Research Methodology

and Biostatistics
This page intentionally left blank
 Research Methodology
and Biostatistics
A Comprehensive Guide for Health Care Professionals

SURESH K. SHARMA PhD, FCCN, RN

Professor-cum-Principal
College of Nursing
All India Institute of Medical Sciences
Rishikesh, Uttarakhand
RELX India Pvt. Ltd.
Registered Office: 818, 8th Floor, Indraprakash Building, 21, Barakhamba Road, New Delhi 110001
Corporate Office: 14th Floor, Building No. 10B, DLF Cyber City, Phase II, Gurgaon-122002, Haryana, India

Research Methodology and Biostatistics: A Comprehensive Guide for Health Care Professionals,
Suresh K. Sharma

Copyright © 2017, by RELX India Pvt. Ltd.

All rights reserved.

ISBN: 978-81-312-4442-5
e-Book ISBN: 978-81-312-4648-1

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or me-
chanical, including photocopying, recording, or any information storage and retrieval system, without permission
in writing from the publisher. Details on how to seek permission, further information about the Publisher’s per-
missions policies and our arrangements with organizations such as the Copyright Clearance Center and the
Copyright Licensing Agency can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other
than as may be noted herein).

Notice

Knowledge and best practice in this field are constantly changing. As new research and experience broaden
our understanding, changes in research methods, professional practices, or medical treatment may become
necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and
using any information, methods, compounds, or experiments described herein. In using such information or
methods they should be mindful of their own safety and the safety of others, including parties for whom they
have a professional responsibility.
With respect to any drug or pharmaceutical products identified, readers are advised to check the most
current information provided (i) on procedures featured or (ii) by the manufacturer of each product to be
administered, to verify the recommended dose or formula, the method and duration of administration, and
contraindications. It is the responsibility of practitioners, relying on their own experience and knowledge of
their patients, to make diagnoses, to determine dosages and the best treatment for each individual patient, and
to take all appropriate safety precautions.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any
liability for any injury and/or damage to persons or property as a matter of product liability, negligence or
otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the
material herein.
Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this
publication does not constitute a guarantee or endorsement of the quality or value of such product or of the
claims made of it by its manufacturer.
Please consult full prescribing information before issuing prescription for any product mentioned in this
publication.

Manager—Content Strategy: Nimisha Goswami

Sr Manager—Education Solutions: Shabina Nasim
Content Development Specialist: Subodh Kumar
Project Manager: Ranjjiet Varhmen
Sr Production Executive: Ravinder Sharma
Sr Cover Designer: Milind Majgaonkar

Typeset by GW India
Printed in India by
 Foreword

Evidence-based practice is the need of the hour in a research question, separating the two approaches
health care disciplines. Thus, health care profes- when discussion on the fundamental interests of
sionals such as doctors, dentists, nurses, physio- research results in a polarized view. Therefore, this
therapists, occupational therapists, dieticians, book has an added strength in that it will facilitate
clinical psychologists, as well as health care tech- the readers to understand quantitative as well as
nologists are continuously engaged in conducting qualitative research methods beginning from the
research to generate new evidence or refine old conception of research idea to the analysis of data.
evidence to continuously improve their practices I found this book very reader-friendly to the
for better patient outcome. Research is taught as complex topics, so that beginners can grasp the
support or introductory subject in several ways in fundamentals of approaching research and expe-
most health-related disciplines. The core philo- rienced health care professions can use research in
sophical base for this book comes from the au- practice. I also found that this textbook is pre-
thor’s conviction that although these health- sented in an uncluttered, straightforward manner.
related disciplines vary in content but their board The text uses many bulleted lists to make the
approach towards research enquiry remains simi- content visually interesting as figures and tables
lar. Therefore, this book is addressed to all the further contribute towards this purpose. Thus,
health care professionals, namely doctors, den- this text facilitates the reader to grasp the infor-
tists, nurses, physiotherapists, occupational thera- mation quickly and to read efficiently. I hope this
pists, dieticians, clinical psychologists and health textbook will prove to be an excellent ready refer-
care technologists. ence on research methodology and biostatistics
It is observed that some disciplines place greater for medical, dental, nursing, physiotherapy, occu-
emphasis on the quantitative methods of research, pational therapy, therapeutic dietetics, clinical
while others pay more focus towards qualitative psychology, medical technology, and other health
research methods. However, it is strongly sug- care students and practitioners.
gested by prominent health science researchers
that combined quantitative–qualitative research Sanjeev Misra
methods are more powerful to yield better em- MS, MCh, FRCS (Eng), FRCS (Glasgow),
pirical evidence. I also observed that this text cov- FICS, FACS (USA), FAMS
ers the research process with an integrated discus- Professor Surgical Oncology
sion of both quantitative and qualitative research Director
methods/traditions. Most health care researchers All India Institute of Medical Sciences
have to consider all paradigms when approaching Jodhpur, Rajasthan, India
This page intentionally left blank
 Preface

Research Methodology and Biostatistics: A Com- logic and techniques of data generation and man-
prehensive Guide for Health Care Professionals is agement, and the fundamentals and implications
a definitive, comprehensive guide for understand- of various sampling techniques and alternative
ing and performing research in health care disci- statistical methodologies.
plines. This is designed for undergraduate and This book is organized in 12 easy-to-read chap-
postgraduate medical, dental, nursing, physio- ters; the text emphasizes the importance of clearly
therapy, clinical psychology and other allied health defined research questions and well-constructed
care professional students, faculty and health care hypothesis, effectively chooses research designs,
scientists involved in translational research. This sampling techniques, sample size, data collections
indispensable reference also addresses the unique methods/tools and data analysis statistical tech-
challenges and demands of health science research niques for informing methods and in guiding data
and offers clear guidance in becoming a more suc- interpretation. It is thoroughly reviewed by prom-
cessful member of a medical research team and inent medical scientists and methodologists who
critical reader of the medical research literature. have extensive personal experience in biomedical
This book covers the entire research process, investigation and in teaching key aspects of re-
beginning with the conception of the research and search methodology to medical students, physi-
research problem to publication of findings. Re- cians and other health professionals. The author
search Methodology and Biostatistics: A Compre- expertly integrates theory with examples and uses
hensive Guide for Health Care Professionals com- language that is clear and easy for a general health
prehensively and concisely presents concepts in a care audience. A major contribution to the meth-
manner that is relevant and engaging to read. The odology literature, Research Methodology and
text combines theory and practical application to Biostatistics: A Comprehensive Guide for Health
familiarize the reader with the logic of research Care Professionals is an authoritative resource for
design and hypothesis construction, the impor- all individuals who perform research, plan to per-
tance of research planning, the ethical basis of form it or wish to understand it better.
human subjects research, the basics of writing a
medical research protocol and scientific paper, the Suresh K. Sharma
This page intentionally left blank
 Reviewers

Raj Kumar, MS, Mch, PhD, FRCS, MRCS Ravinder K. Soni, PhD
Former Director Professor
All India Institute of Medical Sciences Department of Social and Preventive Medicine
Rishikesh, Uttarakhand Dayanand Medical College and Hospital
Ludhiana, Punjab
Vartika Saxena, MD
Additional Professor Sonu Goel, MD
Department of Community and Family Assistant Professor
Medicine School of Public Health
All India Institute of Medical Sciences Postgraduate Institute of Medical Education
Rishikesh, Uttarakhand and Research
Chandigarh
Saurabh Varshney, MS
Professor and Head Jagdish Prasad Goyal, MD
Department of Otolaryngology, Head & Neck Assistant Professor
Surgery Department of Paediatrics
All India Institute of Medical Sciences All India Institute of Medical Sciences
Rishikesh, Uttarakhand Rishikesh, Uttarakhand

Rajoo Singh, MD, DM Ashi Chug, MDS

Professor of Gastroenterology and Dean Assistant Professor
Academics Department of Dentistry
Dayanand Medical College and Hospitals All India Institute of Medical Sciences
Ludhiana, Punjab Rishikesh, Uttarakhand

Pratima Gupta, MD Pankaj Bhardwaj, MD

Professor and Head Associate Professor
Department of Microbiology Department of Community and Family Medicine
All India Institute of Medical Sciences AIIMS, Jodhpur
Rishikesh, Uttarakhand
Geetika Khanna Bhattacharya, MD
Jaya Chaturvedi, MS Professor
Professor and Head Department of Pathology
Department of Obstetrics and Gynecology Vardhman Mahavir Medical College
All India Institute of Medical Sciences and Safdarjung Hospital
Rishikesh, Uttarakhand New Delhi
x Reviewers

Sucheta P. Dandekar, PhD Dorothy Deena Theodore, MScN

Professor and Head Principal
Department of Biochemistry Narayana Hrudayalaya College of Nursing
Seth G.S. Medical College and KEM Hospital Bangalore, Karnataka
Mumbai, Maharashtra
 Contents

Chapter 1. Introduction to Medical Chapter 3. Literature Review, 46

Research, 1 Meaning of literature review, 47
Meaning of research and medical Importance of a literature review, 47
research, 2 Purposes of a literature review, 48
Research terminologies, 3 Types of literature reviews, 48
Need and purposes of research in health Sources of a literature, 50
sciences, 5 Steps of a literature review, 54
Importance and significance of research in Points to be considered for a literature
health sciences, 5 review, 59
Historical evolution of clinical research in
India, 6
Characteristics of a good research, 7
Chapter 4. Research Designs, 62
Types of research, 8 Meaning of research design, 63
Research process, 11 Selection of a research design, 63
Validity of a research design, 64
Types of research designs, 66
Chapter 2. Research Problem and Quantitative research designs, 66
Hypothesis, 21 Qualitative research designs, 90
Research problem, 22 Mixed methods/triangulation, 103
Meaning of research problem, 22
Identification of a research problem, 22 Chapter 5. Randomized Controlled
Sources of a research problem, 23
Criteria for a good research problem, 24
Trials, 108
Formulating a problem statement, 26 Meaning of randomized controlled trials, 109
Variables, 28 History of randomized controlled trials, 109
Operations definitions, 30 Characteristics of randomized controlled
Writing research objectives, 32 trials, 109
Hypothesis, 34 Randomization in randomized controlled
Meaning of hypothesis, 36 trials, 110
Importance of hypotheses in research, 36 Types of randomized controlled trials, 110
Characteristics of a good hypothesis, 37 Use of placebo in randomized controlled
Sources of hypotheses, 38 trials, 112
Types of hypotheses, 38 Difficulties in conducting randomized
Formulating hypotheses, 39 controlled trials, 113
Delimitations, 41 Advantages of randomized controlled
trials, 113
Disadvantages of randomized controlled
trials, 113
xii Contents

Chapter 6. Sampling and Sample Size Chapter 9. Descriptive Statistics, 214

Determination, 116 Meaning of statistics and
Terminologies used in sampling, 117 biostatistics, 215
Purpose of sampling, 118 Uses of application of statistics, 215
Characteristics of a good sample, 118 Terms related to biostatistics, 215
Sampling process, 118 Classification of statistics, 216
Factors influencing a sampling process, 119 Frequency distribution, 216
Types of sampling techniques, 120 Measures of central tendency, 223
Probability sampling, 122 Measures of dispersion, 227
Nonprobability sampling, 127
Sample size estimation, 133 Chapter 10. Measures of Relationship
Probability and sampling errors, 143
Problems of sampling, 145 and Inferential
Statistics, 231
Normal probability curve, 232
Chapter 7. Measurements and
Correlation coefficient, 233
Methods of Data Scatter diagram method, 235
Collection, 148 Product moment correlation, 237
Scales of measurements, 149 Regression analysis, 239
Concept of data collection, 150 Inferential statistics, 242
Questionnaire, 153 Parametric tests, 244
Interview as method of data collection, 157 t-Test, 244
Composite scales/attitude scales, 161 Z-test, 246
Likert scale, 162 ANOVA, 247
Semantic differential scale, 163 Nonparametric tests, 250
Visual analog scale, 164 Chi-square test, 250
Observation as method of data Sign test, 251
collection, 165 Wilcoxon signed rank test, 252
Rating scale, 168 Mann–Whitney test, 253
Checklist, 170
Biophysiological methods, 172 Chapter 11. Writing Proposals, Reports
Projective technique, 174
Q-sorts, 176 and Publications, 258
Vignettes, 177 Criteria for communicating research, 259
Validity of research tool, 179 Methods of communicating research, 259
Reliability of research tools, 180 Writing a research report, 261
Pilot study, 183 Format of a thesis or dissertation, 263
Writing references/bibliography, 265
Writing scientific articles, 274
Chapter 8. Analysis of Data, 187 Critical review of research work, 275
Definitions of data analysis, 188 Developing and presenting a research
Analysis of quantitative data, 188 proposal, 279
Parametric tests, 194 Uses of computer in medical
Nonparametric tests, 196 research, 284
Software analysis of quantitative data, 198
Analysis of qualitative data, 200
 Contents xiii

Chapter 12. Ethical Issues in Compensation for participation in

research, 302
Biomedical Research Conflict of interest in research, 302
in India, 289 General and specific principles for clinical
Principles of ethics in biomedical evaluation of drugs/clinical trials
research, 290 guidelines in India, 303
Ethics and regulation in biomedical research Institute ethical committee, 307
– international guidelines, 291
Ethics and regulation in biomedical research Appendices, 311
– national guidelines, 297
Patient information document and informed Index, 331
consent, 299
This page intentionally left blank
CHAPTER OUTLINE
Introduction to
Medical Research 1
INTRODUCTION HISTORICAL EVOLUTION OF CLINICAL RESEARCH
IN INDIA
MEANING OF RESEARCH AND MEDICAL RESEARCH
CHARACTERISTICS OF A GOOD RESEARCH
RESEARCH TERMINOLOGIES
TYPES OF RESEARCH
NEED AND PURPOSES OF RESEARCH IN HEALTH
SCIENCES RESEARCH PROCESS

IMPORTANCE AND SIGNIFICANCE OF RESEARCH IN MULTIPLE CHOICE QUESTIONS

HEALTH SCIENCES

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Recall historical evolution of clinical research in India.
• Understand the concept of research and medical • Identify characteristics of good medical research.
research. • Classify the types of research and recite relevant
• Define research and medical research. examples.
• Define basic medical research terms. • Compare the basic elements of qualitative and quantitative
• Appraise the need, purpose, importance and significance research.
of research in health sciences. • Describe quantitative and qualitative research process.

KEY TERMS

• Applied research 11 • Construct 04 • Pilot study 05

• Basic research 10 • Data 03 • Probability sampling 04
• Deductive reasoning 18 • Demographic variables 03 • Reliability 04
• Medical research 02 • Dependent variables 03 • Representative sample 04
• Qualitative research 08 • Extraneous variables 03 • Research study setting 04
• Quantitative research 08 • Hypothesis 04 • Research variables 03
• Research 02 • Independent variables 03 • Sample 04
• Abstract 03 • Limitations 04 • Sampling 04
• Accessible population 04 • Literature review 04 • Target population 04
• Analysis 05 • Manipulation 04 • Validity 05
• Assumption 04 • Nonprobability sampling 04
• Concept 04 • Operational definition 04
2 Research Methodology and Biostatistics

means ‘to search again’ or ‘carefully examine’.

INTRODUCTION Thus, research means close and careful examina-
Research in the health sciences has helped to tion of facts and their relationship to each other,
develop a large amount of new knowledge for to discover new knowledge.
improving the health of people, increasing their The term ‘research’ may be defined as a process
lifespan and raising living standards. The major of answering unanswered questions or solving
beneficial effects of biomedical research can be unsolved problems. This is done carefully using a
seen on the traditional health indicators such as scientific, orderly and systematic process. The ap-
morbidity and mortality rates. Now, people are proaches and objectives for researches conducted in
living longer than before and are less likely to suc- the different fields/disciplines of health sciences may
cumb to many of the scourges of the past. Today be different but the ultimate aim of any research is to
the health care fraternity has better and improved refine old knowledge and discover new knowledge.
diagnostic and treatment modalities. The most
recent breakthrough of biomedical research is the
MEANING OF RESEARCH
isolation of human stem cells; the mapping of the
human genome, have just begun to demonstrate AND MEDICAL RESEARCH
its vast therapeutic potential. Research is defined as a systematic and scientific
Research is a systematic inquiry which uses process of answering questions about facts and
disciplined methods to answer questions or solve the relationship between the facts. It is an activ-
problems. The ultimate goal of research is to de- ity involved in seeking answers to unanswered
velop, refine and expand a body of knowledge. questions.
Research strives to garner knowledge and make Research seeks to generate an answer to a problem
discoveries that aid our daily lives. Health care as well as suggesting additional questions in
professionals are increasingly engaged in disci- need of further inquiry.
plined studies that benefit the profession and pa- Research is systematic method of exploring, de-
tients. Biomedical research is a systematic inquiry scribing, explaining, relating or establishing the
designed to develop knowledge about the issues of existence of phenomenon, the factors that cause
importance to health care professionals, including change in the phenomenon and how the phe-
health care practices, health science education and nomenon influences other phenomena.
health care administration. Research is a systematic inquiry that uses disci-
Research in health care is conducted with an aim plined methods to answer questions or solve
of gaining a deeper understanding into the funda- problems. The ultimate goal of research is to
mentals of disease causation factors, normal ana- develop, define, refine and expand a body of
tomical, physiological, biochemical and molecular knowledge for a discipline.
variation and pathological process of disease condi- Research is a scientific, systematic, controlled,
tions at the molecular, cellular, organic and organ- orderly and objective investigation to develop,
ism level; and to refine the existing techniques and refine and expand the body of knowledge.
find out newer techniques/interventions of disease Research essentially is a problem-solving process,
diagnosis, treatment, care and rehabilitation of the a systematic, intensive study directed towards
individual, family and community. full scientific knowledge of subject studies.
The word research is derived from the French ­—Ruth M. French (1968)
term ‘recerchier’, a compound word composed of Research may be defined as the systematic and ob-
a prefix ‘re’ and a verb, ‘search’. ‘Re’ means once jective analysis and recording of controlled obser-
again, anew or afresh; ‘search’ means to look for vation that may lead to the development of
something or examine closely and carefully, to generalizations, principles, theories, resulting in
look for information, to test and try or to prove. prediction and possible ultimate control of events.
Langford (2001) also mentioned that ‘research’ —J.W. Best (1969)
 Introduction to Medical Research 3

Research is the process of systematic obtaining of Clinical research can be defined as the research
accurate answers to significant and pertinent that takes place in a hospital or clinical setting
questions by the use of scientific methods of and is focused on treating specific human and
gathering and interpreting information. animal diseases and other ailments. Clinical re-
—Clover and Balsley (1979) search builds upon the knowledge learned
Research may be defined as the planned, system- through applied and basic research. Clinical
atic search for information for the purpose of research is conducted on human beings and
increasing the total body of humanity’s knowl- takes the shape of treatments and drugs that
edge. It involves looking for information which directly improve human health care.
at the time is not available or for which there is
no generally accepted evidence.
—Archold Lancaster (1982)
RESEARCH TERMINOLOGIES
Scientific research is a systematic, controlled, em- l Abstract: A clear, concise summary of a study
pirical and critical investigation of natural phe- that communicates the essential information
nomena guided by theories and hypotheses about about the study. In research journals, it is usu-
the presumed relations among such phenomena. ally located at the beginning of an article.
—Fred N. Kerlinger (1986) l Data: Units of information or any statistics, facts,
Biomedical research is the broad area of science that figures, general material, evidence or knowledge
looks for ways to prevent and treat diseases that collected during the course of the study.
cause illness and death in people and in animals. l Variables: Attributes or characteristics that
This general field of research includes many areas can have more than one value, such as height
of both the life and physical sciences. Utilizing or weight. In other words, variables are quali-
biotechnology techniques, biomedical researchers ties, quantities, properties or characteristics of
study biological processes and diseases with the people, things or situations that change or
ultimate goal of developing effective treatments vary.
and cures. Biomedical research is an evolutionary  Dependent variables: Variables that change

process requiring careful experimentation by as the independent variable is manipulated

many scientists, including biologists and chem- by the researcher; sometimes called criterion
ists. Discovery of new medicines and therapies variables.
requires careful scientific experimentation, devel-  Independent variables: Variables that are

opment and evaluation. purposely manipulated or changed by the re-

—New Jersey Association for Biomedical Research searcher; also called manipulated variables.
Biomedical research is the area of science devoted to  Research variables: These are the qualities,

the study of the processes of life, the prevention properties or characteristics that are observed
and treatment of disease, and the genetic and en- or measured in a natural setting without ma-
vironmental factors related to disease and health. nipulating and establishing cause-and-effect
—Organisation for Economic Cooperation relationship.
and Development  Demographic variables: The characteristics

Medical research is the broad area of science that and attributes of the study subjects are con-
involves the systematic and orderly process of sidered demographic variables, e.g., age, gen-
investigating the normal biological process, dis- der, educational status, religion, social class,
ease causative factors, pathological processes and marital status, habitat, occupation, income
techniques/interventions of disease prevention, and medical diagnosis.
disease diagnosis, treatment, care and rehabilita-  Extraneous variables: Extraneous variables

tion of the individual, family and community are the factors that are not part of the study
through careful experimentation, observation, but may affect the measurement of the study
laboratory work, analysis and testing. variables.
4 Research Methodology and Biostatistics

l Operational definition: The way by which a re- l Manipulation: An intervention or treatment

searcher clarifies and defines the variables under introduced by the researcher in an experimental
investigation. In addition, the researcher must or quasi-experimental study; the researcher ma-
also specify how the variables will be observed nipulates the independent variable to assess its
and measured in the actual research situation. impact on the dependent variable.
l Concept: A word picture or mental idea of a l Population: The entire set of individuals or
phenomenon. Concepts are the words or terms objects having some common characteristic(s)
that symbolize some aspects of reality, e.g., selected for a research study (e.g., patients
stress, pain or love. Concepts are the building admitted to intensive care units); sometimes
blocks of theory. referred to as the universe of the research study.
l Construct: A highly abstract, complex phenome-  Target population: The entire population in

non (concept) is denoted by a madeup or con- which the researchers are interested and to
strued term. A construct term is used to indicate a which they would like to generalize the re-
phenomenon that cannot be directly observed but search findings.
must be inferred by certain concrete or less-ab-  Accessible population: The aggregate of cases

stract indicators of the phenomenon. For example, that conform to designated inclusion or ex-
wellness, mental health and self-esteem are con- clusion criteria and that are accessible as sub-
structs, and they can only be measured through jects of the study.
definable and measurable concepts, e.g. wellness l Research study setting: The study setting is the
can only be assessed through laboratory data. location in which the research is conducted – it
l Proposition: A proposition is a statement or could be natural, partially controlled or highly
assertion of the relationship between concepts. controlled. Natural or field setting is an uncon-
For example, there is a relationship between trolled real-life situation. In a partially con-
level of anxiety and performance; or virus and trolled situation, the environment is partially
acute illness. Propositions are drawn from theo- modified to control extraneous variables, while
ries or empirical data. in highly controlled situations, the study envi-
l Conceptual framework: Interrelated concepts ronment is fully controlled to combat the effect
or abstractions that are assembled together in of extraneous variables.
some rational scheme by virtue of their rele- l Sample: A part or subset of population selected
vance to a common theme; sometimes referred to participate in the research study.
to as a conceptual theoretical framework if based l Representative sample: A sample whose char-
on the concepts of an existing theory or theories. acteristics are highly similar to that of the popu-
l Assumption: A basic principle that is accepted lation from which it is drawn.
as being true on the basis of logic or reason, l Sampling: The process of selecting a sample
without proof or verification. from the target population to represent the en-
l Hypothesis: A statement of the predicted rela- tire population.
tionship between two or more variables in a re- l Probability sampling: The selection of subjects or
search study; an educated or calculated guess by sampling units from a population using a random
the researcher. procedure, e.g., simple random sampling, strati-
l Literature review: A critical summary or re- fied random sampling and systematic sampling.
search on a topic of interest, generally prepared l Nonprobability sampling: The selection of
to put a research problem in context or to iden- subjects or sampling units from a population
tify gaps and weaknesses in prior studies so as to using nonrandom procedures, e.g., convenient,
justify a new investigation. purposive and quota sampling.
l Limitations: Restrictions in a study that may l Reliability: The degree of consistency or accu-
decrease the credibility and generalizability of racy with which an instrument measures the
the research findings. attribute it is designed to measure.
 Introduction to Medical Research 5

l Validity: The degree to which an instrument l Enhance their professional identity as research
measures what it is intended to measure. is an essential component of any profession.
l Pilot study: Study carried out at the end of the l Define the parameters of different health care
planning phase of research in order to explore disciplines, which will help health care profes-
and test the research elements to make relevant sionals identify boundaries of their respective
modifications in research tools and methodology. discipline.
l Analysis: Method of organizing, sorting and l Refine and eliminate old knowledge so that it
scrutinizing data in such a way that the research helps in eliminating actions/interventions that
question can be answered or meaningful infer- have no effect on the achievement of desired
ences can be drawn. client outcomes.
l Identify health care practices that make a differ-
ence in health care status of individuals and are
NEED AND PURPOSES OF RESEARCH
cost-effective.
IN HEALTH SCIENCES l Enhance accuracy of different biomedical
Health care professionals engage in research for a educational and administrative techniques.
number of reasons. As professionals, they seek to l Develop and refine theories and principles
establish a scientific base of knowledge for health applicable in different health care disciplines.
care for prevention of disease, diagnosis of dis- l Solve problems or answer questions related to
ease, treatment, care and rehabilitation of the in- health care practices, medical education and
dividual, family and community. Additionally, health care administration.
the systematic accrual of biomedical information
enables health care professionals to better define
IMPORTANCE AND SIGNIFICANCE OF
the parameters of health science, document the
unique contribution health care professionals RESEARCH IN HEALTH SCIENCES
makes to health care, determine the effectiveness Research must be an integral part of any disci-
of their actions, develop theoretical frameworks pline. It provides the power to health care profes-
for clinical practices, and enhance more informed sionals to develop, define and refine medical
clinical decision-making. There is growing con- knowledge for providing effective patient care,
sensus that knowledge of biomedical research is medical education and health care administra-
needed to enhance the professional practices of all tion. Medical research promotes evidence-based
health care professionals. practice (EBP), ensures credibility of the medical
Research is a fundamental essential prerequi- profession, provides accountability for medical
site for any profession. The specific purpose of practices and enables availability of cost-effective
biomedical research includes identification, de- care. There are several arguments that state the
scription, exploration, explanation, prediction importance and significance of medical research.
and control of facts. Therefore, research enables Some of the facts that prove the importance or
health care professionals to do the following: significance of medical research are as follows:
l Develop, refine and extend the scientific base of l The development and utilization of medical

knowledge, which is required for quality health care, knowledge is essential for continued develop-
medical education and health care administration. ment in patient care; health care professionals
l Enhance the body of professional knowledge in can develop or refine this knowledge with the
health sciences. help of medical research.
l Provide a foundation for evidence-based medi- l Health care professionals are increasingly ex-

cal (EBM) practices. pected to adopt a research-based or evidence-

l Help in the expansion of knowledge, which is based practice, using research findings to guide
essential for continued growth of health care their decisions, actions and interactions with
professions. patients, patients’ families and community.
6 Research Methodology and Biostatistics

Research is the only way which can provide - validate and refine the existing health care
health care professionals first-hand experiences practice and
on which they can build their EBM practices. - make health care efficient and cost-effective.
l In a fast-developing world, health care profes-
sionals are realizing the need to base specific
HISTORICAL EVOLUTION OF CLINICAL
medical interventions and decisions on evi-
dence indicating that the actions are clinically RESEARCH IN INDIA
appropriate, cost-effective and result in positive Historical records shows that evolution of clinical
patient outcome. This can be achieved through research has a long and fascinating journey which
research conducted in the health discipline. moved from the investigations on the effectiveness
l Another major reason for which health care of dietary therapy to the efficacy of drugs. Infor-
professionals must get involved in research is mation about the first clinical trail may be traced
rapid advancement in medical science and tech- back to the Bible in 500 BC. The major milestones
nology; to keep pace with this advancement, in the evolution of clinical research in India are as
health care professionals must conduct research described here:
to learn about the efficacy of a particular tech- l Evidence of clinical research in the form of de-

nology in reference to positive patient outcome. tailed observations of disease and in-depth guid-
l Medical research is essential for health care pro- ance on remedies may be found in the ancient
fessionals to understand the varied dimensions textbooks of the Indian system of medicines.
of their profession. However, there is lack of documentation on sys-
l Research enables health care professionals to tematic clinical experimentation in ancient liter-
study characteristics of a particular health care ature.
situation about which little is known, explain l 15 November 1911: The first meeting of the

the phenomena that must be considered in Governing Body of the Indian Research Fund
planning health care, predict the probable out- Association (IRFA) was held at the Plague Labo-
come and initiate activities to promote desired ratory, Bombay, under the Chairmanship of Sir
patient behaviour. Harcourt Butler.
l Health care professionals need to get involved in l In 1912: The second meeting of the Governing

research because the cost of care is increasing Body of the IRFA was held, where historic deci-
day by day. Therefore, they must find a solution sion was taken to start a journal for Indian
for cost-effective care through research. medical research.
l Research also helps health care professionals to l 1918–1920: Various clinical research projects on

eliminate health care actions that do not achieve malaria, beriberi, kala-azar and indigenous
the desired care outcome for patients. Further- drugs were undertaken.
more, medical research findings may help health l In 1945: The first research unit was established

care professionals identify the practices that at the Indian Cancer Research Centre, Bombay.
improve health care outcome of patients and l In 1949: IRFA was redesignated as the Indian

remain cost-effective. Council of Medical Research (ICMR).

To summarize, the basic importance of medical l Several national level research centres in the fields

research is that it helps in identifying and of tuberculosis, nutrition, leprosy, viral diseases,
describing a phenomenon in the health care cholera, enteric diseases, reproductive disorders,
profession, exploring and explaining the facts toxicology, cancer, traditional medicine, gas disas-
related to this phenomenon and their relation- ter, genetics, AIDS, etc. were started by the ICMR.
ship to each other, and predicting its occur- l 10 September 1996: The first meeting of the Cen-

rence. Therefore, in a nutshell, research helps to tral Ethical Committee of ICMR on Human
- build and expand the body of health profes- Research was conducted under the Chairmanship
sional knowledge, of Hon’ble Justice (Retired) M.N. Venkatachaliah.
 Introduction to Medical Research 7

l Several subcommittees were constituted to con- l Begin with clearly defined purposes: A re-
sider ethical issues in specific areas, e.g., epide- search can only be conducted in an effective
miological research, clinical evaluation of prod- manner if it is started with clearly defined pur-
ucts to be used on humans, organ transplantation poses. In the absence of clarity about objectives,
and human genetics. The committee released research may not yield very good results.
Ethical Guidelines for Biomedical Research on l Emphasize developing, refining and expanding
Human Participants in 2000 which were revised professional knowledge: The main purpose of
in 2006. research is to develop new knowledge or refine
l 1988: The schedule Y of Drugs and the Cosmetics the existing professional knowledge. An ever-
Act come into force and the regulatory guidelines expanding professional body of knowledge is the
for clinical trial (CT) permission were established, most important need of any profession. Therefore,
and then revised in January 2005. This has also a good research is always directed towards expan-
legalized the Indian GCP guidelines of 2001. sion of the professional body of knowledge.
l On 20 July 2007, the Clinical Trials Registry – l Directed towards development or testing of
India (CTRI) was launched, which is a free theories: Theory development and testing is a
online public record system of CTs being systematic process of enquiry in a discipline.
conducted in India. Since then, the registration Therefore, a good research must make it possi-
of each CT has been mandatory on the CTRI ble to develop or test a theory.
website (http://ctri.nic.in/Clinicaltrials) by the l Finding solutions to problems: Finding an-
Drugs Controller General of India (DCGI) swers or solutions to professional problems is
(www.cdsco.nic.in). Editors of biomedical jour- another important purpose of a research. There-
nals in India also make it mandatory that each fore, a good research is always directed towards
research article of CT submitted for publication finding an answer or solution for pertinent
must include the record of registration of CT emerging professional problems.
with CTRI. Thus, CTs evolved into a standard- l Dedicated to develop empirical evidence: An-
ized procedure, focusing on scientific assess- other important purpose of research is to gener-
ment of efficacy and guarding patient safety. ate empirical evidence, which can be used to
improve the professional practices. Therefore, a
good research always strives to develop empiri-
CHARACTERISTICS OF A GOOD
cal evidence.
RESEARCH l Strive to collect first-hand information/data:
Research employs scientific methods. Good re- A successful research is conducted by collecting
search is systematic, logical, empirical and also data directly from subjects by different methods
replicable. In detail, one expects that good re- like questioning, interviews or observation.
search must satisfy the following criteria: l Objective and logical process: Research infor-
l Orderly and systematic process: A research can mation collected through subjective means or in
only be considered good if it is conducted in an a haphazard manner will never lead to satisfac-
orderly and systematic way, because this is the tory results. Therefore, a good research empha-
most important criterion for a scientific re- sizes on an objective and logical research process.
search process. l Generate findings to refine and improve profes-
l Based on current professional issues: Resolv- sional practices: A research cannot be considered
ing current issues of any discipline is one of successfully conducted until and unless it contrib-
the fundamental purposes of any research. utes towards the refinement and improvement in
Therefore, a good research must be based on professional practices as per the need of the times.
current professional issues, so that a particular l Use of appropriate methodology: Selection of
discipline can be up to date with solutions to a methodology in any research depends on sev-
professional concerns. eral factors, because each variable and research
8 Research Methodology and Biostatistics

has unique features. Therefore, a good research Box 1.1. Qualities of a good biomedical researcher
always employs the most appropriate and suit-
able methodology. R – Research-oriented
E – Efficient
l Conducted on representative sample: General-
S – Scientific
ization of research findings is only possible if E – Efficient
the study is conducted on a sample which has A – Active
characteristics similar to the population of the R – Resourceful
study. Therefore, a good research is conducted C – Creative
on a representative sample. H – Honest
l Conducted through appropriate use of meth- E – Economical
R – Religious
ods and tools of data collection: Based on the
nature of phenomenon and subjects under study,
methods and tools of data collection will be dif- a good research cannot be imagined. Box 1.1
ferent. Therefore, only the most appropriate also presents the qualities of a good biomedical
and suitable methods and tools of data collection researcher.
can lead to successfully conducted research. For l Adequately communicated: A research cannot
example, a questionnaire cannot be used for be considered good until the information gen-
illiterate subjects or a performance can only be erated is adequately disseminated to its users.
measured through the observation method. Therefore, a good research activity strives to
l Use of valid and reliable data collection tools: communicate findings as widely as possible.
Evidence generated from a research activity can
only be considered valuable if they are gener-
TYPES OF RESEARCH
ated through valid and reliable research tools.
Therefore, a good research always depends on Research can be classified based on several criteria.
the validity and reliability of research tools. However, basically, research is classified based on
l Carefully recorded and reported: A research either an approach of studying the variable (quan-
can only be considered good if research infor- titative and qualitative research) or the purpose of
mation is carefully recorded and reported; in conducting the research (basic and applied re-
the absence of careful recording and reporting, search). These classifications are further discussed
the evidence generated during research may be in the following sections (Fig. 1.1).
covert in nature. Therefore, to generate quality
QUANTITATIVE AND QUALITATIVE RESEARCH
empirical evidence, it is essential to carry out
(Table 1.1)
recording and reporting very carefully.
l Adequately and appropriately analysed re- Quantitative Research
search: A research activity is incomplete and It is an inquiry into an identified problem, based
poor until gathered data are adequately and ap- on testing a theory composed of variables, mea-
propriately analysed using standardized and ac- sured with numbers and analysed using statistical
cepted methods of data analysis. techniques. In this type of research, data are
l Patiently carried out activity: Research cannot collected in numerical form and analysed by using
be carried out in a hurry; it needs lots of time descriptive or inferential statistics. Quantitative
and patience. Therefore, patience and endur- research involves analysis of numerical data.
ance are foundations of good research. Quantitative research is often an iterative pro-
l Researcher’s expertise, interest, motivation and cess whereby evidence is evaluated, theories and
courage: It is largely believed that a good research hypotheses are tested and technical advances are
activity needs expertise, interest, motivation and made. The two major types of quantitative re-
courage among researchers. In the absence of these search designs are experimental and observational
attributes among researchers, accomplishment of research. The experimental research is further
 Introduction to Medical Research 9

Types of research

Based on approach Based on purpose

Qualitative Quantitative Basic Applied

research research research research

Experiential Observational
research research

Descriptive research
Case report
Randomized Quasi-experimental Case series
control trial (RCT) research Cross-sectional
Case–control studies
Cohort studies
Prospective cohort studies
Historical cohort studies
Ambispective cohort studies

Grounded
Phenomenology Ethnography Case study Historical
theory
research research research research
research

Fig. 1.1. Types of research.

classified into randomized control trial (RCT) and Qualitative research is the examination, analysis
quasi-experimental research. The observational and interpretation of observations for the purpose
research is further categorized as descriptive, of discovering underlying meanings and patterns
case–control and cohort research. of relationships, including classifications of types of
phenomena and entities in a manner that does not
Example of quantitative research
involve mathematical models. Qualitative research
A descriptive study on the prevalence of hospital- has the following main features:
acquired infection among patients admitted in l Qualitative research is concerned with the opin-
intensive care units. In this research study, the ions, experiences and feelings of individuals
researcher deals with numerical data.
producing subjective data.
l Qualitative research describes social phenom-
Qualitative Research ena as they occur naturally.
Qualitative research is a field of inquiry that cross- l Understanding of a situation is gained gradually

cuts disciplines and subject matter. It involves an in- through a holistic perspective.
depth understanding of human behaviour and the l Qualitative data are collected through direct in-

reasons that govern human behaviour. Here, data are teraction with individuals, through one-to-one
collected in descriptive form rather than numerical or group interviews or by observation. Data
form and analysed by descriptive coding, indexing collection is time-consuming.
and narrations. Qualitative research involves analysis l The intensive and time-consuming nature of data

of data, such as words, pictures or objects. collection necessitates the use of small samples.
10 Research Methodology and Biostatistics

Table 1.1. Comparison of quantitative and qualitative research

Parameters Quantitative Research Qualitative Research
General nature Objective approach to seek precise Subjective approach to seek in-depth
measurement in numerical form description in narrative form
Knowledge of Variables are clearly understood and Researcher may have only a rough idea
study variables defined in advance by the researcher about variables in advance
Aims To identify and count the variables To collect complete, detailed description
to test or refine theories and of variables to develop theories and
hypotheses assumptions
Research problem Problem is deductively reasoned and Problem is inductively reasoned and is
does not change over the course of constantly reformulated during the
the study course of the study
Planning All aspects of the study are carefully The design emerges as the study unfolds
designed before data are collected
Research design Experimental, quasi-experimental and Phenomenology, grounded theory, ethnogra-
nonexperimental phy, case study and historical research
Sample Quantitative studies are conducted Qualitative research studies are in-depth
on large representative samples to studies and, therefore, have small-size
generalize the research findings samples; thus, generalization of findings is
generally not possible
Data Quantitative data are easier to collect Qualitative data are ‘richer’, time-consuming
and facilitate the testing of hypothe- and less generalizable
ses, but may miss contextual details
Tools Structured or semistructured tools to Less formally structured instruments like par-
collect numerical data, such as ticipative observation, in-depth interviews
questionnaires and opinion polls and focused group discussions (FGDs)
Methods Researcher tends to remain objectively Researcher tends to become subjectively
separated from the subject matter immersed in the subject matter
Analysis Analysis of numerical data through Analysis of data such as words, pictures
descriptive and/or inferential statistics or objects through descriptive coding,
indexing, narrations, integrative diagrams
and thematic analysis

l Criteria used to assess reliability and validity BASIC AND APPLIED RESEARCH
differ from those used in quantitative research. Basic Research
The five major types of qualitative research de-
Basic (fundamental or pure) research is performed
signs used are phenomenology, ethnography,
without a specific purpose in mind. Rather, it is
grounded theory, case study and historical research.
primarily concerned with generation of new
knowledge. It is used to generate and expand theo-
Example of qualitative research
ries that describe, explain or predict a phenome-
A researcher is conducting a study on experiences non of interest to the discipline without regard to
of women who have undergone female feticide in its immediate use. In other words, it is a formal
selected rural and urban areas in Punjab. In this
and systematic process leading to the development
instance, the researcher is not handling any numeri-
cal data, rather he or she is collecting descriptive
of theories. There is no obvious immediate com-
narrations verbalized by study subjects. mercial value of the result that is generated through
basic research, as its main motivation is to expand
 Introduction to Medical Research 11

the body of knowledge. Basic research is usually

carried out for some of the following aims: RESEARCH PROCESS
l Gather and generate information. A research study begins with a problem or a ques-
l Expand the body of knowledge to improve tion, and ends with generation and dissemination
understanding about a discipline. of an answer for that problem or question. The
l Develop or refine theories and principles. research process is a sequential process, where
there are chances of overlapping or interchange-
ability of different steps. Furthermore, the steps in
Examples of basic research
quantitative and qualitative research processes are
• A researcher carries out a study on the effect of not similar; there exist differences in both the pro-
participatory culture in an organization on work
cesses. Therefore, they are discussed separately in
performance of employees. Here, the researcher
observes that workers’ performances have im- this chapter.
proved in those organizations that have participa-
tory culture. Thus, a researcher gives a basic theory, QUANTITATIVE RESEARCH PROCESS:
‘workers’ participation in management increases OVERVIEW
their performance’. Quantitative research is a formal, objective and
• A researcher may perform an in-depth study to systematic process for generating information
understand the process of adaptation to stress in about the world. Quantitative research is con-
a rural tribal population, without having explicit
immediate implications in mind.
ducted to describe new situations, events or con-
cepts. In quantitative research study, variables are
preselected and defined by the investigator, the
Applied Research data are collected and quantified (i.e., translated
into numbers), and then statistically analysed,
The term applied research refers to those studies
often with the view to establish cause-and-effect
that have functional purposes and practical use or
relationship among the variables. This process is
application. They focus on finding an immediate
an orderly way of dealing with a research problem,
solution to an existing problem. In this, the re-
where variables are generally studied in numerical
searcher scientifically collects data to be used in
(quantitative) form. The research process starts
the clinical, administrative or instructional area to
with the identification and formulation of a re-
find the solution to the problem, evaluate prac-
search problem and ends with the dissemination
tices and identify needs of the patient. Applied
of research findings, as shown in Fig. 1.2.
research may be conducted to resolve a number of
l Formulation of research problem: The fore-
issues, some of which are as follows:
l Solve problems
most step of a research process is to discover an
l Make decisions
interesting and research-worthy problem. It is
l Evaluate something of interest
such an important step that the whole process
l Develop something new for immediate use
can get wasted if the research problem is not
clearly understood or defined. Therefore, any
good research needs a good research problem. It
Example of applied research generally starts with a broad topic and later it is
A researcher has planned to conduct a study to as- narrowed down to a specific topic for the study.
sess the efficacy of two different antihypertensive For instance, a researcher wants to conduct a
drugs in patients with primary hypertension. Here, study on the morning sickness symptoms of
the researcher is interested to know the answer for antenatal mothers; this is the broad topic; later,
the question, ‘which of the following drugs is more it is delimited to a specific morning sickness
effective in the treatment of primary hypertension’.
The knowledge generated through this research will
symptom during the first trimester among ante-
have immediate implications. natal mothers. To formulate a topic (i.e., effects
of ginger on morning sickness among pregnant
12 Research Methodology and Biostatistics

1 2 3

Formulation of Determining study Review of

research problem objectives literature

6 5 4

Specifying the Selecting research Formulating hypothesis/

population approach /design assumptions

7 8 9

Developing tool(s) for Establishing ethical Conducting pilot study/

data collection considerations tryout of tool

12 11 10

Preparing data for

Data collection Sample selection
analysis

13 14

Analysis and Disseminating the

interpretation of data research findings

Fig. 1.2. Steps of the quantitative research process.

mothers in the first trimester) on interesting, There may be primary and secondary objectives
creative, noteworthy and indicative research for a research project.
problem, the researcher must ensure features of This step of the research process also includes
originality, solvability and feasibility of research writing operational definitions of the variables
problem as well as the need to take into account under study. These objectives and operational
the following dimensions: definitions not only help in delimiting the re-
 Substantive dimension: Is the research prob- search problem but also provide a defined direc-
lem clinically or theoretically important? tion to research process, where researchers have
 Methodological dimension: What is the best clarity about what to be studied under a particular
possible way to study this problem? research problem.
 Practical dimension: Are adequate resources l Review of literature: A literature review is a

available to conduct the study? summary of previous knowledge generated on

 Ethical dimension: Would this problem re- the topic of study. It is believed by researchers
spect the rights of the study subjects? that research cannot be conducted in a vacuum,
l Determining study objectives: There must be a bereft of the previous knowledge available, but
clear direction to every research project, and that it requires a foundation of existing knowl-
objectives certainly serve this purpose. There- edge to plan and conduct a good research. This
fore, researchers must have a list of objectives knowledge base can only be acquired through
for the study that provides the guidelines for the literature review from several sources, such as
researcher as to what exactly must be done dur- books, journals, research reports, unpublished
ing the course of a particular research study. theses, newspapers, magazines and other current
 Introduction to Medical Research 13

popular electronic information sources. Review observation of a phenomenon without manip-

of literature helps researchers to understand what ulation and control over variables.
is already known about a topic and what needs to Basically, there are two main designs for quanti-
be further investigated. An effective literature re- tative researches, i.e., observational and ex-
view needs certain basic skills in researchers, such perimental research designs. However, these
as critical thinking, brainstorming, reading, com- designs can be further subclassified into dif-
prehending, analysis, synthesizing and summa- ferent categories:
rizing. Literature review plays an important role - Observational research designs: Observa-
in the development of a research project; more- tional studies are researches where the
over, researchers develop greater insight into the study subjects are observed without im-
research problem and gain information on what posing any intervention or one can say that
has been already investigated regarding a particu- the research variables are observed without
lar topic under study. manipulating them in the natural setting
l Formulating hypothesis: Hypothesis is an as- for the purpose of description, exploration,
sumed statement suggesting an answer to a explanation or identification of the rela-
question, which may or may not be true. In tionship between two or more variables.
simple words, it translates the statement of the Observational studies could be further
problem into a clearly understandable and easy- classified as descriptive research study de-
to-comprehend prediction of what is expected signs (which could include case report,
to be the outcome of the study, which is either case series and cross-sectional studies),
accepted or rejected based on the empirical data case–control study design (backward look-
generated at the end of the research process. ing) studies, cohort study design (forward
For example, consider this research statement, looking) and ecological study design. The
‘A prospective study on incidence of cirrhosis case–control studies could be traditional
among alcoholics residing is a selected com- case–control studies or nested case–control
munity’. Here, the hypothesis will be ‘Alcohol- studies; similarly the cohort studies also
ics will have higher incidence of cirrhosis’. could be further classified as perspective
l Selecting research approach/design: A research cohort design, historical cohort design and
design is a systematic plan to obtain answers to ambispective cohort design.
research questions. In a broader sense, research - Experimental research designs: Experi-
design is the blueprint of research study, which mental studies involve the implementation
enables the researcher to know on whom, what, of intervention on an experimental group
when, where and how the study will be con- and outcome measurement of the experi-
ducted. ential and control group (if applicable).
 Who: Specifies the subject(s) under study, Experiential studies could be randomized
e.g., patients, families and groups. control trials (RCT) or quasi-experiential
 What: Specifies the variables that are to be research studies.
focused upon and measured. (a) RCT involves the random assignment of
 When: Specifies the time of the study, dura- subjects, an availability of a control
tion and frequency of the observations. group to compare with the experimental
 Where: Specifies the setting of the study, i.e., group, and manipulation of indepen-
where the researcher will be conducting the dent variable to observe effect on depen-
study. dent variable in an experimental group.
 How: Specifies how the data will be collected. (b) Quasi-experimental research studies in-
For example, through manipulation of the vari- volve manipulation of independent
ables in an experimental study under random variables to observe the effect on de-
assignment or simple questioning, interview or pendent variables, but usually they do
14 Research Methodology and Biostatistics

not exert complete control over extra- researcher can address ethical issues and may
neous variables in manipulation and follow the ethical considerations by taking the
randomization. Research studies on following measures:
humans, in either field or clinical set-  Taking informed consent from participants.

tings, make it impossible to strictly  Avoiding any violation of rights of research

imply the random assignment of sub- participants and errors in data collection.
jects or control of manipulation and  Obtaining the permission from competent

extraneous variables. Therefore, health authority to conduct the study.

care professionals mainly conduct  Doing justice and treating fairly the partici-

quasi-experimental research. pants of the study.

l Specifying the population: In quantitative re-  Maintaining confidentiality of the informa-

search studies, it is essential to learn about the tion and anonymity of subjects.
characteristics the subjects possess and clearly  Providing a fair opportunity to the partici-

define the group of subjects or objects under pants to withdraw from the study any time
study. Research population is an aggregate of all without facing any discrimination in the fu-
the subjects or objects with specific characteris- ture treatment/care.
tics. For example, in a study on ‘the prevalence of l Conducting pilot study/tryout of tool: Pilot
hospital-acquired infection among patients ad- study is a kind of small-scale rehearsal on the
mitted to intensive care units’, the population is subjects, but these subjects are not a part of the
patients admitted to intensive care units, where actual study. Pilot study is conducted to ensure
all the patients are living in a similar situation, the feasibility of the study and revise methodol-
i.e., they are all admitted to intensive care units. ogy and tool(s) of the study in case any short-
Population specification helps the researcher to coming is found.
plan and develop an effective methodology and l Determining the sample size and selecting the
tools for the development of empirical evidence. sample: It is not practically possible to conduct
l Developing tool(s) for data collection: This is study on the entire population. Therefore, re-
the most important and crucial step of the re- searchers must select a representative part of the
search process. It is essential to devise methods population. The number of participants in-
and tools to measure research variables as ob- cluded in the study is considered as sample size;
jectively and accurately as possible. The plan it should not be so small that the researcher can-
and design of the tools for data collection de- not generalize the research findings to the par-
pend on several factors, such as type of research ticular study population. Neither should the
design, variables, subjects, researcher’s exper- study sample be so large that it unnecessarily
tise, available resources and time for study. The wastes the resources. Therefore, sample size is
researcher could use existing standardized re- determined by using different statistical princi-
search tools or may develop new tools, which ples and computations (may refer to page
must be used after establishing their validity 133–142 for details). A sample can be selected
and reliability. In quantitative research, usually, by using either probability or nonprobability
structured or semistructured tools are used for sampling techniques, where the choice of the
data collection, which requires a lot of construc- sampling technique depends on several factors
tive efforts, such as review of relevant content, like nature of the study, availability of time,
brainstorming and expert suggestions. money, resources and researchers’ knowledge
l Establishing ethical considerations: In health about population, etc. However, researchers al-
sciences, most of the research is conducted ei- ways strive to select a representative sample
ther on humans or to some extent on animals, which adequately reflects the population’s trait.
where it becomes more important to strictly fol- l Data collection: It is the most time-consuming
low ethical principles. During research studies, a step of the research process that involves direct
 Introduction to Medical Research 15

or indirect interaction with respondents to QUALITATIVE RESEARCH PROCESS: OVERVIEW

gather information pertaining to the phenom- Qualitative research is a systematic and subjective
enon under study. It must be carefully planned approach to describe events or life experiences,
and implemented to collect relevant informa- and to give them intended meanings. Qualitative
tion by using preplanned methods, techniques research is an interdisciplinary, transdisciplinary
and tools of data collection. Data collection re- and sometimes counterdisciplinary field. It is
quires adequate planning, patience, communi- committed to the naturalistic perspective to the
cation, interpersonal relationship and recording interpretative understanding of human experi-
skills. Data could be collected through ques- ences. It not only helps to unearth hidden facts
tioning, interviewing or observation methods. but also aids in exploring attitude, emotions, be-
l Preparing data for analysis: In quantitative liefs, sensitive issues, opinions, concepts processes
studies, careful checking of every tool for its and relationships of people. The process of quali-
completeness and coding is the main activity tative research is relatively less formally planned,
during this step of the research process. It must where planning and execution of research moves
be ensured that one code specifies only one hand in hand. Therefore, the steps of the qualita-
piece of information, and it should be main- tive research process are slightly different from
tained carefully to avoid any error. Coding can that of quantitative research as shown in Fig. 1.3.
be carried out manually on a paper sheet or on l Identifying research problem area: In qualita-

a computer grading sheet (e.g., MS Excel spread- tive research, initially a broad research area is
sheet) or directly in statistical software. In this identified, and then focus may be narrowed
technology driven era, manual paper sheet cod- down as the researcher gets more familiarity and
ing has become an outdated method and most experience in the research setting. Qualitative
researchers prefer computer coding in MS Excel research begins with a general topic of the re-
spreadsheet. search problem and the group or population
l Analysis and interpretation of data: In quanti- expected to be studied. Furthermore, researchers
tative research studies, numerical data must be have less understanding about the phenomenon
organized in an orderly and sequential manner, and the population under study; therefore, it
and processing is required because research becomes difficult to have a formal plan of re-
questions cannot be answered in numerical search study in advance. For instance, in a quali-
form. In quantitative research, data may be ana- tative study, an initial statement of problem
lysed by using descriptive or inferential statis- could be ‘a study on the experiences of tsunami
tics. Furthermore, data may be analysed either survivors in coastal regions of the state of Tamil
by manual calculations or by using statistical Nadu, India’.
software programmes, like Statistical Program l Formulating broad study objectives: Objec-

for Social Sciences (SPSS). Data are presented tives serve as guidelines to conduct a research
through tables, graphs and charts to facilitate study; however, in qualitative studies, only
the interpretation of data. broad objects may be planned, which could
l Disseminating the research findings: Re- later be modified based on the need of the
search may fail in achieving its objectives, if research.
findings are not disseminated to its users. Re- l Review of literature: There are different schools

search findings may be communicated through of thoughts about literature review in qualitative
writing of a research thesis, article or present- studies. Some eminent researchers believe that
ing an oral research report at scientific profes- literature review in qualitative studies may be-
sional conferences. Research findings must be come a source of bias, since qualitative research is
communicated in a standardized format ac- a subjective approach of studying a phenome-
cording to international, national or institu- non. However, others argue that literature review
tional guidelines. is a mandatory step in every research process,
16 Research Methodology and Biostatistics

1 2 3

Identifying research Formulating broad Review of

problem area study objectives literature

6 5 4

Select a small Selecting research Entry in research

sample approach/design setting

7 8 9

Establishing ethical Plan tools for data

Data collection
considerations collection

12 11 10

Disseminating the Analysis and Organize data for

research findings interpretation of data analysis

Fig. 1.3. Steps of the qualitative research process.

where researchers may learn about the work action research and philosophical inquiry.
that has been already done on the particular phe- Choice of a research design in qualitative study
nomenon under study. Moreover, in qualitative depends on the nature of the phenomenon
studies, understanding the phenomenon or pop- under study. For example,
ulation under study is very minimal; therefore,  Phenomenological design: It may be used if

literature review would help the researchers gain the researcher is interested in a description of
some amount of prior information to have an experiences of people, e.g., a study on experi-
effective future planning and execution of the ences of people who survived in the tsunami
research project. disaster in the coastal states of India.
l Entry in the research setting: As in qualita-  Ethnographical design: It involves the collec-

tive study, the researcher is not much aware tion and analysis of data about cultural
about population and phenomenon, and thus groups. In ethnographic research, researchers
this is an important step for further planning live and become part of their culture to ex-
and execution of the research process. Entry plore the sociocultural practices and ritual
in research setting requires prior permission customs of the population under study,
from competent authorities as well as con- e.g., an ethnographic study on the features,
tacts with key people in the area of interest. critical attributes, processes and benefits
In ethnographic studies, these people are of self-help groups of women living with
called key informants. These key people may chronic alcoholic husbands in selected vil-
be the important sources of information lages of district Belgaum, Karnataka.
about the phenomenon and subjects under  Grounded-theory approach: This approach

study. may be used by researchers to collect data,

l Selecting research approach/design: There e.g., if they are interested in learning about
are a few common designs used in qualitative the experiences that the survivors went
studies, namely phenomenology, grounded the- through to resolve their crisis during the
ory, ethnography, case study, historical research, tsunami disaster.
 Introduction to Medical Research 17

 Case study: Case studies are in-depth exami- tape-recording, focus-group interviews, photo-
nation of people, objects or institutions, e.g., graphs, reflective journals, field notes, tran-
Modelling of Individual and Organizational scripts, cumulative diaries, and so on.
Factors Affecting Traumatic Occupational In- l Collecting data: In qualitative research, data col-

juries Based on the Structural Equation Mod- lection is the most crucial and time-consuming
eling: A Case Study in Large Construction step of the research process. Interview and ob-
Industries. servation are the two most common types of
 Historical research: This approach of qualita- data collection methods used in qualitative stud-
tive research is concerned with identifying, ies; they provide greater flexibility in data collec-
locating, evaluating and synthesizing data tion, which is of prime importance in qualitative
from the past, e.g., a study on the evolution of studies. Rapport building and confidence estab-
medical research in India. lishment are the two essential tasks a researcher
 Action research design: This design may be must accomplish before starting data collection.
used if, for example, the researcher is inter- Generally, data are collected either by tape-
ested in learning about how health care pro- recording the session, detailed notes on the in-
fessionals worked with tsunami survivors to formation, or both.
reach a resolution of the crisis situation. The use of focus-group interviews has also in-
l Selecting a small sample: Qualitative studies are creased in medical qualitative studies. This method
in-depth inquiries; therefore, generally a quite saves time, since a group of participants are inter-
small sample (10–15 subjects) is selected for the viewed together by a moderator on a particular
study. However, there are no defined set of rules topic. However, each participant may not get an
for exact sample size for a qualitative study; even equal opportunity to express or may be reluctant
a single case can be a sample in case study design to answer in a group.
of qualitative research. Generally the size of the l Organizing data for analysis: Before analysis,

sample in qualitative research is always smaller data are organized for data analysis and inter-
than the one in quantitative research. In qualita- pretation by the following techniques:
tive research, depth and quality of data are more  Listing

important than quantity of data.  Categorizing

l Establishing ethical considerations: Same as  Comparing

quantitative study, qualitative study also devel-  Laddering

ops plans to address the ethical aspects related l Analysis and interpretation of data: In qualita-

to the participant. This is a more important tive studies, analysis of data actually begins with
concern in qualitative study because of the more data collection. Some researchers argue that
intimate nature of the relationship that typically there is no separate step of analysis of data in
develops between the researcher and the study qualitative research. Analysis of data in qualita-
participant. Ethical issues can be addressed and tive research usually involves an examination of
established by taking informed consent from words rather than numbers, as is the case in
participants, avoiding errors in data collection, quantitative research. In qualitative studies,
doing justice to participants in analysing data large amount of data is collected in the form of
and maintaining the confidentiality of the in- words, which could be several thousand pages
formation and anonymity of the subject. of notes. Therefore, data analysis in qualitative
l Planning tools for data collection: Tools used studies is a challenging task.
for data collection in qualitative research are There are no universally accepted rules for anal-
generally semi-structured or unstructured. Most ysis of qualitative data. However, qualitative
of the items are open-ended, the answers of data can be analysed by using descriptive cod-
which should be taken in the form of narrations. ing, indexing, narrations, integrative diagrams,
Data may be collected through video-taping, thematic analysis, etc.
18 Research Methodology and Biostatistics

l Disseminating the research findings: Re- conferences. Research findings must be com-
search findings may be communicated by writ- municated in a standardized format according
ing a research thesis, article or presenting an to international, national or institutional
oral research report at scientific professional guidelines.

MULTIPLE CHOICE QUESTIONS

1. The most objective means of obtaining medical 7. The final step of the research process for a researcher
knowledge is through: is to:
(a) Trial and error (a) Analyse the data
(b) Tradition (b) Interpret the findings
(c) Scientific research (c) Communicate the findings
(d) Authority (d) Utilize the findings

2. Which form of reasoning is the process of drawing a 8. Which of the following is a characteristic of qualita-
specific conclusion from a set of premises? tive research?
(a) Rationalism (a) Generalization to the population
(b) Deductive reasoning (b) Random sampling
(c) Inductive reasoning (c) Unique case orientation
(d) Probabilistic (d) Standardized tests and measures

3. Which of the following is not true about the charac- 9. A process in which no one, including the researcher,
teristics of good research? can link subjects with the data they provide, is
(a) Orderly and systemic process known as:
(b) Conducted using large amount of funds (a) Confidentiality
(c) Finding solution to a problem (b) Connubiality
(d) Begin with clearly defined purposes (c) Unknown data collection
(d) Anonymity
4. Basic research is conducted with which of the fol-
lowing as the aim? 10. A meeting with research participants that ensures
(a) To make a decision their understanding of the reasons and the justifi-
(b) To develop knowledge for immediate use cation for the procedures used in the study is
(c) To develop or refine theories and principles known as:
(d) To provide a solution for new problems (a) Debriefing
(b) Research meeting
5. The development of a solid foundation of reliable (c) Explanation
knowledge is typically built from which type of research? (d) Justification meeting
(a) Basic research
(b) Action research 11. Which of the following is ‘least’ likely to be provided
(c) Evaluation research for subjects in a research study?
(d) Orientational research (a) Informed consent
(b) Anonymity
6. Quantitative research is a process to: (c) Confidentiality
(a) Evaluate theories and hypothesis (d) Singularity
(b) Develop theories and hypothesis
(c) Describe social phenomenon
(d) Obtain experiences, feelings and beliefs
 Introduction to Medical Research 19

12. What is the name of the process used to obtain a 14. Which of the following is only the step of qualitative
9-year-old child’s agreement for participation in research process?
the study? (a) Entry in research setting
(a) Assent (b) Selecting a large sample size
(b) Consent of a minor (c) Developing structured tool for data collection
(c) Minor agreement (d) Data analysis using inferential statistics
(d) None of the above

13. Which of the following is not a step of quantitative

research process?
(a) Formulation of research problem
(b) Review of literature
(c) Formulation of hypothesis
(d) Meta-analysis of research study

Answers of Multiple Choice Questions

1. c; 2. b; 3. b; 4. c; 5. a; 6. a; 7. d; 8. c; 9. a; 10. a;, 11. d; 12. a; 13. d; 14. a.

REFERENCES
Arslanian-Engoren, C., & Scott, L. D. (2003). Issues in pulmonary Center for Disease Control and Prevention. (n.d.). The Tuskegee
nursing. The lived experience of survivors of prolonged me- timeline. ,http://www.cdc.gov/nchstp/od/tuskegee/time.htm/.
chanical ventilation: A phenomenological study. Heart & Lung: Accessed 18.04.06.
Journal of Acute Critical Care, 32(5), 328–334. Cesario, S., Morni, K., & Santa-Donato, A. (2002). Evaluating the
Ayres, L., Kavanaugh, K., & Knafl, K. A. (2003). Within-case and level of evidence of qualitative research. Journal of Obstetric
across-case approach to qualitative data analysis. Qualitative Gynecologic, and Neonatal Nursing, 31(2), 708–714.
Health Research, 13(6), 871. Clover, Vernon T., Balsley, H.L. (1979) Business Research meth-
Bayley, E. M., MacLean, S. L., Desy, P., & McMahon, M. (2004). ods. 2d ed. Columbus, OH: Grid.
ENAs Delphi study on national research priorities for emer- Code of Ethics and Professional Conduct. Indian Nursing Council;
gency nurses in the United States. Journal of Emergency Nurs- 2006.
ing, 30, 12–21. Daly, P. (2005). Mothers living with suicidal adolescents: a phe-
Blythe, J., & Royle, J. A. (1993). Assessing nurses’ information nomenological study of their experience. Journal of Psycho-
needs in the work environment. Bulletin of the Medical social Nursing and Mental Health Services, 43(3), 22–28.
Library Association, 81, 433–435. DeSantis, L., & Ugarriza, D. N. (2000). The concept of theme as
Bonifazi, W. L. (2004). Cruelty and courage: nurses in the Nazi used in quantitative nursing research. Western Journal of Nurs-
era. Nurse Week November, 6, 24–26. ing Research, 22, 351–372.
Bostrom, J., & Suter, W. N. (1993). Research utilisation: making DiCenso, A., Cullum, N., & Ciliska, D. (1998). Implementing
the link to practice. Journal of Nursing Staff Development, 9, evidence-based nursing: some misconceptions. Evidence-Based
28–34. Nursing, 1, 38–40.
Bowker, R. R. Ulrich’s international periodicals directory. New York, DiCenso, A., Guyatt, G., & Ciliska, D. (2005). Evidence-based nursing:
NY: Bowker, 1996. a guide to clinical practice. St. Louis, MO: Mosby.
Burns, N., & Grove, S. K. (2005). The practice of nursing research: Donalek, J. G, & Soldwisch, S. (2004). An introduction to qualita-
conduct, critique and utilization (5th ed.). St. Louis, MO: Elsevier. tive research methods. Urologic Nursing, 24, 354–356.
Burns, N., & Grove, S. K. (2008). Understanding nursing Donalek, J. G. (2005). The interview in qualitative research. Urologic
research: building and evidence-based practices (4th ed.). Nursing, 25, 124–125.
St. Louis, MO: Elsevier. Erlen, J. A. (2000). Conflict of interest: an ethical dilemma for the
Caelli, K. (2003). The changing face of phenomenological re- nurse researcher. Orthopedic Nursing, 19, 74–77.
search: traditional and American phenology in nursing. Quali- Estabrooks, C. A. (1998). Will evidence-based nursing practice make
tative Health Research, 10(3), 366–377. practice perfect? Canadian Journal of Nursing Research, 30, 15–36.
Calmorin, L. P., & Calmorin, M. A. (1995). Methods of research and Fitzpatrick, J. J. (2004). Making a commitment to go public
thesis writing (p. 390). Manila: Rex Book Store. (editorial). Applied Nursing Research, 17, 223.
20 Research Methodology and Biostatistics

Flemming, K. (1998). Asking answerable questions. Evidence- Melney, B. M., & Fineout-Overholt, E. (2005). Evidence-based
Based Nursing, 1, 36–37. practices in nursing and health care: a guide to best practice.
Heater, B. S., Becker, A. M., & Olson, R. (1988). Nursing interven- Philadelphia, PA: Lippincott Williams & Wilkins.
tions and patient outcomes. A meta-analysis of studies. More, J. M. (2005). Evolving trends in qualitative research: ad-
Nursing Research, 37, 303–307. vanced in mixed-method design. Qualitative Health Research,
Houser J. (2011). Nursing research: reading, using and creating 15, 583–585.
evidence. Sudbury: Jones and Bartlett Publishers. Munhall, P. L. (2001). Nursing research: A qualitative perspective
Indian Council of Medical Research. Ethical guidelines for bio- (3rd ed.). Boston: Jones and Bartlett.
medical Research on human participants. (2006). ,www.icmr. National Institute Nursing Research. (n.d.). About the NINR.
nic.in. Accessed 17.10.13. January 16, 2006. ,http://ninr.nih.gov/ninr/about.html.
Jacelon, C. S., & O’Dell K. K. (2005). Case and grounded theory as Accessed 10.10.06.
qualitative research methods. Urologic Nursing, 25, 49–52. Nightingale, F. (1859). Notes on nursing: what it is and what it is
Jacobson, S. F., Chu, N. L., Pascucci, M. A., & Gaskins, S. W. (2004). not. Philadelphia, PA: Lippincott Williams & Wilkins.
Characteristics of nursing research race, ethnicity and culture. Nightingale, F. (1946). Notes on nursing what it is and is not. London:
Journal of Multicultural Nursing & Health, 10(3), 6–12. Churchill Livingstone. (First published in 1859)
Jadad, A. R., & Haynes, R. B. (1998). The Cochrane Collaboration – Pearcey, P. A. (1995). Achieving research-based nursing practice.
advances and challenges in improving evidence-based decision Journal of Advanced Nursing, 22, 33–39.
making. Medical Decision Making, 18, 2–9. Polit, F. D., & Beck, C.T. (2008). Nursing research: generating and as-
Kader, P. (1997). Nursing research: principles, process and sessing evidence for nursing practices (8th ed.). Philadelphia, PA:
issues. New York, NY: Palgrave Macmillan. Lippincott Williams & Wilkins.
Kerlinger, E. N., & Lee, H. B. (2000). Foundation of behavioral Polit, F. D., & Hungler, B. P. (1995). Nursing research: principles
research (4th ed.). Fort Worth, TX: Harcourt. and methods (5th ed.). Philadelphia. PA: J. B. Lippincott &
Ketefian, S. (1975). Application of selected nursing research find- Company.
ings into nursing practice: a pilot study. Nursing Research, 24, Royle, J. A., Blythe, J., Ingram, C., DiCenso, A., Bhatnager, N., &
89–92. Potvin, C. (1996). The research utilization process: the use of
Lake, E. T. (2006). Multilevel model in health outcome research: guided imagery to reduce anxiety. Canadian Oncology Nursing
part-I, theory, design and measurement. Applied Nursing Re- Journal, 6, 20–25.
search, 19, 51–55. Sackett, D. L., Rosenberg, W., Gray, J. A. M., & Haynes, R. B.
Langford, R. (2001) Navigating The Maze Of Nursing Research. (1996). Evidence-based medicine: what it is and what it isn’t.
St. Louis: Mosby. The British Medical Journal, 312, 71–72.
Luker, K. A., & Kenrick, M. (1992). An exploratory study of the Saini, R., & Singh, S. Clinical trials in India: need for bioethics.
sources of influence on the clinical decisions of community Chandigarh: The Tribune, Monday, August 9, 2010.
nurses. Journal of Advanced Nursing, 17, 457–466. Scott, K., & McSherry, R. (2009). Evidence based nursing: clarify-
Macnee, C. L. (2004). Understanding nursing research: reading and ing the concepts for nurses in practice. Journal of Clinical Nurs-
using research in practice. Philadelphia, PA: Lippincott Williams ing, 18(8), 1085–1095.
& Wilkins. Stommel, M., & Wills, C. E. (2004). Clinical research: concepts and
McKibbon, K. A., & Marks, S. (1998). Searching for the best evidence. principle for advanced practice nurses. Philadelphia, PA:
Part 1: where to look. Evidence-Based Nursing, 1, 68–70. Lippincott Williams & Wilkins.
McKibbon, K. A., & Marks, S. (1998). Searching for the best evi- Streubert, H. J., & Carpenter, D. R. (2002). Qualitative research
dence. Part 2: searching CINAHL and Medline. Evidence-Based in nursing: advancing the humanistic imperative (3rd ed.).
Nursing, 1, 105–107. Philadelphia, PA: Lippincott Williams & Wilkins.
Meah, S., Luker, K. A., & Cullum N. A. (1996). An exploration Whittemore, R. (2005). Combining evidence in nursing research:
of midwives’ attitudes to research and perceived barriers to methods and implications. Nursing Research, 54(1), 56–62.
research utilisation. Midwifery, 12, 73–84.
CHAPTER OUTLINE
Research Problem
and Hypothesis 2
RESEARCH PROBLEM HYPOTHESIS
• Meaning of Research Problem • Meaning of Hypothesis
• Identification of a Research Problem • Importance of Hypotheses in Research
• Sources of a Research Problem • Characteristics of a Good Hypothesis
• Criteria for a Good Research Problem • Sources of Hypotheses
• Formulating a Problem Statement • Types of Hypotheses
VARIABLES
• Formulating Hypotheses

DELIMITATIONS
OPERATIONS DEFINITIONS
MULTIPLE CHOICE QUESTIONS
WRITING RESEARCH OBJECTIVES

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Describe meaning, characteristics, needs and types of
• Define a research problem or research question. research objectives.
• Determine the sources of a research problem. • Demonstrate the ability to formulate the research objectives.
• Recognize the criteria for selecting a good research problem. • Define hypothesis and delimitation.
• Demonstrate skills in formulating the research problem. • Determine the sources of hypothesis and classify the
• Explain the concept of variables and classify the different hypotheses.
types of variables. • Recognize the importance and characteristics of hypothesis.
• Discuss the meaning and essential characteristics of • Enumerate and explain the sources of hypotheses.
operational definition. • Demonstrate skills in formulation of hypotheses.

KEY TERMS

• Bivariate study  30 • Operational definition  30 • Directional hypothesis  39

• Critical appraisal  23 • Research objectives  32 • Hypothesis  34
• Declarative  28 • Research problem  22 • Nondirectional hypothesis  39
• Demographic variable  29 • Research question  24 • Null hypothesis  39
• Dependent variable  29 • Research variable  29 • Objectivity  37
• Ethical considerations  26 • Univariate study  29 • Research hypothesis  39
• Extraneous variable  29 • Associative hypothesis  38 • Simple hypothesis  38
• Independent variable  28 • Causal hypothesis  39 • Specificity  37
• Interrogative format  28 • Complex hypothesis  38 • Testability  37
• Multivariate study  30 • Delimitations  41
22 Research Methodology and Biostatistics

We must find out what to do’, i.e., the solution can

RESEARCH PROBLEM be found out only after an investigation.
A research problem is a question that a re- In other words, ‘a research problem is an area
searcher wants to answer or a problem that a of concern where there is a gap in the knowledge
researcher wants to solve. Identification and for- base needed for professional practices’.
mulation of a research problem is the first step of The research problem and problem statement
the research process. However, it is considered as are frequently seen as synonymous, but these two
one of the most challenging and difficult phase aspects are really quite different. A problem state-
of any research project. It is believed that selec- ment is not merely a research problem or purpose
tion of a good research problem is a discovery of the study, but a well-structured formulation
in itself. of what actually the problem is, what is not well
Selection of a research problem depends on understood and what is lacking, which is to be
several factors such as the researcher’s knowl- discovered to solve the problem. A problem state-
edge, skills, interest, expertise, motivation and ment, especially in quantitative studies, generally
creativity with respect to the subject of inquiry. has six components:
In addition, a researcher needs to ensure that a l Relevance of the study

selected problem has high significance and impli- l Title of the study

cation for his or her profession. It should also be a l Operational definitions of the variables

suitable, feasible, testable and solvable research l Objectives of the study

problem. l Delimitations of the study

It is believed that most good research studies l Scope and limitation of the study

need lots of time for selection of a research prob- The problem statement actually articulates the
lem. Initially, every researcher faces the predica- problem to be addressed and indicates the need for
ment of identifying, selecting and formulating a the study through the development of an argu-
good research problem. This chapter will answer ment. In other words, the problem statement pres-
these pertinent questions. ents the topic under study, provides a rationale for
the choice of topic, represents a synthesis of fact
MEANING OF RESEARCH PROBLEM and theory, and directs the selection of design.
A research problem is an enigmatic, troubling
area or topic that needs an in-depth enquiry to IDENTIFICATION OF A RESEARCH PROBLEM
get a better understanding or an ultimate solu- Identification of a research problem is the first
tion. For example, a solution to female feticide in and most important step in the research process.
India or finding solutions to the increasing inci- Generally, a broad area is selected and then a
dence of hospital-acquired infections could be broad topic is delimited or narrowed down to a
enigmatic or troubling topics of the present time, specific one-sentence statement of the problem.
and could become important research problems This step of the research process is considered as
for a biomedical researcher; there could be many the most difficult and challenging, and needs lots
more such problems that a researcher may con- of time.
sider to study. A research problem may come from several
According to Kerlinger, ‘A problem is an inter- sources; however, basically, it depends on the in-
rogative sentence or statement that asks what rela- terest of the researcher. It is generally believed by
tion exists between two or more variables. The researchers that their research topics are to be
answer to questions will provide what is being thought of by themselves. However, that is not
sought in the research.’ true; since no one lives in a vacuum, we would
R.S. Woodworth defines problem as ‘a situation expect that the ideas a researcher collects to iden-
for which we have no ready and successful re- tify a research problem are influenced by a variety
sponse by instinct or by previous acquired habit. of sources.
 Research Problem and Hypothesis 23

could be rich sources of ideas to identify a signifi-

Sources of a Research Problem cant research problem.
The common sources from which a researcher l Critical appraisal of literature: When we criti-
may find ideas to identify and formulate a re- cally study books and articles relating to the
search problem include (Fig. 2.1): subject of our interest, including research re-
l Personal experiences: The day-to-day personal ports, opinion articles and summaries of clini-
experience of a researcher may serve as a good cal issues, pertinent questions may arise in our
source of ideas to formulate a research problem. mind. These may strike our mind indirectly by
For example, a researcher observes domestic stimulating imagination and directly by stating
violence suffered by wives of alcoholic hus- what additional research is needed.
bands. This experience may provide ideas to For example, a health care professional reads an
identify several research problems related to article on the prevalence of pin site infection
domestic violence against women. There may among patients with external fixators. While
be many such life experiences which could be reading this article, he/she learns that there is
used to develop a research problem. lack of consensus about pin site care. This
l Practical experience: Health care professionals information may serve as a basis to formulate
get plenty of ideas to formulate research problems a research problem.
from their clinical experiences. Every curious Another type of literature that acts as a source
health care worker has several questions that are of good research ideas is the requests for
encountered during clinical experience and that proposals (RFPs) that are published by gov-
need to be answered. For example, a health care ernment agencies and some companies. Typ-
worker finds that unrestricted visiting hours in ically, RFPs describe the problems that need
surgical wards reduces the analgesic demand addressing, the contexts in which they oper-
among postoperative patients. In another in- ate, the approach they would like you to take
stance, a resident doctor observes that application to investigate and address the problem, and
of ice at the site of a heparin injection reduces the the amount they would be willing to pay for
chances of ecchymosis. Such clinical experiences such research.
l Previous research: A body of knowledge should
be developed on a sound foundation of research
findings. Usually at the end of a research, fur-
ther research problems are suggested on the
Personal basis of the shortcomings of previous research,
experience Practical which can be investigated. In the Indian sce-
Consultations experience
nario, health care discipline still has a lot of is-
sues that need to be researched; therefore, this
Brainstorming Critical appraisal profession needs researchers who are willing to
of literature
replicate or repeat other studies on different
Sources of
a Research samples and settings where all the essential ele-
Previous Problem Exposure to field ments of the original study are held intact. Fur-
research situations ther refinements may be made in the existing
treatment methods, or more appropriate out-
Existing Social come measures may be identified.
theories issues l Existing theories: Research is a process of the-
Intuitions
ory development and theory testing. Health care
professionals use many theories from other dis-
ciplines in their practices. If an existing theory is
Fig. 2.1. Sources of a research problem. used in developing a researchable problem, a
24 Research Methodology and Biostatistics

specific statement from the theory must be iso- with other members of the health care team,
lated. Generally, a part or parts of the theory are researchers and clinicians.
subjected to testing in the clinical situation. The l Intuition: Traditionally, intuitions are consid-
testing of an existing theory is definitely needed ered good sources of knowledge as well as
in health care-specific application to the particu- sources to find new research problems. It is be-
lar scenario; therefore, they serve as good sources lieved that the reflective mind is a good source
of research problems. of ideas, which may be used to formulate a good
l Consumer feedback: Research problems may be research problem.
generated from the results of activities aimed at l Folklores: Common beliefs could be right or
soliciting patient feedback. For example, at the wrong. For example, it is generally believed that
time of discharge of patient after angiography, a studying just before a test decreases the score.
health care worker obtains a feedback from the We believe that we should not study just before
patient. The patient verbalizes that it was a won- a test to relax our mind. Researchers can con-
derful experience except for the discomfort dur- duct a research study on whether one should
ing removal of the pressure dressing at the femo- study before the test or not.
ral puncture site. This feedback provides a l Exposure to field situations: During field expo-
concept for health care workers to research on sure, researchers get a variety of experiences,
the efficacy of available alternative means of which may provide plenty of ideas to formulate
dressing at puncture sites to minimize the dis- research problems. For example, while working
comfort among patient undergoing angiography. in the field, a researcher observes a specific tra-
l Performance improvement activities: Perfor- ditional practice for cure of a disease condition,
mance improvement activities, also known as which can be used as a research problem to in-
quality improvement activities, are used to im- vestigate its efficacy.
prove processes and outcomes to meet regula- l Consultations with experts: Experts are believed
tory requirement. In the process of performance to have sound experience of their respective field,
activities, several issues immerge that require which may suggest a significant problem to be
answers through research. Thus, performance studied. In addition, experts may help in finding
improvement activities also serve as an impor- a current problem of discipline to be solved,
tant source of research problems. which may serve as a basis for formulation of a
l Social issues: Sometimes, topics are sug- research problem.
gested by more global contemporary social
or political issues of relevance to the health CRITERIA FOR A GOOD RESEARCH
care community. For example, HIV/AIDS, PROBLEM
female feticide, sexual harassment, domestic One commonly used acronym to define the criteria
violence and gender equality in health care for a good research problem or question is FINER,
and in research are some of the current social where F stands for feasible, I for interesting, N for
and political issues of concern for health care novel, E for ethical and R for relevant. However,
professionals. An idea for a study may stem there are many factors that should be considered
from a familiarity with social concerns or while deciding credibility of a particular research
controversial social issues. question for a scientific investigation. Some of the
l Brainstorming: Brainstorming sessions are good most essential factors are discussed here:
techniques to find new questions, where an in- l Significance to the profession: A problem that

tensified discussion among interested people of a researcher is selecting should have significance
the profession is conducted to find more ideas to to the profession or it will not serve any pur-
formulate a good research problem. For example, pose. A research problem is significant for a par-
ideas for studies may emerge from reviewing re- ticular profession when it is directed to develop
search priorities by having a brainstorming session or refine the body of professional knowledge.
 Research Problem and Hypothesis 25

A research problem could be considered signifi- study, an accurate determination of the

cant for a particular profession if it fulfils the needed equipment and supplies should be
following criteria: ensured. Some questions that should be an-
 Benefits the health care profession and pa- swered before beginning a research project
tients, and health care fraternity through the include the following:
study - What is the equipment that will be needed?
 Improves good clinical practices through the - Is this equipment available and in proper
results working order?
 Promotes theory development or testing - Is there a qualified operator of the equip-
 Provides solutions to current health care ment?
practice needs - Are the necessary supplies available or can
 Generates information that have practical they be obtained?
implications for the health care profession If the researcher takes into consideration
l Original: It is fundamentally considered that equipment and supplies in the early
every research problem should be new and phases of a research project, there are less
unique in itself. Therefore, it is the key respon- chances of the project being revised or
sibility of a researcher that an innovative knowl- discarded later because of equipment or
edge is used for selecting a research problem, supply problems.
so as to extend the growth of the existing body  Administrative support: Many research proj-
of knowledge in a profession. ects require administrative support. The health
l Feasible: Feasibility is an essential consideration care researcher may find it very difficult to
of any research project. Regardless of how signifi- conduct research independently. Financial as
cant or researchable a problem may be, its feasi- well as psychological support from administra-
bility in reference to time, availability of subjects, tion is very helpful. Knowing that your superi-
facilities, equipment and money, and ethical con- ors support your research efforts can be a very
siderations should be checked. This checking will powerful motivating force.
help the researcher decide whether the selected  Peer support: Many research ideas have never
problem is appropriate or inappropriate and been developed because potential researchers
study can be actually carried out or not. received no support from their peers. One of
 Time: A health care professional might be the best ways to determine a researchable
interested in studying sibling relationships problem is through interactions and discus-
among quintuplets. Knowledge of the inci- sions with other health care professionals. A
dence of quintuplet births would certainly climate of shared interest in health care re-
discourage anyone considering research on search is essential among the members of the
this particular population unless the re- health care profession.
searcher plans to make this a lifetime project.  Availability of subjects: A researcher may
So time is always a factor to be considered. It believe that study subjects are readily avail-
is wise to allow more time than seems to be able for the study. But this may not be the
needed because unexpected delays frequently case. Potential subjects may not meet the
occur. study criteria, may be unwilling to participate
 Cost: All research projects cost money; some or may already be participating in other stud-
studies are much more expensive than others. ies. Therefore, availability of subjects must be
The researcher must consider realistically the ensured well in advance.
financial resources available.  Researcher’s competence: A research problem
 Equipment and supplies: All research projects can only be feasible if it is in accordance with the
require some type of resources. Therefore, researcher’s competence, where the researcher is
before making the final decision to conduct a capable of handling a given research problem.
26 Research Methodology and Biostatistics

 Ethical considerations: A researcher must en- l Interesting: A research problem can only be
sure that the research problem can be approved considered good if it is in accordance with the
by the ethical committee without undue hur- researcher’s field of interest. A research prob-
dles. A very important topic of research cannot lem must be as per the motivation of the
be considered a feasible unit unless it is in researcher and should be fascinating to the
accordance with ethical guidelines. researcher, so that the research is conducted
l Solvable/researchable: The problem selected is with full enthusiasm and not merely for its
considered good only if it is solvable so that the accomplishment.
chances of insolvability of the problem can be
minimized. Thus, relevant results can be en- FORMULATION OF A RESEARCH PROBLEM
hanced. For example, a researcher selects a re- Formulation of a research problem is a complex
search problem to know the existence of God in and long process which includes the following
this universe. These sorts of problems are am- steps (Fig. 2.2):
biguous and impossible to solve. Therefore, the 1. Selection of a research area: Formulation of a
researcher must ensure that the research prob- research problem begins with selection of a
lem selected for study is solvable. broad research topic from personal experience,
l Current: A good research problem must be based literature, previous research and theories in
on the current problems and needs of a profes- which the researcher is interested and has sig-
sion, so that results generated will be of more use. nificance for the health care profession. For
Furthermore, more number of professionals example, a researcher gets an idea to conduct a
will be interested in the research conducted on study on female feticide. Therefore, he or she
current issues of their profession. initially begins with this broad research topic.

STEPS EXAMPLE

Selection of a research Initially a researcher selects a broad area for

topic study such as Hospital-Acquired Infection

To know what already has been done in this

Reviewing literature area, and to develop an original and unique
and theories research problem, an extensive review of
literature and theories is done at this stage

HAI is a very broad area so at this stage, the

Delimiting the research
researcher delimits the topic from Hospital-
topic
Acquired Injection to the incidence of VAP

At this stage, the researcher evaluates the

Evaluating the research
research problem for feasibility and assesses the
problem
substantive, methodological and ethical dimensions.

Finally, a statement of the problem is formulated

keeping in mind the basic components of the
Formulating the final statement
research problem. Example, ‘A descriptive study on
of the research problem
the incidence of VAP among patients on MV
admitted in the ICU of AIIMS, Rishikesh’.

VAP: Ventilator-Associated Pneumonia; HAI: Hospital-Acquired Infection; MV: Mechanical Ventilator

Fig. 2.2. Steps for formulation of a research problem.

