 319b, 345
(TCPS), 66; TCPS2, 55–56, 70–71 virtual ethnography, 235b
trust, 148, 213 virtual outputs, 290–92
Turnbull, P., 19–20 virtual sampling, 174
turn-taking, 302–3 visual ethnography, 238–40
twin studies, 38 visual objects, 286–87, 286b
Type I and Type II errors, 193–94, 194f volunteers, 37

underage study participants, 61 Wajcman, J., 351b

unique case, 48 Walsh et al., 36–37
unit, 154b Webb et al., 145b
univariate data analysis, 184–87, 185t, 186f Weber, Max, 12
unobtrusive measures, 145b, 347 websites and web references, 290–92
unobtrusive observation, 143b web surveys, 124–26
unstructured data, 84–85 Weins et al., 170
unstructured interviews, 43–44, 208; overview of, 252b; qualita- welfare, 61–66
tive interviews and, 251–62 Whiting et al., 319
unstructured observation, 143b Wilkinson, Lori, 165
Wilson, B., 18–19
validity, 22, 283; of case study designs, 47–48; external, 35, 42, Wolf, D., 230–31
47–48, 80, 91, 170; face, 87; internal, 90; of qualitative re- Woodhall-Melnik, Julia, 318
search, 212–13. See also measurement validity words, 292–93, 341
value positions, 293 Wright et al., 252b
values, 2, 19–20 writing up research: abstract, 368; acknowledgments, 367; appendi-
Van den Hoonaard, W. C., 66 ces, 372; checklist for, 371–72; conclusion, 370, 372; discussion,
van den Scott, Lisa-Jo, 231b, 237, 245b 370; feedback on, 366; introduction, 368; language in, 366, 368;
variables, 5, 30, 95; categorizing, 184, 184f; coding, 293–94, literature review and, 368–69, 370b; references, 372; research
380–81; concepts and, 81; contingency table for, 164; de- methods, 369; results, 369; starting early, 366; title page, 367
pendent, 32, 81, 187–88, 350; explained variance and, 192;
independent, 32, 80, 187–88, 350; interval/ratio, 181, 184t; youth mental health, 170
intervening, 197; manipulation of, 32, 42–44; nominal, 179, youth shelters, 217b
184t; ordinal, 179, 181, 184t; qualitative research and, 350;
in quantitative data analysis, 179–84; relationships and, Zimbardo, P., 68–69, 70–71

bel65796_idx_430-446 439 12/24/21 08:23 AM

bel65796_idx_430-446 440 12/24/21 08:23 AM
bel65796_idx_430-446 441 12/24/21 08:23 AM
bel65796_idx_430-446 442 12/24/21 08:23 AM
bel65796_idx_430-446 443 12/24/21 08:23 AM
bel65796_idx_430-446 444 12/24/21 08:23 AM
bel65796_idx_430-446 445 12/24/21 08:23 AM
bel65796_idx_430-446 446 12/24/21 08:23 AM