 Research Problem and Hypothesis 27

2. Reviewing literature and theories: After get- and prevention of female feticide in selected
ting a broad idea for research, he or she needs rural communities of district Shimoga, Karna-
to review nursing literature and theories. Lit- taka. In this stage, the researcher clearly identi-
erature is reviewed to know what has already fies the variables, population and setting of the
been done in this selected area of research. research study. Furthermore, at this stage, the
Therefore, a researcher can plan a research researcher is quite clear about the phenome-
topic to further expand the existing body of non to be studied, where, and on whom.
knowledge in the respective area of research. 4. Evaluating the research problem: Once the
Review of existing theories provides an oppor- researcher is clear about the specific research
tunity for the health care researcher to plan a problem, next the research problem must be
research problem to contribute towards either carefully evaluated for its significance, research-
testing or development of a theory/conceptual ability and feasibility. Feasibility of the research
model. problem should be evaluated for time, cost,
3. Delimiting the research topic: In this step, the availability of subjects and resources, adminis-
researcher proceeds from a general research trative and peer support, ethical consideration
area of interest to a more specific topic of re- and researcher’s competence and interest.
search to conduct a study. For example, ini- Guidelines for evaluations of a research prob-
tially, a researcher decides to conduct a study lem may be referred from Table 2.1.
on female feticide; later, in this stage, the re- 5. Formulating the final statement of the research
searcher limits it to a specific research topic: A problem: After establishing the significance,
study on the perception of women about causes researchability and feasibility, the researcher

Table 2.1. Guidelines for evaluation of a research problem

Substantive Issues
Does the problem have significance to the health care profession?
Does the problem spell out the purpose of conducting the study?
Does the problem have any relationship with existing theories and models?
Does the problem flow from prior scientific information and/or experience in the topic area?
Methodological Issues
Does the problem statement clearly identify the research variables in measurable terms?
Does the problem identify the nature of population which is to be studied?
Does the problem clearly define the type or design of the study?
Does the problem describe the research settings, place and time?
Does the problem comprise appropriate phrases?
Stylistic Issues
Is the problem statement complete, concise and clear in a declarative form and clearly suggesting answers to
be questioned?
Is the problem statement introduced promptly?
Ethical Issues
Does the research problem require the use of human subjects?
Does the problem imply potential risks for those subjects?
Practical Issues
Has the researcher appropriately delimited the scope of the problem, or is the problem too big or complex for
a single investigation?
Is the researcher able to manage the resources, people, money, material and appropriate permissions from
authorities?
28 Research Methodology and Biostatistics

finally formulates a final statement of the re- This is an example of a declarative form of the
search problem. A statement of the research statement of a problem, where it is precisely and
problem could be in declarative or interrogative concisely stated and includes most of the required
format. components of a research statement, which are as
 Declarative format: In this format, a research follows:
problem is stated in declarative statement, e.g.: l Research study design: Descriptive

- A descriptive study on the prevalence of l Variable(s): It has only one research variable

bedsores among unconscious patients ad- ‘prevalence of pin site infection’

mitted in the intensive care units of l Population: Patients with external skeletal fixation

PGIMER, Chandigarh l Research setting: Orthopaedic wards of Nehru

- An exploratory study on contributing fac- Hospital, PGIMER, Chandigarh

tors of maternal mortality in selected rural
communities of district Ludhiana, Punjab
 Interrogative format: In the interrogative
VARIABLES
format, a research problem is stated in ques- Variables are qualities, properties or characteris-
tion form, e.g.: tics of persons, things or situations that change or
- What is the influence of the level of haemo- vary. Chinn and Kramer stated that variables are
globin on pin site infection among patients concepts at different levels of abstraction that are
with external skeletal fixators admitted in the concisely defined to promote their measurement
orthopaedic wards of AIIMS, New Delhi? or manipulation within a study. Variables are clas-
- What is the effect of ginger on morning sified on the basis of their nature, actions and
sickness symptoms on pregnant women in effects on other variables. The main types of the
selected urban communities of district variables are as follows (Fig. 2.3):
Hissar, Haryana? l Independent and dependent variables: These

The choice of either of these two types of format are two variables that are interrelated and
for formulation of a research problem de- mainly observed in analytical studies, i.e. ex-
pends on the researcher’s preference and insti- perimental studies, case–control studies and
tutional policies; however, declarative format cohort studies.
is much more popular among researchers.  An independent variable: It is a stimulus or

Whether declarative or interrogative, for the activity that is manipulated or varied by the
formulation of a research problem, it is prefer-
able if it fulfils the following features:
- The research problem is clearly, precisely
Independent
and concisely articulated. variables
- It clearly states the variables, population
and research setting under study.
- Variables are expressed in measurable Extraneous Dependent
terms. variables variables
- The type of study also may be included in Types of
research
the statement of the research problem. variables

Example of a Well-Formulated Research

Problem
Demographic Research
‘A descriptive study on the prevalence of pin site variables variables
infection among patients with external skeletal
fixation admitted in the orthopaedic wards of
Nehru Hospital, PGIMER, Chandigarh.’ Fig. 2.3. Types of variables in Research.
 Research Problem and Hypothesis 29

researcher to create an effect on the depen- Examples of variables along with the state-
dent variable. ment of research problems in different types
 A dependent variable: It is the outcome or of quantitative and qualitative research
response due to the effect of the independent studies are presented in Tables 2.2 and 2.3,
variable, which the researcher wants to pre- respectively.
dict or explain. l Extraneous variables: Extraneous variables are

For example, ‘A quasi-experimental study on the the factors that are not part of the study but may
effect of 4-hourly chlorhexidine mouthwash affect the measurement of the study variables.
on prevention of ventilator-associated pneu- These variables are usually recognized and con-
monia (VAP) among patients admitted in se- trolled by researchers in quasi-experimental and
lected ICUs of DMCH, Ludhiana, Punjab’. In experimental studies; however, in other research
this study, 4-hourly chlorhexidine mouthwash studies also, the researcher may do so wherever it
is an independent variable and ventilator- is possible. Generally, specific research design and
associated pneumonia (VAP) is the dependent sample inclusion and exclusion criteria are used
variable. to control the influence of extraneous variables.
l Research variables: In descriptive, exploratory, For example, ‘A study is conducted to assess the
comparative and qualitative research studies, effect of two different pin site care protocols
variables are observed or measured in the natu- on prevention of pin site infection among
ral setting as they exist, without manipulating patients with external skeletal fixation’. In
or imposing the effect of intervention or treat- this study, pin site care protocols are inde-
ment. Here no independent variable is manipu- pendent variables and pin site infection is the
lated and no cause-and-effect relationship is dependent variable. However, the dependent
examined; these variables are considered as re- variable, pin site infection may also be influ-
search variables. Therefore, research variables enced by other factors such as low haemoglo-
can be defined as qualities, attributes, properties bin level or higher blood sugar level among
or characteristics that are observed or measured these patients; these factors are considered as
in a natural setting without manipulating and extraneous variables, which may have an un-
establishing cause-and-effect relationship. wanted effect on the dependent variables or
For example, in a research study, ‘A cross-sectional research variables. In this instance, these
study on factors contributing to sleep distur- extraneous variables may be controlled by
bance among patients admitted in selected in- excluding patients with anaemia or diabetes.
tensive care units of AIIMS, New Delhi’, a factor Research studies may have one, two or many vari-
contributing to sleep disturbance is a research ables under study; on the basis of the number of
variable, which is observed in the natural set- variables under study, research studies are called
ting without manipulating it. univariate study (one variable is studied), bivari-
l Demographic variables: In most research stud- ate study (two variables studied) and multivariate
ies, researchers make an attempt to study sam- study (more than two variables are studied
ple characteristics and present them in research together).
findings. In addition, sometimes, researchers l Univariate study: In descriptive and compara-

even try to establish relations between the de- tive studies, generally a single variable is stud-
mographic variables and the research variables. ied. In these studies, this variable is called the
These characteristics and attributes of the study research variable, e.g., ‘A descriptive study on the
subjects are considered as demographic vari- biophysical profile of nursing personnel work-
ables. Common demographic variables are age, ing in the outpatient department of selected
gender, educational status, religion, social class, hospitals of Bangalore’. This is a univariate
marital status, habitat, occupation, income, study, which includes only one research variable
medical diagnosis, etc. (biophysical profile).
30 Research Methodology and Biostatistics

Table 2.2. Examples of statement of research problems and variables in quantitative research studies
Type of Study Example of the Research Problem Statement Variables
Descriptive A descriptive study on the prevalence of Research variable: prevalence
anaemia among adolescent girls in selected of anaemia
villages of district Gulbarga, Karnataka
Cross-sectional A cross-sectional study on contributing Research variable: contributing
factors of anaemia among adolescent girls factors of anaemia
in selected villages of district Gulbarga,
Karnataka
Cohort A cohort study on smoking and lung cancer Independent variable:
among slum dwellers of city Mumbai smoking
Dependent variable:
lung cancer
Randomized Control A randomized control trial on the efficacy Independent variable:
Trial of oral morphine in the management oral
of chronic cancer pain among advanced morphine
stage cancer patients admitted in a Dependent variable:
hospice at Chandigarh chronic
cancer pain
Quasi-experimental A quasi-experimental study of the effect of Independent variable:
needle gauge on pain perception among needle
patients receiving intramuscular injection gauge
in the outpatient department of Christian Dependent variable:
Medical College and Hospital (CMCH), pain perception during intra-
Vellore, Tamil Nadu muscular (IM) injection

l Bivariate study: Experimental, quasi-experimental,

interventional, case–control and cohort studies OPERATIONAL DEFINITIONS
generally include two variables in research, e.g., ‘A In a research study, each variable or concept
cohort study on smoking and lung cancer among must be operationally defined. An operational
families in urban slums of city Kolkata’. This cor- definition of a variable provides the theoretical
relational study is a bivariate study because it in- or conceptual meaning of the variable under
cludes two variables, one independent (smoking) study. This definition of the variables may be
and the other dependent (lung cancer). derived from the conceptual or theoretical
l Multivariate study: In these studies, more than framework on which the study is based or may
two variables are studied together. Qualitative be stated in a form in which the variables are
studies are usually multivariate studies, which expected to be observed or measured in the
are becoming increasingly common in allied present study, e.g., in the research problem, ‘A
fields of health science research studies, e.g., ‘A quasi-experimental study on the effect of ginger
cross-sectional study on factors contributing to on morning sickness among pregnant women,’
noncompliance of treatment among tuberculo- there are two variables, ginger and morning sick-
sis patients in selected rural communities of ness, which need to be operationally defined.
district Patiala, Punjab’. Here the researcher However, sometimes, researchers also make an
considers more than two variables that are re- attempt to operationally define the population
sponsible for noncompliance among tuberculo- of the study such as pregnant women in this
sis patients. problem statement.
 Research Problem and Hypothesis 31

Table 2.3. Examples of statement of research problems and variables in qualitative research studies
Type of Study Example of the Research Problem statement Variables
Phenomenological A phenomenological study on live experiences Research variable: live experiences
of tsunami victims in selected villages of
state Tamil Nadu
Ethnographical An ethnographic study on the features, critical Research variable: features, critical
attributes, processes and benefits of self- attributes, processes and benefits
help groups of women living with chronic of self-help groups
alcoholic husbands in selected villages of
district Belgaum, Karnataka
Grounded theory A study on the response and adaptation Research variable: response and adap-
process of patients diagnosed with cancer tation process
in Nehru Hospital, PGIMER, Chandigarh
Case study A case study on availability and utilization of Research variable: availability and
emergency services in a public hospital at utilization of emergency services
Ludhiana, Punjab
Historical study A historic study on experiences of ward facili- Research variable: experiences of ward
ties for psychiatric nurses during the 1960s facilities
and 1970s in National Institute of Mental
Health and Neurosciences, Bangalore
Action research A study on practicability of integrating the Research variable: practicability of
nursing service and education in selected integrating the nursing service and
premier nursing institutions of India education

Operational definitions are not the same as The example illustrated in Box 2.1 provides five
conceptual definitions; an operational definition different operational definitions of constipation,
does not only give precise indications as to what one of which may be considered by a particular
the fundamental characteristics of a concept are; it researcher on the basis of the way and means by
also gives precise indications about how to ob- which the concept/variable or phenomenon is be-
serve or even measure the characteristics under ing observed or measured in a particular study. In
study. In other words, an operational definition is this example, the conceptual definition does not
based on the observable characteristics of an ob- indicate how the variable will be measured, but in
ject or phenomenon and indicates what to do or the operational definition, it is very clear how the
what to observe in order to identify these charac- variable will be measured.
teristics. The example given in Box 2.1 provides A unique operational definition is more useful,
clarity about difference in conceptual and opera- since it discriminates accurately between elements
tional definition. that possess the specific characteristics under in-
Thus, it is clear that conceptual definition pre- vestigation and those that do not. In short, the
cisely indicates the fundamental characteristics of most important feature of operational definition
a concept/variable or phenomenon. However, op- is its ability to indicate what to measure and how
erational definition is different from conceptual to measure including the initial ideas about assess-
definition as it precisely indicates the observable ment techniques to determine the level of the
characteristics of a concept/variable or phenome- variable defined.
non. Therefore, different research may have differ- For example, in the research problem, ‘A quasi-
ent operational definition for the same concept. experimental study on the effect of ginger syrup on
32 Research Methodology and Biostatistics

Box 2.1. ​Conceptual and operational definition of constipation

Conceptual definition (precisely indicates the funda- Constipation is defined as passes of hard stools, infre-
mental characteristics of a concept) quent stools, the need for excessive straining, a sense
of incomplete bowel evacuation and excessive time
spent on the toilet or unsuccessful defecation
Operational definition (precisely indicates observ- 1. Bowel movements less than three times a week
able characteristics of a concept) 2. Straining during .25% of bowel movements in
Five different operational definitions of constipation the last 12 weeks
are presented here 3. Lumpy or hard stool for .25% bowel movements
in the last three months
4. Pebble-like, hard stools for a majority of bowel
movements for at least 2 weeks or firm stools
#2 times per week for at least 2 weeks
5. Two or more of these symptoms for at least
12 weeks in the preceding 12 months during
.25% bowel movements: (a) straining while defe-
cation, (b) passing hard stool, (c) sensation of in-
complete evacuation, (d) sensation of anorectal
blockage and (e) manual evacuation of rectum

morning sickness symptoms among pregnant properties possessed by the variable rather
women’, there are two variables, ginger syrup and than by characteristics lacked by the variable.
morning sickness symptoms, which need to be op- For example, a book cannot adequately be de-
erationally defined. However, sometimes, research- scribed as being an object without specific co-
ers also make an attempt to operationally define lour and weight.
the population of the study such as pregnant l Clear and precise: An operational definition
women in this problem statement. should be stated precisely, clearly with unequiv-
ocal terms to avoid different interpretation.
ESSENTIAL CHARACTERISTICS OF l Valid: The operational definition should be valid
OPERATIONAL DEFINITIONS as per the fact. Thus, it is clear that the researcher
l Stated in measurable and observable terms: operationalizes research concepts, but the opera-
An operational definition must denote the tional definitions must have a validity compo-
distinctive characteristics of a variable in mea- nent supported by facts.
surable and observable terms. For example, a l Reliability: The operational definition must
resident doctor is a person who has completed also have a reliability characteristic so that there
a professional medical degree, i.e., MBBS, and will be consistency in interpretation of the op-
has registered with the Medical Council of erational definition (in the same way) by all
India currently practising as a registered prac- those who have to use it.
titioner.
l Nonambiguous language: An operational defi-
nition should not be circular and ambiguous. In WRITING RESEARCH OBJECTIVES
other words, it should not describe something Research is an organized investigation of a prob-
by the same concept. For example, an impatient lem in which there is an attempt to gain solutions
person should not be defined as a person who to a problem. To get the right solution to a right
lacks patience. problem, clearly defined objectives are very im-
l Stated positively: An operational definition portant. Clearly defined objectives enlighten the
should be stated positively, expressing the way in which the researcher has to proceed.
 Research Problem and Hypothesis 33

Therefore, clearly defined objectives are an im-

portant feature of a good research study without
which the researcher is aimless and directionless Specific
in conducting the study, resulting in wastage of Unequivocal Measurable
resources. Without focused objectives, no replica-
ble scientific findings can be expected from any
type of research. Once the need for research infor-
mation has been clearly defined, the researcher Observable Attainable

must specify the objectives of the proposed re- Characteristics

search and develop a specific list of information of research
objectives
needed. Objectives formulated in quantitative and
qualitative studies are quite similar; however, ob- Logical Realistic
jectives formulated in qualitative studies are broad
and include variables that are more complex and
abstract in nature. Feasible Time-bound
Relevant
MEANING OF RESEARCH OBJECTIVES
‘A research objective is a clear, concise, declarative
statement, which provides direction to investigate
Fig. 2.4. Specific characteristics of research objectives.
the variables.’ Generally, research objectives focus
on the ways to measure the variables, such as ways
to identify or describe them. Sometimes, objec-
tives are directed towards identifying the relation- l The research objectives are the specific accom-
ship or difference between two variables. plishments the researcher hopes to achieve by
‘Research objectives are the results sought by the study.
the researcher at the end of the research process, l The objectives include obtaining answers to
i.e., what the researcher will be able to achieve at research questions or testing the research hy-
the end of the research study.’ potheses.
The objectives of a research project summarize
what is to be achieved by the study. Objectives NEED OF RESEARCH OBJECTIVES
should be closely related to the statement of the The formulation of research objectives will help
problem. the researcher to do the following:
l Focus: With clearly defined objectives, the re-
CHARACTERISTICS OF RESEARCH OBJECTIVES searchers can focus on the study. If they know
l Research objective is a concrete statement de- what they have to achieve ultimately, they will
scribing what the research is trying to achieve. concentrate on the ways to achieve it. The
A well-worded objective will be SMART, i.e., formulation of research objectives helps in nar-
specific, measurable, attainable, realistic and rowing down the study to its essentials. It will
time-bound (Fig. 2.4). avoid unnecessary findings, which otherwise
l Research objective should be relevant, feasible, lead to wastage of resources, people, money and
logical, observable, unequivocal and measur- material.
able. l Avoid: The formulation of research objectives

l Objective is a purpose that can be reasonably helps the researcher avoid the collection of data
achieved within the expected timeframe and which are not strictly necessary for understand-
with the available resources. ing and solving the problem that he or she has
l The objective of a research project summarizes defined. For example, a researcher is only inter-
what is to be achieved by the study. ested in identifying the presence or absence of
34 Research Methodology and Biostatistics

variables but not in the detailed description of during the same study. Examples of secondary
related factors, where the research objective will objectives are given as follows:
help the researcher avoid collation of unwanted  To determine the prevalence of morning sick-

information. ness symptoms among pregnant women

l Organize: The formulation of objectives orga-  To compare the severity of morning sickness

nizes the study in clearly defined parts or phases. symptoms among primigravida and multi-
For example, if a researcher is conducting a re- gravida
search study on the effects of ginger on morning  To compare the effects of ginger on morning

sickness among pregnant women, the objectives sickness in multigravida and primigravida
are formulated as (a) to assess the prevalence of  To determine the fetotoxic effects of ginger

morning sickness among pregnant women and

(b) to identify the effect of ginger on morning METHOD OF STATING OBJECTIVES
sickness symptoms among pregnant women. While stating the objectives of the study, the fol-
These objectives help organize the study results lowing guidelines must be taken care of:
in two main parts: the first one focuses on l The objectives should be presented briefly and

the prevalence of morning sickness; the second concisely.

part will focus on the effect of ginger on morn- l They cover the different aspects of the problem

ing sickness. and their contributing factors in a coherent way

l Directions: Properly formulated, specific ob- and in a logical sequence.
jectives will facilitate the development of re- l They are clearly phrased in operational terms,

search methodology and help to orient the specifying exactly what the researcher is going
collection, analysis, interpretation and utiliza- to do, where and for what purpose.
tion of data. l They are realistic considering local conditions.

l They use action verbs that are specific enough

TYPES OF RESEARCH OBJECTIVES to be evaluated.
Research objectives are of two types: Examples of action verbs are as follows: to as-
l Primary objectives: Primary objectives are the sess, to identify, to find out, to determine, to com-
broad and main goals to be achieved during pare, to find, to verify, to calculate, to describe, to
the research study. analyse and to establish. (A detailed list of action
l Secondary objectives: Secondary objectives are verbs is given in Table 2.4.) Avoid the use of vague
for short term and are narrow in focus; they are nonaction verbs such as to appreciate, to under-
not the actual intent of the study but planned to stand and to study. Table 2.5 also illustrates the
be achieved during the study to save time and commonly used action verbs in different types of
resources. research methods.
A researcher should always keep in mind that
Example of Primary and Secondary Objectives when the project is evaluated, the results will be
Research statement: A quasi-experimental study compared to the objectives. If the objectives have
on the effects of ginger on morning sickness not been spelled out clearly, the project cannot be
among pregnant women in selected cities of considered complete.
Kerala.
Primary objective: The general broad and main
objective of this study is HYPOTHESIS
 To assess the effects of ginger on morning A hypothesis is a formal tentative statement of the
sickness among pregnant women. expected relationship between two or more vari-
Secondary objectives: The researcher may also plan ables under study. A hypothesis helps to translate
the following objectives along with the main pri- the research problem and objectives into a clear
mary objective to obtain additional information explanation or prediction of the expected results
 Research Problem and Hypothesis 35

Table 2.4. A list of action verbs used to formulate research objectives

Knowledge Comprehension Application Analysis Synthesis Evaluation
Count Associate Add Analyse Categorize Appraise
Define Compute Apply Arrange Combine Assess
Describe Convert Calculate Breakdown Compile Compare
Draw Defend Change Combine Compose Conclude
Identify Discuss Classify Design Create Contrast
Label Distinguish Complete Detect Drive Criticize
List Estimate Compute Develop Design Critique
Match Explain Demonstrate Diagram Devise Determine
Name Extend Discover Differentiate Explain Grade
Outline Extrapolate Divide Discriminate Generate Interpret
Point out Generalize Examine Illustrate Integrate Judge
Quote Give examples Graph Infer Modify Justify
Read Infer Interpolate Outline Order Measure
Recite Paraphrase Manipulate Relate Organize Rank
Recognize Predict Modify Select Plan Rate
Record Rewrite Operate Separate Prescribe Support
Repeat Summarize Prepare Subdivide Propose Test
Reproduce Produce Utilize Rearrange
Select Show Solve Reconstruct
State Solve Reorganize
Write Subtract Revise
Translate Summarize
Use Specify

or outcomes of the research study. A clearly stated Hypotheses are used in quantitative research
hypothesis includes the variables to be manipu- studies to predict the relationship of two or more
lated or measured, identifies the population to be variables but are not appropriate for qualitative
examined and indicates the proposed outcome of research studies and some quantitative studies
the study. Hypothesis also influences study design, such as descriptive and explorative research stud-
sampling methods, data collection process and ies because these studies do not involve interplay
interpretation of the research findings. of dependent and independent variables. In quan-
The formulation of a hypothesis plays an im- titative research studies, hypothesis serves as a
portant part in the growth of knowledge in every guiding source of the research process including
science. A hypothesis converts the question posed analysis, interpretation and drawing a conclusion.
by the research problem into a declarative statement However, qualitative research studies are carried
that predicts an expected outcome. For example, out to explore new concepts and ideas about little
smoking causes lung cancer. known concepts or to discuss new meaning of
36 Research Methodology and Biostatistics

Table 2.5. Commonly used active verbs Polite and Beck define hypothesis as a statement
in different research methods of researcher’s expectations about the relationship
Quantitative Methods Qualitative Methods between the variables under investigation.
Descriptive Ethnography
A hypothesis is an assumption statement about
the relationship between two or more variables
Identify Assess
that suggest an answer to the research question.
Assess Describe
Find out Examine
IMPORTANCE OF HYPOTHESIS IN RESEARCH
Describe Explain
Compare Discover The hypothesis is the basis of a scientific investiga-
Contrast Drive tion, and it plays a pivotal role in the research
Correlational Grounded theory process. Hypothesis and theories are reciprocal to
Determine Develop each other. The verified hypothesis becomes the
Examine Extend basis of theory; theory also serves as a source
Identify Identify of hypothesis. Some important characteristic fea-
Discriminate Validate tures of hypothesis are given here:
l Hypothesis enables the researcher to objectively
Experimental Phenomenological
Determine Describe investigate new areas of discovery. Thus, it provides
Examine Develop a powerful tool for the advancement of knowledge.
Investigate Generate l It provides objectivity to the research activity. The

Measure Record formulation of a hypothesis provides a study with

Detect Express focus and guides the researcher regarding the spe-
cific aspects of research problems to investigate.
l It provides insight to the researcher about what

data to collect and what not to collect, thereby

concepts. Thus, use of hypothesis in qualitative providing the focus to the study.
research studies is not appropriate and should be l It also provides directions to conduct research

avoided to prevent researchers’ bias. such as defining the sources and relevance of data.
l It provides clear and specific goals to research-
MEANING OF HYPOTHESIS ers. These clear and specific goals provide the
‘Hypothesis is a tentative prediction or explana- investigator with a basis for selecting sample
tion of the relationship between two or more and research procedures to meet these goals.
variables.’ It implies that there is a systematic rela- l It provides a link between theory and actual

tionship between an independent and a depen- practical research.

dent variable. For example, dietary compliance l It provides a bridge between theory and reality.

will be greater in diabetic patients receiving diet l It suggests which type of research is likely to be

instruction in small groups than in diabetic pa- most appropriate.

tients receiving individualized diet instructions. l As it is a tentative statement of anticipated re-

Good and Hatt define hypothesis as a shrewd sults, it guides the researcher towards the direc-
guess or inference that is formulated and provi- tion in which the research should proceed.
sionally adopted to explain observed facts or con- l It stimulates the thinking process of the re-

ditions and to guide in further investigation. searcher as the researcher forms the hypothesis
Walter R. Brog states that hypothesis reflects by anticipating the outcome.
the research workers’ guess as to the probable out- l It also determines the most appropriate research

comes of their experiment. Hypotheses place clear designs and techniques of data analysis.
and specific goals before the researchers and pro- l It provides understanding to the researchers

vide them with a basis for selecting samples and about what to expect from the results of the
research procedures to meet these goals. research study.
 Research Problem and Hypothesis 37

l It serves as a framework for drawing conclu- study, operational definitions, existing evidence
sions of a research study. and related theories.
l Without hypothesis, research would be like l Testability: Hypothesis should be testable and
aimless wandering. should not be a moral judgement. It must be
directly/indirectly observable and measurable.
CHARACTERISTICS OF A GOOD HYPOTHESIS The researcher can set up a situation that per-
A good hypothesis must be written in declarative mits one to assess if it is true or false. It must
statements, using present tense. It must contain be verifiable, e.g., a statement such as ‘bad
the variables, population under study and should partners produce bad children,’ cannot be
be relevant to the research problem and objec- tested.
tives, which can be easily tested. The main charac- A testable hypothesis clearly states the manipu-
teristics of a good hypothesis are as follows: latable independent variables and measur-
l Conceptual clarity: The hypothesis should con- able dependent variables in a specific popula-
sist of clearly defined and understandable con- tion; it provides a clear idea about an
cepts. It should be stated in very clear terms, interventional protocol and whether it will be
the meaning and implication of which cannot implemented precisely and consistently as a
be doubted. To facilitate the conceptual clarity, treatment in the study; it also provides an
the hypothesis can be stated in declarative state- understanding about how the dependent
ment, in the present tense. variable will be accurately measured. This
l Empirical referents: Research must have an ul- hypothesis states the causal link between in-
timate empirical referent. No usable hypothesis dependent and dependent variables, which is
can embody moral judgements. A good hypoth- later evaluated by using inferential statistical
esis must have empirical basis from the area of tests. A clearly stated hypothesis is usually
enquiry. Therefore, hypothesis generally must tested through inferential statistical tests to
be formulated on the basis of the concepts that determine the effect of independent variables
are empirically tested. on dependent variables.
l Objectivity: The hypothesis must be objective, l Consistency: A hypothesis should be consistent
facilitate objectivity in data collection and keep with an existing body of theories, research find-
the research activity free from researcher value- ings and other hypotheses. It should correspond
judgement or bias. The objectivity of hypoth- with existing knowledge.
eses provides an effective guidance to the l Simplicity: A hypothesis should be formulated
researcher about what to measure and how to in simple and understandable terms. It should
measure. require few conditions and assumptions.
l Specificity: A good hypothesis should be specific, l Availability of techniques: Researchers must
not general, and should explain the expected re- make sure that methods are available for testing
lations between variables. For example, regular their proposed hypotheses.
yoga reduces stress. Specificity of hypothesis l Purposiveness: The researcher must formulate
helps to avoid ambiguity and clearly specifies the only purposeful hypotheses, which has rele-
intended influence of an independent variable on vance with research problem and objectives.
the dependent variable. l Verifiability: A good hypothesis can be actually
l Relevant: The hypothesis should be relevant to verified in practical terms.
the problem being studied as well as the objec- l Profundity of effect: A good hypothesis should
tives of the study. In addition, the hypothesis have profound effect upon a variety of research
must have relevance with the theory under test variables.
in a research process. Furthermore, hypotheses l Economical: The expenditure of money and
of the study must be relevant to the statement of time can be controlled if the hypothesis under-
research problem, objectives of the research lying the research undertaken is good.
38 Research Methodology and Biostatistics

finding to formulate a hypothesis as ‘Obese pa-

SOURCES OF HYPOTHESES tients have increased risk for development of
Hypotheses in research studies cannot be devel- coronary artery disease’.
oped merely with wild guesses or assumptions; l Real-life experiences: Real-life experiences also
they are generated from a variety of sources, such contribute in the formulation of hypotheses for
as theoretical or conceptual frameworks, previous research studies. For example, Newton had a
research findings, real-life experiences and aca- life-changing experience of the falling of an ap-
demic literature, as also depicted in Fig. 2.5. ple and formulated a hypothesis that earth at-
l Theoretical or conceptual frameworks: The tracts all mass towards its centre, though several
most important sources of hypotheses are the- researches were conducted before generating a
oretical or conceptual frameworks developed law of central gravity.
for the study. Through a deductive approach, l Academic literature: Academic literature is
these hypotheses are drawn from theoretical or based on formal theories, empirical evidence,
conceptual frameworks for testing them. For experiences, observations and conceptualiza-
example, Roy’s adaptation model is used in a tions of academicians. These literatures may
research study where a hypothesis can be drawn serve as a good source for formulating hypoth-
from a concept of the theoretical mode that eses for research studies.
‘Patient’s adaptation to a chronic illness de-
pends on availability of social support for him TYPES OF HYPOTHESIS
or her’. l Simple and complex hypotheses:
l Previous research findings: Findings of previ-  Simple hypothesis: It is a statement that re-

ous studies may be used for framing the hypoth- flects the relationship between two variables.
eses for another study. For example, in a small For example, the lower the level of haemoglo-
sample descriptive study, a researcher finds that bin, the higher the risk of infection among
a number of patients admitted with coronary postpartum women.
artery disease had increased body mass index. In  Complex hypothesis: It is a statement that

another research study, a researcher may use this reflects the relationship between more than
two variables.
For example, satisfaction is higher among
patients who are older and dwelling in rural
Theoretical
or
areas than those who are younger and
conceptual dwelling in urban areas.
frameworks l Associative and causal hypotheses:
 Associative hypothesis: It reflects a relation-

ship between variables that occurs or exists in

natural settings without manipulation. This
Academic Sources of Previous hypothesis is used in prospective or retro-
literature hypotheses research spective research studies. Associative hypoth-
esis may be stated as follows:

Examples of Associative Prediction of Hypothesis

Hypothesis
Real-life
Communication skills of Predicts relationship
experiences
health care providers among variables
and cost of care are but not the type of
related to the satisfac- relationship
Fig. 2.5. Sources of hypotheses. tion of patients
 Research Problem and Hypothesis 39

The more the communi- Predicts positive rela- it does not specify the anticipated direction
cation from health tionship and nature of relationship such as positive
care providers, the or negative. It only indicates the existence of a
higher is the satisfac- relationship between the variables.
tion among patients For example, there is a relationship between
The lower the blood Predicts positive rela- years of nursing experience and job satis-
sugar level, the lesser tionship faction among nurses.
the risk of infection l Null and research hypotheses:
among diabetic pa-  Null hypothesis (H ): It is also known as the
0
tients
statistical hypothesis and is used for statistical
The lower the cost of Predicts negative rela-
testing and interpretation of statistical out-
care, the higher the tionship
satisfaction among
comes. It states the existence of no relation-
patients ship between the independent and dependent
variables.
For example, there is no relationship between
 Causal hypothesis: It predicts the cause-and- smoking and the incidence of coronary ar-
effect relationship between two or more de- tery disease.
pendent and independent variables in experi-  Research hypothesis (H ): It is also known as
1
mental or interventional setting, where the the alternative hypothesis. It states the exis-
independent variable is manipulated by the tence of a relationship between two or more
researcher to examine the effect on the depen- variables.
dent variable. The causal hypothesis reflects For example, there is a relationship between
the measurement of the dependent variable to smoking and incidence of lung cancer.
examine the effect of the independent vari- A research hypothesis could be simple, com-
able, which is manipulated by the researcher(s). plex, directional, nondirectional, associative or
For example, prevalence of pin site infection causal. However, it is commonly recommended
is lower in patients who receive pin site care to use directional research hypothesis in re-
with hydrogen peroxide as compared to search studies.
patients who receive pin site care with povi-
done–iodine solution. FORMULATING HYPOTHESES
l Directional and nondirectional hypotheses: Hypotheses are usually formulated in quantitative
 Directional hypothesis: It specifies not only research studies which investigate the interplay of
the existence, but also the expected direction dependent and independent variables such as ex-
of the relationship between variables. Direc- perimental and correlation research studies; in
tional hypothesis states the nature of the rela- other studies, use of hypothesis is inappropriate.
tionship between two or more variables such There is no single way to develop an excellent hy-
as positive, negative or no relationship. To pothesis. However, researchers employ inductive
express the direction of relationship between and deductive reasoning on available theory and
variables, directional terms are used to state knowledge to formulate a hypothesis. In addition,
the hypothesis such as positive, negative, less, there are several factors that must be considered
more, increased, decreased, greater, higher while formulating an excellent research hypothe-
and lower. sis. Some essential points to be considered are
For example, there is a positive relationship discussed here:
between years of nursing experience and l Hypotheses are always written before starting the

job satisfaction among nurses. study and should not be changed after the study
 Nondirectional hypothesis: It reflects the re- results are examined. They are primarily derived
lationship between two or more variables, but from the statement of the research problem,
40 Research Methodology and Biostatistics

study variables, operational definitions and rele- ill patients who receive continuous nasogastric
vant theories. feeding experience fewer gastrointestinal prob-
l The researcher uses the null form as well as the lems as compared to patients who receive bolus
alternative form for formulating research hy- nasogastric feeding’.
pothesis. Statistical logic requires stating the l Hypothesis must be stated in an empirically test-

hypothesis in null form, but several research able form. For example, consider the hypothesis
thinkers argue that it is generally more desir- ‘Nurse practitioners are better health care pro-
able to state the hypothesis in directional and viders than physicians’. In this example, the hy-
alternative forms, when there is reasonable pothesis is not empirically testable because better
basis for them because these hypotheses clarify is a value word that cannot be objectively mea-
the study framework and demonstrate that re- sured. Therefore, to make this hypothesis em-
searchers have critically thought about the phe- pirically testable, it may be stated as, ‘Nurse
nomenon under study. Furthermore, a review practitioners spent more time in teaching their
of current nursing research journals shows that clients about preventive health care practices
the research hypothesis has replaced the statis- than do physicians’. In this, the term time can be
tical null hypothesis as the preferred way of measured empirically.
expressing predictions for study. Thus, this l It is better that one hypothesis contains only a

argument supports that it is better to use direc- single predication about relationships of study
tional alternative forms of research hypotheses variables.
for research studies. l Ideally, the number of hypothesis should be re-

l The following four main components of a hy- stricted to 6 or less because more number of
pothesis must be considered while formulating hypotheses leads to confusion and increased
research hypothesis (Box 2.2): work in data collection, analysis and interpreta-
 Dependent and independent variables tions of data.
 Nature of relationship between dependent A hypothesis can be evaluated using the criteria
and independent variables given in Box 2.3.
 Subjects in population being studied

 The level of significance for accepting/reject-

ing the hypothesis Box 2.3. Criteria for evaluation of research

l Hypothesis should be stated using declarative sen- hypotheses
tences and present tense. For example,’ Critically
1. Is each hypothesis clearly and precisely worded?
2. Is the hypothesis written as a declarative
sentence?
Box 2.2. ​Example of components of a hypothesis 3. Is the hypothesis written in present tense?
4. Is the hypothesis derived from and linked to the
For example, patients who underwent angiography research problem, study variables, operational
receiving transparent film dressing have fewer bleed- definitions and relevant theories?
ing complications compared to patients who receive 5. Does the hypothesis present dependent and
pressure dressing. independent variables?
Dependent variable: Bleeding complication 6. Does the hypothesis present the population
Independent variable: Type of dressing at angiogra- under study?
phy site (transparent film vs. pressure dressing) 7. Is each hypothesis stated as a directional research
Relationship between dependent and independent hypothesis?
variables: Predicting fewer bleeding complications 8. Is each hypothesis empirically testable?
among patients receiving transparent film dressing 9. Does each hypothesis contain only single predi-
as compared to patients with pressure dressing cation about relationships of study variables?
Subjects in population being studied: Patients who 10. Is each hypothesis stated with the level of
underwent angiography significance for testing the hypothesis?
 Research Problem and Hypothesis 41

Example of a well-stated hypothesis Box 2.4. Sample phrases that express

the delimitations of a study
A well-stated research hypothesis is written in de-
clarative directional–alternative form, which predicts The study does not cover the . . .
the nature of the relationship between dependent The researcher limited this research to . . .
and independent variables and also includes the This study is limited to . . .
population under study.
There is a significant positive correlation be-
tween years of experience and level of nurses’ job setting, sample size, the variables studied, the
satisfaction. theoretical perspectives, the instruments, the
Male patients have higher satisfaction with nurs- generalizability, etc.
ing care as compared to their female counterparts. l Delimitations are the choices made by the re-
searcher that should be mentioned in the research
study. They describe the boundaries that the re-
DELIMITATIONS searcher has set for the study. In delimitations, the
The terms delimitations and limitations are inter- following parameters are identified and addressed
changeably used by researchers. However, delimi- to clearly define the scope of the study and to
tations are boundaries set by the researcher make the study more feasible. Box 2.4 presents
before starting the study to clearly define the sample phrases that are commonly used to
scope of the particular study. On the other hand, express delimitations of a study.
limitation refers to challenges faced by the re-  Confinement to a particular geographical

searcher beyond his or her control. Delimitations area or study setting/study centre (and why
are mentioned in the initial chapter of the dis- this restriction was considered): For example,
sertation and limitations are mentioned in the this study was limited to the single health care
last chapter of the dissertation/thesis, which are centre of the city because of limited time and
essential to inform the readers about drawbacks funds for study.
on the credibility of the evidence generated  The population, which the researcher is not

through the particular research. The concept of studying (and why not): For example, this
delimitation may be clearly understood from the study will be restricted to doctors working in
following discussion: tertiary care centres. Doctors who are work-
l Delimitations are boundaries that are set by the ing in other than tertiary care hospitals will
researcher in order to control the range of a not be included in the present study because
study. They are created before any investigations the researcher only intended to include doc-
are carried out, in order to reduce the amount tors who have exposure of working in tertiary
of time spent in certain areas that may be seen care hospitals to maintain the homogeneity of
to be unnecessary, and perhaps even unrelated the sample.
to the overall study.  Defining the restriction about the sampling

l In other words, delimitations are restrictions technique/sample size (and specifying the
that researchers impose prior to the inception reason): For example, in the present study, the
of the study to narrow the scope of the study. sample was not drawn using the random
For example, the study might be delimited to sampling technique because of nonavailabil-
secondary schools of a city or to a survey of ity of the sampling frame. The present study
rural female students only. was carried out with a small sample because
l Delimitations are those characteristics identi- of time constraints.
fied by the researcher to define the boundaries  Particular literature will not be reviewed

of the study. The researcher makes conscious (and why not): For example, in the present
exclusionary and inclusionary decisions regard- study, only locally published studies were
ing the geographical location of the study, study considered for literature review and for the
42 Research Methodology and Biostatistics

discussion of the study findings, because of about the credibility and generalizability of the
the inability to access international literature. research findings
 Methodological procedures will not be used
(and why will they not be used): For example, Types of Delimitations
in the present study, randomized control trial Delimitations are restrictions of the study because
design was not used because of the quasi of theoretical or methodological reasons; they may
nature of the independent variable. decrease the credibility and generalizability of the
 Particular tool(s) was not used for the study research findings. Usually, there are two types of
(why it was not used): For example, the re- delimitations in research studies which may reduce
searcher was limited to data collected through the credibility and generalizability of the research
telephonic method because of limited time findings; these are described as follows:
for the present study. l Theoretical delimitations: They restrict the

 Things that researchers are not doing (and ability of research findings to generalize because
specify why they are not doing them): For of the use of specific theoretical concepts in the
example, the researcher limited the follow-up study, or the limitation of the study of variables
of patients for only five days because as per through operational definitions.
hospital protocol, patients who underwent l Methodological delimitations: They usually re-

laparotomy are discharged on the fifth post- sult from some methodological factors such as
operative day. unrepresentative sample, weak design, single
setting, limited control over extraneous vari-
USES OF DELIMITATIONS ables, poor implementations of treatment pro-
The delimitation of a study is restricting the study tocol, research tools with limited reliability and
by geographic location, age, sex, population traits, validity, poor data collection procedure, ineffec-
population size or other similar considerations. tive use of statistical analysis, etc.
Delimitation is used to make a study better and
more feasible and not just for the interest of the POINTS TO REMEMBER WHILE WRITING
researcher. It also identifies the constraints or DELIMITATIONS
weaknesses of the study which are not within the Delimitations of a study are generally mentioned
control of the researcher. Delimitation is used in the initial chapter of the thesis/dissertation and
in research studies because it helps to do the in the beginning of the discussion of the results in
following: a research paper. When discussing the delimita-
l Clearly define the scope of the study to avoid tions of a research study, the researcher must
parametric ambiguities in the research study ensure to do the following:
l Make the research study more feasible and prac- l Describe each delimitation in detailed but con-

ticable cise terms

l Define the boundaries of the selected study l Explain the reasons for not considering certain

parameters such as geographic location, age, parameters in study

gender, population traits, population size, sam- l Mention why a particular delimitation is con-

ple size, study instruments or other similar sidered for a research study
considerations l While stating delimitations, reasons for not

l Identify the constraints or weaknesses of the considering certain parameters in the study
study so that the strengthening of evidence gen- must be clearly explained
erated through study may be determined l Identify and mention the intended impact of

l Researchers usually mention delimitations of each delimitation in relation to the overall find-
their study so that the readers can have an idea ings and conclusions of a study
 Research Problem and Hypothesis 43

MULTIPLE CHOICE QUESTIONS

1. Sources of researchable problems can include: 7. An operational definition specifies:
(a) Researchers’ own experiences as educators (a) The data analysis techniques to be used in the
(b) Practical issues that require solutions study
(c) Theory and past research (b) The levels of measurement to be used in the
(d) All of the above study
(c) How a variable or concept will be defined and
2. A condition or characteristic that can take on differ- measured in the study
ent values or categories is called a: (d) How the outcome of the research objectives for
(a) Constant the study will be measured
(b) Variable
(c) Cause-and-effect relationship 8. Which of the following statements meet the criteria
(d) Descriptive relationship for a researchable question?
(a) Is the use of normal saline to cleanse wounds
3. A variable that is presumed to cause a change in harmful to patients?
another variable is called a(n): (b) Do generalist registered nurses meet the men-
(a) Categorical variable tal health needs of general patients?
(b) Dependent variable (c) What are the patients’ perceptions of the
(c) Independent variable effectiveness of preoperative education for
(d) Intervening variable total hip replacement?
(d) Do palliative care patients have spiritual needs?
4. Which of the following cannot be considered as
a component of the research problem? 9. A variable that changes due to the action of another
(a) Research study tool(s) variable is known as the:
(b) Variable(s) (a) Independent variable
(c) Study population (b) Extraneous variable
(d) Research setting (c) Dependent variable
(d) Complex variable
5. How can you tell if your research questions are really
good? 10. Variables that are not part of the study but may have
(a) If they guide your literature search an effect on the measurement of the study variables
(b) If they are linked together to help you construct are known as:
a coherent argument (a) Independent variables
(c) If they force you to narrow the scope of your (b) Extraneous variables
research (c) Dependent variables
(d) All of the above (d) Complex variables

6. The statement to compare the severity of morning 11. A statement of the expected relationship between
sickness symptoms among primigravida and multi- two or more variables is known as the:
gravida is best described as a research: (a) Concept definition
(a) Aim (b) Hypothesis
(b) Objective (c) Problem statement
(c) Question (d) Research question
(d) Hypothesis
44 Research Methodology and Biostatistics

12. Which of the following statements represents 16. Research hypotheses are
a complex hypothesis? (a) Formulated prior to a review of literature
(a) There is a significant positive correlation be- (b) Statements of predicted relationships between
tween years of experience and level of nurses’ variables
job satisfaction (c) Stated such that they can be confirmed or
(b) Male patients have higher satisfaction with refuted
nursing care as compared to their female coun- (d) Belief about any natural phenomenon
terparts
(c) Social support, balanced diet and regular exer- 17. Hypotheses in qualitative research studies usually:
cise decrease the incidence of postpartum de- (a) Are very specific and stated prior to beginning
pression the study
(d) Music therapy reduces pain perception during (b) Are often generated as the data are collected,
surgery interpreted and analysed
(c) Are never used
13. Which of the following pairs contain hypotheses (d) Are always stated after the research study has
that are similar to each other? been completed
(a) Simple and complex
(b) Associative and causal 18. Which of the following is not a component of
(c) Directional and causal hypothesis?
(d) Null and complex (a) Dependent variables
(b) Independent variables
14. ‘Relaxation therapy is more effective than standard (c) Extraneous variables
patient education alone in decreasing preoperative (d) Study population
anxiety’ is an example of a:
(a) Null hypothesis 19. The main purpose of considering delimitation in
(b) Nondirectional hypothesis research study is
(c) Complex hypothesis (a) To predict the relationship between variables
(d) Directional hypothesis (b) To define the scope of research study
(c) To express the beliefs and ideas about the
15. ‘There is no difference in the incidence of phlebitis phenomenon under study
around intravenous cannulae changed every (d) To define the variable in measurable terms
72 hours and those changed at 96 hours’ is an ex-
ample of a:
(a) Null hypothesis
(b) Directional hypothesis
(c) Nondirectional hypothesis
(d) Simple hypothesis

Answers of Multiple Choice Questions

1. d; 2. b; 3. c; 4. a; 5. d; 6. d; 7. c; 8. c; 9. c; 10. b; 11. b; 12. c; 13. b; 14. d; 15. a; 16. b; 17. b; 18. c; 19. b.
 Research Problem and Hypothesis 45

REFERENCES
Anonymous. (n.d.). Research ideas, critiquing research and Moore, S. M., & Dolansky, M. A. (2001). Randomized trial of a
hypothesis. ,http://www.carlrenold.com/files/Chapter_2.pptx. home recovery intervention following coronary artery bypass
Accessed 21.02.10. surgery. Research in Nursing & Health, 24(2), 93–104.
Anonymous. (n.d.). Student learning objectives (SLO) action verb Munhall, P. L. (2001). Nursing research: A qualitative perspective
list, suggested verbs to use in each level of thinking skills. (3rd ed.). Boston, MA: Johen and Bartlett.
,http://www.emnu.edu/academics/assessment/faculty/man- Needham, I., Abderhalden, C., Halfens, R. J., Dassen, T., Haug, H.
nual/verb_list.html. Accessed 21.02.10. J., & Fischer, J. E. (2005). The effects of a training course in
Burns, N., & Grove, S. K. (2005). The practice of nursing research: aggression: a cluster randomized trial. International Journal of
Conduct, critique and utilization (5th ed.). St. Louis, MO: Nursing Studies, 42(6), 649–655.
Elsevier. Neeraja, K. P. (2007). Textbook of nursing education (pp. 158–160).
Chinn, P. L., & Kramer, M. K. (2011). Integrated theory and knowl- New Delhi: Jaypee Brothers Medical Publishers.
edge development in Nursing. (8th ed.). St Louis: Elsevier Parse, R. R. (2006). Feeling respected: a parse method study. Nursing
Mosby. Science Quarterly, 19(1), 51–56.
Daly, P. (2005). Mothers living with suicidal adolescents: a phe- Polit, D. F. & Beck, C. T. (2011). Nursing Research: Generating &
nomenological study of their experience. Journal of Psychoso- assessing evidence for nursing practice. (9th ed.). New York:
cial and Nursing Mental Health Services, 43(3), 22–28. Wolter Kluwer, Lippincott.
Farrugia, P., Petrisor, B. A., Farrokyar, F., & Bhandari, M. (2010). Polit, D. F., & Hungler, B. P. (1989). Essentials of nursing research:
Research questions, hypothesis and objectives. Canadian Jour- Methods, appraisal and utilization (p. 102). Philadelphia, PA: J. B.
nal of Surgery, 54(4), 278–281. Lippincott and Company.
Fitzpatrick, M. L. (2001). Historical research: the method. In P. L. Reishtein, J. L. (2005). Relationship between symptoms and func-
Munhall (Ed.), Nursing research: A qualitative perspective (3rd tional performance in COPD. Research in Nursing & Health,
ed., pp. 403–415). Boston, MA: Johen and Bartlett. 28(1), 39–47.
Germain, C. P. (2001). Ethnography: the method. In P. L. Munhall Sharma, B. S. (1983). Research methods in social sciences. New Delhi:
(Ed.), Nursing research: A qualitative perspective (3rd ed., Sterling Publishers Private Ltd.
pp. 277–306). Boston, MA: Johen and Bartlett. Sidhu, K. S. (1985). Methodology of research in education. New Delhi:
Ghosh, B. N. (1984). Scientific methods and social research. Sterling Publishers Private Ltd.
New Delhi: Sterling Publishers Private Ltd. Steele, S. K., & Porche, D. J. (2005). Testing the theory of planned
Goode, W. J., & Hatt, P. K. (1952). Methods in Social Research. behaviour to predict mammography intentions. Nursing
New York: Mc Graw Hill. Research, 54(5), 332–338.
Jacelon, C. S., & O’Dell, K. K. (2005). Case and grounded theory Stommel, M., & Wills, C. E. (2004). Clinical research: Concepts and
as qualitative research methods. Urologic Nursing, 25, 49–52. principle for advanced practice nurses. Philadelphia, PA: Lippincott
Jocobson, S. F., Chu, N. L., Pascucci, M. A., & Gaskins, S. W. (2004). Williams & Wilkins.
Characteristics of nursing research race, ethnicity and culture. Trivedi, R. N. (1991). Research methodology. Jaipur: College Book
Journal of Multicultural Nursing and Health, 10(3), 6–12. Depot.
Kahn, C. R. (1994). Picking a research problem: the critical Vrrkevisser, C. M., Pathmanathan, I., & Ann, B. (2003). Designing
decision. The New England Journal of Medicine, 330(21), and conducting health systems research projects. Proposal de-
1530–1533. velopment and fieldwork. ,http://www.idrc.ca/en/ev-56599-
Kerlinger, F. N., & Lee, H. B. (2000). Foundation of behavioral re- 201-1-DO_TOPIC.html. Accessed 21.02.10.
search (4th ed.). Atworth, TX: Harcourt College. Walter, R. B., & Meredith, D. G. (1971). Educational Research:
Melney, B. M., & Fineout-Overholt, E. (2005). Evidence-based An Introduction. New York: Devid McKay.
practices in nursing and healthcare: A guide to best practice. Williams, M. A. (1980). Editorial: assumptions in research.
Philadelphia, PA: Lippincott Williams & Wilkins. Research in Nursing & Health, 3(2), 47–48.
 CHAPTER OUTLINE
Literature Review 3
INTRODUCTION SOURCES OF A LITERATURE

MEANING OF LITERATURE REVIEW STEPS OF A LITERATURE REVIEW

IMPORTANCE OF A LITERATURE REVIEW POINTS TO BE CONSIDERED FOR A LITERATURE REVIEW

PURPOSES OF A LITERATURE REVIEW MULTIPLE CHOICE QUESTIONS

TYPES OF LITERATURE REVIEWS

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Identify the steps for conducting literature review.
• Define the concept of literature review. • Differentiate between good and poor written literature
• Recognize the importance of literature review. review.
• Discuss the purposes of literature review. • Explain the essential points to be considered while
• Mention and explain the types of literature review. writing a literature review.
• Enumerate the sources of literature review.

KEY TERMS

• Boolean operators 55 • Literature review 47 • Printed sources 53

• Chronological organization 57 • Meta-analysis 49 • Search engine 51
• Database 51 • Metasynthesis 49 • Secondary sources 50
• Empirical 47 • Methodological 47 • Systemic literature review 49
• Electronic sources 50 • Preview 56 • Traditional literature review 49
• Key search terms 54 • Primary sources 50 • Thematic organization 57

done and the knowledge and ideas that have been

INTRODUCTION already established on a particular topic of re-
Review of literature is one of the most important search. A literature review is an account of the
steps in the research process. It is an account of previous efforts and achievements of scholars and
what is already known about a particular phenom- researchers on a phenomenon. Actually, it is a piece
enon. The main purpose of literature review is to of discursive prose, and not a list describing or
convey to the readers the amount of work already summarizing one piece of literature after another.
 Literature Review 47

Literature review is a laborious task, but it is Second, a literature review seeks to describe,
essential if the research process is to be successful. summarize, evaluate, clarify, and/or integrate
Research studies are usually undertaken within the content of primary reports.
the context of an existing knowledge base, because —H.M. Cooper (1988)
research cannot be conducted in an intellectual l A literature review is an evaluative report of

vacuum. Before starting any research, a literature information found in literature related to a se-
review of previous studies and experiences related lected area of study. The review describes, sum-
to the proposed investigations has to be done. One marizes, evaluates and clarifies this literature. It
of the most satisfying aspects of the literature re- gives a theoretical base for the research and
view is the contribution it makes to the new helps to determine the nature of research.
knowledge, insight and general scholarship of the —Queensland University (1999)
researches. Health science research may be consid- l A literature review is a body of text that aims to

ered a continuing process in which knowledge review the critical points of knowledge on a
gained from earlier studies is an integral part of particular topic of research.
research in general. —ANA (2000)
l A literature review is an account of what has

been already established or published on a par-

MEANING OF LITERATURE REVIEW ticular research topic by accredited scholars and
A review of literature is a description and analysis researchers.
of the literature relevant to a particular field —University of Toronto (2001)
or topic. It provides an overview of what work l Literature review is defined as a broad, compre-

already had been carried out, who are the key hensive, in depth, systematic and critical review
researchers who did that work, which of the ques- of scholarly publication, unpublished printed
tions are already answered regarding a particular or audio visual materials and personal commu-
area of research interest, what methods and meth- nications.
odologies were used to answer the particular ques- —S.K. Sharma (2005)
tions and what are the prevailing theories and Literature review could just be an annotated bibli-
hypotheses. ography, which purely describes the facts and may
l Literature review may be defined as a selection add comments to bring out themes and trends; or
of available documents, both published and literature review may be a critical analysis of facts of
unpublished on the topic of research interest, the field of research interest, which identifies the
which contain information, ideas, data, and evi- strengths, weaknesses and gaps in existing literature
dence written from a particular standpoint to and identifies the questions which further need to
fulfil certain aims or express certain views on be answered. The analytical review not only pres-
the nature of the topic and how it is to be inves- ents the summary of the facts but evaluates the
tigated and the effective evaluation of these relationships and contrast findings of the facts in
documents in relation to the research being literature; thus the key themes are identified.
proposed.
l A literature review uses as its database, reports

of primary or original scholarship and does

IMPORTANCE OF A LITERATURE
not report new primary scholarship itself. The REVIEW
primary reports used in the literature may Literature reviews provide a practical guide to a
be verbal, but in the vast majority of cases, particular topic. For health care professionals, they
reports are written documents. The types are useful reports that keep them updated with
of scholarship may be empirical, theoretical, what is present in the field. For scholars, the depth
critical/analytic, or methodological in nature. and breadth of the literature review highlights the
48 Research Methodology and Biostatistics

credibility of the writers in their respective fields. Literature review allows the reader to be updated
Literature reviews also provide a solid background with the state of research in a field and any contra-
for a research study. Comprehensive knowledge of dictions that may exist which challenge findings of
the literature of the field is essential to most other research studies. Furthermore, literature re-
research studies. Review of relevant literature can view enhances researchers’ knowledge. It helps to
help in the fulfilment of the following objectives: develop research-investigative tools and to improve
l Identification of a research problem and devel- research methodologies. It also provides knowledge
opment or refinement of research questions about the problems faced by previous researchers’
l Generation of useful research questions or projects/ while studying the same topic. Besides enhancing
activities for the discipline researchers’ knowledge about the topic, writing a
l Orientation to what is known and not known literature review helps to do the following:
about an area of inquiry to ascertain what l Place each in the context of its contribution to

research can best contribute to knowledge the understanding of the subject under review
l Determination of any gaps or inconsistencies in l Describe the relationship of each study to other

a body of knowledge research studies under consideration

l Provision of evidence that a selected research l Identify new ways to interpret and shed light on

problem is of importance any gaps in previous research

l Discovery of unanswered questions about l Resolve conflicts among seemingly contradictory

subjects, concepts or problems previous studies

l Determination of a need to replicate a prior l Identify areas of prior scholarship to prevent

study in different study settings or different duplication of effort

samples or sizes or different study populations l Point a way forward for further research

l Identification of relevant theoretical or concep- l See what has and has not been investigated

tual framework for research problems l Develop general explanation for observed varia-

l Identification or development of new or tions in a behaviour or phenomenon

refined clinical interventions to test through l Identify potential relationships between con-

empirical research cepts and identify researchable hypothesis

l Description of the strengths and weaknesses of l Learn how others have defined and measured

designs/methods of inquiry and instruments key concepts

used in earlier research work l Identify data sources that other researchers have

l Development of hypothesis to be tested in a used

research study l Develop alternative research projects

l Assistance in planning the methodology of the l Discover how a research project is related to the

present research study work of others

l Assistance in the development of research l Place one’s original work (in case of thesis or dis-

instruments sertation) in the context of the existing literature

l Identification of suitable design and data collection

methods for a research study

l Assistance in interpreting study findings and in
TYPES OF LITERATURE REVIEWS
developing implications and recommendations Literature review may be carried out for different
purposes using different methodologies. There-
fore, on the basis of purposes and methodologies,
PURPOSES OF A LITERATURE REVIEW literature review would be broadly classified into
The purpose of a literature review is to convey to the following categories:
the reader previous knowledge and facts established 1. Traditional or narrative literature review:
on a topic, and their strengths and weaknesses. This type of literature review presents the
 Literature Review 49

summary of literature and draws conclusions evidence of reliability and validity of litera-
about the topic in question. In this, the body ture for readers, it is suggested to present the
of literature is composed of a summary of precise criteria used for certain aspects of
facts on the subject area from primary and the literature review, i.e., (i) formulation of
secondary sources. The main aim of this type the research question, (ii) inclusion and ex-
of literature review is to gather a volume of clusion criteria for literature, (iii) selection
literature in a specific subject area, and sum- and access of the literature, (iv) criteria to
marize and synthesize it. assess the quality of literature reviewed and
The primary purpose of traditional literature (v) analysis, synthesis and dissemination of
review is to offer a comprehensive back- the findings.
ground of the subject under study to under- In contrast to narrative literature review,
stand the subject area, identify gaps, incon- systemic review uses explicit and rigorous
sistencies in the body of knowledge and criteria to identify, critically evaluate and
highlight the significance of a new research synthesize all the published and unpublished
question. Thus, traditional literature is es- literature available on a particular topic of
sential for getting in-depth insight into the interest. However, traditional or narrative
subject area, refining the research question literature only summarizes the facts and
and hypotheses and supporting the impor- findings of a number of studies.
tance of new research through identifying 3. Meta-analysis: Meta-analysis is an advanced
the gaps and inconsistencies in existing form of systemic literature review, which
literature. It is also believed that traditional includes a large body of findings from quan-
literature review is useful for developing con- titative studies and computes the statistical
ceptual and theoretical frameworks. Further- analysis for drawing integrated and cumula-
more, it is argued that narrative literature tive inferences and conclusions. The main
may also be carried out for the nonresearch aim of meta-analysis is to involve the find-
reasons presented in Table 3.1. ings of several quantitative studies on a
2. Systemic literature review: Systemic litera- single subject area and carry out statistical
ture review is significantly different from computations on them using standardized
the traditional or narrative literature review; statistical techniques and procedures. Thus,
it includes a more systemic and rigorous to draw cumulative and integrated inferences
approach to literature review in a specific and conclusions to improve the quality of
subject area. The main aim of systemic lit- evidence, a particular area of research can be
erature review is to find the answers for served as the highest level of evidence for
well-focused questions of clinical practice. generating evidence-based practices for the
It is suggested that a systemic literature discipline.
review should include the methods used 4. Metasynthesis: Contrary to meta-analysis,
for searching, evaluating and synthesizing metasynthesis is the nonstatistical technique
the literature. In addition to presenting the used to integrate, evaluate and interpret the
findings of multiple qualitative research studies.
Table 3.1. ​Nonresearch reasons for undertaking Findings from phenomenological, grounded
a literature review theory or ethnographic studies may be inte-
• As an academic assignment
grated and used. Such studies may be combined
• To update current knowledge and practice on to identify their common core elements and
a discipline/field themes. Unlike meta-analysis, where the ultimate
• To develop and refresh practice guidelines intention is to reduce findings, metasynthesis
• To construct specific procedures and policies involves analysing and synthesizing key elements
50 Research Methodology and Biostatistics

in each study, with the aim of transforming indi- interpretation of the primary work; it is neces-
vidual findings into new conceptualizations and sary then to review the primary source whenever
interpretation. possible to ensure accuracy. However, secondary
sources may be used when primary sources are
not available or if researchers want external
SOURCES OF A LITERATURE opinions on an issue or problem or even the re-
Literature can be reviewed from two main sources, sults of their own research.
i.e., primary and secondary sources; however, in
research studies for literature review, researchers
mostly rely on primary sources. Example of a Secondary Source
A literature review on patient experiences in the ICU:
PRIMARY SOURCES Stein-Parbury, J. & Mckinley, S. (2000). Patient experi-
Primary sources are original, peer-reviewed and ences of being in an intensive care unit: a select litera-
published research journal articles reported by ture review. American Journal of Critical Care, 9, 20–27.
original researchers. In other words, primary
sources are the research reports which are de-
scription of studies written by researchers who The main sources from where literature can be
conducted them. A primary source is written by a searched are as follows (Fig. 3.1):
person who developed the theory or conducted 1. Electronic sources: Computer-assisted litera-
the research, or is the description of an investiga- ture search has revolutionized the review of
tion written by the person who conducted it. literature. These searches, however, for a vari-
Most primary sources are found in published lit- ety of reasons may not provide the desired
erature; for example, a medical research article. references. Electronic literature search through
A credible literature review reflects the use of Web may be very useful, but sometimes it can
mainly primary sources. be time consuming and unpredictable because
there are many websites and web pages that can

Example of a Primary Source

An original qualitative study on patient experiences
in the ICU: Sharma, S.K. (2005). Feeling safe: the Electronic Books
psychosocial needs of ICU patients. Indian Journal database
of Psychiatry, 32, 361–367.
Magazines
and Journals
newspapers
SECONDARY SOURCES Resources
of literature
Secondary source research documents are de- review
scription of studies prepared by someone other Conference
Research
than the original researcher. They are written by reports papers
people other than the individuals who developed
the theory or conducted the research. Secondary
sources include comments and summaries Encyclopaedia
and Theses
of multiple research studies on one topic, e.g., dictionary
systemic reviews, meta-analysis and metasynthe-
sis. Secondary sources are usually paraphrased
and may be based on the secondary author’s Fig. 3.1. ​Sources of literature review.
 Literature Review 51

lead to information overload and confusion. (www.pubmed.com or http://www.ncbi.

However, currently it is one of the most impor- nlm.nih.gov/pubmed) and new citations
tant sources of literature searches. are added every week.
General literature search can be conducted - MEDLINEPlus (http://www.medlineplus.
through search engines, like Yahoo (www. gov/): It is a free electronic database, which
yahoosearch.com), Google (www.google. has a large amount of research databases
com), MSN search, Lycos, WebCrawler, Alta from the United States National Institutes of
Vista or Excite; and scientific health care lit- Health and other valid health-related sources.
erature may be located at several available - PubMed Central (http://www.pubmed-
database sources. It is appropriate to define central.nih.gov/): Pubmed Central (PMC)
the database and to indicate how it is different is a free full-text archive of biomedical and
from a search engine. As mentioned earlier, life sciences journal literature of the United
Google is a search engine and MEDLINE is a States National Institutes of Health, which
database. is primarily developed to take the lead in
A search engine is a collection of software preserving and maintaining, unrestricted
programmes that collect information from access to the electronic literature.
the Web, index it and put it in a format that - PubMed (http://www.ncbi.nlm.nih.gov/
is accessible visually on screen at an on-site pubmed/): This is a free online search en-
library or in downloadable, readable writ- gine, which is produced by the United
ten format. In contrast, a database is an States National Library of Medicine, which
organized body of related information primarily provides free access to abstracts
arranged for speed of access and retrieval. and selected full articles of biomedical
In other words, a database is a storage loca- sciences and health.
tion like a library, where information is - Cochrane database of system reviews:
stored, catalogued, maintained and up- Health care–related literature can be
loaded systematically. Two main types of searched from this source, available at
databases are available, i.e., bibliographic http://www.cochrane.org.
and full-text databases. The bibliographic - ERIC: The ERIC database is the largest
database gives only direction on where to source of education information. A free
find the information, whereas the full-text search may be carried out at http://www.
database contains the information itself eric.ed.gov.
(full research articles in a downloadable - PsycINFO: The PsycINFO database belongs
format). However, the most useful and rel- to the American Psychological Association,
evant databases of biomedical sciences and and covers literature from psychological or
health are as follows: related disciplines. It may be searched at
- MEDLINE (Medical Literature Analysis http://wwwpsychinfo.com.
and Retrieval System Online): It is the - World Cat (http://www.worldcat.org/): It
United States National Library of Medi- is a free database of more than 10,000
cine’s database, which was initially named libraries, which facilitates investigators to
as MEDLARS. It includes the bibliographic identify and locate books, research reports,
database of academic research articles from journal articles, synopsis, thesis and
the fields of medicine, nursing, dentistry, websites.
pharmacy, veterinary medicine, the health - WHO Countries: country-by-country
care system and preclinical biomedical sci- links to health data (http://www.who.int/
ences. This database is freely accessible on countries/en/): It is a free database, which
the Internet via the PubMed interface is popularly known as WHOSIS (WHO
52 Research Methodology and Biostatistics

Statistical Information System). It provides database provides more than 1,500,000

the most comprehensive, recent and up- citations of rehabilitation research
dated data related to health and health in- published from 1990 onwards; however,
dicators of about 193 WHO member states it does not include the research of
including India. This database obtains the the USA.
data of a particular country available - Health Evidence (http://www.healthevi-
through Health Ministries and Central dence.org/search-login.aspx): It is a free
Statistics Offices. online database developed by McMaster
- TOXNET (Toxicology Literature Online) University, Australia, which provides a
(http://toxnet.nlm.nih.gov/cgi-bin/sis/ database of systemic reviews on public
htmlgen?TOXLINE): It is a free biblio- health.
graphic database, which is archived by the - National Center for Complementary and
United States National Library of Medicine Alternative Medicine (http://nccam.nih.
with references to literature on toxicological, gov/): This is a database of scientific
pharmacological, biochemical and physio- research of complementary and alterna-
logical effects of drugs and other chemicals. tive medicine developed by the National
- TestLink (http://www.ets.org/test_link/): It Center for Complementary and Alterna-
is a free collection of standardized tests and tive Medicine (NCCAM), USA.
research tool/instruments worldwide - Networked Digital Library of Theses &
mainly from the USA, Canada, Great Brit- Dissertations (http://www.ndltd.org/):
ain and Australia. It has an archive of about It is a free database of full-text theses and
more than 20,0000 tests, research instru- dissertations from many universities
ments and other measurement devices. worldwide.
- African Index Medicus (http://indexmedi- - PEDro (Physiotherapy Evidence Database)
cus.afro.who.int/): It is a free online data- (http://www.pedro.org.au/): It is a free
base of African biomedical information database of more than 18,000 research stud-
and health literature produced by WHO to ies in the field of physiotherapy. Abstracts as
promote global exposure to untapped well as full-text articles are available in this
information in books, reports and research database.
studies from the African countries. - PLOS Medicine (http://medicine.plos-
- Bioline International (http://www.bio- journals.org/perlserv/?request5index-
line.org.br/): It provides a free database of html&issn51549-1676): The Public
peer-reviewed biomedical sciences jour- Library of Science is an open access
nals published in developing countries, venue of peer-reviewed research studies
which has open access. of medicine and health.
- BioMed Central (http://www.biomedcen- - POPLINE (http://www.popline.org/): It
tral.com/): It provides free open access of is the world’s largest free database consist-
more than 250 peer-reviewed journals in ing of over 300,000 research articles
fields of medical, health and biomedical on reproductive health.
research, which is developed by BioMed - PQDT Open (http://pqdtopen.proquest.
centre; all the original research articles com/): It is a free database of full-text open
published by this organization are made access to theses and dissertations at the
freely available as soon as they are pub- postgraduate and doctoral level.
lished. - REHABDATA (http://www.naric.com/
- CIRRIE Database of International research/rehab/): It is a free database of
Rehabilitation Research (http://cirrie. more than 65,000 research abstracts on
buffalo.edu/search/): This free source rehabilitation and disability.
 Literature Review 53

- CINAHL (Cumulative Index to Nursing the free open online access for literature
and Allied Health Literature): Accessible review):
at http://www.cinahl.com, it contains cita- - http://www.aidsinfo.nih.gov (HIV/AIDS
tions of nursing literature published after information)
1988. Even full-text articles are available on - http://www.hazmap.nlm.nih.gov (infor-
CINAHL Plus, a paid web page. mation on hazardous agents)
- British Nursing Index: The British Nursing - http://www.child.nih.gov (combined health
Index (BNI) is a leading UK nursing data- information database)
base providing bibliographic references to - http://www.toxinet.nlm.nih.gov (toxicol-
journal articles from all the major British ogy database network)
nursing and midwifery journals, as well as 2. Printed sources: Printed sources are also used for
a selection of English-language interna- literature review. Printed research summary may
tional journals. It is available from the be located from published abstracts, such as Medi-
Royal College of Nursing Library e-library cal Research Abstract, Psychological Abstracts, Dis-
database: www.rcn.org.uk/elibrary. sertation Abstract International, Masters Abstract
- Registry of Nursing Research: Sigma Theta International, etc. References to other printed
Tau International Honor Society of Nursing sources may be located through indices, such as
makes this database available through its Index Medicus, Abridged Index Medicus, Index
Virginia Henderson International Nursing Medicus Chronology, Index Medicus for the East-
Library. Access to this database has been re- ern Mediterranean Region (IMEMR) and Index
designed and made easier for users to obtain Medicus of the South-East Asian Region
evidence and scientific findings from more (IMSEAR). Following are the main printed sources
than 2200 research articles and conference that can be used to review relevant literature:
abstracts. The research abstracts are searched  Journals: There are several national and inter-

via key words, author and title of the research national journals that can be used to review
study. Free access to this database may be research-related literature. The published
found at http://www.nursinglibrary.org. journal articles may be reviewed either in the
- Other online databases: Many other online form of hard copy or electronically in the
databases can be searched for free by health form of soft PDF copy.
care professionals from the following  Research reports: There are national scien-

websites (Table 3.2 also presents some of tific bodies/organizations that fund research

Table 3.2. Free online access to scientific/health science literature

Although complete open access to research results is not yet available, there are sources that health care
professionals can use to access free evidence for practice. In the USA, the National Institutes for Health
operates a website (www.guideline.gov) that publishes a broad range of practice guidelines. Some of the
international open access resources include the following websites, which provide free access to many
professional journals:
• The Alliance for Taxpayer Access: www.taxpayeraccess.org
• BioMed Central: www.biomedcentral.com
• Budapest Open Access Initiative: www.soro.org/openaccess
• DOAJ (The Directory of Open Access Journals): www.doag.org
• PLOS: Public Library of Science: www.plos.org
• PubMed Central: http://www.pubmedcentral.nih.gov/
• SciDev Net: www.scidev.net
• SPARC (The Scholarly Publishing and Academic Resource Coalition): www.arl.org/sparc
54 Research Methodology and Biostatistics

projects such as the Indian Council of Medi-  Books: Books are primarily written from the
cal Research (ICMR), Department of Science knowledge gathered from life experiences or
and Technology (DST), etc., where on com- previously published literature in journals,
pletion of an extramural research project, the periodicals, books etc. Therefore, sometimes
research report is submitted by the principal they may be a good source for literature
investigator. These reports may serve as a review.
good source of literature review.
 Unpublished dissertations and theses: It is
observed that many postgraduate and doc-
STEPS OF A LITERATURE REVIEW
toral theses are not published by the research- Literature review is a systemic and orderly process,
ers; however, they may be a good source of which is undertaken for various reasons, such as a
literature review. step of the research process or eliciting information
 Magazines and newspapers: Magazines and for developing policies and evidence-based prac-
newspapers also may be a source of literature tices. It is identified that researchers find
review where the news or article published literature review a daunting task and frequently
may be the area of interest for the particular have many questions ranging from where to start,
researcher. how to select the themes, where to search the litera-
 Conference papers and proceedings: ture, how to search the literature, how many articles
Scientific conference generally attracts many to include, how to synthesize and organize the lit-
research papers, which are compiled and erature and how to write the literature. The follow-
published in the form of conference souve- ing discussion presents the steps of literature review
nirs or proceedings; this could serve as a in a research process and answers many of the
source of literature review. afore-mentioned pertinent questions related to the
 Encyclopaedias and dictionaries: Some his- methods of literature review (Fig. 3.2):
torical information, definitions and concepts Step 1. Understanding the concepts of the re-
may be referred from encyclopaedias and/or search problem and identifying the key search
dictionaries. terms: The journey of literature search begins

Writing the literature

review
V Analysing and
synthesizing the
literature
IV
Searching the
literature

III
Identifying the
relevant sources
II Understanding the concepts
of the research problem and
identifying the key search terms
I

Fig. 3.2. ​Step of literature review.

 Literature Review 55

with understanding and examining the con- databases available for research; thus it is very
cepts of the research problem so that essential essential to judiciously decide the relevant elec-
key search terms may be identified. The key tronic database for literature search.
search terms serve as a foundation for the over- Step 3. Searching the literature: Having selected
all search of literature. These search terms may relevant sources of literature review, the next
be elements that are included in the problem step is to identify in a structured way, the ap-
statement, variables that appear in the research propriate and related information. The tradi-
question and characteristics of the population tional method of literature search is manual
of interest. Boolean operators (AND, OR, NOT) search from the hard copy of journals that are
are then used to combine the search terms into specially related to the topic of interest or those
a search statement. A detailed discussion about that are likely to cover the topic. In this method,
the use of Boolean operators is given in Step 3 the initial reference of a journal is obtained
of the literature review. Examples of a research from indices developed for this purpose, such
problems broken into key search terms are as Index Medicus and/or from the reference/
shown in Table 3.3. bibliography of the initially located articles of
Step 2. Identifying the relevant sources: After related topic. This is a slow and tedious task but
understanding the concepts of the research often a rewarding way of searching for articles.
problem and identification of key search terms, However, this may be facilitated through identi-
now it is time to locate the relevant sources for fying the initial related reference from the elec-
literature review. The common sources used for tronic reference/abstract database, such as
literature review by health care professions are PubMed (www.pubmed.com) and then search-
discussed in the previous topic of this chapter. ing full articles in hard copies of related pub-
Journals are considered more superior and up- lished journals.
to-date source than books and other resources, Nowadays searches are undertaken most
because books are considered outdated by the commonly using computer and electronic da-
time they are published after a long process. tabases or search engines. Therefore, identify-
However, the contribution of books can not be ing and using keywords is the most critical act
ignored in literature review. But in the present in literature search. However, keywords need
scenario, electronic literature sources are the careful consideration in order to select terms
most commonly preferred resources for litera- that will generate the data being sought. For
ture review because of easier and quicker access American databases, such as MEDLINE, the
to a vast amount of research database than keywords used to identify terms may differ
manual search. There are a vast variety of from the British in spelling and meaning

Table 3.3. Identification of key search terms from the research problem
Research Problem Examples of Relevant Key Search Terms
A descriptive study on metabolic complications of trau- • Traumatic AND brain injury
matic brain injury in children admitted in JPN Trauma • Metabolic response AND trauma
Center, AIIMS, New Delhi • Paediatric AND brain injury
• Brain injury AND children
• Hypothermia AND brain injury
• Hypothermia AND paediatric
A comparative study on postoperative length of stay in • Bariatric surgery AND exercise
hospital among ambulatory vs. bedside peddler patients • Effects AND postoperative AND exercise
admitted in the surgical department of AIIMS, Jodhpur • Exercise AND length of post-op stay
• Ambulation AND length of stay in hospital
56 Research Methodology and Biostatistics

(e.g., tumour/tumor, paediatric/pediatric, summary system (or a combination of

transcultural/multicultural). It is a good idea to both) to assist the review process. There are
consider alternative keywords with similar different criteria suggested for the indexing
meanings that might elicit further information; and summarization but generally they in-
for example, if you are undertaking a review in clude the author, year of study, title of arti-
an aspect of pressure ulcers, you would need to cle, purpose and methodology of study,
include terms, such as pressure sores and decu- main findings and outcome of study. In ad-
bitus ulcers, to access older material. Some of dition, it is important to include the re-
these alternative keywords can be gleaned from viewer’s key comments and thoughts about
the database thesaurus. Another strategy is the reviewed article. It is suggested to keep a
‘Boolean operators’ where keywords are com- source and full reference to ensure better
bined to facilitate the search. The most com- record keeping and to avoid the problem of
monly used Boolean operators are ‘AND’, ‘OR’ later searching for references/source of in-
and ‘NOT’. They are presented in Table 3.4 with formation. A review may include not only
their purposes and relevant examples. primary sources but also secondary and
Step 4. Analysing and synthesizing the literature: nonresearch articles. Possible headings,
After collecting the relevant literature, it is the adapted from appraisal tools for various
stage to analyse and synthesize the collected types of literature are outlined in Table 3.5.
literature. There are several substages or systems The reading is an ongoing process during
to achieve this purpose. The most popular and answering the afore-specified questions for
simple method is referred to as the preview, each article, which finally ends with writing
question, read, summarize (PQRS) system. This a short summary of each article and may
system not only keeps the reviewer focused and include key thoughts, comments, strengths
consistent but also ultimately facilitates easy and weaknesses of the article/publication. It
identification and retrieval of the relevant should be written in the researcher’s own
literature. words to facilitate understanding of the
(a) Preview/Overview: This begins with an ini- literature, and it also forms a good basis for
tial quick overview or preview of the arti- the writing of the review.
cles through the abstract or summary of the Step 5. Writing the literature review: This is the
articles given at the beginning of each arti- final step of literature review after completing
cle to get a sense of what the article is about the adequate appraisal of literature. It is time to
and see if it is worthy of inclusion for fur- give a serious thought about how to write
ther reading. a draft of the literature review. It is essential
(b) Question, read and summarize: In the to avoid long and confusing words and jargon
question stage, questions are asked of each and keep sentences short with one clear mes-
article under review. Here, several writers sage. The spelling and grammar must get due
have suggested using an indexing or attention, consistency must be maintained and

Table 3.4. Examples of Boolean operators and their purposes

Command Purpose Example
AND Look for articles that include all the identified Brain injury AND head injury
keywords
OR Look for articles that include any of the identified Corticosteroids OR steroids OR
keywords glucocorticoids
NOT Exclude articles that contain this specific keyword Viral bronchitis NOT bacterial bronchitis
 Literature Review 57

Table 3.5. Summary of information required in review

Primary Source Secondary Source – Review Nonresearch Literature
Title Title Title
Author (year) Author (year) Author (year)
Full reference Full reference Full reference
Purpose of study Review question/purpose Purpose of paper
Type of study Key definitions Credibility
Research setting Review boundaries Quality
Methods of data collection Appraisal criteria Coherence
Main findings Synthesis of studies Recommendations
Recommendations Summary/conclusions Key thoughts/comments
Key thoughts/comments Key thoughts/comments (e.g., strengths/weaknesses)
(e.g., strengths/weaknesses) (e.g., strengths/weaknesses)

English used. The organization of the literature a good and poor literature review, and Table 3.8
in an objective manner and the structure of presents examples of better and bad written
the review are crucial to its comprehensiveness literature.
(approaches of organizing the literature are (a) Introduction: While writing the introduction,
given in Table 3.6). Nonetheless, it is imperative the following steps should be taken care of:
to be logical and there are some important ele- - Initiate the introduction with definition
ments that need to be included in all literature of the general topic and issues under
reviews. Primarily, an introduction, body, and study, thus offering the background con-
conclusion should be included in the written text to the literature.
report (Fig. 3.3). The length of literature reviews - Highlight the trends, similarities, con-
varies and word limits and specific criteria for trasts and gaps in facts, mythologies and
the overall construction depend on the guide- procedures of available literature, identi-
lines prescribed by a particular institute/univer- fying the importance of the new topic
sity. Table 3.7 illustrates the characteristics of under study.

Table 3.6. Approaches of organizing the literature review

Approaches Description
Thematic organization (dividing the Literature is organized under different themes identified from the
literature into themes or categories) literature and brainstorming; it is the most popular approach of or-
ganizing the literature; it allows integration of theoretical and em-
pirical (research) literature; however, care must be taken to
ensure that the themes are clearly related to the literature
Chronological organization (presenting Literature is divided into time periods; it is useful when examining
the literature chronologically) the emergence of a phenomenon over a time period
Theoretical vs. methodological organiza- Discussion of theoretical literature is followed by exploration of
tion (exploring theoretical and methodological literature that would give some indication of
methodological literature) why a particular research design might be appropriate for
investigating the topic; it is useful when the body of literature is
largely theoretical with little or no empirical (research) literature
Theoretical vs. empirical organization Reviewer has both theoretical and empirical literature, and they are
(examining theoretical literature and discussed separately; the purpose of the researcher is to describe
empirical literature in two sections) rather than critically review
58 Research Methodology and Biostatistics

Table 3.8. Example of a better and a bad written

literature review
Introduction
Example of a Better Review
The victims of domestic violence suffer a range of
consequences, from lowered self-esteem and loss
of self-confidence to withdrawal from social in-
teraction, changed career goals and depression
Elements

(Sharma, Saini and Arora, 2003; Rao and Bhai,

Co

2004; Joshi and Mathew, 2009); for example,

dy
nc

Bo
lu

Sharma, Saini and Arora, (2003) noted that 26%

sio
n

of women said that they avoided going to social

functions or interacting openly with family mem-
bers because of the risk of domestic violence
Fig. 3.3. ​Elements of written review literature.
Example of a Bad Review
Domestic violence has many consequences;
Table 3.7. Characteristics of a good and a poor Sharma, Saini and Arora (2003) found that
literature review some women said that they avoided going
to social functions or interacting openly with
Good Literature Review Poor Literature Review family members because of the risk of
• It is a synthesis of • It is an annotated domestic violence; they also found that
available research bibliography rural and urban women reacted differently;
• It is a critical evaluation • It is confined to their research was conducted through a
description survey of 2000 rural and urban women
• It has appropriate • It is narrow and dwellers; Rao and Bhai’s study in domestic
breadth and depth shallow violence among families (2004) lists many
• It has clarity and • It is confusing and problems created by domestic violence; in
conciseness long winded their excellent book, Joshi and Mathew
• It uses rigorous and • It is constructed in an (2009) give a long list of problems faced by
consistent methods arbitrary way victims of domestic violence

- Point out thoughts and comments of the literature, remembering that space
reviewer’s view points and explain denotes significance.
the criteria of critical appraisal and - Use ‘umbrella’ sentences at the beginning
organization of literature review. of paragraphs, signpost throughout and
(b) Body: The following measures need to be brief ‘so what’ summary sentences at inter-
undertaken while writing the body of the mediate points in the review to aid in un-
literature. derstanding comparisons and analysis.
- Group research studies and other types of (c) Conclusion: The points to be taken care of
literature (reviews, theoretical articles, in the conclusion are as follows:
case studies) according to common de- - Mention the main contribution of
nominators, such as qualitative versus important studies to the body of knowl-
quantitative approaches, conclusions of edge for the particular topic under study.
authors, specific purposes or objectives, - Specify the gaps, study findings and
chronology and so on. inconsistencies in methodologies, theo-
- Summarize individual studies or ries and procedures of the literature.
articles with as much as or as little - Conclude by mentioning the most
detail as each merit according to significant findings and pertinent issues
its comparative importance in the related to the topic under study.
 Literature Review 59

bibliography/reference list. The common prac-

POINTS TO BE CONSIDERED
tice is that the reviewer does not list references
FOR A LITERATURE REVIEW in the bibliography that are not directly cited in
Specifically, here are some points to address when the literature review or elsewhere in the paper/
writing a literature review. While reviewing litera- thesis.
ture, one needs to assess it (clearly understand and l Organization of literature review: A literature
state the claim) and analyse it (evaluate its reli- review is organized by subtopic, not by indi-
ability, usefulness, validity). The following points vidual reference. In a typical literature review,
are to be considered when assessing, analysing and the writers may cite several references in the
writing a literature review: same paragraph and may cite the same refer-
l Be specific: It is always better to mention only ence in more than one paragraph, if that
specific features of the study methodology and source addresses more than one of the subtop-
findings of the study, rather than have an ics in the literature review. Typically, discus-
in-depth analysis of each point and a long sion of each source is quite brief. The
description of the study. contribution the present reviewers make is
l Be selective: The researcher should narrow organizing the ideas from the sources into a
down a lot of information into a small space for cogent argument or narrative that includes
literature review. Just the most important points their perspectives.
(i.e., those most relevant to the review’s focus) l Referring the original source: The reviewer
must be mentioned in each work of review. should focus on citing the material that origi-
l Focus of current topics: The researcher needs nates with each reference. This may require a
to analyse some points, such as if it is a current careful reading of the reference. If the reference
article, and if not, how old it is; has its claims, author refers to another source whose ideas are
evidence or arguments been superseded by relevant or interesting, it is better to track and
more recent work; if it is not current, then if it use that original reference.
is important for historical background, etc. l Write in own language: It must be ensured by
l Ensure evidence for claims: The researcher the researcher that the literature is written in his
should focus on what support is given for or her own words; paraphrasing must be avoided.
claims made in the literature. What evidence However, the opinion of the original author
and what type (experimental, statistical, anec- must be accurately presented in the written lit-
dotal, etc.) of evidence is offered? Is the evi- erature.
dence relevant and sufficient? What arguments l Simple and accurate sentence structure: A
are given? What assumptions are made, and researcher should use simple sentences and
are they warranted? must avoid errors of grammar and punctua-
l Focus on sources of evidence: Researchers tion. Express one idea in a simple sentence.
should ensure the reliability of the sources of Ensure that all sentences have a subject, verb
the evidence or other information – if they are and object.
from the author’s own experiments, surveys, l Group the ideas: Group the sentences that
historical records, government documents, etc. express and develop one aspect of the particular
He or she should check how reliable those topic. Use a new paragraph for another aspect
sources are. or another topic.
l Account of contrary evidence: Does the author l Effective use of transition words: Use words
take into account contrary or conflicting evi- that link paragraphs and which show contrast
dence and arguments? How does the author and development to the argument, e.g., hence,
address disagreements with other researchers? therefore, but, thus, as a result and in contrast.
l Reference citation: Any references cited in l Avoid abbreviations: Avoid technical terms,
the literature review must be included in the jargons and abbreviations.
60 Research Methodology and Biostatistics

MULTIPLE CHOICE QUESTIONS

1. A literature review is 6. The literature review should occur:
(a) Everything that is known about a subject (a) Near the end of the research process
(b) An analytical summary of research findings (b) Shortly before the analysis of the problem
(c) All approved data on a research topic (c) Early in the research process
(d) A compilation of all positive results of research (d) None of the above

2. When conducting a review of literature on a partic- 7. A database differs from a search engine in the
ular subject, the researcher should: following manner:
(a) Read all available material on the subject (a) A database stores the information
(b) Read the whole journal article and then decide (b) A search engine takes you to the information
whether or not it is useful (c) Databases are specialized by area of knowledge
(c) Read strategically and critically (d) All of the above
(d) Read fully only those texts that appear to agree
with his/her point of view 8. A database that is considered the premier bibliogra-
phy database for providing access for biomedical
3. The purpose of a literature review is to: literature is
(a) Use the literature to identify the present knowl- (a) Google
edge and what is unknown (b) MEDLINE
(b) Assist in defining the problem and operational (c) Yahoo
definition (d) MSN search
(c) Identify strengths and weaknesses of previous
studies 9. The main difference between a research article and
(d) All of the above a nonresearch article is
(a) A research article reports statistics on survey
4. Why do you need to review the existing literature? and a nonresearch article does not
(a) To make sure you have a long list of references (b) A research article describes the research by the
(b) Because without it, you could never reach the original author
required word count (c) A nonresearch article describes the methods of
(c) To find out what is already known about your how the study was conducted
area of interest (d) A nonresearch article conducts analysis and
(d) To help in your general studying statistical testing on the data presented in the
article
5. To read critically means:
(a) Taking an opposing point of view to the ideas 10. When conducting a literature review, it is advisable to:
and opinions expressed (a) Seek most of the information from internet
(b) Skimming through the material because most (b) Gather literature from books
of it is just padding (c) Gather literature from journals
(c) Evaluating what you read in terms of your own (d) Seek assistance from a librarian
research questions
(d) Being negative about something before you read it

Answers of Multiple Choice Questions

1. a; 2. c; 3. d; 4. c; 5. c; 6. c; 7. d; 8. b; 9. b; 10. c.
 Literature Review 61

REFERENCES
Afolabi, M. (1992). The review of related literature in Coughlan, M., Cronin, P., Ryan, F. (2007). Step-by-step guide to
research. International Journal of Information and Library critiquing research. Part 1: quantitative research. British Journal
Research, 4(1), 59–66. of Nursing, 16(11), 658–663.
Anson, C. M., & Schwegler, R. A. (2000). The Longman handbook for Cronin, P., Ryan, F., & Coughlan, M. (2008). Undertaking a litera-
writers and readers (2nd ed.). New York, NY: Longman. ture review: a step-by-step approach. British Journal of Nursing,
Beecroft, C., Rees, A., Booth, A. (2006). Finding the evidence. In K. 7(1), 38–43.
Gerrish & A. Lacey (Eds.), The research process in nursing Ely, C., & Scott, I. (2007). Essential study skills for nursing. Edinburgh:
(5th ed., pp. 90–106). Philadelphia, PA: Blackwell Publishing. Elsevier.
Beyea, S., & Nicholl, L. (1998). Writing an integrative review. Hart, C. (1998). Doing a literature review. London: Sage
AORN Journal, 67(4), 877–880. Publications.
Bourner, T. (1996). The research process: four steps to success. In Hek, G., & Langton, H. (2000). Systematically searching and re-
T. Greenfield (Ed.), Research methods: guidance for postgradu- viewing literature. Nurse Researcher, 7(3), 40–57.
ates. London: Arnold. Hek, G., & Moule, P. (2006). Making sense of research: an introduc-
Bruce, C. S. (1990). Information skills coursework for postgradu- tion for health and social care practitioners (3rd ed.). London:
ate students: Investigation and response at the Queensland Sage Publications.
University of Technology. Australian Academic and Research Hendry, C., & Farley, A. (1998). Reviewing the literature: a guide
Libraries, 21(4), 224–232. for students. Nursing Standard, 12(44), 46–48.
Bruce, C. S. (1993). When enough is enough: or how should re- Jones, R., Bizzaro, P., & Selfe, C. (1997). The Harcourt Brace guide
search students delimit the scope of their literature review? In to writing in the disciplines. New York, NY: Harcourt Brace.
Challenging the conventional wisdom in higher education: Lamb Sandra, E. (1998). How to write it: a complete guide to every-
Selected contributions presented at the Nineteenth Annual thing you’ll ever write. Berkeley, CA: Ten Speed Press.
National Conference and Twenty-First Birthday Celebration of Leedy, P. D. (1997). Practical research: planning and design
the Higher Education Research and Development Society of (6th ed.). Upper Saddle River, NJ: Merrill.
Australasia Inc., HERDSA, University of New South Wales, Libutti, P., & Kopala, M. (1995). The doctoral student,
Sydney, Australia, pp. 435–439. the dissertation, and the library: a review of the literature.
Bruce, C. S. (1994). Research student’s early experiences of the Reference Librarian, 48(5), 5–25.
dissertation literature review. Studies in Higher Education, Mauch, J. E., & Birch, J. W. (2003). Guide to the successful thesis and
19(2), 217–229. dissertation: a handbook for students and faculty (5th ed.).
Bruce, C. S. (1994). Supervising literature reviews. In O. Zuber- New York, NY: Marcel Dekker.
Skerritt & Y. Ryan (Eds.), Quality in postgraduate education. Newell, R., & Burnard, P. (2006). Research for evidence-based practice.
London: Kogan Page. Oxford: Blackwell Publishing.
Bruce, C. S. (1997). From neophyte to expert: counting on reflec- Paniagua, H. (2002). Planning research: methods and ethics. Practice
tion to facilitate complex conceptions of the literature review. Nursing, 13(1), 22–25.
In O. Zuber-Skerritt (Ed.), Frameworks for postgraduate edu- Parahoo, K. (2006). Nursing research – principles, process and issues
cation. Lismore, NSW: Southern Cross University. (2nd ed.). Basingstoke: Palgrave Macmillan.
Burns, N., & Grove, S. K. (2007). Understanding nursing Patrick, L. J., & Munro, S. (2004). The literature review:
research – building an evidence-based practice (4th ed.). demystifying the literature search. The Diabetes Educator,
St. Louis, MO: Elsevier. 30(1), 30–38.
Carnwell, R., & Daly, W. (2001). Strategies for the construction of Polit, D. F., & Beck, C. T. (2004). Nursing research: principles and
a critical review of the literature. Nurse Education in methods (7th ed.). Philadelphia, PA: Lippincott Williams &
Practice, 1, 57–63. Wilkins.
Caspers, J. S. (1998). Hands-on instruction across the miles: using Polit, D., & Beck, C. (2006). Essentials of nursing research: methods,
a web tutorial to teach the literature review research process. appraisal and utilization (6th ed.). Philadelphia, PA: Lippincott
Research Strategies, 16(3), 187–197. Williams and Wilkins.
Cohen, G. (1990). Memory. In I. Roth (Ed.). The Open University’s Rosen Leonard, J., & Behrens, L. (2000). The Allyn and Bacon
Introduction to psychology (Vol 2, pp. 570–620). Milton Keynes: handbook (4th ed.). Boston, MA: Allyn & Bacon.
Lawrence Erlbaum. Timmins, F., & McCabe, C. (2005). How to conduct an effective
Colling, J. (2003). Demystifying the clinical nursing research process: literature review. Nursing Standard, 20(11), 41–47.
the literature review. Urologic Nursing, 23(4), 297–299. Troyka, L. Q. (2002). Simon and Schuster handbook for writers.
Cooper, H. M. (1988). The structure of knowledge synthesis. Upper Saddle River, NJ: Prentice Hall.
Knowledge in Society, 1, 104–126. Younger, P. (2004). Using the internet to conduct a literature
Cooper, H. M. (1989). Integrating research: a guide for literature search. Nursing Standard, 19(6), 45–51.
reviews (2nd ed.). Newbury Park, CA: Sage Publications.
CHAPTER OUTLINE
Research Designs 4
MEANING OF RESEARCH DESIGN • Qualitative Research Designs
SELECTION OF A RESEARCH DESIGN
• Mixed Methods/Triangulation
MULTIPLE CHOICE QUESTIONS
VALIDITY OF A RESEARCH DESIGN

TYPES OF RESEARCH DESIGNS

• Quantitative Research Designs

LEARNING OBJECTIVES

• Define the concept of research designs. • Describe the types of quantitative and qualitative
• Discuss the elements of research design and factors research designs.
influencing the selection of designs.
• Recognize the internal and external validity of research
designs.

KEY TERMS

• Action research 102 • Grounded theory 95 • Prospective cohort design 84

• Case report  80 • Historical cohort design  85 • Qualitative research
• Case series  80 • Historical research 98 design 90
• Case study 102 • Latin square design 75 • Quantitative research
• Case–control studies 82 • Manipulation 69 design 66
• Cohort studies 83 • Meta-analysis 87 • Quasi-experiential research
• Cross-sectional design 81 • Methodological studies 87 design 76
• Crossover design 74 • Observational research • Randomization 71
• Descriptive design 80 design 79 • Randomized block design 74
• Ethnographic research 94 • Operational research 90 • Retrospective cohort
• Evaluation studies 89 • Outcome research 89 design 85
• Experimental research • Phenomenological research 92 • Secondary data analysis 88
design 66 • Pre-experimental research • Survey research design 81
• Factorial design 73 design 78
 Research Designs 63

and comparison between groups. The approach of

MEANING OF RESEARCH DESIGN a research study depends on several factors, but
A research design is the framework or guide used for primarily on the nature of the phenomenon un-
the planning, implementation and analysis of a der study. At this stage of the research study, con-
study. It is a systematic plan of what is to be done, ceptual framework may or may not be incorpo-
how it will be done and how the data will be anal- rated.
ysed. Research design basically provides an outline of
how the research will be carried out and the methods
that will be used. It includes descriptions of the re- SELECTION OF A RESEARCH DESIGN
search approaches, dependent and independent Research designs are plans and procedures for
variables, sampling design and a planned format for research that span decisions from broad assump-
data collection, analysis and presentation. tions to detailed methods of data collection and
It is a methodology for answering research analysis. In order to meet the aims and objectives
questions or the hypotheses that may arise. Differ- of a study, researchers must select the most ap-
ent types of questions or hypotheses demand dif- propriate design. The selection of a research de-
ferent types of research designs; so it is important sign largely depends on the nature of the research
to have a broad preparation and understanding of problem, the resources available (cost, time, ex-
the different types of research designs available. In pertise of the researcher), accessibility of subjects
addition, as a single research design may fall short and research ethics. However, the main factors
of answering all the research questions or hypoth- that affect the selection of research design are as
eses, investigators may use a combination of dif- follows.
ferent research designs.
The research design is the master plan specifying FACTORS AFFECTING SELECTION
the methods and procedures for collecting and ana- OF RESEARCH DESIGN
lysing the needed information in a research study. The selection of research design may be influ-
Research design can be defined as a blueprint enced by several factors of which some of the im-
to conduct a research study, which involves the portant ones are listed as follows.
description of the research approach, study set- l Nature of the research problem: This is the

ting, sampling size, sampling technique, tools and most important factor which helps the re-
methods of data collection and analysis to answer searcher decide on a particular research design.
specific research questions or for testing research On the basis of the nature of the research prob-
hypotheses. lem or phenomenon, researchers decide whether
Research design is a plan of how, when and it should be investigated through an experimen-
where data are to be collected and analysed. tal, quasi-experimental or nonexperimental
Research design is the researcher’s overall plan approach.
for answering the research questions or testing the l Purpose of the study: A study may be con-

research hypotheses. ducted for the purpose of prediction, descrip-

In arrow concept, the research design also tion, exploration or correlation of the research
termed as research approach. It involves the de- variables. Therefore, the purpose of the research
scription of the plan to investigate the phenome- study helps the researcher choose a suitable re-
non under study in a structured (quantitative), search design.
unstructured (qualitative) or a combination of the l Researcher’s knowledge and experience: Selec-

two methods (quantitative–qualitative integrated tion of the research design is largely influenced
approach). Therefore, the approach helps to de- by the researchers’ knowledge and experience
cide the presence or absence of as well as manipu- because they avoid using those designs wherein
lation and control over variables. In addition, it they lack confidence, relevant knowledge or ex-
also helps to identify the presence or absence of perience. Hence the awareness of and expertise in
64 Research Methodology and Biostatistics

existing research designs are important factors l Possible control on extraneous variables: An
pertaining to researchers. efficient design can maximize results, decrease
l Researcher’s interest and motivation: Interest errors and control pre-existing or impaired
and motivation levels help researchers decide conditions that may affect the outcome of the
about a particular research design(s). Motivated study. The maximized efforts of the researcher
researchers always analyse most aspects of a re- should maximize control. Therefore, possible
search design before selecting one or a combina- control over extraneous variables may affect the
tion of designs, while casual and callous research- selection of a research design. For example, a
ers may choose research design(s) that may lead researcher wants to conduct a study through a
to failure. randomized control trial or true experimental
l Research ethics and principles: The incorpora- design; but because of the inability to control se-
tion and application of ethical and legal princi- lected extraneous variables, other similar design
ples in the research design are essential. This has to be opted for, such as quasi-experimental
includes moral obligations, such as respect for research design.
participants and their rights, informed consent l Users of the study findings: A research design
and protection from harm, including any ad- also involves various methods of data collection
verse effects to educational progress, health and and data analysis. Therefore, while choosing a
well-being. research design, the researcher must ensure that
Selection of a research design is significantly the research design is as appropriate for the us-
influenced by the ethics of the research study. ers of the study findings as possible so that
For example, a researcher may be willing to maximum advantage of the results can be ob-
conduct a research study through a certain tained.
experimental approach, but problems of ethi-
cal approval may stop the researcher to do so
and the researcher may have to settle for an- VALIDITY OF A RESEARCH DESIGN
other available possible research design. There are two important criteria for evaluating
l Subjects/participants: The number and avail- the credibility and dependability of the research
ability of study subjects may influence the selec- results: internal and external validity.
tion of research design. If only a few subjects are
involved, an in-depth qualitative research de- INTERNAL VALIDITY
sign study may be a chosen method; but in case It validates whether the independent variables
of a large sample, the researcher may opt for a actually made a difference. Did the intervention
quantitative research design. or treatment lead to the results, or were the results
l Resources: No research can be conducted with- a response to other factors (extraneous variables)?
out resources, such as money, equipment, facili- Campbell and Stanley (1963) used the term
ties and support from colleagues. However, internal validity to refer to the extent to which it
some studies require more amounts of resources is possible to make an inference that the indepen-
as compared to others. Therefore, the selection dent variable is truly influencing the dependent
of a research design may be affected by the variable. In internal validity, the independent
availability of resources for the research study. variable is responsible for the variation in the
l Time: Time is also a major deciding factor for dependent variable. Internal validity demands a
the selection of research design. For example, a tighter control over the study to maximize the
researcher needs more time to conduct longitu- effectiveness of the results.
dinal studies, while cross-sectional studies may Internal validity is helpful in making the infer-
be conducted in a shorter time. Therefore, time ence that the independent variable influences
is also a significant contributing factor in selec- the dependent variable. So internal validity an-
tion of a research design. swers the question, ‘Is the connection between the
 Research Designs 65

independent variable and dependent variable experiment is conducted for a very long period,
clear to make causal inferences?’ Is an indepen- it is difficult to make out whether the effect on
dent variable, in fact, responsible for variation in weight and height is due to maturation or the
the dependent variable? Or, ‘did the treatment nutritional protocol.
cause the effect?’ If the experiment can clearly es- l Testing: It refers to the effect of taking a pretest
tablish that the treatment caused an effect, the of subjects’ performance post-test. The effect of
experiment has internal validity. For example, a taking a pretest may sensitize an individual and
study of the effect of guided imagery on the pain improve the score of the post-test. Individuals
perception among chronic cancer pain. In this generally score higher when they take a test a
study, the guided imagery is an independent vari- second time regardless of the treatment.
able and pain is a dependent variable. It follows l Instrumentation change: Another threat re-
the principle of cause and effect. Experiments will lated to measurement is that of instrumenta-
have high degree of internal validity if they involve tion. This bias reflects changes in measuring
randomization of different groups, which enables instruments or methods of measurements be-
the researcher to rule out competing explanations. tween two points of data collection. Instru-
According to Campbell and Stanley (1966), six ments like thermometer, sphygmomanometer,
major extraneous variables have been identified weighing scale and tape measure should be
that can jeopardize the internal validity. They are checked for their accuracy at regular intervals, and
known as threats to the internal validity and are as the same instruments should be used throughout
follows: the study to minimize the instrument-related
l History: The threat of history occurs when some error of the internal validity.
event beside the experimental treatment occurs l Mortality: Mortality is the loss or dropout of
during the course of study, and this event even study subjects during the course of study. If the
influences dependent variables. For example, you subjects who remain in the study or join later
are conducting a health teaching programme on are not similar to those who dropped out, the
the importance of breast self-examination (BSE), results could be affected. The longer the period
while recently a famous film actress was diag- of the study, the more the chances of subject
nosed to be suffering from breast cancer. It catches mortality. For example, a researcher conducting
media attention. Medical experts are interviewed a longitudinal study wherein a subject who par-
and the importance of BSE is supported. All ma- ticipated in the first round of the data collection
jor television channels and newspapers start re- may not be available for the second or other
porting on the importance of BSE. While you find rounds of data collection.
that the BSE activity has improved, you as a re- l Selection bias: If the subjects are not selected
searcher may not be able to conclude if the change randomly for participation in groups, then there
in behaviour is the result of your teaching pro- is a possibility that the groups that will be com-
gramme or is a result of the diagnosis of the pared may not be equivalent. The effect on the
affliction of the movie actress and the subsequent dependent variable may be due to some other
media coverage. factors. For example, if two different classes are
l Maturation of subjects: When experimental re- used to test the effects of two types of lecture
search is carried on for a long period of time methods or if subjects are selected in a nonran-
over a group of subjects, there may be changes in dom way, the effect on the dependent variables
the subjects in different ways, like in children could be because of other heterogeneous factors
there is increase in height and weight. So matu- rather than the types of lecture methods.
ration is a threat to internal validity. For example,
a researcher is interested in assessing the effect of EXTERNAL VALIDITY
a particular nutritional protocol on the weight It refers to the extent to which the results can be
and height of malnourished children. If this generalized to a large population. External validity
66 Research Methodology and Biostatistics

researches under what conditions and in which methods of doing things. Once treatment is
type of subjects the same results can be expected to more familiar and as the novelty wears off,
be replicated, or whether the same intervention will results might differ.
work in another setting and with different subjects. l People: For example, people of a specific race,
External validity explores the generalization such as whites have high prevalence of coronary
beyond specific experiment(s) to check if the re- artery disease compared to the blacks. There-
sults and findings come out to be the same with fore, a generalization made for whites will not
other settings or with other subject population, be applicable for blacks. Hence, this is a threat to
but with related variables. For example, Newton’s external validity.
law of gravitation is applicable universally, irre- l Place: For example, people living at high alti-
spective of whether it is an American or an Indian tudes have high haemoglobin (Hb) levels be-
apple. The factors that may affect external validity cause at higher altitudes the requirement of
are Hawthorne effect, experimental effect, reactive oxygen is more, due to which there is more
effect of pretest, novelty effect, people, place and production of red blood cells (RBCs). However,
time. the Hb level of people living on the plains is
l Hawthorne effect: Subjects may behave in a lower in comparison, so a generalization for
particular manner because they are aware of people of hilly areas is not applicable for people
them being observed; this is called the living on plains.
Hawthorne effect. Subjects have the knowledge l Time: If a research was carried out on a com-
that they are involved in a research study, thus munity in 1990 and then again in 2000, the re-
affecting the result. sults of these two researches would be different.
l Experimental effect: Experimental effect is a Therefore, older results cannot be generalized
threat to study results when the researcher’s over periods of time as societies and circum-
characteristics, mannerisms or behaviour may stances constantly change.
influence subject behaviour. Examples of re-
searcher’s characteristics or behaviour are facial
expressions, clothes, age, gender and body built. TYPES OF RESEARCH DESIGNS
Thus, the way the researcher dresses up or the There is a lack of consensus on the uniform clas-
researcher’s gender can influence the way in sification of research designs; each author has an
which respondents answer research questions. own style of classifying research designs. The fol-
l Reactive effect of pretest: The reactive effect of lowing discussion presents the most practical and
the pretest occurs when subjects have been sen- simplified classification, which is practised by
sitized to the treatment because of taking a pre- most prestigious research institutions and emi-
test. People might not respond to the treatment nent research scholars. Generally, research designs
in the manner they finally do if they had not are classified into three broad categories and sev-
received the pretest. For example, a researcher eral subtypes.
wants to conduct a study to assess the effect of a l Quantitative research designs (classification

health education programme on the awareness and main features are depicted in Table 4.1)
of HIV/AIDS among people. In this instance, the l Qualitative research designs (classification and

researcher conducts a pretest to collect baseline main features are depicted in Table 4.2)
data before health education. This pretest may l Mixed methods (triangulation)

sensitize the subjects to learn about the HIV/

AIDS irrespective of whether the health educa- EXPERIMENTAL RESEARCH DESIGNS
tion is provided to the subjects or not. Experimentation is the most scientifically sophis-
l Novelty effect: When a treatment is new, sub- ticated research method. It is defined as ‘observa-
jects and the researcher might behave in differ- tion under controlled conditions’. Experimental
ent ways. They may be enthusiastic about new research differs from nonexperimental design in
 Research Designs 67

Table 4.1. Types of quantitative research designs

Broad Categories Types of Research Designs Main Features
I. Experimental 1. True experimental design/RCT Three main characteristics: intervention/
research designs • Post-test-only control design manipulation, randomization and
• Pretest–post-test control group control
design
• Solomon four-group design
• Factorial design
• Randomized block design
• Crossover design
• Latin square design
2. Quasi-experimental design Manipulation of independent variables,
• Nonrandomized control group but absence of either randomization
design or control group
• Time-series design
3. Pre-experimental designs Manipulation of independent variables,
• One-shot case design but limited control over extraneous
• One-group pretest–post-test design variables, no randomization and
control group
II. Observational 4. Descriptive research design Accurate description of characteristics of
research designs • Case report individual, situation or group, and the
• Case series frequency with which a certain phe-
• Cross-sectional design nomenon occurs in a natural setting
• Univariant design without imposing any control or
• Exploratory design manipulation, where generally sub-
• Comparative design jects are contacted not more than
once for data collection.
5. Case–control research design The causes of a disease are investigated
• Nested case–control design after the occurrence of a disease.
6. Cohort research design A longitudinal approach is used to inves-
• Prospective cohort design tigate the occurrence of a disease in a
• Historical cohort design cohort with existing presumed causes
• Ambispective cohort design compared with homogenous controls.
III. Other miscella- 7. Methodological studies Research conducted to develop, test and
neous research evaluate research instruments and
designs methods
8. Meta-analysis Quantitatively combining and integrat-
ing the findings of multiple research
studies on a particular topic
9. Secondary data analysis A research design in which the data col-
lected in one research are re-analysed
by another researcher, usually to test
new hypotheses
10. Outcome research Outcome research involves the evalua-
tion of care practices and systems in
place. It is used in health care to
develop evidence-based practice
and improve health care services
Continued
68 Research Methodology and Biostatistics

Table 4.1. Types of quantitative research designs—cont’d

Broad Categories Types of Research Designs Main Features
11. Evaluation studies It is a research design that involves the
judgement about success of pro-
grammes, practices, procedures or
policies
12. Operational research Operational research involves the study
of complex human organizations and
services to develop new knowledge
about institutions, programmes, use
of facilities and personnel in order to
improve working efficiency of an
organization

Table 4.2. ​Types of qualitative research designs

Types of Research Designs Main Features
Phenomenological research Phenomenological research examines human experiences through the
descriptions provided by the people involved
Ethnographic research Ethnographic research involves the information collected from certain
cultural groups, by living with people of those groups and from their key
informants, who are believed to be most knowledgeable about the
selected culture
Grounded theory From a collection of data gathered by a participant observer, theory is devel-
oped accordingly
Case studies Research on a phenomenon by studying in-depth about any case example.
It could be any event, a group, a distinct person or an institution
Historical research Orderly gathering of data and their objective evaluation belonging to past
occurrences to examine hypotheses related to causes, effects or trends of
these cases which might help in explaining current events and expecting
future events
Action research Action research seeks action to improve practices and study the effect of the
action that was taken

one important aspect. The researcher using an independent variable; but a true experiment also
experimental design is an active agent rather than consists of the principles of randomization and
a passive observer. control. The application of control is difficult
Experimental research designs are concerned when studies are conducted in natural settings
with examination of the effect of the indepen- on human subjects. Therefore, in many of the
dent variable on the dependent variable, where allied health care disciplines, experimental re-
the independent variable is manipulated through search design is not feasible to use, and quasi-
treatment or intervention(s), and the effect of experimental, pre-experimental or nonexperi-
these interventions is observed on the dependent mental studies are more frequently conducted.
variable. All experimental researches have a According to Riley, experimental research de-
common characteristic, i.e. manipulation of the sign is a powerful design for testing hypotheses of
 Research Designs 69

causal relationship among variables. Ideally, in the In another example, suppose we hypothesize
experimental design, the investigator throws in a that gentle massage is effective as a pain relief mea-
sharp relief of explanatory variables in which he sure for elderly nursing home residents. Providing
or she is interested, controlling and manipulating gentle massage to elderly in the experimental
the independent variable, observing its effect on group and withholding for others in the control
the dependent variable and minimizing the effect group is considered manipulation of independent
of extraneous variables, which might confound variables, where the effect of this manipulation is
his or her results. Experimental research design is observed on the pain level in both the groups.
further classified into true experimental designs,
quasi-experimental designs and pre-example re- Independent Gentle massage (as manipulation,
search design. variable which can be given to some
patients and withholding it from
TRUE EXPERIMENTAL DESIGNS others in the control group)
True experimental research designs, also known as Dependent Pain level
randomized control trials, are those where re- variable
searchers have complete control over the extrane-
ous variables and can predict confidently that the
observed effect on the dependable variable is only Control
due to the manipulation of the independent vari- Control is another essential element of a true ex-
able. A detailed discussion of randomized control perimental design. Control refers to the use of a
trials is presented in Chapter 5. control group and controlling the effects of extra-
neous variables on the dependent variable in
Essential Characteristics which the researcher is interested.
A true experimental research design must essen- The subjects in the control and experimental
tially consist of the following three characteristics: groups are similar in number and characteristics,
manipulation, control and randomization. but the subjects in the control group receive no
experimental treatment or any intervention at all.
Intervention or Manipulation of Independent The experimental group receives the planned treat-
Variable ment or intervention, and a comparison is made
Manipulation refers to a conscious control of the with the control group to observe the effect of this
independent variable by the researcher through treatment or intervention. However, generally in
treatment or intervention(s) to observe its effect health care and other allied health care research, it
on the dependent variable. In other words, it is a is not ethically feasible to keep a control group de-
conscious act by the researcher, where he/she var- prived of interventions; however, existing conven-
ies the independent variable and observes the ef- tional methods of interventions may be compared
fect that the manipulation has on the dependent with experimental interventions. The control group
variable of interest. could be of the following three types:
For example, a researcher is conducting a study l Negative control: In this type of control group,

on the efficacy of chlorhexidine mouthwash on the subject receives neither a placebo nor any
the prevention of ventilator-associated pneumo- other type of treatment or intervention.
nia (VAP) among patients admitted in ICUs. In l Clear control: The subjects in this type of con-

this example, chlorhexidine mouthwash is the in- trol group receive a placebo.
dependent variable, which is manipulated by the l Positive control: The subjects in the control

researcher and is used as an intervention for the group receive other treatment or experimental
experimental group, while the control group is intervention.
kept deprived of it to observe its effect on the in- The control of the effects of extraneous vari-
cidence of VAP. ables on the dependent variables can be
70 Research Methodology and Biostatistics

ensured by using several conventional mea- l Counterbalancing: The use of ‘counterbalanc-

sures, but none of them is very convenient to ing’ is another way to exert active control over
use. Some of these measures can, however, extraneous variables. Counterbalancing is used
control extraneous variables, which include in studies in which the researcher is concerned
the following: with the orders in which treatments or inter-
l Matching: A conscious ‘matching’ of the subject ventions administered influence the study re-
characteristics in both the groups. It is a weak sults. With counterbalancing, all the subjects
but common method of control over extraneous receive all the treatments, but the order of ad-
variables. In matching, the researcher identifies ministration of treatment is varied. For example,
one or more extraneous variables to be con- counterbalancing can be used to study the ef-
trolled, which are supposed to have an effect on fects of several different relaxation techniques
the dependent variable. As soon as subjects are by varying the order in which they are adminis-
recruited for one of the treatment groups, the tered to subjects.
researcher tries to find subjects for the other l Homogeneity by statistical test: The researcher
group, similar to the subjects of the first group may also exert control by conducting a statistical
based on the specific matching variables. For test to find no significant difference in the basic
example, if age and gender are the matching vari- characteristics of subjects that may affect the
ables of interest in a two-group study, and a dependent variables among different groups un-
65-year-old woman is recruited for the first der study. For example, to ensure the homogene-
group, the researcher would try to find another ity of the demographical characteristics among
65-year-old woman to be included in the second two groups under study, a chi-square test may
group. However, matching is not always feasible be applied on the frequency distribution of se-
because of several reasons, such as not being able lected characteristics in two groups; there should
to find subjects with similar characteristics in not be a significant statistical difference in
the population under study may affect depen- demographical characteristics of the two groups
dent variable(s); the need to match more than to establish adequate control, as depicted in
three or four characteristics simultaneously may Table 4.3. However, finding such statistical
make it even more difficult. difference is not easy because a researcher needs

Table 4.3. ​Statistical comparison of sociodemographic characteristics in experimental and control groups
Groups
Sociodemographic Variables Experimental (n 5 600) Control (n 5 600) x2
Age (years)
18–30 287 (47.8) 299 (49.8) 0.675
31–45 188 (31.3) 174 (29.0) df 5 3
46–60 86 (14.3) 87 (14.5) p 5 0.9NS
.60 39 (06.5) 40 (06.7)
Gender 0.483
Male 329 (54.8) 317 (52.8) df 5 1
Female 271 (45.2) 283 (47.2) p 5 0.483NS
Marital status
Unmarried 86 (14.3) 68 (11.3) 6.706
Married 495 (82.5) 507 (84.5) df 5 3
Divorced/separated 08 (01.3) 04 (00.7) p 5 0.82NS
Widow/widower 11 (01.8) 21 (03.5)
Notes: NS 5 Experimental and control groups are not statistically different in age, gender and marital status.
 Research Designs 71

to have prior knowledge about the characteris- Table 4.4. ​Sample of a random table
tics of subjects which may affect the dependent
82 47 42 55 93 48 54 53 52 47 18 61 91 36 74
variables and must collect relevant information
for statistical comparison before starting the ex- 53 35 24 42 76 75 12 21 17 24 74 62 77 37 07
periment. In the examples in Table 4.3, experi- 82 64 12 28 20 92 90 41 3141 32 39 21 97 63
mental and control groups are not statistically 13 57 4172 00 69 90 26 37 42 78 46 42 25 01
different in age, gender and marital status, and 29 59 38 86 27 94 97 21 15 98 62 09 53 67 87
therefore, adequate control can be ensured
70 0141 50 21 41 29 06 73 12 71 85 71 59 57
among these groups.
37 23 93 32 95 05 87 00 11 19 92 78 42 63 40
Randomization 18 63 73 75 09 82 44 49 90 05 04 92 17 37 01
Randomization means that every subject has an 05 32 78 21 62 20 24 78 17 59 45 19 72 53 32
equal chance of being assigned to the experimen- 95 09 66 79 46 48 46 08 55 58 15 19 11 87 82
tal or control group. This is also known as random
assignment of subjects, which involves the place-
ment of study subjects on a random basis. Through groups. Computer-assisted random sequences
random assignment of subject under the experi- also may be used for the random assignment of
mental or control group, chances of systemic bias the subjects.
are eliminated. l Computer-assisted randomization: The most

Randomization is used in true experimental re- modern, accurate and easy method of random-
search designs to minimize the threat of internal ization is computer-assisted randomization.
validity of the study and to eliminate the effect of The random number may be generated using
extraneous variables on dependent variables. several free websites, such as Research Random-
Through randomization, on average, the character- izer (http://www.randomizer.org/form.htm).
istics of the subjects in experimental and control Using the random assignments, in a study of
groups are similar; thus influence of extraneous 16 subjects and 2 groups, the first person to enter
variables on the dependent variable is eliminated the study would be assigned to group I, the second
by dispersing the variability of the subject charac- would be assigned to group II and so on. This
teristics equally in both the groups. method ensures that the sample sizes for both the
groups being compared are equal (Table 4.5).
Methods of randomization
l Random assignment of subjects may be done Symbolic Presentations
with a simple flip of a coin for each subject; if the To facilitate the understanding of research de-
coin lands on its ‘head’, subjects are assigned to signs, some symbols are used to present the differ-
the first group and with ‘tail,’ subjects are as- ent stages of experimentation in the literature.
signed to the second group. Some of the symbols used are as follows:
l Another possible method is to write the names of R 5 Random assignment of subjects to groups
the subjects on slips of paper, put the slips into a O 5 Observation or measurement of the depen-
bowl and then draw lots. The first designated dent variable
numbers of subjects are placed in one group and X 5 Experimental treatment or intervention
the rest are assigned under another group. The subscript number appearing after X (X1,
l Thirdly, a random table (Table 4.4) may be used X2, X3, . . . ) indicates the treatments, such as first
to facilitate the randomization process. In this treatment, second treatment, third treatment
method, blindfolded subjects choose a number and so on. The subscript number appearing after
from a table of numbers horizontally (row) or an O (O1, O2, O3, . . . ) indicates the observation
vertically (columns) till a requisite number is or measurement, such as first observation, sec-
reached for both experimental and control ond observation, third observation and so on.
72 Research Methodology and Biostatistics

Table 4.5. ​Example of randomization procedure Post-Test-Only Control Design

Name Group This design is composed of two randomly
of Subject Number Assignment assigned groups, i.e. experimental and control,
Kiran 1 I
neither of which is pretested before the imple-
mentation of treatment on the experimental
Amandeep 2 II
group. In addition, while treatment is imple-
Tanveer 3 I mented on the experimental group only, post-test
Loverpreet 4 II observation is carried out on both the groups to
Reena 5 II assess the effect of manipulation (Fig. 4.1). This
Tara 6 I design can be helpful in situations where it is not
possible to pretest the subjects, for example, to
Shria 7 II
study the effect of an educational intervention
Amarpreet 8 I related to urinary incontinence on the subsequent
Shidhika 9 II help-seeking behaviour of older adults.
Rajbeer 10 I
Gurpreet 11 I Pretest–Post-Test-Only Design
Aarvinder 12 II
In this research design, subjects are randomly as-
signed to either the experimental or the control
Nish 13 II
group. Effect of the dependent variable on both the
Pardeep 14 I groups is seen before the treatment (pretest). Later,
Tamanpreet 15 I the treatment is carried out on the experimental
Mandeep 16 II group only, and after-treatment observation of the
dependent variable is made on both the groups to
examine the effect of the manipulation of the inde-
Few Types of True Experimental Designs pendent variable on the dependent variable (Fig. 4.2).
Six types of true experimental designs are discussed For example, such a design could be used for ‘an
in this chapter. There are a few other types of true experimental study to assess the effectiveness of
experimental designs, but they are rarely used in cognitive behavioural therapy interventions for
research conducted in health care settings. patients with breast cancer’.

True experimental designs

Solomon Four-Group Design
There are two experimental groups (experimental
group 1 and experimental group 2) and two con-
trol groups (control group 1 and control group 2).
Initially, the investigator randomly assigns sub-
Post-test-only Crossover jects to the four groups. Out of the four groups,
only experimental group 1 and control group 1
receive the pretest, followed by the treatment to the
experimental group 1 and experimental group 2.
Finally, all the four groups receive post-test, where
Randomized
Pretest–post- the effects on the dependent variables of the study
block
test-only
are observed and comparison is made of the four
groups to assess the effect of the independent vari-
able (experimental treatment) on the dependent
Solomon Factorial
4 groups
variable. In this, experimental group 2 was ob-
served at one occasion, and that score should be
similar to averaged scores of those in experimental
 Research Designs 73

Exp. group Treatment Post-test

Random
assignment

Control group Post-test

Fig. 4.1. ​Schematic diagram of post-test-only control design.

Exp. group Pretest Treatment Post-test

Random
assignment

Control group Pretest Post-test

Fig. 4.2. ​Schematic diagram of pretest–post-test-only design.

and control groups. To estimate the amount of treatment. Unfortunately, this research design re-
change in experimental and control group 2, the quires a large sample and statistical analysis, and,
average test scores of experimental and control therefore, it is not commonly used by health care
groups 1 are used as baseline (Fig. 4.3). researchers.
The Solomon four-group design is believed to
be the most prestigious experimental research de- Factorial Design
sign because it minimizes the threat to internal and In factorial design, the researcher manipulates two
external validity. This design controls not only all or more independent variables simultaneously to
the threats to internal validity but also the reactive observe their effects on the dependent variables.
effects of the pretests. Any differences between the This design is useful when there are more than
experimental and the control groups can be two independent variables called factors to be
more confidently attributed to the experimental tested. For example, a researcher wants to observe

Exp. group 1 Pretest Treatment Post-test

Control group 1 Pretest Post-test

Random
assignment

Exp. group 2 Treatment Post-test

Control group 2 Post-test

Fig. 4.3. ​Schematic diagram of Solomon four-group design.

74 Research Methodology and Biostatistics

Protocols of the mouth care

the effects of three different antihypertensive drugs
Frequency of
mouth care
on patients with hypertension. In this example, to
Chlorhexidine (α 1) Saline (α 2) ensure the homogeneity among the subjects under
4 hourly (β1) α1...β1 α2 ...β1 treatment, the researcher randomly places the sub-
jects in homogeneous groups (blocks) like patients
6 hourly (β2) α1...β2 α2 ...β2 with primary hypertension, diabetic patients with
8 hourly (β3) α1...β3 α2 ...β3 hypertension and renal patients with hypertension
(Fig. 4.5).
Fig. 4.4. ​Schematic diagram of factorial design. This design looks similar to a factorial design in
structure; but out of two factors, one factor is not
experimentally manipulated; like in the given ex-
the effects of two different protocols of mouth ample, there are two factors: type of antihyperten-
care on prevention of VAP when performed at dif- sive drugs and type of patients with hypertension,
ferent frequencies in a day (Fig. 4.4). where only the type of drug is manipulated and
This design also facilitates the testing of sev- types of patients with hypertension are simply
eral hypotheses at a single time. Typical factorial grouped in different blocks with similar charac-
design incorporates 2 3 2 or 2 3 3 factorial; but teristics to ensure homogeneity.
it can be in any combination. The first number
(a) refers to the independent variables or the Crossover Design
types of experimental treatment, and the second In this design, subjects are exposed to more than
number (b) refers to the level or frequency of the one treatment, where they are randomly assigned
treatment. to different orders of treatment (Fig. 4.6). It is also
known as ‘repeat measures design’. This design is
Randomized Block Design more efficient in establishing the highest possible
Control of inherent differences between experi- similarity among subjects exposed to different
mental subjects and differences in experimental conditions, where the groups compared obviously
conditions is one of the difficult problems faced by have equal distribution of characteristics. Although
researchers in biological sciences. When there are a crossover design is considered an extremely pow-
large number of experimental comparison groups, erful research design, sometimes it is not effective
the randomized block design is used to bring ho- because when subjects are exposed to two different
mogeneity among selected different groups. This is conditions, their responses to the second condition
a simple method to reduce variability among the may be influenced by their experience in the first
treatment groups by a more homogeneous combi- condition.
nation of the subjects through randomized block For example, when we compare the effectiveness
design. For example, a researcher wants to examine of the chlorhexidine and saline mouth care protocol,

Blocks
Type of
antihypertensive Patients with primary Diabetic patients with Renal patients
drugs hypertension hypertension with hypertension
(I) (II) (II)

A A, I A, II A, III

B B, I B, II B, III

C C, I C, II C, III

Fig. 4.5. ​Schematic diagram of randomized block design.

 Research Designs 75

Groups Protocols of the mouth care design. The advantages of Latin square design are
that it helps in controlling more variation than
Group I Chlorhexidine (α 1) Saline (α 2) completely randomized (CR) or randomized com-
plete block (RCB) designs; its two-way stratifica-
Group II Saline (α 2) Chlorhexidine (α 1)
tion results in a smaller mean square for error, and
makes the data analysis simple even with instances
Fig. 4.6. ​Schematic diagram of crossover design.
of missing plots. However, it does have some disad-
vantages, such as in this design, the number of
first we administer the chlorhexidine mouth care treatments is limited to the number of replicates,
protocol on group I and the saline mouth care pro- which seldom exceeds 10; and if it has less than
tocol on the subjects of group II. Later, the treatment 5 treatments, the df for controlling random varia-
is swapped, where group I receives the saline mouth tion is relatively large; the df for error is small.
care and group II receives chlorhexidine. In such An example of a Latin square design with four
studies, subjects serve as their own control. treatments (ABCD) is presented in Fig. 4.7, where
the numbers in the upper left corner are the plot
Latin Square Design numbers for each row and the letters are the
Latin square designs differ from randomized com- treatments followed.
plete block designs in that the experimental units
are grouped in blocks in two different ways, i.e. by Advantages and Disadvantages
rows and columns. Therefore, two different sources of True Experimental Design
of variation can be isolated. For example, an ex- Advantages
periment can be installed as a Latin square with the l Experimental research designs are considered
rows isolating the type of drugs used in treatment the most powerful designs to establish the causal
of irritable bowel syndrome (IBS) and columns relationship between independent and depen-
isolating the dose of a drug that also has a doses of dent variables.
drugs running across the different drugs. A restric- l For the purpose of research as explanation,
tion in the assignment of treatments in a Latin causal relationship may be established among
square is that each treatment can occur once and variables by experimentation, especially in stud-
only once in each row and column so that each row ies involving physical objects, where the vari-
and each column is like a complete block. A re- ables are more easily controlled than that in
quirement of the Latin square is that the number of human studies.
treatments, rows and number of replications and l In these studies, the controlled environment in
columns must be equal; therefore, the total number which the study is conducted can yield a greater
of experimental units must be a perfect square. For degree of purity in observation.
example, if there are four treatments, there must be
four replicates, or four rows and four columns. This Columns
is a 4 3 4 Latin square that gives a total of 16 ex- 1 2 3 4
perimental units. Because of this restriction, Latin 1 2 3 4
square experiments can become large and unman- 1 A B C D
ageable very readily. Also, if the number of treat-
1 2 3 4
ments is too small, there are too few degrees of Rows 2 B C D A
freedom (df) for error so that the most common 1 2 3 4
squares are in the range of 5 3 5 – 8 3 8. Since two 3 C D A B
sources of variation are isolated in rows and col- 1 2 3 4
umns, the mean square for error will be smaller 4 D A B C
than for the same data analysed as either com-
pletely randomized or randomized complete block Fig. 4.7. ​Latin square design.
76 Research Methodology and Biostatistics

l Conditions not found in a natural setting can little opportunity to study the effects as they occur.
be created in an experimental setting, where Only by a difficult and costly procedure, can these
the independent variable is manipulated by discharged patients be followed in their homes to
the investigator. observe the experimental results. However, it is not
l In the experimental approach, we can often cre- always possible because of several constraints.
ate conditions in a short period of time that l Another disadvantage of the experimental re-

may take years to occur naturally. For example, search design is that it is very difficult to obtain
in genetic studies, we can breed strains in a very cooperation from the study participants be-
small time, which would take a long time to cause it may involve medical or surgical treat-
occur naturally. ments or interventions, which may make the
l When an experiment is conducted in a labora- prospective subjects reluctant to participate in
tory, experimental unit or other specialized re- the research study.
search setting, it is removed from the pressure, l Because the sample size for experiments involv-

and problems of real-life situations and the re- ing human beings is often kept small, there is a
searcher can pursue his or her studies in a more question as to how representative the findings of
leisurely, careful and concentrated way. such studies can be. If the target population is
diverse, if the extraneous variables are numerous
Disadvantages or if these extraneous variables can exert spuri-
l Most of the times, the results of experimental ous influence on dependent variables when
research designs cannot be replicated in studies uncontrolled, the question arises as to whether a
conducted on human beings due to ethical small number of study subjects can provide
problems. meaningful results for a large population, even
l For certain research problems, because of the if they have been carefully selected and the
danger to physical and psychological health of conditions are kept intact during the study.
the human subjects, it is not possible to conduct l Although theoretically experimental designs can

experiments on human beings. yield insight into the causes of certain phenomena,
l Many of the human variables neither have valid in research involving human beings, these phe-
measurable criteria nor instruments to mea- nomena are usually so complex and driven from
sure them. For example, patient welfare or level such a plurality of causes that the simple experi-
of wellness cannot be measured on any scale or mental approach does not help. A more complex,
by any instrument. In these situations, if a re- multivariate and nonlinear approach is required,
fined experimental design is used, there may be which may only be possible with great difficulties
a mismatch of research design and the variable- through the experimental approach.
measuring instruments. The limitations inherent in the experimental
l In experimental studies conducted in natural set- approach are greatest when the experimental sub-
tings like hospitals or community, it is not pos- jects are human beings. Since most conceivable
sible to impose control over extraneous variables. patient-care studies involve humans, the use of
l Experiments are often impractical when the effect the experimental approach in such research is
of the independent variable may require a lengthy limited. However, the quasi-experimental or pre-
period of time before it can emerge as a response experimental designs could be more feasible on
on the criterion measures. This situation exists for human subjects.
many of the variables in allied health care disci-
plines. One of the main drawbacks in conducting
QUASI-EXPERIMENTAL RESEARCH
experiments on the effects of allied health care
interventions care on acutely ill, hospitalized pa- DESIGN
tients is that the patients are discharged from Quasi-experimental research design involves the
hospitals in such short periods of time that there is manipulation of the independent variable to
 Research Designs 77

Exp. group Pretest Treatment Post-test

Control group Pretest Post-test

Fig. 4.8. ​Schematic diagram of nonrandomized control group design.

observe the effect on the dependent variable, but no random assignment of subjects in experimental
it lacks at least one of the two characteristics of the and control groups. In this design, experimental
true experimental design: randomization or a and control groups are selected without random-
control group. ization, and dependent variables are observed in
In other words, quasi-experimental designs experimental as well as control groups before the
have an element of manipulation but lack at least intervention. Later, the experimental group
one of the other two properties that characterize receives treatment and after that, post-test obser-
true experiments: randomization or a control vation of dependent variables is carried out for
group. Quasi-experimental designs are generally both the groups to assess the effect of treatment on
used to establish the causality (effect of an the experimental group (Fig. 4.8).
independent variable on a dependent variable) in For example, this method was used to study ‘the
situations where researchers are not able to consequences of combined care on the nature of
randomly assign the subjects to groups or for work in nursing homes: a quasi-experiment’. The
various reasons no control group is available for aim of this study was to check the application of
an experimental study. combined care in the nursing homes and its con-
sequence on the kind of work of the caregivers.
MAIN CHARACTERISTICS A nonequivalent pretest–post-test control group
l Manipulation of independent variables to ob- design was used in this study. Data were collected
serve the effects on dependent variables through a questionnaire. The result showed that
l Lack of at least one of the two other essential intervention seemed to be outstanding on the so-
characteristics of the true experiment, i.e. ran- matic wards. The caregivers in these wards were
dom assignment of subject or a control group very successful in developing a home-like envi-
l Quasi-independent variables are used instead of ronment for their residents. As regards the conse-
true independent variables where the indepen- quences of the intervention on the traits of work
dent variable cannot be manipulated in com- factors, the result included an increase of social
pletely controlled situations. support by supervision. The intervention on the
psychosomatic wards was unsuccessful, although
TYPES OF QUASI-EXPERIMENTAL DESIGNS the introduction of integrated care on the somatic
Quasi-experimental designs which are also known wards was fairly successful.
as nonrandomized control trials may be of several
types; but two most frequently used designs are Time-Series Design (Fig. 4.9)
discussed in this chapter. l This design is useful when the experimenter
wants to measure the effects of a treatment over
Nonrandomized Control Group Design a long period of time.
It is also known as the ‘nonequivalent control group l The researcher would continue to administer
design’. This design is identical to the pretest– the treatment and measure the effects a number
post-test control group design, except that there is of times during the course of the experiment.
78 Research Methodology and Biostatistics

Exp. group O1 O2 O3 Treatment O1 O2 O3

Fig. 4.9. ​Schematic diagram of time-series design.

l Generally, it is a single-subject research in which l The absence of a control group or a lack of con-
the researcher carries out an experiment on an trol over the research setting makes the results
individual or on a small number of individuals of this design less reliable and weak for the
by alternating between administering and then establishment of causal relationship between
withdrawing the treatment to determine the independent and dependent variables.
effectiveness of the intervention.
PRE-EXPERIMENTAL RESEARCH
Examples DESIGN
• A researcher may evaluate pain levels of a group This research design is considered very weak
of patients having lower back pain. After assess- because the researcher has very little control over
ing for pain as long as 3 weeks, subjects are edu- the experiment.
cated about special exercises to lessen that pain.
In the next 3 weeks, pain levels would again be TYPES OF PRE-EXPERIMENTAL RESEARCH
evaluated. DESIGN
• Evaluating a child’s school work on a weekly basis,
bringing out a new teaching method and then This chapter discusses two types of pre-experimental
again evaluating it on a weekly basis. research designs: one-shot case design and one-
group pretest–post-test design.
One-shot case design: In this research design, a
single experimental group is exposed to a treat-
ADVANTAGES AND DISADVANTAGES
ment and observations are made after the imple-
OF QUASI-EXPERIMENTAL DESIGN
mentation of that treatment. There is no random
Advantages assignment of subjects to the experimental group
l Quasi-experimental designs are more frequently and no control group at all (Fig. 4.10).
used because they are more practical and feasi- One-group pretest–post-test design: It is the
ble to conduct research studies in health care simplest type of pre-experimental design where
disciplines, wherein the absence of a large sam- only the experimental group is selected as the study
ple size, randomization and/or availability of subject. A pretest observation of the dependent
control groups are not always possible. variables is made before implementation of the
l This design is more suitable for a real-world treatment to the selected group, the treatment is
natural setting than true experimental research administered and finally a post-test observation
designs. of dependent variables is carried out to assess the
l It allows researchers to evaluate the impact of effect of treatment on the group (Fig. 4.11). Some
quasi-independent variables under naturally researchers also argue that this design is a
occurring conditions. subtype of quasi-experimental research design.
l It may be able to establish causal relationship, However, in the absence of both randomization and
wherein some of the hypotheses are practically control group, this design ethically cannot be placed
answered through this design only.

Disadvantages Exp. group Treatment Post-test

l There is no or very little control over extraneous
variables influencing the dependent variables. Fig. 4.10. ​Schematic diagram of one-shot case design.
 Research Designs 79

Exp. group Pretest Treatment Post-test

Fig. 4.11. ​Schematic diagram of one-group pretest–post-test design.

under the classification of quasi-experimental l It has a higher threat to internal validity of re-
research design. search and may have a selection bias, which can
This design measures the effect on the experi- be a very serious threat in using this particular
mental group on the basis of their state before design.
the beginning of the experiment (pretest) and
the difference achieved at the end of the experi-
ment (post-test). There is no control group in OBSERVATIONAL RESEARCH DESIGN
this design. The observational research design is also known
For example, in ‘a study on the effect of inter- as epidemiological or noninterventional or non-
ventions on the stress-coping resources of associ- experimental research design. It is one of the
ate degree health sciences students’, a pretest and broad categories of research designs in which the
post-test design was used to examine three groups researcher observes the phenomena as they occur
in health sciences programmes. Each group com- naturally and no external variables are introduced.
pleted the instrument – coping resources inven- It is a research design in which neither the vari-
tory for stress (CRIS) – at the beginning and at the ables are deliberately manipulated nor the setting
end of the first year of the particular health sci- is controlled.
ence course. The modified curriculum group re- In observational research, researchers collect
ceived interventions for the development of stress- data without making changes or introducing
coping resources in their initial course, the second treatments. Data obtained are analysed and the
group experienced self-directed interventions and results may lead to the formation of a hypothesis
the third group received no intervention. This that can then be tested experimentally. Within a
study concluded that curriculum intervention for quantitative framework, the observations are rep-
first-year degree nursing students resulted in a resented by numbers that can be statistically anal-
measured increase of stress monitoring and ysed. Areas of research interest that have not been
tension-control coping resources. examined in depth may be best understood by
conducting a descriptive study. Data in observa-
ADVANTAGES AND DISADVANTAGES OF tional research are generally collected through the
PRE-EXPERIMENTAL DESIGN
use of questionnaires, interviews, observations,
Advantages literature reviews and critical-incident technique.
l It is very simple and convenient to conduct
these studies in natural settings, especially in NEED FOR OBSERVATIONAL RESEARCH DESIGN
allied health science disciplines. Observational research design is frequently used
l It is the most suitable design for beginners in by biomedical researchers. Some of the study situ-
the field of experimental research. ations where only observational designs can be
used to conduct a study are as follows:
Disadvantages l The studies in which the independent variables

l It is considered a very weak experimental design cannot be manipulated

to establish causal relationships between inde- l The studies in which it is unethical to manipu-

pendent and dependent variables because it late the independent variable, i.e. manipulation
controls no threat to internal validity. may cause physical or psychological harm to
l It has a very little control over the research. subjects
80 Research Methodology and Biostatistics

l The studies or research situations where it is not l Descriptive design may be used to develop theo-
practically possible to conduct experiments ries, identify problems with current practices,
l Descriptive-type studies that do not require any justify current practices, make judgments or
experimental approaches determine other practices in similar situations.
l In descriptive studies, bias is prevented through
TYPES OF OBSERVATIONAL RESEARCH operational definitions of variables, large sam-
DESIGNS ple size, random sampling techniques, valid and
There is a lack of uniformity in classification of reliable research tools, and formal data collec-
observational research designs. However, the follow- tion methods.
ing classifications of the observational research l Descriptive designs include identification of
designs are the most suitable to improve the a phenomenon of interest, identifying the
understanding about observational research designs. variables within the phenomenon, develop-
l Descriptive research design ing operational definitions of the variables
 Case report and describing the variables. The description
 Case series of variables leads to an interpretation of the
 Cross-sectional design theoretical meaning of the findings and the
- Univariant design development of hypotheses.
- Exploratory design
- Comparative design TYPES OF DESCRIPTIVE RESEARCH DESIGN
l Case–control research design Five basic descriptive research designs are dis-
 Nested case–control design cussed in this chapter: case report, case series,
l Cohort research design univariant descriptive design, exploratory design
 Prospective cohort design and comparative design.
 Historical cohort or retrospective cohort design

 Ambispective cohort design Case Report

It is descriptive observational study of a single pa-
tient or case with rare, specific and unique features
DESCRIPTIVE RESEARCH DESIGNS and presentation, which is not a routine phenome-
The purpose of descriptive studies is to observe, non prevailing in practices or documented in the
describe and document aspects of a situation as it literature, for example, acute pancreatitis following
naturally occurs, and sometimes to serve as a medical termination of pregnancy: a case report.
starting point for hypothesis generation or theory Acute pancreatitis is not a common complication of
development. pregnancy or abortion; thus a researcher may pre-
pare a cumulative anecdote of the complete case
MAIN FEATURES with clinical presentation of the case, diagnostics test
l Descriptive designs are used to observe, docu- carried out, specific findings of the tests, treatment/
ment and describe a phenomenon occurring in interventions implemented and patient’s response to
its natural setting without any manipulation or the treatment/intervention. Then, the researcher will
control. try to infer the possible cause of acute pancreatitis in
l Descriptive studies are designed to gain more such a rare case comparing it with established causes.
information about characteristics within a par- Such researches generate new hypothesis which can
ticular field of inquiry. be later tested through analytic observational
l Descriptive studies provide an impression of a researches such as case–control or cohort studies.
situation as it occurs in a natural setting.
l Descriptive studies do not involve the manipu- Case Series
lation of variables and variables are studied as It is a descriptive observational study in which a
they exist in the real world. group of patients with similar clinical presentation,
 Research Designs 81

diagnosis or who are exposed to similar Subjects with

intervention(s) are observed over a period of time, intended outcome
for example, the medical and ethical challenges of
fertility preservation in teenage girls: a case series of
sickle cell anaemia patients prior to bone marrow
Subjects
transplant. In this case series, gonadotrophin ovar- selected for
ian stimulation was given to eight consecutive cases, the study
mature oocyte retrieval and oocyte cryopreserva-
tion was done for each patient with the objective to
describe the feasibility of performing ovarian stim- Subjects
without
ulation, and the ethical challenges faced, in teenage intended
girls, aged 14–18 years, prior to undergoing bone outcome
marrow transplant for sickle cell anaemia.
Case series is the most commonly conducted
research available in literature but they provide very Fig. 4.12. ​Conceptual framework of cross-sectional
weak empirical evidences because of lack of true study.
comparability of cases and lack of control on the
experiment; thus, no causal inferences should be studies are relatively inexpensive and efficient for
made from case series research about the efficacy of including potentially large number of subjects.
a particular treatment/intervention. Cases series are They are also ethical in terms of typically involving
best sources to generate new hypothesis which later minimal risk to subjects. However, they are rela-
can be tested through analytic researches including tively inefficient for studying rare diseases in that
randomized controlled trials. an extremely large sample would be required to
capture enough cases to support analysis.
Cross-Sectional Design In a cross-sectional study design, the researcher
It is a type of descriptive study of the relationship may carry out the survey about health behaviour in
between disease and other factors at one point in a certain population, can also perform an overview
time in a defined population. Cross-sectional studies of patients with a certain diagnosis (age, sex, clinical
are also known as prevalence studies or epidemio- data) or may conduct prevalence studies – to esti-
logical studies in which the researcher collects data at mate the percentage of people with a given disease in
a particular point of time (one period of data collec- the population. For example, a researcher is inter-
tion). Cross-sectional studies analyse data collected ested in assessing the awareness on swine flu among
on a group of subjects at one time rather than over a people of a selected geographical area. Furthermore,
period of time. Cross-sectional studies are designed the cross-sectional studies could be of the following
to determine ‘what is happening?’ right now. For types: univariant cross-sectional design, exploratory
example, in ‘a study on level anxiety experienced by design and comparative design
medical students on first day of their ward posting,’
the researcher interacts only once to collect awareness- Univariant Cross-Sectional Design
related data from respondents. These studies are easier Univariant descriptive designs are undertaken to
and more convenient to carry out. describe the frequency of occurrence of a phe-
The defining characteristics of cross-sectional nomenon. This design does not necessarily focus
studies are the exposure and outcome measured at on the study of a single variable; there may be one
the same point of time (Fig. 4.12). This design is or more variables involved in the study.
equipped to identify factors that are associated For example, a researcher is interested in assessing
with the health phenomenon; but causality cannot the experiences of patients suffering from rheuma-
be established without proper temporal order of toid arthritis. In this study, the researcher may
exposure first and outcome later. Cross-sectional describe the frequency of different symptoms
82 Research Methodology and Biostatistics

experienced by the patients and the type of treat-

ment they received during the course of disease, etc.
CASE–CONTROL RESEARCH DESIGN
There are multiple variables in this research study. In this design, causes of a disease are investigated
The basic purpose of the study is not only to after the occurrence of a disease (Fig. 4.13). For
describe each of the variables but also to establish example, a researcher investigates the history of
the relationship between those variables. This smoking in patients diagnosed with lung cancer.
design is mainly used to identify and describe the Some of the specific characteristics of case–
perception, awareness, behaviour, attitude, knowl- control study design are as follows:
edge and practices of people. For example, this l This study design uses true exposure and out-

design was used for ‘a descriptive study of the per- come measures that are anchored in measured
ceived causes of female feticide among women in time. The study begins with the outcome mea-
selected rural and urban communities of district sure and relies on participant’s memories or
Jaipur, Rajasthan’. There are mainly two types of medical records to go back to measure the po-
descriptive studies in the field of epidemiology: tential exposure.
incidence and prevalence studies. For example, l A unique feature of this design is that individu-

descriptive study of the prevalence of pin-site als with the outcome of interest (case) are com-
infection among patients with external skeletal pared to individuals who do not have the out-
fixation can be carried out with this experimental come but were at risk for it (control) to
design. Or it can be used for ‘a descriptive study determine the difference in past exposure to the
on the incidence of urinary tract infection among potential cause of the health phenomenon.
catheterized patients admitted in intensive care
units (ICUs)’.

Exploratory Design Cause Effect

Exploratory design is used to identify, explore and
describe existing phenomenon and its related fac-
tors. In other words, it is not only a simple descrip-
tion or observation of the frequency of occurrence Exposed
of a phenomenon but also its in-depth exploration
and study of its related factors to improve further
Cases
understanding about a less-understood phenome-
non, for example, an exploratory study to assess the
multifactorial dimensions of falls and home safety Unexposed
measures for elderly people living in selected com-
munities in the city Ludhiana.
Exposed
Comparative Design
Comparative design involves comparing and con- Controls
trasting two or more samples of study subjects on
one or more variables, often at a single point
of time. This design is used to compare two Unexposed
distinct groups on the basis of selected attributes,
such as knowledge level, perceptions and atti-
tudes; physical or psychological symptoms; and so
on, for example, ‘A comparative study on health Direction of study/inquiry
problems among rural and urban older people in Fig. 4.13. ​Conceptual framework of case–control
district Bikaner, Rajasthan’. studies.
 Research Designs 83

l The unit of analysis in case–control studies is Box 4.1. ​Examples of research studies with
the individual; and in this design, only one mea- case–control research design
sure of association between exposure and out-
• A case–control study on the lifestyle practices of
come is appropriate, i.e. odd ratio.
patients diagnosed with coronary artery disease
l Case–control studies are especially valuable for
admitted in Hero DMC Heart Institute (HDHI),
studying rare or emerging diseases. Ludhiana
l This type of design begins with selecting the cases • A case–control study on the alcohol consumption
after operationally defining the cases. The opera- pattern among patients diagnosed with alcoholic
tional definition should not be too narrow or too liver cirrhosis at Gastroenterology Department,
broad because too narrow an operational defini- CMCH, Ludhiana
tion of the case makes it difficult to choose the
sufficient size sample in a given period and too
broad a definition seriously minimizes the validity the extent to which the sample represents the popu-
of study results. Therefore, a balance must be en- lation of individuals with the outcome cannot be
sured in operational definition of cases to avoid established because the sample was purposively se-
the afore mentioned problems. lected to include a large proportion of cases.
l In this study design, cases and controls are iden- Another weakness of the case–control design is
tified from either clinical or general population uncertainty about the true temporal order of expo-
sources. Clinical sources, like hospitals and med- sure and outcome due to potential weakness of
ical clinics, are convenient sources to obtain available historical data. In addition, the quality of
cases; but it must be ensured that when cases are data in this study design largely depends on the past
taken from a hospital, control also must be iden- memories of the study subjects and sincerity, and
tified from the same or similar hospital, which capability of person collecting the data. Examples
did have that specific disease manifestations. For of research studies with case–control design are
example, if cases of lung cancer are selected from presented in Box 4.1.
hospital A, control treated in hospital A for
illness other than cancer should be chosen. NESTED CASE–CONTROL DESIGN
However, patients with illness too similar to lung The nested case–control design is same as the tra-
cancer, such as emphysema, should be avoided as ditional case–control design except that only a
controls due to similar disease aetiology. subset of controls from the cohort are compared
l Cases may also be chosen from the community to the incident cases. The nested case–control
setting through other previous disease registry design is more suitable when expression of inter-
cross-sectional surveys. For these cases, con- est is difficult or expensive to obtain and when
trols should be chosen from the same neigh- outcome is rare. In this design, by only measuring
bourhood, community or region. Friends, rela- the covariate in as many participants as necessary,
tives and co-workers of cases can also be used the cost and effort of exposure assessment are
as controls. reduced.
l It must be ensured that whenever possible and

practical, controls should be similar to cases

in important outcome-relevant characteristics,
COHORT RESEARCH DESIGN
such as age, gender, occupation, medical history, A cohort is a group of people who have something
residence and so on. in common and who remain part of a group over
The potential weakness of case–control studies an extended time. In health care, the subjects in
is an inability to calculate the incidence or preva- cohort studies are selected by some defining char-
lence of the outcome. Incidence is impossible to acteristic (or characteristics) suspected of being a
measure because the study travels back in time; so precursor to or risk factor for a disease or health
cases cannot be measured as they emerge. Similarly, effect. In this design, a longitudinal approach is
84 Research Methodology and Biostatistics

used to investigate the occurrence of a disease in a typically needed to maximize retention rates and
cohort (group with similar characteristics for a subsequently the validity of the study sample. This
time period) with existing presumed causes com- design is also inefficient for rare outcome and for
pared with homogenous controls. For example, a those with a very long latency period (time from
researcher longitudinally observes a homogenous exposure to outcome).
cohort of smokers and nonsmokers for the devel-
opmental of lung cancer. PROSPECTIVE COHORT DESIGN
Some of the specific characteristics of cohort A prospective cohort study is an observational
study design: longitudinal study, where a cohort (a group of
l Cohort studies, also called incidence studies, are subjects with similar characteristics) are followed
designed to measure the exposure and outcome over a time period; but cohorts differ in certain
in the context of time. factors under study to determine how these fac-
l In this study design, individual subjects are fol- tors affect the rate of certain dependent outcomes
lowed over time to measure the exposure when (Fig. 4.14). For example, a researcher may follow a
it happens (in real time or historical); then they cohort of middle-aged nurses who vary in terms
measure the outcome at a point in time after of body mass index, to the hypothesis that the
exposure. Because the outcome is measured ex- 15-year incidence rate of coronary artery disease
actly or approximately when it happens, the in-
cidence of new cases can be determined in co-
hort studies.
l The strength of this design is the ability to dem-
With intended
onstrate the temporal order of the exposure and outcome
outcome – a necessary criterion to determine
causality. Subjects
(exposed)
l While case–control studies are limited to odd

ratios for measuring association between expo- Without

intended
sures and outcomes, cohort studies allow the outcome
use of risk differences, risk ratios and odd ratios Cohort
for evaluating associations. The concept of ‘risk’ selected
is meaningful because the outcome or disorder for study
is measured as incidence. With intended
l Cohort studies could be prospective, histori- outcome
cal/retrospective or ambispective. The retrospec-
tive cohort study design differs from the Controls
(unexposed)
case–control study design; in retrospective
cohort study, the subjects chosen are at high risk
Without
for outcome, naturally including those with or intended
without the outcome; however, in case–control outcome
studies, subjects are purposively chosen with the
outcome; then a group of control without the
disorder from the same setting is used to
compare the history of exposure with cause.
The main disadvantage of cohort studies is that Onset of study Time
they tend to be very expensive and time consum-
ing. The study is substantively weakened with Direction of the study
high attrition of subjects between measurement Fig. 4.14. ​Conceptual framework of prospective cohort
points. Considerable resources and staff time are studies.
 Research Designs 85

Box 4.2. Examples of research studies with the researcher merely collects the data now, and
prospective cohort design establishes the risk of developing a disease if
exposed to a particular risk factor (Fig. 4.15).
• A prospective cohort study on the effect of earth-
For example, one researcher wanted to study
quakes on migration of people of Gujarat
• A prospective cohort study to assess socioeco- outcomes in men with prostate cancer treated
nomic effect of tsunami in the tsunami-affected with a specific type of radiation therapy. He
areas of Tamil Nadu plans to examine existing records to look at sur-
• A prospective cohort study on the complication vival and tumour recurrence in 1607 men who
in development pattern among patients with were treated between 2009 and 2016 and had at
diabetes type I visiting the endocrinology OPD of least four prostate-specific antigen measure-
Nehru Hospital, PGIMER, Chandigarh
ments after radiation. This approach to a study
is possible if the records on follow-up are com-
plete and adequately detailed and if the investi-
will be highest among obese nurses, followed by gators can ascertain the current status of the
overweight and the healthy weight. patients. In retrospective cohort studies, a risk
Prospective cohort studies are helpful in de- ratio or odd ratio gives an assessment of relative
termining the aetiology of different diseases and risk of a disease or disorder.
disorders. The important feature of the cohort
prospective study is that subjects with homoge-
nous characteristics are enrolled in the cohort
and a specific intended independent variable is With intended
observed in subjects longitudinally for a time outcome
period to determine the effect of the hypothe-
sized risk factor on the causation of a particular Subjects
(exposed)
disease or disorder. For example, a researcher
might follow a cohort of middle-aged shopkeep- Without
intended
ers who vary in terms of smoking habits, to test outcome
the hypothesis that the 20-year incidence rate of Records
lung cancer will be highest among heavy smok- selected
ers, followed by moderate smokers and then for study
nonsmokers. Other examples of research studies With intended
with prospective cohort design are presented in outcome
Box 4.2.
Controls
(unexposed)
HISTORICAL COHORT/RETROSPECTIVE
COHORT DESIGN Without
The historical cohort design is also known as intended
outcome
retrospective cohort design, where one can also
undertake a cohort study by using information
collected in the past and kept in records or file.
In the case of a retrospective cohort study, the
investigator collects data from past records and
does not follow up patients as is the case with a Onset of study Time
prospective study. Thus, in retrospective cohort
study, all the events – exposure, latent period Direction of the study
and subsequent outcome (e.g. development Fig. 4.15. ​Conceptual framework of historical cohort
of disease) – have already occurred in the past; studies.
86 Research Methodology and Biostatistics

AMBISPECTIVE COHORT DESIGN l Trend studies: These help to investigate a sam-

The ambispective study design moves both for- ple from a general population over a time with
ward and backward in time. With this design, the respect to some phenomenon. Trend studies
exposure is measured twice – historical and in real permit researchers to examine pattern and rate
time – during the study period. The design allows of changes, and to make prediction about future
a more precise measure of exposure in terms of direction on the basis of previously identified
time. It can be determined if the exposure is of patterns and rates of changes.
longer or shorter terms or variable with respect to l Panel studies: A panel in research is referred to
coming and going. For example, a study about hip the sample of people involved in a study. In
replacement (outcome in the future) can start panel studies, same people are involved; and
with adults aged 20–40 years, and their current over a period of time, they become more infor-
and future physical activities (contemporary mative on the phenomenon than the subjects in
exposure) as well as their school and other his- trend studies because the researcher cannot only
torical records can be measured to determine their examine the patterns of change but also the
childhood experience with organized sports type, reasons for change. The same selected people
intensity and duration. are contacted two or more times to collect
In another example of an ambispective cohort further data.
study: ‘An ambispective cohort study of the natu- l Follow-up studies: These are undertaken to
ral history of HIV infection among former determine the subsequent states of subject(s)
unsafe commercial blood and plasma donors’, with a specified condition or for those who have
HIV/AIDS cases were found and confirmed prior received a specific intervention.
to 24 July 2006 being former commercial blood
donors. Plasma donors were selected and data
regarding infection, incidence, death and influ-
ADVANTAGES AND DISADVANTAGES
encing factors were collected. Other examples of OF OBSERVATIONAL RESEARCH
research studies with ambispective design are DESIGN
presented in Box 4.3. ADVANTAGES
Longitudinal research design is used to collect l Observational research designs tend to be clos-
data over an extended time period (long-time est to real-life situations. Unlike many experi-
study). Its value is in its ability to demonstrate mental studies, descriptive research designs,
change over a period of time. Longitudinal studies case–control and cohort designs are rarely
are generally classified into three types: trend criticized for their artificiality.
studies, panel studies and follow-up studies. l Observational research designs are most suitable
for many of the health care research studies. In
fact, these research designs are extremely useful
Box 4.3. Examples of research studies with to enhance our understanding about the existing
ambispective cohort design real-world settings around us.
• An ambispective study to assess correlation be-
l Numerous human characteristics are inherently
tween the physical activities of childhood as well not subject to experimental manipulation
as later age and risk for the need of hip replace- (e.g. blood type, personality, health beliefs and
ment among adults residing in selected communi- medical diagnosis); therefore, the effects of
ties of Bangalore these characteristics on other phenomena can-
• An ambispective study to assess correlation be- not be studied experimentally.
tween the exposure to childhood as well as later l There are many variables that could technically
age intoxicants and risk of developing degenera-
tive brain disease among healthy middle-aged be manipulated, but manipulation is forbidden
adults residing in selected areas of New Delhi on ethical grounds. In such cases, it is fair to
carry out observational research.
 Research Designs 87

l There are many research situations in which it Methodological studies are conducted to develop,
is simply not practical to conduct a true ex- validate, test and evaluate the research instruments
periment. Constraints might involve insuffi- and methods. Basically, a methodological research
cient time, lack of administrative approval, includes the following steps; however, these steps
shortages of funds, excessively inconvenient, require sound, specified and exhaustive literature
etc. In such cases, observational researches are review to identify the theories and understand the
most suitable. construct.
l Defining the behaviour or construct to measure
DISADVANTAGES l Formulating the items for tools

l The major disadvantage of observational re- l Developing instruments for users and respondents

searches is that the results obtained and the l Testing the reliability and validity of the

relationship between dependent and indepen- research tool

dent variables can never be absolutely clear For example, a researcher may conduct ‘a meth-
and error-free. The sheer presence of a relation- odological study to develop a pressure sore risk
ship – a strong one – among variables is not assessment tool for patients admitted in ortho-
sufficient to certify the result that one variable paedic wards’.
induced the other.
l Observational studies are conducted for com- META-ANALYSIS
parative purposes using nonrandomly selected Quantitatively combining and integrating the
groups, which may not be homogeneous and findings of multiple research studies on a particu-
tend to be dissimilar in different traits or char- lar topic is known as meta-analysis. In other words,
acteristics that may affect the authenticity and it is a method of integrating quantitative research
generalizability of the study results. findings statistically. In 1976, Glass coined the
term meta-analysis; it indicates the analysis of
analysis.
OTHER MISCELLANEOUS RESEARCH
A meta-analysis statistically merges the out-
DESIGNS comes of various studies that hint of a research
There are some research designs that cannot be hypothesis that is shared. Just as individual studies
categorized under either experimental or observa- summarize data collected from many participants
tional research designs. Some of such research in order to answer a specific research question
designs are discussed in the following sections. These (i.e. each participant is a separate data point in the
include methodological studies, meta-analysis, sec- analysis), a meta-analysis summarizes data from
ondary data analysis, outcome research, evaluation several individual studies that concern a specific
studies and operational research. research question (i.e. each study is a separate data
point in the analysis).
METHODOLOGICAL STUDIES Meta-analysis is considered as the statistical
The main problem in health care research is to find analysis of a large amount of analysed results from
appropriate measurement tools or techniques to individual studies for the purpose of integrating
measure health-related psychosocial phenomena. the findings. It is believed that the findings from
Health care professionals frequently use tools devel- each study individually may not be effective
oped by other disciplines, such as psychology and or powerful enough to allow for decisions affect-
sociology. These tools from other disciplines can be ing clinical practices. However, when results of
useful if they are found appropriate to measure several similar studies are analysed together, the
health-related psychosocial phenomena, but most finding of such studies may be more effective or
of the time these tools are only used because of powerful. Meta-analysis studies start with select-
the unavailability of reliable and valid tools to mea- ing studies with similar variables and population
sure the health-related psychosocial phenomena. samples, followed by identifying and coding study
88 Research Methodology and Biostatistics

characteristics, and finally statistically analysing behaviour of people with cancer, which are
and reporting the assessed findings of the meta- conducted on similar age groups, and later carry
analysis study. out a statistical analysis on the results of the
selected studies.
Steps in Conducting a Meta-Analysis Literature review and meta-analysis are not
There are generally five steps in conducting the synonymous; meta-analysis is a statistical analysis
meta-analysis. of results of previous researches with similar vari-
1. Define hypothesis: First, there must be a well- ables and population, while literature review is
defined statement of the relationship between only a descriptive presentation of different previ-
the variables under investigation so that the ous research studies on a similar topic without any
researcher can carefully define the inclusion analysis and interpretation of results of the previ-
and exclusion criteria when locating potential ous studies.
studies.
2. Locate the studies: A meta-analysis is only SECONDARY DATA ANALYSIS
informative if it adequately summarizes the Secondary data analysis is a research design in
existing literature, so a thorough literature which the data collected by one researcher are re-
search is critical to retrieve every relevant study, analysed by another researcher, usually to test new
such as database searches, ancestry approaches, hypotheses. Sometimes researchers collect lots of
descendancy approaches, manual searching data in a study, out of which some of the data are
and the invisible college (i.e. network of left unused or unanalysed that is later taken up and
researchers who know about unpublished utilized by another researcher. This is called
studies, conference proceedings, etc.). secondary data analysis research. In this research
3. Input data: Empirical findings are gathered design, researchers analyse data collected in previ-
from primary studies (e.g. p-value and effect ous studies, but they may have a new research
size) and input into statistical database. Not question or may test new hypotheses. These stud-
every study provides sufficient statistical infor- ies are considered as most convenient, time saving
mation to calculate the effect size statistics. and cost-effective because data collection is what is
4. Calculate effect sizes: The overall effect is calcu- usually considered as the most difficult, time con-
lated by converting all statistics to a common suming and costly affair in a research activity.
metric, making adjustments necessary to correct A secondary analysis can be performed with
for issues, like sample size or bias, and then cen- both quantitative and qualitative data.
tral tendency (e.g. mean effect size and confi-
dence intervals around that effect size) and vari-
ability (e.g. heterogeneity analysis) are calculated. Quantitative Study Example
5. Analyse variables: If heterogeneity exists, you A researcher utilized data from a past study of pa-
may want to analyse moderating variables in tients having chronic obstructive pulmonary disease
the database and analysing either mean differ- to test an analytical model of factors affecting func-
tional status in this disorder.
ences (for categorical variables) or weighted
regression (for continuous variables) to see if
the variable accounts for the difference in the
effect size. Note: Even if heterogeneity does not Qualitative Study Example
exist, some researchers insist that analysing From a qualitative study of clients with end-stage
moderating variables is appropriate. renal disease, a researcher did a secondary analysis of
For example, this design can be used for ‘a data. The initial study was devised to evaluate how
meta-analysis study on the coping behaviour of and by whom were the decisions about treatment
the people with cancer diagnoses’. In this study, modality done; the secondary study evaluated fac-
tors influencing those decisions.
researcher selects quantitative studies with coping
 Research Designs 89

OUTCOME RESEARCH
Examples of Outcome Research
Outcome research involves the evaluation of
• An outcome research on the effectiveness of the
care practices and systems in place. It is used Quality Assurance Programme (QAP) implemented
in health care to develop evidence-based prac- in Nehru Hospital, PGIMER, Chandigarh and
tice and to improve health care services. It is Safdarjung Hospital, New Delhi
a research that is planned to assess or record the • An outcome research study on patients’ satisfac-
end result of health care services. These studies tion with the existing health care model in DMCH,
are conducted in response to the increasing Ludhiana
demand from the public to justify care practices
and systems that improve patient treatment
outcome and reduce costs of care. The focus of EVALUATION STUDIES
the outcome research is predominantly on l In the era of evidence-based practice and client-
patients’ health status and cost of care; but there centred care, evaluation studies assume great
has been growing interest among health care importance. Evaluation research studies are an
professionals and related personnel to study applied form of research design, which involve the
patient outcomes in relation to health care judgement about how well a specific programme,
services care. practice, procedure or policy is working. Evalua-
tion studies may also be used to determine the
effectiveness or value of processes, personnel,
equipment and the material used in a particular
Structures
(having the right things) setting.
l Health care professionals can evaluate their
practices by reflecting on what they do. Evalua-
tion studies are generally carried out when the
Processes researcher wants to find out how and to what
(doing the right things)
extent the objectives of a particular activity
have been or are being met. For example,
Sharma (2012) carried out an evaluation study
Outcomes to assess the effectiveness of the Anganwari
(having the right results) Scheme on the nutritional level of beneficiary
children in selected villages of district Sikar,
Shifting paradigm of allied health science research
Rajasthan, where the researcher evaluates how
far the predetermined aims and objectives of
the particular scheme is achieved.
Outcome research aims to document effec- l The aim of the evaluation studies is to improve
tiveness of health care services by a global as- a particular activity; it makes sense that the
sessment of nursing and health care services. original aims and objectives of an activity pro-
The policy makers, insurers and public demand vide the benchmark against which effectiveness
improved patient outcomes and costs. This re- or success is evaluated.
search methodology is the result of demands by l Evaluation research is not similar to audit. It
the professional standards review organizations uses the original aims and objectives as bench-
for quality assessment and quality assurance mark to assess the effectiveness of a particular
of health care services. Outcome research essen- activity. Audit, on the other hand, is a cyclic
tially justifies existing care practices and process that involves setting standards for
systems in place. Focus on health status and practice, monitoring that practice, comparing
costs associated with medical care studies have actual practice with the standards set, if neces-
resulted in better patient outcomes. sary making changes to practice, and then
90 Research Methodology and Biostatistics

remonitoring practice to see if the agreed practices to improve health care services. It may be
standards are attained. helpful in effectively designing a hospital or ward
l Evaluation research could be either formative for an efficient flow of man and material, con-
evaluation or summative evaluation research. structing communication systems at low cost,
 Formative evaluation research refers to the as- computerizing patient information storage for ef-
sessment of a programme as it is being imple- ficiency, etc.
mented; the focus is on the evaluation process In operational research studies, researchers as-
of a programme rather than the outcome. sess the effectiveness of different programmes, fa-
 Summative evaluation refers to the assessment cilities and personnel models in a natural setting as
of the outcome of a programme that is they are implemented to improve work efficiency
conducted after the completion of the of an organization.
programme.
l The main limitation of evaluation studies lies in
the fact that they are aimed at understanding a Operational Research Study Examples
specific practice, programme, policy or event. • An operational research study on patients’ satisfac-
Their contribution to knowledge, in general, tion with, and awareness of, single- and double-
and research methodology, in particular, corridor ward designs in selected hospitals at New
remains a secondary objective. Delhi
• An operational research study on the patients’
perception about different patient assignment
Examples of Evaluation Studies methods for administering nursing care in chosen
hospitals of Hyderabad
• An evaluation study on the implementation of
the National Rural Health Mission Programme in
selected districts of Rajasthan
• An evaluation study on the effectiveness of new
infection control programme at AIIMS, Rishikesh, QUALITATIVE RESEARCH DESIGNS
India
Qualitative research approaches have somewhat
short and less well-defined plans. In a qualitative
OPERATIONAL RESEARCH research study design, elements typically evolve
Operational research involves the study of com- over the course of the project. As one study un-
plex human organizations and services to develop folds, decisions are made in the field about how
new knowledge about institutions, programmes, best to obtain information, and from whom; how
use of facilities and personnel in order to improve to schedule data collection; and how long each
working efficiency of an organization. data collection session should last.
Operational research is the application of the Qualitative research as emergent design: The
scientific method of investigation to the study of design for a qualitative study is emergent in na-
complex human organizations and services. While ture, i.e. a design that emerges as the researcher
conducting operational research, one is concerned makes ongoing decisions while reflecting on what
all the time with the activities of a group of people has already been learnt. It has been mentioned by
with the purpose of inducing some changes. eminent researchers and thinkers that an emer-
The main objectives of operational research are gent design in qualitative studies is not the result
to develop new knowledge about institutions, pro- of researchers’ sloppiness or laziness in not having
grammes, use of facilities and personnel in order a design in place before the commencement of the
to improve working efficiency of an organization. study, but rather it is their desire to base their
Operational research translates research concepts enquiry on the realities and viewpoints of those
into measurable phenomena. It is very useful in under study – realities and viewpoints that are not
health care studies as we require evidence-based known or understood at the outset.
 Research Designs 91

IMPORTANCE OF QUALITATIVE RESEARCH

DESIGNS Emerge as
l Qualitative methodologies contribute a great study
deal in health care studies, as it is a discipline advances Flexible and
elastic
that is building a knowledge base in the process
of clarifying how health care sciences should be Ongoing
data analysis
developed; therefore, health care scientists have
appreciated this contribution. Characteristics Multiple
l Qualitative research methods enable researchers of qualitative strategies of
to study social and cultural phenomena; so they research design data collection

also contribute to the social sciences. Researcher

l Qualitative research is an inductive approach as instrument
for discovering or expanding knowledge. In
this, it includes researcher’s involvement in Intensive Holistic
the identification of meaning or relevance of involvement
a particular phenomenon to the individual. of researcher
Qualitative strategies are useful for exploring
facts and developing concepts about an area
Fig. 4.16. ​Characteristics of qualitative research design.
of interest that has received little research
attention.
l Ongoing data analysis: Qualitative research
CHARACTERISTICS OF QUALITATIVE RESEARCH design requires data analysis for the formation
DESIGN of subsequent strategies and to determine when
A number of different disciplines have been guided further fieldwork should be done.
by qualitative enquiry; and each one of them has
produced research methods that best suit its disci- PHASES OF QUALITATIVE RESEARCH DESIGN
pline. Some of the general characteristics of a quali- The exact form of a qualitative study design can-
tative research design tend to apply across most not be known and specified in advance. The natu-
disciplines. Qualitative research designs basically ralistic enquiry processes go through three broad
have the following characteristics (Fig. 4.16). phases when the researcher is in the field. The
l Emerge as study advances: Generally, it is three main phases of qualitative research designs
believed that qualitative research designs emerge are as follows:
as study advances. l Orientation and overview phase: This is the
l Flexible and elastic: They are flexible and elas- first phase of qualitative research design.
tic, and can be adjusted to the information At this stage, the researchers only presume
being gathered during data collection. the type of knowledge that is expected
l Multiple strategies of data collection (triangu- to be obtained by conducting this particular
lation): Qualitative designs typically merge the qualitative study. However, they are not
various strategies of data collection. familiar with the phenomenon that will
l Holistic: They tend be holistic, striving for an drive the enquiry forward. Therefore, initially,
understanding of the whole. researchers get an overview of the salient
l Intense researcher’s involvement: They features of interest. This orientation and
require intense involvement of the researcher, overview phase enables them to plan further
i.e. commitment for longer periods in the field for the research study.
of study. Sometimes qualitative research l Focused exploration: This is the second phase

designs also require the researchers to become in qualitative research design. The salient
research instruments for data collection. aspects of the phenomenon are more focused in
92 Research Methodology and Biostatistics

this phase, and then an in-depth exploration of ethnography research, grounded theory, histori-
the salient aspects of the phenomenon is carried cal research and action research. A brief discus-
out. In this phase, a variety of people related to sion of the main types, meaning and description
the field are invited to participate in the study of qualitative research designs may be perused
and questions are asked from them to gather from Table 4.6.
more information about the phenomenon.
Questions are asked based on the understand- PHENOMENOLOGICAL RESEARCH
ing developed during the first phase of the Phenomenology is a movement in philosophy
design. Therefore, at this stage of the design, that has been adapted by certain sociologists to
focus is on exploration of the salient aspects of raise an understanding of the relationship be-
the phenomenon under study. tween states of individual consciousness and
l Confirmation and closure: This is the third and social life. As an approach within sociology, phe-
final phase of the qualitative design. Efforts are nomenology attempts to uncover how human
undertaken to establish that the findings gath- awareness is implicated in the production of so-
ered are trustworthy. The qualitative researchers cial action, social situation and social world. The
regroup and discuss their understanding with aim of the phenomenological approach to quali-
the participants of the study. Therefore, the tative research is to trace out precisely the live
qualitative researchers confirm their findings by experiences of people and generate theories or
analysing and discussing with study participants models of the phenomena being studied.
the authenticity and correctness of their find- According to G.W.F. Hegel, ‘Phenomenology is
ings; and then finally the study is closed. an approach to philosophy that begins with an
exploration of a phenomenon (what presents itself
MAIN TYPES OF QUALITATIVE DATA to us in conscious experience), logical, ontological
COLLECTION and metaphysical spirit that is behind phenome-
Those who are not familiar with qualitative meth- non. This is called a “dialectical phenomenology”.’
odology may be surprised by the sheer volume of According to Edmun Husserl, ‘Phenomenol-
data and the detailed level of analysis that result ogy is an approach to philosophy that takes the
even when research is confined to a small number perceptive experience of phenomenon (what
of subjects. Following are the three main methods presents itself to us in phenomenological reflec-
of data collection. tion) as its origin and attempts to distillate from
it the main traits of experiences and the essence
1. Written depiction People are expected to of what we experience’, as it is shown diagram-
by participants write down in detail their matically in Fig. 4.17.
understanding of the
phenomenon Characteristics of the Phenomenological
2. Observation Detailed observations of Approach
both verbal and nonverbal l Phenomenology tends to withstand the accep-
behaviours
tance of those circumstances that are unobserv-
3. Interactive People are asked to verbally able, and is a grand system erected in speculative
interviewing mention their understand-
thinking.
ing of the event
l Phenomenology tends to oppose naturalism,
i.e. objectivism and positivism.
l Positively speaking, phenomenology tends to
TYPES OF QUALITATIVE RESEARCH
justify knowledge with reference to awareness of
DESIGNS a substance itself, as disclosed in the most com-
In this chapter, the main qualitative research de- prehensive, distinct and suitable way for some-
signs discussed are phenomenological research, thing of its kind.
 Research Designs 93

Table 4.6. ​Main types of qualitative research designs

Qualitative Research Design
Categories Definition and Narration
1. Ethnographic research Concentrates on the sociology of meaning through close field observation of
sociocultural phenomena. Basically, the ethnographer checks on a community
2. Grounded theory Theory is developed inductively from a corpus of data gathered by a participant
observer
3. Historical research Systematic accumulation and objective measurement of data relating to earlier
occurrences to measure hypotheses pertaining to causes, effects or trends of
these events which might facilitate in explaining present events and antici-
pate future events
4. Case study Trying to concentrate on a phenomenon by checking in depth an individual case
example. The case can be a group, an individual, an event or an institution
5. Action research Action research is a form of applied research that tries to empower people
through a process that constructs and uses knowledge. It tries to find practical
solutions to problems existing in the framework of an organization. It
increases understanding of how change in one’s actions or practices can
mutually benefit health care professionals within an organization
6. Phenomenological Details the structure of experiences as they present themselves to awareness,
research without recourse to assumptions, theory or deduction from other methods

Approach to Table 4.7. Data collection issues in

philosophical phenomenology
context
Features of
experience Phenomenology
Intuitive experience Issues Research
of phenomenon
among people
Essence of Methods of data Primarily in-depth interviews,
what is collection sometimes diaries or other
experienced written material
Units of data Individual subject
Fig. 4.17. ​Schematic presentation of Husserl’s view collection
about phenomenology.
Data collection Mainly cross-sectional
points
Length of time for Typically moderate
l Phenomenology tends to seize/grasp that en- data collection
quiry ought to emphasize upon ‘encountering’ Data recording Interview notes and/or audio-
as it is directed at objectives and correlatively tape recording
upon objectives as they are encountered. Salient features Bracketing one’s view, build-
l The data collection issues in phenomenology are ing rapport, encouraging
shown in Table 4.7. The primary sources of data candour, listening while
collection are the real-life situations of the indi- preparing what to ask next,
viduals being studied, wherein in-depth inter- keeping discussion on
views, i.e. semistructured, audiotaped inter- ‘track’ and handling
views with participants, are the most common emotions
94 Research Methodology and Biostatistics

means. Furthermore, emerging themes are fre- grounding for the world and the positive sci-
quently validated with participants because ences of it, for example, social beliefs, positions
their meanings of that live experience are cen- and practices.
tral to phenomenological study. People who 3. Existential phenomenological research:
have recently had the experience are selected as This is concerned with topics, such as ac-
participants. For example, a recently diagnosed tions, conflicts, desires, finitude, oppression
HIV patient or a woman who has had a stillborn and death.
baby in the last 3 months can be interviewed for 4. Hermeneutical phenomenological research:
a study of psychological impacts of these medi- It utilizes vital experiences as a device for bet-
cal conditions. Bracketing is a technique used by ter consideration of the political, social,
researchers to assist participants in describing cultural or historical aspect in which those
live experiences, while at the same time setting experiences happen. Hermeneutic enquiry al-
aside their own personal feelings. Statements most always concentrates on interpretation
that would be helpful in eliciting a participant’s and meaning: how historically and socially
description should be a part of such an inter- conditioned individuals portray their world
view. The researcher must use intuiting to de- within a given context.
velop an awareness of the live experiences with-
out forcing prior expectations or knowledge Example of Phenomenological Research
into the process. Then after classifying, the re-
What is the vital experience of adults who are com-
searcher attempts to reduce categories until
bining hearing loss into their lives? The sample com-
overlap, vagueness and irrelevancy are elimi- prises 32 adults with mild-to-profound extent of
nated. The final listing of categories should re- hearing loss. Data were gathered through semistruc-
flect the live experience captured in a coherent, tured, audiotaped interview with the participants.
meaningful manner. Analysis took into consideration the identification of
l Data analysis: The process of analysis involves major and core themes present in the data, and the
the difficult task of contrasting and comparing authorization of the findings with selected partici-
pants. Through the participants’ perception of mov-
the final data to determine what pattern, themes
ing, awkwardly or gracefully, with the changes
or threads emerge. In the final analysis, a wise needed due to hearing loss, with them never being
researcher seeks further knowledge about that sure of the next step, the core theme of dancing was
live experience in a concise manner. If the elegantly secured. The dominant themes were danc-
knowledge is to be of relevance to other re- ing with (a) fluctuating feelings, (b) loss and fear,
searchers, it must be understandable and clear, (c) an altered life perspective and (d) courage amid
detailing the relationship that exists. change.
It gave the researchers a nice detailing of what it
Types of Phenomenological Research was like to live with hearing loss.
These outcomes provide health care profession-
There are basically four types of phenomenologi- als a crystal-clear apprehension of a phenomenon
cal researches, as discussed here. that they can utilize in their correspondence with
1. Realistic phenomenological research: It fo- people living with hearing loss.
cuses on gathering the universal abstract of
various types of information, including human
actions, motives and results. Some examples of phenomenological research
2. Constitutive phenomenological research: This design are presented in Box 4.4.
includes the philosophy of natural sciences. This
procedure involves suspending acceptance of ETHNOGRAPHIC RESEARCH
the pregiven position of conscious life as some- Ethnography, associated with the field of anthro-
thing that exists in the world, and is carried out pology, is a branch of human enquiry, which
in order to obtain an ultimate intersubjective concentrates on the culture of a group of people
 Research Designs 95

Box 4.4. Examples of phenomenological rather than using a formal tool for data
research studies collection.
l Information on three major aspects of cultural
• A phenomenological study on live experiences life is sought in the ethnography studies:
of tsunami victims in selected villages of cultural behaviour (what members of the
Tamil Nadu
culture do), cultural artefacts (what members of
• A phenomenological study on the psychological
experiences of the victims of the Bhopal gas the culture make and use) and cultural speech
(what people in the cultural group say).
tragedy at Bhopal, Madhya Pradesh
l Ethnographers rely on various sources of data

collection such as in-depth interviews, record

analyses and observation of physical evidence
with an effort to interpret their worldview. (photographs, diaries, letters, etc.).
Ethnographic studies are involved in the collec- For example: ‘An ethnographic study focusing
tion and analysis of data about cultural groups. on blood donation beliefs of women living in
Ethnography is basically classified into two rural areas of district Ludhiana, Punjab’. Some
types: macro- and microethnography. Macroeth- more examples of ethnographic research design
nography is a study of broadly defined culture, are presented in Box 4.5.
while microethnography is a study of more narrow
aspects of a culture. GROUNDED THEORY
Ethnographic research is a method of con- Grounded theory is an inductive technique devel-
ducting enquiry of a life process by studying oped for health-related topics by Glaser and
individuals, artefacts or documents in their Strauss (1967). It emerged from the discipline of
natural setting. It includes both anthropology sociology. The term grounded theory means that
and historical forms of research. In health care the theory developed from the research is
research, ethnography provides access to health ‘grounded’ or has its roots in the data from which
beliefs and health care practices in particular it was derived.
cultural or subcultural groups. Therefore, Grounded theory has become an important
ethnographic enquiry facilitates understanding research method for the study of health sciences
cultural behaviour and practices affecting theories of phenomena relevant to health care
health of people. professionals. It is an approach to study of social
process and social structures. It is different from
Characteristics of Ethnographic Research other methods because of its specific access to
l Ethnographers learn about cultural groups in theory development. As per grounded theory, there
which they are interested through extensive should be an ongoing interplay between data col-
fieldwork. lection and data analysis. The main focus is on
l Ethnographic research is a labour-intensive and
time-consuming endeavour, where even months
or years of fieldwork can be involved. Box 4.5. Examples of ethnography research
l A certain level of intimacy with cultural group studies
members is required to study culture. Intimacy • An ethnographic study on the features, critical at-
can develop over time and by working together tributes, processes and benefits of self-help groups
directly with those cultural group members of women living with chronic alcoholic husbands
who are active participants. in selected villages of district Belgaum, Karnataka
l Researchers use themselves as instruments in • An ethnographic study on the sociocultural beliefs
these ethnography studies, where they spend of people about antenatal, natal and postnatal
time with group members to collect data care of women in selected tribal communities of
Rajasthan
through informal interactions and observations
96 Research Methodology and Biostatistics

developing social experience, the social and

Data collection
psychological stages and phases that characterize a
particular event or episode.
Martin and Jurner (1986) have mentioned
that ‘grounded theory is an inductive theory Note taking
discovery methodology that allows the researcher
to develop a theoretical account of the general
features of a topic while simultaneously ground-
ing the account in the empirical observation Coding
or data.’

Characteristics of the Grounded Theory

Memoing
l Overview and phases of grounded theory:
Grounded theory begins with a research situa-
tion. Within that situation, the task of the re-
searcher is to understand what is happening Sorting
there (core variable), and how the players man-
age their role. After each hour of data collection,
the researcher notes down key issues; this is
Writing
known as note taking. Constant comparison lies
at the heart of the process. First, the researcher
compares one interview (or other data) to an- Fig. 4.18. ​Phases of grounded theory.
other (or other comparative data). From this
comparison, a theory emerges. Researchers com- inductive and deductive process is utilized
pare the initial data to the theory, and the results (deductive means what is expected to be found
of this comparison are written in the margin of in social life). For example, a community health
the notes taken as ‘codes’. The researcher’s task is nurse is using grounded theory methodology
to identify categories and their properties from to study scapegoating in dysfunctional families.
these codes. In the researcher codes, there may be The nurse limits the study to those families that
links between categories or a core category may have children (under 10). Observation for the
emerge (a category which appears central to the purpose of collecting data occurs primarily in
study), and proceeding to this step provides the the homes of the participants. Family interac-
researcher with a final theory. The researcher tion and communication patterns are assessed
writes further notes about this theory, which is through observation and interviews as well as
called memoing. If this process saturates, it is the through role-playing (manipulation) and ob-
time for sorting. The researcher groups memos, servation of family members in their work or
line-by-line, and sequences them in whatever school environment. As data collection pro-
order will make his/her theory clearer. Over time ceeds, the nurse begin to form some hypotheses
a grounded theory works through the mostly regarding scapegoating activity. During this
overlapping phases (Fig. 4.18). process, the nurse began to categorize the data
l Methodology: The steps of the grounded the- and code it; coding helps her in thinking and
ory research occur simultaneously. An impor- conception of the families. To verify the theory,
tant methodological technique in grounded the nurse plans to study other dysfunctional
theory is constant comparison. Although this families who are scapegoating.
method is frequently considered to be an in- l Sources of data collection: For data collection,
ductive means of developing theory (from the the researcher will use all his/her senses. There is
data actually gathered), in reality a combined a lot to be learned just by observing; some of it is
 Research Designs 97

Table 4.8. Data collection issues in grounded

theory Audio/video
tape
Issues Grounded Theory recording
Diaries
Methods of data Primarily individual inter-
Formal/
collection views, sometimes group informal
interviews, observation, interviews
participant journals,
documents, self-reports Sources of
data Case
Units of data Individual subjects collection studies
collection
Data collection Cross-sectional or longitudinal Documents
points
Length of time for Typically moderate Artwork
data collection Participant
observation
Data recording Interview notes/audiotaping/
memoing/observation
notes
Fig. 4.19. ​Sources of data collection in grounded theory.
Salient features Building rapport, encourag-
ing candour, listening
while preparing what to
ask next, keeping ‘on track’ etc., are read in search of the answer to the
and handling emotions repeated question, ‘What is this about? What
is being referred to here?’
 These labels can refer to anything like hospitals,
evident within minutes of entering a situation. information gathering, friendship and social
Sources of data collection vary with the focus of loss. Part of the analysis process is to identify
enquiry, the purpose of investigation and the the more general categories that these things
guidelines suggested by the research approach are a part of, such as institutes, work activities,
being utilized. The data collection issues for social relations and social outcomes. For exam-
grounded theory are shown in Table 4.8. ple, with regards to friendship, we might ask
Some possible sources that could be used in col- about its duration, its closeness or its impor-
lecting data relevant to the life processes have tance to each party. It is important to have a
been shown in Fig. 4.19. fairly abstract category in addition to every
l Data analysis: Typically, grounded theory re- concern, as the abstract categories help to gen-
search projects in health care studies tend to erate general theory. Considering pain, for ex-
have a sample of 25–50 people, and are con- ample, the researcher views pain as having dif-
ducted by in-depth interviews, data collection ferent properties, like intensity (it varies from a
notes, typed interview transcripts or videotaped/ little to a lot), consequences (when it hurts a lot,
audiotaped conversations that contain multiple we do not want to get out of our beds, do not
pieces of data to be stored and analysed. The feel like doing anything), agents of pain relief
process is initiated by coding and categorizing (drug), duration of pain relief (could be tempo-
the data. There are several types of coding: open, rary) and effectiveness (could be partial).
axial and selective coding.  One can see that this sort of analysis has a
 Open coding: It is the part of the analysis re- very emic cast to it; even though most
lated to categorizing, identifying, naming and grounded theories believe they are theorizing
describing phenomenon as found in the text. about how the world is rather than how re-
Essentially, each paragraph, line, sentence, spondents see it, coding can be done very
98 Research Methodology and Biostatistics

formally and systematically or quite infor- Box 4.6. Examples of grounded theories
mally. In this it is done informally. As the research studies
codes are developed, it is useful to write
• A study on the response and adaptation process of
memos known as code notes that discuss the patients diagnosed with cancer in Nehru Hospital,
codes. These memos become source for later PGIMER, Chandigarh
development into reports. • A grounded theory research on the adaptation
 Axial coding: It is the process of connecting levels of people with postinjury permanent physi-
codes (properties and categories) to each cal disabilities at AIIMS, New Delhi
other through deductive and inductive think-
ing. In order to simplify this process, besides
looking at other kinds of relations, grounded theory that came from their research: ‘The fam-
theorists point out causal relationships and fit ilies of chronically ill children establish specific
things into a main frame of generic relation- ways of coping to meet the requirements of all
ship. It seems obvious from the previous ex- their members. If support is available, equilib-
ample that the phenomenon of interest is rium is achieved. If needs increase dramatically
pain, causal condition is arthritis, the action or support changes, there is lack of equilibrium.’
strategy is taking drugs and the consequence Some more examples of the grounded theory
is pain relief. Note that grounded theorists do research design are presented in Box 4.6.
not show much interest in the consequences
of the phenomenon itself. HISTORICAL RESEARCH DESIGN
 Selective coding: Selective coding is the act of The systematic collection and critical evaluation
selecting one category as the core category of data relating to past occurrences of a particular
and connecting all categories to that category. phenomenon is also a tradition that relies primar-
The main idea is to find out a single storyline ily on qualitative data. Historical research is un-
around which everything else is draped. It is dertaken to answer questions concerning causes,
about finding the driver that impels the story effects or trends relating to past events that may
forward. shed light on present behaviour or practices.

Types of Grounded Theory Characteristics of Historical Research Design

There are mainly two types of grounded theories. l It is important not to confuse historical research
l Substantive theory: It is grounded in data on a with a review of literature about historical
specific substantive area, such as postpartum events. Historical research involves the careful
depression. study and analysis of data about past events.
l Formal theory: Substantive theory serves as a l Historical research is a critical investigation of
springboard for developing a higher, more events, their development, experiences of the
abstract level of theory from a complicated past, the careful weighing of evidence of the
substantive grounded theory study regarding a validity of courses of information from the past
particular phenomenon. Kearney (1998) used and the interpretation of the weighed evidence.
an interesting analogy to differentiate substan- l The purpose is to gain a clearer understanding
tive theory (custom-tailored clothing) and of the impact of the past on the present and
formal theory (ready-to-wear clothing). future events related to the life process. It
l For example: A grounded theory study was con- involves detailed analysis of what has been writ-
ducted by a group of researchers, which looked ten or done and is used to describe, explain
at how parents coped with difficult times when or interpret these events.
caring for a chronically ill child. They audio- l Generally, historical research involves the re-
taped interviews of 30 families who used their view of written materials, but may include oral
clinic. The following is a brief statement of documentation as well.
 Research Designs 99

l Historical research typically relies on available actual witnesses of the same. These sources are
data. Data for historical research are usually in consciously produced for the purpose of trans-
the form of written narrative records of the mitting information to be used in the future.
past, diaries, letters, newspapers, minutes of  Documents classified as primary sources are

meetings, reports and so on. constitutions, characters, laws, court rulings,

l The results of the historical research studies official records, autobiographic letters, diaries
contribute to a clearer understanding of past, contracts, deeds, wills, licenses, declarations,
present or future events as they relate to health certificates, bills, receipts, newspapers, maga-
care and the life process. For example, a zines, advertisements, maps, inscriptions, dia-
researcher or investigator may be interested in grams, books, pamphlets, films, pictures,
studying the professional nurse’s role as an paintings, recordings and research reports.
advocate for pregnant women since the Second l Secondary sources: These are the reports of
World War. Knowledge of how the advocate’s people who related the testimony of an actual
role has been fulfilled in the past may provide witness of an event or actual participants in the
nurses with helpful information regarding how same. The writer of the secondary source was
that role might best be carried out today. not on the scene of the event. Secondary sources
l Historical method of research also covers cate- of data are usually of limited worth because of
gories, such as historical, legal, documentary, the usual errors that result when information is
bibliographical, biographical, ideational, insti- passed on from one person to another, for ex-
tutional and organizational. ample, most history books and encyclopaedia.
l Important existing data sources for biomedical
researchers are hospital records, nursing charts, Second Step: Criticism of the Data
physicians, order sheets and care plan state- The second step necessitates a comprehensive re-
ments. They all constitute rich data sources. view of gathered materials. This is an arduous task
Records are an economical and convenient requiring a sense of scepticism about the accuracy
source of information. of the documents in question. Christy (1975)
describes the analytic process of document review
Steps of Historical Research as a two-pronged activity.
Use of the historical research approach is high- l External criticism: The establishment of valid-

lighted by discrete steps. ity by determining the authenticity of the

source.
First Step: Data Collection  External criticism is covered basically with the

Comprehensive gathering of data is undertaken. authenticity and genuineness of the data. It is

Historical sources of data are usually classified a preliminary and preparatory step, providing
into two main categories. data to be used in the second phase known as
l Primary sources: Primary sources are first- internal criticism.
hand information that includes the following:  External criticism primarily deals with data

 Remains or relics associated with persons, relating to form and appearance rather than
groups, periods or events. meaning of contents, while internal criticism
 Fossils, skeletons, tools, weapons, utensils, weighs the testimony of the document in
clothing, buildings, furniture, pictures, paint- relation to the truth.
ings, coins and art objects are examples of the  The researcher ascertains that all the documents

remains that were not deliberately intended are original. First-hand oral and written
for use in transmitting information or to be accounts are accepted as valid.
used as records.  Signatures and date on materials are care-

 Oral or written testimony or the records kept fully reviewed; when each document selected
and written by actual participants in an event, or for review clearly meets the criteria for
100 Research Methodology and Biostatistics

authenticity, it is accepted as appropriate for involves logical processes rather than statistical
further analysis and interpretation. ones, and, therefore, the possibility of subjectiv-
l Internal criticism: The determination of reli- ity arises.
ability by correctly interpreting the contents of l Historical composition is a synthetic and con-
the document. structive process that involves the mechanical
 The use of original, authentic sources; aware- problems of documentation, the logical prob-
ness of one’s own biases; the substantiation of lem of arrangement of topics and subtopics,
the document in question by another collabo- and the philosophical problem of inter­
rating source are a few of the safeguards used pretation.
to ensure that interpretations are correct. l The earlier practice in historical writing pre-
 To ensure reliability, the researcher examines sented a strictly chronological type of organiza-
each document to make sure that the meaning tion. Such organization involved a series of
of facts and statement is clearly understood. events broken up into short time units. The
This process entails seeking collaboration from writing used to be in the form of calendars and
individuals who may have witnessed the events dates, events and names. The chapters were
in question as well as determining the meaning marked off by a period of a few months or a few
of words, phrases and colloquialisms unique to years, depending upon the changes of some
that time period. important nature or occurrence of some signifi-
 After the authenticity or genuineness of a cant events.
historical document or relic has been estab- l The organization of historical material can also
lished, the next question is to establish the be done in a topical, thematic or functional
validity of its contents or to determine the arrangement.
accuracy and value of the statement made. l A well-written history of education provides
 In performing internal criticism, historians information on the conditions of education in
must make several determinations, which the past. Even the stories written in the past
require historical knowledge beyond the show purpose and the meaning of education in
perusal of the materials in question. historical times, and provide background for
 Evidence bearing on the accuracy of historical better understanding of current educational
data might include one of the following: problems.
1. Comparison with other people’s accounts l The writing of history demands careful avoid-
of the same event to determine degree of ance of the following factors:
agreement  Oversignifying facts

2. Knowledge of time at which the document  Overgeneralizations from insufficient evi-

was produced dence

3. Knowledge of the points of view or biases  Failure to distinguish between significant and

of the written and oral documents trivial facts

4. Knowledge of the degree of competence of  Tendency to use secondary data

the writer to record events authoritatively  Tendency to accept statements as essentially

and accurately authentic when several observers agree

 Personal bias

Third Step: Presentation of the Facts  A dull and colourless style

After evaluating the authenticity and accuracy of his-  Failure to interpret words and expressions in

torical data, the researcher must bring the material the light of their usage in earlier times
together to analyse it and test the research hypotheses. l Determining the interrelationship among
l Historical researchers must be extremely careful facts and the meaning of the data gathered,
at this point since the analysis of historical data reviewed and recorded is the final step in
 Research Designs 101

historical research. The three steps identified Areas of Historical Study

are central to discovering the truth about the l Periods: Historical studies often focus on
past and how it can relate to health care pro- events and developments that occurred during
vided today. For example, a biomedical re- particular blocks of time in the past. The his-
searcher is interested in examining the influ- torical researcher gives these periods of time
ence of early Western health care leaders in names in order to allow the organization of
Indian modern health care system from 1890 ideas and the classificatory generalization to be
to 1950; a historical research approach would used by these historical researchers. The names
be appropriate in generating the evidence. given to a period can vary with geographical
The researcher could interview descendents, location as can the dates of the start and end of
review professional documents, read diaries a particular period.
and letters, and examine photographs, among l Centuries and decades are commonly used peri-
other sources. ods, and the time they represent depends on the
Scrutiny of these materials should assist in the dating system used. Most periods are con-
construction of the ‘story’ of facts regarding structed retrospectively and so reflect value
these leaders. Ultimately, the researcher would judgement made about the past. The way peri-
synthesize the material and make a determina- ods are constructed and the names are given to
tion regarding the contribution of these leaders them can affect the way they are viewed and
to health care as well as suggestions for future studied.
change. l Geographical locations: Particular geographi-
Another example of historical research is cal locations can form the basis of historical
that researchers have done intensive analysis of study, for example, continents, countries and
the changing image of different health care pro- cities.
fessionals through the examination of various l Military history: Military history concen-
documents. Documents examined have in- trates on the study of conflicts that have hap-
cluded film clips, novels, newspaper clippings pened in human society. This includes exam-
and magazine articles, among others. ining wars, battles, military strategies and
Intensive examination of these historical ma- weaponry. In addition, the effect of these
terials has allowed the researcher to identify events on society and health may also be
varying images during different time periods. examined.
The result of this analysis has been a systematic
evaluation of the image problems currently ex- Methods and Tools Used in Historical Research
isting within the profession. Some more exam- l Contemporaneous corroboration: Contempo-
ples of historical research design are presented in raneous corroboration is further support pro-
Box 4.7. vided by the existing evidence about past events.
It is a powerful method generally used by the
historical researcher to establish facts beyond
their limited lifespan.
Box 4.7. Examples of historical research l Photography: Photography is a methodological
studies tool for the collection of all known information
• A historical research on the development of phys- about individuals in a given period.
iotherapy in India l Historical revisionism: Historical revisionism
• A historical study on inception and development is traditionally used in a completely natural
of clinical research in India. sense to describe the work or idea of a historian
• A historical research on the development of medi- who has revised a previously accepted view of a
cal education in India
particular topic.
102 Research Methodology and Biostatistics

l Change log: Log or record of changes made to a Data Collection Methods

project, such as a website or software project, is Action research involves the following methods of
called change log. data collection:
l Human evolution: Human evolution is the pro- l Interview

cess of change and development or evolution by l Observation

which human beings emerged as distinctively l Storytelling

special. l Sociodrama
l Social changes: Changes in nature, social l Drawing and painting

institutions, social behaviour, or social rela- l Plays and skits

tions of a society or community of people l Other creative ways, like explore lives, tell stories

are social changes. For example, ‘The histori- and recognize own strengths.
cal evolution and political/moral contents
of technological instruments in obstetric Importance of Action Research in Health Sciences
surveillance and diagnosis, and the use of Research has the ability to exert power, especially
those instruments by health care professionals if research involves collaboration between study
signify social change. participants and the researcher. The research pro-
cess can improve health care practice and health
Example of Historical Study Using Records care researchers. The habit of thinking, the ability
to work harmoniously with others and the profes-
A researcher investigated the medical records of
100 patients who had in-hospital cardiopulmonary
sional spirit will improve. Health care knowledge,
arrest. The intention of the study was to find out pre- actions and consciousness will benefit from action
arrest physiologic changes in the patient and devise research, as it will help to recognize strengths and
factors that affected the resuscitation outcome. areas of improvement.
Action research is of particular importance to
the health care field, especially in the Indian setup,
ACTION RESEARCH as its objectives are as follows:
Action research is a form of applied research l To solve a problem by enriching the field of ap-

that tries to empower people through a process plication of a discipline

that constructs and uses knowledge. It tries to l To collaborate with several disciplines for solv-

find practical solutions to problems existing in ing the problem

the framework of an organization. It increases l To study the individual cases without the objec-

understanding of how change in one’s actions tive to generalize

or practices can mutually benefit health care l To recognize that other variables are constantly

professionals within an organization. Changes changing

are planned, which can be locally implemented l To try to say how things can be changed

in a particular organizational setup, and, there- l To report in common language

fore, are not universally applicable. Examples of the action research design are pre-
A branch of action research known as partici- sented in Box 4.8.
patory action research (PAR) emerged in the
1940s. It was started by Kurt Lewin, a social CASE STUDY
psychologist, and is closely related to critical l Case studies are in-depth examination of people,
research. The problem is researched, changed and places or institutions. Robson defines case study
reresearched within the research by the partici- as the development of detailed, intensive knowl-
pants and researchers. It is actively coresearched edge about a single ‘case’ or a small number of
by those who are hoped to be helped. Taking related cases. This strategy is of particular interest
action through planning and fact-finding brings if a researcher wishes to gain a rich
about future and organizational change. understanding of the context of the research and
 Research Designs 103

Box 4.8. ​Examples of action research studies Box 4.9. ​Examples of case study design
• An action research on the practicability of Western • A case study on the availability and utilization of
treatment practice protocols in public and private the emergency services in Civil Hospital, Ludhiana,
Indian health care organizations Punjab
• An action research on the use of proposed well- • A case study on the organization and functioning
being support programme to solve somatic mor- of the Civil Hospital, Jalandhar, Punjab
bidity among chronic mentally ill patients in a
selected mental hospital in North India
end-of-life issues of the terminally ill cancer
patient. The researcher may conduct a case
its processes. H. Odum defines case study method interviewing the terminally ill patient with
as a technique by which individual factor whether cancer or health care workers working in
it be an institute or just an episode in the life of hospices. In case studies, data collection
an individual or a group is analysed in its might be both qualitative and quantitative.
relationship to any other in the group. For example, in case study, a researcher may
l Case study is often used when the selected case obtain the pain on a numerical pain mea-
might offer insight into a unique situation or surement scale, respiration rate during pain
when a researcher wants to know more about a (quantitative) and/or may ask patient to
particular phenomenon within a real-life con- describe person’s experience of pain and
text. Using case study methodology, researchers discomfort (qualitative).
are able to maintain the richness and complex- l In health care sciences, case study methodology

ity of the concept of interest, while simultane- has been used since a long time for in-depth
ously highlighting specific details. study of a single patient or a group of patients to
l On the basis of the purpose of carrying out case generate knowledge for solving health care prob-
studies, they are classified into three major lems of patients suffering with specific disease
types, i.e. descriptive, exploratory and explana- conditions. For example, ‘How health care profes-
tory case studies: sionals and pregnant women manage pain dur-
 Descriptive case studies allow researchers to de- ing delivery process?’ Basically, case studies are
scribe characteristics, features and qualities of a considered as qualitative research studies; how-
yet unstudied person, institution or situation. ever, they could be considered either as a quanti-
 Exploratory case studies offer an opportunity tative or qualitative research study based on the
to clarify key concepts, ask more relevant purpose of the study and phenomenon under
questions and better understand a concept of study. Some other examples of case study re-
interest, often as a prelude to understand a search design are presented in Box 4.9.
large study of the issue. Case study data are collected by observation or
 Explanatory case studies can be used as pilot by the personal interview method. Generally,
studies in the development of a conceptual analysis in case study design does not involve so-
framework, which can serve to organize the phisticated quantitative and statistical techniques.
data collection in a large study. Explanatory In case studies, the stress is more on the qualitative
case studies give new or refined meaning to analysis of collected data.
previously explored concepts, while the re-
searcher holds extent to theories and prior
assumptions in abeyance. TRIANGULATION (MIXED METHODS)
Each type of case study uses multiple data A number of terms are used interchangeably in the
sources and data collection methods to literature to describe a combination of methods in
obtain a broad view of the phenomenon. the same study. Some of the common ones are
For example, a researcher is interested in mixed methods, multiple method, methodological
104 Research Methodology and Biostatistics

pluralism and triangulation. The letter is often an are very familiar with the concept of the
abused term to describe any study that uses more family as the ‘unit of care’; person triangula-
than one method or a study, which combines tion is a similar concept. The patient experi-
qualitative and quantitative research methods. encing a phenomenon might have a very dif-
Therefore, triangulation is one form of combining ferent perspective on that phenomenon than
more than one method in the same study that is that of his/her spouse, which may be different
primarily qualitative and/or quantitative. The ben- from the perspectives of sibling, parents,
efit of convergence of qualitative and quantitative caregivers or friends. Collecting data from
methods can include increased confidence in study many of the parties, therefore, widens the
data and expansion of understanding of a phe- perspective on the phenomenon of interest.
nomenon, the revelation of unique data and inte- l Methods triangulation: It is the combination

gration of theories. of qualitative and quantitative data collection

Denzin (1989) defines triangulation as the techniques, which allows the researcher to fully
combination of two or more theories, data understand a phenomenon and link dimen-
sources, methods or investigators in the study sions of the phenomenon. It also allows the
of a single phenomenon. Triangulation is a pro- researcher to sort through competing ideas
cess where the research process along with the about factors that might explain the phenome-
data is analysed from several different direc- non of interest. For example, a researcher is in-
tions. This process is employed to decrease the terested to assess the burn out among health
potential for biases within the research process care professionals working in critical care units
and the analysis of the data while validating the of AIIMS, New Delhi. In this instance, the re-
results obtained. searcher may use the structured quantitative
Triangulation might be used in four basic ways data measurement tool(s) as well as open-
in a study: (a) data triangulation, (b) methods ended qualitative data measurement tool(s) so
triangulation, (c) researcher triangulation and that the breadth and depth of evidence may be
(d) theory triangulation. The most commonly obtained.
used triangulation is the methods triangulation as The triangulation of methods could be of
in recent research studies. two main types, i.e. concurrent and sequential
l Data triangulation: This involves the use of type. In concurrent type, quantitative and quali-
multiple data sources to validate a conclusion. tative methods are equally prioritized, inte-
It includes at least three types – time, space grated and implemented. However, in sequential
and person triangulation. In time triangula- type, qualitative or quantitative methods follow
tion, the researcher collects data on the same one another and one of the methods dominates
phenomenon or about the same informant at while prioritizing, integrating and implement-
different times. The interval between data ing the research methods.
collection points may be hours or months. l Researcher triangulation: In this, several ob-

The goal of time triangulation is to observe servers are used rather than a single observer.
how the informant or phenomenon changes More than one researcher may collect data or
over time. In space triangulation, the re- analyse and interpret data as part of the same
searcher collects data on the same phenome- study. This technique is often used in qualita-
non at different sites, which allows for assess- tive studies as a way to ensure the validity of the
ment of consistency of data across sites. research. The goal of researcher triangulation is
Finally, person triangulation involves data to minimize the occurrence of researcher bias
collection from different types of people or and to enhance depth and breadth of data
informants. Public health care professionals analysis.
 Research Designs 105

l Theory triangulation: In this, the researcher for the purpose of cross-validation can raise
uses more than one theory or hypothesis as more questions than solutions. Therefore, use of
data are analysed and interpreted. This strat- triangulation methods for richness of data and
egy allows for flexibility in analysis and inter- to enhance the breadth and depth of evidence
pretation. Theory triangulation also helps the may be effectively justified. Combining methods
researcher to avoid allowing preconceived and approaches should be a well thought-out
ideas to guide data analysis and interpretation process backed by a strong rational discussion,
and the resulting conceptualization of the the purpose and benefits of such decision,
phenomenon. and the choice of methods. Sale et al. (2002)
Maggs-Rapport (2000) mentions that in a believed that mixed-method research is now be-
multimethod research when multiple findings ing adapted uncritically by a new generation of
are produced it becomes very difficult to make researchers who have overlooked the underlying
decisions about the validity of one set of find- assumptions behind the qualitative–quantitative
ings over others. Therefore, use of triangulation debate.

MULTIPLE CHOICE QUESTIONS

1. What is a research design? 4. ‘Internal validity’ refers to
(a) A way of conducting research that is not (a) Whether or not there is really a causal relation-
grounded in theory ship between two variables
(b) The choice between using qualitative or quanti- (b) Whether or not the findings are relevant to the
tative methods participants’ everyday lives
(c) The style in which you present your research (c) The degree to which the researcher feels that
findings, e.g. a graph this was a worthwhile project
(d) A framework for every stage of the collection (d) How accurately the measurements represent
and analysis of data underlying concepts

2. What is a cross-sectional design? 5. Which of the following items distinguishes true ex-
(a) A study of one particular section of society, e.g. perimental research from quasi-experimental research?
the middle classes (a) Size of sample
(b) One that is devised when the researcher is in a (b) The use of a nonprobability sample
bad mood (c) Random assignment of subjects to groups
(c) The collection of data from more than one case (d) The introduction of an experimental treatment
at one moment in time
(d) A comparison of two or more variables over a 6. Which of the following designs control for the sen-
long period of time sitization of subjects to a pretest?
(a) Pretest–post-test control group design
3. Survey research is cross-sectional and therefore (b) Solomon four-group design
(a) High in reliability but low in internal validity (c) One-shot case study
(b) High in internal validity but low in reliability (d) Time-series design
(c) High in ecological validity but low in external
validity 7. Ex post facto studies are also called
(d) None of the above (a) Exploratory
(b) Correlational
(c) Longitudinal
(d) Descriptive
106 Research Methodology and Biostatistics

8. Consider this title: ‘The Live Experience of Surviving 10. The group that does not receive the experimental
a Tsunami’. This title would indicate which of the treatment condition is called.
following types of qualitative research? (a) Experimental group
(a) Ethnographic (b) Control group
(b) Phenomenological (c) Treatment group
(c) Action (d) Independent group
(d) Historical

9. When a researcher starts with the dependent vari-

able and moves backwards, it is called
(a) Predictive research
(b) Retrospective research
(c) Exploratory research
(d) Descriptive research

Answers of Multiple Choice Questions

1. d; 2. c; 3. a; 4. a; 5. c; 6. b; 7. b; 8. b; 9. b; 10. b.

REFERENCES
Bayley, E. M., Maclean, S.L., Desy, P., & McMahon, M. (2004). Glaser, B. G., & Strauss, A. (1967). The discovery grounded
ENAs Delphi study on national research priorities for emer- theory: strategies for qualitative inquiry. Aldin, Chicago.
gency nurses in the United States. Journal of Emergency Nursing, Ingersoll, G. L., McIntosh, E., & Williams, M. (2000). Nurse-sensitive
30, 12–21. outcome of advanced practice. Journal of Advanced Nursing,
Beck C. T. (2005). Replication strategies for nursing research. In 32(5), 1272–81.
Burns, N., Grover, S.K (Ed.), The practices of nursing research: Jacelon, C. S., & O’Dell, K. K. (2005). Case and grounded theory as
conduct, critique and utilization (5th ed.). Philadelphia, PA: qualitative research methods. Urology Nursing, 25, 49–52.
Saunders. Johnson, R. L. (2005). Gender differences in health-promoting life-
Burns, N., & Grove, S. K. (2005). The practice of nursing research: style of African Americans. Public Health Nursing, 22(2), 130–37.
conduct, critique and utilization (5th ed.). St. Louis: Elsevier Kader, P. (1997). Nursing research: principles, process and
Saunders. issues. New York: Palgrave Macmillan.
Burns, N., & Grove, S. K. (2008). Understanding nursing research: Macnee, C. L. (2004). Understanding nursing research: reading and
building and evidence-based practices (4th ed.). Noida: Elsevier. using research in practice (1st ed.). Philadelphia, PA: Lippincott.
Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi- Maggs-Rapport F. (2000). Combining methodological approaches
experimental designs for research on teaching. In N. L. Gage in research: Ethnography and interpretive phenomenology.
(Ed.), Handbook of research on teaching (pp. 171–246). Journal of Advanced Nursing;31(1):219–25.
Chicago, IL: Rand McNally. Martin, P. Y., & Turner, B. A. (1986). Grounded theory and orga-
Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi- nizational research. The Journal of Applied Behavioral Science,
experimental designs for research. Boston: Houghton Mifflin 22(2), 141–157.
Company. Orem, D. E. (2001). Nursing concept of practice (6th ed.). Philadelphia,
Christy T. E (1975). The methodology of historical research. PA: Mosby.
Nursing Research, 24(3):189–192. Polit, F. D., & Beck, C. T. (2008). Nursing research: generating and as-
Denzin N. K. (1989). The Research Act. A Theoretical Instruction sessing evidence for nursing practices (8th ed.) Philadelphia, PA:
to Sociological Methods. (3rd ed.). Englewood Cliffs NJ, Pren- Lippincott.
tice Hall. Polit, F. D., & Hungler, B. P. (1995). Nursing research: principles
Donalek, J. G. (2004). Demystifying nursing research: phenome- and methods (5th ed.). Philadelphia, PA: Lippincott.
nology as a qualitative research method. Urology Nursing, 24, Sharma, S. K. (2012). A Study to assess the effectiveness of the
516–17. Anganwari Scheme on the nutritional level of beneficiary chil-
Donohue, R. (2003). Nurse practitioner–client interaction as re- dren in selected villages of district Sikar, Rajasthan. Nursing &
sources exchange in a women’s health clinic. Journal of Midwifery Research Journal, 8(2), 269–278. July 2012.
Advanced Nursing, 12(5), 717–25. Sharma, S. K., & Choudhary, A. (2008). Substance abuse related
Dormire, S. L., & Yarandi, H. (2001). Predictors of risk for adoles- high risk factors among traumatic head injury patients.
cent childbearing. Applied Nursing Research, 14(2), 81–86. Nursing & Midwifery Research Journal, 4(3), 96–101.
 Research Designs 107

Sharma, S. K., & Kaur, K. (2009). A descriptive study on dietary Streubert, H. J., Carpenter, & D. R. (2002). Qualitative research
pattern and biophysical profile among nursing students. in nursing: Advancing the humanistic imperative (3rd ed.).
Nursing & Midwifery Research Journal, 5(2), 80–88. Philadelphia, PA: Lippincott.
Sharma, S. K., Gupta, J., Walia, I., & Sen, R. (2005). Prevalence of Ulseter, A. A., & Antle, B. J. (2005). In the darkness there can be
pin site infection among patients with external skeletal fixation. light: a family’s adaptation to a child’s blindness. Journal of
Nursing & Midwifery Research Journal, 1(2), 124–30. Visual Impairment & Blindness, 99(4), 209–18.
Sharma, S. K., Kaur, K., & Garg, R. (2007). Factors affecting bowel Velsor-Friendrich, B., Pigott, T., & Louloudes, A. (2004). The ef-
movements in critically ill patients. Nursing & Midwifery fect of a school based intervention on the self-care and health
Research Journal, 3(2), 71–78. of African–American inner city school children with asthma.
Sharma, S. K., Kaur, K., & Kaur, M. (2006). Awareness of HIV/ Journal of Pediatric Nursing, 19(4), 247–56.
AIDS among newly admitted nursing students. Nursing & Williams, S. A., & Schreier, A. M. (2005). The role of education in
Midwifery Research Journal, 2(1), 32–38. managing fatigue, anxiety and sleep disorder in women under-
Slade, M., Kuipers, E., & Priebe, S. (2002). Mental health services going chemotherapy for breast cancer. Applied Nursing
research methodology. International Review of Psychiatry, 14(1), Research, 18(3), 138–47.
12–18.
 Randomized Controlled

CHAPTER OUTLINE
Trials 5
INTRODUCTION USE OF PLACEBO IN RANDOMIZED CONTROLLED TRIALS

MEANING OF RANDOMIZED CONTROLLED TRIALS DIFFICULTIES IN CONDUCTING RANDOMIZED

CONTROLLED TRIALS
HISTORY OF RANDOMIZED CONTROLLED TRIALS
ADVANTAGES OF RANDOMIZED CONTROLLED TRIALS
CHARACTERISTICS OF RANDOMIZED CONTROLLED TRIALS
DISADVANTAGES OF RANDOMIZED CONTROLLED TRIALS
RANDOMIZATION IN RANDOMIZED CONTROLLED TRIALS
MULTIPLE CHOICE QUESTIONS
TYPES OF RANDOMIZED CONTROLLED TRIALS

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Identify the advantages and disadvantages of random-
• Understand the concept of randomized controlled trials. ized controlled trials.
• Determine the criteria and types of randomization. • Describe uncontrolled, laboratory and field trials.
• Classify the types of randomized controlled trials.
• Appraise the difficulties incurred in conducting
randomized controlled trials.

KEY TERMS

• Randomized Controlled trials 109 • Unequal randomization 110 • Controlled trials 112
• Randomization 109 • Preclinical trials 110 • Open trials 112
• Simple randomization 110 • Clinical trials 110 • Blind trials 112
• Block randomization 110 • Field trials 111 • Placebos 112
• Stratified randomization 110 • Uncontrolled trials 111

pointing out that there is a scarcity of evidence

INTRODUCTION generated through RCTs. There are several reasons
Health care professions are in increasing need of for lack of RCTs; primarily because most medical
randomized controlled trials (RCTs) to justify researchers lack the experience and exposure of
their practices on sound evidence against day by conducting RCTs and secondly, stringent rules
day increasing number and type of diseases. RCTs and regulation for conduction of RCTs make it a
are considered as the gold standard to generate less preferred design. The present chapter attempts
high-level evidence. There are a number of authors to demystify some basic concepts of RCTs.
 Randomized Controlled Trials 109

relationship exists between an intervention and an

MEANING OF RANDOMIZED
outcome. For this reason, they are often used in
CONTROLLED TRIALS medicine, especially when new or alternative inter-
A randomized controlled trial is a study in which ventions are being compared to usual standard
subjects are allocated at random to receive one of care. Also, they are frequently used to assess the
several clinical interventions. One of these inter- cost-effectiveness of an intervention.
ventions is the standard of comparison or control.
The control may be a standard practice, a placebo
HISTORY OF RANDOMIZED
or no intervention at all.
In other words, RCTs are quantitative, com- CONTROLLED TRIALS
parative and controlled experiments in which a The chronicle of clinical trials can be assumed to
group of investigators study two or more inter- date back to approximately 600 BC when Daniel
ventions in a series of participants allocated ran- of Judah supervised what is perhaps the oldest
domly to each intervention group. recorded clinical trial. Comparison was made
l Experiments: Investigators can influence num- concerning the health effects of a vegetarian diet
ber, type and regime of interventions. with a royal Babylonian diet for a 10-day span.
l Quantitative: Measure events rather than inter- Credit for the modern randomized trial usually
pret them in a natural setting. goes to Sir Austin Hill. The Medical Research
l Comparative: Compare two or more interven- Council trials on streptomycin for the treatment
tions or an intervention is compared with a of pulmonary tuberculosis are considered land-
placebo or no intervention. mark trials that ushered in a new era of medicine.
l Randomization: Random allocation of subjects Since that time, the randomized controlled trial
to intervention groups and the control group procedure has been accepted to a greater degree
(comparison group) so that all participants and there has been a significant growth in the
have equal chance of being allocated to each number of randomized controlled trials.
intervention and comparison group.
l The RCT is a random allocation of subjects to
CHARACTERISTICS OF RANDOMIZED
intervention groups so that all participants have
equal chance of being allocated to each inter- CONTROLLED TRIALS
vention group. These are the most straightfor- l The main feature of RCT is randomization,
ward of studies to design and interpret. which is the most crucial step in RCTs. The
l RCTs are often considered to be the ‘gold stan- main purpose of this is to stop selection bias by
dard’ of clinical and epidemiological studies. distributing the features of patients which might
This is because, if they are conducted properly, influence the result randomly between the
it is often possible to be fairly sure that the re- groups so that any discrepancy in the result can
sults are correct, at least for the type of patients be detailed only by the treatment.
who took part in the study. l Randomization is an effective method to bal-
l The intervention to be tested is called the ex- ance confounding factors between treatment
periment and the group receiving intervention groups and can eliminate the influence of con-
is called the experimental group. The other in- founding variables.
tervention is regarded as the standard of com- l It enables the investigator to assess whether the
parison or control and the group of participants intervention itself, as opposed to other factors,
which receives this standard is called the control caused the observed outcomes.
group. The control can be a conventional prac- l From a well-defined point, time zero or baseline
tice, a placebo or no intervention at all. for the study, each participant must be followed
l RCTs are considered to be the most rigorous scien- till the end point of the study (expected out-
tific method of determining whether a cause–effect come as defined in the study).
110 Research Methodology and Biostatistics

l There is experimental group(s) and a control (e.g. early patients are sicker than later), early
group, which are compared for intended out- imbalances cannot be corrected. Block ran-
come variables. domization is often used to fix this issue.
l Outcome variable must be estimated for all the  Example: Consider two treatments of A, B

enrolled participants. If any participant leaves For the block sizes 2 3 2 5 4

the study then it also must be recorded along Some of the probable allocations for treat-
with most probable reason. It is very important ment within every block are as follows:
to note the compliance of the newer interven- (1) AABB, (2) BBAA, (3) ABAB, (4) BABA,
tion. Sometimes the newer intervention could (5) ABBA, (6) BAAB
be very effective but if the compliance rate is The block size depends on the total number
very poor, then it cannot be an effective method. of treatments.
Hence, it is important to keep records of non-  The block size should be at least two times the

compliance, loss to follow-up and side effects. number of treatments.

l Stratified randomization
 An RCT may not be considered valid if it is
RANDOMIZATION IN RANDOMIZED
not well-balanced across prognostic factors.
CONTROLLED TRIALS  For example, age group: ,40, 41–60, . 60;

Randomization is the random allocation of sub- sex: M, F.

jects into experimental and control groups. Ran-
Total number of strata 5 3 (three of age group)
domization is necessary to guard against any use of
3 2 (two of particular sex) 5 6
judgment or systematic arrangements leading to
one treatment getting ‘better conditions’ to succeed  Stratification can balance subjects on baseline
and also to provide a basis for standard methods of covariates and tends to produce comparable
statistical analysis, such as the significance test. groups with respect to certain characteristics
(e.g. gender, age, race, disease severity); thus,
CRITERIA FOR RANDOMIZATION it produces valid statistical tests.
There are three main criteria for randomization. l Unequal randomization
l Unpredictability: Every participant has got simi-  Most randomized trials allot same number of

lar options of getting any of the interventions. patients to control and experimental groups,
l Balance: Groups are alike in all aspects, i.e. treat- as it maximizes statistical power for a given
ment groups are of a same constitution and size. total sample size. Hence, this is the most sta-
l Simplicity: It is easy for the investigator to tistically efficient randomization ratio.
implement.  This might not be economically/practically/

ethically feasible and efficient.

TYPES OF RANDOMIZATION
l Simple randomization
 This method is equivalent to tossing a coin for
TYPES OF RANDOMIZED
each subject that enters a trial, such as heads 5 CONTROLLED TRIALS
active, tails 5 placebo. l Types on the basis of purpose:
 Treatment assignment is totally unpredictable  Preclinical/ pharmacological and toxicity/

as it is very easy and simple to implement. laboratory trials: These are conducted on
 It can get imbalanced in treatment assign- nonhuman subjects for examining the effi-
ments, especially in smaller trials. Imbalanced cacy, toxicity and pharmacokinetic informa-
randomization reduces statistical power. tion of a drug. Preclinical trials are conducted
l Block randomization in more controlled conditions.
 Simple randomization does not guarantee  Clinical trials: These are done on human sub-

balance in numbers during trial. Especially, if jects beginning from the phase 0 trials, where
the patient characteristics change with time pharmacokinetics of the drugs, especially the
 Randomized Controlled Trials 111

oral bioavailability and half-life of the drugs, community trials. The reason for this differ-
are examined only on few target volunteers. entiation is that many interventions can be ap-
Generally, phase 0 is missed to jump directly plied only at the community level and it is
on to phase I. Clinical trials have the follow- difficult to apply them at the individual level.
ing phases: For example, health education through hoard-
- Phase I: The drug is tested on healthy volun- ing, banners, radio, TV programmes will ad-
teers in subtherapeutic dose with ascending dress the whole community together. Efficacy
doses to determine the dose ranging and to of changes in the quality of water supply and
know whether the drug is safe to check for fortification of salt are other examples, which
efficacy. can only be tested through community trials.
- Phase II: During this phase of the trial, the The biggest challenge of conducting a field trial
drug is tested on patients with a specific is its resource intensiveness. It requires enor-
disease to assess the efficacy (expected effect mous time, money and manpower. Therefore,
of drug/intervention under ideal circum- even though they are the best way of testing
stances) and safety of the drug. hypothesis, only limited numbers of studies are
- Phase III: During this phase of trial, the conducted.
drug is tested on patients with a specific l Types on the basis of study design: One classifi-
disease to assess the effectiveness (beneficial cation of RCTs is on the basis of study design.
effects of drug/intervention under real clin- From the most to the least common in medical
ical situations) and safety. literature, the major categories of RCT study
- Phase IV: This is postmarketing surveillance designs are as follows:
to monitor the adverse effects/reactions of  Parallel group: As every participant is allotted

the drug. to a group at random, all the participants in the

 Field trials: Field trials are studies with a large group draw (or do not get) an intervention.
sample size, most of the time conducted in  Crossover: Over time, every participant either

their natural setting. Usually these trials are receives or does not receive an intervention
conducted with the objective of evaluation of randomly.
preventive methods/strategies/vaccine/treat-  Split body: Different parts of the body of every

ment modalities, etc. Field trials usually involve participant (e.g. the left and right sides of the
disease-free subjects, or patients with diseases face) are randomized to either receive or not
of public health importance with mass impli- receive an intervention.
cations. A field trial involves evaluation of  Cluster: Some pre-existing groups of partici-

whether an agent or procedure reduces the risk pants (e.g. villages, schools) are selected at
of developing disease among those free from random to either receive or not receive an
that condition at the time of enrolment. It can intervention.
also assess efficacy of the blanket treatment in  Factorial: Every participant is assigned ran-

cases of diseases gripping masses. As these tri- domly to a group that draws a specific consoli-
als commonly involve healthy rather than dis- dation of interventions or noninterventions
eased people, they tend to be logistically more (e.g. group 1 draws vitamin X and vitamin Y,
difficult to carry out than clinical trials. They group 2 gets vitamin X and placebo Y, group 3
generally have to be conducted in the ‘field’ gets placebo X and vitamin Y and group 4 gets
rather than in hospitals or clinics. Field trials placebo X and placebo Y).
can be carried out among individuals or groups l Types based on the control: Based on the con-
of people. In the first case, the unit of interven- trol, trials could be of the following types:
tion is the individual, whereas in the second  Uncontrolled trials: Where there is no control

case, intervention is applied to the whole group to compare with the experimental
group. Field trials in which the whole com- group but the same group or individual is
munity is the unit of intervention are called compared before and after the intervention.
112 Research Methodology and Biostatistics

 Controlled trials: Where there is a control group statistically significant approach. Moreover, some
to compare with the experimental group, the RCTs are noninferiority trials and ‘determine if a
control could be of the following types: new treatment is no worse than a reference treat-
- Placebo concurrent control: The control ment’. Other RCTs are equivalence trials in which
group receives an identical appearing treat- the hypothesis is that two interventions are indis-
ment to compare the outcome measure- tinguishable from each other.
ments with the experimental group. l Other miscellaneous types are as follows:
- No-treatment concurrent control: The con-  Open trial: In an open trial, also called an

trol group receives no treatment; thus, open-label trial, the researcher knows the full
blinding is not possible in this type of trials. details of the treatment and so does the patient.
- Dose response concurrent control: The con- These trials are open to challenge for bias, and
trol group receives a different dose from the do nothing to reduce the placebo effect.
experiential group.  Blind trial or blind experiment: Blinding is a

- Active (positive) concurrent control: The method to reduce the bias in the outcome of
control group receives the existing treat- the study because of being associated with the
ment and is compared with the newer treat- study at the level of study participant, investi-
ment given to the experimental group. gator/clinician or study analyst. Blinding
- External control (including historical control): could be done at different levels.
When the control is not available concur- - Single-blind trial: In a single-blind trial, the
rently with the experimental group, then a researcher knows the details of the treat-
group of similar patients treated previously ment but the patient does not. Because the
(historical control) or treated in another patient does not know which treatment is
centre/ setting is considered as control. being administered (the new treatment or
- Multiple control: When more than one con- another treatment), there might be no pla-
trol are considered in the study including cebo effect. Only the patient is unaware of
the active and the placebo. the treatment.
l Types on the basis of interest (efficacy versus - Double-blind trial: This trial is a clinical trial
effectiveness): One can classify RCTs as ‘prag- design wherein neither the study staff nor
matic’ or ‘explanatory’. With highly selected the participating individuals have the
participants and under highly controlled con- knowledge of the participants receiving the
ditions, explanatory RCTs measure efficacy in a treatment and those that are getting a pla-
research setting. On the other hand, pragmatic cebo. This kind of double-blind trial is pre-
RCTs measure effectiveness in everyday prac- ferred, as they tend to give the most accurate
tice with comparatively unselected participants results.
and under flexible conditions. Thus, ‘decisions - Triple-blind trial: The patient, investigator
about practice’ can be informed by pragmatic and data analyst are blind. The statistician
RCTs. can only be partially blinded since he/she
l Types on the basis of hypothesis (superiority has to know the patients that are in the
versus noninferiority versus equivalence): There same treatment group.
is another classification method in which RCTs
categorize them as superiority trials, noninferior-
USE OF PLACEBOS IN RANDOMIZED
ity trials and equivalence trials, which differ in
reporting and methodology. As most RCTs CONTROLLED TRIALS
are superiority trials, hence one intervention is It is another strategy to control extraneous vari-
hypothesized to be superior to the other in a ables. The idea of receiving a new treatment can
 Randomized Controlled Trials 113

itself make some people feel better, called the Examples:

placebo effect. If some subjects in an experiment - Treatments where a convincing placebo
have high expectations of a new drug or other treatment would be too dangerous to be
form of treatment being tested, this can affect ethically acceptable (e.g. surgery).
the results of the study. To overcome the possible - Treatments wherein some sort of active
suggestive effect of the new intervention, a participation of the patient is required (e.g.
placebo – a substance that has no pharmacologi- dieting and physical therapy).
cal or therapeutic property – can be administered
along with the new drug or treatment. It is
ADVANTAGES OF RANDOMIZED
assumed that in all medical interventions, there
are placebo effects to some extent, including CONTROLLED TRIALS
the effect of the therapeutic setting, perception l Randomized experiment designs completely re-
of the therapist by the patient and credibility of move any accusations of conscious or subcon-
medication itself (colour, size, shape, test, etc.). scious bias from the researcher and practically
Therefore, generally a placebo intervention is guarantee external validity.
designed to overcome this placebo effect, for ex- l Randomized controlled trials completely remove
ample, the use of placebo in an RCT comparing effect of extraneous variables.
the effectiveness of amethocaine gel with a similar l They are easier to blind/mask than observa-
resembling placebo gel in the management of pro- tional studies.
cedural pain in neonates. However, it is often not l Their results can be analysed with well-known
possible to design a placebo to match many treat- statistical tools.
ment interventions, especially psychosocial inter- l Populations of participating individuals are
ventions, such as counselling. For this reason, the clearly identified.
use of placebo is uncommon in RCTs conducted
in allied health sciences disciplines; most of the
DISADVANTAGES OF RANDOMIZED
RCTs are conducted to test best-known (usual)
intervention against new interventions. CONTROLLED TRIALS
l Volunteer biases: The population that partici-
pates may not be representative of the whole.
DIFFICULTIES IN CONDUCTING l RCTs may be limited by ethical considerations.
RANDOMIZED CONTROLLED TRIALS (For example, if patients are given an interven-
l Outside pressure: A major difficulty in dealing tion thought to be inferior to other treatment,
with trial results comes from commercial, the trial could be considered unethical.)
political and/or academic pressure. l Patient recruitment may be difficult in some
l Expensive: Most trials are expensive to run. RCTs. (Once an intervention has become wide-
l Time consuming: RCTs are time consuming in spread and well known to the public, potential
planning and implementation. study participants may be unwilling to be in a
l Chances of statistical error: The result that study looking at alternative treatments.)
comes out of the trial may be false–negative or l Some potential study subjects do not like the idea
false–positive, which may vary from the hy- that they will not receive any treatment, as is the
pothesis assumed. case with some trials with a placebo–control arm.
l Blinding: Ideally, studies should be blinded by l Costly and time consuming.
giving placebo treatments to avoid bias caused l Hawthorne effect – This is the alteration of
by placebo effects. However, for some treat- behaviour by the subjects of a study due to their
ments, placebos are not possible. awareness of being observed.
114 Research Methodology and Biostatistics

MULTIPLE CHOICE QUESTIONS

1. Randomized controlled trials (RCTs) are 7. What can be observed under phase II clinical trials
(a) Not required to be based on the concept of of randomization?
equipoise (a) Drug is tested on healthy volunteers in sub-
(b) Always have a control arm that uses placebo therapeutic dose with ascending doses to de-
(c) Considered to be the ‘gold standard’ for deter- termine the dose ranging and to know whether
mining the efficacy and safety in clinical re- drug is safe to check for efficacy.
search (b) The drug is tested on patient with specific dis-
(d) Always double-blinded ease to assess the effectiveness (beneficial
effects of drug/intervention under real clinical
2. A true cause and effect relationship can be identified situations) and safety
through (c) The drug is tested on patient with specific
(a) Survey disease to assess the efficacy (expected effect of
(b) Cross-sectional studies drug/intervention under ideal circumstances)
(c) Case studies and safety of the drug
(d) Randomized controlled trials (d) Postmarketing surveillance is done to monitor
the adverse effects/reactions of the drug
3. A randomized controlled trial is characterized by
(a) Randomization, manipulation, control 8. In a randomized controlled trial, the following type
(b) Randomization, control only of bias is reduced by randomization:
(c) Manipulation, control only (a) Ascertainment bias
(d) Randomization, manipulation only (b) Selection bias
(c) Recall bias
4. The major purpose of random assignment in a (d) Bias in handling dropouts
clinical trial is to
(a) Help ensure that study subjects are representa- 9. To avoid experimenter bias, neither the experi-
tive of the general population menter nor the participant is aware of which group
(b) Help ensure double blinding the participant is in; this is known as
(c) Help ensure that the study groups are compa- (a) Null hypothesis
rable on baseline characteristics (b) Random assignment
(d) Facilitate measurement of the outcome variables (c) Variable manipulation
(d) Double blind
5. All of the following are true in a randomized controlled
trial (RCT), except: 10. What is the name of the effect that occurs when an
(a) Baseline characteristics of intervention and experimental group gets better simply because they
control groups should be similar are being given a pill and this leads them to expect
(b) Investigator’s bias is minimized by double to get better?
blinding (a) The domino effect
(c) The sample size required depends on the hy- (b) The butterfly effect
pothesis (c) The placebo effect
(d) The dropouts from the trial should be excluded (d) The expectancy effect
from the analysis
11. All the following are potential benefits of a ran-
6. Randomized controlled trial is considered the most domized controlled trial, except:
valid of the epidemiologic study designs because of (a) Randomized controlled trials completely
what aspect of design? remove effect of extraneous variables
(a) Randomization (b) Self-selection for particular treatment is eliminated
(b) Observation (c) External validity of the study is increased
(c) Manipulation (d) Assignment of the next subject cannot be
(d) Control predicted
 Randomized Controlled Trials 115

12. Randomization is the best approach in designing 14. Which method of controlling confounding extrane-
a clinical trial in order to ous variables takes precedence over all other methods?
(a) Achieve predictability (a) Matching individual participants
(b) Achieve unpredictability (b) Holding extraneous variables
(c) Achieve blinding (c) Counterbalancing
(d) Limit confounding (d) Randomly assigning research participants in
group
13. The group that receives the experimental treatment
condition is the 15. Which of the following is not true about field trials?
(a) Experimental group (a) Usually involves healthy rather than diseased
(b) Control group people
(c) Participant group (b) Always conducted in hospital or clinic
(d) Independent group (c) Tends to be logistically more difficult to carry
out than clinical trials
(d) It can be carried out among individuals or
groups of people
Answers of Multiple Choice Questions
1. c; 2. d; 3. a; 4. c; 5. c; 6. a; 7. c; 8. b; 9. d; 10. c; 11. c; 12. b; 13. a; 14. d; 15. b.

REFERENCES
Chalmers, T. C., Smith, H. Jr., Blackburn, B., Silverman, B., Schro- Phase 0 trials: a platform for drug development? (2009). Lancet
eder, B., Reitman D., et al. (1981). A method for assessing the 374(9685), 176.
quality of a randomized control trial. Controlled Clinical Trials, Piaggio, G., Elbourne, D. R., Altman, D. G., Pocock, S. J., Evans,
2(1), 31–49. S. J.; & CONSORT Group. (2006). Reporting of noninferiority
DeMets, D., Friedman, L., & Furberg, C. (2010). Fundamentals of and equivalence randomized trials: an extension of the
clinical trials (4th ed.). New York: Springer. CONSORT statement. Journal of the American Medical Associa-
Dirnagl, U., Hakim, A., Macleod, M., Fisher, M., Howells, D., Alan tion, 295(10), 1152–1160.
S. M., et al. (2013). A concerted appeal for international coop- Ranjith, G. (2005). Interferon-a-induced depression: when a ran-
eration in preclinical stroke research. Stroke, 44, 1754–1760. domized trial is not a randomized controlled trial. Psychotherapy
Haahr, M. T., & Hróbjartsson, A. (2006). Who is blinded in ran- Psychosomatics, 74(6), 387.
domized clinical trials? A study of 200 trials and a survey of Rosenberger, W. F., & Lachin, J. M. (1993). The use of response-
authors. Clinical Trials, 3(4), 360–365. adaptive designs in clinical trials. Controlled Clinical Trials,
Lachin, J. M. (1988). Statistical properties of randomization in 14(6), 471–484.
clinical trials. Controlled Clinical Trials, 9(4), 289–311. Schmidt, S. (2009). Shall we really do it again? The powerful con-
Landis, S.C., Amara, S. G., Asadullah, K., Austin, C. P., Blumen- cept of replication is neglected in the social sciences.
stein, R., Bradley E. W., et al. (2012). A call for transparent Review of General Psychology, 13, 90–100.
reporting to optimize the predictive value of preclinical Schulz, K. F., & Grimes, D. A. (2002). Generation of allocation
research. Nature, 490, 187–191. sequences in randomised trials: chance, not choice. Lancet,
Mathieu, S., Boutron, I., Moher, D., Altman, D. G., & Ravaud, P. 359(9305), 515–519
(2009). Comparison of registered and published primary out- Shamoo, A. E. (2008). The myth of equipoise in phase 1 clinical
comes in randomized controlled trials. Journal of the American trials. Medscape Journal of Medicine, 10(11), 254.
Medical Association, 302(9), 977–984. Willmann, R., De Luca, A., Benatar, M., Grounds, M., Dubach,
Norfleet, E., & Gad S. C. (2009). Phase I clinical trials. In J., Raymackers J. M., et al. (2012). Enhancing translation: guide-
S. C. Gad (Ed.), Clinical trials handbook (p. 247). Hoboken, lines for standard pre-clinical experiments in mdx mice.
NJ: Wiley. Neuromuscular Disorders, 22, 43–49.
Sampling and Sample Size
Determination
CHAPTER OUTLINE
6
INTRODUCTION NONPROBABILITY SAMPLING

TERMINOLOGIES USED IN SAMPLING PROBABILITY SAMPLING

PURPOSE OF SAMPLING SAMPLE SIZE ESTIMATION

CHARACTERISTICS OF A GOOD SAMPLE PROBABILITY AND SAMPLING ERRORS

SAMPLING PROCESS PROBLEMS OF SAMPLING

FACTORS INFLUENCING A SAMPLING PROCESS MULTIPLE CHOICE QUESTIONS

TYPES OF SAMPLING TECHNIQUES

LEARNING OBJECTIVES

• Define the basic terms related to population, sample • Distinguish between probability and nonprobability
and sampling. sampling strategies.
• Determine the purposes and process of sampling. • Discuss the approaches to determine sample size.
• Discuss the elements and factors influencing the • Enumerate the errors, bias and problems in sampling.
sampling process.
• Describe the different types of probability and
nonprobability sampling techniques.

KEY TERMS

• Accessible population 117 • Purposive sampling 128 • Sampling frame 117

• Cluster/multistage sampling 126 • Quota sampling 131 • Sampling plan 118
• Consecutive sampling 130 • Representative sample 118 • Sequential sampling 127
• Convenience sampling 130 • Sample 117 • Simple random sampling 122
• Element 117 • Sample size 120 • Snowball sampling 132
• Nonprobability sampling 127 • Sampling 117 • Stratified random sampling 123
• Population 117 • Sampling bias 118 • Systematic random sampling 125
• Probability sampling 122 • Sampling error 118 • Target population 117

in day-to-day life, to arrive at an understanding

INTRODUCTION about societies, opinions or situations. For ex-
Sampling is a process of selecting representative ample, a grain merchant takes a handful of
units from an entire population of a study. grains to get an idea about an entire bag of
Sampling is not a new development, but in re- grains; or for instance, a chef tests a few morsels
cent times, it is used by people in all fields, even of a cooked dish to determine if the taste and
 Sampling and Sample Size Determination 117

flavours are right; similarly a pathologist takes l Accessible population: It is the aggregate of
only a few millilitres of blood to measure the cases that conform to the designated criteria
haemoglobin levels of an individual. and are also accessible as subjects for a study.
Similarly, in research studies, it is not always For example, a researcher is conducting a study
possible to study an entire population; therefore, on the resident doctors working in All India
the researcher draws a representative part of a Institute of Medical Sciences, New Delhi. In this
population through a sampling process. In other case, the population for this study is all the resi-
words, sampling is the selection of some part of an dent doctors working in AIIMS, New Delhi, but
aggregate or a whole on the basis of which, judge- some of them may be on leave and may not be
ments or inferences about the aggregate or mass accessible for research study. Therefore, the ac-
are made. It is a process of obtaining information cessible population for this study will be resident
regarding a phenomenon about an entire popula- doctors who meet the designated criteria and
tion by examining a part of it. who are also available for the research study.
l Sampling: Sampling is the process of selecting a
representative segment of the population under
TERMINOLOGIES USED IN SAMPLING study.
Some of the main terms used in a sampling pro- l Sample: Sample may be defined as the represen-
cess are as follows (Fig. 6.1): tative unit of a target population, which is to be
l Population: Population is the aggregation of all worked upon by researchers during their study.
the units in which a researcher is interested. In In other words, a sample consists of a subset of
other words, population is the set of people or units which comprise the population selected
entities to which the results of a research are to be by investigators or researchers to participate in
generalized. For example, a researcher needs to their research project.
study the problems faced by postgraduate doc- l Element: The individual entities that comprise
tors in India; in this, the ‘population’ will be all the samples and population are known as ele-
postgraduate doctors who are Indian citizens. ments. An element is the most basic unit about
l Target population: A target population consists whom/which information is collected. An ele-
of the total number of people or objects which ment is also known as a subject in research. The
meet the designated set of criteria. In other most common element in medical and bio-
words, it is the aggregate of all the cases with a medical research is an individual. The sample or
certain phenomenon (or phenomena) about population depends on the phenomenon under
which the researcher would like to make a gener- study.
alization. For example, a researcher is interested l Sampling frame: It is a list of all the elements or
in identifying the complication of diabetes mel- subjects in the population from which the sam-
litus type-II among people who have migrated ple is drawn. The sampling frame can be freshly
to Ludhiana, India. In this instance, the target prepared by the researcher or an existing frame
population are all the migrants in Ludhiana with may be used. For example, a researcher may
diabetes mellitus type-II. prepare a list of all the households of a locality

Target Accessible
Population Sample
population population
Subjects

Fig. 6.1. ​Schematic representation of sampling.

118 Research Methodology and Biostatistics

which have pregnant women or may use a reg- is always easier to establish better rapport with a
ister of pregnant women for antenatal care sample than the whole population and thus to
available with the local Anganwadi worker. collect more accurate data. Thus, a sample helps
l Sampling error: There may be fluctuations in to generate precise and accurate data in a re-
the values of the statistics of characteristics search study.
from one sample to another, or even those
drawn from the same population.
l Sampling bias: Distortion that arises when a CHARACTERISTICS OF A GOOD SAMPLE
sample is not representative of the population There are various qualities and characteristic fea-
from which it was drawn. tures that make a sample good. To generalize the
l Sampling plan: The formal plan specifying a findings for an entire population, a good sample
sampling method, a sample size and the proce- for the research study must have the following
dure of selecting the subjects. characteristics:
l Representative: A representative sample is that

whose key characteristics are closely related to

PURPOSES OF SAMPLING those of the population. Representativeness of a
l Economical: In most cases, it is not possible or sample makes it possible to generalize the find-
economical for researchers to study an entire pop- ings for the population.
ulation. With the help of sampling, the researcher l Free from bias and errors: A good sample is one

can save lots of time, money and resources to which is free from deliberate selection of the sub-
study a phenomenon. Therefore, sampling pro- jects for study. The sample should be free from
vides an economical option for the researcher to simple random sampling errors or sampling bias.
generate empirical evidences. l No substitution and incompleteness: A sample

l Improved quality of data: It is a proven fact is said to be good if once a subject is selected for
that when a person handles less amount of work the study, it is neither replaced nor is it incom-
or less number of people, it is easier to ensure plete in any aspect of the researchers’ interest.
the quality of the outcome. Similarly, in re- l Appropriate sample size: Generally, it is be-

search, when a researcher handles information lieved that in quantitative studies, the larger
from only a part of the population under study, the sample size, the better the probability of the
it is easier to maintain the quality of the re- sample being good. However, in qualitative
search work, which would not be possible in studies, this notion is not considered important.
case the entire population was involved.
l Quick study results: Studying an entire popula-
tion itself will take a lot of time, and generating SAMPLING PROCESS
research results of a large mass will be almost Sampling is a process of selecting a part of the
impossible as most research studies have time assigned population to represent the entire popula-
limits. But with a sample, it is possible to gener- tion. The procedure of sampling should be system-
ate study results faster, which is one of the im- atically organized. This helps to eliminate the data
portant objectives of every researcher. bias that can occur during sample selection, which
l Precision and accuracy of data: Conducting a is mostly carried out inconsistently. The sampling
study on an entire population leaves researchers process entails the formulation of a specific criteria
with voluminous data, and maintaining preci- for selection, which ensures that the characteristics
sion of that data becomes a cumbersome task, of the phenomenon of interest will be present in all
while carrying out a study on a part of the popu- the units being studied. The sampling process con-
lation (sample) helps the researcher to generate sists of the following stages (Fig. 6.2):
more precise data, where formulation of the in- l Identifying and defining the target popula-

terpretations of the data becomes much easier. It tion: The first step of the sampling process is
 Sampling and Sample Size Determination 119

Identifying and defining the target units helps the researcher eliminate confusion
population while selecting the sample.
l Specifying sample selection methods: It is one
Describing the accessible population and of the important stages of the sampling process,
ensuring sampling frame where the researcher decides whether a sample
will be drawn from the population by using
Specifying the sampling unit
probability or nonprobability sampling tech-
niques. Selection of the specific method for se-
lection of the sample depends on several factors,
Specifying sample selection methods such as types of population, kind of phenome-
non under study, availability of resources and
knowledge of the researcher.
Determining the sample size l Determining the sample size: It is very essential
to determine the size of the sample, so that the
researcher can plan the implementation of the
Specifying the sampling plan sampling process accordingly.
l Specifying the sampling plan: Before the selec-
tion of a particular sample, the researcher must
Selecting a desired sample have a final sampling plan, so that the sampling
process can be implemented without any undue
Fig. 6.2. ​Stages of the sampling process. problems.
l Selecting a desired sample: Finally, a researcher
draws a representative sample from the accessi-
the identification and defining the target popu- ble population. This requires the implementa-
lation. Target population consists of the total tion of the plan of the sampling process, and
group of people or objects which are meeting ultimately the selection of a representative part
the designated set of criteria of interest of the of the population, which is used by the re-
researcher. Target population is the aggregate of searcher for data collection in research study.
cases about which the researcher would gener-
alize the information. Therefore, it is the first
FACTORS INFLUENCING A SAMPLING
and the most essential stage of the sampling
process. PROCESS
l Describing the accessible population and en- Following are some of the factors which may af-
suring sampling frame: It is not always pos- fect the sampling process:
sible to have access to each and every subject
included in the target population. Therefore, NATURE OF THE RESEARCHER
the researcher must establish a description l Inexperienced investigator: If the investigator
about the accessible population that is readily lacks adequate knowledge and experience about
available for research. In addition, after estab- the conditions of the research and the method-
lishing a complete description of the accessi- ology to be applied in a research process, the
ble population, researchers must have a sam- sample selection may be adversely affected.
pling frame available to select a sample from l Lack of interest: Lack of self-motivation and ap-
the accessible population. preciation for carrying out tasks or establishing
l Specifying the sampling unit: Next, the re- research methodology on the part of the re-
searcher must establish the specific inclusion searcher also affects the drawing of the sample.
and exclusion criteria to select a particular sam- l Lack of honesty: Lack of honesty will affect the
pling unit. Therefore, specifying the sampling sampling process in research. The researcher
120 Research Methodology and Biostatistics

should be honestly involved in each step of the and implementation of the sampling process.
research process. Otherwise, the sampling process gets adversely
l Intensive workload: Lack of adequate resources affected.
and ability to carry out the research process re- l Large geographic area: A large geographic area
sults in inadequate selection and application of needs lots of time and resources to accomplish
all resources, which results in adverse handling the sampling process. In addition, large geo-
of a research process including the sampling graphical areas can also lead to mental and
process. physical exhaustion and thus the sampling pro-
l Inadequate supervision: There should be adequate cess can get adversely affected.
supervision of the research activity to ensure the l Lack of cooperation: During the sampling process,
appropriate implementation of the research pro- the researcher needs cooperation from compe-
cess including the sampling process. tent authorities as well as from the study subjects.
In the absence of cooperation from the requisite
NATURE OF THE SAMPLE authorities and study subjects, the sampling pro-
l Inappropriate sampling technique: If a researcher cess may be adversely affected.
fails to choose an appropriate sampling tech- l Natural calamities: Sometimes the sampling
nique, the whole sampling process may get af- process is affected by natural calamities, such as
fected. floods and other natural disasters, deaths or
l Sample size: Very large samples become het- other environmental constrains.
erogeneous and do not exhibit characteristics
of the whole population in general; also there
are always chances of a biased sample. If a TYPES OF SAMPLING TECHNIQUES
sample is too small, a researcher may not be Sampling, as mentioned earlier, is the process of
able to generalize the study findings to the selecting a representative part of the population.
whole population. Thus, a carefully carried out sampling process
l Defective sampling frame: Defective sampling helps to draw a sample that represents the charac-
frame leads to a faulty sampling process. The teristics of the population from which the sample
researcher should have adequate knowledge is drawn. There are several methods or techniques
about the population under study to have an of sampling (Fig. 6.3); however, basically sampling
appropriate sampling frame. techniques are classified into two broad catego-
ries: probability and nonprobability sampling
CIRCUMSTANCES techniques. A brief review of probability and non-
l Lack of time: Adequate time should be available probability sampling techniques is presented in
with the researcher to have adequate planning Tables 6.1 and 6.6, respectively.

Types of sampling techniques

Probability sampling technique Nonprobability sampling technique

1. Simple random sampling 1. Purposive sampling

2. Stratified random sampling 2. Convenient sampling
3. Systematic random sampling 3. Consecutive sampling
4. Cluster/multistage sampling 4. Quota sampling
5. Sequential sampling 5. Snowball sampling

Fig. 6.3. ​Types of sampling techniques.

 Sampling and Sample Size Determination 121

Table 6.1. ​Brief review of probability sampling techniques

Types with Description Prerequisites and Uses Advantages Disadvantages
1. Simple random • Population should be • Most reliable and unbi- • Needs up-to-date
sampling homogeneous ased method complete list of all the
Every member of the • Must have a list • Requires minimum members of the popu-
population has an of the elements/ knowledge of the lation
equal chance of members of the study population • Expensive and time
being selected as accessible population • Free from sampling consuming
subject • Random selection errors/bias
either by lottery, ran-
dom table or computer
2. Stratified random • Used for heteroge- • Ensures representative • Requires complete
sampling neous population sample in heteroge- information of popula-
Dividing a heteroge- • Division of heteroge- neous population tion
neous population neous population • Comparison is possible • Large population is
into strata based on into strata based on between two groups required
selected traits, such selected traits, such as • Chances of faulty
as age, gender, habi- age, gender, religion, classification of strata
tat and then per- socioeconomic status,
forming random diagnosis, education,
selection of a sam- geographical region
ple from each strata
3. Systematic random • List of the target popu- • Convenient and simple • Less-representative
sampling: lation must be available to carry out sample if subjects
Selecting every Kth • Random distribution • Distribution of sample are nonrandomly
case from the group, of subjects rather than over entire population distributed
such as every 10th segregation • Sometimes may result
person on a patient • The first subject in biased sample
list or every 100th number is chosen with
person the help of the ran-
dom number table
4. Cluster or multistage • When population is • Cheap, quick and easy • Possibility of high
sampling very large, such as in for a large population sampling error
In very large popula- Asia, it is broken up • Population parameters • Chances of least–
tion, random selec- into geographic clus- can be estimated for representative sample
tion of geographic ters and then random sample size due to overrepre-
cluster is done and selection of geographic sented or underrepre-
then random selec- clusters is done sented cluster
tion of subjects from • Random selection of
these clusters subjects from selected
clusters
5. Sequential sampling • Sample size is not • Study on best-possible • Not possible to study a
The investigator ini- fixed; continue till smallest sample phenomenon that
tially selects small inferences are drawn • Facilitates inferences needs to be studied at
sample and tries to of study one point of time
make inferences; if • Requires repeated
not able to draw re- entry into the field to
sults; he/she then collect the sample
adds subjects until
clear-cut inferences
can be drawn
122 Research Methodology and Biostatistics

with each subject chosen independently of the

PROBABILITY SAMPLING
other members of the population.
TECHNIQUES There are two essential prerequisites to implement
It is based on the theory of probability. It in- the simple random technique: the population
volves random selection of the elements/ must be homogeneous and the researcher must
members of the population. In this, every subject have a list of the elements/members of the ac-
in a population has equal chance to be selected as cessible population.
a study sample. Probability sampling techniques The first step of the simple random sampling
are used to enhance the representativeness of the technique is to identify the accessible popula-
selected sample for a study. In probability sam- tion and prepare a list of all the elements/
pling techniques, the chances of systematic bias members of the population. The list of the sub-
are relatively less because subjects are randomly jects in the population is known as the sampling
selected. frame (Table 6.2). A sample is drawn from sam-
pling frame by using the following methods:
FEATURES OF PROBABILITY SAMPLING  The lottery method: There are a variety of

l Probability sampling techniques involve samples methods to perform simple random sam-
to be gathered in a process that provides equal pling. The oldest and most mechanical one is
chances to all the individuals in the population the lottery method. Each member of the pop-
of getting selected. ulation is alloted a unique number. Each
l In this sampling technique, the researcher has to number is placed inside a hat or bowl and the
guarantee that each individual has an equal op- bowl or hat is shook in a thorough manner.
portunity for getting selected. This is feasible The blind-folded researcher then chooses
only if the researcher uses randomization. numbered tags from the hat. All the individu-
l The absence of both sampling and systematic als who bear the numbers that are chosen by
bias is the advantage of utilizing a random the researcher become the subjects for the
sample. If random selection is made in a proper study. This may be carried out by using re-
manner, the sample is representative of the placement method, where chosen numbers
whole population. are replaced back in the container or by using
l The effect of this is an absent systematic bias or nonreplacement method, where chosen num-
a minimal one, which is the difference between bers are not replaced back in container.
the results from the sample and those from the  The use of a table of random numbers: This is

population. Since the subjects are randomly se- the most common and accurately used
lected, sampling bias is eliminated. method in simple random sampling. As de-
picted in Table 6.3, random tables present
TYPES OF PROBABILITY SAMPLING several numbers in rows and columns. The
Probability sampling techniques are further classi- researcher initially prepares a numbered list
fied into five subtypes: simple random sampling, of the elements/members of the population,
stratified random sampling, systematic random and then with a blindfold chooses a number
sampling, cluster or multistage sampling and se- from the random table. The same procedure
quential sampling. is continued until the desired number of
subjects is got. However, if repeatedly simi-
I. Simple Random Sampling Technique lar numbers are encountered, they are ig-
This is the most pure and basic probability sam- nored and the next numbers are considered
pling design. In this type of sampling design, until the desired number of subjects is
every population member has a similar chance achieved.
of being picked as the subject. The whole pro-  The use of computer: Nowadays, random ta-

cess of sampling is carried out in a single step, bles may be generated from the computer,
 Sampling and Sample Size Determination 123

Table 6.2. ​Sampling frame of a study conducted on college students

No. Name of Student No. Name of Student No. Name of Student
1 Ramandeep 14 Harpreet 27 Shruti
2 Kulwinder 15 Hardeep 28 Navpreet
3 Navdeep 16 Mandeep 29 Gurdeep
4 Deepinder 17 Kirandeep 30 Gagandeep
5 Kusum 18 Manjeet 31 Jaswinder
6 Kamini 19 Kulveer 32 Tamanpreet
7 Lalita 20 Kuldeep 33 Sharika
8 Jasbir 21 Kamaldeep 34 Sidhika
9 Jayamala 22 Shivani 35 Tanveer
10 Savita 23 Shalini 36 Prabhjot
11 Amandeep 24 Lakhwinder 37 Anita
12 Jasveen 25 Shalony 38 Bindu
13 Tara 26 Sharmila 39 Rajbeer

and subjects may be selected as described in This method of sampling is free from sam-
the use of the random table. For popula- pling errors.
tions with a small number of members, it is  Sample errors can be easily computed and the

advisable to use the first method, but if the accuracy of estimate easily assessed.
population has many members, a computer- Demerits
aided random selection is preferred. The re-  One of the most obvious limitations of the

search randomizer (https://www.randomizer. simple random sampling method is the re-

org/) or GraphPad Software: Quickcalcs quirement of a complete and up-to-date list
(http://graphpad.com/quickcalcs/randomN1. of all the members of the population. This list
cfm) are free resource software that can be is usually not available for large populations.
used by researchers to generate random In such cases, it is wiser to use other sampling
number tables and also may be used for the techniques.
randomization of subjects in experimental  This method does not make use of knowledge

research. about a population which the researcher may

Merits already have.
 One of the best things about simple ran-  Lots of procedures need to be performed be-

dom sampling is the ease of assembling the fore sampling is accomplished.

sample.  It is an expensive and time-consuming

 It is also considered a fair way of selecting a process.

sample from a given population since every
member is given equal opportunity of being II. Stratified Random Sampling
selected. This method is used for heterogeneous popula-
 Simple random sampling requires minimum tions. Stratified sampling is a probability sam-
knowledge about the population in advance. pling technique wherein the researcher divides
 This is one of the most unbiased probability the entire population into different homoge-
methods of sampling. neous subgroups or strata, and then randomly
124 Research Methodology and Biostatistics

Table 6.3. ​Extract from a table of random numbers

10 85 73 97 99 31 01 04 39 07 09 90 92 25 76
14 17 93 94 89 66 42 57 89 58 12 54 79 48 21
37 63 20 33 47 12 54 79 48 21 66 42 57 89 58
08 75 26 64 06 09 90 92 25 76 31 01 04 39 07
09 90 92 25 76 10 85 73 97 99 10 85 73 97 99
12 54 79 48 21 14 17 93 94 89 14 17 93 94 89
66 42 57 89 58 37 63 20 33 47 37 63 20 33 47
31 01 04 39 07 08 75 26 64 06 08 75 26 64 06
09 90 92 25 76 10 85 73 97 99 14 17 93 94 89
12 54 79 48 21 14 17 93 94 89 37 63 20 33 47
66 42 57 89 58 37 63 20 33 47 09 90 92 25 76
31 01 04 39 07 08 75 26 64 06 12 54 79 48 21
10 85 73 97 99 31 01 04 39 07 31 01 04 39 07
14 17 93 94 89 66 42 57 89 58 10 85 73 97 99
37 63 20 33 47 12 54 79 48 21 08 75 26 64 06
08 75 26 64 06 09 90 92 25 76 66 42 57 89 58
14 17 93 94 89 66 42 57 89 58 10 85 73 97 99
37 63 20 33 47 12 54 79 48 21 14 17 93 94 89
09 90 92 25 76 10 85 73 97 99 37 63 20 33 47
12 54 79 48 21 14 17 93 94 89 08 75 26 64 06
31 01 04 39 07 08 75 26 64 06 09 90 92 25 76
10 85 73 97 99 31 01 04 39 07 12 54 79 48 21
08 75 26 64 06 09 90 92 25 76 66 42 57 89 58
66 42 57 89 58 37 63 20 33 47 31 01 04 39 07

selects the final subjects proportionally from the in proportion to the size of the total popula-
different strata. tion. When viewed against the entire popula-
The strata are divided according to selected traits tion, the sample size of each stratum in this
of the population, such as age, gender, religion, technique seems proportionate to the popula-
socioeconomic status, medical diagnosis, edu- tion size of the stratum.
cation, geographical region, type of institution, In this technique, the thing that is important
type of care, type of specialties, clinical special- to remember is to utilize similar sampling
ization and site of care. fraction for each stratum irrespective of
According to the weightage of the sample and the differences in the strata’s population
proportion, stratified random sampling is fur- size. It is like assembling a smaller popula-
ther divided into two categories: proportion- tion that is particular to the relative pro-
ate and disproportionate stratified random portions of the subgroups within the
sampling. population.
 Proportionate stratified random sampling: In For example, a researcher has three strata
this, the sample chosen from each stratum is containing 100, 200 and 300 population
 Sampling and Sample Size Determination 125

sizes, respectively. The researcher then se- Merits

lects a sampling fraction that is ½. For this,  It ensures representation of all groups in a

the researcher should randomly sample 50, population.

100 and 150 subjects from each stratum,  Researchers also employ stratified random

respectively (Table 6.4). sampling when they want to observe existing

 Disproportionate stratified random sampling: relationships between two or more subgroups.
In this subtype, the sample chosen from each Therefore, comparison is possible between
stratum is not in proportion to the size of the subgroups in this sampling technique.
total population in that stratum. Sampling  The researcher can representatively sample

fraction is the only difference between pro- the most inaccessible subgroups and even
portionate and disproportionate stratified the smallest in the population with strati-
random sampling. The different strata have fied sampling. This permits the researcher
dissimilar sampling fractions with dispropor- to sample the rare extremes of a provided
tionate sampling. population.
The exactness of this design is mainly depen-  There is a higher statistical precision com-

dent on the sampling fraction attribution pared to simple random sampling with strati-
of the researcher. If the researcher makes fied sampling technique. This is due to the
mistakes in allocating sampling fractions, fact that the variability within the subgroups
a stratum may either be underrepresented is less compared to the variations while deal-
or overrepresented, resulting in skewed ing with the entire population.
results.  As this technique has high statistical precision,

For example, if a researcher wants to study the it should also be understood that it needs a
biophysical profiles of students in a college smaller sample size that can save much time,
that contains 100 students from Himachal money and effort of the researchers.
Pradesh, 200 students from Haryana and Demerits
300 students from Punjab, the researcher  Proportionate stratification requires accurate

will choose different sampling fractions and information on the proportion of population
then randomly select a sample of 50 sub- in each stratum.
jects from each stratum (Table 6.5).  Large population must be available from

which to select subjects.

 There is always a possibility of faulty classifica-
Table 6.4. ​Example of proportionate stratified tion and hence increase in variability.
random sampling

Stratum A B C III. Systematic Random Sampling

Population size 100 200 300 Systematic random sampling can be likened to an
arithmetic progression, wherein the difference
Sampling fraction 1/2 1/2 1/2
between any two consecutive numbers is the
Final sample size 50 100 150 same. It involves the selection of every Kth case
from a list of the group, such as every 10th per-
son on a patient list or every 100th person from
Table 6.5. ​Example of disproportionate stratified a phone directory. Systematic sampling is some-
random sampling times used to sample every Kth person entering
Stratum A B C a bookstore, or passing down the street or leav-
Population size 100 200 300 ing a hospital, and so forth. Systematic sampling
can be applied so that an essentially random
Sampling fraction 1/2 1/4 1/6
sample is drawn. If we had a list of subjects or
Final sample size 50 50 50 sampling frame, the following procedure could
126 Research Methodology and Biostatistics

be adopted. The desired sample size is estab- random sampling cannot be imagined when the
lished at some number (n) and the size of the population in discussion is the entire popula-
population must be known or estimated (N). tion of Asia.
Number of subjects in target popu
ulation(N )
Cluster sampling means random selection of sam-
K 5 N / n or K 5 pling units consisting of population elements.
Size of sample(n)
Then from each selected sampling unit, a sam-
For example, a researcher wants to choose about ple of population elements is drawn by either
100 subjects from a total target population of 500 simple random selection or stratified random
people. Therefore, 500/100 5 5. Thus, every 5th sampling. This method is used in cases where
person will be selected. the population elements are scattered over a
In this method, a list of the subjects is prepared wide area, and it is impossible to obtain a list of
for the target population (sampling frame) all the elements. The important thing to re-
and then the first subject is randomly selected; member about this sampling technique is to
later, every Kth subject is selected from the give all the clusters equal chances of being
sampling frame. Before selecting a sample selected.
through systematic random sampling tech- Geographical units are the most commonly used
nique, the researcher must keep in mind that ones in research, e.g. a researcher requiring to
subjects/elements of the population must be in survey the academic performance of Indian
random order, which means a mixture of the high school students.
elements rather than segregation, and the first  He/she can divide the entire Indian popula-

subject is chosen by the help of a random tion into different clusters (e.g. cities).
number table.  Depending on his/her research through sim-

Merits ple or systematic random sampling, the re-

 Researchers find this technique convenient searcher then selects a number of clusters.
and simple to carry out.  Then the researcher can either include all the

 Distribution of sample is spread evenly over high school students as subjects or he/she can
the entire given population. chose a number of subjects from each cluster
 Less cumbersome, time consuming and through simple or systematic random sam-
is cheaper than simple random sampling pling from the selected clusters (randomly
technique. selected cities).
 Statistically more efficient and provides a bet- Types of cluster samples
ter representative sample when population  One-stage cluster sample: Recall the example

elements are randomly distributed. given above; one-stage cluster sample occurs
Demerits when the researcher includes all the high
 If the first subject is not randomly selected, school students from all the randomly se-
then it becomes a nonrandom sampling lected clusters as sample.
technique.  Two-stage cluster sample: In this, initially the

 Sometimes this may result in a biased sample. researcher lists all the clusters appearing in
 If the sampling frame has nonrandomly dis- the population. Then, first the clusters are
tributed subjects, this sampling technique selected normally by simple random sam-
may not be appropriate to select a representa- pling (SRS). Then usually by simple random
tive sample. sampling or often by systematic sampling,
the elements or units in the selected clusters
IV. Cluster or Multistage Sampling of the first stage are then sampled in the sec-
When simple random sampling is not possible ond stage.
due to the size of the population, cluster  Multistage cluster sample: In multistage cluster

random sampling is carried out. A simple sample, the sampling is done at more than
 Sampling and Sample Size Determination 127

two levels by initially identifying clusters as V. Sequential Sampling

population at different levels and selecting This method of sample selection is slightly differ-
them using the simple random sampling ent from the other methods. Here the sample
technique and finally, selecting units (ele- size is not fixed. The investigator initially selects
ments) using simple random sampling or a small sample and tries to make inferences; if
systematic sampling. not able to draw results, he or she then adds
 Probability proportion to size cluster sam- more subjects until clear-cut inferences can be
pling: It is a variant of cluster sampling, drawn.
when the size of clusters is not the same and For example, a researcher is studying association
there is risk of oversampling from the between smoking and lung cancer. Initially, the
smaller size clusters and undersampling researcher takes a smallest sample and tries to
from the larger size clusters. It is most useful draw inferences. If unable to draw any infer-
when the sampling units vary considerably ences, he or she continues to draw samples until
in size because it assures that those in larger meaningful inferences are drawn.
sites have the same probability of getting
into the sample as those in the smaller sites,
and vice versa. For example, a researcher is Having lung
interested in identifying the prevalence of Number of Nonsmokers cancer
HCV infection in rural and urban commu- subjects Smokers (A) (B) (A) (B)
nities of Punjab state. The most suitable 20 7 12 2 1
method of sampling will be probability pro- 30 18 22 5 3
portional to size (‘PPS’) cluster sampling
50 28 22 10 4
(see detailed example in Appendix XI).
Merits
 This sampling technique is cheap, quick and The aforementioned table depicts the progres-
easy for a large population. sively increasing sample sizes until inferences
 Large populations can be studied, and require are drawn. It can be said that in these values, out
only a list of the members. of 50 subjects, 28 smokers had almost double
 Enables investigators to use existing division, incidence of lung cancer compared to 22 non-
such as districts and villages/towns. smokers.
 Same cluster can be used again for study. Merits
Demerits  Facilitates conducting a study on best-possible

 This technique is the least representative of smallest representative sample.

the population from the various kinds of  Helps in ultimately finding inferences from

probability sampling. The tendency of indi- the study.

viduals within a cluster is to get similar Demerits
characteristics. Moreover, there is a chance  With this sampling technique, it is not pos-

that the researcher can have an overrepre- sible to study a phenomenon which needs to
sented or underrepresented cluster that can be studied at one point of time.
skew the results of the study with a cluster  Requires repeated entries into the field to col-

sample. lect samples.

 There is a possibility of high sampling error.

Since limited clusters are included in the sam-

ple, a significant proportion of the population NONPROBABILITY SAMPLING
is unsampled. Nonprobability sampling is a technique wherein
 If a small homogeneous population is under the samples are gathered in a process that does
study, this technique is not at all useful. not give all the individuals in the population
128 Research Methodology and Biostatistics

equal chances of being selected in the sample. l It can also be utilized when the researcher tar-
In other words, in this type of sampling, every gets to make a qualitative, pilot or exploratory
subject does not have equal chance to be se- study.
lected because elements are chosen by choice l It can also be used when randomization is im-
not by chance through nonrandom sampling possible like when the population is almost
methods. limitless.
Generally, it is believed that nonrandom meth- l Moreover, when the research does not aim to
ods of sampling are more likely to produce a bi- produce results that will be utilized to generate
ased sample rather than random methods. In generalizations pertaining to the entire popula-
nonprobability sampling, certain elements have tion, it can be used.
more probability to be the part of the sample, l In addition, when the researcher has limited
while others may have no chance of being in- budget, time and workforce, it is also of use.
cluded in the sample. This restricts the generaliza- l Used as a randomized, probability sampling,
tion that can be made about the study findings. It this technique can also be used in an initial
is generally observed that despite the limitations study (pilot study) and can be carried out
of nonprobability, they are the most frequently again.
used sampling methods by the researchers.
TYPES OF NONPROBABILITY SAMPLING
FEATURES OF NONPROBABILITY SAMPLING Nonprobability sampling techniques are classified
l Nonprobability sampling is a technique wherein into several categories, such as purposive sam-
the samples are gathered in a process that does pling, convenient sampling, quota sampling, con-
not give all the individuals in the population secutive sampling and snowball sampling, as listed
equal chances of being selected. In any form of in Table 6.6.
research, true random sampling is always diffi-
cult to achieve. I. Purposive Sampling
l Most researchers are bound by time, money and Purposive sampling is more commonly known as
workforce, and because of these limitations, it is judgmental or authoritative sampling. In this
almost impossible to randomly sample the entire type of sampling, subjects are chosen to be part
population. It is then often necessary to employ of the sample with a specific purpose in mind.
the nonprobability sampling technique. In purposive sampling, the researcher believes
l In contrast with probability sampling, nonprob- that some subjects are fit for research compared
ability sample is not a product of randomized to other individuals. This is the reason why they
selection processes. Subjects in a nonprobability are purposively chosen as subjects.
sample are usually selected on the basis of their In this sampling technique, samples are chosen by
accessibility or by the purposive personal judge- choice not by chance, through a judgement
ment of the researcher. made by the researcher based on his or her
l The downside of this is that an unknown propor- knowledge about the population. The researcher
tion of the entire population is not sampled. This might decide purposely to select subjects who
entails that the sample may or may not represent are judged or believed to be typical or represen-
the entire population accurately. Therefore, the tative of the accessible population. Particularly,
results of the research cannot be used in general- experts who have in-depth knowledge about the
izations pertaining to the entire population. accessible population under study may be asked
to provide help to select the sample in purposive
USES OF NONPROBABILITY SAMPLING sampling.
l This type of sampling can be utilized when For example, a researcher wants to study the lived
showing that a particular trait is existent in the experiences of postdisaster depression among
population. people living in earthquake-affected areas of
 Sampling and Sample Size Determination 129

Table 6.6. ​Brief review of nonprobability sampling techniques

Types with Description Prerequisites and Uses Advantages Disadvantages
1. Purposive sampling • Requires in-depth • Simple to draw a • Requires considerable
Subjects are chosen to knowledge about the sample knowledge about the
be part of the sam- accessible population • Saves resources as it population
ple with a specific • Used when a limited requires less fieldwork • Conscious biases may
purpose in mind number of individuals occur
possess the trait of
interest
2. Convenience • Convenient accessibil- • Easiest, cheapest and • Chances of sampling
sampling ity and proximity to least time consuming bias
Subjects are selected study subjects • Helps in saving time, • Nonrepresentative
because of their • Used in case of limited money and resources sample
convenient accessi- availability of time and • Findings cannot be
bility and proximity resources generalized
to the researcher
3. Consecutive • Used for continuously • Ensures more represen- • Limitless option about
sampling changing population, tative sample the sample size
Picks up all the avail- such as hospital • Convenient and less • Does not always
able subjects who patients time consuming. guarantee the repre-
meet the preset in- sentative sample
clusion and exclu-
sion criteria
4. Quota sampling • The bases of the quota • Economically cheap • Does not always guar-
Equal or proportionate are usually age, gen- • Suitable where field- antee representative
representation of der, education, race, work has to be done sample
subjects from each religion, socioeco- like studies related • Chances of sampling
quota nomic status, etc. to market and public bias
opinion polls
5. Snowball sampling • Used by researchers to • Facilitates sampling for • Little control of
Locating the initial identify potential sub- people difficult to researcher over the
subject, and then jects in studies where locate sampling method
taking assistance subjects are hard to • Cheap, simple and • Representativeness
from the subject to locate cost-efficient of the sample is not
identify people with • Needs little planning guaranteed.
a similar trait of and lesser workforce • Chances of poor
interest coverage of entire
population

Gujarat. In this case, a purposive sampling tech- the population as per the knowledge of the
nique is used to select the subjects who were the researcher.
victims of the earthquake disaster and have suf- Uses of purposive sampling: Purposive sampling
fered postdisaster depression living in earth- design is usually used when a limited number
quake-affected areas of Gujarat. For this study, of individuals possess the trait of interest. It
the researcher will select only those people who is the only viable sampling technique in
fulfil the criteria as well as particular subjects obtaining information from a very specific
that are the typical and representative part of group of people. It is also possible to use
130 Research Methodology and Biostatistics

purposive sampling if the researcher knows a Subjects are chosen simply because they are
reliable professional or authority that he or easy to recruit.
she thinks is capable of assembling a repre- For example, if a researcher wants to conduct a
sentative sample. study on older people residing in Delhi, and the
Merits researcher observes that he can meet several
 Simple to draw sample and useful in explor- older people coming for morning walk in a park
ative studies. located near his residence in Delhi, he can
 Saves recourses, requires less fieldwork. choose these people as his research subjects.
Demerits These subjects are readily accessible for the re-
 Requires considerable knowledge about the searcher and may help him to save time, money
population under study. and resources.
 It is not always a reliable sample, as conscious Uses of convenience sampling: Researchers use
biases may exist. convenience sampling not just because it is
 The two important weaknesses of purposive easy to use, but also because it has other re-
sampling stay with the authority and in the search advantages. In pilot studies, conve-
sampling process, both of which relate to the nience sample is usually used because it al-
bias and reliability that accompanies the sam- lows the researcher to obtain basic data and
pling technique. trends for his or her study without the com-
 Unfortunately, there is usually no way to deter- plications of using a randomized sample. In
mine the reliability of the expert or the author- addition, this sampling technique is also use-
ity. To minimize sampling errors, it is best to opt ful in documenting a particular quality of a
for the most experienced and best authority in substance or phenomenon that occurs within
the field of interest. a given sample. Such studies are also very use-
 It is generally biased as no randomization is ful for detecting relationships among differ-
utilized in getting the sample when it comes ent phenomena.
to the sampling process. It should also be Merits
noted that the members of the population  This technique is considered easiest, cheapest

had no equal chance of getting selected. The and least time consuming. It may help in sav-
result of this is the misrepresentation of the ing time, money and resources.
entire population that will thereafter limit Demerits
generalizations of the results of the study.  The most obvious criticism about conve-

nience sampling is sampling bias.

II. Convenience Sampling  It does not provide a representative sample

Convenience sampling is probably the most from the population of the study.
common of all sampling techniques because  Findings generated from these samples can-

it is fast, inexpensive and easy, and the sub- not be generalized on the population.
jects are readily available. In convenience
sampling, which is a nonprobability sampling III. Consecutive Sampling
technique, subjects are selected for their con- Consecutive sampling is very similar to conve-
venient accessibility and proximity to the re- nience sampling except that it seeks to include
searcher. The subjects are chosen just because all accessible subjects as part of the sample. This
of the fact that they are very easy to recruit for nonprobability sampling technique can be con-
the study. Moreover, the researcher does not sidered as the best of all nonprobability samples
opt for choosing subjects that are representa- because it includes all the subjects that are avail-
tive of the entire population. Sometimes, it able, which makes the sample a better represen-
is also known as an accidental sampling. tation of the entire population.
 Sampling and Sample Size Determination 131

It is also known as total enumerative sampling. In selected without using the random process from
this sampling technique, the investigator picks different quotas.
up all the available subjects who meet the preset Steps and uses of quota sampling
inclusion and exclusion criteria. This technique  Dividing the population into exclusive sub-

is generally used in small-sized populations. For groups is the initial step in nonprobability
example, if a researcher wants to study the ac- quota sampling.
tivity pattern of post–kidney transplant pa-  Thereafter, the researcher should recognize

tients, he can select all the post–kidney trans- the proportions of these subgroups in the
plant patients who meet the designed inclusion population; this same proportion will be em-
and exclusion criteria, and who are admitted in ployed in the sampling process.
the posttransplant ward during a specific time  Finally, the researcher choses subjects from

period. various subgroups when taking into account

Merits the proportions noted in the previous step.
 There is very little effort on the part of the  The sample is representative of the entire

researcher when performing this sampling population and is ensured by the final step. It
technique. also permits the researcher to study charac-
 It is not expensive, time consuming or work- teristics and traits that are significant for each
force intensive. subgroup.
 Ensures more representativeness of the se-  The main reason for choosing quota samples

lected sample. is that they permit researchers to sample a

Demerits subgroup that is of very large interest to the
 The researcher has no set plans about the study. This type of sampling is the best tech-
sample size and sampling schedule. nique to investigate a trait or a characteristic
 It does not always guarantee the selection of a of a certain subgroup. Additionally, quota
representative sample. sampling also permits researchers to con-
 Results from this sampling technique cannot tinue relationships between subgroups. In
be used to reach conclusions and interpreta- some studies, traits of a certain subgroup
tions pertaining to the entire population. have been found to be interacting with other
traits of other subgroups. In such cases also,
IV. Quota Sampling it is essential for the researcher to utilize this
Quota sampling is a nonprobability sampling type of sampling technique.
technique wherein the researcher ensures equal Merits
or proportionate representation of subjects, de-  Economically cheap, as there is no need to ap-

pending on which trait is considered as the basis proach all the candidates
of the quota. The bases of the quota are usually  Suitable for studies where fieldwork has to be

age, gender, education, race, religion and socio- carried out, like studies related to market and
economic status. public opinion polls
For example, if the basis of the quota is year in Demerits
college and the researcher needs equal represen-  Although this kind of sampling technique is

tation, with a sample size of 100, he or she must seen to be wholly representative of the popu-
select 25 first-year students, another 25 second- lation, in some cases it is not so – this is be-
year students, 25 third-year and 25 fourth-year cause only the chosen traits of the population
students. are taken into consideration while forming
The quota sampling technique seems similar to the subgroups.
the stratified random sampling technique; how-  Other traits in the sample may be overrepre-

ever, in this sampling technique, the sample is sented in the process of sampling these
132 Research Methodology and Biostatistics

subgroups. The final sample may have a created by the completion of the desired
skewed representation of race, age, educa- sample (Fig. 6.4).
tional attainment, marital status and a lot Exponential nondiscriminative snowball sam-
more in a study that considers gender, socio- pling: In this, each sample member is asked to
economic status and religion as the basis of provide reference to at least two similar sub-
the subgroups. jects, because of which the size of the sample
 It is not possible to estimate errors. grows exponentially and a large sample size
 Bias is possible, as the investigator/interviewer can be achieved (Fig. 6.5).
can select persons known to him or her. Exponential discriminative snowball sampling: In
this, initially, one sample is selected and asked
V. Snowball Sampling for two references to similar subjects, out of
Snowball sampling is a nonprobability sam- which at least one subject must be active to
pling technique that is used by researchers to provide further references and another could
identify potential subjects in studies where be nonactive in providing references. Simi-
subjects are hard to locate, such as commer- larly, each active reference subject is further
cial sex workers and drug abusers. For ex- asked for two references to similar subjects;
ample, a researcher wants to conduct a study out of which one should be active for further
on the prevalence of HIV/AIDS among com- references (Fig. 6.6).
mercial sex workers. In this situation, snow-
ball sampling is the best choice to select a
sample.
Therefore, researchers use this sampling method if
the sample for studies are very rare or are lim- Fig. 6.4. ​Linear snowball sampling.
ited to very small subgroups of the population.
This type of sampling technique works like a
chain referral. Therefore, it is also known as
chain referral sampling. The researcher asks for
assistance from the subject to identify people
with a similar trait of interest after observing
the initial subject.
Asking subjects to nominate another persons
with the same trait is similar to the process of
snowball sampling. Until a sufficient number
of subjects is obtained, the researcher observes
the nominated subjects and continues in the Fig. 6.5. ​E xponential nondiscriminative snowball
same way. sampling.
For example, the researcher may choose to use
snowball sampling while obtaining subjects for
a study to observe a rare disease. It is also pos-
sible that patients with similar disease may have
a support group. Observing one of the members
as the initial subject, leads to more subjects for
the study.
Types of snowball sampling
Linear snowball sampling: In this, each se-
lected sample is asked to provide reference to
only one similar subject; a linear chain is Fig. 6.6. ​Exponential discriminative snowball sampling.
 Sampling and Sample Size Determination 133

Merits problems which are under study. There are sev-

 The chain referral process permits the re- eral statistical formulas to calculate sample size.
searcher to reach populations that are difficult This chapter presents some of the simplest ways
to sample. of calculating the sample size for a research study
 The process is simple, cheap and cost- to facilitate inferences and generalization of the
efficient. study findings.
 This sampling technique requires lesser work-

force and little planning compared to other SAMPLE SIZE IN QUALITATIVE STUDIES
methods. Sample size in qualitative studies is not guided by
Demerits rules. It largely depends on the several purposes
 The researcher has less control over the sam- of the study, the quality of the informants and
pling method. The subjects that the researcher the type of sampling strategy used. It also de-
can get depend chiefly on the previous subjects pends upon the variety of characteristics we are
that were observed. looking for. The more the variety, the bigger is
 Sample representativeness is not guaranteed. the sample size needed. In case less number of
The researcher has no idea regarding the true characteristics is expected from the sample, the
distribution of the population and of the sample size will be kept at a level where it attains
sample. saturation, and there is no further possibility of
 When using this sampling technique, sam- more views to be accommodated. While if the
pling bias is also a matter of great concern number of characteristics under study is more,
for researchers. Initial subjects generally then a larger sample has to be reached until satu-
nominate people who they are aware of. Due ration to responses to those characteristics is
to this, it is very much possible that the sub- achieved.
jects share similar characteristics and traits. Therefore, the thumb rule to decide the sam-
Thus, the sample that the researcher will ple size in qualitative studies is based on data
obtain is possibly a small subgroup of the saturation, which is a point at which no new
entire population. information is obtained and redundancy is
reached. Data saturation depends on the num-
ber of factors/scope of the research question.
SAMPLE SIZE DETERMINATION The broader the scope, the larger the number of
Research is a systematic way of approaching a subjects required.
problem to find answers to questions. The re- In ethnography, 25–50 key informants are se-
search problem needs to be broken into a set of lected as a study sample, while in phenomenol-
questions that will extract the desired answers ogy, usually 10 or less subjects who have typically
from a chosen unit of population. Since it is not experienced the phenomenon under study are
feasible to study the whole population, a sample considered as sample. However, in grounded the-
is selected out of the population. The sample ory, generally 20 to 30 informants, who can best
drawn out should reflect population traits. Sam- contribute information to develop a theory are
ple size should be adequate to represent the selected.
population.
Often researchers face the problem of deter- SAMPLE SIZE IN QUANTITATIVE STUDIES
mining how many individuals they should take in The quantitative researcher needs to pay careful
their study as sample; sample size is considered attention to the number of subjects required to
one of the major aspects in the conduct and test a research hypothesis adequately. The only
evaluation of research. Sometimes, researchers thing which researchers have to keep in mind is to
are confused as to what would be the size of the choose the largest sample possible because sample
sample which would be sufficient to answer the error is inversely proportional to sample size. As the
134 Research Methodology and Biostatistics

Table 6.7. Number of patients and average represents the population. The result so revealed
number of visits by the sample should stand closest to the true
population value. However, power analysis can be
No. of Patients Average No. of Visits used to estimate an accurate sample in quantitative
50 100 studies.
15 50
THUMB RULES FOR ESTIMATING SAMPLE SIZE
25 75
35 85
Thumb rules for estimating sample size lack sci-
entific basis and many scientists dislike them.
40 95
However, when no baseline information for com-
putation of sample size is available and constraints
do not permit a pilot study either, the following
thumb rules can be used:
sample size increases, the probability of getting a A large-sized health science trial should include
markedly deviant sample diminishes. A large sam- nearly 300 subjects in each group; a mid-sized
ple provides an opportunity to counterbalance trial, nearly 100 subjects in each group and a
atypical values. small-sized trial and postgraduate thesis, at least
Let us illustrate it with an example of number 30 subjects in each group. In a multicentric study,
of patients who visit an ostomy clinic in a year. the sample must include the same number of sub-
The population comprises all patients visiting jects from each centre. The same norms may be
that clinic. Now, for our convenience, we take a applied for retrospective or case–control studies.
sample as follows. However, in prospective studies, the sample should
Table 6.7 shows a total population of 50 pay- be large enough that in the end at least 30 subjects
ing 100 visits to an ostomy clinic. It is also ob- are available to measure the outcome in each
served that as the number of patients included in group. In case of descriptive studies, at least
the study increases from 15 to 25, the average 200 subjects are required.
shifts towards the average number of visits for
the total population. As it goes from 25 to 40, it SAMPLE SIZE DETERMINATION USING
comes even closer to the average for the popula- A TABLE
tion, so we can appreciate that as the sample size Various formulas are available and recommended
becomes larger, the sample becomes more repre- for calculation of sample size, but it is not an
sentative and is less deviated compared to a easy task to calculate sample size using these for-
smaller sample. mulas. Therefore, a table is given by research
However, a large sample does not always en- advisors (2006) based on a sample size calcula-
sure accuracy. For instance, we carry out a study tion formula devised by Krejcie and Morgan
on the number of people who go to KFC restau- (1970). It is quite easy to determine a sample size
rants. In this study, most of the people included for a research study of an experimental or con-
would be from the higher socioeconomic strata, trol group of study using Table 6.8. In nursing
excluding poor people, as the number of visits research studies, generally 95% confidence level
from poor people would be less. Therefore, how- and 5% margin error is considered ideal. For
soever high the number of subjects in the sample example, a nurse researcher wants to study pa-
may be, the poor strata will be missing, and thus tient satisfaction in a hospital, where 5000 pa-
the results would not be accurate. tients visit the facility and therefore, wants to
Sample size determination in a quantitative study sample a sufficient number to generate a 95%
is a diligent job. A sample is the representation of confidence interval and 62.5% margin error.
a given population and it is worth it if it actually He would need to interview about 176 patients
 Sampling and Sample Size Determination 135

Table 6.8. ​Table for sample size determination

Population Size Confidence 5 95% Margin of Error Confidence 5 99% Margin of Error
5.0% 3.5% 2.5% 1.0% 5.0% 3.5% 2.5% 1.0%
10 10 10 10 10 10 10 10 10
20 19 20 20 20 19 20 20 20
30 28 29 29 30 29 29 30 30
50 44 47 48 50 47 48 49 50
75 63 69 72 74 67 71 73 75
100 80 89 94 99 87 93 96 99
150 108 126 137 148 122 135 142 149
200 132 160 177 196 154 174 186 198
250 152 190 215 244 182 211 229 246
300 169 217 251 291 207 246 270 295
400 196 265 318 384 250 309 348 391
500 217 306 377 475 285 365 421 485
600 234 340 432 565 315 416 490 579
700 248 370 481 653 341 462 554 672
800 260 396 526 739 363 503 615 763
1,000 278 440 606 906 399 575 727 943
1,200 291 474 674 1067 427 636 827 1119
1,500 306 515 759 1297 460 712 959 1376
2,000 322 563 869 1655 498 808 1141 1785
2,500 333 597 952 1984 524 879 1288 2173
3,500 346 641 1068 2565 558 977 1510 2890
5,000 357 678 1176 3288 586 1066 1734 3842
7,500 365 710 1275 4211 610 1147 1960 5165
10,000 370 727 1332 4899 622 1193 2098 6239
25,000 378 760 1448 6939 646 1285 2399 9972
50,000 381 772 1491 8056 655 1318 2520 12455
75,000 382 776 1506 8514 658 1330 2563 13583
100,000 383 778 1513 8762 659 1336 2585 14227
250,000 384 782 1527 9248 662 1347 2626 15555
500,000 384 783 1532 9423 663 1350 2640 16055
1,000,000 384 783 1534 9512 663 1352 2647 16317
2,500,000 384 784 1536 9567 663 1353 2651 16478
10,000,000 384 784 1536 9594 663 1354 2653 16560
100,000,000 384 784 1537 9603 663 1354 2654 16584
300,000,000 384 784 1537 9603 663 1354 2654 16586
136 Research Methodology and Biostatistics

visiting the hospital chosen through simple ran- Box 6.1. ​Za critical values on normal distribution
dom sampling (Table 6.8). In other types of sam-
pling and designs, the design effect is considered. a-error 5% 1% 0.1%
For example, in cluster sampling, the design ef- Two-sided (two-tailed 1.96 2.5758 3.2905
fect is considered as 2; therefore, sample size will test)
be double than in the case of the simple random One-sided (one-tailed 1.65 2.33
the sampling technique. test)

STATISTICAL DETERMINATION
OF THE SAMPLE SIZE one-tailed test) and Z12a/2 (for two-tailed test)
As mentioned earlier, one of the most difficult values may be referred from Box 6.1, where Z is
questions to answer in research is: what should be constant (set by convention according to the ac-
sample size for a research study? As usual, there is cepted a error and whether it is a one-sided or
no simple answer to this question. The calculation two-sided effect).
of an adequate sample size is the most crucial step l Power of the study: It is also known as type-II
in biomedical research studies so that true infer- error or false-negative observation, where the
ences about the population can be made from the researcher fails to detect the difference due to
results obtained through the study. Sample size is intervention that actually occurred; it is statically
a function of a number of things: the research represented as b. The ‘power’ of study is equal to
design being used; the variability of the key vari- (1 – b), which is the probability of failing to de-
ables; if we are trying to estimate population tect a difference when actually there is a differ-
value; or if we are testing hypotheses; the size of ence. The power of a study increases as the
the difference when testing hypotheses. The re- chances of committing a type-II error decreases.
searcher tries to set a sample size that minimizes An 80% power of study is usually accepted in
making two types of errors when drawing conclu- most studies, which means that we are accepting
sions from the data: (a) to claim that variables are that there is a 20% chance of missing the real
related when they are not (called type-I error) or difference; however, sometimes 90% power is
(b) to conclude the opposite that two variables are required in large studies to reduce the possibility
unrelated when in fact they are related (called of false-negative results to 10%. The critical values
type-II error). Resolving these issues and deter- of Zb may be referred from Box 6.2, where Z is a
mining sample size is not an easy task. constant set by convention according to power of
The statistical determination of sample size is the study.
considered as the most precise and accurate way l Effect size: The difference between the value of
of estimating the sample size. Generally, the sam- the variable in the control group and that in the
ple size for any study depends on the following: test drug group is referred to as effect size. For
l Level of significance: It is also known as type-I example, the reduction of systolic BP of a pa-
error or acceptance of false-positive results, tient in the experimental group is 30 mm Hg
which are believed to be by chance and not be- and in the control group, it is 10. Thus, we can
cause of particular intervention. The level of say that the effect size in this scenario is 20. The
significance is a predetermined value in a par- effect size may be expressed in absolute differ-
ticular study, denoted by symbol p or Z12a (for ence or relative difference, e.g. the absolute
one-tailed test) and Z12a/2 (for two-tailed test).
When one says that we will accept p,0.05 as
Box 6.2. Zb critical values on normal distribution
significant, it means that we will accept 5%
chances of false-positive results. The excepted Power 80% 85% 90% 95%
level of significance in research studies in gener- Values 0.8416 1.0364 1.2816 1.6449
ally either 0.05 (5%) or 0.01 (1%). Z12a (for
 Sampling and Sample Size Determination 137

effect size in the afore-mentioned example may determining the sample size for a variable ex-
be 20 and relative reduction with yoga interven- pressed as a percentage is as follows:
tion is 20/30 or 66.67%. The effect size may be
(1n / N )t 2 ( p q )
estimated by referring to data from previous n
d2
studies. It is pertinent to mention that the larger
the effect size, the smaller the need of sample where
size and in the cases where the effect size is not n 5 sample size
known or is small, the required sample size will N 5 size of the eligible population
be larger. t2 5 square value of the SD score that refers to the
l Variability of measure under study (standard de- area under a normal distribution of values
viation): It is important for the researcher to p 5 percentage category for which we are com-
know the standard deviation (SD) of the par- puting the sample size
ticular parameter in the study population. Stan- q512p
dard deviation is a measure of dispersion or d2 5 square value of one-half of the precision in-
variability in the data. The anticipation of the terval around the sample estimate
variance (SD) in the measures that are being Expressing in words, the formula states that
studied is essential to know the homogeneity sample size is an expression of a finite population
of the population under study. The greater correction factor (1 2 n/N) times the probability
the variance or SD, the larger the sample size, level for this sample occurrence times the variable
because of heterogeneity in the population, or the variability of our variable in the population
whereas the smaller the variance or SD of the divided by the size of the confidence interval that
measure under study, the smaller the size of the we want for our estimate. Let us talk more in
sample for study. The anticipation of the SD depth about these components and then look at
value may be obtained for similar previous an example.
studies/pilot studies conducted in similar or
nearly similar population. Finite Population Correction
l Prevalence rate: It is important to know the esti- The finite population correction (fpc) is an ad-
mated prevalence rate of the particular event or justment factor that becomes part of the formula
phenomenon under study in the study popula- when sample elements are selected without re-
tion. Prevalence rate of the particular event or placement. However, the fpc has very little effect
phenomena may be estimated from previous on the end result when the size of the sample is
reported studies or the report of the pilot study less than 5% of the total population. In most
or a rough guess by logical reasoning. sample surveys, this is the case and fpc is excluded.
l Other factors that can be considered while calcu- In the example given on p. 139, we exclude the fpc;
lating the final sample size include the expected we have more to say about the relationship be-
drop-out rate, an unequal allocation ratio, and tween sample size and population size in a later
the objective and design of the study. section.
This textbook presents different statistical
formulas and examples to estimate the sample Probability Level
size for observational descriptive studies, expe- The formula now has three components to solve
riential studies/clinical trials and diagnostic the equation. We set the value for two components
studies. (probability level and confidence interval) and we
approximate the third (variance). The value for t
SAMPLE SIZE DETERMINATION FOR probability level is the SD score that expresses the
OBSERVATIONAL DESCRIPTIVE STUDY percentage of a variable’s value that falls within a
In a hypothesis testing descriptive study, an ap- set interval when the variables are normally dis-
proximate formula (W.G. Cochran, 1977) for tributed. One SD includes approximately 68% of
138 Research Methodology and Biostatistics

the sample value and its score is 1.0; two SDs in- estimating the variance is to conduct a small pilot
clude approximately 95% of the sample value and study. Depending on one’s resources, a short
its score is 1.96 and three SDs include approxi- phone call to a random sample of 25–50 house-
mately 99% of the value and its score is 2.8. By holds should provide a reasonable estimate. In
using these values and their score, we assume our this example, p would be the proportion of adults
variables are normally distributed. When we set t, who smoke and q would be 1 – p (minus that per-
we are setting a probability level such that out of centage of adults who smoke).
100 samples of the size that we are going to con-
duct, we want to be x per cent confident (we set x) Confidence Interval
that our sample confidence interval (d) includes The third component of the formula, which is
the population value. determined by the researcher, is the confidence
interval or d. This interval is the margin of error
Variance that we require or will tolerate. The only way,
The formula is for a variable expressed as a per- theoretically, to estimate a population value with-
centage. Typically, this means that we have a vari- out sampling error is to include or interview every
able expressed as two categories; those who do element in the population. When we interview a
and those who do not or those who have and sample of cases, we can only estimate the popula-
those who do not have. For example, we may want tion value within a range of values. The compo-
to estimate the percentage of the population that nent d is expressed as plus or minus and repre-
is suffering from diabetes, hypertension, smokers sents one-half of the range. Thus, if we assume
and use public transport. Each of these variables is that 30% of the adults in Ludhiana smoke, d is
expressed as a percentage. We may be interested in how large or small we need the estimate range to
a variable that has response categories of strongly be to reach 30%. As we show shortly the smaller
agree, disagree and strongly disagree. We can di- the range, the larger is the required sample size.
chotomize this variable by deciding what category Let us go back over the components of the for-
or categories we are interested in determining the mula and then go through an example. The vari-
sample size for. If we were interested in the pro- ance is the value we must guess or estimate. The
portion that agree with a statement, we would percentage of population with characteristics is
combine into one category must sum to 1.0. The p and the remaining percentage of the population
variance of percentage variables is the product of is q. This is at probability level t, which we select,
the two percentages. and that represents in how many repeat samples of
If we want to carry out a study to estimate the 100 the total of the sample size occurs and the
number of adults smoking in Ludhiana, to deter- population value is likely to fall within the specified
mine the sample size, we need to come up with a interval. The confidence interval d is the width of
good guess or estimate of the percentage of smok- the range in which we want our estimates to fall.
ers before we do the study. This estimate is a nec- Let us end the example estimating the number
essary component of the formula. While this of adult smokers in Ludhiana. The question is
seems incongruous with our task, there are a ‘what sample size do we need if we assume the
number of methods that can be used to make this following components’. Let us assume that Ludhi-
estimate. Another method is to make a conserva- ana is similar to Delhi. The population in the
tive guess at the proportion of smokers. As we proportion of adults who smoke is based on our
show shortly, the variance component does not af- assumptions that Ludhiana is home to a number
fect the solution of the formula as much as the of major industries and has lots of migratory la-
confidence interval. Thus, an incorrect guess does bour population. We assume that these factors
not affect the required sample size as much as add to the smoking population, and believe that
the setting of the precision of the confidence in- the percentage of adult smokers in Delhi is a good
terval around the estimate. The third method for estimate for Ludhiana too.
 Sampling and Sample Size Determination 139

We assume p 5 0.30 and, therefore, q is 1 – 0.30 5 t 5 2.8; it is the SD score for 99% set interval (as men-
0.70. We want our probability or confidence level to tioned earlier in probability level)
be 95%, i.e. to include the population value 95 of p 5 assumed or estimated proportion of adults hav-
every group of 100 in the sample size. And we want ing diabetes in Ludhiana as estimated, 20% (0.20)
the confidence interval to be 65%. In summary, the based on the findings of previous study in a similar
city like Delhi
problem is stated as follows: what sample size do we
q 5 1 – p (1 – 0.20 5 0.80)
need if we want to be 95% confident that the popu- d 5 confidence interval, i.e. 1% (0.01)
lation value lies in the interval between 25% and
35%? Substituting these values into the formula (2.8)2 (0.20  0.80) 7.84  0.16
gives n 
(0.01)2 0.0001
(1n / N )t 2 ( p q ) 1.2544
n   12,544
d2 0.0001

where The researcher will require a sample size of 12,544 sub-

(1 2 n/N): We are not taking this into consider- jects from the population. This estimate of the sample
ation, because it has very little effect on sample may vary with the following situations:
size estimation • At least 10% extra sample is considered to take
care of any errors in estimation, because the
t 5 1.96; it is the SD score for a 95% set interval estimate is based on a Delhi study, which
p 5 assumed or estimated proportion of adults itself could be in error. This makes it 12,544 1
who smoke in Ludhiana as estimated, 30% (0.30) 1255 513,799.
q 5 1 – p (1 – 0.30 5 0.70) • If the researcher expects that 25% subjects will be
d 5 confidence interval, i.e. 5% (0.005) nonrespondent or dropouts, then the sample size
After putting the afore-mentioned values in the inflates to 13,799 1 25% 5 17,249.
formula, calculations are as follows: • If the researcher is interested in identifying the
separate prevalence among male and female, then
(1.96)2 (0.30  0.70) 0.8067 the sample would be approximately double the
n   323 estimation, i.e. 34,498; this is a rough estimate; to
(0.05)2 0.0025
have an exact estimate, the researcher has to find
The number of interviews we need is 323. Re- separate studies for prevalence of diabetes among
member that our estimate of the variance was a male and female.
guess or a rough estimate. If we do 323 interviews • The sample size further has to multiply if the
and find that our estimate is wrong, what effect researcher wants to get separate estimates of
prevalence for different age groups. Further-
does that have on our calculations?
more, if more number of variables are present
in the study, the sample size has to be inflated
accordingly.
• This estimated sample uses a random sampling
Example of Estimating Sample Size
technique to select the sample. If the sampling
for a Descriptive Study
technique is not simple random due to heteroge-
A researcher wants to conduct a descriptive study on neity of the population, the sample size will be
the prevalence of diabetes type-II in city Ludhiana. further multiplied considering a design effect of
Estimate the sample size the researcher will consider to 23 estimated sample.
generalize the findings.

(1n / N )t 2 ( p q )
n
d2
SAMPLE SIZE ESTIMATION FOR EXPERIMENTAL
where STUDY/CLINICAL TRIAL
(1 – n/N): We are not taking this into consideration be-
cause it has very little effect on sample size estimation
In experimental study while testing the hypoth-
esis that the difference between two groups, such
140 Research Methodology and Biostatistics

as experimental group and control group, is sig- Z12b : The power the researcher considers for de-
nificant or not, the following formula may be tecting the difference, which is considered as
used: 0.10 in this study. Z0.90 5 1.28 from the Gaussian
table
( Z1  2  Z1 )2 s1 5 SD of the outcome variable in group I 5 11.3
n  ( 1  2 )2 mm Hg
(m1 m2 )2
s2 5 SD of the outcome variable in group II 5 13.0
mm Hg
where m1 – m2 5 mean difference of the outcome variable
Z12a/2: The level of significance for two tails, in group I and group II 5 3.0
when the researcher considers that intervention
may have a positive or negative effect on the (1.96 1.28)2
n  (11.3 13.0)2 347
outcome variable. Otherwise, one tail effect is (3.0)2
considered only in one direction (Z12a). Refer
Therefore, a minimum of 347 subjects in each group
to Box 6.1.
would be required to be included in this study. This calcu-
Z12b: The power the researcher considers for de- lation assumes the subjects must be randomized between
tecting the difference, which is generally consid- the two groups. There may not be any nonresponse in
ered 80% or 90%. Z0.90 5 1.28 from the Gaussian such kind of trials. Other considerations will be as per that
table (refer to Box 6.2). mentioned in the example of estimating sample size for
s1 5 SD of the outcome variable in group 1 descriptive study.
s2 5 SD of the outcome variable in group 2
m1 5 mean of the outcome variable in group 1
m2 5 mean of the outcome variable in group 2

SAMPLE SIZE ESTIMATION FOR DIAGNOSTIC

STUDIES
Example of Estimating Sample Size for
an Experimental Study
New diagnostic tests are developed by biomedical
researchers to achieve improved accurate disease
A researcher wants to conduct an experimental study diagnosis and reduce cost, risk, inconveniences
to assess the effect of lowering sodium in diet on and time of test, where estimation of sensitivity
systolic blood pressure (SBP) among hypertensive
and specificity are two important components
patients visiting the OPD of AIIMS, Rishikesh, India.
Based on a similar research conducted at a differ-
that measure the inherent validity of a diagnostic
ent place or a pilot study, the SD of SBP in the group test compared to the gold standard. The new tests
with a low sodium diet is 11.3 mm Hg, while in that are developed and their sensitivity and specificity
with high sodium diet, it was 13.0 mm Hg. If a 5 0.05 are measured in cases where the gold standard test
and b 5 0.10, how many persons from each group is more invasive, riskier, takes more time to pro-
should be studied, if one wants to detect 3.0 mm Hg duce results and is not practically convenient to be
difference in SBP in the two groups? adapted in practice. For example, the conformity
To estimate the sample size the researcher will
consider to generalize the findings:
diagnosis of pancreatic cancer can only be estab-
lished by performing laparotomy of alive or au-
( Z1  2  Z1 )2 topsy of dead patient; which is practically not
n  ( 1  2 )2
(m1 m2 )2 convenient to be adapted in routine practices. In
such cases, it becomes necessary to develop a new
where test to be used in place of the gold standard test.
Z12a/2: The level of significance, which is considered However, it is essential to have an adequate sample
as 0.05 for two trials because the researcher con-
size to ensure that the study yields estimated sen-
siders that salt intake increases or decreases the
SBP. Z0.95 5 1.96
sitivity and specificity with acceptable level of
precision. For such diagnostic studies, the sample
 Sampling and Sample Size Determination 141

size may be calculated using Buderer formula as by R.K. Malhotra and A. Indrayan (http://www.ojo.
given here: in/text.asp?2010/58/6/519/71699).
Z12  2  SN (1 SN )
Sample size (n) based on sensitivity 
L2  Prevalence
Example of Estimating Sample Size
for a Diagnostic Study
Z12  2  SP (1 SP )
Sample size (n) based on specificity  A biomedical researcher wants to conduct a diag-
L2  (1 Prevalence)
nostic study to estimate the sensitivity and specific-
ity of the GeneXpert (a new diagnostic test) against
where Sputum culture for ABF (Gold standard test) for the
n 5 required sample size diagnosis of tuberculosis in patients visiting the OPD
of AIIMS, Jodhpur, India.
SN 5 anticipated sensitivity
The researcher assumes anticipated sensitivity of
Sp 5 anticipated specificity the new test (SN) 5 0.80, precision (L) 5 0.03 with
a 5 size of the critical region (1 2 a is the confi- 95% confidence level (two-tailed) and expected
dence level) prevalence of TB 5 0.20.
Z1 – a/2 5 standard normal deviate corresponding
Z12  2  SN (1 SN )
to the specified size of the critical region (a) Sample size (n) based on sensitivity 
L2  Prevalence
L 5 absolute precision desired on either side
(half-width of the confidence interval) of sensi- where
tivity or specificity Z21 – a/2 5 level of significance, which is considered as
Note: 0.05 for two trials
l Anticipated sensitivity or specificity of the diag-
Z0.95 5 1.96
SN 5 anticipated sensitivity, i.e. 0.80
nostic test can vary from 0.70 to 0.97. A test L 5 absolute precision, i.e. 0.03
with anticipated sensitivity or specificity less Prevalence of TB 5 20% (0.20)
than 0.70 may not be worthy of investigations.
l Absolute precision could be generally consid- (1.96)2  0.80 (1 0.80)
  3413
ered as 0.03, 0.05, 0.07 and 0.10. (0.03)2  0.20
l Anticipated sensitivity should be such that after
Furthermore, researchers assume anticipated speci-
adding the required precision, it does not ex- ficity of new test (SP) 5 0.80, precision (L) 5 0.05 with
ceed 1. For example, when anticipated sensitiv- 95% confidence level (two-tailed) and expected
ity is 0.96, a researcher cannot select precision prevalence of TB 5 0.20.
to be more than 0.04.
Z12  2  SP (1 SP )
l The sample size may be calculated considering Sample size (n) based on specificity 
L2  (1 Prevalence)
either anticipated sensitivity or specificity or
both, depending on the intent of the diagnostic where
study. However, when the researcher considers Z21 – a/2 5 level of significance, which is considered as
both sensitivity and specificity important for a 0.05 for two trials
particular study, he must take the larger sample Z0.95 5 1.96
SP 5 anticipated sensitivity, i.e. 0.80
size calculated either for sensitivity and specific-
L 5 absolute precision, i.e. 0.05
ity. For example, calculated sample came out to Prevalence of TB 5 20% (0.20)
be n 5 140 based on anticipated sensitivity and
n 5 290 based on specificity; then the researcher (1.96)2  0.80 (1 0.80)
  854
must consider 290 as minimum sample size, if (0.03)2  (1 0.20)
considering both sensitivity and specificity im-
portant for the particular diagnostic study. Therefore, a minimum of 3413 patients is needed for
estimating the sensitivity and 854 patients for estimating
The sample size for diagnostic studies also may be the specificity of the new diagnostic test. If the researcher
estimated using the simple normogram developed
142 Research Methodology and Biostatistics

is interested in both sensitivity and specificity, then he or sampling strategy. Quantitative studies require a
she must consider a minimum sample size of 3413. This larger sample than qualitative studies. The sam-
calculation assumes that the patients will be selected ple size variation is also considered in qualita-
using random sampling techniques and must be ran- tive studies. For example, a longitudinal study
domly allocated under both the tests. Other factors design requires larger sample size than cross-
such as drop-out rate, an unequal allocation ratio and sectional studies, because longitudinal studies
the study design effect also must be considered for
may lose subjects over time. Stratified random
estimating the final sample size as mentioned in the
example of estimating sample size for observational sampling and quota sampling techniques, for
descriptive study. example, allow the researcher to use smaller
sample size than would be needed in studies
employing simple random or convenience sam-
SAMPLE SIZE CALCULATION USING COMPUTER pling, because some of the representativeness is
The beginner health care professional may find already build into the stratification procedure.
the sample size using different sample size calcu- l Resources available: A large sample may ensure
lators. For example, he or she may use the user precision, but it may prove to be costly as it may
friendly, free webpage called Raosoft (http:// make the researchers realize at some time or the
www.raosoft.com/samplesize.html) for sample other that they are running short or out of
size calculation. This page gives an example of money and other resources. So before carrying
how preformatted software may be used to deter- out a study with a large sample, researchers
mine the sample size for new and uninitiated need to decide if they have sufficient resources.
health care professionals (Fig. 6.7). The clinical l Nature of study: The sample size depends upon
researchers may use the sample size calculator the type of study to be carried out. Small sam-
(http://www.sample-size.net/about-this-site/) of ples can make intensive and longitudinal studies
UCSF Clinical and Translational Science Institute successful, but for extensive and one-time stud-
for sample size calculations in advanced level ies, large samples are required.
clinical research studies. l Sampling methods used: Smaller but efficiently
selected samples prove to be far better than
FACTORS AFFECTING SAMPLE SIZE badly selected, spurious large samples. Probabil-
l Design effect: The study design factors influenc- ity sampling has been able to prove its worth
ing the sample size include the type of study and in research more, compared to nonprobability
the number of variables under study and the sampling.

Fig. 6.7. ​A facsimile of the software Raosoft, which may be used freely by health care professionals.  (Courtesy:
Shanti R. Rao and Potluri M. Rao, ‘Sample Size Calculator’, Raosoft Inc., 2009. ,http://www.raosoft.com/samplesize.html.)
 Sampling and Sample Size Determination 143

l Homogeneity: If the population is homoge- that the process has on the outcome. Data collec-
neous, then a small sample may be sufficient. tion instruments in which measurement error is
Since in a homogeneous population, answers minimal are said to be precise. The less precise
are going to be close, taking a large sample the instrument, the larger the sample size needed.
would be unwise. Because interval-level data are generally more
l Effect size: If the relationship between indepen- precise, from a sample size perspective, it is best
dent and dependent variables is strong, then a to measure at that level if at all possible, because
small sample will be sufficient. For example, to a smaller sample size may be used.
find the effect of drug abuse on family disorga-
nization, a sample that is small will be enough.
PROBABILITY AND SAMPLING
l Degree of accuracy desired from the estimate:
Precision is the limit of tolerable errors in the ERRORS
sample estimates. For example, a study is con- Sampling is an act of extracting a representative
ducted on the number of patients satisfied with part of population for determining characteristics
the diet provided at the hospital. In this study, of the whole population. A sample is expected to
all that the researcher wants is for the number of represent the population just like salt’s taste on
people satisfied to be within . 50 of the true everybody’s taste buds. However, it may not be
population (patients on hospital diet). necessary that each sample will be representative
l Degree of confidence: Confidence refers to our of the population it is derived from. This differ-
desire that we do not want the error to go ence is the sampling error which arises when just
beyond a certain limit. Therefore, the higher the sample is under study and not the population
the degree of confidence, the larger is the sam- itself. Sampling error is the deviation of the selected
ple size required. For example, if we need a de- sample from the true characteristics, traits, behav-
gree of confidence of 98%, then it means we do iours, qualities or figures of the entire population.
not want the error to be beyond 2%.
l Cooperation and attrition: Attrition refers to REASONS FOR SAMPLING ERRORS
the condition when people initially willing to l Sampling process error occurs because research-
participate fail to continue participation and ers draw different subjects from the same popula-
leave the sample. If the data are collected from tion, but the subjects have individual differences.
multiple points and there is less chance of com- Keep in mind that when a researcher selects a
munication of the researcher with the sample sample, it is only a subset of the entire population;
due to long time gap, there can be a possibility therefore, there may be a difference between the
of attrition, so sample size should be selected sample and the population.
including expected refusals. l The most frequent cause of the said error is a
l Subgroup analysis: If the hypotheses are to be biased sampling procedure. Every researcher
tested not only for the population, but also for must seek to establish a sample that is free from
specific subgroups, then the sample size should bias and is representative of the entire popula-
be large enough to make the generalization of the tion. In this way, the researcher is able to mini-
results possible to those subgroups too, for ex- mize or eliminate sampling error.
ample, if we want to study the use of an internet- l Another possible cause of this error is chance. The
equipped library for nursing students, and then process of randomization and probability sam-
utilize the same results for final-year students pling is done to minimize sampling process error,
who are expected to carry out internet searching but it is still possible that all the randomized sub-
exercises. jects are not representative of the population.
l Measurement factors: Measurement factors l The most common result of sampling error is
that influence the sample size include the sensi- systematic error wherein the results from the
tivity of the research instruments and the effect sample differ significantly from the results from
144 Research Methodology and Biostatistics

the entire population. It is logical that if the sam- l Exclusion bias: This type of bias happens when
ple is not representative of the entire population, some people of the group are eliminated from
the results from it will most likely differ from the the study as the instance of day scholars men-
results taken from the entire population. tioned earlier.
Two basic reasons for sampling error: l Healthy user bias: This type of bias occurs
l Chance error: The error occurs by chance. For when the sample selected has more likelihood to
example, someone did a comparative study on be healthier compared to the general popula-
malnutrition in fewer than five children in two tion. For example, if the sample is selected from
cities A and B. Unfortunately, city B had a large students who take food from the mess and not
number of slum dwellers, so it comprised a the canteen where junk food is available.
large number of malnourished children, thus
skewing the result. SAMPLE SIZE AND SAMPLING ERRORS
l Sampling bias: Sampling bias is a tendency to Given two exactly same studies, same sampling
favour a selection of sample units that possess methods, same population, the study with a larger
particular characteristics. It may occur in the sample size will have less sampling process error
form of overrepresentation bias. compared to the study with smaller sample size
l For example, a study is done on nursing stu- (Fig. 6.8). Keep in mind that as the sample size
dents’ satisfaction with staying at a hostel or as increases, it approaches the size of the entire
a paying guest. This study is biased towards population. Therefore, it also approaches all the
those students who are hostellers or paying characteristics of the population, thus, decreasing
guests, but it excludes students who come from sampling process error.
their own homes or who are day scholars.
l Another way to understand this is if we do a MINIMIZING SAMPLING ERRORS/BIAS
study on the time spent by students on internet. There is only one way to eliminate this error. The
The students who participate will be those who solution is to eliminate the concept of sample, and
have an email address. This study is biased to- to test the entire population. In most cases, this is
wards those who have an email address, but the not possible; consequently, what a researcher must
ones who do not have an address are excluded. do is to minimize sampling process error. This can
be achieved by a proper and unbiased probability
TYPES OF SAMPLING BIAS sampling and by using a large sample size. How-
l Self-selection bias: This type of bias happens ever, to minimize sampling bias, the following
in a situation when the participants in the things should be done:
study have some kind of control over the study l Avoid convenient or judgement sampling.

to participate or not. For example, if a study is l Ensure that the target population is well defined

conducted on the number of people who can and the sample frame matches it as much as
carry a load of 10 kg for 20 min, then only possible.
well-built people will have a preference to par- l When the complete population cannot be sam-

ticipate. pled, care should be taken that the population

Sampling Sampling
Sample error Sample error

Population Population

Fig. 6.8. ​The larger the sample, the lesser the chances of sampling errors.
 Sampling and Sample Size Determination 145

that is excluded is not taking away the desired to poor sample selection and ultimately affects
features, which were supposed to be measured the study results.
from the population. l Lack of resources: Resource allocation espe-
cially for nursing research studies is very poor in
a country like India, where it becomes very dif-
PROBLEMS OF SAMPLING ficult for a researcher to draw a representative
There are several problems encountered by the sample in the absence of adequate resources.
researcher during the sampling process; some of l Lack of cooperation: On several occasions, it is
the important problematic issues regarding the observed that study subjects do not cooperate
sampling process are as follows: with researchers, which ultimately affects the
l Sampling errors: Sampling error is the deviation sampling process and creates a problem for the
of the selected sample from the true characteris- researchers. For example, a researcher selected a
tics, traits, behaviours, qualities or figures of the sample, with simple random sampling and
entire population. Sampling process error occurs found that several subjects are not interested in
because researchers draw different subjects from participating in the study; this ultimately dis-
the same population, but the subjects have indi- turbs the sampling process and causes an addi-
vidual differences. A sample is only a subset of tional burden to the researcher.
the entire population; therefore, there may be a l Lack of existing appropriate sampling frames
difference between the sample and population. for larger population: It is generally observed
l Lack of sample representativeness: Selection of that in developing countries including India, it is
a representative sample is not an easy task; fur- very difficult to get existing sampling frames,
thermore, it becomes more difficult when the which generally leads to problems in the sam-
population under study is heterogeneous and pling process. For example, a nurse researcher is
the sampling process is carried out without con- interested in sample selection from nurses pres-
sidering the nature of the population. ently practising in Punjab, but the Punjab Nurses
l Difficulty in estimation of sample size: There is Registration Council does not make provision of
lack of availability of population parameters, es- renewal of registered nurses licences, and there-
pecially in our own scenario. In the absence of fore, one cannot get a list of registered nurses
population parameters, it becomes very difficult presently practising as nurses in Punjab.
to estimate the sample size. Studies conducted on l Callous approach of the researcher towards
an undetermined sample size do not facilitate the sampling process: Generally, it is observed
generalization of the findings for the particular that researchers pay very little attention for ap-
population. propriateness of the sampling method for a
l Lack of knowledge about the sampling pro- particular phenomenon under study. Further-
cess: It is generally observed that researchers more, they casually choose a particular type of
usually carry superficial knowledge about sam- sampling technique to select a sample rather
pling methods and sampling process. This leads than use an effective sampling process plan.

MULTIPLE CHOICE QUESTIONS

1. The small group selected from a larger group to 2. Another term used for cluster sampling is
participate in a study is known as the (a) Systematic sampling
(a) Study population (b) Quota sampling
(b) Sample population (c) Multistage sampling
(c) Sample (d) Network sampling
(d) Element
146 Research Methodology and Biostatistics

3. Which of the following studies would require the 7. Which of the following principles is followed to
largest sample size? determine the sample size for qualitative research?
(a) Correlational (a) The ‘thumb rule’
(b) Ethnography (b) The ‘rule of 30’
(c) Grounded theory (c) Data saturation
(d) Phenomenological (d) Power analysis

4. A researcher has decided to conduct a satisfaction 8. In quantitative research studies, the power analysis
survey among all the patients who presented to the is used to
emergency department over a 2-month period of (a) Set the level of significance
time. This is an example of (b) Test the internal validity
(a) Stratified random sampling (c) Estimate the sample size
(b) Cluster sampling (d) Determine the reliability
(c) Convenience sampling
(d) Purposive sampling 9. The process of selecting a subset of population for
a survey is known as
5. A total of 40 nurses are randomly selected from a (a) Survey research
random sample of five large hospitals of a state. This (b) Representation
is an example of (c) Triangulation
(a) Simple random sampling (d) Sampling
(b) Convenience sampling
(c) Cluster sampling 10. In which of the following nonrandom sampling
(d) Purposive sampling techniques does the researcher ask the research
participants to identify other potential research
6. The sample drawn from which of the following sam- participants?
pling techniques will be the least likely to be the (a) Snowball
representative of the overall population? (b) Convenience
(a) Quota sampling (c) Purposive
(b) Stratified random sampling (d) Quota
(c) Systematic sampling
(d) Convenience sampling

Answers of Multiple Choice Questions

1. c; 2. c; 3. a; 4. c; 5. c; 6. d; 7. c; 8. c; 9. d; 10. a.

REFERENCES
Buderer, N. M. (1996). Statistical methodology: I. Incorporating Heckathorn, D. D. (2002). Respondent-driven sampling II: deriv-
the prevalence of disease into the sample size calculation for ing valid estimates from chain-referral samples of hidden
sensitivity and specificity. Academic Emergency Medicine, 3(9), populations. Social Problems, 49, 11–34.
895–900. Kadam P, & Bhalerao S. (2010). Sample size calculation. Interna-
Burden, N. T. (2005). Phenomenon of mothering with HIV. tional Journal of Ayurveda Research, 1(1), 55–57.
Journal of the Association of Nurses in AIDS Care, 16(4), 3–13. Kerlinger, F. N., & Lee, H. B. (2000). Foundation of behavioural
Burns, N., & Grove, S. K. (2005). The practice of nursing research: research. New York, NY: Harcourt Brace.
conduct, critique and utilization (5th ed.). St. Louis, MO: Elsevier. Kindy, D., Petersen, S., & Parkhurst, D. (2005). Perilous work:
Goodman, L. A. (1961). Snowball sampling. Annals of Mathemat- nurses’ experiences in psychiatric units with high risk of assault.
ical Statistics, 32, 148–170. Archives of Psychiatric Nursing, 19(4), 169–175.
Heckathorn, D. D. (1997). Respondent-driven sampling: a new Krejcie R.V. & Morgan D.W. (1970). Determining sample
approach to the study of hidden populations. Social Problems, size for research activities. Educational and psychological
44, 174–199. measurements, 30, 607–610.
 Sampling and Sample Size Determination 147

Malhotra, R. K., & Indrayan, A. (2010). A simple normogram for Simpson, K. R., & James, D. C. (2005). Effect of immediate versus
sample size for estimating sensitivity and specificity of medical delayed pushing second-stage labor on fetal well-being: a ran-
tests. Indian Journal of Opthalmology, 58, 19–22. ,http://www.ojo. domized clinical trial. Nursing Research, 54(3), 149–157.
in/text.asp?2010/58/6/519/71699/. Accessed 22.01.16. Solutions for your life. University of Florida. <http://edis.ifas.ufl.
Meeker, M. A. (2004). Family surrogate decision making at the edu/pd006> Accessed 06.02.14.
end of life: seeing them through with care and respect. Qualita- Tabo, L. A. (1995). Principles and practices of nursing research.
tive Health Research, 14, 204–225. Boston, MA: Mosby.
More, J. M. (2000). Determining sample size. Qualitative Health Ukande, U. (2001). Population, sampling and data collection. In:
Research, 10(1), 3–5. Proceedings of Workshop of Nursing Research Society of India.
Rew, L. (2003). A theory of taking care of oneself grounded White, D., & Nichols, B. (2003). Prediction of nurses’ acceptance
in experience of homeless youth. Nursing Research, 52(4), of an intravenous catheter safety device. Nursing Research,
234–241. 52(4), 249–255.
Salganik, M. J., & Heckathorn, D. D. (2004). Sampling and estima- Wongvatunyu, S., & Porter, E. J. (2005). Mother’s experience of
tion in hidden populations using respondent-driven sampling. helping young adult with traumatic brain injury. The Journal of
Sociological Methodology, 34, 193–239. Nursing Scholarship, 37(1), 48–56.
CHAPTER OUTLINE
Measurements and
Methods of Data
Collection 7
INTRODUCTION BIOPHYSIOLOGICAL METHODS

SCALES OF MEASUREMENTS PROJECTIVE TECHNIQUE

CONCEPT OF DATA COLLECTION Q-SORTS

QUESTIONNAIRE VIGNETTES

INTERVIEW AS METHOD OF DATA COLLECTION STRATEGIES TO MINIMIZE MEASUREMENT ERRORS

COMPOSITE SCALES/ATTITUDE SCALES VALIDITY OF RESEARCH TOOL

• Likert scale RELIABILITY OF RESEARCH TOOLS
• Semantic differential scale
• Visual analog scale PILOT STUDY
OBSERVATION AS METHOD OF DATA COLLECTION MULTIPLE CHOICE QUESTIONS
• Rating scale
• Checklist

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Discuss the in vivo and in vitro biophysiologic methods
• Identify the concept of data collection and factors of data collection.
influencing the selection of data collection methods. • Determine the use of projective technique, Q-sorts and
• Recognize the different measurement scales. vignettes as methods of data collection.
• Describe the interview, questionnaire, composite scales • Appraise the importance of pilot study.
as methods of data collection.
• Appraise the methods of data collection and tools used
for observation, such as rating scale, checklist anecdotes,
videotapes, films and CCTV.

KEY TERMS

• Anecdotes 152 • Interview schedule 160 • Rating scales 168

• Attitude scales/composite • Likert scale 162 • Reliability 180
scales 161 • Observation 165 • Semantic differential scale 163
• Checklists 170 • Opinionnaire 152 • Validity 179
• Data 149 • Pilot study 183 • Vignettes 177
• Focus group discussion 178 • Projective technique 174 • Visual analogue scale 165
• In vitro 173 • Q-sorts 176
• In vivo 172 • Questionnaire 153
 Measurements and Methods of Data Collection 149

another in some way. In other words, they have

INTRODUCTION no quantitative values.
Data are the observable and measurable facts that l Its categories are mutually exclusive and
provide information about the phenomenon un- exhaustive.
der study. In research studies, two types of data are
collected: primary and secondary. However, bio- Examples of Nominal-Level Measurement
medical research studies rely mainly on primary
data. A systematic collection and analysis of data • Gender
Male
are most vital to any empirical research. It is most
Female
essential for a researcher to have appropriate meth- • Habitat
ods of data collection so that planned study vari- Urban
ables can be adequately measured and accurate Rural
conclusions can be drawn. A variety of data collec- Slums
tion methods are used in biomedical research
studies, such as interview, questioning, observa-
tion, biophysical measurements, psychosocial ORDINAL-LEVEL MEASUREMENT
measurement scales and record analysis. The ordinal-level measurement is described as
follows:
l It ranks objects based on their relative standing
SCALES OF MEASUREMENTS on a specific attribute.
Measurement is central and essential to the process l The order of ranking is imposed on categories.

of obtaining data. The term measure means ‘to as- l The numbers must preserve some level of order

certain the dimension of quantity or capacity to but do not need to be precise.

mark off’, usually with reference to some unit of l It reflects only magnitude and does not equal

measurement. Measurement is the assignment of intervals or an absolute zero point.

numbers to objects according to specific rules, to
characterize quantities of an attribute. Measure- Examples of ordinal-level measurement
ment consists of a set of rules for assigning numeric
values to objects to represent varying degrees of • Health status
some attributes that are being measured. It is a sys- Poor
Fair
tem of classifying measurements according to the
Good
nature of the measurement and the type of math- Excellent
ematical operations to which they are amenable. It • Income status
is very important to understand the scales or levels Low income
of measurements before learning about the analysis Middle income
of data. There are four levels of measurements: Upper income
nominal, ordinal, interval and ratio levels.

NOMINAL-LEVEL MEASUREMENT INTERVAL-LEVEL MEASUREMENT

The nominal-level measurement is described as The interval-level measurement is described as
follows: follows:
l It is the lowest of the four levels of measure- l There is specification of ranking of objects on an

ments. attribute of the distance between those objects.

l It involves the assignment of numbers to repre- l There is, more or less, equal numerical distance

sent categories or classes of things. between intervals.

l It consists of categories that are not more or l There is no absoluteness because there is no

less than each other but are different from one zero point or absence of concept.
150 Research Methodology and Biostatistics

Examples of interval-level measurement CONCEPT OF DATA COLLECTION

• Temperature Data are the observable and measurable facts that
10–80°
provide information about the phenomenon under
40–50°
study. For example, when physical growth of an in-
fant is the phenomenon under study, the data re-
RATIO-LEVEL MEASUREMENT quired to measure physical growth would be the
The ratio-level measurement is described as body weight, height, chest and head circumferences.
follows: In other words, using the data and information, in-
l It is the highest level for measurement.
dicators are developed, which are the reflection of
l It has all the three attributes: magnitude, equal
indirect or partial measures of a real situation. The
intervals and absolute zero point. aim of gathering and summarizing data is to trans-
l It represents continuous values.
form this into meaningful information in order to
identify and measure variables, describe behaviour
and compute empirical evidence that is objective,
Examples of ratio-level measurement reliable and valid.
• Biophysical parameters
There is a pertinent question that always strikes
Weight: ……… cm the researcher’s mind: How to determine what
Height: ……… kg data are to be collected? The following discussion
Volume: ……… L provides the answer to this question. For example,
Blood pressure: ……… mm Hg to study the quality of perinatal care services
Hb level: ……… g/dL provided in a rural area, the first step is to opera-
tionally define perinatal care and its different com-
ponents. The perinatal period for a mother is
Examples of nominal- to ratio-level measurements divided into three phases: antenatal, natal and
I. What is your place of habitat? postnatal periods. The indicators of antenatal care
Rural are early detection of pregnancy, tetanus toxoid
Suburban (TT) immunization, identification of high-risk
Urban mothers, regular antenatal check-ups and supply
This is a nominal-level measure because there is no of supplements. Therefore, data collection should
way to quantify the answer. aim at identification of observable and measurable
II. Do you consider yourself:
Teenager
facts or variables that would relate to these indica-
Young adult tors. The questions that need to be answered are as
Middle-aged adult follows: What are the indicators? How will they be
Senior citizen measured or recorded? What is the analysis
This is an ordinal-level measure because data are planned? The five ‘Ws’ of data collection are as
ranked one category higher or lower than the follows:
other. l What data are to be collected?
III. How old are you? l From whom data are to be collected?
25–34
l Who will collect data?
35–44
l From where the data will be collected?
45–54
55 or older l When are the data to be collected?

This is an interval-level measure because numerical

distances are equal on the scales. DATA SOURCES
IV. What is your height and weight? Information collected from different research
Height: ……… cm; weight: ……… kg
studies generally depends on various sources.
This is a ratio-level measure.
However, a quality research study requires that
 Measurements and Methods of Data Collection 151

Table 7.1. ​Sources of data collection

Primary Sources Secondary Sources
Internal Sources (Private Documents) External Sources (Public Documents)
Published Records Unpublished Records
• People, objects, pro- Biographies Journals and magazines Unpublished thesis
grammes, institutions, etc. Diaries Newspapers Unpublished dissertations and
(Primary data are collected Letters Government reports reports
through interviews, Memoirs Statistical abstracts Official or patient records
questioning, observation, Census reports
biochemical measure- Mass communication
ments and psychosocial Commission reports
measurement scales)
Note: Nursing research studies primarily depend on primary sources.

highly reliable and valid data are collected; there- biographies, personal diaries, letters, memoires
fore, diligence and application of the researcher and so on.
can be of high importance. Sources of data collec-
tion in different research studies largely depend METHODS AND TOOLS OF DATA COLLECTION
on several factors, such as type of research study, The methods and tools of data collection are the
phenomenon under study and purpose of the strategies and system used for gathering informa-
study. However, basically, sources of data are gener- tion from study subjects/participants. Let us un-
ally categorized into two broad categories, namely derstand terms related to methods, techniques
primary and secondary sources (Table 7.1), as dis- and tools of data collection.
cussed here: l Methods of data collection: The various steps or

l Primary sources: Primary data are directly col- strategies used for gathering and analysing data
lected from the research units, which may be in a research investigation are known as meth-
individuals, objects, programmes or institutions. ods of data collection.
Primary sources provide first-hand information l Techniques of data collection: The means of

collected by the researcher directly from the re- gathering data with the use of specific tools used
spondents or situations. They may be collected in given methods are known as techniques of
through interviews, questioning, observation, data collection.
biochemical measurements and psychosocial l Instruments/tools for data collection: A re-

measurement scales. search instrument is a device used to measure

l Secondary sources: Secondary data are collected the concept of interest in a research project that
from either internal or external secondary a researcher uses to collect data.
sources. External sources involve existing materi- A brief classification of different types of methods
als, such as published or unpublished records. and tools of data collection is depicted in Table 7.2.
Published records may include journals, maga-
zines, newspapers, government reports, statistical SELECTION OF METHODS OF DATA COLLECTION
abstracts, census reports, constituents of mass Complete and appropriate answer to a research
communication and commission reports. Un- question largely depends on the selection of appro-
published records may include official records, priate methods of data collection, which generally
patient records, thesis, dissertations and reports. begin at the time of selection of a research design.
In addition, internal secondary sources – A research question or hypotheses may be an-
also known as private documents – may include swered by using one or more number of methods
152 Research Methodology and Biostatistics

Table 7.2. ​Types of methods and tools of data collection

Types of Method/ Techniques Tools for Data Collection
1. Interview Interview schedule
Opinionnaire
2. Questioning (self-report) Questionnaire
Opinionnaire
Attitude scales/composite scales (e.g. Likert scale and semantic differential scale)
Visual analogue scale
3. Observation Rating scales
Checklists
Anecdotes
Videotapes/films
Closed circuit TV
4. Biophysiologic methods In vivo biophysiologic measurements
In vitro biophysiologic measurements
5. Psychosocial methods Projective technique
Q-sorts
Vignettes

of data collection. However, selection of the appro- objects or institution, questionnaires or inter-
priate method of data collection is influenced by views may not be possible at all, and research-
several factors as discussed here: ers will have to depend mostly on observation
l The nature of the phenomenon under study: to collect relevant data.
The nature of the phenomenon under study l The type of research study: Quantitative and
largely influences the choice of the method of qualitative research studies need different meth-
the data collection. Each research phenomenon ods of data collection. For example, in qualitative
has its particular characteristics and, therefore, research, more in-depth information is required;
needs different approaches and methods of data therefore, focused group interviews or unstruc-
collection. For example, some phenomenon can tured participatory interviews are feasible for
only be studied appropriately through observa- data collection, while for quantitative research
tion, such as clinical practices or processes in studies, more structured interviews, questioning
particular biomedical procedures. Similarly, or observation is used for data collection. How-
knowledge of a group of patients can only be ever, in biomedical research, patients’ data are
assessed through questioning or interviews. mostly biochemical in nature and are mostly col-
Therefore, the nature of the phenomenon under lected through biophysiological methods of data
study significantly affects the selection of a par- collection.
ticular method of data collection. l The purpose of the research study: The pur-
l Type of research subjects: Data collection pose of the study also influences the choice of
methods are also influenced by the type of the methods of data collection, such as in a study
subjects under study. For example, data collec- conducted with the purpose of exploring a phe-
tion from physically or psychologically dis- nomenon, in-depth interviews may be needed
abled subjects can be done either by interviews for data collection, while studies conducted with
or through observation; data collection the purpose of description or correlation of
through questionnaires is not feasible. On the study variables may need more structured meth-
other hand, if data have to be collected from ods of data collection.
 Measurements and Methods of Data Collection 153

l Size of the study sample: When a study is con- of questionnaires may not demand these skills;
ducted on a small sample, interviews or direct however, for the development and construction
observation may be possible, while these methods of a good questionnaire, good writing skills may
can be tedious for large samples. For larger sam- be required.
ples, questionnaires can be better and more pref-
erable methods of data collection. Interviews and CRITERIA OF EVALUATION/ASSESSMENT
observation methods will also be cost-effective OF DATA COLLECTION METHODS
and easy for smaller groups, while questionnaires The appropriateness of the data collection method
will be convenient, easier and cost-effective meth- may be evaluated or assessed by using following
ods of data collection for larger samples. criteria:
l Distribution of the target population: If the tar- l Accuracy and completeness of data collection:

get population is spread in a large geographical Researcher must ensure that the data collection
area, it will not be possible to carry out interviews methods used will yield accurate and complete
or observation, and therefore, mailed question- data to answer research questions or test
naires may be a better option; they will be more hypotheses
convenient and cost-effective in such conditions. l Compatibility with educational level, sociocul-

l Time frame of the study: If a research is con- tural values and beliefs of the subjects
ducted for the long time, it may permit the re- l Cost-effectiveness and speed in data collection

searcher to use less-structured methods of data procedure

collection to gain in-depth information, while l In accordance with the nature of the phenome-

short time-frame studies may not allow the re- non, type, purpose, time frame and resources
searcher to use unstructured methods of data available for the study
collection, where he/she gets very little time for l Furthermore, the following questions may be

data collection and analysis. Therefore, struc- considered while evaluating or assessing the
tured methods of data collection are used more method for data collection:
for short-term research designs.  Is the data collection method complete in all

l Literacy level of the subjects: Illiterate subjects the aspects for study of the study variables?
put constraints on the use of self-responding  Are the data collection methods thoroughly

methods of data collection, such as question- described?

naires. For illiterate subjects, interviews conducted  Are the data collection methods in accor-

in the native language is one of the few possible dance to the research questions/hypotheses to
methods of data collection used, while more var- be tested?
ied and numerous options are available for liter-  Are validity and reliability of the data collec-

ate subjects. tion methods established?

l Availability of resources and manpower: Some  Is the number of methods used for data

of the methods of data collection require more collection sufficient for complete coverage
quantities of resources and manpower, such as of research data or are additional methods
conducting interviews and observation com- required for data collection?
pared to the use of questionnaires. Therefore,  Are anonymity and confidentiality assured?

availability of resources and manpower also af-  Are instruments described in detail?

fects the selection of methods of data collection.  Were the criterion measures or scoring meth-

l Researcher’s knowledge level and competence: ods clearly established?

The researcher’s knowledge and competence also
affects the selection of methods of data collection;
for example, conducting an interview or observa- QUESTIONNAIRE
tion may require special social and psychological Questionnaires are one of the most popular
knowledge, skills and competence, while the use methods of conducting research. They provide a
154 Research Methodology and Biostatistics

convenient way of gathering information from a appropriate answer. Closed-format questions have
target population. They are cost-effective and the following characteristics:
easy to construct and analyse. A questionnaire is l Facilitate easy statistical calculation of data

a planned self-reported research instrument de- l Provide easy preliminary analysis

signed to elicit information though written or l Can be asked to different groups at different

verbal responses of the subjects. In other words, intervals

a questionnaire is a research instrument used for l Facilitate efficient tracking of opinion

data collection. It is a set of questions that may or

may not provide response alternatives to be an- Subtypes of Closed-Format Questions
swered by subject. A. Dichotomous questions: These require the
respondent to make a choice between two
DEFINITIONS OF QUESTIONNAIRE responses, such as Yes/No or Male/Female.
A questionnaire is a structured instrument consist- Example: Q. Have you ever been hospitalized?
ing of a series of questions prepared by the researcher (a) Yes (b) No
that a research subject is asked to complete, to gather B. Multiple choice questions: These questions
data from individuals about knowledge, attitude, require respondents to make a choice between
beliefs, and feelings. more than two response alternatives.
The instrument is called a questionnaire or Example: Q. Which of the following diseases is
sometimes a self-administered questionnaire (SAQ), sexually transmitted?
when respondents complete the instrument them-
selves, usually in a paper and pencil format but (a) Diabetes mellitus (c) Syphilis
occasionally directly onto a computer. (b) Hypothyroidism (d) Hypertension
A questionnaire is a structured self-reported C. Cafeteria questions: These are a special type
paper and pencil instrument that a research sub- of multiple choice questions that ask respon-
ject is asked to complete. dents to select a response that most closely
corresponds to their views.
TYPES OF QUESTIONS Example: Q. What do you think about hormone
Open-Format Questions replacement therapy?
Open-ended questions are those questions that (a) It is dangerous, should be avoided.
provide an opportunity to the respondents to (b) One should be cautious while using it.
express their opinions and answers in their own (c) I am uncertain about my views.
way. Open-format questions have the following (d) It is beneficial, should be promoted.
characteristics: D. Rank order questions: These questions ask
l There is no predetermined set of responses. respondents to rank their responses from most
l They provide true, insightful and unexpected favourable to least favourable.
suggestions. Example: Q. What according to you is most im-
l An ideal questionnaire contains open-ended portant for your life? Rank from most favour-
questions towards the end of all the questions able to least favourable.
that would ask respondents to provide sugges- (a) Money (c) Family
tions for changes or improvements. For exam- (b) Education (d) Health
ple: State your opinion about the quality of health
E. Contingency questions: A question that is
care services in Punjab?
asked further only if the respondent gives a
Closed-Format Questions particular response to a previous question.
Example: Q. Are you stressed?
These questions offer respondents a number of
alternative replies from which the subjects must (a) No (b) Yes. If yes, what are the
choose the one that most likely matches the reasons?
 Measurements and Methods of Data Collection 155

F. Rating questions: These questions ask respon- I like going [] [] [] [] [] I like watching
dents to judge something along an ordered for walks movies
dimension. The respondent is generally re-
quired to provide rating to a specific element J. Matrix questions: Matrix questions include
on a scale selecting among the points from multiple questions, and identical response
poor to good. They may provide a number of categories. Questions are placed one under
choices. another, forming a matrix. Response catego-
Example: Q. How do you rank the education ries are placed along the top.
quality in India? Example: Q. Please let us know your weekly
schedule of the following:
1 2 3 4
Good Fair Poor Very poor Mon Tue Wed Thurs Fri Sat Sun
Gym (weight
G. Importance questions: In this, respondents training)
are asked to rate the importance of a particu- Aerobics
lar issue on a rating scale of 1–5. This helps to
Eating (dinner/
determine the things/issues that are important
lunch)
to a respondent.
Drink (alcoholic
Example: Q. Exercising every day is ……… for
beverages)
health.
1 Extremely important
GUIDELINES FOR DESIGNING A GOOD
2 Very important QUESTIONNAIRE
3 Somewhat important General Points
4 Not very important l The questionnaire must be developed exactly in
5 Not at all important accordance with study objectives.
l The questionnaire should begin with instruc-
H. Likert questions: Likert questions help to de- tions for the respondents to provide responses.
termine how strongly the respondent agrees l The drafting of the questionnaire should be
with a particular statement. These questions concise, precise and brief because lengthy
help to assess how the respondent feels to- questionnaires may lead to boredom among
wards a certain issue/service. respondents.
Example: Q. Is this community a good place to l The language of the questionnaire should be
raise children? one that the respondent knows.
1 Strongly agree l Questions outside the respondent experience
should not be asked.
2 Agree l In asking questions about past events, too much
3 Uncertain reliance should not be placed on the respon-
4 Disagree dents’ memory.
5 Strongly disagree
l Questions that are likely to lead to bias in the
respondents should be avoided.
I. Bipolar questions: Bipolar questions are ques- l Questions should be very clear and simple.
tions that have two extreme answers. The re- Avoid professional jargons.
spondent has to mark his/her response between l As far as possible, open-ended questions should
two opposite ends of the scale. be avoided.
Example: Q. What is your balance of preference l Avoid questions with difficult concepts, which
here? are not easily understandable for respondents.
156 Research Methodology and Biostatistics

l Controversial and ambiguous questions should Question Construction (Fig. 7.1)

be avoided. l Use statements that can be interpreted in the
l The structure of the questionnaire should be same way by all subjects.
according to the form in which the responses l Use such kind of statements where persons may
are to be recorded. have different opinions or traits and may give
l Cross-check the respondent by asking the same different answers.
information in two different ways. l Use only one aspect of the construct in which
l A mailed questionnaire should be accompanied you are interested.
by an introduction to the study, purpose and l Avoid asking double-barrelled questions con-
directions to fill the questionnaire. taining two distinct ideas or concepts. For
l Abrupt ending of the questions and question- example, are you satisfied with pay and fringe
naire should be avoided. benefits?
l Avoid leading, loaded, ambiguous and long
Sequence of Questions
questions.
l There should be a logical sequence of the ques- l Avoid negative and double negative questions.
tions in the questionnaire. l Use positive statements.
l The researcher has to make sure that the answer l Assumptions should not be made about
to a question is not derived from a previous the respondent.
question. l Comprehensible and clear wording should be
l There should be a flow of questions in the trend used.
from general to more specific. l Use correct spelling, grammar and punctuation.
l Moreover, there should be a flow of questions
from least to most sensitive.
l Sandwich theory states that a questionnaire Methods of Questionnaire Administration
should generally start with the demographic A questionnaire may be administered in the
profile of subjects, followed by specific ques- following ways; each method has its unique ad-
tions according to the objectives of the study. vantage as listed here.

The sample Who are you going to ask?

The method How are you going to ask them?

The questionnaire What are you going to ask them?

The result What will you do with the information?

The cost How much do you want to pay for the answer?

The time scale By when do you need the information?

Fig. 7.1. ​Questions researcher must consider before developing a questionnaire.

 Measurements and Methods of Data Collection 157

Postal situation. Although the interview is generally associ-

• Low cost ated with counselling or psychotherapy, it is used
• Not labour intensive effectively to collect useful information about indi-
• Anonymous
viduals in many research situations.
Interviewing is a popular way of gathering in-
Phone formation in qualitative research. Interview pro-
• High speed vides a wide range of data. It provides information
• Rapport with respondents
• Detailed questions
about person’s feelings, perceptions and opinions.
• High response rate Interviewers should not make preassumptions
about the respondents. Interviews require high-
Electronic level questioning skills and active participation.
• Low cost Interviewing requires face-to-face contact or
• High speed
• Anonymous
contact over telephone and calls for interviewing
• Not labour intensive skills. Interviews may be used either as a main
method or as a supplementary one in collection of
Personally administered research-related data.
• Detailed questions They are most often used to gather detailed,
• High response rate
qualitative data. Interviews are conducted one-on-
one, while focus group interviews are conducted in
small groups. Both are usually conducted with tar-
ADVANTAGES OF QUESTIONNAIRE geted samples of staff, administrators, youth, fami-
l Questionnaires are cost-effective. lies, community members and so on. Interviews
l They are easy to analyse. can also be conducted either face-to-face or tele-
l They require less time and energy to administer. phonically. Most questions are open ended in which
l These offer the possibility of anonymity. responses are provided in notes or in a transcrip-
l They reduce bias as the interviewer is not present. tion manner. In some cases, interviews generally
l Questionnaires are used for large sample size. employ structured quantitative response methods.
l Questionnaires are less intrusive than phone or Interviewing is the only suitable method for gath-
face-to-face interview. ering information from illiterate or less-educated
respondents. It is superior to other data-gathering
DISADVANTAGES OF QUESTIONNAIRE methods. People are usually more willing to talk
l Questionnaires are not suitable for all, e.g. children, than to write. Once rapport is established, even
the blind and illiterates. confidential information can be obtained.
l They have a low response rate.
l Those sent by mail may be filled by someone DEFINITIONS OF INTERVIEW
other than the intended person. A method of data collection in which one person
l They provide only superficial information. (interviewer) asks questions to another person
l Probing of response is not possible. (respondent) conducted either face-to-face or
l There are chances of misinterpretation. telephonically.
l People can lie and answer the question vaguely. The conversation among more than one people
(interviewer/s and interviewee) where questions
are asked by the interviewer/s to obtain informa-
INTERVIEW AS A METHOD OF
tion from the interviewee is known as an interview.
DATA COLLECTION Interview is a two-way systematic conversation
The interview is a process of communication or in- between an informant and an investigator. It is
teraction in which the subjects or interviewee gives generally intended to get information that is rele-
the needed information verbally in a face-to-face vant to a particular purpose. It requires not only
158 Research Methodology and Biostatistics

conversations but also learning from the respon- People who would normally ignore a questionnaire
dent’s gestures, facial expressions and pauses, and are willing to talk with an interviewer, which elicits
his/her environment. interesting information, e.g. hospitalized patients.
l Clarifies misunderstandings: When conduct-
CHARACTERISTICS OF INTERVIEW ing an interview, a researcher can be sensitive to
l The participants, the interviewer and the the subject’s misunderstanding of the questions,
respondent are strangers. and can provide further clarification on the
l The relationship between the participants is a topic under discussion.
transitory one. l Asks questions at several levels: A researcher can
l Interview is a mode of obtaining verbal answers plan to ask questions at several levels to gather
to questions put verbally. maximum information from the subjects.
l The investigator records information furnished l Helps to gather other supplementary informa-
by the respondent in the interview. tion: The interviewer can gather other supple-
l Interview is not a mere casual conversational mentary information, such as the economic
exchange, but a conversation with a specific level and living conditions.
purpose. l Allows use of special devices: The interviewer can
l The interaction between the interviewer and the use special scoring devices, visual material, etc. in
respondent need not necessarily be face-to-face order to improve the quality of interviewing.
because interviews can be conducted over tele- l Accuracy can be checked: The accuracy and
phone also. dependability of the answers given by the
l Although interview is usually a conversation respondent can be checked by observation.
between two persons, it is not always limited to l Flexible and adaptable: Interviews are flexible
a single respondent. and adaptable to individual situations. The req-
l Interview is an interactional process – the inter- uisite amount of control can be exercised over
action between the interviewer and the respon- the interview situation.
dent depends on how they perceive each other. l For people who are unable to write their re-
l Interview is not a standardized process; it can be sponses, interview is very advantageous, e.g.
modified according to the situation. patients with eye patches or in tractions.

BENEFITS OF INTERVIEW TYPES OF INTERVIEWS

l Provides in-depth and detailed information: Basically, interview can be classified into the fol-
The greater value of this method is the depth lowing six broad categories.
and detail of information that can be secured.
When used with a well-conceived schedule, Structured Interview (Directive Interview)
an interview can obtain a great deal of infor- Structured interview is a means of data collection in
mation. which the interviewer has an interview schedule in
l Permits greater depth of response: The inter- which the questions are listed in the order in which
view permits greater depth of response, which is they are to be answered. These interviews are also
not possible through any other means. It also known as standardized interviews. In other words,
enables an interviewer to get information con- this is an interview made with a detailed, standard-
cerning feelings and attitudes related to the ized schedule. The same questions are put to all the
phenomenon under research. respondents and in the same order. Interviewers are
l Data from illiterate subjects: Interview is ad- not permitted to change the specific wording of the
vantageous for illiterate subjects who do not interview schedule questions. Structured interview
write as fluently as they speak. consists of the following characteristics:
l Higher response: Interview method results in bet- l Interview schedule is formalized and has a

ter, more elaborate responses than questionnaires. limited set of questions.

 Measurements and Methods of Data Collection 159

l The aim is to ensure that each interview is Merits

presented with exactly the same questions in the l It is less prone to interviewer’s bias.
same order. l It provides greater opportunity to explore prob-
l It increases the reliability and credibility of the lems in an unrestricted manner.
research data. l It is useful for gathering information on sensi-
l It minimizes context effects, wherein answers tive topics, like divorce, social discrimination
given to survey questions depend on the nature and drug addiction.
of preceding questions.
Demerits
Merits l The data obtained from one interview are not
l Data from one interview to the next are easily comparable to the data from the next.
comparable. l Time may be wasted in unproductive conversation.
l Recording and coding data do not pose any l There is no order or sequence in this interview.
problem. l It requires more skills on the part of the
l Attention is not diverted to irrelevant and time- researcher.
consuming conversation.
Semistructured Interview
Demerits Semistructured interview is a flexible method that
l It tends to lose the spontaneity of natural allows new questions to be brought up during the
conversation. interview, depending upon the situation during
l The way the interview is structured may be such the interview. Semistructured interviews have the
that the respondent’s views are minimized and following characteristics:
the investigator’s own biases regarding the  The interviewer prepares an interview guide,

problems under study are assessed. which is an informal list of topics and ques-
l The scope of exploration of information of data tions the interviewer can ask in different ways
is limited. to different participants.
 Interview guide helps the researcher focus on
Unstructured Interview (Nondirective Interview) the topics at hand without constraining them
These interviews are also known as nonstandardized to a particular format.
interviews. Unstructured interview is a method wherein
the questions can be changed to meet the respondent’s In-Depth Interview
intelligence, understanding and beliefs. Questions are This is an intensive and investigative interview
changed based on how each individual person re- conducted and aimed at studying the respondent’s
sponds to the questions. In this type of interview, the opinion and emotions based on an interview
interviewer encourages the respondent to talk freely guide. It requires more training and interpersonal
about a given topic with minimum prompting. In this, skills than structured interviewing. An in-depth
no preplanned schedule is used. Unstructured inter- interview deliberately aims at eliciting uncon-
view consists of the following characteristics. scious as well as extremely personal feelings and
l Interview schedule is not formalized and has emotions. This is designed to encourage free ex-
open-ended questions, where there is opportu- pressions of affectively charged information.
nity to ask questions not planned before.
l The aim is to explore the information from the Focused Group Interview
respondents. Focused group interview is an unstructured group
l It increases the reliability and credibility of interview technique where 8–12 members are
research data. brought together under the guidance of a trained
l It facilitates the natural unobstructed proceed- interviewer to focus on a specific phenomenon. Its
ings of the interview procedure. purpose is to gain insight from the appropriate
160 Research Methodology and Biostatistics

sample or target population. The researcher tries to Telephone Interview

focus his/her attention on a particular aspect of a Telephone interviewing is a nonpersonal method
problem. It is a carefully planned discussion where of data collection. This method of collecting
the interviewer guides discussion according to a information consists of contacting respondents
written set of questions or topics to be covered. In on telephone. It is used widely in industrial sur-
focused group interviews, special attention is given veys, particularly in developed regions.
to specific topics or ideas. An attempt is made to
concentrate the discussion on the real effects of a Merits
provided experience to which the respondents have l It is more flexible and a quick way of obtaining
been asked to respond. The interview is aimed to information.
analyse the subjective experience of the respondent l Cost is relatively less and rate of response is high.
that is his/her opinion or emotional response to the l Access to the respondent is easy through a tele-
topic of research. The basic characteristics of this phone directory.
type of interview are as follows: l No field staff is required and wider coverage of
l It consists of 8–12 members homogeneous sample is possible.
in terms of demographic and socioeconomic
Demerits
characteristics.
l Duration of the interview is 1.5–2 h.
l It is difficult to establish a rapport between
l All information including facial expressions and
the respondent and the interviewer.
l Surveys are limited to respondents having
body language is recorded.
l It has a moderator to lead the discussion.
telephone facilities.
l It permits the interviewer to obtain details of
l The respondent’s characteristics, facial expres-
sions and environment cannot be observed.
personal reactions to an experience.
l It is free from inflexibility of formal methods.
l Possibility of bias of the interviewer is more.
l There are a large number of opportunities for INTERVIEWING PROCESS
respondents to present their views. The interviewing process includes the following steps.
l The interviewer can freely choose the order of

questions and ensure the extent of asking the Preparation for the Interview
questions. l Fix appointment: The interviewer should fix the
topic and time of interview with the subject ahead
Merits of time. It would save the time of the interviewer.
l Wide range of information is collected. l Appearance and dress: The interviewer should
l It stimulates new ideas and creative concepts. dress in a fashion fairly familiar and with due
l A person’s comments may trigger a response regards to the sensibilities of the people who will
from another person, leading to in-depth be interviewed. The interviewer should be pleas-
discussion on the topic. ant and should communicate a genuine interest.
l It increases excitement in a group. l Familiarity with interview schedule: The inter-
l Participants feel more comfortable to answer in view schedule must be studied carefully and the
a group with similar interests. interviewer must practise reading it aloud to be
l It involves many participants at one time. very familiar with all the questions. The inter-
viewer must also be able to list items to respon-
Demerits dents without errors.
l There are chances of client and researcher bias.
l It is difficult to moderate. Preinterview Introduction
l It is difficult to code, analyse and interpret. l The researcher must introduce himself or her-
l Focus group interview is not a representative of self to all respondents respectfully as per their
the general population. cultural practices.
 Measurements and Methods of Data Collection 161

l Identify the respondent by name. l A tape recorder should be used to record the
l Select a setting with as little disturbance as responses of the respondent.
practicable. l Shorthand should be used to record responses.
l Describe the intention of the interview.
l Mention the confidentiality terms. Closing the Interview
l Describe the interview structure. After the interview is over, a polite leave of the
l Give a hint as to how long the interview would respondent should be taken, thanking him or her
go on. with a friendly smile and saying goodbye.
l Ask them if they have any questions.
ADVANTAGES OF INTERVIEW
Developing Rapport l Interviews are useful to obtain information
After primary introduction, the researcher starts about people’s feelings, perceptions and
developing rapport with the respondents. This is opinions.
developed through confidence building and l Interviews allow more detailed questions to be
understanding the concerns of the respondents by asked.
beginning the conversation on general topics. l High response rate is achieved.
l Respondent’s own words are recorded.
Carrying the Interview Forward l Ambiguities can be clarified and incomplete
After establishing rapport, the technical task of answers followed up.
asking questions from the interview schedule l Interviews are not influenced by others in the
starts. While interviewing the respondent, the fol- group.
lowing guidelines should be used: l Meaning of questions can be clarified.
l Start the interview.

l Ask only one question at a time. DISADVANTAGES OF INTERVIEW

l Repeat a question if necessary. l Interviews are time consuming and costly af-
l Try to make sure that the subject understands fairs.
the questions. l Different interviewers may understand and
l Listen carefully to the subject’s answer. translate interviews in different ways.
l Observe the subject’s facial expressions, gestures l There are high degree chances of interviewer’s
and tone of voice. biases.
l Allow the subject sufficient time to answer the

questions.
l Do not show signs of surprise, shock or anger.
COMPOSITE SCALES/ATTITUDE
l Maintain a neutral attitude with respect to
SCALES
controversial issues during the interview. A scale is a device designed to assign a numeric
l Take note of answers that seem to be vague or score to people to place them on a continuum
ambiguous. with respect to attributes being measured, like a
l Ask additional questions to follow-up clues or scale for measuring attitude or weight.
to obtain additional information. These rating scales can also be used to assess
l Do not hurry the interview. If silence is too the attitudes or feelings of self-concept. The ex-
prolonged, introduce a stimulus. pressions of any point of view are accounted as
measurements towards any item, object or con-
Recording the Interview cept; they show the person’s positive or negative
l It is essential to record responses as they take attitude towards any concept. Measuring the score
place. between two opposite concepts tells us about the
l Good information can be obtained by note- attitude, feeling and perception of a person or
taking. study subject towards the directions of positive or
162 Research Methodology and Biostatistics

negative attitude in a scale. Visual scales illustrate variables. However, in recent times, one can even
visual depiction of any culture; photographs are observe the Likert scale with four points (strongly
given to get the feelings, beliefs, opinions about agree, moderately agree, disagree and uncertain) to
contrast, texture, colour and elements. seven points (very strongly agree, strongly agree,
l Composite scales are socio-psychologic measure- agree, uncertain, disagree, strongly disagree and
ments directed towards quantifying qualitative very strongly disagree) scaling categories. An ex-
attributes, such as feelings, attitude, self-concept, ample of a five-point Likert scale may be perused
perceptions, beliefs, etc. in Table 7.3.
l Composite scales are socio-psychologic rating

scales generally used for qualitative measures Definitions of the Likert Scale
and discriminate people with different attri- The Likert scale is a composite measurement scale
butes, perception, fears, motives, personality used to measure attitude, values, and feelings of
traits, and needs that are often incorporated people that involve summation of scores on a set
into interviews or questionnaires. of positive and negative declarative statements
This chapter discusses some of the commonly regarding measuring variables to which respon-
used composite scales generally used by health sci- dents are asked to indicate their degree of agree-
ence researchers to measure the sociopsychologic ment or disagreement.
issues related to health care practices, such as the The Likert scale is a composite measure of at-
Likert scale and the semantic differential scale. titudes that involve summation of scores on a set
of items (statements) to which respondents are
LIKERT SCALE asked to indicate their degree of agreement or
The Likert scale was named after the psychologist disagreement.
Rensis Likert, who developed it in 1932 as a psy-
chologic concept measurement scale. It is one of Uses of the Likert Scale
the most commonly used scaling techniques. It l Basically, it is used to measure the attitudes,
was developed to measure the attitudes, values values and feelings of people about specific
and feelings of people. Primarily, the original ver- concepts, such as situations, people, places,
sion was developed with five-point scales (strongly objects, programmes, practices and policies.
agree, agree, uncertain, disagree and strongly dis- l This scale is used to have quantified measure-
agree) containing a mixture of positive and nega- ment of qualitative attributes of people, such as
tive declarative statements regarding measuring feelings, values and attitudes.

Table 7.3. ​Example of a five-point Likert scale to assess attitudes regarding HIV/AIDS
Statement Strongly Agree Agree Uncertain Disagree Strongly Disagree
1. Person with multiple sex partners is
at high risk of AIDS
2. You can get AIDS by sharing utensils
3. You may get HIV by sharing needles
with others
4. Only gay men can get AIDS
5. One way of getting AIDS is infected
blood transfusion
6. AIDS is a curable disease
Note: Please tick () in appropriate column for each statement.
 Measurements and Methods of Data Collection 163

l It may also be used to assess the opinions of respondents get higher score if there is agreement
respondents about a particular abstract concept. with the statement. However, in case of a negative
l It collects the opinions of people with various statement, the respondent gets a higher score if
attitudes, emotions and feelings towards a par- there is a disagreement with the statement or vice
ticular concept. versa, as depicted in Table 7.4.
Characteristics of the Likert Scale Advantages of the Likert Scale
The main characteristics of the Likert scale are as l It is relatively easy to construct this scale.
follows: l The Likert scale is considered as a reliable and
l Psychologic measurement tool: The Likert scale
valid tool to measure psychosocial variables.
is basically a psychologic measurement tool to l It is easy to administer since respondents only
assess the attitudes, values and feelings of the have to tick in the spaces provided against each
people about a specific concept. statement.
l Illustrative in nature: This scale is generally il- l It is less time consuming during construction
lustrative in nature, where each statement in a and administration.
scale is stated in an explicitly illustrative way so
that a person can make a clear judgement about Disadvantages of the Likert Scale
the degree of agreement or disagreement with l In the Likert scale, respondents may feel forced
the particular item on the scale. to answer questions against all preplanned items
l Neutral statements: This must contain neutral and their categories.
statements without incorporation of any bias of l Feelings of the respondents may not be fully as-
the researcher. sessed due to the researcher’s preplanned state-
l Bipolar scaling method: This is composed of ments and categories.
alternative positive and negative declarative l Difficulty in justifying the selection of the num-
statements so that the respondent’s casual re- ber of categories and numerical assignment to
sponse bias can be eliminated. Here positive these categories.
statement gets a high score with agreement and l Casual approach of the respondent in these
negative statements gets a high score with dis- scales may provide misleading data.
agreement with the statement.
l Measurement of the specific number of scaling
SEMANTIC DIFFERENTIAL SCALE
categories: This scale was originally developed Semantic differential scale is most effective and most
with five scaling categories; but later scales were widely used technique nowadays. In 1967, Osgood,
developed with even four, six and seven scaling Suci and Tannenbaum introduced this method for the
categories. first time in their book, The Measurement of Meaning.
Although the original purpose of the scale was not
Scoring of the Likert Scale necessarily to measure the assessment of attitude, the
Scoring of the Likert scale is done on the basis of procedure was well adopted for attitude assessment.
type of statement and level of respondents’ agree- People’s attitudes regarding objects, stimulus words,
ment with the statement. For a positive statement, and concepts are measured by semantic differential

Table 7.4. ​Scoring of statements on the Likert scale

Statement Strongly Agree Agree Uncertain Disagree Strongly Disagree
Positive statement: Persons with multiple 5 4 3 2 1
sex partners are at high risk of AIDS
Negative statement: You can get AIDS by 1 2 3 4 5
sharing utensils
164 Research Methodology and Biostatistics

questions. This comprises a series of differential adjec- Table 7.5. ​Adjective pairs used for construction of
tive pairs, such as beneficial–harmful, and good–bad, a semantic differential scale
which are mentioned on the two ends of a bipolar
scale. With different cultures, age groups and lan- Evaluation Potency Activity
guages, many studies have indicated that semantic Good–bad Hard–soft Active–passive
differential questions can work efficaciously. As these Kind–cruel Strong–weak Fast–slow
questions are very easy to structure and administer and Wise–foolish Heavy–light Hot–cold
as these render reliable quantitative data and are rea-
Beautiful–ugly Deep–shallow Motivated–
sonably valid, these questions are considered popular. aimless
Definition of Semantic Differential Scale Happy–sad Potent– Moving–still
impotent
Semantic differential scale is a type of rating
Sociable– Large–small Excitable–calm
scale designed to measure the connotative
unsociable
meaning of objects, events, and concepts.
Friendly– Simple– Alive–dead
These connotations are used to derive the
unfriendly complex
respondents’ attitude towards the objects,
events and concepts. Willing– Difficult–easy Emotional–
unwilling unemotional
Usages of Semantic Differential Scale Honest– Submissive– Bright–dim
dishonest assertive
Semantic differential scale is most commonly
used in:
l Patient satisfaction survey

l Customer satisfaction survey

l Employee survey Disadvantages of the Semantic Differential Scale

l Marketing survey l It is difficult to select the relevant concepts that
l Operational research are appropriate for any given investigation.
l Personality measurements l It is time consuming to find the appropriate

l Clinical psychology adjective pairs.

An example of a semantic differential scale to assess
Designing a Semantic Differential Scale the belief about HIV/AIDS is presented in Fig. 7.2.
For the preparation of a semantic differential scale,
adjective pairs are used as depicted in Table 7.5. VISUAL ANALOGOUS SCALE
l The bipolar adjective pairs can be used for a It is a widely used psychomotor response scale,
wide variety of subjects. which is a measurement instrument for subjective
l The adjective pairs are selected according to the characteristics or attitudes that cannot be directly
objectives of the survey. measured. The scale has 10 cm lines and the ends
l The adjective pairs can be grouped into three mark semantic opposites (e.g. alert–drowsy).
large categories; each survey question usually in- The visual analogous scale (VAS) is used to mea-
cludes a few points from each category. These sure the intensity of certain sensations and feelings,
categories are evaluation, potency and activity. such as pain, discomfort, anxiety, alertness, quality of
sleep, severity of clinical symptoms, functional abil-
Advantages of the Semantic Differential Scale ity and attitude towards environmental conditions.
l A convenient method to assess beliefs, attitudes
and values in quantitative form Definitions of Visual Analogous Scale
l Easy to administer The visual analogous scale is a tool that is used to
l Provides reasonable valid and reliable quantita- assist a person in rating the intensity of certain
tive data feelings and sensations (e.g. pain).
 Measurements and Methods of Data Collection 165

HIV/AIDS
Cure 1 2 3 4 5 6 7 Death

No punishment 1 2 3 4 5 6 7 Punishment

Social acceptance 1 2 3 4 5 6 7 Social rejection

Affordable treatment 1 2 3 4 5 6 7 Expensive treatment

Normal life 1 2 3 4 5 6 7 Miserable life

Fig. 7.2. ​Example of a semantic differential scale to assess the belief about HIV/AIDS.

The VAS is a self-reported measure consisting Advantages of Visual Analogous Scale

simply of a 10 cm (100 mm) line with a statement l It is a reliable and valid tool to measure the
at each end representing one extreme of the intensity of certain sensations and feelings.
dimension being measured. l Quantitative assessment is a highly subjective
The VAS is a bipolar scale used to determine phenomenon that is possible by this scale.
the degree of stimuli a patient experiences. One l This scale is of most value when looking at
side of the scale expresses the absence of stimuli, changes among individuals.
while the other expresses the presence of stimuli.
The visual analogous scale is a 100 mm hori- Disadvantages of Visual Analogous Scale
zontal or vertical line with right angle stops at ei- l This scale cannot be used for comparing results
ther ends. Between these two extreme ends (e.g. across a group of individuals at the same time.
best and worst, no pain and severe pain, alert and l While measuring pain, the scale does not

drowsy), the respondent provides his response for provide the patient a chance to document an
a particular phenomenon, which is later measured increase in pain. Thus, it creates a treatment
by using a ruler from the left end. bias.
For example, a straight line with no pain at one l In measuring pain or such sensations, the reli-

end and the worst pain imaginable at the other ability of this scale is low owing to the patient’s
end indicates the visual analogue scale for pain. As need to recall his/her initial pain severity.
per the amount of pain he/she feels, a patient usu- An example of a horizontal VAS is depicted in
ally marks a point on the line that matches with Fig. 7.3.
his/her feelings.
OBSERVATION AS A METHOD OF
No pain Worst pain ever DATA COLLECTION
Observation is a way of gathering data by watching
Uses of Visual Analogous Scale behaviour, events or noting physical characteristics
l To measure the level of pain, anxiety and mood in their natural setting. Observation is a method of
l To measure the alertness, quality of sleep, sever- data collection that can be used to gather informa-
ity of clinical symptoms, functional ability and tion such as characteristics and conditions of an
attitude towards environmental conditions individual, verbal and nonverbal communication.
l To choose the right dose of medicine for the Observation method of data collection is particu-
treatment of the disease larly well suited to many biomedical research stud-
l To determine the drug’s effect on the patient ies. There are several situations that require the
l For rating soft variables, such as skills, confi- biomedical researcher to use the observation
dence and quality of team work method for data collection, such as healing of a
166 Research Methodology and Biostatistics

Scale A: Horizontal visual analogue scale

How much shortness of breath are you having right now?

Please indicate by marking the line. If you are not experiencing any
shortness of breath at present, circle the marks at the left end of the line.

No shortness Shortness of breath

of breath as bad as can be

Fig. 7.3. ​Example of a horizontal visual analogous scale.

wound, assessing a physical sign of a disease condi- USES OF THE OBSERVATION METHOD
tion, behaviour and attributes of the patients, their Observation as a method of data collection is used
families, and hospital staff. in the following situations:
As everyone is aware that observations are being l To collect data related to physical sings of disease

noted, these can be overt or covert (the subjects are condition and examining recovery: The main
not aware that they are being observed and the source of data collection in biomedical research are
observer is hiding). The advantage in covert obser- either the patients or the samples collected from the
vation lies in the fact that people most likely behave patients. The assessment of signs of the patients suf-
naturally as they are unaware that they are being fering from a particular disease condition and dis-
noticed. Howsoever, one will specifically require order and their body samples require observation to
employing overt observations due to ethical prob- examine the severity level of the clinical condition
lems linked to hiding the observation. and the effects of a particular intervention.
There can be either direct or indirect observa- l To understand an ongoing process or situation:

tions too. Whenever you see processes, interac- Through observation, a process or situation can
tions or behaviours as they happen, these are be monitored and evaluated as it occurs at a par-
direct observations; for example, noticing a ticular time. For example, a researcher wants to
teacher teaching a lesson from a written docu- assess the existing ward management practices in
ment to ascertain whether they are doing it with public and private hospitals. In this situation,
accuracy. When you see the results of processes, observation is the best method of data collection.
interactions or behaviours, these are indirect l To gather data on individual behaviours or inter-

observations; for example, determining how actions between people: Observation allows re-
much waste is left in the plates by students in searchers to watch peoples’ behaviours and interac-
a school cafeteria to ascertain if a new food is tions directly, or watch for the results of behaviours
acceptable to them. or interactions; for example, how a doctor respond
to agitated patients in an emergency setting.
DEFINITIONS OF OBSERVATION l To know about a physical setting: Observing

Observation is a technique of collecting data the environment or place where something hap-
or acquiring information through occur- pens can facilitate increased understanding of
rences that can be observed through the the activity, event or situation that the researcher
senses with or without mechanical devices. is evaluating. For example, a researcher can ob-
It is a two-part process to collect data for serve whether a classroom or training facility is
study that includes an observer (someone conducive to learning.
who is observing) and the observed (there is l Data collection where other methods are not

something to observe). possible: If respondents are unable to provide data

 Measurements and Methods of Data Collection 167

through questionnaires or interviews or are un- an extended period. In addition, the observer
willing to do so, observation is a process that needs becomes an accepted member of the group or
little from the individuals for whom we require community. The level of participation of the
data. For example, a researcher is conducting a observer may range from a completely absorbed
study on the satisfaction of deaf and dumb illiter- state to a minimal state of participation in the
ate children residing in a rehabilitation centre. In group or community under study. During par-
this situation, the researcher has only the observa- ticipatory observation, unstructured tools are
tion method to observe the subjects’ nonverbal used to collect data, such as logbooks, field
gestures and behaviour to assess their satisfaction. notes, field diary, tape and video recording.
l Nonparticipant observation: In nonparticipatory
TYPES OF OBSERVATION observation, the observer works as an eavesdropper
Observation method of data collection can be (secretly listening or observing), where an attempt
classified into four broad categories. is made to observe people without interacting with
l Structured observation: In structured observa- them. In this technique, the observer is not a par-
tion method, the researcher prepares a structured ticipant in the setting but is merely viewing the
or semistructured tool in advance to observe the situation. In addition, people are observed without
phenomenon under study. Structured observa- their knowledge. This method is routinely used by
tion is guided by the preplanned structured or psychologists to study children or animals. How-
semistructured instrument, where the researcher ever, this method possesses ethical problems due to
observes only specific attributes or behaviour in observation of people without their consent.
accordance with planned observation guidelines.
This helps the researcher to be on the track while PLANNING FOR OBSERVATION
carrying out an observation’ analysis of data col- l Focus determination: The evaluation question(s)
lected during this method of observation re- you want to be answered through observation are
mains easy. Structured observation is generally to be visualized first. Then choose a few focus
carried out by using the following tools: points for your data collection. For example, you
 Checklist may need to know how nicely an HIV awareness
 Rating scale syllabus is being followed in the classroom. The
 Category system focus areas might be the teachers’ skills, knowl-
l Unstructured Observation: Unstructured obser- edge and behaviours in teaching this particular
vation is generally used in qualitative studies, topic and interactions between teachers and
where observation is made with minimally struc- students.
tured or researcher-imposed categories. It is used l Data collection design: You need to think about
for complete and nonspecific observation of phe- the specific items for which you want to gather
nomenon, which is very well known by the re- data once you have focused your evaluation.
searcher. Unstructured observation is carried out Then ascertain the method as to how you are
by using unstructured tools, and the following going to acquire the information that you re-
tools are generally used by the researchers: quire. The following three primary methods of
 Log and field notes collecting observation data can also be com-
 Anecdotes bined to ensure your data collection needs:
 Field diary  Checklists and recording sheets are some of

 Video recording the accepted ways of gathering observation

l Participant observation: Historically, field and data. This includes both responses and preset
ethnographical researches have been associated questions. These forms are usually employed
most strongly with participant observation, to collect data that can be described in ad-
where the observer may live or work in the field vance easily (e.g. topics that are to be covered
and actively participate in ongoing activities for in an HIV-prevention course).
168 Research Methodology and Biostatistics

 Observation guides list the processes, interac- DISADVANTAGES OF OBSERVATION

tions or behaviours to be checked with space l Observer’s bias
to enlist open-ended descriptive data. l ‘Hawthorne effect’ susceptibility: People usually
 Field notes are the least formal ways of gath-
do better when they come to realize that they
ering observation data. They do not have re- are being noticed, although this may decrease in
sponses or preset questions. Field notes are indirect observation
generally open-ended descriptive data and are l Can be time consuming and expensive as against
dictated or written onto a tape recorder. other data collection techniques
l Site selection: An adequate number of sites l Does not enhance the clear-cut understanding
need to be selected to ensure that they represent of why people behave as they do
the larger population. This gives a picture of the
plot you are noticing.
l Observer selection: You may prefer to be the RATING SCALE
only observer or you may add other persons Rating is the term used to express the opinion or
while conducting observations. Other profes- judgement regarding some performance of a person,
sional staff members, stakeholders, interns and object, situation or character. The rating scale has
graduate students and volunteers generally act qualitative description of some aspects or traits of a
as potential observers. person or a thing. When rating scales are used, we
l Observer training: It is important that observ- value an object by absolute terms as against some
ers are well trained in the data collection process specific criteria (e.g. we rate the properties of objects
in order to ascertain consistent and high-quality without any reference to such similar objects).
data. The training level will fluctuate depending
on the data collection complexity and the capa- DEFINITIONS OF RATING SCALE
bilities of the individual observers. Rating scale refers to a scale with a set of opinion,
l Appropriate observation timings: A sequence of which describes varying degrees of the dimen-
events generally precedes programmes and pro- sions of an attitude being observed.
cesses. It is important that you strategize your Rating scale is a device by which judgements
observations to observe the activity components may be qualified or an opinion concerning a trait
that will respond to the evaluation questions. can be systematized.
Advance planning is needed for this process. Rating scale is a tool in which one person sim-
ply checks off another person’s level of perfor-
ADVANTAGES OF OBSERVATION mance. It could be a three-point, five-point or
l Collects data as to when and where an activity seven-point rating scale, as presented in Figs 7.4
or event is happening. and 7.5.
l Does not depend on people’s ability or willing- Rating scale measures how much or how well
ness to give information. something happened, where generally quantita-
l Directly allows you to notice what people do tive and qualitative terms are used to judge the
rather than checking what people tell that they did. performance.

No pain 2 3 4 5 6 7 8 9
Worst pain

Fig. 7.4. Numerical rating scale.​

Always Sometimes Never

Fig. 7.5. ​A three-point rating scale.

 Measurements and Methods of Data Collection 169

Table 7.6. ​Common content of appraisal in rating scales

Quantity of work Volume of work Quality of work Dependability
Reliability Initiatives Judgement Neatness
Accuracy of work knowledge Cooperation Ability to work Willingness

For example: For example: How much are you satisfied with
Q. How good you felt during your stay in this hospital? noise control in your ward?
Q. How many times did doctors promptly respond to
your calls? Least Most
The common content which can be judged using
rating scale is illustrated in Table 7.6.
2. Descriptive rating scales: This type of rating
TYPES OF THE RATING SCALE scales do not use number, but divide the assess-
ment into a series of verbal phrases to indicate
1. Graphic rating scale: In this scale, the perfor-
the level of performance.
mance is printed horizontally at various points
For example: Q. Judge the level of performance of
from lowest to highest. It includes the numeri-
housekeeping personnel working in the medical
cal points on the scale. It is anchored by two
ICU
extremes presented to respondents for evalua-
tion of a concept or object.

Nursing Personnel in a Ward Level of Clinical Performance

Very Active Active Moderately Active Passive
1. Amandeep
2. Jasveen
3. Tara
4. Kirandeep

3. Numerical rating scale: It divides the evaluation each statement is generally assigned a numerical
criteria into a fixed number of points, but defines score ranging from 1 to 10, or even more.
only numbers at the extremes. In these scales, For example: Pain assessment numerical scale.

No pain 2 3 4 5 6 7 8 9
Worst pain

4. Comparative rating scale: In this type of rating must have prehand knowledge about the se-
scale, the researcher makes a judgement about lected attributes of the people with whom the
an attribute of a person by comparing it with subjects are supposed to be compared.
that of similar other person(s).
For example, Mr Ram’s decision-making abili- CHARACTERISTICS OF RATING SCALE
ties closely resemble those of Mr Shyam and l Rating scales are value judgements about the
Mr Gopal. In this type of rating scale, researcher attributes of one person by another person.
170 Research Methodology and Biostatistics

l Theses scales are the most commonly used tools DISADVANTAGES OF RATING SCALE
to carry out structured observations. l It is difficult or dangerous to fix a rating about
l Rating scales are generally developed to make quan- many aspects of an individual.
titative judgements about qualitative attributes. l Misuse can result in decrease in objectivity.
l Rating scales provide more flexibility to judge l There are chances of subjective evaluation;
the level of performance or presence of attri- thus the scales may become unscientific and
butes among subjects. unreliable.
l Guilford (1954) identified that a rating scale
must have the following basic characteristics,
which must be taken care of while constructing CHECKLIST
a rating scale. It is one of the most commonly used instruments
 Clarity: The rating scale must be constructed
for performance evaluation. A checklist enables
using short, concise statements in simple and the observer to note whether or not a trait is pres-
unambiguous language. ent. It consists of a listing of steps, activities and
 Relevance: The statements designed in the rat-
behaviours the observer records when an incident
ing scale should be relevant to the phenome- occurs. The observer has to judge whether certain
non and should be exactly in accordance with behaviour has taken place.
the variables under study. In the health care setting, checklist is com-
 Variety: While developing a rating scale,
monly used to list patients’ behaviours associated
monotony in the statements must be avoided with particular nursing interventions, and the as-
and a variety in different statements must be sessor checks or ticks within a space provided
ensured. whether or not those particular behaviours oc-
 Objectivity: Statements in the rating scale must
curred. An example of a checklist is depicted in
be objective in nature, so that it is convenient Table 7.7.
for the researcher to judge the attributes or
performances of the subjects under study. DEFINITIONS OF CHECKLIST
 Uniqueness: Each statement constructed in

the rating scale must be unique in itself, so A checklist is a simple instrument consisting of a
that attributes can be judged appropriately. prepared list of expected items of performance or
attributes, which are checked by a researcher for
ADVANTAGES OF RATING SCALE their presence or absence.
l A rating scale is easy to administer and score the Checklists are constructed by breaking a per-
measured attributes. formance and the quality of a product into mul-
l Rating scales have a wide range of application in tiple steps, which specifies the presence or absence
nursing research. of an attribute or trait; they are then ‘checked’ by
l Graphic rating scale is easier to make and the rater/observer.
requires less time.
l Rating scales can be easily used for a large CHARACTERISTICS OF CHECKLISTS
group. l Observe one respondent at one time.
l It can also be used for quantitative methods. l Clearly specify the characteristics of behaviour
l It may be used for the assessment of interests, to be observed.
attitudes and personnel characteristics. l Use only carefully prepared checklist to avoid
l Used to evaluate performance, skills and prod- more complex traits.
uct outcomes. l The observer should be trained how to observe,
l Rating scales are adaptable and flexible research what to observe and how to record the observed
instruments. behaviour.
 Measurements and Methods of Data Collection 171

Table 7.7. ​Checklist for evaluation of student’s performance during surgical dressing
Behaviour Place for () Tick Mark Remarks
Yes No
1 Explains procedure
2 Collects necessary equipment
3 Arranges equipment for convenient use
4 Prepares patient
5 Washes hand
6 Maintains aseptic technique
7 Removes dressing
8 Observes condition of wound
9 Cleans wound
10 Applies dressing
11 Removes equipment
12 Makes patient comfortable
13 Completes charting
14 Takes care of equipment

l Use checklists only when you are interested in ADVANTAGES OF CHECKLIST

calculating a particular characteristic. l Checklist allows interindividual comparisons.
l It provides a simple method to record observation.
CONSTRUCTION OF CHECKLISTS l It is adaptable to subject matter areas.
While constructing or preparing a checklist, the l It is useful in evaluating learning activities.
following points should be kept in mind: l It is helpful in evaluating procedural work.
l Express each item in a clear, simple language. l Properly prepared checklists allow the observer
l An intensive survey of the literature is made to
to contain the direct attention.
determine the type of checklist to be used in a l It has the objectivity to evaluate characteristics.
particular research. l It is useful for evaluating those processes that
l The list of items in the checklist may be con-
can be subdivided into a series of actions.
tinuous or divided into groups of related l It decreases the chances of error in observation.
items.
l These lists of items are formulated on the basis

of judgement of experts, and each item is evalu- DISADVANTAGES OF CHECKLIST

ated in respect of the number of favourable and l Checklists do not indicate the quality of perfor-
unfavourable responses. mance, so usefulness of the checklist is limited.
l Avoid negative statements whenever possible. l Only a limited component of the overall clinical
l Avoid lifting statements verbatim from the text. performance can be evaluated.
l Ensure that each item has clear responses: Yes or l Only the presence or absence of an attribute,
No, True or False, and the like. behaviour or performance parameter may be
l Review the items independently. assessed. However, degree of accuracy of perfor-
l Checklist must have the quality of completeness mance cannot be assessed.
and comprehensiveness. l It has limited use in qualitative research studies.
172 Research Methodology and Biostatistics

BIOPHYSIOLOGIC METHODS documenting the ways in which medical/health

The research community views biophysiologic interventions affect the patient’s biophysiologic
measures as objective data. Researchers may use the outcome. For example, the effect of diathermy
biophysiologic data collection processes either heat application on treatment of muscular spasms
alone or in combination with other methods. While of the pelvis region among patients visiting the
suitably identifying the project’s measurement physiotherapy clinic at AIIMS, Raipur.
collection ways, the investigator should first enlist
l To evaluate medical/health interventions: These
different interesting variables in the study to in- methods involve a new intervention that is being
clude them in the hypotheses or research questions. tested usually in comparison with the standard
After the data collection methods are selected, the methods of care. These methods involve a hy-
researcher should be cognizant of the limitations of pothesis stating that the innovative medical/
the specific data collection method opted for and health intervention result in improved biophysi-
put to use methods to limit the difficulties. ologic outcome of patients. For example, a study
This trend is an expanded use of measures to on the effect of music on heartbeat rate and
assess the physiologic status of a person. Specific blood pressure of cardiac surgery patients.
technical instruments and equipment are used
l To study the correlation of physiologic function-
to measure the physiologic and physical vari- ing in patients with health problems: These
ables, and specialized training is needed for the methods also may be used for the comparison of
interpretation of results. Data on physiologic the physiologic functioning of patients with sev-
activity or dysfunction can often be gathered eral health problems. For example, a researcher
through direct observation. For example, phe- may compare the relationship of selected physi-
nomena, such as vomiting, cyanosis, postcardi- ologic parameters, such as haemoglobin, haema-
otomy delirium, oedema and wound status can tocrit, serum albumin, blood sugar and incidence
be observed for presence or absence and inten- of hospital-acquired infections.
sity. Other biophysiologic data can be gathered
TYPES OF BIOPHYSIOLOGIC METHODS
by asking people directly. Biophysiologic phe-
nomenon requires the use of specialized techni- Biophysiologic methods are basically classified
cal apparatus that yield quantitative measures. into two categories:

DEFINITION OF BIOPHYSIOLOGIC METHODS 1. In vivo bio- Performed directly to measure

Biophysiologic method involves the collection of physiologic processes occurring internally
biophysiologic data from subjects by using special- methods within living organisms through
medical or surgical instruments,
ized equipment to determine biological and physi-
e.g. BP monitoring, ECG
cal status of subjects. For example, blood pressure and USG
measurement by using special equipment, such as
2. In vitro bio- Physiologic processes are
sphygmomanometer and stethoscope.
physiologic measured and analysis is done
PURPOSES OF BIOPHYSIOLOGIC METHODS methods outside the organism, e.g.
biochemistry, microbiological
The purposes of biophysiologic measures are as and pathological tests
discussed here:
l To study basic physiologic process: Such studies in-

volve the basic physiologic processes that are relevant In Vivo Biophysiologic Methods
to nursing care. This involves healthy and normal In in vivo biophysiologic methods, measurements
subjects. For example, a study of the person related are directly performed over the organism or study
to biophysiologic reactions during period of stress. subject by using specialized instruments or equip-
l To study physiologic outcome of medical/health ment. In other words, it is the measurement of the
intervention: Used in studies exploring and biophysiologic attributes of the subjects by using an
 Measurements and Methods of Data Collection 173

instrument, which can then be interpreted by the and signal processors are used to amplify the de-
researchers. For example, measurement of blood sired physiologic signals from transducers.
pressure using sphygmomanometer, measuring l Display equipment: The display device converts

cardiac conduction using ECG machine and mea- the modified electrical signals into visual or
suring temperature using thermometer. auditory outputs; generally, computers or elec-
The in vivo measuring instrument system in- tronic devices are used for this purpose.
volves six components listed as follows: (Fig. 7.6): l Recording equipment: For research purpose, it is

l Stimulus: Biophysiologic measurements need a usually necessary to obtain a permanent record

stimulus to begin them. The stimulus could be of the physiologic measurements for subsequent
external or internal to the subject under study. scrutiny and analysis.
For example, in electromyography (EMG), the Physiologic measurements are good examples of
monitoring system requires the stimulus com- in vivo biophysiologic measurements. This includes
ing from the brain and passed to the nerves and measurements of the physiologic attributes of an indi-
to the muscles. vidual, such as temperature, pulse, respiration, blood
l Subject: The human body consists of chemical, pressure, physiologic volumes and pressures in the
electrical, mechanical, thermal and various body, ECG, ultrasonography, X-ray, MRI and CT scan.
other systems, which become subject to bio-
physiologic measures. In Vitro Biophysiologic Methods
l Sensory equipment (transducers and electrodes): In in vivo biophysiologic methods, measurements
It consists of one or more transducers or elec- are carried out outside the organism or study sub-
trodes, which converts one form of energy into ject by using specialized equipment. In other
another that can be measured and interpreted. words, it is the measurement of the biophysiologic
l Signal conditioning (amplifiers and signals proces- attributes of the subjects, which is carried out
sors): Generally, two types of signals are produced through collection of a sample of data related to
by the subjects: key physiologic signals and inter- physical, biochemical, microbiologic, pathologic
ference signals. Key physiologic signals are signals, and anatomic status by using some technical
such as electrocardiographic signals. Amplifiers instruments and by sending them to a laboratory
for their analysis and interpretation. For example,
1 a blood sample is collected from a patient to assess
Stimulus
the blood electrolyte status of the patient.
2 3
Subject Electrodes The following measurements can be carried
out using in vitro biophysiologic methods (Fig. 7.7).
l Radiologic measurements: This includes the

4 measurements of the radiologic attributes of the

Amplifier and subject, such as evaluation of the body tissues
signal processes
by the use of X-rays, CT scan, magnetic reso-
To triceps
nance imaging (MRI) and positron emission
To biceps
To dry earth strap
tomography (PET) scan.
l Biochemical measurements: This includes mea-

5 surements of the biochemical attributes of an in-

Display dividual, such as hormonal levels, blood sugar
levels, haemoglobin, serum electrolytes, liver func-
6
Recording
tion tests, renal function tests and serum markers.
equipment l Microbiologic measurements: This includes mea-

surements of the microbiologic attributes of an

Fig. 7.6. ​Example of an in vivo measuring instrument individual, such as bacterial counts, culture and
system. sensitivity tests.
174 Research Methodology and Biostatistics

l The results produced by the biophysiologic

X-ray, CT scan, MRI and PAT scan measurement instruments may be affected by
the environment. For example, axillary temper-
Radiological ature recording in a room with or without air

urea, electrolytes, etc.

measurements

Hormone levels,
conditioning may have different readings.
Tissue biopsies

measurements

measurements
Biochemical
Cytological

l The use of biophysiologic measurement instru-

In vitro
measurements ments may cause fear and anxiety among par-
ticipants. For example, the collection of blood
Microbiological sample for biophysiologic measurements may
measurements have traumatic impact on the subjects.
l The use of some biophysiologic methods may
Culture and sensitivity test have a harmful effect on the participants, such
as repeated exposure to X-ray increases the
Fig. 7.7. ​In vitro measurements. health risk for study subjects.

PROJECTIVE TECHNIQUE
l Cytologic measurements: This includes measure-
This involves a wide array of games and tasks in
ments of the cytologic attributes of an individ-
which respondents need to participate during an
ual, such as tissue biopsies.
interview or group discussion; it is schemed to
ADVANTAGES OF BIOPHYSIOLOGIC METHODS
extend, facilitate or enhance the discussion nature.
Basically taken in a psychotherapeutic setting and
l Biophysiologic measures are relatively more accu- being loosely based on different approaches, some
rate and errorless. For example, to assess the oxygen are called ‘projective’ techniques. They are based
saturation, a pulse oxymeter is used, which is be- on the idea that someone will ‘project’ his/her own
lieved to be more accurate in providing the reading. beliefs or feelings onto an imaginary situation or
l Biophysiologic measures are more objective in person. It enables the respondents to mention
nature. For example, a health care professional more about the research subject than they can say
measuring temperature or blood pressure is a spontaneously, accessing meanings, thoughts or
good example of objective measurements. feelings that are not readily available.
l Biophysiologic methods provide valid measures for
targeted variables. For example, the thermometer DEFINITION OF PROJECTIVE TECHNIQUE
will only record temperature and not blood pres- These are methods for measuring psychologic
sure. This means biophysiologic measurement tools attributes (value, attitudes and personality) by
only measure what they are supposed to measure. providing respondents with unstructured stimuli
l Easy access to most of the instruments used for to which they respond.
biophysiologic measurements. The instruments
used for biophysiologic measurements are easily TYPES OF PROJECTIVE TECHNIQUES
available in hospital settings, which are used for Lindzey proposed a five-way classification of
routine patient care. projective methods based on types of response.

DISADVANTAGES OF BIOPHYSIOLOGIC Association Technique (Word Association

METHODS Technique)
l Some of the instruments used for biophysio- This technique requires the subject to respond at
logic measurements are very expensive. the presentation of stimulus (word) with the first
l The use of biophysiologic measurement instru- thing that comes to mind, so as to extract infor-
ments requires significant amount of training, mation regarding words that have maximum
knowledge and experience. association (meaning). In this sort of test, the
 Measurements and Methods of Data Collection 175

respondent is asked to mention the first word that l Rosenzweig test: This test uses a cartoon format
comes to mind, e.g. interviewer says cold and the wherein we have a series of cartoons with empty
respondent may say hot. This technique is mostly (no words inserted in) ‘balloons’. The respon-
used in marketing research to find the quality that dents are asked to put their own words in the
is mostly associated with a brand of a product. empty balloon space.
CAN’T YOU DO

Completion Techniques ANYTHING RIGHT?

Completion projective techniques supply the sub-

ject with stimulus that is incomplete and the sub-
ject is required to complete it as he/she wishes.
l Sentence completion test: The subject is required

to complete a sentence. It reveals his or her at-

titude towards the topic. This technique permits Picture for Rosenzweig Test
the testing not only of words but of ideas as
well, and thus helps in developing hypothesis l Rorschach test: This test consists of 10 cards hav-
and construction of questionnaires. In addition, ing prints of inkblots, wherein the design happens
this technique is also quick and easy to use. to be symmetrical but meaningless. Respondents
For example, working with others all the time are asked to describe what they perceive in them.
is ………. l Holtzman inkblot test (HIT). It is a modification
Subjects might respond with words, such as of the Rorschach test and consists of 45 cards,
mad, nervous, sick, tired (negative effect) or which are based on colour, movement, shading
contended, good, happy (positive effect). and other factors. One response per card is in-
l Story completion test: The researcher contrives terpreted at three levels of appropriateness.
stories instead of sentences and asks the infor- l Tomkins–Horn picture arrangement test: This test
mants to complete them. is designed for group administrations with five
plates, each containing three sketches that may
Construction Techniques be arranged in different ways to portray se-
Here the focus is on the product as it is related to quences, which participants consider reasonable.
the subject. The subject is required to produce or
Expressive Technique
construct something at direction, usually a story
or a picture. There are several pictorial techniques A subject is required to draw, role play, act or paint
used in construction technique. The important a particular situation or concept. Expressive tech-
ones include the following: niques generally concentrate on the way in which
l Thematic apperception test (TAT). The TAT con- something is constructed by the subject, as against
sists of a set of pictures of a normal every day what it represents.
event and the respondents are asked to describe
Ordering Choice
what they think the pictures represent.
This type of projective technique is most frequently
used in quantitative researches. It is generally used
informally in qualitative researches when the sub-
jects explain why specific things are ‘most’ or ‘least’
important, or to order, rank or categorize some fac-
tors related to topics under research.

Miscellaneous Types
l Drawing a picture: Each participant is given a
Picture for TAT blank piece of paper and a pencil, and is asked
176 Research Methodology and Biostatistics

to draw a picture that might offer some insights languages, do not require reading and in some
on behaviour and attitudes. cases do not even involve speech at all. Thus,
l Creating families: Using a family analogy, this their scope is wider than that of rating proce-
technique is used to uncover relationships be- dures or self-reporting.
tween the target category and other categories.
l Using fantasy and daydreams: Participants are DISADVANTAGES OF PROJECTIVE TECHNIQUE
asked to describe topic-related fantasies and l Interpretative skills are needed on the part of
daydreams in order to bring to the surface char- the researchers to handle the complexity of in-
acteristics or features that are important to the terpretation of projective techniques. Research-
study. ers who are skilled and trained in interpretation
l Clay modelling: Participants are provided with of this type of information are required because
different coloured clays and are requested to the responses have little meaning without care-
give shape to them to reflect their feelings and ful interpretation.
thoughts. l In the projective techniques interpretation of
responses, a considerable degree of subjectivity
ADVANTAGES OF PROJECTIVE TECHNIQUE is involved and experts frequently do not agree
l The amount, richness and accuracy of the infor- among themselves.
mation collected are reliable to an appreciating l As highly skilled research staff have to be em-
extent. ployed, projective techniques are expensive to
l If the research is related to personality, beliefs, administer.
values, motivation or other aspects related to l With statistically significant samples, it is diffi-
individuals, their behaviours and unique cogni- cult to perform projective tests. This restricts
tions, projective techniques can make a specific generalization of the results to the population as
contribution. a whole.
l These techniques give a view of the total func- l To project themselves into the roles that the re-
tioning of individuals. searchers want them to do, it may be difficult to
l In the aspect of conventional focus group get subjects. Some subjects may disagree to take
discussions or intensive individual interviews, part in role-playing exercises.
different projective techniques are generally
employed. Once the subjects are comfortable
with each other and the moderator, these tech- Q-SORTS
niques may be employed at various stages in the The history of the use of Q-sorts in research
discussion process. is relatively new. It was developed in the 1930s by
l At the start of a qualitative research project, the British physicist–psychologist William Ste-
projective techniques may be employed as a phenson. The idea behind the development of
basis of ‘breaking the ice’ in a focus group this method was to inquire into the subjectivity of
discussion. the human mind. Examples of such subjectivity
l Rendering new energy to a group discussion are limitless and include aesthetic judgement,
and enlightening the tone or mood of the pro- appreciation of art, preferences for music,
ceedings, subjects enjoy projective techniques. experiences of family after tragic events and
l The tasks provided to the subjects are usually attitudes towards political groups. These were
unique and unstructured. They are encouraged difficult, if not impossible, areas that could not
to respond with a wide range of ideas; there are be measured and reported scientifically by the
no right or wrong answers to the exercise. conventional quantitative methods available at
l In terms of academic or literacy skills, most of that time.
these techniques make very less demand. They, Q-sorts are powerful tools in which the par-
generally, are largely independent of particular ticipants are presented with a card arrangement.
 Measurements and Methods of Data Collection 177

Participants are provided prewritten cards with l Rather than superficial study of a large number
words, phrases or statements, and are asked to of cases, it concentrates on in-depth analysis of
arrange these cards in an order along with a a small number of cases.
specific bipolar dimension. There are approxi-
mately 60–100 such cards that can be sorted out DISADVANTAGES OF Q-SORTS
in 9–11 piles with a few number of cards placed l Q-sorts takes much time and is an exhaustive
in each pile. procedure.
l It is not very cost-effective.
PURPOSES OF Q-SORTS l Only a small sample can be studied.
l Q-sorts can be used in a variety of ways, de- l It cannot be used to test hypothesis.
pending on the interest of the researchers. As l Population and sample are not rigidly distributed.
one of the major variables in health science
research is often the subjective viewpoint of
individuals, Q-sorts can be a perfect tool in a VIGNETTES
variety of settings. With appropriate modifica- In order to explore people’s beliefs, perceptions
tions, the scope of the research can be ex- and meanings about some particular situations,
panded to answer many difficult issues. For vignettes act as a valuable technique; they are es-
example, to analyse the doctor–patient rela- pecially helpful for important inquiry areas that
tionship in greater depth, data derived from may not be promptly accessible through any other
the Q-sort can be used to categorize major means. However, the problematic relationship
models of doctor–patient relationship, to between action and belief must be heeded when
determine the physicians’ qualities that are utilizing this technique, either in isolation or
important for patients and to identify the den- as part of a multimethod approach. In social
tists’ attributes that are associated with poor research, vignettes may be used for three primary
evaluation from patients. purposes: (1) to clarify people’s judgements,
l Q-sorts can also be applied before and (2) to permit actions in context to be explored and
after an intervention to evaluate the effective- (3) to give a less threatening and less personal way
ness of intervention. For example, we can utilize of exploring sensitive topics. Vignettes facilitate
the Q-method to evaluate whether attitudes or participants to qualify the situation in their own
perceptions of doctors change following imple- terms in qualitative research. The questions posed
mentation of a course on doctor–patient rela- to respondents after vignettes are either open-
tionship. Thus, the Q-method has the potential ended (How do you describe the quality of life of
to be used in diverse situations depending upon these cancer patients?) or close-ended. (Rate the
the interest of the researchers and the area of quality of life of this cancer patient on a 10-point
study. scale given as follows.)

ADVANTAGES OF Q-SORTS DEFINITIONS OF VIGNETTE

l Q-sorts provides in-depth study of small- The vignette technique is a method that can
sample populations elicit perceptions, opinions, beliefs and atti-
l It facilitates exploratory research tudes from responses or comments to stories
l A well-developed theoretical literature supports depicting scenarios and situations. Vignettes
and guides its usage. are employed in different ways and for different
l Through a person’s self-reference, it captures purposes.
subjectivity in operation. Vignettes are ‘short stories about hypo-
l There is no need for participants to be ran- thetical characters in specified circum-
domly selected. stances to whose situation the interviewee is
l Internet facilitates its administration. invited to respond.’ (Finch, 1987)
178 Research Methodology and Biostatistics

Vignettes are short scenarios in written become increasingly popular with social re-
or pictorial form, intended to elicit re- search. However, although it is often used as a
sponses to typical scenarios. (Hill, 1997) warm-up exercise to get participants to start
talking to each other, little has been done and
Examples: (1) Vignettes can be used in exploring written about the use of vignettes in groups.
ethical frameworks by mentioning the thinking of l Used in multimethod approach: Alongside
children regarding family issues. (2) Research into other data collection methods, vignettes have
postdivorce family life uses vignettes as a beneficial been used as a complementary technique.
method of exploring moral codes of young people They can be used either to generate data not
and their contingent status within various con- tapped by other research methods (such as
texts. observation or interviews) or to enhance ex-
isting data; for example, in order to explore
USES OF VIGNETTES children’s experiences and perceptions of
l Tapping general beliefs and attitudes: About a long-term foster care using vignettes, response
specific situation, vignettes are used to elicit cards, games, unfinished sentences, postal
cultural norms derived from respondents’ boxes and other visual stimuli to achieve an
beliefs and attitudes. These are also used to insider’s position on children’s value systems
explore participants’ moral codes and ethical and perceptions.
frameworks.
l Sensitive topics exploration: Whenever par- DESIGNING GOOD VIGNETTES
ticipants find it difficult to discuss, vignettes To guide the design of research that uses vignettes,
can be useful in exploring topics that are po- some principles can be distilled from literature:
tentially sensitive. Giving comments on a l Stories must appear real and plausible to par-

story is less personal than talking about direct ticipants. Some researchers have directly built
experiences; hence it is often regarded by par- their vignettes based on actual experiences, for
ticipants as less threatening. Vignettes offer example, by utilizing situations given by partici-
participants the opportunity to have greater pants in the pilot stage of the research, or indi-
control over the interaction by enabling them rectly by using agony aunt letters in teenage
to ensure at what stage, if at all, they intro- magazines. However, if the participant is in a
duce their own experiences to highlight their similar position, it is crucial to keep in mind
abstract responses. that telling participants about a potentially up-
l Comparison of perceptions of disparate setting scenario based on a real experience may
groups: Vignettes offer the possibility of exam- be distressing.
ining different interpretations of a groups’ ‘uni- l Eccentric characters and disastrous events must

form’ situation. Barter and Renold (1999) used be avoided; the vignette should reflect ‘mun-
vignettes to explore residential care workers’, dane’ occurrences instead. Although some re-
young people and managers’ evaluations and searchers also state that it can be advantageous
perceptions regarding different forms of vio- in some circumstances to incorporate some
lence between children in their children’s ‘unusual occurrences’ into the storyline.
homes research. Responses were then analysed l Vignettes should contain sufficient context for

to mention the level of discord or harmony respondents to understand the situation being
between child and adult evaluations, thus giv- depicted. It should be vague enough to ‘force’
ing benchmarks for understanding differences participants to provide additional factors influ-
in interpretation. encing their decisions.
l Used in focus group discussions: Vignettes can l Ability of the participants may be enhanced

be used differently with participants within a to engage with the story if they have personal
‘focus’ group or individually – a way that has experience of the situation described.
 Measurements and Methods of Data Collection 179

l Vignettes should be presented in an appropriate l Volume of information can be difficult to record,

format. This may be particularly important for can be very large and may lessen data clarity.
research with young people and children. Vi- l Cover limited sample size.
gnettes have also been presented on computers, l Involve subjective interpretations to a wider ex-
videotape, and through music videos and music; tent.
although written narratives are most common.
l Initially, participants might give socially desir-
able responses; only after checking that they will VALIDITY OF A RESEARCH TOOL
not be threatened or laughed at, will they men- Validity of an instrument refers to the degree to
tion how they actually believe they would re- which an instrument measures what it is supposed
spond to the situation. Some studies query to be measuring. For example, a tool designed for
participants to elaborate how they feel the per- measuring the burnout among critical physicians
son in the story would act and why, and then must only measure the burnout; it cannot be con-
how they would respond and why. If the charac- sidered a valid instrument if it measures an attri-
ter acts in a socially unacceptable way, by reduc- bute other than burnout. Similarly, if a researcher
ing the pressure to provide socially desirable develops a tool to measure pain, and if it also in-
responses, participants can be given the ‘free- cludes items to measure anxiety, it cannot be con-
dom’ to mention how they think they would sidered a valid tool. Therefore, a valid tool should
actually respond in that situation. only measure what it is supposed to be measuring.
l It is important that the stories presented in the
vignettes be easily comprehensible, internally DEFINITIONS OF VALIDITY
consistent and not very complex. According to Treece and Treece, ‘Validity refers to
l To see if any significant differences emerge, in an instrument or test actually testing what it is
some situations it may be desirable to include a supposed to be testing’.
control vignette. For example, looking at abused According to Polit and Hungler, ‘Validity refers
and nonabused children conceptual systems re- to the degree to which an instrument measures
lated to personal body safety and including a what it is supposed to be measuring’.
control about a road traffic accident and finding According to American Psychological Founda-
significant differences with respect to body tion, ‘Validity is the appropriateness, meaning,
space and none in relation to the control. fullness, and usefulness of the interference made
from the scoring of the instrument’.
ADVANTAGES OF VIGNETTES Validity is the appropriateness, completeness,
l Allow face-to-face contact with respondents. and usefulness of an attribute measuring research
l Give opportunity to explore topics in depth. instrument.
l Allow interviewer to elaborate or help clarify
questions, thus increasing the likelihood of TYPES OF VALIDITY
useful responses. Basically, validity is classified into the following
l Make it easy for respondents to reply to the four categories:
concerned topic. l Face validity: Face validity involves an overall

l Provide detailed data. look of an instrument regarding its appropri-

ateness to measure a particular attribute or
DISADVANTAGES OF VIGNETTES phenomenon. Though face validity is not con-
l Interviewee may distort information through sidered a very important or essential type of
selective perceptions, recall error, desire to validity for an instrument, it may be taken
please interviewer. into consideration while assessing for other
l Interviewer clarifications resulting in inconsis- aspects of validity of a research instrument. In
tencies. simple words, this aspect of validity refers to
180 Research Methodology and Biostatistics

the face value or the outlook of an instrument. less-than-perfect criterion because the rating
For example, a Likert scale designed to mea- found by empirical and supervisory methods
sure the attitude of health care professionals may be computed mathematically, which can
towards patients admitted with HIV/AIDS: a correlate the score of the instrument with
researcher may judge the face value of this scores of the criterion variable. Here the range
instrument by its appearance, that is, whether of the coefficient should be 0.70. Criterion-
it looks good or not; but it does not provide related validity may be differentiated by pre-
any guarantee about the appropriateness and dictive and concurrent validity.
completeness of the research instrument with  Predictive validity: It is the degree of forecast-

regard to its content, construct and measure- ing judgement; for example, some personality
ment score. tests on academic futures of students can
l Content validity: It is concerned with the scope be predictive of behaviour patterns. It is the
of coverage of the content area to be measured. differentiation between performances on
More often it is applied in tests of knowledge some future criterion and an instrument’s
measurement. It is also mostly used in measur- ability. An instrument may have a predictive
ing complex psychologic tests of a person. It is a validity when its score significantly correlates
case of expert judgment about the content area with some future criteria.
included in the research instrument to measure  Concurrent validity: It is the degree of the

a particular phenomenon. Judgement of the measures at present. It relates to the present

content viability may be subjective and is based specific behaviour and characteristics; hence
on the opinion of previous researchers and ex- the difference between predictive and concur-
perts, about the adequacy, appropriateness and rent validity refers to the timing pattern of
completeness of the content of the instrument. obtaining measurements of a criterion.
Generally, this viability is ensured through the l Construct validity: A construct is founded in
judgments of experts about the content. this type of validity, such as a researcher may
l Criterion validity: This type of validity is a have designed an instrument to measure the
relationship between measurements of the in- concept of pain in amputated patients. The pain
strument with some other external criteria. For pattern may be due to anxiety; hence the results
example, a tool is developed to measure the pro- may be misleading. Construct validity is a key
fessionalism among physicians and to assess the criterion for assessing the quality of a study, and
criterion validity, physicians were separately it has most often been addressed in terms of
asked about the number of research papers they measurement issues. The key construct validity
published and the number of professional questions with regard to measurements are:
conferences they have attended. Later a correla- What is this instrument really measuring? Does
tion coefficient is calculated to assess the crite- it adequately measure the abstract concept of
rion validity. This tool is considered strong in interest? Construct validity gives more impor-
criterion validity if a positive correlation exists tance to test relationship predicted on theoreti-
between the score of the tool measuring profes- cal measurements. The researcher can make
sionalism and the number of research articles predictions in relation to other such types of
published and the professional conferences at- constructs. One method of construct validation
tended by the physicians. is known as group technique.
The instrument is valid if its measurements
strongly respond to the score of some other
valid criteria. The problem with criterion-re- RELIABILITY OF A RESEARCH TOOL
lated validity is to find a reliable and valid The quality and adequacy of the quantities of
external criterion. Mostly, we are to rely on a data can only be assessed by establishing the
 Measurements and Methods of Data Collection 181

reliability of an instrument. Reliability is the times. Stability is estimated to make sure that the
degree of consistency with which the attributes research instrument is consistent in providing
or variables are measured by an instrument. similar results with repeated administration.
For example, a blood pressure-measuring in- It is also known as the reliability of the test–
strument gave a reading of 120 mm Hg systolic retest function. To measure test–retest reliability,
blood pressure; after some time, when the blood the test is given twice at two different points in
pressure is again measured for the same subject, time. It is used to measure the consistency of a
it gave a reading of 160 mm Hg systolic blood test across time. This type of reliability supposes
pressure. In this situation, this instrument is that there will be no change in the quality or
not considered reliable. However, if a research construct being assessed. Test–retest reliability is
instrument yields similar or close to similar best used for things that are strong over time
results on repeated administration of the instru- (e.g. intelligence). Reliability will be, generally,
ment, it is considered as a highly reliable re- higher when little time has been taken between
search instrument. tests.
Reliability pertains to the consistency of a mea- The test–retest method is a relatively easy and
sure. A test is thought to be reliable if we get the straightforward approach to establish reliability.
same result in a repeated manner. For example, if It is used for questionnaire, observation check-
a test is designed to scale a trait (e.g. introversion), list, observation-rating scales and physiologic
each time the test is put on a subject, the result measurement tools. However, it has one basic
should be similar. Although it is not possible to problem – that there are many traits that may
calculate the exact reliability, there are various change with time, such as attitude, behaviour,
ways to measure reliability. mood knowledge, satisfaction and physical
conditions.
DEFINITIONS OF RELIABILITY
Reliability is the degree of consistency and accu-
racy with which an instrument measures the Statistical Calculation (Test–Retest Method)
attribute for which it is designed to measure. The procedure for calculating test–retest reliabil-
‘Reliability is defined as the ability of an instru- ity (stability) of research instrument involves the
ment to create reproducible results.’ Therefore, following steps.
reliability is concerned with consistency of mea- l Administration of a research instrument to a

surement tools. A tool can only be considered reli- sample of subjects on two different occasions
able if it measures an attribute with similar results l Scores of the tool administered at two different

on repeated usage. occasions are compared and calculated by using

Formula 7.1 for correlation coefficient.
MEASURING RELIABILITY l The correlation coefficient reveals the magni-

There are several ways to measure the reliability tude and directions of relationship between
of research tools, which depends on several fac- scores generated by a research instrument at
tors, such as the nature of the instrument two separate occasions.
as well as aspects of reliability the researcher l Interpretation of results: The results of the correla-

wants to measure. The main aspects of reliabil- tion coefficient range between 21.00 through
ity considered important in quantitative re- 0.0 and 11.00, and the results are interrelated as
search include stability, internal consistency follows:
and equivalence.  11.00 score indicates prefect reliability.

Stability: The stability aspect of reliability  0.00 score indicates no reliability.

means that the research instrument provides the  A score above 0.70 indicates an acceptable

same results when used consecutively two or more level of reliability of a tool.
182 Research Methodology and Biostatistics

Formula 7.1: Karl Pearson’s correlation coefficient Formula 7.2

formula for estimation of reliability:
N XY – ( X ) ( Y ) r5
(
∑ x–x y– y )( )
r
( ) ( )
2 2
N X 2 – ( X 2 ) N Y 2 – ( Y )
2 ∑ x–x –∑ y– y

In the aforementioned formula r 5 correlation l In split-half, to overcome the underestimation

coefficient, N 5 number of pairs of scores, oXY 5 of reliability of the entire scale, as Formula 7.2
sum of the products of paired scores, oX 5 sum has estimated reliability of only half the items,
of X scores, oY 5 sum of Y scores, oX25 sum of the following formula (Formula 7.3) is used to
squared X scores and oY2 5 sum of squared Y estimate the reliability of the entire test.
scores. Formula 7.3
Internal consistency: It is also called homogene-
ity. Internal consistency ensures that all the sub- 2r
r1 
parts of a research instrument measure the same 1 r
characteristics. For example, a patient’s satisfaction
measurement scale developed to measure his or her where r 5 the correlation coefficient computed on
satisfaction with physician care must include all the the split-halves with Formula 7.2.
subparts related to the measurement of satisfaction r1 5 the estimated reliability of the entire test
l The split-half technique is frequently used to
with physician care only. It should not include the
patient’s satisfaction measurement scale with other estimate the internal consistency; however, a
aspects of care; including a subpart related to pa- more preferred method, Cronbach’s alpha or
tient’s satisfaction with other team members of coefficient alpha, may be calculated by using the
health care would be inappropriate in this scale. following formula (Formula 7.4):
A research tool can only be considered inter- Formula 7.4
nally consistent if all the subparts of the tool are k  ∑  12 
measuring the same characteristics or phenom- r  1
k 1   y 2 
ena. One of the most primitive approaches of as-
sessing the internal consistency is the split-half
technique. However, to cover its drawback as an where r 5 the estimated reliability
estimation tool of different correlation coefficient k 5 the total number of items in the test
results with different combinations of split-half s21 5 the variance of each individual items
(odd–even, first-half–second-half, etc.), this sy2 5 the variance of the total test scores
method is replaced by formulas that compensate o5 the sum of
l Equivalence: This aspect of reliability is esti-
this deficiency, such as Cronbach’s alpha and the
Kuder–Richardson formula 20. mated when a researcher is testing the reliability
of a tool which is used by two different observ-
Statistical Calculation (Split-Half Method) ers to observe a single phenomenon simultane-
Procedure for calculating the split-half reliability of a ously and independently or two presumably
research instrument involves the following steps: parallel instruments are administered to an in-
l Divide items of a research instrument into two dividual at about the same time. For example, a
equal parts through grouping either in odd rating scale is developed to assess the severity
number question and even number question or level of pin site infection among patients with
first-half and second-half item groups. external skeletal fixation; this rating may be ad-
l Administer two subparts of the tool simultane- ministered to observe the severity level of pin
ously, score them independently and compute the site infection among patients with external fixa-
correlation coefficient on the two separate scores tion by two different observers simultaneously
by using Formula 7.2. but independently.
 Measurements and Methods of Data Collection 183

This is also known as interrater or interobserver application of an intervention. Pilot results

reliability, which is estimated by the administra- can depict feasibility and check modifications
tion of tools to observe a single event simultane- required in the design of a larger, ensuing
ously and independently by two or more trained hypothesis testing study.
observers. This reliability may be computed by l A pilot study can also be utilized to measure the
using the following equation: feasibility of assessment procedures, recruit-
Number of agreements
ment, randomization, retention, new methods
r and implementation of novel interventions.
Number of agreements  Number of disagreements
l One of the merits of conducting a pilot study is
that it might provide advance warning about
where research protocols may not be followed,
PILOT STUDY where the main research project could not work,
l In biomedical research, the term pilot study is or whether proposed methods or instruments
utilized in two different ways. It refers to the so- are inappropriate or too complicated.
called feasibility studies that are trial run or
small-scale versions, which are done in prepara- DEFINITIONS OF PILOT STUDY
tion for the major study. However, a pilot study Pilot study is a trial study carried out before a
can also be the ‘trying out’ or pretesting of a research design is finalized to assist in defining
specific research instrument. the research question or to test the feasibility, reli-
l A pilot study is referred to a small-scale pre- ability, and validity of the proposed study design.
liminary tryout of the method to be used in an Pilot study is a smaller version of a study
actual large study; it acquaints the researcher carried out before the actual investigation is
with the problems that need to be corrected in done. Researchers use information gathered in
proportion for the larger research study; a pilot pilot studies to refine or modify the research
study is done to provide the researcher with an methodology for a study and to develop large-
opportunity to try out the procedure, methods scale studies.
and tools of data collection. Pilot study is a small-scale study conducted to
l In other words, the pilot study is a miniature test the plan and method of a research study.
trial run of the methodology planned for the Pilot study is a small study often carried out to
major research study, which facilitates to im- help in preparing a larger and more comprehen-
prove the methodology of the study, can assess sive investigation.
the feasibility of the study and may identify the Pilot study is a trial run or a small preliminary
problems that may be faced by the researcher in test of an intervention of an evaluation activity
the actual large research project. (e.g. an instrument or sampling procedure). The
l A small-scale test of the methods and procedures results of the pilot are utilized to facilitate the pro-
to be used on a larger scale is called a pilot study. gramme or evaluation procedure being piloted
The main intention of conducting a pilot study is before it is used on a larger scale.
to ascertain the feasibility of a method that is re-
quired to ultimately be used in a large-scale PURPOSES OF PILOT STUDY
study. This includes all types of research studies. There are several reasons or purposes to under-
l A fundamental phase of the research process is take a pilot study; besides these reasons, pilot
represented by a pilot study. The intention of studies are also done to convince funding bodies
conducting a pilot study is to check the feasibil- that their research proposal for the main study is
ity of a method intended to be utilized in a worth funding. Pilot studies are employed for a
large-scale study. range of different purposes or reasons:
l A pilot study is the required initial step in ex- l Assessing feasibility and practicability of the

ploring a unique intervention or an innovative research study

184 Research Methodology and Biostatistics

l Assessing the availability of study subjects used to evaluate the feasibility of recruitment,
l Establishing validity and reliability of the randomization, retention, assessment proce-
research tools dures and implementation of the novel inter-
l Ensuring the appropriateness of methods and vention, and each of these can be quantified;
procedures of data collection study components that are deemed infeasible or
l Understanding the study variables and other unsatisfactory should be modified in the subse-
confounding variables quent trial or be removed altogether
l Estimating the actual time and potential
problems the researcher may encounter during PROBLEMS AND ISSUES OF PILOT STUDY
the actual large research study It should be recognized that pilot studies may
l Refining the study methodology also have a number of problems, issues and limi-
l Planning for the data analysis and interpreta- tations. Following are some of the significant
tion of the final larger research problems or issues of pilot study, which a re-
l Developing and testing sufficiency of research searcher may consider while carrying out the pi-
instruments lot study:
l Assessing the practicality of a full-scale study/ l It may be possible that the researcher makes

survey inaccurate predictions or assumptions on the

l Assessing whether the research protocol is prac- basis of pilot data, specifically when the data of
ticable and workable the pilot study are used for estimating the sam-
l Establishing that the technique and sampling ple size. Thus it is always better to use the find-
frame are effective ings of a previous full-scale similar study con-
l Assessing the probable success of the proposed ducted for estimating a sample size rather than
study subject recruitment approaches using the data from a pilot study for sample size
l Logistical problems identification by using estimation.
proposed methods l Contamination is another concern. This may
l Estimating variability in outcomes to help in arise in two ways: (1) where pilot participants
determining sample size are included in the main study, but new data are
l Preliminary data collection collected from these people and (2) where data
l Determining the resources (e.g. finance, staff) from the pilot study are included in the main
required for a planned study results. To avoid the risk of contamination, the
l Assessing the proposed data analysis techniques following is considered:
for potential problems  The data collected during pilot study are not in-
l Developing a research plan and research question cluded in analysis of the main study data. The
l Training a researcher in various elements of the other reason that the pilot study data are not in-
research process cluded in main study data analysis is that after
l Convincing funding bodies that the research the pilot study, there may be changes in data col-
team is knowledgeable and competent lection tool(s) and study methodology. However,
l Convincing funding bodies that the main study if there is no change in the study methodology
is worth funding and feasible and data collection methods and tool(s), the pi-
l Convincing other institutes and stakeholders lot study data may be included in the main study
that the main study is worth supporting results.
l Containing the basic information to plan the  The main study is generally not conducted in

sample size for the larger study the same area where the pilot study was con-
l Pilot study results can guide in the designing ducted to avoid the risk of contamination.
and implementation of larger scale efficacy So to sum up, pilot study is the preresearch activity
studies; there are several aspects of randomized to refine research methodology and plan for ap-
control trial (RCT) feasibility that are informed propriate use of resources with sound beforehand
by conducting a pilot study; a pilot study can be information about the main study parameters.
 Measurements and Methods of Data Collection 185

MULTIPLE CHOICE QUESTIONS

1. Which among the following is not a type of 6. Which of the following is not a method of data
reliability? collection?
(a) Split-half (a) Interviews
(b) Content (b) Experiments
(c) Internal consistency (c) Questionnaires
(d) Test–retest (d) Observations
2. Which among the following statements describes 7. Development of a questionnaire is based on several
test–retest reliability? principles. Which among the following is not one
(a) Measure of consistency of test scores over time of those principles?
(b) Measure of consistency with which a test (a) Do not use ‘leading’ or ‘loaded’ questions
measures a single construct or concept (b) Avoid double negatives
(c) Measure of consistency of scores obtained from (c) Avoid double-barrelled questions
two equivalent halves of the same test (d) Avoid using multiple items to measure a single
(d) Measure of degree of agreement between two or construct
more scorers, judges or raters
8. Which type of interview among the following
3. Which of the following types of reliability indicates allows the questions to emerge from the immediate
the consistency of test scores over time? context or course of things?
(a) Equivalent forms reliability (a) Closed quantitative interview
(b) Interscorer reliability (b) Informal conversational interview
(c) Split-half reliability (c) Interview guide approach
(d) Test–retest reliability (d) Standardized open-ended interview
4. Identify the following terms that most closely refers 9. Which among the following is not one of the six
to a judgement of the extent to which scores from a major methods of data collection that are used by
test can be utilized to infer, or predict the examinees’ educational researchers?
performance in some activity: (a) Observation
(a) Face validity (b) Questionnaires
(b) Criterion-related validity (c) Interviews
(c) Inference validity (d) Checklists
(d) Content reliability
10. A biomedical researcher decides to collect the data
5. Which type of reliability indicates the consistency of from medical ward nurses; the nurses get them-
a group of individuals’ scores on two equivalent selves transferred to the medical ward to be able to
forms of a test designed to measure the same observe and collect data about the behaviour of
characteristic? nurses towards patients with chronic illness. The
(a) Test–retest researcher is using which observation role?
(b) Split-half (a) Complete participant
(c) Split-forms (b) Participant-as-observer
(d) Equivalent forms (c) Complete observer
(d) Observer-as-participant

Answers of Multiple Choice Questions

1. b; 2. a; 3. d; 4. b; 5. d; 6. b; 7. d; 8. b; 9. d; 10. a.

REFERENCES
Barter, C., & Renold, E. (1999). Physical and sexual violence Bellak, L. (1992). Projective techniques in the computer age.
amongst children in residential settings. ,http://www.brunel. Journal of Personality Assessment, 58(3), 445–453.
ac.uk/dept/law/vrp.. Accessed 12 March 2014.
186 Research Methodology and Biostatistics

Berkman, H. W., & Gilson, C. (1986). Consumer behavior: con- Leierer, S. J., Strohner, D. C., Leclere, W. A., Cornwell, B. J., &
cepts and strategies (3rd ed.). Boston: Kent. Whitten, S. L. (1996). The effect of counsellor disability,
Bhaduri, A. (1998). Method of data collection. Nursing Research attending behaviour and client problem of counselling.
Society of India workshop proceedings. 4th National Work- Rehabilitation Counselling Bulletin, 40, 82–95.
shop from 9th to 13th September, 1996. New Delhi: Nursing Levy, S. J. (1981). Interpreting consumer mythology: a structural
Research Society of India. approach to consumer behaviour. Journal of Marketing, 45, 49–61.
Burns, L. D., & Lennon, S. J. (1993). Social perception: methods for Loudon, D. L. & Della-Bitta, A. J. (1993). Consumer behaviour:
measuring our perception of others. International Textile and concepts and application (4th ed.). New York: McGraw-Hill.
Apparel Association Special Publication, 5, 153–159. MacAuley, C. (1996). Children in long term foster care: emotional
Carlson, B. E. (1996). Dating violence: student beliefs about con- and social development. Hampshire: Avebury.
sequences. Journal of Interpersonal Violence, 11, 3–18. Macfarlane, J. W., & Tuddenham, R. R. (1951). Problems in the
Catterall, M., & Ibbotson, P. (2000). Using projective techniques validation of projective techniques. In H. H. Anderson & G. L.
in education research. British Educational Research Journal, Anderson (Eds.), An introduction to projective techniques. En-
26(2), 245–247. glewood Cliffs, NJ: Prentice-Hall.
Churchill, G. A. (1991). Marketing research methodological foun- Maclean, C. (1999). Children, family, community and work: an
dations (5th ed.). Chicago: Dryden. ethnography of the oil and gas industry in Scotland, (Depart-
Cohen D., & Strayer, J. (1996). Empathy in conduct-disordered and ment of Management Studies). ,http://www.abdn.ac.uk..
comparison youth. Developmental Psychology, 32, 988–998. Accessed 12 March 2015.
Crumbaugh, J. C. (1990). The meaning of projective techniques. Miller-Perrin, C. L., Wurtele, S. K., & Kondrick, P. A. (1990). Sexu-
In J. C. Crumbaugh, J. Graca, R. R. Hutzell, M. F. Whiddon, & ally abused and non-abused children’s conceptions of personal
E. C. Cooper (Eds.), A primer of projective techniques of psycho- body safety. Child Abuse and Neglect, 14, 99–112.
logical assessment. San Diego: Libra. Neale, B. (1999). Post divorce childhoods. ,http://www.leeds.ac.
Dillon, W. R., Madden, T. J., & Firtle, N. H. (1998). Marketing re- uk/family.. Accessed 12 March 2015.
search in a marketing environment. St Louise: Times Mirror. Peterson, D. L., & Pfost, K. S. (1989). Influence of rock videos on
Faia, M. A. (1979). The vagaries of the vignette world: a document on attitudes of violence against women. Psychological Reports, 64,
Alves and Rossi. American Journal of Sociology, 85, 951–954. 319–322.
Finch, J. (1987). The vignette technique in survey research. Sociology, Piotrowski, C., Keller, J. W., & Ogwa, T. (1993). Projective techniques: an
21, 105–114. international perspective. Psychological Reports, 72, 179–182.
Frary, R. B. (n.d.). Hints for designing effective questionnaire. Polit, D.F., Beck, C.T, & Hungler, B.P. (2001). Essentials of nursing
,http://www.Eric.Ed.gov.. Accessed 12 March 2015. research methods appraisal and utilization (5th ed.). New York:
Gordon, W., & Langmaid, R. (1988). Qualitative market Lippincott.
research. Aldershot: Gower. Rahman, N. (1996). Caregivers’ sensitivity to conflict: the use of vi-
Graham, J. R., & Lilly, R. S. (1984). Psychological testing. gnette methodology. Journal of Elder Abuse and Neglect, 8, 35–47.
Englewood Cliffs, NJ: Prentice-Hall. Sampson, P. (1986). Qualitative research and motivation research.
Green, J. (1984). Approaching the core of consumer behaviour. In R. M. Worcester & J. Downham (Eds.), Consumer market
Marketing Research, 16(3), 19–20. research handbook (3rd ed.). Amsterdam: Elsevier.
Guilford, J.P. (1954). Psychometric methods (2nd ed.). New York: Solomon, M. R. (1994). Consumer behaviour. Buying, having and
McGraw-Hill. being (2nd ed.). Boston: Allyn & Bacon.
Haire, M. (1950). Projective techniques in marketing research. Stolte, J. F. (1994). The context of satisficing in vignette research.
Journal of Marketing Research, 14(5), 649–656. Journal of Social Psychology, 134, 727–733.
Harden, J. (1999). Impact of risk and parental risk anxiety on the Treece, E.W, & Treece, J.W. (1986). Elements of research in nursing
everyday worlds of children. ,http://www.esrc. ac.uk/ (4th ed.). St Louis: The CV Mosby Company.
curprog.html.. Accessed 12 March 2015. Vitkoritch, M., & Tyrrell, L. (1995). Sources of disagreement in
Hazel, N. (1995). Elicitation techniques with young people, Social object naming, Quarterly Journal of Experimental Psychology,
Research Update, 12, Surrey, UK: Department of Sociology, Section A: A Human Experimental Psychology, 48A, 822–848.
University of Surrey, ,http://www.soc.surrey.ac.uk/sru/ Wade, A. (1999). New childhoods? Children and co-parenting after
SRU12.html.. Accessed 12 March 2015. divorce. ,http://www.leeds.ac.uk/family.. Accessed 12 March 2015.
Hill, M. (1997). Research review: participatory research with Wagner, E. E. (1995). A logical analysis of projective techniques
children. Child and Family Social Work, 2, 171–183. based on independence of items and latitude of response.
Hughes, R. (1998). Considering the vignette technique and its ap- Perceptual and Motor Skills, 81, 868–870.
plication to a study of drug injecting and HIV risk and safer Webb, J. R. (1992). Understanding and designing marketing
behaviour. Sociology of Health and Illness, 20(3), 381–400. research. London: Academic Press, 1992.
Janda, L. H. (1998). Psychological testing: theory and applications. Wilkinson, S. (1998). Focus group methodology: a review. Inter-
Boston: Allyn & Bacon. national Journal of Social Research Methodology Theory and
Kassarjian, H. H. (1974). Projective methods. In R. Ferber (Ed.), Practice, 1(3), 181–203.
Handbook of marketing research. New York: McGraw-Hill. Will, V., Eadie, D., & Macaskill, S. (1996). Projective and enabling
Kline, P. (1983). Personality: measurement and theory. London: techniques explored. Marketing intelligence and planning 14(6),
Hutchinson. 38–44. (Special issue: Qualitative Market Research.)
 CHAPTER OUTLINE
Analysis of Data 8
INTRODUCTION NON-PARAMETRIC TESTS

DEFINITIONS OF DATA ANALYSIS SOFTWARE ANALYSIS OF QUANTITATIVE DATA

ANALYSIS OF QUANTITATIVE DATA ANALYSIS OF QUALITATIVE DATA

PARAMETRIC TESTS

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Choose the relevant statistical methods and test the
• Define data analysis and other related quantitative data analysis.
terminologies. • Recognize the appropriate methods of qualitative data
• Discuss the steps of quantitative data analysis. analysis.
• Describe the steps of process of data analysis. • Interpret the data analysis and conclude the data analysis.

KEY TERMS

• Classification 189 • Editing 188 • Parametric tests 194

• Coding 188 • Inferential statistics 191 • Recontextualizing 200
• Compilation 188 • Level of significance 191 • Synthesizing 200
• Comprehending 200 • Measures of central • Tabulation 189
• Confidence interval 193 tendency 191 • Test of significance 194
• Data analysis 188 • Measures of dispersion 191 • Theorizing 200
• Degree of freedom 193 • Measures of relationship 191 • Type-I and type-II errors 191
• Descriptive statistics 190 • Nonparametric tests 196

pretation of data includes compilation, editing,

INTRODUCTION coding, classification and presentation of data.
Analysis and interpretation of data is the most im- The collected data are known as raw data; raw
portant phase of the research process. It involves data are meaningless unless certain statistical treat-
the computation of certain measures along with ment is given to them. The analysis of data means to
searching for patterns of relationship that exists make raw data meaningful or to draw some results
among data groups. Data collection is followed by from the data after proper statistical or analytical
the analysis and interpretation of data, where the treatment. In other words, the analysis and interpre-
collected data are analysed and interpreted in ac- tation of data represents the application of deductive
cordance with study objectives. Analysis and inter- and inductive logic to the collected raw data. Data
188 Research Methodology and Biostatistics

are often classified into subgroups and then analysed

and synthesized in such a way that hypothesis may ANALYSIS OF QUANTITATIVE DATA
be verified or rejected. Analysis of quantitative data deals with information
The analysis of data includes comparison of the collected during research study, which can be quan-
outcomes of various interventions/treatments tified, and statistical calculations be computed.
upon the several groups and the making of a deci- The data analysis process includes the follow-
sion as to the achievement of the goals of research. ing four steps:
Data relevant to each hypothesis must be assem- I. Data preparation (cleaning and organizing
bled in a quantitative form and tested to deter- data for analysis)
mine whether there is a significant difference in Data preparation involves logging or checking
the results obtained from the controlled groups. the data in, checking the data for correctness,
Usually, the analysis develops as a comparison entering the data into the computer, trans-
between groups. However, sometimes the type of forming the data and documenting as well as
data obtained tends to make them show the exist- developing a database structure to integrate
ing difference better by contrast or by summing different measures. Data preparation thus
up. The purpose of analysing the data collected in involves the following steps:
a study is to describe the data in meaningful - Compilation: The compilation process in-
terms, as the raw data collected do not answer the cludes gathering together all the collected
research questions or test research hypothesis. The data in a manner that a process of analysis
data used are to be systematically analysed so that can be initiated. While compiling data,
trends and patterns of relationships can be de- care is to be taken to arrange all the data in
tected. The analysis and interpretation of qualita- an order so that editing and the coding
tive and quantitative data follow different paths, process can be implemented with ease.
which are discussed in this chapter distinctly in - Editing: Editing implies the checking of
detail. The main purposes for analysis of data are the gathered data for accuracy, utility and
as follows: completeness. If the raw data are errone-
l To make the raw data meaningful ous or inconsistent, these deficiencies will
l To test null hypotheses be carried through all subsequent stages of
l To test the statistical significance of data or re- processing and will greatly distort the re-
lated data sult of any enquiry, so the editor or project
l To draw inferences and make the generalization director must see that none of the ques-
l To estimate parameters tions has been skipped, all the answers
have been recorded and all the replies are
internally consistent with each other.
DEFINITIONS OF DATA ANALYSIS - Coding: Coding is important for analysis as
Analysis is the process of organizing and synthe- numerous replies can be reduced to a small
sizing data so as to answer research questions number of classes through coding. The origi-
and test hypothesis. nal data are transformed into symbols com-
Analysis is referred to as a method of organizing patible with manual or computer-assisted
data in such a way that research questions can analysis. Coding can be carried out before or
be answered and hypothesis be tested. after the actual data are collected. Code is an
Analysis is the process of breaking a complex topic abbreviation, a symbol, a number or an alpha-
into smaller parts to gain a better understanding bet, which is assigned by the researcher to
of it. every schedule item and response category.
It is also defined as the process of systematically For coding, some important criteria should
applying statistical and logical techniques to be considered, such as categories should be
describe, summarize and compare data. exhaustive, so that all the responses can be
 Analysis of Data 189

classified in one or the other category; codes - Quantitative classification: Quantitative

should be mutually exclusive, which means data, such as age, height, weight, income,
that one code should be specific to only one Hb level, are classified based on quantita-
kind of information and not overlapping the tive classes such as:
criteria of other codes. Separate categories - Monthly income in rupees: ,5000,
can be created for recording ‘no response’ 5001–10,000, .10,000
and ‘no-knowledge responses’. - Tabulation: Tabulation is the recording of
Coding decisions may be taken at the designing the classified data in accurate mathemati-
stage of a research tool, which is helpful for cal terms; for example, marking and
computer tabulation. In case of manual cod- counting the frequency tallied. In simple
ing, some standard methods may be used, tabulation, there is an orderly arrange-
such as coding in the margin with a coloured ment of data in columns and rows. The
pencil or transcribing the data instrument to arrangement of the assembled data has to
a coding sheet. Care must be taken to avoid be done in concise and logical order. Tabu-
errors in the coding method. lation is a tedious process where raw data
- Classification: Data has to be classified, as are to be summarized and displayed in
many researches result in large volumes of compact form, which can be done manu-
raw data which must be reduced to homoge- ally or by computer. Tabulation helps in
neous groups. In the process of classification, detection of errors, omissions, comparison
we divide and arrange the entire data into and various statistical computations that
different categories, classifications, groups or may be carried out on the data.
classes on the basis of common characteris- - A table is a tabular representation of statis-
tics. When data are classified based on quali- tical data. Tabulation is the first step before
ties and attributes that are incapable of data can be used for further statistical
quantitative measurements, the classification analysis and interpretation. Tabulation
is said to be according to attributes, and means the systematic presentation of the
when data are classified on the basis of quan- information contained in the data in
titative measurements, the classification is rows and columns in accordance with
said to be according to class intervals. The some common features and characteris-
classification of data could be: tics. Rows are horizontal and columns are
- Geographic classification: This is a type of clas- vertical arrangements. Basically, tables are
sification in which data are classified based on of four types:
geographical categories, for example: - Frequency distribution tables: These tables
- Area of residence: Urban, semi-urban, present the frequency and percentage
rural, etc. distribution of the information collected,
- Districts: Sikar, Jaipur, Bikaner, etc. where an attribute is grouped into a num-
- Chronological classification: In this classifi- ber of classes, which may vary between
cation, data are classified based on time three and eight classes. Too many or too
period, such as days, months and years. few groups or classes may fail to reveal the
- Weekly profit of the company: Monday, salient features of the data. It is also kept in
Tuesday, Wednesday, Thursday, Friday, mind that class or group intervals are kept
Saturday constant. Table 8.1 presents an example of
- Qualitative classification: The data are clas- a frequency distribution table.
sified based on certain attributes, such as - Contingency tables: Tables that report
gender, religion and type of diseases. on the frequency distribution of two
- Type of gender: Male and female nominal variables simultaneously and
- Religion: Hindu, Muslim, Sikh, etc. that include the totals are known as
190 Research Methodology and Biostatistics

Table 8.1. ​Sociodemographic profile of patients

establish the relationship or association
(n 5 60) between them. These tables could be
2 3 2, 2 3 3 and 3 3 3 depending on
Sociodemographic Variables f (%) the number of variables on which the
Age (in years) subjects are cross-classified. The number
20–40 18 (30.0) of subjects in a cell is called the cell
41–60 42 (70.0) frequency. These tables are generally
Gender used in a chi-square test. Table 8.2 is an
Male 39 (65.0) example of a 2 3 2 contingency table.
Female 21 (35.0) - Multiple response tables: When classifica-
tion of the cases is to be done into catego-
Educational status
Illiterate 08 (13.3) ries that are neither exclusive nor exhaus-
Middle 17 (28.3) tive, a multiple response table is used. For
Senior secondary 20 (33.3) example, a patient can have two or more
Graduate and above 15 (25.0) complaints, while say only the major ones
Marital status may be listed. In such cases, the total sum
Married 52 (86.7) of frequencies would exceed the total
Unmarried 08 (13.3) number of subjects and may lead to con-
Occupation fusion. Therefore, the total number of
House wife/nonworking 28 (46.7) subjects in cases of multiple responses is
Agriculture/labour 15 (25.0) given as a base, and from this we calculate
Service 09 (15.0) the percentages. Example of a multiple
Business 08 (13.3)
response table is presented in Table 8.3.
Habitat - Miscellaneous tables: These tables are
Urban 37 (61.7) used to present data other than fre-
Rural 23 (38.3)
quency or percentage distributions, such
Religion as mean, median, mode, range and stan-
Sikh 31 (51.7)
dard deviation. A table is called miscel-
Hindu 29 (48.3)
laneous when the presentation of data
cannot be classified under the frequency
contingency tables. The categories con- distribution table, contingency table or
sidered should be mutually exclusive as multiple response tables.
well as exhaustive (i.e. observations can- II. Describing the data (descriptive or summary
not be beyond these categories). Contin- statistics):
gency tables are also known as cross- Descriptive statistics is used to describe the basic
tables, which present the frequency features of data and to provide simple sum-
distribution of two or more variables to maries about the sample and the measures

Table 8.2. ​Type of ventilation and daily bowel movements among patients
Mode of Ventilation
Bowel Movements Spontaneous Ventilation; f (%) Mechanical Ventilation; f (%) Total f x2 Value
Present 391 (64.0) 32 (29.4) 423 45.87a
Absent 220 (36.0) 77 (70.6) 297
Total 611 109 720 df 5 1
a
Significant (p , 0.05).
 Analysis of Data 191

Table 8.3. ​Factors contributing to sleep

or more variables with the help of paramet-
deprivation among patients (n 5 60) ric and nonparametric statistical tests. The
most commonly used inferential statistical
Factorsa f (%) tests are Z-test, t-test, analysis of variance
Blood sampling 35 (58.3) [ANOVA], chi-square tests. An inference is
Diagnostic tests 33 (55.0) a conclusion or judgement based on evi-
Medication 33 (55.0) dence. Statistical inferences are made cau-
tiously and with great care.
Vital signs monitoring 32 (53.3)
Inferential statistics is concerned with popula-
Noise 32 (53.3)
tions and uses sample data to make an infer-
Bright lights 30 (50.0) ence about the population or to test the hy-
a
Each patient has more than one factor. potheses considered at the beginning of the
research study. Inferential statistics helps the
used in a study. It is also used to describe the researcher to determine if the difference
main features of a collection of data in quan- found between two or more groups, such as
titative terms. Percentages, means of central an experimental and control group, is a real
tendency (mean, median and mode) and difference or only a chance difference that
means of dispersion (standard deviation, occurred because an unrepresentative sam-
range and mean deviation) are examples of ple was chosen from the population. Before
descriptive statistics. learning about the test of significance, let us
Descriptive statistics is used to organize and sum- understand a few of the important concepts
marize the data to draw meaningful interpreta- related to inferential statistics. Inferential
tions. It also allows the researcher to interpret statistics is explained in detail in Chapter 10.
the data meaningfully, so that research ques- Figure 8.1 illustrates the decision tree for the
tion can be answered completely and appropri- choice of inferential statistical tests.
ately. Descriptive statistics may be categorized - Type-I and type-II errors: Type-I error oc-
in several ways; however, this chapter presents curs where the null hypothesis is rejected,
the most simplified classification of descriptive when it should have been accepted. It is also
statistics that includes: called alpha error. Type-II error occurs
- Measures to condense data (frequency where the null hypothesis is accepted when
and percentage distribution through tabu- it should actually have been rejected. It is
lation and graphic presentations): For also known as beta error. These errors gen-
details, please refer to Chapter 9 (page no. erally occur due to unrepresentative sample
218–223). drawn from the population (Table 8.4).
- Measures of central tendency: For details, - Level of significance: The probability of
please refer to Chapter 9 (page no. 223– making type-I error is called level of signifi-
227). cance. It is represented by a or p. In other
- Measures of dispersion: For details, please refer words, level of significance is the probabil-
to Chapter 9 (page no. 227–228). ity of rejecting the null hypothesis when it
- Measures of relationship (correlation is true. In health sciences, we generally con-
coefficient): For details, please refer to sider the level of significance at either 1%
Chapter 10 (page no. 233–241). (0.01) or 5% (0.05). A significance level of
III. Drawing inferences from data (inferential 0.05 means that the researcher is willing to
statistics) take a risk of being wrong 5% of the times,
Inferential statistics helps in drawing infer- or 5 times out of 100, when rejecting the
ences from data, e.g. finding the differences, null hypothesis. If the decision needs to be
relationship and association between two much more accurate, such as deciding if a
192 Research Methodology and Biostatistics

No. of Type of No. of Type of Levels of Same or

Parametric
dependent dependent independent independent independent different
measures?
variables? variables? variables? variables? variables? participants?

Independent t-Test or
DIFFERENT YES
Point–Biserial Correlation

TWO NO Mann–Whitney Test

YES Dependent t-Test

SAME

Categorical NO Wilcoxon Matched-Pairs Test

One-Way Independent
DIFFERENT YES
ANOVA
More than
NO Kruskal–Wallis Test
TWO
ONE
One-Way Repeated
SAME YES
Measures ANOVA

NO Friedman’s ANOVA

Pearson Correlation or
Continuous YES
Regression

Spearman’s Correlation or
Continuous NO
Kendall’s Tau

Categorical DIFFERENT YES Independent Factorial

ANOVA/Multiple Regression

Factorial Repeated
SAME YES
Measures ANOVA
Two or BOTH YES Factorial Mixed ANOVA
More
ONE
Continuous YES Multiple Regression

Multiple Regression /
Both YES
ANCOVA

Pearson Chi-Square
ONE Categorical DIFFERENT
or Likelihood Ratio
Logistic Regression or
Categorical Continuous Biserial / Point-Biserial
Correlation
Two or
Categorical DIFFERENT Loglinear Analysis
More

Continuous Logistic Regression

Both DIFFERENT

Two or
Continuous ONE Categorical YES MANOVA
More
Two or
Categorical YES Factorial MANOVA
More

Both YES MANCOVA

Fig. 8.1. ​Inferential statistical decision tree.

 Analysis of Data 193

CI; however, medical literature has had a

Table 8.4. ​Explanation about type-I and type-II
errors comparatively larger use of the CI.
To be 95% confident about his or her
Decision estimation, a researcher should calcu-
Reality Accept H0 Reject H0 late as follows:
H0 (True) Correct decision Type-I error - First estimate the standard error of the
H0 (False) Type-II error Correct decision mean by using the following formula:
Notes: H0: Null hypothesis. 3 3
SE 5 5 5 0. 3
100 10
nursing intervention is effective or not, the
level of significance might be set at 0.01 or - Next insert 0.3 into the formula for
even at 0.001. With a 0.01 level of signifi- obtaining the 95% CI (lower limit
cance, the researcher stands the risk of be- and upper limit).
ing wrong 1 time out of 100 when rejecting
the null hypothesis. At the 0.001 level of Lower limit  X X  1.96 (SE )
significance, the risk is 1 time out of 1000.  130 1.96(0.3)  130  0..588  129.412
However, most nursing studies consider
0.05 level of significance to take a decision Upper limit  X X 1.96 (SE )
regarding accepting and rejecting the null
130 1.96(0.3)  130  0..588  130.588
hypotheses.
- Confidence interval: Confidence interval
Here the researcher can say that he or she
(CI) is a range of values that with a speci-
is 95% confident that the per capita in-
fied degree of probability is thought to
come of the household in Chandigarh is
contain the population value. The re-
between Rs 129.421 and 130.588.
searcher asserts with some degree of con- - If a researcher wants to be 99% confi-
fidence that the population parameter lies
dent that his or her estimates are cor-
within those boundaries. CI contains a
rect, he or she has to replace the 1.96 in
lower and an upper limit. For example, a
the formula with 2.58 as follows:
resource allocation management (RAM)
company conducted a study on 100 Lower limit  X X  2.58 (SE )
households to assess the per capita income  130  2.58(0.3)  130  0..774  129.226
of households at Chandigarh and found
that the mean per capita income is Rs 130
with a standard deviation of Rs 3. Experts Upper limit  X X  2.58 (SE )
may ask the researcher if a similar study is  130  2.58(0.3)  130  0..774  130.774
conducted on the sample population then
will it provide similar results. In this case, Here the researcher can say that he or
the researcher may calculate the CI with she is 99% confident that the per capita
an upper and lower level of per capita in- income of the household in Chandigarh
come so that he or she is 95% or 99% is between Rs 129.226 and 130.774.
confident with the limits that if the study - Degree of freedom: The interpretation of a
is repeated, there is 95% or 99% probabil- statistical test depends on the degree of
ity that per capita income will fall in these freedom. It is denoted by the abbreviation
limits. After the level of significance, it is df and a number (e.g. df 5 3). Degree of
one of the important parameters, but un- freedom indicates the number of values
fortunately in Indian and western nursing that are free to vary. Although degree of
literature, very little use has been made of freedom indicates the number of values
194 Research Methodology and Biostatistics

that can vary, the focus is actually on the - Strategies for effective interpretations:
number of values that are not free to vary. - Interpretations must be made in the light of
The procedure to calculate the degree of the research problem, objectives, conceptual
freedom differs from test to test. framework, hypotheses and assumptions.
- Test of significance: There are several - Critical examination of each element
parametric (t-test, Z-test, ANOVA) and of the study results before framing the
nonparametric tests (chi-square test, me- interpretations.
dian test, McNemar test, Mann–Whitney - Careful consideration and recognition of
test, Wilcoxon test, Fisher exact test) avail- the limitations of the research study so
able to establish statistical significance. that inappropriate interpretation can be
The decision tree depicted in Fig. 8.1 may avoided.
be used to decide about the type of infer- - Interpretations must be based on the
ential statistical test of significance that is study results only so that chances of mis-
required for analysis of particular data. interpretations or overinterpretations of
IV. Interpretation of data unstudied facts can be avoided.
Interpretation of data refers to the critical - Each part, aspect and segment of the
examination of the analysed study results to analysed result must receive close at-
draw inferences and conclusions. Interpre- tention, so that misrepresentation can
tation of the research findings of a study be avoided.
involves a search for their meaning in rela-
tion to the research problem, objectives,
conceptual framework and hypotheses. PARAMETRIC TESTS
This is an activity of critical thinking, which These tests are also known as normal distribution
is done carefully through brainstorming to statistical tests. The statistical methods of inference
infer the condensed and statistically com- make certain assumptions about the populations
puted data so that research questions can be from which the samples are drawn. Parametric tests
answered and hypothesis can be tested. are a type of inferential statistics tests that assume
Interpretation is one of the most essential tasks that data have come from a type of normal distri-
in a research process to frame the recommen- bution and makes inferences about the parameters
dations of a research study so that the empiri- of the distribution. For example, the assumptions
cal evidence generated through the research may be that the populations are normally distrib-
process can be used during the implementa- uted and have the same variance; population val-
tion phase of the research study. Interpreta- ues, as we have seen, are known as parameters; the
tion is a very subjective activity; it is not statistical techniques that make assumptions about
guarded with scientific methods and proce- the parameters are called parametric techniques.
dures. Therefore, it is liable to bias and errors. They are said to have more statistical power than
To present this adverse consequence, the re- the nonparametric tests. However, they may be
searcher must be very careful and critical misleading, if the idea behind the assumptions are
while examining the study results to draw the incorrect. The commonly used parametric tests are
interpretations (Fig. 8.2). presented in Table 8.5 along with a description

Analysed study results in Drawing

• Tables Careful and critical
comparative and
• Graphs examination of study
contrasts
• Statistical computations results
relationships

Fig. 8.2. ​Interpretation process.

 Analysis of Data 195

Table 8.5. ​Commonly used parametric tests

Commonly Used
S. No. Parametric Test Basic Assumptions/Considerations
1. Paired t-test It is used to compare two quantitative measurements taken from the
same group of individuals
Used when the following criteria are fulfilled:
Randomly selected homogeneous sample
Quantitative data (measured on an interval or ratio scale)
Variability normally distributed
Sample size less than 30
2. Unpaired t-test It is used to compare means between two distinct/independent
groups
Used when the following criteria are fulfilled
Randomly selected homogeneous sample
Quantitative data (measured on an interval or ratio scale)
Variability normally distributed
Sample size less than 30
3. Z-test It is used to compare the difference in population mean and a sample
mean or the difference between two independent sample means
Used when the following criteria are fulfilled:
The sample or samples must be randomly selected
The data must be quantitative in nature measured on an interval or
ratio scale
The variability is assumed to follow normal a distribution in the
population
The sample size must be larger than 30
Population variances are known
4. One-way ANOVA It is used to compare means between three or more distinct/indepen-
dent groups but may be used for more than two repeat measures of
the same group
Used when the following criteria are fulfilled:
The sample must be randomly selected
The data must be quantitative in nature measured on an interval or
ratio scale
The variability is assumed to follow normal distribution in the
population
5. Pearson coefficient It is used to estimate the degree of relationship/association between
of correlation two quantitative variables
Used when the following criteria are fulfilled:
The sample must be randomly selected
The data must be quantitative in nature measured on an interval or
ratio scale
The variability is assumed to follow normal distribution in the
population
196 Research Methodology and Biostatistics

about the basic assumptions and considerations analysis options making considerations about
to be kept in mind while using these tests. However, computation speed unnecessary.
in general, paramedic tests are employed effectively l Nonparametric procedures may be applied

in the following conditions: when the data being analysed consist merely of
l The sample has drawn from the study popula- ranking or classifications. That is, data may not
tion using probability sampling techniques. be based on a measurement scale strong enough
l The population is normally distributed from to allow the arithmetic operations necessary for
where the sample is drawn for study. carrying out parametric tests.
l The measurement of data is on an interval scale l However, the use of nonparametric tests with

or, may be sometimes, on time ratio scale. data that can be handled with a parametric test
l The population of the study has been clearly results in a waste of data, and application of
defined and complete information about popula- some nonparametric tests may be laborious for
tion is known. large samples.
Nonparametric tests may be effectively used in
the following situations:
NONPARAMETRIC TESTS l The sample is selected using either probability

Researchers in the field of health sciences many or even nonprobability sampling techniques.
times may not be aware about the nature of the l The population distribution is not known and

distribution or other required population param- may not be normally distributed.

eters. In addition, the sample may be too small to l The measurement of data is generally on nomi-

test the hypotheses and generalize the findings for nal or ordinal scales but may even be on interval
the population from which the sample is drawn. or ratio scales.
Furthermore, many times in the observations pre- l The population of the study has not been clearly

sented in numerical figures, the scale of measure- defined and complete information about the
ments may not be really numerical, such as grad- population is not known.
ing bedsores, or ranks given to analgesic drug’s Table 8.6 presents commonly used nonparametric
effectiveness in cancer pain management. In these tests along with the basic considerations and
situations, parametric tests may not be suitable, assumptions while using these tests.
and a researcher may need different types of tests
to draw inferences; those tests are known as non- COMPARING PARAMETRIC VERSUS
parametric tests. Nonparametric tests have the NONPARAMETRIC TESTS
following advantages: The following points briefly compares parametric
l They allow for the testing of hypotheses that are and nonparametric statistical tests. Table 8.7 also
not statements about population parameter val- presents a comparison of parametric and non-
ues. Some of the chi-square tests of goodness of parametric statistical tests.
fit and the test of independence are examples of l Parametric and nonparametric tests are the two

tests possessing this advantage. main classifications of statistical procedures.

l Nonparametric tests may be used when the dis- l Parametric tests are used only when it is

tribution of population parameters is unknown. assumed that there is normal distribution of

l Nonparametric tests tend to be computationally parameters in the population from where the
easier and consequently more quickly applied sample is drawn. However, nonparametric
than parametric procedures. This can be a de- tests can even be used when study parameters
sirable feature in certain cases but when time is are not normally distributed in the study
not at a premium, it merits a low priority as population.
criteria for choosing a nonparametric test. l When collecting data from a sample, which is

Indeed, most statistical software packages drawn from a population where those parameters
now include a wide variety of nonparametric are not usually distributed, the use of parametric
 Analysis of Data 197

Table 8.6. ​Commonly used nonparametric tests

Commonly Used
S. No. Nonparametric Test Basic Assumptions/Considerations
1. Chi-square test It is used to find out the association between two nominal or ordinal sets of
data/variables.
Used when the following criteria are fulfilled:
Preferably random sample but not necessarily
Distribution of population parameters is unknown or non-normally distributed
Qualitative data measured on nominal or ordinal scale
Sample size should be more than 30
Lowest expected frequency not less than 5
2. The sign test It is used as an alternative test to t-test where the median is compared rather
than mean.
Used to test null hypothesis about population median with single sample or
paired data
Used when population parameters are not known or not normally distributed
Used when the data available are on ordinal scale rather than interval or
ratio scale
The underlying assumption of the test is that the distribution of the variable
of interest is continuous
3. Median test It is used to test the null hypothesis that two independent samples have been
drawn from populations with equal median.
The following assumptions are considered:
The samples are selected independently and at random from respective
populations
The populations are of the same form and differ only in location
The variance of interest is continuous
The level of measurement must be at least ordinal
The samples do not have to be equal in size
4. Mann–Whitney The median test does not make full use of all the information measured on or-
test dinal scale; therefore, the Mann–Whitney test is used for better use of data.
The following assumptions are considered:
Median is used as a measure of central tendency to compare two samples
The two samples of size n and m, respectively, available for analysis have
been independently and randomly drawn from their respective population
The measurement scale is at least ordinal
The variable of interest is continuous
If population differs at all, then it does only with respect to their median
5. Wilcoxon signed If a small size sample (n < 30) is drawn from a grossly non-normally distributed pop-
rank test ulation and t-test and Z-test cannot be applied, then a best alternative nonpara-
metric test is the Wilcoxon signed rank test; because the sign test may be used
when data consists of a single sample or there is paired data. However, for analy-
sis of interval scale data, the sign test may be undesirable because it will make full
use of information in the data. Thus, the Wilcoxon sign rank test is most appropri-
ate as it tests the magnitude of differences between measurements and a
hypothesized location parameter rather than just the signs of the difference.
The following assumptions are considered:
The sample is random
The variable is continuous
The population is symmetrically distributed about its mean m
The measurement scale is at least interval
Continued
198 Research Methodology and Biostatistics

Table 8.6. ​Commonly used nonparametric tests—cont’d

Commonly Used
S. No. Nonparametric Test Basic Assumptions/Considerations
6. Spearman rank A nonparametric test used to estimate the degree of correlation between two
correlation variables measured on ordinal scales.
The following assumptions are considered:
Preferably random sample but not necessarily
Qualitative data measured on ordinal scale
Distribution of population parameters is unknown or non-normally
distributed

Table 8.7. ​Comparison of parametric and nonparametric statistical tests

Items Parametric Nonparametric
Assumed distribution Normal Any
Assumed variance Homogeneous Any
Typical data Ratio or interval Ordinal or nominal
Data set relationships Independent Any
Usual central measure Mean Median
Benefits Can draw more conclusions Simplicity; less affected by outliers
Choosing Tests Choosing Parametric Tests Choosing Nonparametric Tests
Correlation test Pearson coefficient of correlation Spearman coefficient of correlation
Independent measures, 2 groups Independent-measure t-test Mann–Whitney test
Independent measures, .2 One-way, independent-measures Kruskal–Wallis test
groups ANOVA
Repeated measures, 2 conditions Paired t-test Wilcoxon test
Repeated measures, .2 conditions One-way, repeated-measures Friedman test
ANOVA

tests for statistical analysis of such data leads to the

SOFTWARE FOR ANALYSIS
wrong outcome.
l Analogous nonparametric test may be used for OF QUANTITATIVE DATA
statistical analysis if assumptions of parametric Traditionally, health care biostatistics has been
tests are not satisfied. handling data manually; it involves laborious
l The parametric assumption of normality is work needing the scrolling of a lot of paper. The
considered doubtful, and it is better to utilize results often became available too late and were
the best possible nonparametric test for a small subjected to different types of errors. The advent
size sample (n , 30). of computers and their popularity in almost every
l If data are truly normally distributed, it is believed sphere of life have significantly changed the way
that nonparametric tests have less power for the biostatistics is now applied to health sciences.
same sample size than the corresponding para- Computer hardware and appropriate software
metric test. Moreover, it is assumed that interpre- have optimized the various processes involved in
tation of nonparametric procedures can be more the application of biostatistics by making the data
difficult than for parametric procedures. analysis and storage easier, faster and errorless.
 Analysis of Data 199

On the other hand, it has also improved the com- at Stanford University to help solve problems in
prehension of data through the use of sophisti- social sciences. SPSS now stands for Statistical
cated graphic presentation techniques. Product and Services Solutions and is among
However, the use of computer in statistics has the most comprehensive and popular statistical
its share of disadvantages as well. Just as a car can- packages. In addition to its provisioning for
not be expected to behave sensibly by itself under standard procedures needed in descriptive sta-
an incompetent driver, the usefulness of the re- tistics, it can also provide for regression analysis
sults from a computer depends critically upon the and ANOVA. It is also capable of multivariate
statistical knowledge and competence of the user. analysis involving sophisticated techniques,
Several computer packages are available for namely cluster analysis and time series analysis.
statistical analysis; among them commonly used (Fig. 8.3). The product is also capable of deter-
packages are Microsoft Excel, SPSS, Epi-info, SAS, mining sample size and estimating power of a
Minitab, Stata, Systat, NCSS, etc. statistical procedure. To know more about SPSS,
l Microsoft Excel: Microsoft Excel is a very popu- visit the website www.spss.com.
lar and useful spreadsheet program that can be l SAS: The Statistical Analysis System (SAS) is a
used for data entry. It has the capacity to gener- very comprehensive software developed by
ate random numbers, and it can be used for the North Carolina State University. This software is
computation of many standard statistical ap- divided into many modules and its licencing is
plications like computation of mean, range and flexible, based upon the need for functions. This
standard deviation. It can also be used for rela- system contains a very large variety of statistical
tively sophisticated work like computing regres- methods and is the software of choice of many
sion and ANOVA. However, statistical experts major businesses, including the entire pharma-
point out that there are problems in the algo- ceutical industry. SAS has also developed a PC
rithms of some versions of this program, as a SAS, which is compatible with the personal
result of which the results could be erroneous. computer and has a user-friendly windows in-
For example, in earlier versions, blank rows or terface. The more useful components of SAS are
columns were treated as zero instead of being Base SAS, SAS/STAT, SAS/GRAPH, etc. An ad-
ignored in the calculations. The pseudorandom vantage of SAS is its capability to transport data
number generators are also known to be faulty. files in various formats and convert them to SAS
It is, therefore, better to export Excel data files to data set without much effort. To learn more,
other packages, such as SPSS or SAS before do- visit the website www.asa.com.
ing statistical analysis. However, Excel may be l Minitab: This is another statistical package de-
used for generating graphs, such as bar charts, signed to facilitate the teaching of statistical
pie chart and scatter plots. methods by using the computer. The Minitab is
l SPSS: Originally known as Statistical Package also a very user-friendly product with well-
for Social Sciences, SPSS was developed in 1960 designed documentation facilities that are being

SPSS Window: Data view SPSS Window: variable view

Fig. 8.3. ​SPSS Windows showing data and variable views.

200 Research Methodology and Biostatistics

used widely in educational institutions. To learn following are the four processes that play a role in
more about Minitab, visit the website www. qualitative analysis:
minitab.com. l Comprehending: Early in the analysis process, a

qualitative researcher strives to make sense of the

data and learn ‘what is going on’ by reading the
ANALYSIS OF QUALITATIVE DATA voluminous information collected in qualitative
Qualitative analysis is a time-consuming, detail- research. Once he or she comprehends the topic
oriented and seemingly overwhelming task that at hand, the researcher is able to prepare a thor-
provides ways of discerning, examining, compar- ough and rich description of the phenomenon,
ing and interpreting meaningful patterns or which is completed only when a complete and
themes. It deals in words and is guided by fewer accurate report of it has been written.
universal rules and standardized procedures. In l Synthesizing: Synthesizing involves sifting of

contrast to quantitative analysis, numbers and the data, and putting pieces together. At this
what they stand for are the material of analysis. stage, the researcher gets a sense of what is typi-
Therefore, qualitative analysis is an intensive and cal and what is different about the phenome-
time-consuming activity that involves clustering non. At the end of the synthesis process, the re-
together of related types of narrative information searcher can make some generalized statements
into a coherent scheme. about the phenomenon and about the study
participants.
PROBLEMS IN QUALITATIVE DATA ANALYSIS l Theorizing: Theorizing involves a systematic

Qualitative data analysis is very challenging task sorting of the data. During the theorizing pro-
because of the following reasons: cess, the researcher develops alternative expla-
l There are no universally accepted rules and nations of the phenomenon under study and
methods for analysis of qualitative data. then analyses those explanations to determine
l There is lack of standard analysis methods and their fit with the data. The theorizing process
procedures for qualitative data. Therefore, va- continues to evolve until the best and most
lidity of analysed data is always a controversial relevant explanation is obtained.
issue. l Recontextualizing: The process of recontextu-

l Voluminous data in the form of several page alization involves the further development of
narrations collected during qualitative studies the theory such that its applicability to other
require a lot of hard work before being able to settings or groups is explored. In qualitative
draw meaningful conclusions. inquiries, the ultimate goal of which is theory
l Qualitative data cannot be condensed too much development, the theory must be recontextual-
like quantitative data because they may lose ized and generalized.
their richness of content and evidentiary value.
STEPS OF QUALITATIVE DATA ANALYSIS
QUALITATIVE DATA ANALYSIS PROCESS In qualitative data analysis, data are obtained
The analysis of qualitative data is carried out typi- through loosely structured interviews with
cally in active and inactive processes. Insights open-ended questions, focus group discussions
and theories cannot spring forth from the data (FGDs), observations, and projective and partici-
unless the researcher is completely familiar patory approaches. Irrespective of the purpose
with the whole data. It is noted that qualitative of the qualitative study and any of the afore-
analysis is a process of fitting data together or mentioned techniques of data collection, the
making the invisible obvious by linking and at- researcher ends up with a substantial number of
tributing consequences of antecedents. It is a pro- pages of written text that needs to be analysed.
cess of conjecture and verification, of correction Though procedures and outcome of the qualita-
and modification, of suggestion and defence. The tive data analysis differ from those of quantitative
 Analysis of Data 201

data analysis, the principles are not so different. In they were provided (adding the question-
both the cases, the researcher will have to carry naire number in order to avoid losing
out the following procedures: the connection with the informants’ other
l Describe the study sample characteristics. data).
l Order and reduce/code the data (data processing). ii. Next read the answers carefully, remem-
l Summarize the data in such a way that interpre- bering the purpose of the question. For
tation become easy, e.g. by preparing compila- example, the question ‘why are you smok-
tion sheet, descriptive narrates, flow charts, ing’ was supposed to help nursing stu-
diagrams or matrices. dents develop an intervention against
l Draw conclusions, relate these to the other data smoking.
sets of the study, and decide how to integrate the iii. Make rough categories of answers that
data in the report. seem to belong together and code them
l If required, develop strategies for further testing with respective keywords. For example,
or confirming the qualitative data in order to answer 3 (It gives me pleasure) and an-
prove their validity. swer 14 (I like to blow smoke rings) could
1. Description of study sample characteristics: A be labelled with the term pleasure, which
useful first step in data processing as well as could be abbreviated to the code ‘please’.
reporting the finding is a description of the iv. Again list all the answers but now per
study subjects. If numbers allow, relevant back- code, so that you get some five to seven
ground data may be tabulated, e.g. on age, short lists, for example:
gender, occupation, education or marital sta-
tus, as the practice in quantitative studies. Pleasure Being Sociable Giving Self-
However, as qualitative data originate from (please) (soc) Confidence (selfc)
small samples, more information is required to 2. I like the feel 10. All my 6. Because
place the data in their context. For example, of the ciga- friends I feel confi-
who were the key informants, what made the rette in my are dent and
researcher choose them, who took part in the hand smokers recharged
FGDs, how were the participants of the group 3. It gives me 11. It helps when
selected, and how representative are they for pleasure to make smoking
the study population? For observation: Under 5. I like to blow people 7. It helps me
the smoke more to think
what circumstances were they carried out? through my friendly better
Who were observed and by whom? Unless this mouth and and com- 18. It helps me
type of information is provided, interpretation nose fortable, to reduce
of data may appear haphazard. 14. I like to blow when the pressure
2. Ordering and coding of data: During ordering smoke rings offering a and tension
or coding of qualitative data, the researcher 15. I like the cigarette at work
handles two types of data: taste 17. It helps me
 Responses or answers acquired through
to relax
open-ended questions.
 More elaborate narratives from loosely structured v. Interpret each list, and end up with some
interviews or FGDs. five to seven meaningful categories, each
Ordering and coding for the answers acquired with a characteristic keyword, e.g. giving
through open-ended question involve the status, giving self-confidence, pleasure, be-
following steps: ing sociable, addiction and defiance. There
i. A first, basic step in the analysis of answers may be a discussion on the requirement
to open-ended questions is to list the an- of dividing some categories or merging
swers of a sample of 20–25 informants as others with few answers. For example,
202 Research Methodology and Biostatistics

answers 17 and 18 could be put in a differ- descriptive text. When interpreting the texts,
ent category, that is, reducing stress. In such we generally find out that no matter how
a case, there would be seven categories. The good our rules for the discussion were, the
defiance category may have two answers: I data hold useful information but also a num-
do not see why I must give up smoking! And ber of less-essential details. In addition, since
why not? The exclamation mark signifies informants may hop from one topic to the
defiance rather than lack of knowledge, other, the data are usually not presented in the
which forms the basis for the answer. These order we need for our analysis, so we have to
answers would have been difficult to encode order and reduce the data in order to make the
without this addition by the interviewer. analysis easier. Ordering is best done in rela-
According to the assumed willingness of tion to the discussion topics and objectives.
informants to change their behaviour, It is helpful to follow a number of steps
now a researcher can form a tentative in- systematically as follows:
terpretation. It might be most easy to give (a) Reread research objectives and discussion
up smoking for those who smoke to so- topics.
cialize or for pleasure. Those who feel they (b) Read the FGDs, interviews or narrative
derive status from smoking might form a observations carefully and analyse them.
middle category along with those who are Number the material as per the broad
addicted but are trying to quit, whereas discussion topic it concerns; by using a
for those who smoke to reduce stress and coloured marker, highlight specifically
heighten their self-confidence or who are illustrative remarks. The margins can be
very defiant at the question why they used to define the subtopics.
smoke, it might be very difficult to quit. For example, in a gender and leprosy study
vi. Now check if the labels work by trying a done in some countries, it seemed that
batch of 20–25 answers. It is possible that the discussion topic stigma had to
at this stage, some labels will still be be categorized as per different social
altered or new categories may be included settings in which it happened: among
or merged with others. spouses, in-laws, close relatives (parents–
vii. Make a final list of the labelled categories children) and community members. Fur-
and code all the data, including the data thermore, a differentiation had to be done
you have processed earlier with the ab- between self-stigmatization (e.g. a wife
breviated codes. diagnosed as a leprosy patient encourag-
Then discuss if you will adhere to your tempo- ing her husband to marry for a second
rary analysis of the data and what this means time to get rid of divorce, or a patient not
for the content of the messages to address being able to attend community meet-
various reasons for smoking. This content ings for the sheer fear of being ignored)
analysis is a very important use of the analy- and stigmatization by others. In stigma-
sis. The researcher will get insights so as to tization, different degrees of severity
how common the different reasons are by could also be differentiated that could
counting the answers under each label. vary from partial avoidance to total
Ordering and coding for the answers acquired expulsion. You would mark everything
through elaborated narratives involve the fol- related to stigma within the margin if
lowing steps: stigma would be a topic in your discus-
The data from interviews with FGDs or key in- sion list, and add keywords, such as
formants are by rule bulkier than answers spouse, in-laws, self-stigma, comm., in
to open-ended questions. The transcribed the margin, as well as keywords, like
tapes and field notes may comprise pages of divorce or sleep (ing) sep (arately), giving
 Analysis of Data 203

an indication of the gravity of the mothers, young mothers or male and female pa-
stigma. tients, are tabulated on different sheets. If the
(c) Make a list of all keywords belonging to a topics covered in those subgroups are not similar,
certain topic in the subcategories that they it is important to be orderly and follow roughly
have been developed under, e.g. everything the similar order of topics for each category of
belonging to stigma could be listed and sub- informants. The information entered is summa-
divided into the four major social settings in rized in key sentences and keywords, sufficiently
which stigma was found to manifest itself. clear to keep in mind the statements informants
(d) Make an interpretation of the data, e.g. made. (As the number of each study unit is in-
differentiate the major forms in which serted in the compilation sheet, it is always practi-
stigma manifests itself in these many so- cal to go back to the original data and provide the
cial settings, make an attempt to list a full statement, for example, in a research report or
ranking order of severity and link it with a presentation.)
other variables (such as socioeconomic Now an overview of all data per study popula-
status, degree of deformity) to interpret tion on one or more compilation sheet(s) is
differences in stigma. available. You have a list of answers of all
(e) Then in this way, code all your qualitative group members on a certain (sub) topic on
data. If required, your coding scheme can the columns. By reading horizontally, per
be adapted as you code, order and inter- informant, you can relate various topics with
pret more data. In this case, you should each other or to the informant’s personal
again read and probably recode the mate- features. Moreover, by comparing the com-
rial that has already been processed. pilation sheets, it is easier to compare the
Note: In order to adjust and deepen your inter- answers of different groups on certain issues.
view guides or topic lists, you may already For example, the personal data of leprosy patients
have analysed and coded your qualitative (recently declared cured) and a number of top-
data in the field. In that case, it may be prac- ics and subtopics discussed with them are pre-
tical to develop your final coding list in one sented in Table 8.8. Stigma actually experi-
cycle instead of two. However, you may also enced, which earlier was a topic, has been
refine your interpretation as you record your further subdivided into four major social set-
roughly coded, summarized data in compila- tings in which stigmatization may happen in
tion sheets without developing a very de- the compilation sheet: marriage, close blood
tailed coding system on your rough data. relatives, wider circle of spouse’s relatives and
3. Data summarization in compilation sheets: Af- community. Still finer distinctions can be made
ter ordering data, start their summarization. A in each of those (e.g. community can be friends,
helpful first step is to summarize all data of each workmates, neighbours, schoolmates or distant
study unit per study population on different community members). These may all be in-
compilation sheets. Compilation sheets for quali- serted under the heading ‘community’ as sam-
tative data comprise a number of columns with ples are small. Codes (italics) can be assigned to
the topics covered by the study as headings like the statements given in keywords, for example,
the master sheets for quantitative data. These may worry and fear under the heading ‘first reac-
be, in turn, subdivided into smaller themes that tion’. Out of the three examples presented, it
you recognized and coded when ordering the looks as if (confirmed by the analysis of all data
data (Table 8.8). Each FGD, interview or observa- in all four countries), in general, the stigma
tion is numbered and is entered successively in feared when patients get to know the diagnosis
that order on the relevant compilation sheet. The of leprosy is bigger than the stigma experienced
data for different categories of informants within in reality. Patient (12) in this respect is an
one study population, e.g. an older generation of exceptional case. In irony, the husband who
Table 8.8. ​Example of a compilation sheet (gender and leprosy)
Personal Data Symptoms Stigma in Reality Experienced (11)
Economical At Diagnosis Now Ec./Domestic Perception of
S. No. Sex Age Education Married Status (6) (8) First Reaction (10) Spouse Relatives In-Laws Community Act (12) Cure (25)
1 M 40 6 years Yes Farmer • Patches g Big fear • Remains supportive • Children sup- Not told, • Not told Hires No (still signs:
Shopkeeper • Painful None • Wife will run away • Helps more in shop portive; small hiding • Thinks labour drop foot)
nerves Still • Community will isolate • He decided to ab- ones not friends do (no force
• Drop foot him stain from sex (8 aware not know to farm) g
(2) • Fingers and toes will drop months) • Parents and • Behave as Income
off Self-stigma brother/sister before
• No longer able to work visit as before Hiding
and sustain family No stigma?
10 F 21 8 years Yes Husband Patches None Knew little; worried • Fiance inquired at HC Parents very First wife of All villagers came Does every- Yes (no signs)
farms (teacher Bad disease • If curable supportive husband to marriage thing
Father big saw and Fiancé will break off marriage • If she could get chil- told No stigma
farmer referred proceedings dren whole
her to HC) • Marriage postponed village
till patches subsided
• She now has child
Stigma reversed
12 F 60 – Yes, but Small Patches g Worried Husband kicked her out Son took her in Daughter- Avoided big Small trade to Yes, cured
now farming • Bad disease Divorce Supportive in-law meetings but earn bus (but
divorced + trade • Husband angry Supportive now OK fee for husband
Self-stigma treatment died)
reversed
 Analysis of Data 205

divorced her died from another disease by the of young mothers pertaining to the intro-
time she was declared cured from leprosy. duction of soft foods and those of mothers
Horizontal comparison of the data of patient above childbearing age. They then summa-
(1) shows that it is very unlikely that the man’s rized these answers in a matrix (Table 8.9).
friends do not have any idea that he has the This type of display made it easy for
disease, as even after he has been declared the researchers to reach the following
cured, he has some visible signs. In this case, the conclusion:
researchers had to interview the friends to as- - Younger mothers start providing soft
certain if in fact this man was (or had not been) foods, on average, 2.5 months earlier
stigmatized at all by the community. than the generation of their own
You may notice that interpretation of data and mothers.
labelling indeed becomes easy when using - Compared to women in the previous
compilation sheets, as a researcher can visu- generations, younger mothers utilize a
alize all aspects of his or her informants even larger variety of soft weaning foods.
if he or she looks at only one aspect at a time - For the same reasons as their mothers
for the whole study population. did, younger mothers also provide soft
The next step in summarizing may be the com- foods to their babies more often.
bining, contrasting or further analysis of im- Matrices considerably facilitate data analy-
portant topics through graphical displays, such sis. They have the very common form of
as matrices, diagrams, flow charts and tables. displaying qualitative data graphically.
4. Further summarizing of data in narrations, They can be utilized to compare and or-
matrices, figures and quasi-statistical tables: der information in a variety of ways, for
Further summarization of data may be achieved example, as per the following features:
through narrative analysis, matrices, figures - Time order (of methods being exam-
and quasi-statistical tables. ined in different periods, for example)
(a) Matrices: Matrices can be utilized for qual- - Different types of informants (as in
itative as well as quantitative data compari- the previous example)
son. In qualitative data, we may analyse - Data collection location (to visualize
different data sets or groups on significant discrepancies between urban and rural
variables that are presented in keywords. A populations)
matrix is a chart that appears like a cross- (b) Diagrams: A diagram is a figure with boxes
table. But it contains words as well as num- that contains arrows and variables, indicat-
bers. In an FGD on altering weaning prac- ing the relationships between these vari-
tices, the researchers enlisted the answers ables. While analysing the problems you

Table 8.9. ​Matrix on introduction of soft baby foods among mothers of different age groups
Age Groups Onset of Soft Food Type of Food Frequency of Soft Food/Day
Young mothers Range: 4–7 months • Soft porridge 2–4 times daily
(20–30 years) Average: 6 months • Soft porridge with pounded • Depends on availability
groundnuts of mother and caretaker
• Mashed potatoes, mashed • Depends on appetite of
fruits, soaked biscuits child
Mothers past Range: 5–11 months • Soft porridge 1–2 times daily
childbearing Average: 8.5 months • Soft fruit • Depends on availability of
age (.45) mother and caretaker
• Depends on appetite of child
206 Research Methodology and Biostatistics

wanted to examine during the protocols’ decisions or actions. The preceding figure
development, most groups create a diagram. (Fig. 8.5) is an example of a flow chart as it
Similarly, diagrams can be formed to sum- indicates the successive steps in protocol
marize findings of a study (Figs. 8.4and 8.5). development. Flow charts are especially help-
You might utilize a diagram to emphasize a ful in providing a summary of different flows
significant issue in your study by merging all of events that are mutually linked. For ex-
available quantitative and qualitative data ample, a counselling team in Bulawayo, Zim-
gathered. Diagrams, such as matrices, can be babwe interviewed some 95 HIV-positive
of great help in giving a holistic view of the persons in-depth for 2 years and provided a
data gathered and in guiding data analysis. summary in roughly 100 pages of interview
(c) Flow charts: Flow charts are specific kinds of material for every informant by drawing five
diagrams that mention the logical order of lines (Fig. 8.6). With crises and periods of

Twins Poor nutritional

status of mother

Cultivating Disease in
‘Insufficient milk’
season mother

EARLY
Job of Mother partly Diarrhoea ‘Bad quality
INTRODUCTION
mother absent in child of milk’
OF SOFT FOODS

Recommended by
Recommended
husband, mother,
by health staff
friends, etc.

Fig. 8.4. ​Reasons for early introduction of soft foods by young mothers.

Fields near Good nutritional

Plenty of milk
home status of mother

Convient for LATE Child refuses

Lack of time for
mother to INTRODUCTION porridge
extra cooking
breastfeed OF SOFT FOODS

Scarcity of land Lack of money

Last-born
and other to buy additional
child
resources foods
Lack of
information Mother, friends or
husband not in favour of
Scarcity of additional foods early introduction of soft
(except maize flour) foods

Fig. 8.5. ​Reasons for late introduction of soft foods by young mothers.
 Puts money Tries to Migrates to
Good income aside for child find new neighbouring Economic
job abroad country coping
(good job)

Hides news from

Own family family Tries to disclose to
strong links Girlfriend (pregnant brother but does Still no one Social
Wife left from other man) not dare. Decides knows coping
Girlfriend ± leaves him to abstain from sex
M, 36

Headmaster PS Genital Proves to Still genital Sores State of

sores be HIV+ sores improve symptoms
Divorced now;

girlfriend
Asks HIV Does test, Medical
test. Doctor finds Finds supportive care
refuses counselling doctor sought

Worried Very angry Joins healing

(reads about Calm at with ex-wife church (without Mood high Copes with Emotional
HIV) result and unfaithful disclosing HIV (miracle!) religion coping
girlfriend status)
0 3 6 9 12
months months months months
Time frame

Example of a relatively well-to-do man who copes in solitude despite supportive relatives because he is too ashamed to disclose his HIV+ status.

Fig. 8.6. ​Flow chart on coping situations of HIV-positive persons with their condition over time. Adapted from Meursing, K. (1997). A world of silence:

Analysis of Data
living with HIV in Matabeleland, Zimbabwe. Amsterdam: Royal Tropical Institute.

207
208 Research Methodology and Biostatistics

relative well-being, one central line gave the existing ward designs in public and private
development of the disease over time. An- hospitals. In this example, the collected
other line mentioned various forms of medi- data in narrative forms are analysed
cal care obtained; a third, the flaws in eco- through quasi-statistics as explained here.
nomic status linked to the disease (e.g. seeking In this, initially specific themes are identi-
employment elsewhere, loss of job); a fourth, fied from the narrations until saturation of
the probable alteration in social status, like themes is achieved. Later, the frequency of
(re) marriage or divorce; while a fifth line these identified themes is computed in a
rendered the patient’s emotional status con- table (Table 8.10).
nected to events happening in the four other (e) Narrative analysis: It is the study of
fields (e.g. depression, positive coping). These an individual’s speech. It is the most
flow charts were very helpful while compar- common approach of qualitative data
ing data per informant and between various analysis. In this approach, the stories of
groups of informants (e.g. single/married, respondents are narrated. The story is
males/females). They highlighted the effect of what a person shares about himself or
the disease on the lives of various groups of herself. A researcher always tries to com-
patients and their method of dealing with it. pare ideas presented by respondents
(d) Quasi-statistical tables: In this technique of about self. Narrative analysis could
qualitative data analysis, the researcher us- involve study of literature or diaries
ing a quasi-statistical style typically begins or folklore and autobiographies and
with some preconceived ideas about the comparsion between them.
analysis and uses those ideas to sort the data. For example, a study was conducted on
This approach is sometimes referred to as the long-term effect of political
manifest content analysis. In this, the re- violence: a narrative inquiry across a
searcher reviews the content of the narrative 20-year period, wherefore interview
data and searches for particular words or transcripts were read repeatedly to iden-
themes that have been specified in advance. tify common themes. They were subse-
The result of the search is information that quently categorized into tree and tree
can be analysed statistically; hence, the name nodes and attributes, and under these
quasi-statistic. For example, the analyst categories, narrations of different
can count the frequency of occurrence of subjects were depicted to provide the
specific themes or can cross-tabulate the evidence for the categories as presented
occurrence of certain words. in Table 8.11.
For example, a qualitative study was con- (f) Integrating quantitative and qualitative
ducted on the nurses’ perception about the data: So far we have gone through qualita-

Table 8.10. ​Nurses’ perception about existing ward designs

Themes f
Good visibility of beds from nursing station 1
Adequate open space in ward 12
Provision of privacy for patients 10
Closeness of supportive services 07
Appropriately located sanitary facilities 05
Adequate space for each bed 02
Notes: Computed from primary qualitative data.
 Analysis of Data 209

Table 8.11. ​Examples of narrative analysis

Theme: Individual factors in coping
Category 1: Gender differences in coping
Narrations: ‘The death of my father had a great effect on me as I was at that impressionable age of 11 or 12, so
I rebelled till I reached my late teens and it still really never goes away.’
‘Initially, it was like it occurred and that was it; we just moved on, and then as we grew older and started to real-
ize that he wasn’t there, it was hard to digest. I remembered you were to be alright for your mummy, to keep
her fine.’
Category 2: Long-term consequences as psychological problems, such as sleeping problems, anxiety, restless-
ness and suicidal ideas.
Narrations: ‘I find it difficult to get a good sleep. Often I wake during the nights because I think I heard a noise.’
‘After his death I went numb and could not even cry. After that I was hysterical. I don’t remember much for
about six months. I had to be drugged. A year later I took an overdose of tablets and ended up in a mental
hospital for a few months.’
Theme: Social factors in coping
Category 1: Traumatic events had detrimental financial and economic consequences
Narrations: ‘Sometimes I don’t really know how to manage. Money is always tight and it is difficult to make
ends meet.’
‘My husband was a businessman and he and I ran a business together. So, I had to close the business. I had a
13-year-old daughter, and I had to cope with bringing her up. I did not want to go on. I thought the world
came down around me, my business; I lost my business, I lost my husband. I just didn’t want to go on.’
Category 2: Traumatic events caused complete break-up of family and social relationship and consequently,
participants suffered intense loneliness.
Narrations: ‘Loneliness is really the worst of it all. I feel so alone since he was killed. He was there for me.’
‘I had one son at that time (of my husband’s death). He was nine years of age. He is forty now and I have not
seen him for seven years. I am also a grandmother. I do not see the child either. All these things I have lost for
the loss of my husband.’

tive data analysis as a different activity. But, it was revealed that they assumed adoles-
if a research team has collected quantitative cent girls and young women keep their
as well as qualitative data, which is the case skin patches hidden, due to shameful
in most health science research studies, it association with ugliness and dirt. This
would be wise to look at them in combina- gave the incentive for a successive break-
tion, as this can instigate more rewarding down of the quantitative data. It revealed
and deeper analysis. that the M/F difference in reporting was in
- When analysing the registration data of fact most pronounced in the 15–34 age
4500 new leprosy patients who had regis- group, and levelled off above 35. Therefore,
tered over the past 5 years, the Indonesian reason(s) for this comparatively vast gen-
‘gender and leprosy’ research team ob- der difference in the younger age groups
served that the M/F ratio was not favour- was then further identified.
able in the age group of 15–44 years. This 5. Drawing conclusions and verifying: Drawing
was not a smooth finding, as in Nepal, conclusions and verifying is the gist of data analy-
where women in this age group were found sis. It is not a separate activity either. When we
much better (though still less than men). start making a summary of our data in matrices,
After having in-depth interviews with staff, compilation sheets, flow charts or diagrams, we
210 Research Methodology and Biostatistics

persistently draw conclusions, and reject or mod- l Representativeness of data analysis: Although
ify quite a number of them as we advance. Writ- in qualitative research, informants have gener-
ing facilitates generating new ideas. Hence, writ- ally not been selected at random, they must
ing should be started as soon as practicable, right have been picked in a systematic manner, as per
from the beginning of data processing and analy- previously established rules. Ascertain if you
sis, even if it is only for ourselves. There should have in fact interviewed all categories of infor-
not be any loss to any kind of creative insights. mants required to get a total picture of your
Note: Qualitative data collection, processing, topic (not believing excessively on talkative au-
analysis and reporting are intricately inter- thorities). Be sure that you do not make a gen-
twined and are not (as is the case with quan- eralization from unrepresentative events.
titative data) different successive steps. It may l Biasness checking: Because of the influence of
often be important to get back to the actual the researcher on the research situation, observer
field notes and verify conclusions, gather ex- bias, the results can be drastically affected.
tra data if current data seem controversial and l Cross-checking data with evidence from other
ask feedback from all parties concerned. independent sources: These sources may in-
6. Reporting the data: In fact, there are two ways clude different research techniques to investi-
of reporting qualitative data that constitute part gate the same topic or results from other similar
of a study in which many research techniques kind of studies with some different independent
were used. One way is by summarizing the ma- informants. The data should not contradict
jor qualitative results in a different section of the each other; it should confirm. Active cross-
findings along with quotations and examples, checking of data, i.e. looking at independent
following the objectives that provided guidance evidence or counter-evidence, is one of the very
of the collection of this specific data. The results significant methods to heighten the authenticity
would then be elaborated in the ‘Discussion’, of research data. For example, responses of hus-
along with the results of other, more quantita- bands and wives (and other related informants)
tive data collection formats and would then be must conform with each other on issues like
reflected in the summary of the recommenda- who decides whether and what family planning
tions and findings. ways should be utilized, who checks whether
Another probability is to completely mix various daughters should have circumcision, or what
data sets in the chapter of findings, ordered as has altered in husband–wife relationships after
per the objectives of the entire study. If qualita- finding leprosy or another dreadful disease in
tive and quantitative data have been analysed one of the spouses.
and sometimes even gathered in an integrated l Comparing and contrasting data: Through as-
way, it would be logical to introduce them in an sociating various categories of informants,
integrated manner. Attention should be given comparison will often have been developed into
so that no valuable data are lost. So a rough the research design. If we need to confirm, for
draft of all significant findings is needed in any example, that variable A (high level of educa-
case, after which it can be considered to present tion) has influence on variable B (use of family-
the data either in different sections or chopped planning methods), we need to compare a
up for merging with other data. group of mothers having high education with a
group of mothers having low education on their
ESTABLISHING THE VALIDITY OF QUALITATIVE use of family-planning methods. Data compari-
DATA son and contrasting is significant if you are try-
Data generated after the qualitative study may ing to recognize your variables and also to con-
often generate questions concerning its validity. firm linkings among variables.
Therefore, to establish the validity of the qualita- l Using extreme (groups of) informants to the
tive data, the following strategies must be applied. maximum: We stated that it may be helpful to
 Analysis of Data 211

look for categories of informants that represent research. Not only is this important for ethical
the extremes on a certain variable in the discussion reasons or because it will ameliorate the chances
of study design and sampling. For example, you that the results will be implemented, but also
may check it to be most useful to study ‘dropouts’ because it will escalate the quality of your data,
and regular attendees of TB services, leaving out study design and the results obtained from these
the category of attendees who are not regular. This data. Additional information collected during
may be the most capable method of recognizing feedback sessions and suggestions will definitely
the important variables that have influence on the step up the quality of your research report.
attendance behaviour of TB patients.
l Undertaking additional research while testing SOFTWARE FOR ANALYSIS OF QUALITATIVE
the findings of your study: The outcomes of DATA
your study may be so intriguing that you plan to Strategies for analysing qualitative data by com-
do a follow-up study later. Such a study may be puter have been/are being developed with the
done for a variety of reasons: ever-increasing significance of computers in re-
 To have a replication of certain findings search. There are many practicalities, from normal
 To ignore (or identify) probable intervening word-processing programs to highly sophisticated
variables qualitative data management software, including
 To rule out competitors’ explanations by practicalities for statistical testing of associations.
investigating them As numbers are usually small in health sciences
 To look for any negative approach research, content analysis, which can be done
Additional studies carried out for one or manually, is most likely more significant than
more of these reasons may help in making testing of associations, which will not be elabo-
the outcomes of your original study more rated here. Contrarily, we can refer interested
convincing. students to anthropology or psychology depart-
l Information from your informants: A qualita- ments at universities that have got experience with
tive researcher needs to get involved with all programs like Qualitan or SPSS-17 for qualitative
parties concerned at different stages of the data processing.

MULTIPLE CHOICE QUESTIONS

1. What does it mean when you calculate a 95% CI? 3. The use of the laws of probability to make inferences
(a) The process you used will capture the true pa- and draw statistical conclusions about populations
rameter 95% of the time in the long run based on sample data is referred to as
(b) You can be ‘95% confident’ that your interval (a) Descriptive statistics
will include the population parameter (b) Inferential statistics
(c) You can be ‘5% confident’ that your interval will (c) Sample statistics
not include the population parameter (d) Population statistics
(d) All of the above statements are true
4. What is the median of the following set of scores: 18,
2. A type-II error is also known as a 6, 12, 10, 14?
(a) False positive (a) 10
(b) False negative (b) 14
(c) Double negative (c) 18
(d) Positive negative (d) 12
212 Research Methodology and Biostatistics

5. Which of the following is not a measure of variability? 8. Qualitative data analysis involves the categorization
(a) Median and indexing of texts or videos. The labels that are
(b) Variance given to sections or chunks of the text and video
(c) Standard deviation material are called
(d) Range (a) Data points
(b) Units of analysis
6. Mr Ram is conducting a study on patient satisfac- (c) Codes
tion with nursing care, where total satisfaction score (d) Memos
ranges between 0 and 100. If Mr Ram decides to
statistically compare the male and female patients 9. If data have significant variability, which of the fol-
on the level of satisfaction (highly satisfied, satisfied, lowing measures of central tendency is most ap-
dissatisfied and highly dissatisfied), he should use propriate to be considered?
which of the following tests? (a) The mean
(a) Parametric (b) The median
(b) Nonparametric (c) The mode
(c) Correlation coefficient (d) The range
(d) None of the above
10. When distribution of data in not known or cannot
7. Which of the following measure(s) of central ten- be considered as normal distribution, which of the
dency is/are appropriate for use with nominal data? following inferential test of significance is consid-
(a) The median only ered to compare the difference between sets of data?
(b) The mode only (a) t-test
(c) The median and the mean (b) Chi-square test
(d) The median and the mode (c) ANOVA
(d) Z-test

Answers of Multiple Choice Questions

1. d; 2. b; 3. b; 4. d; 5. a; 6. b; 7. b; 8. d; 9. b; 10. b.

REFERENCES
Altman, D. G. (1992). Practical statistics for medical research. London: Kirkwood, B. R. (1988). Essentials of medical statistics. London:
Chapman & Hall. Blackwell Scientific Publications.
Armitage, P. (1989). Statistical methods in medical research. London: Kish, L. (1987). Statistical designs for research. New York, NY: Wiley
Blackwell Scientific Publications. & Sons.
Ashish, B., Devendra, B. G., & Gaurisankar, R. (1977). Population Lancaster, H. O. (1974). Introduction to medical statistics. New
statistics in India. New Delhi: Vikas Publishing House. York, NY: Wiley & Sons.
Bahn, A. K. (1972). Basic medical statistics. New York, NY: Grune Mahajan, B. K., & Khankal, A. B. (2010). Methods in biostatistics
& Sratton. for medical students and research workers (7th ed.). New Delhi:
Bland, M. (1989). An introduction to medical statistics. London: Jaypee Brothers Medical Publishers.
ELBS. Mathew, D. E. (1988). Using and understanding medical statistics.
Bourke, G. J., & McGilvary, J. (1985). Interpretation and uses of New Delhi: Karger.
medical statistics. London: Blackwell Scientific Publishers. Miles, M. B., & Huberman, A. M. (1984). Qualitative data analysis, a
Daniel, W. W. (1987). Biostatistics: a foundation for analysis in the sourcebook of new methods. Beverley Hills, CA: Sage Publications.
health sciences. New York, NY: John Wiley & Sons. Negi, K. S. (2010). Biostatistics (2nd ed.). New Delhi: AITBS
Das, R., & Das, P. N. (2009). Instant medical biostatistics. New Publishers.
Delhi: Ane Books Pvt Ltd. Patton, M. Q. (1990). Qualitative evaluation and research methods
Dunn, O. J. (1984). Basic statistics. New York, NY: Wiley & Sons. (2nd ed.). Newbury Park, CA: Sage Publications.
 Analysis of Data 213

Rao, S., & Richard, J. (2006). Introduction to biostatistics and Walker, R. (Ed.). (1985). Applied qualitative research. Hants:
research methods (4th ed.). New Delhi: Prentice-Hall of India. Gower Publishing Company Ltd.
Reid, N. G. (1987). Research methods and statistics in healthcare. Williams, D. G., & Johnson, N. A. (1996). Essentials in qualitative
London: Edward Arnold. research: a notebook for the field. Hamilton: McMaster University.
Ress, D. G. (1989). Essentials of statistics. London: Champman and Woolson, R. F. (1987). Statistical methods for the analysis of
Hall. biomedical data. New York, NY: Wiley & Sons.
Spradly, J. P. (1979). The ethnographic interview. New York, NY: Yin, R. K. (1984). Case study research: Design and methods. Beverly
Holt, Rinehart & Winston. Hills, CA: Sage Publications.
Sprent, P. (1989). Applied nonparametric statistical methods. London:
Chapman and Hall.
 CHAPTER OUTLINE
Descriptive Statistics 9
INTRODUCTION CLASSIFICATION OF STATISTICS

MEANING OF STATISTICS AND BIOSTATISTICS FREQUENCY DISTRIBUTION

USES AND APPLICATIONS OF STATISTICS/BIOSTATISTICS MEASURES OF CENTRAL TENDENCY

TERMS RELATED TO BIOSTATISTICS MEASURES OF DISPERSION

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Organize, tabulate and present data meaningfully.
• Explain the basic concepts related to statistics and • Use descriptive statistics to predict results.
biostatistics.
• Describe the scope of statistics in medicine/health
sciences.

KEY TERMS

• Biostatistics  215 • Graphs  216 • Parameters  216

• Continuous data  215 • Inferential statistics  216 • Qualitative data  215
• Data  215 • Measures of central tendency  223 • Quantitative data  215
• Descriptive statistics  216 • Measures of dispersion  227 • Scattered diagram  222
• Discrete data 215 • Nominal  219 • Statistics  215
• Frequency distribution  216 • Ordinal  219 • Tabulation  216

national income, etc., that is expressed in num-

INTRODUCTION bers. At the outset, it might be useful to point out
The word statistics conveys a variety of meanings two different meanings attached to the term statis-
to people in different walks of life. To some, it tics. It has been used to indicate facts and figures
could be an imposing form of mathematics con- of any kind: health statistics, trade statistics, busi-
taining tables, charts and figures, which one can ness statistics and so on. It is also used to refer to
commonly find in newspapers, journals, books, a body of knowledge known as statistical methods,
reports, speeches, classroom lectures, etc. To oth- developed for handling data in general, particu-
ers, statistics can be the means to gather informa- larly in the field of experimentation and research.
tion about any group, activity, characteristic or Every day we are exposed to a wide array of nu-
a process whether it is production, population, merical information, which often has a profound
 Descriptive Statistics 215

impact on our lives. For example, there are 933 l It facilitates comparisons.
females per 1000 males. On average, for 400 stu- l It simplifies the masses of figures.
dents there is only one book written, published l It helps in formulating and testing hypothesis.
and printed in India every year. The density of l It helps in prediction.
population has gone up from 267 persons in 1991
APPLICATIONS
to 324 persons/km2 in 2001. The literacy rate has
gone up from 18.33% in 1951 to 64.84% in 2001. Applications of statistics in health sciences
Numbers play an essential role in statistics. They can be conceptualized into one of the following
provide the raw material for statistics. The study of categories:
l Defining normal and not normal in the context
statistics involves refining numerical and non-
numerical information into useful data. of various aspects related to health and illness
l Establishing the accuracy of diagnostic proce-

dures
MEANING OF STATISTICS AND l Planning of experiments and analysis of results
BIOSTATISTICS l Observations on the natural history of a disease,

The root of the word statistics comes from the Ital- namely its signs, symptoms, course and variation
ian word statista (meaning ‘statesman’) and the l Assessment of treatment protocol and different

German word statistik (meaning ‘political state’). interventions used for the care and treatment of
It was for the first time used by Gottfried Achen- patients
wall, a professor in Marlborough, in 1949 to refer l Collection, analysis and dissemination of vari-

to the subject matter as a whole. ous population health statistics

Statistics can be defined as numerical data in-
volving variability and treatment of such data.
According to Croxton and Cowden, ‘Statistics is TERMS RELATED TO BIOSTATISTICS
defined as collection, presentation, analysis, and l Data: The set of values recorded on one or more
interpretation of numerical data’. observational units or the factual information
According to Professor Horace Secrist, ‘Statis- collected during research studies are called data.
tics means aggregate of facts affected to a marked l Qualitative data: The variables that yield obser-
extent by multiplicity of causes, numerically ex- vations on which individuals can be categorized
pressed, enumerated or estimated according to according to certain characteristics or qualities
reasonable standards of accuracy, collected in sys- are referred to as qualitative variables or data,
tematic manner for a predetermined purpose and e.g. gender, occupation, marital status and edu-
placed in relation to each other’. cational level.
Therefore, precisely, statistics is the study of l Quantitative data: The variables that yield ob-
techniques and procedures for data collection, servations that can be measured are considered
classification, summarization, analysing and in- to be quantitative data, e.g. height, weight, blood
terpretation of the numerical data. pressure, serum cholesterol and body tempera-
Biostatistics is the method used in dealing with ture. Quantitative data are further divided into
statistics in the field of health sciences, such as discrete and continuous data.
biology, medicine, nursing, public health, den-  Discrete data: Data where are measured in
tistry, pharmacy and physical therapy. whole numbers are called discrete data, e.g.
number of children in a family, pulse rate,
USES AND APPLICATIONS OF ESR, blood sugar and blood pressure.
 Continuous data: Data which can be mea-
STATISTICS/BIOSTATISTICS
sured in fractional values, such as height,
USES weight, body temperature and chest circum-
l It presents facts in a definite form. ference, are called continuous data.
216 Research Methodology and Biostatistics

l Parameters: Characteristics of a population l Measures of dispersion

(e.g. average age of all the nurses in Punjab l Measures of relationship (correlation coefficient)
state) are called parameters.
 Parametric tests: A class of statistical tests

that involve assumption about the distribu- FREQUENCY DISTRIBUTION

tion of the variables and the estimation of a An appropriate presentation of data involves or-
parameter are known as parametric tests. ganization of data in such a manner that mean-
 Nonparametric tests: Tests that do not in- ingful conclusions and inferences can be drawn to
volve stringent assumptions about the distri- answer the research question. Unsorted and un-
bution of critical variables are called nonpara- grouped records do not allow us to draw clear
metric tests. conclusions. Quantitative data are generally con-
densed and frequency distribution is presented
through tables, charts, graphs and diagrams.
CLASSIFICATION OF STATISTICS
Statistics is broadly classified into two categories, TABLES
i.e. descriptive and inferential. Descriptive statistics A table presents data in a concise, systematic man-
deals with the enumeration, organization and ner from masses of statistical data. Tabulation is
graphical representation of data. An example of the first step before data are used for further sta-
descriptive statistics is the decennial census of In- tistical analysis and interpretation. Tabulation
dia, in which all the residents are requested to means a systematic presentation of information
provide information, such as age, gender, religion, contained in the data in rows and columns in ac-
marital status, education and occupation. The cordance with some common features and char-
data obtained in such a census can be compiled acteristics. Rows are horizontal and columns are
and arranged into tables and graphs that describe vertical arrangements.
characteristics of the population at a given time.
Inferential statistics provides procedures to draw General Principles of Tabulation
inferences about the conditions that exist in a A table can be simple or complex, depending upon
large set of observations, i.e. an entire population the number or measurement of a single set or mul-
from a study of a part of that set (sample). This tiple sets of items. Irrespective of it being simple or
branch of statistics is also known as sampling sta- complex, there are certain general principles which
tistics. An example of inferential statistics is to test must be kept in mind while designing a table. The
the efficacy of a new antihypertensive drug in basic principles that must be followed while present-
which the physician will have only a limited num- ing data in the form of a statistical table are as follows:
ber of hypertensive subjects on which efficacy is l A table should be precise, understandable and

tested and inferences are drawn. self-explanatory.

Descriptive statistics are used to organize and l Every table should have a title that is placed at

summarize data to draw meaningful interpreta- the top of the table. The title must describe the
tions. Descriptive statistics also allow the re- content clearly and precisely.
searcher to interpret the data meaningfully, so that l Items should be arranged alphabetically or ac-

research questions can be answered completely cording to size, importance or causal relation-
and appropriately. Descriptive statistics may be ship to facilitate comparison.
categorized in several ways; however, this chapter l Rows and columns to be compared with one

presents the most simplified classification of de- another should be brought together.
scriptive statistics that includes the following: l The content of the table, as a whole as well as

l Frequency distribution and graphical presenta- the items in each column and row, should be
tion (measures of condensation) defined clearly and fully.
l Measures of central tendency l The unit of measurement must be clearly stated.
 Descriptive Statistics 217

l Percentage can be given in parenthesis or can be Too many or too few groups or classes may fail
worked out to one decimal figure to draw the to reveal the salient features of the data. It is
reader’s attention to the fact that the figure is a also kept in mind that class or group intervals
percentage and not an absolute number. are kept constant. Table 9.1 provides an exam-
l Totals can be placed at the bottom of the ple of a frequency distribution table.
columns. 2. Contingency table: Tables that report the fre-
l Reference symbols can be directly placed be- quency distribution of two nominal variables
neath the table for any explanatory footnotes. simultaneously and that include the totals are
l Two or three small tables should be preferred to known as contingency tables. The categories con-
one large one. sidered should be mutually exclusive as well as
exhaustive (i.e. observations cannot be beyond
Parts of a Table these categories). Contingency tables are also
The parts of a table vary from problem to prob- known as cross tables, which present the fre-
lem, depending upon the nature of the data and quency distribution of two or more variables to
the purpose of investigation. A good statistical establish the relationship or association among
table must contain the following: them. These tables could be 2 3 2, 2 3 3 and 3
l Table number: It should be placed at the top of 3 3 depending on the number of variables on
the table.
l Title: Every table must have a suitable title. The

title should be brief and concise and describe TABLE 9.1. Sociodemographic profile of patients
the contents of the table. The content of the ta- (n 5 60)
ble should be self-explanatory.
l Head notes: The head note is given just below
Sociodemographic Variables f (%)
the title in a prominent type usually enclosed in Age (years)
brackets for further description of the contents 20–40 18 (30.0)
41–60 42 (70.0)
of the table.
l Captions and stubs: Captions are the headings Gender
Male 39 (65.0)
designated for vertical columns and stubs are
Female 21 (35.0)
the headings for horizontal rows.
l Body of a table: Arrangement of the data ac- Educational status
Illiterate 08 (13.3)
cording to the description given in the form of
Middle 17 (28.3)
captions and stubs form the body of the table. Senior secondary 20 (33.3)
l Footnotes: When some characteristics or items
Graduate and above 15 (25.0)
of the table are not adequately explained, foot- Marital status
notes are used to explain those items. Married 52 (86.7)
l Source note: Source note is used when second-
Unmarried 08 (13.3)
ary data are used to mention the source from Occupation
which the data for the table or the table itself are Housewife/nonworking 28 (46.7)
retrieved. Agriculture/labour 15 (25.0)
Service 09 (15.0)
Types of Tables Business 08 (13.3)
Basically tables are of four types: Habitat
1. Frequency distribution table: These tables Urban 37 (61.7)
present the frequency and percentage distribu- Rural 23 (38.3)
tion of the information collected, where an at- Religion 31 (51.7)
tribute is grouped into a number of classes, Sikh 29 (48.3)
which may vary between three and eight classes. Hindu
218 Research Methodology and Biostatistics

TABLE 9.2. Type of ventilation and daily bowel movements among patients

Bowel Mode of Ventilation Total x

Movements Spontaneous Ventilation, (%)f Mechanical Ventilation, (%) f Frequency, f Value2
Present 391 (64.0) 32 (29.4) 423
Absent 220 (36.0) 77 (70.6) 297 45.87a
df 5 1
Total 611 109 720
a
Significance: p , 0.05.

which the subjects are cross-classified. The miscellaneous when presentation of data cannot
number of subjects in a cell is called cell fre- be classified under the frequency distribution
quency. These tables are generally used in a table, contingency table or multiple-response
chi-square test. Table 9.2 provides an example of table.
a 2 3 2 contingency table.
3. Multiple-response table: When classification GRAPHICAL PRESENTATION OF DATA
of the cases is done into categories that are The main reasons for using diagrammatic and
neither exclusive nor exhaustive, then it is graphic representation of data are as follows:
called a multiple-response table. For example, a l It is the most convenient and appealing way in

patient can have two or more complaints, but which statistical results may be presented.
only the major ones may be listed. In such l It gives an overall view of the entire data.

cases, the sum total of frequencies would ex- l It is visually more attractive than other ways of

ceed the total number of subjects and may lead representing data.
to confusion. Therefore, the total number of l It is easier to understand and memorize data

subjects in case of multiple responses is given through graphical representation.

as a base and from this, percentages can be l It facilitates comparison of data relating to dif-

calculated. An example of a multiple-response ferent periods of time of different origins.

table is presented in Table 9.3.
4. Miscellaneous table: These tables are used to Constructing Diagrams/Graphs
present data other than frequency or percentage The following features must be essentially kept in
distributions, such as mean, median, mode, mind while constructing a diagram or graph:
range and standard deviation. A table is called l It should have a title and index.

l The proportion between width and height

should be balanced.
TABLE 9.3. Factors contributing to sleep
l The selection of scale must be appropriate.
deprivation among patients (n 5 60)
l Footnotes may be included wherever it is
Factorsa f (%) needed.
Blood sampling 35 (58.3) l Principle of simplicity must be kept in mind.

l Neatness and cleanliness in construction of

Diagnostic tests 33 (55.0)
Medication 33 (55.0) graph must be ensured.
Vital signs monitoring 32 (53.3)
Types of Diagrams and Graphs
Noise 32 (53.3) The commonly used diagrams and graphs in
Bright lights 30 (50.0) the presentation of data of research studies are
a
Each patient has more than one factor. bar diagram, pie diagram, histogram, frequency
 Descriptive Statistics 219

polygon, line graphs, cumulative frequency 60

curve, scattered diagram, pictogram and map dia- 60 Population
Land

Percentage of total world

population and land area
gram. 50
1. Bar diagram: It is a convenient graphical de-
vice that is particularly useful for displaying 40
nominal or ordinal data. It is an easy method 30
20 22 22
adopted for visual comparison of the magni- 20
tude of different frequencies. The length of the 20 14
8 9
bars drawn vertically or horizontally indicates 10
the frequency of a character. Bar charts are
called vertical bar charts (or column charts), if 0
Asia Africa Europe North
the bars are placed vertically. When the bars America
Regions
are placed horizontally, the charts are called
horizontal bar charts. There are three types of Fig. 9.2. ​Multiple bar diagram showing the percentage
bar diagrams: simple, multiple and proportion of world’s population and land area in different regions
bar diagrams (Figs. 9.1–9.3). Some of the of the world.
points to be kept in mind while making a bar
diagram are as follows: researcher must remember that only percentage
 The width of bars should be uniform through- data must be used to prepare pie diagrams. It
out the diagram. gives comparative differences at a glance. The
 The gap between the bars should be uniform size of each angle is calculated by multiplying
throughout. class percentages with 360° or by using the fol-
 Bars may be vertical or horizontal. lowing formula:
2. Pie diagram/sector diagram: It is another use- Class frequency
ful pictorial device for presenting discrete data  360°
Total observation
of qualitative characteristics, such as age groups,
genders and occupational groups in a popula- 3. Histogram: It is the most commonly used
tion (Fig. 9.4). The total area of the circle repre- graphical representation of a grouped fre-
sents the entire data under consideration. The quency distribution. Variable characters of dif-
ferent groups are indicated on the horizontal
line (x-axis) and frequencies (number of ob-
servation) are indicated on the vertical line
(y-axis). The frequency of each group forms a
80 column or rectangle. Such a diagram is called a
70 histogram. The area of the rectangle is propor-
60 tional to the frequency of the corresponding
No. of people

50
class interval and the total area of the histo-
40 72 gram is proportional to the total frequency of
all the class intervals. A histogram may be
30
drawn keeping the following in mind:
20 27  The set of vertical bars have areas propor-
10
tional to the frequencies represented.
0  The difference between a histogram and a bar
Vegetarian Nonvegetarian
diagram is that a bar diagram is one dimen-
Types of dietary pattern
sional and only the length of the bar has sig-
Fig. 9.1. ​Simple bar diagram showing the dietary nificance, while in a histogram, both length
pattern of people. and width matter.
220 Research Methodology and Biostatistics

100% North America

Percentage of population
Europe
80%
Africa
Asia
and land

60%

40%

20%

0%
Population Land

Fig. 9.3. ​A proportionate bar graph showing the percentage of world’s population and land area in different regions
of the world.

70
11% Frequency (no. of males)
10% 26% 60

50
40

30
17%
34% 2% 20
Basketball Computer games 10
Soccer Listening to music 0
Reading 15 20 25 30 35 40
Netball
Age in years
Fig. 9.4. ​Pie diagram showing cultural and leisure Fig. 9.5. ​Histogram depicting number of males in dif-
activities in which urban children participate. ferent age groups.

 When class intervals are equal, the frequencies vertical axis. From the data shown earlier, a
are taken on the y-axis and the variables on the histogram is drawn as in Fig. 9.5.
x-axis, and adjacent rectangles are constructed.
 When the class intervals are unequal, a cor-
4. Frequency polygon: It is the curve obtained
rection for unequal class intervals must be by joining the middle top points of the rect-
made. angles in a histogram with straight lines. It
The following frequency distribution can be rep- gives a polygon, i.e. a figure with many an-
resented graphically in the form of a histogram. gles. In this, the two end points of the line
drawn are joined to the horizontal axis at
Age Group the midpoint of the empty class intervals
(Years) 15–20 20–25 25–30 30–35 35–40 at both ends of the frequency distribution.
No. of Frequency polygons are simple and sketch
Males 15 20 40 60 50 an outline of the data pattern more clearly
than histograms. On the same axis, one can
Here, we will take class boundaries along the plot frequency polygons of several distribu-
horizontal axis and frequencies along the tions, thereby making comparisons possible.
 Descriptive Statistics 221

70

60

Frequency (no. of males)

50

40

30

20

10

0
15 20 25 30 35 40
Age in years
Fig. 9.6. ​Frequency polygon depicting number of males in different age groups.

A frequency polygon can be drawn using the Close the polygon at both ends of the distri-


following steps: bution by extending them to the baseline.

 Hypothetical classes at each end would have
 Draw the histogram of the given data.

 Join the midpoints of the upper horizontal to be included with a frequency of zero.
sides of each rectangle with the adjacent one A frequency polygon can be drawn with the follow-
by a straight line. ing example (Fig. 9.6).

Age Group (Years) 15–20 20–25 25–30 30–35 35–40

No. of Males 15 20 40 60 50

5. Line graphs: In this, variables in the fre- The vertical axis may not start from zero, but
quency polygon are depict by lines. It is at some point from where the frequency
mostly used where data are collected over a starts. The given data may be plotted with
long period of time. On the x-axis, values of reference to x- and y-axes and these consecu-
independent variables are taken and values of tive points or data are then joined by straight
dependent variables are taken on the y-axis. lines (Fig. 9.7).

600
No. of cars sold (in thousands)

500
456 486
402 417
400 387 347 337
328 In Delhi
300 298 342 307
298 In Mumbai
200 203

123
100

0
2001 2002 2003 2004 2005 2006 2007
Years
Fig. 9.7. ​Line graph representing the number of cars sold in Delhi and Mumbai during 2001–2007.
222 Research Methodology and Biostatistics

Years 2001 2002 2003 2004 2005 2006 2007

Sale of Cars in Delhi (in Thousands) 123 203 328 298 337 417 486
Sale of Cars in Mumbai (in Thousands) 456 402 387 347 342 307 298

6. Cumulative frequency curve or ogive: This graph frequencies are then plotted corresponding to the
represents the data of a cumulative frequency upper limits of the classes. The points correspond-
distribution. For drawing an ogive, an ordinary ing to the cumulative frequency at each upper
frequency distribution table is converted into limit of the classes are joined by a free-hand curve.
acumulative frequency table. The cumulative The diagram made is called ogive (Fig. 9.8).

Age Group (Years) 15–20 20–25 25–30 30–35 35–40

No. of Males 15 20 40 60 50
Cumulative Frequency 15 35 75 135 185

7. Scattered or dotted diagrams: It is a graphic Therefore, it is also called a correlation diagram.

presentation that shows the nature of correla- An example of a scattered or dotted diagram is
tion between two variable characters x and y presented in Fig. 9.9.
regarding similar features or characteristics, 8. Pictograms or picture diagram: This method
e.g. height and weight in men of 20 years old. is used to impress the frequency of the
200
180 185
160
Cumulative frequency

140
135
120
100
80
75
60
40
35
20
15
0
15–20 20–25 25–30 30–35 35–40
Age in years
Fig. 9.8. ​Cumulative frequency curve representing number of males in different age groups.
Performance level
Weight

Height Time in day

(A) (B)
Fig. 9.9. ​Scattered diagram showing (A) positive correlation and (B) negative correlation.
 Descriptive Statistics 223

High economic values around a central value is known as central

class tendency. The central value around which there is
a concentration is called measure of central ten-
Middle economic
dency. By calculating the measure of central ten-
class
dency, we can find a single value to represent the
complete data. It also helps us to compare the
Below poverty
line value of two or more groups.
Developing countries Developed countries
Fig. 9.10. ​Pictogram showing the proportion of people Definitions of Measure of Central Tendency
of respective economic classes. According to Ya-Lun-Chou, ‘the average is some-
times described as a typical value in the sense that
occurrence of events to common people, such it is sometimes employed to represent the entire
as attacks, deaths, number of operations, individual in a series of a variable’.
admissions, accidents and discharges in a pop- According to Croxton and Cowden, ‘An average
ulation (Fig. 9.10). value is a single value within the range of the data
9. Map diagram or spot map: These maps are that is used to represent all of the value of series.
prepared to show geographical distribution of Since an average is somewhere within the range of
frequencies of characteristics (Fig. 9.11). data. It is also called measure of central value’.

Limitations of a Graph Objectives of Central Value

l It is confusing (may be false or true). l To get a single value that describes the charac-
l It presents only quantitative aspect. teristics of the entire group
l It gets information only on one aspect or on l To facilitate comparisons
limited characteristics.
l It can present only approximate values. Requisites of Good Averaging
l Easy to understand
l Simple to compute
MEASURES OF CENTRAL TENDENCY l Based on all the items
A researcher collects the values of any variable l Not unduly affected by the extreme observations
under study, like Hb% from a large population, l Rigidly defined
and the observation values cluster around a cen- l Capable of further algebraic treatment
tral value. This feature of concentration of the l Sampling stabilities

Fig. 9.11. ​Map showing distribution of geriatric population.

224 Research Methodology and Biostatistics

Measures of Central Tendency Age in Years (x) No. of Students (f) xf

The important measures of central tendency are 16 35 560
as follows: 17 31 527
l Arithmetic mean

l Median
18 20 360
l Mode 19 14 266
l Geometric mean of 5 100 oxf 5 1713
l Harmonic mean

xf 1713
Arithmetic Mean X   17.13
l The mean or average is probably the most com- f 100
monly used method of describing central
Therefore, mean age of the adolescents in this
tendency.
group is 17.13 years.
l Mean is computed by dividing the sum of all the l Calculating arithmetic mean from a continu-
values by the total number of values. Arithmetic
ous frequency table: In case of a continuous
mean is represented by X
frequency table, mean can be calculated by the
following formula:
Sum of the values ( x )
X
Number of the values (n ) xf
X
f
For example, the haemoglobin levels of 10
women are given here, i.e. 12.5, 13, 10, 11.5, where x 5 midpoint of class interval and
11, 14, 9, 7.5, 10 and 12. The mean haemoglo- f 5 corresponding frequency.
bin level of this sample of women will be Midpoint of the class interval is calculated by
calculated as follows: the following formula:
Sum of the values ( x )
X  Lower limit  Upper limit
Number of the values (n) Midpoint 
12.5  13  10  11.5  11 
2
14  9  7.5  10  12 110.5
  For example, calculate the mean age of the
10 10
following group of people.
X  11.5

Class Interval No. of People Midpoint

The mean haemoglobin of the study sample is of Age (f) (x) xf
11.05.
15–20 15 17.5 262.5
l Calculating arithmetic mean from a discrete
frequency table: In a discrete frequency table, the 20–25 20 22.5 450
mean is calculated using the following formula: 25–30 40 27.5 1100
30–35 60 32.5 1950
xf of 5 135 oxf 5 3762.5
X
f
xf 3762.5
X   27.87
where x 5 corresponding value variable and f 135
f 5 frequency.
For example, the following data give the age of Therefore, mean age of the people in this group
100 adolescent girls. Find the mean age. will be 27.87 years.
 Descriptive Statistics 225

Merits of arithmetic mean l Use the formula:

l It is a simple average to understand and easiest
 n  1
to compute. M 
 2 
l It is affected by the value of each item in a series. 71 8
l It is defined by a fixed mathematical formula M   4th observation or item
2 2
with the result that irrespective of who com-
putes it, the average gets the same answer. l Therefore, the fourth observation in the given
l It can be treated algebraically. data is the median. Thus, 158, will be the me-
l It is a reliable method of calculating the average. dian.
l It calculates a value that is not based on the po- Calculation of median from the discrete fre-
sition of the series. quency table.
Example 2: Calculate the median for the data
Demerits of arithmetic mean in the following frequency table.
l Very small and very large items usually affect
the value of the average. Income/day (x) 100 150 200 250 300 350
l In a distribution with open-ended classes, values Number of
of mean cannot be computed without making Households (f) 05 19 03 11 06 09
assumptions. Cumulative
l It is not always a good measure of central tendency. Frequency (cf) 5 24 27 38 44 53

Median Solution:
Median of a set of values is the middle-most value l In case of the discrete frequency table, one has
when the data are arranged in ascending order of to calculate the cumulative frequency first.
magnitude. The middle value will divide the num- l Then consider the final cumulative frequency as
ber of observations in the data into two equal n, and calculate the median by the following
parts. The median is denoted by M. It is also called formula:
positional average.
l It refers to the middle value of distribution.  n 1 
M 
l One-half of items in the distribution have a  2 
value larger than the median value. 53 1
M
l One-half of items in the distribution have a 2
value smaller than the median value. 54
  27th observation or item
To calculate the median from discrete data, we use 2
the following formula l The 27th observation in the array means the
 n 1 
th 27th cumulative frequency and the x value in
M  observation or item
 2  front of the 27th cumulative frequency is 200.
Therefore, median value will be 200.
Calculation of median from discrete data l In case the exact calculated array value does not

Example 1: The following daata give the weight of match the cumulative frequency in the table, the
the seven persons in pounds: 158, 167, 143, 169, next higher cumulative frequency to the calcu-
172, 146 and 151. Calculate the median of the data lated value is considered the observation for the
given. median.
Solution: Calculation of median from the continuous
l First arrange the data in ascending or descend- frequency table.
ing order of magnitude: 143, 146, 151, 158, 167, Example 3: Calculate the median for the fol-
169, 172 lowing frequency table data.
226 Research Methodology and Biostatistics

Income/day (x) 100–150 150–200 200–250 250–300 300–350 350–400

Number of Households (f) 05 19 03 11 06 09
Cumulative Frequency (cf) 5 24 27 38 44 53

Solution:
the value at the point around which the items tends
In case of a continuous frequency table, median
to be most heavily concentrated. It may also be re-
can be calculated by using the following formula:
garded as the most typical of a series of values. It is
(N /2  m)w denoted by Z. If a set of values has two items that
M l  have the highest frequency, then distribution has two
f
modal values and it is called a bimodal distribution.
where l Calculation of mode for discrete frequency:
l 5 the lowest limit 5 200 For the raw data and discrete frequency table
w 5 the width 5 50 (Table 9.4), the most frequently occurring value
f 5 the frequency of the median class 5 03 is considered as mode.
m 5 the cumulative frequency of a class just For example, data about the amount of the
before the median class 5 24 sugar purchased by the people from a shop: 3,
N 5 o f 5 53 1, 7, 4, 1, 2, 5, 3, 4, 6, 5, 5, 4, 4, 3, 5, 2, 4
(26.5  24)50 In this series, mode (Z) 5 4
M  200  Z 5 value which has the highest frequency, i.e.
03
19 in the values given earlier, and the corre-
2.5  50 125
 200   200  sponding x value is 150, which is mode in the
033 03 given data. Therefore, mode of data in Table
 200  41.67  241.7 9.4 is 150.
l Calculation of mode from continuous fre-
Therefore, the median is 241.7. The value of
quency data: From a continuous frequency
median always lies within the median class.
table, mode can be calculated by:
Merits of median ( f  f1 )w
l It is useful in case of open-ended and unequal Z l
classes. 2 f  f1  f 2
l Extreme values do not affect the median.
l It is the most appropriate average in dealing where
with qualitative data. l 5 lowest limit of the modal class
l The value of median can be determined f 5 frequency of the modal class
graphically. w 5 width of the modal class
f1 5 frequency of class just before the modal
Demerits of median class
l For calculating the median, it is necessary to f2 5 frequency of class just after the modal class;
arrange the data. modal class is the one which has the highest
l Since it is the positional average, the value is not frequency
determined by all the observations.
l It is not capable of further algebraic treatment. TABLE 9.4. Identify the mode in the following
l Median is not calculated for quantitative data. discrete frequency table

Income/day (x) 100 150 200 250 300 350

Mode
It is the value which has the highest frequency in the Number of 05 19 03 11 06 09
households (f)
data. In other words, the mode of the distribution is
 Descriptive Statistics 227

For example, calculate the mode for the following

continuous frequency data:

Income/day (x) 100–150 150–200 200–250 250–300 300–350 350–400

Number of Households (f) 05 19 03 11 06 09

( f  f1 )w
Z l Range
2 f  f1  f 2
It is the difference between the highest and the
where l  150, f  19, w  50, f1  5, f 2  3,
lowest value in the data. If H is the highest and
(19  5)50 14  50 L is the lowest value, then
Z  150  150 
2  19  5  3 38  8 Range (R) 5 H – L
(700) For example, calculate the range for the
Z  150  150  23.33  173.33
30 following data: 3, 5, 7, 8, 9, 12, 15, 17, 19, 20, 21,
23, 26.
Hence, the mode for the tabulated data is 173.33. Range (R) 5 26 – 3 5 23
The value of mode always lies within the modal Therefore, range of these data is 23.
class.
Merits of Range
Merits of mode l Range is very simple to understand.
l It is not affected by extreme values. l It is also easy to calculate.
l It can be used to describe qualitative phenomenon.
l Values of mode can be determined graphically. Demerits of Range
l It is not suitable for deep analysis.
Demerits of mode l It is not suitable in case of extreme values.
l The value of mode cannot always be determined.
l It is not capable of algebraic manipulation. Standard Deviation
l It is not based on each value.
It is the most prominently used measure of dis-
persion. It is denoted by SD or s.
MEASURES OF DISPERSION
Definition of Standard Deviation
The observation deviating from the central value
is different in different sets of values of character. Standard deviation is the positive square root of
In some distributions, the difference may be less, the mean of the squared deviations of values from
whereas in some other, it may be more. This prop- the arithmetic mean.
erty of deviation of the values from the average is The formula for the calculation of the SD from
called variation or dispersion. The degree of varia- discrete data:
tion is indicated by the measure of dispersion. The
( x  x )
2
various measures of dispersion are as follows: SD 
l Range n
l Mean deviation

l Standard deviation The formula for the calculation of the SD from

l Quartile deviation continuous data:
Among these measures of deviation, range and
 ( xx ) f
2
standard deviation are the most commonly used SD 
measures of dispersion. n
228 Research Methodology and Biostatistics

Calculation of the SD from Discrete Data l Compute the SD by using the following formula:
Example: Blood cholesterol levels of 10 persons
are 240, 260, 290, 245, 255, 288, 272, 263, 277 and
( x  x )
2
257. Calculate the standard deviation of the given SD 
discrete data. n
Solution: where ( x  x )  2564.1
2

l Calculate the mean of the data (i.e. x 5 264.7

in this example). and n (number of observation)  10

l Compute the o(x – x ).
2564.1
SD   256.41
10
x (x 2 x) ( x 2 x )2 SD  16.01
240 240 – 264.7 5 –24.7 610.09
Calculation of the SD from Continuous Frequency
245 245 – 264.7 5   19.7 388.09 Data
255 255 – 264.7 5 –9.7 94.09 For example, the following data depict the weight of
257 257 – 264.7 5 –7.7 59.29 adolescent girls. Calculate the standard deviation for
260 260 – 264.7 5 –4.7 22.09 the following continuous frequency data.
263 263 – 264.75 –1.7 2.89
Weight 60–64 64–68 68–71 71–74
272 272 – 264.7 5 7.3 53.29
No. of Subjects 10 09 07 02
277 277 – 264.7 5 12.3 151.29
288 288 – 264.7 5 23.3 542.89
l Calculate the SD by using the following formula:
290 290 – 264.7 5 25.3 640.09
( x  x ) f
2

o (x2x) 5 2 SD 
ox 5 2647 x 5264.7 n
2564.1

Weight (kg) f Midpoint x xf (x 2 x) ( x 2 x )2 ( x 2 x )2 3 f

60–64 10 62 620 –4.14 17.14 171.40
64–68 09 66 594 00 00 00
68–72 07 70 490 3.86 14.90 104.3
72–76 02 74 148 7.86 61.78 123.56

of 5 28 oxf 5 1852 x 5 66.14 2

o ( x 2 x ) 3 f 5 399.26

xf 1852
x    66.14
f 28

( x  x ) f
2
where (x  x )2 f  399.26
SD  n( f )  28
n

n(f )  28
399.26
SD   14.26
28
SD  3.78
 Descriptive Statistics 229

MULTIPLE CHOICE QUESTIONS

1. The goal of ___________ is to focus on summariz- 6. Which of the following is not a measure of vari-
ing and explaining a specific set of data. ability?
(a) Inferential statistics (a) Median
(b) Descriptive statistics (b) Variance
(c) None of the above (c) Standard deviation
(d) All of the above (d) Range

2. The most frequently occurring number in a set of 7. Which of the following is the formula for range?
(a) H 1 L
values is called the ___________.
(b) L 3 H
(a) Mean
(c) L 2 H
(b) Median
(d) H 2 L
(c) Mode
(d) Range
8. Which of the following is an example of nominal
data? (one correct choice)
3. What is the median of the following set of scores: 18,
(a) Number of people on a course
6, 12, 10, 14?
(b) Cancer staging scale
(a) 10
(c) Types of employment among rural dwellers
(b) 14
(d) Heart rate
(c) 18
(d) 12
9. Calculate the approximate mean and standard
deviation (SD) for the following data: 25, 7, 22, 33,
4. ___________ are used when you want to visually
18, 15.
examine the relationship between two quantitative
(a) 20 and 8.2, respectively
variables.
(b) 20 and 8.9, respectively
(a) Bar graphs
(c) 120 and 79.2, respectively
(b) Pie graphs
(d) 120 and 8.9, respectively
(c) Line graphs
(d) Scatter plots
10. As a general rule, the ___________ is the best mea-
sure of central tendency because it is more precise.
5. The ___________ is often the preferred measure of
(a) Mean
central tendency if the data are severely skewed.
(b) Median
(a) Mean
(c) Mode
(b) Median
(d) Range
(c) Mode
(d) Range

Answers of Multiple Choice Questions

1. b; 2. c; 3. a; 4. d; 5. b; 6. a; 7. d; 8. c; 9. a; 10. a.

REFERENCES
Altman, D. G. (1992). Practical statistics for medical research. London: Bland, M. (1989). An introduction to medical statistics. London:
Chapman & Hall. ELBS.
Armitage, P. (1989). Statistical methods in medical research. London: Bose, A., Gupta, D. B., & Raychaudhuri, G. (1977). Population
Blackwell Scientific. statistics in India. New Delhi: Vikas.
Bahn, A. K. (1972). Basic medical statistics. New York: Grune & Bourke, G. J., McGilvray, J., & Daly, L. E. (1985). Interpretation and
Stratton. uses of medical statistics. Oxford: Blackwell Scientific.
230 Research Methodology and Biostatistics

Chou, Ya-Lun. (1975). Analysis: with business and economics Mathew, D. E. (1988). Using and understanding medical statistics.
applications (qualitative methods). Canada: Holt, Rinehart and New Delhi: Karger.
Winston Ltd. Muzumdar R. D., Kulkarni A.P., & Baride J.P. (2003). Manual of
Croxton, F.E., & Cowden, D.J. (1956). Applied general statistics. biostatistics (1st ed.). New Delhi: Jaypee Brothers.
London: Sir Isaac Pitman & Sons Ltd. Negi, K. S. (2010). Biostatistics (2nd ed.). New Delhi: AITBS.
Daniel, W. W. (1987). Biostatistics: a foundation for analysis in the Pillai, R. S. N., Bhagavathi. (2008). Statistics: Theory and Practice,
health sciences. New York: Wiley. New Delhi: S. Chand & Company Ltd.
Das, R., & Das, P. N. (2009). Instant medical biostatistics. New Delhi: Rao Sunder P.S.S. (2012). Introduction to biostatistics and research
Ane Books. methods (5th ed.). New Delhi: Prentice-Hall.
Dunn, O. J. (1984). Basic statistics. New York: Wiley. Rao, S., & Richard, J. (2006). Introduction to biostatistics and
Gupta S.P. (2008). Statistical methods (1st ed.). Delhi: Sultan research methods (4th ed.). New Delhi: Prentice-Hall.
Chand. Reid, N. G. (1987). Research methods and statistics in healthcare.
Kirkwood, B. R. (1988). Essentials of medical statistics. Oxford: London: Edward Arnold.
Blackwell Scientific. Ress, D. G. (1989). Essentials of statistics. London: Chapman &
Kish, L. (1987). Statistical designs for research. New York: Wiley. Hall.
Lancaster, H. O. (1974). Introduction to medical statistics. New York: Sharma, A. K. (2005). The textbooks of Elementary Statistics, New
Wiley. Delhi: Discovery Publishing House.
Mahajan, B. K. (2010). Methods in biostatistics (7th ed.) New Delhi: Sprent, P. (1989). Applied nonparametric statistical methods. London:
Jaypee. Chapman & Hall.
Mahajan, B. K., & Khankal, A. B. (2010). Methods in biostatistics for Woolson, R. F. (1987). Statistical methods for the analysis of bio-
medical students and research workers (7th ed.). New Delhi: Jaypee. medical data. New York: Wiley.
 Measures of Relationship
and Inferential Statistics
CHAPTER OUTLINE
10
NORMAL PROBABILITY CURVE • ‘t’-test
CORRELATION COEFFICIENT
• ‘Z’-test
• ANOVA
SCATTER DIAGRAM METHOD NONPARAMETRIC TESTS
PRODUCT MOMENT CORRELATION • Chi-square test
• Sign test
REGRESSION ANALYSIS • Wilcoxon signed rank test
INFERENTIAL STATISTICS • Mann–Whitney test
MULTIPLE CHOICE QUESTIONS
PARAMETRIC TESTS

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Understand and define the basic terminologies related
• Describe the concept, properties and importance to inferential statistics.
of the normal probability curve. • Draw conclusions of the study and predict the statistical
• Discuss the concept, types and methods of computing significance of the results.
the correlation coefficient. • Demonstrate skills in computing different parametric
• Demonstrate the skills in computation of different measures and nonparametric statistical tests.
of relationship such as the Karl Pearson correlation coeffi- • Understand and explain the advantages and disadvan-
cient, Spearman rank correlation coefficient, scatter diagram, tages of parametric tests.
product moment correlation and regression analysis. • Understand and explain the advantages and disadvan-
• Explain the basic concepts of inferential statistics. tages of nonparametric tests.
• Use inferential statistics to predict results.

KEY TERMS

• Karl Pearson correlation • ANOVA 247 • Parametric tests 244

coefficient  234 • Chi-square test 250 • Sign test 251
• Normal probability curve 232 • Confidence interval 242 • t-Test 244
• Product moment correlation 237 • Degree of freedom 243 • Type-I and type-II errors 242
• Regression analysis 239 • Inferential statistics 242 • Wilcoxon signed rank test 252
• Scatter diagram 235 • Level of significance 242 • Z-test 246
• Spearman rank correlation • Mann–Whitney U-test 253
coefficient 234 • Nonparametric tests 250
232 Research Methodology and Biostatistics

l It is a continuous-type probability curve.

NORMAL PROBABILITY CURVE l The curve is symmetrical and asymptotic
(Z SCORE) (i.e. touches at infinity).
If large values are collected for any character and l All the measures of central tendency are
a frequency table is prepared with small class in- equal and stable on the highest peak axis, i.e.
tervals, the frequency curve of these data will give mean 5 median 5 mode.
a bell-shaped symmetrical curve, which is known l The total area under the curve is equal to unity.
as the Gaussian or normal curve. The shape of this The quartiles Q1 and Q3 are equidistant from the
curve depends on the mean and SD of the data. If mean m, and Q1 5 m 2 0.67s, Q3 5 m 1 0.67s
the standard deviation (variation) is very high, the approximately.
width of the curve is also more. If we move both l The normal curve has two parameters: mean
sides from the midpoint (x 5 mean), the height of (m) and standard deviation (s).
the curve decreases. If the mean and standard de- l On the basis of mean (m) and standard devia-
viation are 0 and 1, respectively, then the normal tion (s), the area of the normal curve is distrib-
curve is called the standard normal curve. If the uted as follows:
total area under the curve is considered as unity,  m 6 1s 5 68.27%

the normal curve is called the normal probability  m 6 2s 5 95.45%

curve (Fig. 10.1).  m 6 3s 5 99.73%

The normal distribution is as follows:  m 6 0.67s 5 50%

 m 6 1.96s 5 95%
( x )2
1   m 6 2.58s 5 99%
f (x ) e 2 2
 2 l The normal curve is mesokurtic.
where l Thus, the total area under the curve above the
m 5 mean distribution (it also may be median x-axis is 1 square unit. This characteristic fol-
and/or more) lows from the fact that the normal distribution
s 5 standard deviation, and its variance is is a probability distribution. Because of the
therefore 2 symmetry already mentioned, 50% of the area is
to the right of a perpendicular erected at the
PROPERTIES OF NORMAL PROBABILITY CURVE mean and 50% is to its left.
l Normal probability curve is a normal bell-
IMPORTANCE OF THE NORMAL PROBABILITY
shaped curve.
CURVE
Normal distribution plays an important role in
sampling theory. It has been found that
l Data obtained from biological measurements

approximately follow normal distribution.

l Binomial and Poisson distribution can be

approximated to normal distribution.

l For a large sample, any statistics (i.e. sample

mean, sample standard deviation) approxi-

2.15% 13.6% 34.1% 34.1% 13.6% 2.15% mately follows normal distribution, and as such
it can be studied with the help of a normal
m – 3s m – 2s m– σ m m +s m + 2s m + 3s
68.2%
curve.
l Normal curve is used to find confidence limits
95.4%
99.7%
of the population parameters.
l Normal distribution also forms the basis for the

Fig. 10.1. ​Normal probability curve. tests of significance.

 Measures of Relationship and Inferential Statistics 233

For example, the mean height of 500 students TYPES OF CORRELATION COEFFICIENT
is 165 cm and the standard deviation is 5 cm. 1. Perfect positive correlation (r 5 11)
Assuming that the heights are normally distrib- 2. Perfect negative correlation (r 5 21)
uted, find how many students will have height 3. Moderately positive correlation (0 , r , 1)
between 153 and 180 cm. 4. Moderately negative correlation (21 , r , 1)
5. Absolutely no correlation (r 5 0)
Solution
Perfect positive correlation (r 5 11) (Fig. 10.2)
Here, mean (m) 5 165 cm, standard deviation X is directly proportional to Y.
(s) 5 5 cm Both variables rise and fall in the same propor-
tion.
X   153 165 12
We have Z    2.4 For example, designation and salary, height and
 5 5 weight.
180 165 15
   3.0 Perfect negative correlation (r 5 –1) (Fig. 10.3)
5 5 Here X and Y are inversely proportional to each
Therefore, the proportion of students whose other.
r 5 21
height is in between 153 and 180 cm
5 Area under the standard normal curve between X is inversely proportional to Y.
Z 5 22.4 and 3.0 For example, leniency of teacher and discipline
5 (Area between Z 5 22.4 and 0) 1 (Area be- among students.
tween Z 5 0 and 3.0) Moderately positive correlation (0 , r , 1)
5 0.4918 1 0.4987 (see Appendix V) 5 0.9905 (Fig. 10.4)
Hence, the total number of students whose For example, the correlation between infant
height is between 153 and 180 cm 5 0.9905 3 500 mortality rate (IMR) and overcrowding, and
5 495 (approximately).

CORRELATION COEFFICIENT
Sometimes two continuous characters are mea-
sured in a series or in a similar pattern such as
weight and cholesterol, weight and height, IQ in Fig. 10.2. ​Perfect positive relationship.
brothers and sisters. This relationship or associa-
tion between two quantitatively measured or
continuous variables is called correlation.
The extent or degree of relationship between
two sets of figures is measured in terms of another
parameter called the correlation coefficient. It is
denoted by r. One can find the correlation
between weight and cholesterol and cholesterol Fig. 10.3. ​Perfect negative relationship.
and coronary artery disease.
Correlation determines the relationship
between two variables, but it does not prove that
one particular variable alone causes change
in the other (–1 # r # 1). Correlation between
overcrowding and tuberculosis (TB) is established,
but it does not mean that the high rate of TB is
due to overcrowding alone. Fig. 10.4. ​Moderately positive relationship.
234 Research Methodology and Biostatistics

correlation coefficient (PMCC). It is denoted by r and

can be computed using the following formula:
∑ ( x  x )( y  y )
r
∑(x x ) ∑( y  y )
2 2

Fig. 10.5. ​Moderately negative relationship.

For example, compute the correlation coefficient
from the following data.
Weight (kg) 60 70 80 90
Cholesterol (mg/dL) 120 130 140 150

Fig. 10.6. ​Absolutely no relationship. Solution

∑ ( x  x )( y  y )
r
∑(x x ) ∑( y  y )
2 2

correlation between the plasma value and

circulating albumin in grams.
Moderately negative correlation (–1 , r , 1)
(Fig. 10.5) 500 500 500
r   1
For example, age and vital capacity in adults, 500 500 2, 50, 000 500
IMR and income.
Absolutely no correlation (r 5 0) (Fig. 10.6) Hence, there is perfect correlation between weight
For example, habit of smoking and housing and cholesterol level of patients. This means that
facility of a person. there is an increase in cholesterol level with in-
crease in weight of patients.
METHODS OF COMPUTING THE CORRELATION
The correlation between variables can be studied Spearman Rank Correlation Coefficient
by the following ways: It is a method of finding the correlation between
l Karl Pearson correlation coefficient two variables by taking their ranks. This method
l Spearman rank correlation coefficient of finding correlation is especially useful in
dealing with qualitative data. It can be used when
Karl Pearson Correlation Coefficient the actual magnitude of characteristics under
It is used to measure the degree of linear relationship consideration is not known, but relative position
between two variables. It is also called product moment or rank of the magnitude is known. It is denoted

x y x2x y2y ( x 2 x )( y 2 y ) ( x 2 x )2 ( y 2 y )2
60 120 –15 –15 225 225 225
70 130 25 25 25 25 25
80 140 5 5 25 25 25
90 150 15 15 225 225 225
∑ ( x  x ) ( y  y )  500 ∑ ( x  x )  500 ∑ ( y  y )  500
2 2
x 575 y 5135
 Measures of Relationship and Inferential Statistics 235

by p. There are two cases for calculating rank cor- l Case 2 – Ranks correlation with a tie between
relation. In the first case, there is no tie among allotted ranks: In this case, any one or more
allotted ranks, i.e. no two numbers are the same; values in the x and y series is repeated. So we
in the second case, there is a tie between allotted have to apply a correlation factor (CF), and
ranks. the correlation coefficient is calculated by using
l Case 1 – Ranks correlation with no tie among the following formula:
allotted ranks: In this case, all the values in the x
6 ∑ d 2  CF
or y series are not repeated. We can use the follow-   1
ing steps for calculating the Spearman rank cor- (n n 2 1 )
relation coefficient with no tie among allotted (m 2
1) 1
ranks: CF  mx
11
or CF 
12
(m3 m )
 Rank the highest value 1, rank the next high-

est value 2 and so on. where m is number of repetition of ranks in the

 Rank x series values and y series values given series.
separately.
 Calculate the difference of ranks in each pair

of values or (d 5 Rx – Ry)
SCATTER DIAGRAM METHOD
 Calculate the sum of the square of the differ- The simplest method for ascertaining the rela-
ence of ranks (Sd2). tionship between two variables by preparing a
 Finally, calculate the correlation coefficient dot chart is called scatter diagram (scatter plot or
using the following formula: dot diagram) method. The scatter diagram is a
tool to determine the potential correlation be-
6∑d2 tween two different sets of variables, i.e. how one
  1
(
n n 2 1 ) variable changes with the other variable. One
variable is plotted on the horizontal x-axis and
For example, find the rank correlation of the the other is plotted on the vertical y-axis. The
marks obtained by five nursing students in pattern of their intersecting points can graphi-
anatomy and medical surgical nursing. cally show relationship patterns between X and
Y. Most often a scatter diagram is used to prove
or disprove cause–effect relationships. The scat-
Anatomy (x) 85 81 77 68 53 ter diagram does not determine the exact rela-
Medical surgical 78 70 72 62 67 tionship between the two variables, but it does
nursing (y) indicate whether they are correlated or not. Also,
scatter diagrams do not predict cause–effect rela-
tionships between these variables.
x y Rx Ry d 5 Rx 2 R y d2
85 78 1 1 0 0 USE OF SCATTER DIAGRAM METHOD
81 70 2 3 –1 1 l Quickly confirms a hypothesis that two vari-
77 72 3 2 1 1 ables are correlated.
68 62 4 5 –1 1
l Provides a graphical representation of the strength
of the relationship between two variables.
53 67 5 4 1 1
l It also helps in an effective examination and
∑d2 5 4 interpretation about cause–effect relationship
to evaluate whether manipulation of the inde-
Hence, this indicates that marks of the two pendent variable (cause) actually produces the
subjects are partially positive correlated. change in the dependent variable (effect).
236 Research Methodology and Biostatistics

STEPS TO MAKE A SCATTER DIAGRAM l If the points are widely scattered over the
The following steps can be followed to make a diagram, it indicates very little relationship
scatter diagram: between the variables, either negative or posi-
l Step 1: On a graph paper or normal paper, draw tive (Figs 10.9 and 10.10).
a horizontal line showing the x-axis and a verti- l If the plotted points lie in a haphazard manner,
cal line showing the y-axis. Make your scale it shows the absence of any relationship be-
units at even multiples of 10, 20, etc. so as to tween the variables (Fig. 10.11).
have an even scale system. l A line of best fit (or ‘trend’ line) is a straight line
l Step 2: On the horizontal axis (from left to that best represents the data on a scatter plot. This
right), place the independent or ‘cause’ variable. line may pass through some of the points, none
l Step 3: On the vertical axis (from bottom to of the points, or all of the points. Depending on
top), place the dependent or ‘effect’ variable. the number of dots touching the line, one can
l Step 4: Plot the data points (dots) at the inter- ascertain the relationship between the variables.
section of X and Y values. For example: X 5 7,
Y 5 2. Go right 7 spaces, and then go up 2
15
spaces to plot the point. The plotted dots on the
graph will generally look scattered; hence, the 10
name scatter plot. Remember that the points are
not connected in scatter diagram. 5
l Step 5: Interpret the data and find the relationship.

0
INTERPRETATION OF THE RELATIONSHIP 0 10 20
l The type of relationship that exists is indicated
by the slope of the diagram. The manner in Fig. 10.8. ​Perfect negative correlation (r 5 –1).
which these points are scattered, suggest the
degree and the direction of correlation.
l The greater the scatter of plotted points on the chart, 20

the lesser the relationship between the two variables. 15

l If the points come more closely to a straight line,
10
falling from the lower left corner to the upper right
corner, the correlation is said to be perfectly posi- 5
tive (r 5 11), as illustrated in Fig. 10.7. 0
l On the other hand, if all the points lie on the 0 5 10 15
straight line falling from the upper left corner to
the lower right corner, correlation is said to be per- Fig. 10.9. ​Weak positive correlation (0.5 , r . 0).
fectly negative (r 5 21), as illustrated in Fig. 10.8.

60 20

15
40
10
20
5

0 0
0 10 20 0 5 10 15

Fig. 10.7. ​Perfect positive correlation (r 51). Fig. 10.10. ​Weak negative correlation (–0.5 , r . 0).
 Measures of Relationship and Inferential Statistics 237

represents the minimum Y value; for a negative

15
correlation, it represents the maximum Y value.
10 MERITS OF SCATTER DIAGRAM
l It is a simple and nonmathematical method of
5
studying correlation between variables.
l Easily understood and a rough idea can be quickly
0
0 5 10 formed as to whether variables are related or not.
l It is not influenced by the size of extreme items,
Fig. 10.11. ​No correlation (r 5 0).
whereas most mathematical methods of finding
correlation are influenced by extreme values.

The more the number of dots touching the line, DEMERITS OF SCATTER DIAGRAM
the stronger is the relationship. l By applying this method, we can get an idea
about the direction of correlation and also
LINEAR RELATIONSHIP whether it is high or low. But we cannot estab-
It means do the data ‘line up’? Linearity has four lish the exact degree of correlation between the
parameters: variables.
1. Correlation: It measures how well the data line l Although this method is simple and gives a
up. The more the data resemble a straight line, rough idea about the existence and the degree of
the higher the correlation with each other. correlation, it is not reliable. As it is not a math-
2. Slope: It measures the steepness of the data. ematical method, it cannot measure the degree
The steeper the data slope, assuming the cor- of correlation (r).
relation is good, the greater the importance of
the relationship. A change in the X variable will
have a larger impact on the Y variable, and you PRODUCT MOMENT CORRELATION
will begin to see a pattern that represents the The PMCC is a measurement of the degree of
moderate correlation diagram (Fig. 10.12). scatter. It is also called the Pearson product mo-
3. Direction: The X variable can have a positive or ment correlation coefficient. It is a measure of the
a negative impact on the Y variable. As one fac- strength of a linear association between two vari-
tor goes up, the other goes down. In a positive ables (say X and Y) and is denoted by r. It is widely
correlation, both factors will move in the same used in the sciences as a measure of the degree of
direction. In the graph examples explained linear dependence between two variables. It was
earlier, you can see that the positive correlation developed by Karl Pearson from a related idea in-
moves from the lower left, towards the upward troduced by Francis Galton in the 1880s.
right. The negative correlation moves from the It gives a value between 11 and 21, where 1 is
lower right, towards the upward left. total positive correlation, 0 is no correlation and
4. Y-Intercept: A line drawn through the data 21 is negative correlation. Basically, a Pearson
crosses the y-axis. For a positive correlation, it product moment correlation attempts to draw a

Strong Weak Weak Strong

−1.0 −0.5 0.0 +0.5 +1.0

Negative Zero Positive

correlation correlation

Fig. 10.12. ​Correlation coefficient showing strength and direction of correlation.

238 Research Methodology and Biostatistics

line of best fit through the data of two variables, Here,

and the PMCC, r, indicates how far away all these - n – the number of data 5 5.
data points are to this line of best fit (how well the - Sx – the sum of all the x values.
data points fit this new model/line of best fit). The - Sx2 – the sum of the squares of the
method is used only for interval and ratio data. x values.
For ordinal data, Spearman rank order correlation - Sy – the sum of all the y values.
or a Kendall tau correlation can be used. - Sy2 – the sum of the squares of the y values.
- Sxy – the sum of each x value multiplied by
ASSUMPTIONS FOR PEARSON PRODUCT its corresponding y value.
MOMENT CORRELATION l Step 2: Calculate Sxy, Sxx and Syy using these values
There are three assumptions that are made with Sxy 5 xy 2 (Sx Sy 4 n), Sxy 5 283 2 (12 3
respect to Pearson correlation: 93/5) 5 59.8
1. The variables must be either interval or ratio Sxx 5 Sx2 2 (Sx Sx 4 n), Sxx 5 40 2 (12 3
measurements. However, both variables do not 12/5) 5 11.2
need to be measured on the same scale (e.g. one Syy 5 Sy2 2 (Sy Sy 4 n), Syy 5 2089 2 (93 3
variable can be ratio and one can be interval). 93/5) 5 359.2
2. The variables must be approximately normally l Step 3: Insert these values into the following equation
distributed. to calculate the product moment correlation (r).
3. There is a linear relationship between the two
Sxy
variables. r5
Sxx . S yy
STEPS OF CALCULATION OF PRODUCT
MOMENT CORRELATION 59.8
r
Let us calculate the product moment correlation 11.2  359.2
using an example in economics. You are the owner r  0.9
of a restaurant. For every 10th customer, you re-
cord the time he/she stayed in your restaurant (x, INTERPRETATION OF PRODUCT MOMENT
in minutes) and the amount spend (y, in dollars). CORRELATION
Is it generally true that the long stayers are also the The PMCC, r, can take a range of values from 11
bigger spenders? This would be a positive correla- to 21. A value of 0 indicates that there is no asso-
tion. Or is it actually the other way around, e.g. the ciation between the two variables. A value greater
richer the client, the lesser the time he takes for his than 0 indicates a positive association; that is, as the
lunch? This would be a negative correlation. In value of one variable increases, so does the value of
order to shed some light on this mystery, you can the other variable. A value less than 0 indicates a
calculate the PMCC. negative association, i.e. as the value of one variable
l Step 1: Summarize the data into the values needed increases, the value of the other variable decreases.
for the calculation, i.e. n, Sx, Sy, Sxy, Sx2, Sy2. The stronger the association of the two vari-
ables, the closer the PMCC (r) to either 11 or 21
x y xy x2 y2 depending on whether the relationship is positive
1 10 10 1 100 or negative, respectively. Achieving a value of 11
3 25 75 9 625 or 21 means that all your data points are included
1 8 8 1 64 on the line of best fit – there are no data points
that show any variation away from this line. The
2 20 40 4 400
closer the value of r to 0 the greater the variation
5 30 150 25 900
around the line of best fit. Different relationships
Sx 5 Sy 5 Sxy 5 Sx2 5 Sy2 5 and their correlation coefficients are shown in the
12 93 283 40 2089 following diagram.
 Measures of Relationship and Inferential Statistics 239

r = 0.7 r = 0.3 r=0

Strong positive correlation Weak positive correlation No correlation

r = −0.7 r = 0.3 r=0

Strong negative correlation Weak negative correlation No correlation

Guidelines to interpreting Pearson product REGRESSION ANALYSIS

moment correlation coefficient
After studying the correlation that two variables are
Coefficient (r) closely related, in regression, one can estimate or
Strength of Association Positive Negative predict the value of one variable by using the value
Small (weak) 0.1–0.3 20.1 to 20.3
of the other variable. For example, if we know that
the yield of rice and rainfall are closely related, we
Medium (moderate) 0.3–0.5 20.3 to 20.5
will be interested to know the amount of rain re-
Large (strong) 0.5–0.9 20.5 to 20.9 quired to achieve the desired production of rice.
Perfect 11 21
Regression analysis attempts to establish the
MERITS nature of relationship between the variables,
i.e. to study the functional relationship be-
l It summarizes in a figure not only the degree of
tween the variables and thereby provide a
correlation but also the direction.
mechanism for prediction, or forecasting.
DEMERITS (Ya-Lun Chou)
The term regression analysis refers to the
l The correlation coefficient always assumes lin-
methods by which estimates are made of
ear relationship regardless of the fact that the
the values of a variable from knowledge of
assumption is correct or not.
the values of one or more other variables and
l Great care must be exercised in interpreting the
in the measurement of errors involved in this
values of the coefficient.
estimation process. (Morris Hamburg)
l The value of the coefficient is unduly affected by
the extreme items. In simple terms, it is a statistical measure that at-
l It is a time-consuming method. tempts to determine the strength of the relationship
240 Research Methodology and Biostatistics

between one dependent variable (usually denoted l Multiple/progestic regression analysis (1:2, 1:3
by Y) and a series of other changing variables or more association of variables)
(known as the independent variables and denoted by
X). For regression, the variables should be related to Linear Regression
each other. It is applicable to data with variables. The Linear regression is an approach to model the rela-
analysis starts with a scatter plot of Y vs. X. tionship between a scalar dependent variable Y and
one or more explanatory variables denoted by X.
VARIABLES l The case of one explanatory variable is called

l X 5 independent/explanatory variable (we are simple linear regression.

using to predict) l For more than one explanatory variable, it is

l Y 5 dependent/explained variable (we are called multiple linear regression.

trying to predict) l In multivariate linear regression, multiple corre-

The dependent and independent variables refer lated dependent variables are predicted rather
to the mathematical and functional meaning of than a single scalar variable.
dependence and do not imply cause–effect rela-
tionship between variables. It simply means that Multiple Regression Analysis
the estimates of values of the dependent variable Y It is a technique that allows additional factors to
may be obtained from the given values of the inde- enter the analysis separately so that the effect of
pendent variable X. But the X variable may or may each can be estimated. It is valuable for quantify-
not be causing change in Y. For example, while es- ing the impact of various simultaneous influences
timating production of rice from the amount of upon a single dependent variable.
rainfall, rice is the dependent variable. However,
there may or may not be casual relationship be- REGRESSION LINE
tween the two in the sense that production of rice In case of two variables such as X and Y, we shall
can only be estimated by the amount of rainfall. have two regression lines, i.e. regression of X on Y
and the regression of Y on X. The regression of X
PURPOSES AND USES OF REGRESSION ANALYSIS on Y gives the values of Y and vice versa. However,
l To estimate the values of a dependent variable in case of perfect positive (r 5 11) or negative
from values of an independent variable – regres- relationship (r 5 21), the regression line will
sion line is a method which is used to accomplish coincide, i.e. there will be one single line. In case
the estimation of no relationship (r 5 0), the regression lines are
l To determine change in Y with change in each at right angles to each other.
of the independent variable, i.e. X Regression lines should cut each other at a
l To determine the measure of error involved in point of average of X and Y. A perpendicular is
using the regression line as the basis of estima- drawn on the x-axis at the point where both lines
tion – This is determined by calculating the cut each other; this will be the mean value of X.
standard error (SE) of the estimate; it is calcu- Similarly, a perpendicular can be drawn on y-axis
lated by measuring the of scatter of observa- which is the mean value of Y.
tions around the regression line
l To find the relationship between variables – The Regression/Estimating/Normal Equations
farther the two regression lines from each other, the A regression equation is the algebraic expression of
lesser the degree of correlation and vice versa the regression lines. Since there are two regression
lines, there should be two regression equations:
TYPES OF REGRESSION ANALYSIS Regression equation of X on Y
Regression analysis is of two types:
l Linear regression analysis (1:1 association of
SX 5 Na 1 bSY
variables) SXY 5 aSY 1 bSY2
 Measures of Relationship and Inferential Statistics 241

Regression equation of Y on X 80 5 8a 1 64b

SY 5 Na 1 bSX 683 5 64a 1 584b
SXY 5 aSX 1 bSX2 To make the equations equal, multiply the 1st
equation by 8,
where
a: Y-intercept (value of point at which the regres- 640 5 64a 1 512b
sion line crosses the y-axis or vertical axis) 683 5 64a 1 584b (here, 64a of both equa-
b: Slope of the line (change in Y variable for a unit tions will get cancelled)
change in X variable) After calculation, b 5 0.59
a and b: numerical constants (called so because Now, after putting the value of b in any equation,
for a given straight line, their values remain the value of a can be calculated as follows:
constant)
SX, SY, SXY, SX2, SY2: Totals which are 80 5 8a 1 64 3 0.59
computed from the observed pair of values of a 5 5.28
variables X and Y Regression equation of Y on X
CORRELATION COEFFICIENT SY5 Na 1 bSX
It is the square root of the product of the two re- SXY 5 aSX 1 bSX2
gression coefficients: 64 5 8a 1 80b
r  bXY bYX 683 5 80a 1 832b
To make equations equal, multiply the 1st
where
equation by 10
bXY: Regression coefficient (X upon Y)
bYX: Regression coefficient (Y upon X) 640 5 80a 1 800b
Value of r is interpreted in the same way as in the 683 5 80a 1 832b (here, 80a of both equa-
Karl Pearson correlation coefficient. tions will get cancelled)
Example: From the given data, obtain two After calculation, b 5 1.34
regression equations and hence find the regression Now, after putting the value of b in any equa-
coefficient. tion, the value of a can be calculated as follows:
X Y XY X2 Y2 64 5 8a 1 80 3 1.34a 5 25.4
10 9 90 100 81
6 4 24 36 16
Regression coefficient ( r )  bXY  bYX

9 6 54 81 36  1.34  0.559 0.88

10 9 90 100 81
12 11 132 144 121
Note:
1. Coefficient of correlation (r) can be only calcu-
13 13 169 169 169
lated if both the regression coefficients, i.e. bXY and
11 8 88 121 64 bYX have the same sign, either positive or negative.
9 4 36 81 16 2. Since the value of the coefficient of correlation
80 64 683 832 584 (r) cannot exceed one, one of the regression
coefficients must be less than 1.
Regression equation of X on Y 3. Coefficient of correlation will have the same
sign as that of the regression coefficients. For
SX 5 Na 1 bSY
example, if regression coefficients are negative,
SXY 5 aSY 1 bSY2 then r will also be negative.
242 Research Methodology and Biostatistics

wrong 5% of the times, 5 times out of 100, when

INFERENTIAL STATISTICS
rejecting the null hypothesis. If the decision needs
Inferential statistics is concerned with populations to be much more accurate, such as deciding
and uses sample data to make an inference about whether a nursing intervention is effective or not,
the population or to test the hypotheses consid- the level of significance might be set at 0.01 or
ered at the beginning of the research study. Infer- even at 0.001. With a 0.01 level of significance, the
ential statistics helps the researcher to determine researcher stands the risk of being wrong 1 out of
whether the difference found between two or more 100 times when rejecting the null hypothesis. At
groups, such as an experimental and a control the 0.001 level of significance, the risk is 1 out of
group, is a real difference or only a chance differ- 1000 times. However, most nursing studies con-
ence that occurred because an unrepresentative sider 0.05 level of significance for taking decision
sample was chosen from the population. Before regarding whether to accept or reject the null
learning about the test of significance, let us un- hypotheses.
derstand a few of the important concepts related to
inferential statistics. Confidence Interval
Confidence interval (CI) is a range of values that
BASIC TERMS RELATED TO INFERENTIAL
with a specified degree of probability is thought
STATISTICS
to contain the population value. The researcher as-
Type-I and Type-II Errors serts with some degree of confidence that the pop-
Type-I error occurs when the null hypothesis is ulation parameter lies within those boundaries.
rejected, when it should have been accepted; it is The CI contains a lower and an upper limit. For
also called alpha error. Type-II error occurs when example, Ram conducted a study on 100 house-
the null hypothesis is accepted, when it should holds to assess the per capita income of households
actually have been rejected. It is also known as at Chandigarh and found that mean per capita in-
beta error. These errors generally occur due to an come is Rs 130 with a standard deviation of Rs 3.
unrepresentative sample being drawn from the Experts may ask the researcher that if a similar
population (Table 10.1). study is conducted on the sample population,
whether the results would be similar. In this case,
Level of Significance the researcher may calculate the CI for an upper
The probability of making a type-I error is called and lower level of per capita income so that he is
the level of significance. It is represented by a or 95% or 99% confident with the limits, and men-
p. In other words, level of significance is the prob- tions that if the study is repeated, there is 95% or
ability of rejecting the null hypothesis when it is 99% probability of the per capita income falling in
true. In health sciences, we generally consider the these limits. Level of significance is one of the im-
level of significance as either 1% (0.01) or 5% portant parameters of statistics; but unfortunately
(0.05). A significance level of 0.05 means that the very little use of it has been made in Indian or West-
researcher is willing to take the risk of being ern nursing studies or nursing literature; however,
medical literature has consistently made large use
of the CI in medical studies.
Table 10.1. ​Explanation about type-I and type-II If the researcher wants to be 95% confident about
errors his or her estimation, calculation will be carried
Decision out as follows:
First, estimate SE of the mean by using the
Reality Accept H0 Reject H0
following formula:
H0 (true) Correct decision Type-I error
H0 (false) Type-II error Correct decision 3 3
SE 5 5 5 0.3
Note: H0: null hypothesis. 100 10
 Measures of Relationship and Inferential Statistics 243

Next, insert 0.3 into the formula for obtaining the populations from which the samples are
the 95% CI (lower and upper limits): drawn. For example, the assumptions may be that
the populations that are normally distributed
Lower limit  X  1.96 ( SE )
have the same variance, etc. Population values, as
 130  1.96 ( 0.3)  130  0.588  129.412 we have seen, are known as parameters; the statis-
Upper limit  X  1.96 ( SE ) tical techniques that make assumptions about the
 130  1.96 ( 0.3)  130  0.588  130.588 parameters are called parametric techniques.
Nonparametric tests: Many times, researchers in
Here the researcher can say that he is 95% con- the field of health sciences may not be aware
fident that the per capita income of the households about the nature of the distribution or other
in Chandigarh is in between 129.421 and 130.588. required population parameters. In addition,
If the researcher wants to be 99% confident that their sample may be too small to test the hy-
estimates are true, he has to replace 1.96 in the potheses and generalize the findings for the
formula with 2.58 as follows: population from which the sample is drawn.
Furthermore, many times in the observations
Lower limit  X  2.58 ( SE )
presented in numerical figures, the scale of mea-
 130  2.58 ( 0.3)  130  0.774  129.226 surements may not be really numerical, such as
Upper limit  X  2.58 ( SE ) grading bedsores or ranks given to analgesic
 130  2.58 ( 0.3)  130  0.774  130.774 drug’s effectiveness in cancer pain management.
In these situations, parametric tests may not be
Here the researcher can say that he is 99% con- suitable and a researcher may need different
fident that per capita income of the households in types of tests to draw inferences; those tests are
Chandigarh is in between 129.226 and 130.774. known as nonparametric tests.
The circumstances in which nonparametric
Degree of Freedom tests are applicable are as follows :
The interpretation of a statistical test depends on When we have to test an assumption about the
the degree of freedom (df). It is denoted by the population distribution with a random sample
abbreviation ‘df ’ and a number (e.g. df 5 3). from the population:
Degree of freedom indicates the number of values (a) Binomial test can be used when the data are
that are free to vary. Although the df indicates in two categories and the sample size is
the number of values that can vary, the concern is small.
really focused on the number of values that are (b) Chi-square test can be used when the data
not free to vary. The procedure to calculate the are in discrete categories (two or more) and
df varies from test to test; further details are the expected frequencies in the sample of
discussed with individual tests of significance and these categories (based on our assumption)
can be referred from there. are sufficiently large.
(c) Kolmogorov–Smirnov test will be applica-
TESTS OF SIGNIFICANCE ble when the variable has a continuous
There are several parametric (t-test, Z-test, analy- distribution.
sis of variance [ANOVA]) and nonparametric When we have to test whether two random sam-
tests (chi-square test, median test, McNemar test, ples have likely come from populations with the
Mann–Whitney U-test, Wilcoxon signed rank test, same mean:
Fisher exact test) available to establish statistical (a) Randomization test can be used with small
significance. samples when the data consist of measure-
Parametric tests: These tests are also known as nor- ments on a scale and the Kolmogorov–
mal distribution statistical tests. Statistical meth- Smirnov test can be made use of with
ods of inference make certain assumptions about weaker measurements.
244 Research Methodology and Biostatistics

(b) Mann–Whitney U-test may be used with l Almost nil or very limited assumptions, which
large samples when the data represent are needed to be made about the data format, are
weaker measurements. needed for nonparametric methods. Therefore,
(c) Median test is used when the data cannot they may be preferred when the assumptions
even be meaningfully ranked and when it is needed for parametric methods are invalid.
possible only to dichotomize the observa- l If samples are of sizes as small as six, there is no
tions as above or below the combined me- alternative to using a nonparametric test unless
dian; the Fisher test must be used to analyse the nature of the population distribution is
the data if the combined sample size is precisely known.
small. l There are suitable tests for treating observations
(d) When more than two samples are consid- from samples drawn from several different
ered, the Kruskal–Wallis test can be applied populations.
to test whether they all belong to the same l Tests are available to treat data that are inher-
population. ently in ranks as well as data whose seemingly
(e) Fisher exact probability test is used when all numerical scores have only the strength of ranks.
the scores from the independent random l Methods are available to treat data that are
samples fall into one or other of mutually simply classificatory.
exclusive classes. The test determines l These tests are much easier to learn and apply
whether the two groups differ in the pro- than parametric tests.
portion with which they fall into the two
classifications. Disadvantages of Nonparametric Tests
When we have to find out the statistical significance The disadvantages of nonparametric tests are as
of differences in matched pair comparisons: follows:
(a) McNemar test is used when the data are l If all the assumptions of the parametric test are

frequencies in different categories. in fact met in the data, and if the measurement
(b) Sign test is used when the data are on a is of the required strength, then nonparametric
variable that has underlying continuity, but tests are wasteful of data.
can be measured only in a gross way. In this l There are nonparametric methods for testing

case, the ranks within the pair are used. If, interactions in the analysis of variance.
however, the differences observed for the l Tables of critical values may not be easily

various matched pairs can be meaningfully available.

ranked, Wilcoxon signed rank test can be l Nonparametric methods may lack power as

applied. compared with more traditional approaches,

(c) When the data are measurements in a nu- especially when sample size is very small.
merical scale, and when the sample size is
sufficiently small, randomization test can
be applied. PARAMETRIC TESTS
When we have to measure the correlation as the In this section, the most commonly used parametric
observations are ranked: tests, i.e. t-test, Z-test and ANOVA, are discussed.
(a) Kendall ‘t’ (tau) is used.
(b) Or Spearman rs is used. t-TEST
It is applied to find the significant difference
Advantages of Nonparametric Tests between two means. This test can be applied when
The advantages of nonparametric tests are as the following criteria are fulfilled:
follows: l Randomly selected homogeneous sample

l Probability statements obtained from most l Quantitative data measures on interval or ratio

nonparametric tests are exact probabilities. scale

 Measures of Relationship and Inferential Statistics 245

l Variability is normally distributed Primary Hypertension Secondary Hypertension

l Sample size less than 30; if sample size is more than (mm Hg) (mm Hg)
30, Z-test is applied. However, sometimes, even 122 132
t-test can be applied in sample size of more than 30.
134 130
Types of t-Tests 120 120
The t-test is further divided into two types: 126 122
Unpaired t-test: It is applied when we obtain data 124 126
from subjects of two independent separated 120 124
groups of people or samples drawn from two 122 120
different populations.
130 124
Paired t-test: It is applied on paired data of inde-
pendent observations made on the same sample 128 126
before and after the intervention. Paired test is 136 –
most commonly used in nursing research studies. n1 5 10 n2 5 9

Steps of Application of Unpaired t-Test

Solution. Calculate the mean and SD of both the
Following are the steps for the application of the
groups.
unpaired t-test.
Calculate the SE by using the following formula. X1 5 126.2; SD1 5 5.4; n1 5 10
X 2 5 124.9; SD2 5 3.9; n2 59
( SE ) 
(SD1 )2  (SD2 )2
n1 n2 Calculate the SE using the following formula:
where
SE 
(SD1 )2  (SD2 )2
SD1: Standard deviation of the first sample
n1 n2
SD2: Standard deviation of the second sample
n1: Number of subjects in the first sample (5.4 )2 ( 3.9)2
SE  
n2: Number of subjects in the second sample 10 9
Calculate the observed difference between the two SE  2.146
means.
Calculate the t-test value using the following formula:
Observed difference  X1  X 2 .
( X1  X 2 )  126.2 – 124.9
where X1 is mean of the first sample and X 2 is t -test 
SE 2.146
the mean of the second sample.
 0.61
Calculate the t-value by using the following formula:
Observed difference ( X1  X 2 ) Degree of freedom for the unpaired sample
t -test  (df 5 n1 1 n2 – 2):
SE

Compute the degree of freedom (df). df 5 10 1 9 – 2 5 17

df 5 n1 1 n2 – 2 (unpaired sample) and n – 1 (paired Refer to the tabulated t-value (see Appendix I)
sample) at 17 df for the 0.05 level of significance.
If the tabulated t-value is more than the
Example of an Unpaired t-Test calculated t-value, accept the null hypothesis
The following data show the effect of hypnosis (p . 0.05). If the tabulated t-value is less than
on the blood pressure among the patients with the calculated t-value, reject the null hypothesis
primary and secondary hypertension. (p , 0.05).
246 Research Methodology and Biostatistics

Example of a Paired t-Test l Degree of freedom for the paired sample

Calculate the paired t-test for the given data of (df 5 n – 1):
hypnosis efficacy on the BP of patients with pri-
df 5 10 – 1 5 09
mary hypertension.
l Refer to the tabulated t-value (see Appendix I)
BP before BP after at 9 df, for the 0.05 level of significance. If the
Intervention Intervention Difference tabulated t-value is more than the calculated
(BP in mm (BP in mm in Paired Square of t-value, accept the null hypothesis (p . 0.05). If
Hg) Hg) Observations Difference the tabulated t-value is less than the calculated
A B (d 5 A 2 B) (d2)
t-value, reject the null hypothesis (p , 0.05).
122 132 210 100
134 130 104 16 Z-TEST
120 120 0 00 When a sample is larger than 30 subjects, and a
126 122 104 16 researcher wants to compare the difference in
population mean and a simple mean or the differ-
124 126 202 04
ence between two sample means, a Z-test is
120 124 204 16 applied. There are the following four prerequisites
122 120 102 04 for application of the Z-test:
l The sample or samples must be randomly selected.
130 124 106 36
l The data must be quantitative in nature mea-
128 126 102 04
sured on interval or ratio scale.
136 120 116 256 l The variability is assumed to follow normal
n 5 10 ∑d 5 ∑d2 5 452 distribution in the population.
(134 2 l The sample size must be larger than 30.
16) 5 18
Z-test for mean has two applications:
l To test the significance of the difference between

Solution: Steps for computation of paired t-test a sample mean and a known value of popula-
l The following formula is used for computing tion by using the following formula:
the paired t-test:
Mean ( X )  Population (  )
Z
d SE of sample mean
t
( )
n ∑ d 2 ( ∑ d )
2

l To test the significance of the difference between

n 1
two sample means or between the experimental
l Calculate d (the difference between the first and the sample mean and the control sample mean by
second paired observations); compute the sum of using the following formula:
the calculated difference (∑d): see the table. Observed difference between two sample meaans
l Calculate d2 and sum of d2 as given in the table. Z5
SE of difference between two sample mean
ns
l Calculate the t-value using the following formula:
d 18 Calculation of SE and mean differences is
t 
)
n ( ∑ d 2 ( ∑ d )
2
10 ( 452 ) (18 )
2 similar to the t-test; however, in Z-test, df is not
computed; and to determine the significance of
n 1 10 1
the Z-value, the probability (p) value is found
18 18 from the following table, constructed on the basis
   0.834
54420  324 21.529 of the normal distribution (Table 10.2).
9 If the calculated Z-value is greater than the tabu-
t 0.834 lated Z-value, we reject the null hypothesis. In another
 Measures of Relationship and Inferential Statistics 247

Table 10.2. ​Probability table for the Z-test l Calculate the sum of squares within the groups
(error sum of squares) by using the following
p-value 0.10 0.05 0.02 0.01
formula: Total sum of squares – sum of squares
Z-value 1.6 2.0 2.3 2.6 between the groups.
l Calculate the df for between and within the groups.
case, if the calculated Z-value is less than the tabulated df for between the groups 5 Number of
Z-value, we accept the null hypothesis. groups – 1
df for within the groups 5 Number of subjects
ANALYSIS OF VARIANCE (ANOVA) TEST in all the groups – Number of groups
When a researcher wants to compare the difference df total 5 Number of subjects in all the
between more than two sample means, t-test will not groups – 1
be useful and the need for an alternative test will be l Calculate the mean of sum of squares by using
felt. This need can be fulfilled by a test known as the following formula:
ANOVA test. Therefore, it is clear that ANOVA is used Mean of sum of squares between the groups
to compare more than two sample means drawn
Sum of squares between the groups
from the corresponding normal population. For ex- 5
df for bettween the group
ample, a researcher wants to examine the difference
in effect of ginger on three different conditions such Mean of sum of squares withiin the groups
as nausea, vomiting and retching; here t-test can not Sum of squares within the groups
5
be applied to examine the difference in the mean df for within the groups
scores because there are more than two groups; there-
fore, ANOVA will be used in this case. l Finally compute the F-ratio by using the
following formula.
Steps of Application of ANOVA
F-ratio of square
l Calculate the total of submissions of all the Mean of sum of squares between the groups
group of observations. 5
Mean of sum of squares witthin the groups
l Calculate the sum squares of all the observations.
l Calculate the total of sum of squares by using Refer to the tabulated F-value (see Appendix II)
the following formula: for horizontal (df for between the groups) and
vertical df (for within the groups) at the specified
 ∑ 2 
( ∑ X )2
level of significance such as 0.05 and 0.01. If
N
the calculated F-value is more than the tabulated
l Calculate the sum of squares between the groups F-value, we reject the null hypothesis. If the calcu-
by using the following formula: lated F-value is less than the tabulated F-value, we
accept the null hypothesis.

( ∑ X1 ) 2  ( ∑ X 2 ) 2  ( ∑ X 3 ) 2  ( ∑ X 4 ) 2 If F-value is less than the tabulated F-value, we
n1 n1 n3 n4 accept the null hypothesis.
For example, find the difference of mean knowl-
 ...sso on
(∑ X ) 2

edge score of swine flu among four different

N
groups of nursing students.

First Year (X1) Second Year (X2) Third Year (X3) Fourth Year (X4) Interns (X5)
18 17 19 18 18
16 19 23 18 19
17 18 13 23 20
Continued
248 Research Methodology and Biostatistics

First Year (X1) Second Year (X2) Third Year (X3) Fourth Year (X4) Interns (X5)
20 15 19 14
12 22 17 19
19 12 32 16
16 14 21
14 23 18
23 17 20
16 16
13 15
21 22
20 19
14 15
14 16
17 15
14 13
13 12
18 18
19 14
16 15
15 15
16 18
13 16
14 14
16 16
20 17
17
15
16
Total 5 492 Total 5 443 Total 5 182 Total 5 59 Total 5 106
Mean 5 16.4 Mean 5 16.40 Mean 5 20.22 Mean 5 19.66 Mean 5 17.66
SD 5 2.7 SD 5 2.83 SD 5 5.21 SD 5 2.88 SD 5 2.24

Solution Calculate the total of sum of squares by using the

Calculate the total number of submissions of all following formula:
the group of observations.
∑ X2 
( ∑ X )  22,404  (1282)
2 2

∑ X  492  443 182 59 106  1282

N 75
16,43,5524
Calculate the sum of squares of all the 75 observations.  22,404 
75
∑ X 2  182 162 17 2  ... 192 162  22,404  22, 404.00  21,913.65  490.35
 Measures of Relationship and Inferential Statistics 249

Calculate the sum of squares between the groups by using the following formula:


( ∑ X1 )2  ( ∑ X 2 )2  ( ∑ X3 )2  ( ∑ X 4 )2  ...soo on  ( ∑ X )2
n1 n2 n3 n4 N


( 492) 2


( 443) 2


(182) 2


(59) 2


(106) 2


(1282) 2

30 27 9 3 6 75
2, 42, 064 1, 96, 249 33,124 3481 11, 236 16, 43, 524
     
30 27 9 3 6 75
 8068.80  7268.48  3680.44 1160.30 1872.66  21,913.65
 22,050.71 21,913.65  137.06

Calculate the sum of squares within the groups (error sum of squares) by using the following formula:
Sum of squares within the groups ( error sum of squares )
 Total sum of squares  Sum of squares between the groups
 490.35 137.06  353.29

Calculate the df for between and within the groups.

df for between the groups 5 Number of groups – 1 5 5 – 1 5 4
df for within the groups 5 Number of subjects in all the groups – Number of groups
5 75 – 5 5 70
df total 5 Number of subjects in all the groups – 1 5 75 – 1 5 74
Calculate the mean of sum of squares by using the following formula:
Sum of squares between the groups
Mean of sum of squares between the groups 5
ween the group
df for betw
137.06
5 5 34.265
4
Sum of squares within the groups
Mean of sum of squares within the groups 5
df for within the gro
oup
353.29
5 5 5.047
70

Finally compute the F-ratio by using the following formula.

Mean of sum of squares between the groups
F-ratio of square 5
hin the groups
Mean of sum of squares with
34.265
F -ratio 5 5 6.789
5.047

Analysis of variance (ANOVA)

Square of variance df Sum of square Mean of sum of square F-ratio
Between the groups 5–154 137.06 34.265 6.789a
Within the groups 75 – 5 5 70 353.29 5.047
Total 75 – 1 5 74
a
Significant (p , 0.05).
250 Research Methodology and Biostatistics

Refer to the tabulated F-value (see Appendix II) for Determine the expected number (E) in each cell
horizontal df (df 54) and vertical df (df 570) at the 0.05 of the table on assumption of null hypothesis.
level of significance. The calculated F-value (6.789) is Column or vertical total  Row or horizontall total
greater than the tabulated F-value (2.52) at 0.05 level of E
Sample total
significance (p 5 0.05). Therefore, null hypothesis is
rejected and it is inferred that there is statistical signifi- Find the difference between the observed and the
cant difference in knowledge score about swine flu expected frequencies in each cell (O – E).
among four different groups of nursing students. Calculate the chi-square value for each cell by the
formula
NONPARAMETRIC TESTS
( O  E )2
Almost nil or very limited assumptions, which are 2 
E
needed to be made about the data format in para-
metric tests, are needed for nonparametric meth- Sum up the x2 values of all the cells to get the total
ods. There are a variety of methods generally used chi-square value.
in different situations; however, some of the com-
monly used methods discussed in this chapter are ( O  E )2
 df2  ∑
chi-square test, sign test, Wilcoxon signed rank E
test and Mann–Whitney U-test.
This indicates the total x2 values at the particu-
CHI-SQUARE TEST lar df indicated at foot of the chi-square.
This is a nonparametric test used to find out the as- Calculate the df from the number of categories in
sociation between two events in binomial or multi- each event.
nomial samples. It is represented by the symbol ‘x2’ Degree of freedom (df) 5 (c – 1) (r – 1); where c
and is used to find out the association between two is the number of vertical columns and r is the
discrete attributes. For example, this test can be used number of horizontal rows.
if one wants to find the relation between smoking in Refer to the x2 table (see Appendix III) to com-
pregnancy and low-birth-weight babies, blood pres- pare the calculated and tabulated x2 values at
sure and renal diseases, obesity and coronary diseases, computed df and specific level of significance such
etc. The two events may have a possibility of affecting as 0.05, 0.02, 0.01 and 0.001. If the tabulated
each other or they may be dependent of one another. x2 value is more than the calculated x2 value, we
This test is also used to find the significance of differ- accept the null hypothesis. On the other hand,
ence in two or more than two proportions. if the tabulated x2 value is less than the calculated
x2 value, we reject the null hypothesis.
Prerequisites of Chi-Square Test For example, compute the chi-square value for
There are four essential requirements to apply the data given in Table 10.3.
x2 test, which are as follows:
l Preferably random sample but not necessarily Null Hypothesis
l Qualitative data measured on nominal or ordi- There is no difference in bowel movements among
nal scale (frequency data, not the means) critically ill patients with positive and negative
l Sample size should be more than 30 24-h fluid balance at 0.02 level of significance.
l Lowest expected frequency not less than 5
Solution
Calculation of Chi-Square Value
Determine the expected number (E) in each cell
Make contingency tables. of the table by using the following formula:
Note the frequencies observed in each class of one
event row-wise and numbers in each group of Column or vertical total  Row or horizontall total
E
the other event column-wise. Sample total
 Measures of Relationship and Inferential Statistics 251

Table 10.3. ​A contingency table showing effect

statistical significant difference in bowel move-
of 24-h fluid balance on bowel movements among ments among critically ill patients with positive
critically ill patients and negative 24-h fluid balance at the 2% level of
significance (p . 0.02).
Bowel 24-h Fluid Balance
Movements Positive Negative Total x2-Value SIGN TEST
Present 403 19 422 The sign test is probably the simplest of all the
Absent 278 20 298 1.696 , a nonparametric methods. This test is utilized in
df 5 1 comparing a single sample with some hypothe-
Total 681 39 720 sized value. Therefore, this test is of use in situa-
a
Not significant (p . 0.02).
tions in which the one-sample or paired t-test
might conventionally be applied. The name sign
test is so given as it attributes a sign, either positive
681 422 39  422 (1) or negative (2), to every observation as per
E1   399.14; E2   22.85
720 720 whether it is lesser or greater than some hypothe-
681 298 39  298 sized value and takes into account whether this is
E3   281.85; E4   16.14
720 720 significantly different from what is expected by
chance. If observations are found to be similar to
Calculate the chi-square value for each cell by the the hypothesized value, they are skipped and
formula: dropped from the sample size.

( O  E )2 Steps in Performing a Sign Test

2 ∑
E The following steps are needed while conducting the
12 
(O1  E1 ) 2


( 403  399.14 ) 2

 0.037
sign test.
l Step I: Mention the hypothesized value
E1 399.114
for comparison and the null hypothesis in
 22 
(O2  E2 )2  (19  22.85)2  0.683 particular.
E2 399.14 l Step II: Attribute a sign (1/2) to every obser-

 32 
(O3  E3 ) 2


( 278  281.85)2
 0.053
vation as per whether it is lesser or greater than
E3 281.85 the hypothesized value. (Observations that are
similar to the hypothesized value are ignored
 42 
(O4  E4 )2  ( 20 16.14 )2  0.923 from the analysis.)
E4 16.14 l Step III: Find out the following:

N1 5 The number of observations that are

Sum up the x2 values of all the cells to get the total greater than the hypothesized value
chi-square value: x2 5 0.037 1 0.683 1 0.053 1 N2 5 The number of observations that are
0.923 5 1.696 lesser than the hypothesized value
Calculate the df from the number of categories in S 5 The smaller of N1 and N2
each event. l Step IV: Determine an appropriate p-value.

Degree of freedom (df) 5 (c – 1) (r – 1); Remember that the sign test only explores the role
df 5 (2 – 1) (2 – 1) 5 1 of chance in describing the relationship; it does
Referring to the chi-square table (see Appendix not give any direct estimate of any effect’s size.
III), at 1 df, the value of chi-square under proba- Although it is sometimes possible to get nonpara-
bility 0.02 is 5.41 and under 0.01 is 6.64. So the metric estimates of effect and the associated CIs in
calculated x2 value is less than the tabulated value principal, the associated methods are complex in
at the 0.02 level of significance. Hence, we accept practice and are not readily available in standard
the null hypothesis and infer that there is no statistical software. This lack of a straightforward
252 Research Methodology and Biostatistics

effect estimate is a significant demerit of nonpara- i.e. lesser than zero (hypothesized values) and
metric methods. the rest eight values were found positive, i.e.
more than zero (hypothesized value).
Solution l Step III: Find out the following:

Apply the sign test on the following data given in N1 5 The number of observations that are
Table 10.4. greater than the hypothesized value 5 8
l Step I: Mention the hypothesized value for com- N2 5 The number of observations that are less
parison in particular and the null hypothesis. than the hypothesized value 5 2
As per the null hypothesis, 6 h of intensive care S 5 The smaller among N1 and N2 5 2
unit (ICU) treatment on SvO2 has no effect. l Step IV: Determine an appropriate p-value.

In contrast, the mean of the differences p 5 0.11 (using the binomial distribution table)
between SvO2 at admission and after 6 h of Refer the binomial distribution probability
admission would be zero under the null hy- table to find out the p-values against n 5 10; if
pothesis. tabulated values are higher than the hypothesized
As per the sign test, almost half of the differ- values, then we accept the null hypothesis;
ences would be above zero (positive), otherwise we reject the null hypothesis.
while the other half would be below zero
(negative). WILCOXON SIGNED RANK TEST
l Step II: Attribute a sign (1/2) to every obser- The sign test is both very simple to perform
vation according to whether it is lesser or greater and intuitive. But it has one demerit that it at-
than the hypothesized value. (Observations that tributes a sign to every observation as per
are similar to the hypothesized value are ig- whether it stays above or below some hypoth-
nored from the analysis.) esized value and does not even consider the
The signs were allocated as tabulated in Table magnitude of the observation. Ignoring infor-
10.4 and only two values were found negative, mation on the magnitude of the observations is
not very efficient and might lessen the statisti-
cal power of the test. In this aspect, it can be
Table 10.4. ​Central venous oxygen saturation at said that a Wilcoxon signed rank test is an al-
the time of admission and 6 h later ternative that takes into account the magnitude
of the observations.
SvO2 (%)
On Steps in Conducting the Wilcoxon Signed Rank Test
Patients Admission After 6 h Difference Sign The Wilcoxon signed rank test comprises five
1 56.0 59.5 3.5 1 basic steps.
2 57.7 71.4 13.7 1 l Step I: Mention the hypothesized value for com-

3 37.8 50.0 12.2 1 parison in particular and the null hypothesis.

l Step II: Ignoring their signs, rank all the
4 33.6 51.3 17.7 1
observations in increasing order of magnitude.
5 58.2 60.7 2.5 1 Moreover, ignore observations that are similar
6 59.1 56.7 22.4 2 to the hypothesized value. If two observations
7 65.3 59.8 25.5 2 have similar magnitude, they are provided an
8 39.7 52.9 13.2 1 average ranking, irrespective of their signs.
l Step III: Attribute a sign (1/2) to each obser-
9 60.6 67.7 7.1 1
vation as per whether it is greater or lesser than
10 56.1 61.9 5.8 1 the hypothesized value (as in the sign test).
 Measures of Relationship and Inferential Statistics 253

l Step IV: Determine Step III: Attribute a sign (1/2) to each observa-
a. R1 5 Addition of all positive ranks tion as per whether it is greater or lesser than
b. R2 5 Addition of all negative ranks the hypothesized value (as in the sign test).
c. R 5 Less than R1 and R2 Signs are allocated (see the Table 10.5).
l Step V: Determine an appropriate p-value. Step IV: Determine
a. R1 5Addition of all positive ranks 5 2 1 3
Solution 1 5 1 6 1 7 1 8 1 9 1 10 5 50
Apply the Wilcoxon signed rank test on the fol- b. R2 5Addition of all negative ranks 5 1
lowing data given in Table 10.5. 1455
Step I: Mention the hypothesized value for com- c. R 5 Less than R1 and R2 5 05
parison, in particular, and the null hypothesis. Step V: Determine an appropriate p-value.
As per the null hypothesis, there is no consequence Similar to the sign test, a p-value for a small sample
of 6 h of ICU treatment on SvO2. Speaking dif- size such as this can be achieved from tabulated val-
ferently, under the null hypothesis, the mean of ues like those shown in Table 10.6. With respect to
the differences between SvO2 at admission and the critical values for p 5 0.10 and p 5 0.05, the
that after 6 h of admission would be zero. calculated value of R (i.e. 5) is either less or equal, but
Step II: Give rank to all observations in increasing is more than that for p 5 0.01. Therefore, it can be
order of their magnitude, without taking into mentioned that the p-value lies between 0.01 and
account their signs. Do not take into consider- 0.05. Speaking differently, there is some proof that
ation any observations that are similar to the suggests that there is a difference between admission
hypothesized value. When two observations and 6 h of SvO2 beyond that expected by chance.
have similar magnitude, irrespective of their
signs, they are provided an average ranking. MANN–WHITNEY U-TEST
Ranks are attributed (see Table 10.5). Both the sign test and the Wilcoxon signed rank test
are helpful nonparametric alternates to the one-sam-
ple and paired t-tests. A nonparametric alternate to
the unpaired t-test is provided by the Wilcoxon rank
Table 10.5. ​Central venous oxygen saturation on sum test, which is also known as the Mann–Whitney
admission and 6 h after admission U-test. This is generally considered when comparison
is done between two independent groups.
SvO2 (%)
On Steps in Conducting the Mann–Whitney U-Test
Patient Admission After 6 h Difference Rank Sign
This is the same as that of the Wilcoxon signed
2 59.1 56.7 22.4 1 2 rank test and comprises the following three steps:
7 58.2 60.7 2.5 2 1 l Step I: Give rank to all observations in increas-

9 56.0 59.5 3.5 3 1 ing order of their magnitudes by ignoring which

10 65.3 59.8 4
group they come from. If two observations are
25.5 2
3 56.1 61.9 5.8 5 1
5 60.6 67.7 7.1 6 1
6 37.8 50.0 12.2 7 1 Table 10.6. ​Wilcoxon signed rank test critical
1 39.7 52.9 13.2 8 1 values (with a sample size of 10)
4 57.7 71.4 13.7 9 1 p-value 0.10 0.05 0.01
8 33.6 51.3 17.7 10 1 Critical value 10 8 3
254 Research Methodology and Biostatistics

of the same magnitude, irrespective of their have been taken from the group of patients
groups, they are provided an average ranking. who stayed for around 3–5 days in ICU. The
l Step II: Sum up the ranks in the smaller of the total dose of propofol given to each patient is
two groups (S). If the two groups are similar in ranked by increasing magnitude, irrespective
size, then either of them can be chosen. of the patient being either in the protocolized
l Step III: Determine an appropriate p-value. or nonprotocolized group. Remember that
two patients had total doses of 21.6 g, and
Solution these are attributed an equal, average ranking
Apply the Mann–Whitney U-test on the following of 7.5.
data given in Table 10.7. l Step II: Sum up the ranks in the smaller of the
l Step I: Give rank to all observations in increas- two groups (S). If the two groups are equal in
ing order of their magnitudes, without consid- size, then either of them can be chosen.
ering which group they have come from. If two There were a total of 9 protocolized patients and
observations have similar magnitude, irrespec- 11 nonprotocolized, and the addition of the
tive of their groups, they are provided an aver- ranks of the smaller protocolized group (S) is
age ranking. 84.5.
A pilot study was chosen to determine whether l Step III: Determine an appropriate p-value.
protocolzing sedative administration lessened For a small sample such as this, a p-value can be
the total dose of propofol administered. Pa- determined from tabulated values. The two
tients were segregated into groups on the basis individual sample sizes are utilized to identify
of their duration of stay. The data given here the appropriate critical values in this case. As
shown in Table 10.8, these are expressed in
terms of a range. The range in each case
shows the addition of the ranks, outside
Table 10.7. ​Total propofol doses in patients with a which the calculated statistic S must fall to
3- to 5-day stay in ICU reach that level of significance. In other words,
Nonprotocolized Group Protocolized Group S must either be less than or equal to 68 or
greater than or equal to 121 for a p-value be-
Dose (g) Rank Dose (g) Rank
low 0.05. In this case, p is greater than 0.05 as
7.2 2 5.6 1 S 5 84.5. This test renders no proof to sup-
15.7 4 14.6 3 port the notion that the group who received
19.1 6 18.2 5 protocolized sedation got lower total doses of
21.6 7.5 21.6 7.5 propofol beyond that expected through
26.8 10 23.1 9
chance. p-values can be determined using a
normal distribution for S for larger sample
27.4 11 28.3 12
sizes (greater than 20 or 30).
28.5 13 31.7 14
32.8 16 32.4 15
36.3 17 36.8 18 Table 10.8. ​Critical values for the Wilcoxon rank
43.2 19 sum test (with sample sizes of 9 and 11)
44.7 20 p-value 0.05 0.01 0.001
S 5 84.5 Critical value 68–121 61–128 53–136
 Measures of Relationship and Inferential Statistics 255

MULTIPLE CHOICE QUESTIONS

1. Correlation is a procedure used to determine if: 6. Which of the following statements is true concern-
(a) X comes before Y ing correlation?
(b) X causes Y (a) A correlation is always between 22 and 2; a zero
(c) X and Y are different value indicates no clustering towards the line
(d) X and Y vary together (b) A correlation is always between 22 and 2; a
zero value indicates all points on the line
2. The value of a correlation is reported by a researcher to be (c) A correlation is always between 21 and 1; a zero
r 5 20.5. Which of the following statement is correct? value indicates no clustering towards the line
(a) The x variable explains 25% of the variability in (d) A correlation is always between 21 and 1; a
the y variable. zero value indicates all points on a horizontal
(b) The x variable explains 225% of the variability line
in the y variable.
(c) The x variable explains 50% of the variability in 7. _______________ is the set of procedures used
the y variable. to explain or predict the values of a dependent
(d) The x variable explains 250% of the variability variable based on the values of one or more
in the y variable. independent variables.
(a) Regression analysis
3. What is the effect of an outlier on the value of a
(b) Regression coefficient
correlation coefficient?
(c) Regression equation
(a) An outlier will always decrease a correlation
(d) Regression line
coefficient.
(b) An outlier will always increase a correlation co-
8. Which of the following correlation coefficients
efficient.
would indicate the strongest relationship between
(c) An outlier might either decrease or increase a
anxiety and the ability to concentrate?
correlation coefficient, depending on where it is
(a) r 5 0.30
in relation to the other points.
(b) r 5 20.30
(d) An outlier will have no effect on a correlation
(c) r 5 20.60
coefficient.
(d) r 5 21.30
4. One use of a regression line is to:
9. Which per cent confidence interval will be the
(a) Determine if any x-values are outliers.
widest (i.e. the least precise) for a particular data set
(b) Determine if any y-values are outliers.
that includes exactly 500 cases?
(c) Determine if a change in x causes a change in y.
(a) 99%
(d) Estimate the change in y for a one-unit change in x.
(b) 95%
5. Past data have shown that the regression line relating (c) 90%
the final exam score and the midterm exam score for (d) None of the above
students who take statistics from a certain professor is
Final exam 5 50 1 0.5 3 Midterm 10. To analyse the relationship between blood pressure
Which of the following is the interpretation of the slope? and height in a sample of students, we could use:
(a) A student who scored 0 on the midterm would (a) Paired t-test
be predicted to score 50 on the final exam. (b) Correlation coefficient
(b) A student who scored 0 on the final exam would (c) Chi-square test
be predicted to score 50 on the midterm exam. (d) Two sample t-test
(c) A student who scored 10 points higher than
another student on the midterm would be pre- 11. Which of the following is a nonparametric test?
dicted to score 5 points higher than the other (a) ANOVA
student on the final exam. (b) Student’s t-test
(d) Students only receive half as much credit (0.5) (c) Chi-square test
for a correct answer on the final exam compared (d) Z-test
to a correct answer on the midterm exam.
256 Research Methodology and Biostatistics

12. Which of the following is a true statement about 16. A type-II error is also known as _______________.
confidence intervals? (a) False-positive
(a) The intervals give a range of values within (b) False-negative
which the true value will lie. (c) Double negative
(b) 95% Confidence interval means that 95% of all (d) Positive negative
observed values fall within that interval.
(c) They indicate the presence or otherwise of a 17. The use of the laws of probability to make infer-
statistical difference between two groups. ences and draw statistical conclusions about popu-
(d) The intervals are larger with smaller sample size. lations on the basis of sample data is known as
_______________.
13. Which of the following is a true statement about (a) Sample statistics
nonparametric tests? (b) Descriptive statistics
(a) They cannot be used if the nature of the distri- (c) Population statistics
bution of the data is unknown. (d) Inferential statistics
(b) Student’s paired t-test is a nonparametric test.
(c) They are used to analyse samples that are not 18. Which of the following is a parametric statistical
normally distributed. test?
(d) They can be used on very small samples. (a) Chi-square test
(b) ANOVA
14. _______________ results if you fail to reject the (c) Wilcoxon signed rank test
null hypothesis when the null hypothesis is actually (d) Mann–Whitney U-test
false.
(a) Type-I error 19. Which among the followings depicts the values that
(b) Type-II error mark the boundaries of the confidence interval?
(c) Type-III error (a) Margin of error
(d) Type-IV error (b) Levels of confidence
(c) Confidence intervals
15. When you calculate a 95% confidence interval, (d) Confidence limits
what does it mean?
(a) You can be ‘95% confident’ that your interval 20. While computing chi-square test for a 2 3 3 contin-
will include the population parameter. gency table, what will be the degree of freedom?
(b) You can be ‘5% confident’ that your interval (a) One
will not include the population parameter. (b) Two
(c) The process you used will capture the true pa- (c) Three
rameter 95% of the time in the long run. (d) Four
(d) All of the above statements are true.

Answers of Multiple Choice Questions

1. d; 2. a; 3. c; 4. d; 5. c; 6. d; 7. a; 8. c; 9. a; 10. b; 11. c; 12. a; 13. c; 14. b; 15. d; 16. b; 17. d; 18. b; 19. d;
20. b.

REFERENCES
Altman, D. G. (1991). Practical statistics for medical research. Armitage, P. (1989). Statistical methods in medical research.
London: Chapman & Hall. London: Blackwell Scientific.
Altman, D. G. (1992). Practical statistics for medical research. Armitage, P., & Berry, G. (1989). Statistical methods in medical
London: Chapman & Hall. research. London: Blackwell Scientific.
Altman, D. G. (1992). Practical statistics for medical research. Ashish, B., Devendra, B. G., & Gaurisankar, R. (1977). Population
London: Chapman & Hall. statistics in India. New Delhi: Vikas Publishing.
 Measures of Relationship and Inferential Statistics 257

Ashish, B., Devendra, B. G., & Gaurisankar, R. (1977). Population Mahajan B.K. (2010). Methods in Biostatistics for Medical Students
statistics in India. New Delhi: Vikas. and Research Workers. (7th ed.). New Delhi: Jaypee Brothers
Bahn, A. K. (1972). Basic medical statistics. New York: Grune & Medical Publishers Ltd.
Sratton. Mathew, D. E. (1988). Using and understanding medical
Bahn, A. K. (1972). Basic medical statistics. New York: Grune & statistics. New Delhi: Karger.
Stratton. Mathew, D. E., & Farewell, V. T. (1988). Using and understanding
Bland, M. (1989). An introduction to medical statistics. London: medical statistics. New Delhi: Karger.
ELBS. Muzumdar R.D., Kulkarni A.P. & Baride J.P. (2003). Manual of
Bland, M. (1989). An introduction to medical statistics. London: biostatistics (1st ed.). New Delhi: Jaypee.
ELBS. Muzumdar, R. D., Kulkarni, A. P., & Baride, J. P. (2003). Manual of
Bourke, G. J., & McGilvary, J. (1985). Interpretation and uses of biostatistics (1st ed.). New Delhi: Jaypee.
medical statistics. London: Blackwell Scientific. Neave, H. R. (1981). Elementary statistics tables. London:
Bourke, G. J., Daly, L. E., & McGilvary, J. (1985). Interpretation and Routledge.
uses of medical statistics. London: Blackwell Scientific. Negi, K. S. (2010). Biostatistics (2nd ed.). New Delhi: AITBS.
Daniel, W. W. (1987). Biostatistics: a foundation for analysis in the Negi, K. S. (2010). Biostatistics (2nd ed.). New Delhi: AITBS.
health sciences. New York: Wiley. Rao Sunder P.S.S. (2012). Introduction to biostatistics and
Daniel, W. W. (1987). Biostatistics: a foundation for analysis in the research methods (5th ed.). New Delhi: PHI Learning.
health sciences. New York: Wiley. Rao, S., & Richard, J. (2006). Introduction to biostatistics and
Das, R., & Das, P. N. (2009). Instant medical biostatistics. New research methods (4th ed.). New Delhi: Prentice-Hall.
Delhi: Ane Books. Rao, S., & Richard, J. (2006). Introduction to biostatistics and
Das, R., & Das, P. N. (2009). Instant medical biostatistics. New research methods (4th ed.). New Delhi: Prentice-Hall.
Delhi: Ane Books. Rees, D. G. (1989). Essentials of statistics. London: Chapman & Hall.
Dunn, O. J. (1984). Basic statistics. New York: Wiley. Reid, N. G. (1987). Research methods and statistics in healthcare.
Dunn, O. J. (1984). Basic statistics. New York: Wiley. London: Edward Arnold.
Gupta S.P. (2008). Statistical methods (1st ed.). Delhi: Sultan Reid, N. G., & Boore, J. (1987). Research methods and statistics in
Chand. healthcare. London: Edward Arnold.
Gupta, S. P. (2008). Statistical methods (1st ed.). Delhi: Sultan Ress, D. G. (1989). Essentials of statistics. London: Chapman
Chand. & Hall.
Kirkwood, B. R. (1988). Essentials of medical statistics. London: Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the
Blackwell Scientific. behavioural sciences (2nd ed.). New York: McGraw-Hill.
Kirkwood, B. R. (1988). Essentials of medical statistics. London: Sprent, P. (1989). Applied nonparametric statistical methods.
Blackwell Scientific. London: Chapman & Hall.
Kirkwood, B. R. (1988). Essentials of medical statistics. Oxford: Sprent, P. (1989). Applied nonparametric statistical methods.
Blackwell Science. London: Chapman & Hall.
Kish, L. (1987). Statistical designs for research. New York: Wiley. Sunder Rao, P. S. S. (2012). Introduction to biostatistics and
Kish, L. (1987). Statistical designs for research. New York: Wiley. research methods (5th ed.). New Delhi: PHI Learning.
Lancaster, H. O. (1974). Introduction to medical statistics. New Woolson, R. F. (1987). Statistical methods for the analysis of
York: Wiley. biomedical data. New York: Wiley.
Lancaster, H. O. (1974). Introduction to medical statistics. New Woolson, R. F., ed. (1987). Statistical methods for the analysis of
York: Wiley. biomedical data. New York: John Wiley & Sons.
Writing Proposals, Reports
and Publications
CHAPTER OUTLINE
11
INTRODUCTION WRITING SCIENTIFIC ARTICLES

CRITERIA FOR COMMUNICATING RESEARCH CRITICAL REVIEW OF RESEARCH WORK

METHODS OF COMMUNICATING RESEARCH DEVELOPING AND PRESENTING A RESEARCH PROPOSAL

WRITING A RESEARCH REPORT USES OF COMPUTER IN MEDICAL RESEARCH

FORMAT OF A THESIS OR DISSERTATION MULTIPLE CHOICE QUESTIONS

WRITING REFERENCES/ BIBLIOGRAPHY

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Prepare correct citations and references for research
• Describe the criteria and methods of communicating reports.
research reports. • Describe the process of development and presentation
• Discuss the format and characteristics of a good research of research proposals.
report including thesis/dissertation.
• List the necessary elements and guidelines for research
critique.

KEY TERMS

• Bibliography 265 • References 265 • Research report 261

• Communicating research 259 • Research article 274 • Thesis/dissertation 263
• Oral reports 260 • Research critique 275 • Utilization of research 286
• Poster presentation 260 • Research proposal 279 • Written reports 260

INTRODUCTION empirical research evidence generated through

A research project cannot be considered com- research study. The dissemination of research
plete until its results are effectively communi- findings is achieved through either written or
cated to its users and consumers. Therefore, oral means. In written means of communication,
communication of research findings is one of the the researcher writes a detailed description of the
essential and final steps of the research process. whole research process, in the form of theses,
Communication of the research findings is car- dissertations, research articles, scientific papers,
ried out through the dissemination of the etc., while oral communication of the research
 Writing Proposals, Reports and Publications 259

findings is achieved through the presentation of will be the leading author and the contrib-
the final results and effects of the research pro- uting authors. The International Commit-
cess to a group of people in a professional scien- tee of Medical Journal Editors (ICMJE)
tific conference through either an oral scientific advices that authorship credits should
paper or poster presentation. Traditionally, oral be given only to those who have made a
presentation has been used most frequently, but substantial contribution to the conceptual
in the last 15–20 years, posters and computer designs of the study, data analysis and in-
slide presentations have increased in popularity. terpretations, drafting or revising of the
Effective communication of the research report manuscript and approving of the final ver-
helps in the following ways: sion of the manuscript.
l Promotion of learning of new knowledge  Deciding about the content: Researchers also

among professionals have to decide how many papers are required

l Expansion of the base for evidence-based prac- to communicate the findings of the research
tices effectively, as sometimes researchers collect a
l Improving health care outcomes or client care huge amount of data that cannot be commu-
outcomes by refining the existing body of pro- nicated in a single paper.
fessional knowledge  Preparing outline of the report: It is impor-

tant that if there are multiple authors of a re-

port, each one has responsibility for different
CRITERIA FOR COMMUNICATING
sections of the manuscript. The advantage of
RESEARCH having an outline is that it can be incorpo-
In order to share or report research findings, the rated into a timeline that sets goals for com-
researcher has to follow some criteria, which are as pleting the manuscript.
follows:  Assemble the material needed to begin a draft

l Selecting proper channel for communicating: and finally start preparing a report with the
The researcher has to select suitable and appro- outlined timeline.
priate channel to disseminate his or her research l Careful selection of a journal for publica-
findings, such as student-related outlets like tion of the report: In selecting a journal for
theses and dissertations; however, professional publication, some important factors must be
academicians generally need publication of re- kept in mind, such as the journal’s goal, audi-
search articles in professional journals, books or ence, its prestige, acceptance and frequency of
oral research presentations in conference publication.
through papers or slides. l Careful review of the author’s manuscript
l Knowing the consumers: Researchers must guidelines: Researchers need to be careful to
know in advance to whom they want to com- review the journal instructions to authors be-
municate their research findings, such as cause each journal has its own guidelines for
health care research findings to health care submission of manuscripts. Most of the jour-
professions, health care educators, health care nals limit manuscripts to 15–20 double-spaced
administrators or even some times general typed pages.
public. This will help in choosing the right
method, mode and content of research to be
METHODS OF COMMUNICATING
communicated.
l Developing an effective plan for writing a
RESEARCH
research report: A plan for writing research in- Communication of research findings must be car-
cludes the following aspects: ried out through methods that are fast and have a
 Decide on authorship: Researchers have wide range of coverage to disseminate the research
to decide among themselves as to who evidence among professionals. Generally, research
260 Research Methodology and Biostatistics

findings are disseminated through the following this problem, oral reports must be supplemented
methods: with the availability of the written reports of the
l Written reports: A written report is considered oral presentation in advance in the form of con-
as the best method for disseminating research ference abstract publications. Oral reports are
findings because this communication medium presented in professional conferences, where the
has the potential to reach the largest percentage researcher is given about 15–30 min to present
of consumers. Health care research is a method his or her findings, with a few minutes for dis-
of building knowledge, and publication is the cussion. Before the conference, authors are in-
medium for sharing this knowledge. In addi- vited to submit the research abstract in a given
tion, the growth of different health care profes- format. However, poster and multimedia pre-
sions largely depends on documented evidence sentations in conferences are a recent trend,
for the expansion and advancement of new pro- where researchers present brief research findings
fessional knowledge. Written reports could be on posters or slides, which are displayed in open
theses, dissertations, published scientific articles halls during conferences. Conference partici-
in journals or periodicals. Publishing a research pants are given special time to visit the posters
article in a scientific journal involves authoring and the researcher must be there to clarify the
a research article in a format suggested for doubts of the participants.
the manuscript and sending it to an appropri- l Poster presentation: Poster presentations at sci-
ate professional journal for publication (see entific conferences have become a popular and
checklist for publishing research results in efficient way to present research findings. Tradi-
Table 11.1). tionally, oral presentations were the primary
l Oral reports: Oral research reports are another method of research dissemination at confer-
popular and useful method for the dissemina- ences; the results of only a few studies could be
tion of knowledge among users. Oral reports are presented in this form due to limitations in
generally written and presented to a group of time. Poster sessions make it possible for a huge
professionals at conferences, which can be ei- number of researchers to present their findings
ther read out or presented through poster or at conferences. The poster is a visual graphic
computer slide presentation. Presenting re- presentation of study results that can be exam-
search results at a conference has advantages ined by an audience with or without interaction
over publishing a scientific research article in a with the investigator. The same information
professional journal. Firstly, oral presentation presented in oral form can be included in the
provides an opportunity to present the recent study poster. A poster follows the abstract, out-
findings almost immediately, while publishing a line, purpose of the research project, methods,
research article in a journal may take time. Sec- results and conclusions. The results section of
ondly, audiences get an opportunity to interact the poster should make use of tables, graphs and
with the researcher(s) and clarify their doubts if other visual illustrations. Ideally, the poster
any. However, there is a lack of permanent re- should be visually interesting enough to attract
cord of the oral reports; therefore, to overcome an interested audience. It should be colourful

Table 11.1. Checklist for publishing research results

• Does the research topic fit the mission of the journal?
• Is the impact factor appropriate for the author?
• Is the journal well known in your area of research?
• Is the journal highly accessible to a large number of its audience?
• Is the reviewing timelines appropriate?
• Is there an issue about cost for publishing?
 Writing Proposals, Reports and Publications 261

and clearly laid out and should not contain too information leads to an overall improvement
much text. The information from purpose to in knowledge and practice among health care
implications should be presented from left to providers.
right on the poster. Some scientific conferences
provide guidelines for the presentation of post- CHARACTERISTICS OF A GOOD RESEARCH
ers in terms of poster size and display format. REPORT
Some important suggestions for successful Research reports are an effective means of com-
poster presentation are as follows: municating research findings to readers and inter-
 The choice of fonts, font size and layout are ested audiences. A well-written research report is
essential in attracting an audience. Colleagues one which effectively, efficiently and widely dis-
can be asked to critique the poster as if they seminates research information among its users. A
are walking through the poster session. good research report has the following character-
 Microsoft PowerPoint and laminated poster istics (Table 11.2):
paper are the best tools and materials for cre- l A research report must have the characteristics

ating a poster. Suggested formats for laying of conciseness, clarity, honesty, completeness
out material can also be downloaded from and accuracy.
numerous sites and used as a guide. l A research report must be long enough to cover

 During the poster session time, the author the subject content and short enough to main-
should be available, standing or sitting near tain interest among its users and consumers.
the poster, to explain the findings and answer l It must be written and presented logically so

questions for the interested audience. Hand- that different pieces of the research process fit
outs of key results and contact information effectively.
can be distributed as well. l Abstract terminology and technical jargons

must be avoided in the research report.

l Presentation of the research report must be
WRITING A RESEARCH REPORT lucid and visually attractive, so that it can be
A research report is the ultimate outcome of the interesting to its users.
research process. A research process cannot be l Research report must reflect its originality.

considered complete until its report has been writ-

ten and disseminated. Writing a report highlights FORMAT OF A RESEARCH REPORT
the research project and helps in the dissemination Format of research reports varies from institution
of the research findings. Dissemination of research to institution and university to university. How-
findings serves scientific, professional and public ever, the basic contents of the reports remain
functions. An effective dissemination of scientific nearly the same. Quantitative research reports

Table 11.2. Tips for writing good research report

• Use short, simple sentences
• Write in active voice
• Use headings and subheadings to guide the readers
• Use bold type in headings to attract attention
• Use tables, graphs to break up print pages
• Use left and right margin justification
• Make sure there are good transitions between sections
• Follow directions as stated in the guidelines for writing report
• Have peer review of the research report, if possible
• Be ready to edit the proposal many times
262 Research Methodology and Biostatistics

typically follow a conventional format referred to and inferences after the rejection or acceptance
as IMRAD format. It involves the organization of of the hypotheses.
the study material into four sections: introduc- l Discussion: A typical discussion section addresses
tion, methods, results and discussion. the following questions:
l Introduction: It is often written in a funnel-  What were the main findings?

shaped structure. The researcher begins broadly  What do the findings mean?

to establish a framework to understand the  How do the results compare with prior

study and narrows it down to the specifics of knowledge on the topic?

what he or she intended to learn. It should in-  What can be concluded about the findings

clude the clear and concise problem statement, vis-à-vis their use in health care practices
its significance, existing literature, conceptual and health care theories and in future health
framework of study, research questions, hy- care research?
potheses, underlying assumptions and the need l Conclusion: Conclusion includes the para-
for the study. So, in a nutshell, introduction graphs that summarize what was done and
generally includes the following: found in the present study as precise as possible.
 Background of the study l Other aspects of the report
 Need for the study  Title: It should include the variables under

 Research problem study, study population and place of the study.

 Objectives The title should be clear and concise, ideally
 Hypothesis/assumptions not more than about 15 words.
 Operational definitions  Abstract: It is an unstructured paragraph of

 Conceptual framework 100–200 words or in a structured form with

 Literature review probably a subhead. It should be written
l Materials and methods: The methods section clearly so that a reader can decide whether to
is often subdivided into several parts, which read the entire report or not.
include stages, such as:  Keywords: Substance, methodology and theo-

 Research design retical terms can be used as keywords; these

 Research setting keywords are used in the index to help others
 Target population locate the study.
 Sampling technique and sample size  References: The reference style should be se-

 Development and description of data collection lected after reviewing the manuscript or re-
tools port. Usually, Vancouver reference style is
 Validity and reliability of research tools used in health sciences literature.
 Methods and procedures of data collection  Acknowledgement: Those people whose con-

 Pilot study and tool tryout tribution does not qualify them for author-
 Feasibility of the study ship are sometimes acknowledged at the end
 Ethical considerations of the report or the beginning of the report.
 Plan for data analysis

 Plan for reference writing STEPS IN WRITING A REPORT

l Results: It is the heart of a research report. If Research reports are the products of slow, pains-
both descriptive and inferential statistics have taking and accurate inductive work. The steps in-
been used, then results start with descriptive volved in writing a report include the following:
statistics to present sample characteristics. l Logical analysis of the subject matter: It is the

Later, analysed data are presented in accor- first step which is primarily concerned with the
dance with study objectives using descriptive development of a subject.
and inferential statistics. Inferential statistics l Preparation of the final outline: It is the next

are generally used to test the preset hypotheses step in writing the research report. ‘Outlines are
 Writing Proposals, Reports and Publications 263

the framework upon which works are con-

FORMAT OF A THESIS
structed. They are an aid to the logical organiza-
tion of the material and a remainder of the OR DISSERTATION
points to be stressed in report.’ A description of the layout and format of a scien-
l Preparation of the rough draft: This step is of tific thesis or dissertation, with all its parts and
utmost importance for researchers as it involves subparts, is given in this section. However, the
actually writing down what they have done in format may slightly vary from one institution to
the context of their research studies. The re- another or university to university.
searchers write down the procedures adopted by
them in collecting the material for their study PRELIMINARY PAGES
along with various limitations, the techniques This section includes the following:
of analysis adopted by them, the broad findings l Cover page

and generalizations, and the various suggestions

they want to offer regarding the problems en- Title of the Research Study
countered. Thesis submitted for the partial fulfilment of the
l Rewriting and polishing the rough draft: This requirement for the degree in
step happens to be the most difficult part of all
formal writing. Careful revision makes the differ- Doctor of Philosophy
in Neurosciences
ence between a mediocre and a good piece of writ- of the AIIMS, Delhi
ing. While rewriting and polishing, one should
check the report for weaknesses in logical develop- Year
ment or presentation. The researcher should Name of the candidate
check the mechanics of writing – grammar, spell- Name of the department and institution
ing and usage.
l Preparation of final bibliography: Next in the l Certification of work from the guide/
order comes the task of preparing the final bib- supervisor and principal
liography. It is a list of books, journal articles,
papers presented, etc. that have contributed or Certificate of Supervisor
are in some way are pertinent to the research This is to certify that Ms/Mr X …… has carried out
which has been done. The bibliography should study titled ‘ … … ‘. This is the original work of
be arranged alphabetically and may be divided the above-said person and is conducted with full hon-
into two parts: the first part may contain the esty and compassion.
names of books and journals, and the second
part may contain the names of magazines and Name of guide/supervisor
Qualification
newspaper articles. Designation
l Writing the final draft: This constitutes the
last step. The final draft should be written in a Name of the Principal/Dean
concise and objective style and in simple lan- Qualification
guage, avoiding vague expressions, such as ‘it Address
seems’ and ‘there may be’. While writing the
final draft, the researcher must avoid abstract l Acknowledgement: Acknowledgement must
terminology. It must be remembered that ev- include the citation of the words of gratitude
ery report should be an attempt to solve some and appreciation for the people who have con-
intellectual problem and must contribute to tributed to the compilation of the research task
the solution of a problem; it must add to the including supervisor, co-supervisor, HOD,
knowledge of both the researcher and the other faculty members, ethical committee, other
reader. authorities who have facilitated the access of the
264 Research Methodology and Biostatistics

data collection sources, study participants, study is based. It explains the way in which
friends, family members and other people who the study relates to larger issues and uses a
have directly and indirectly facilitated the con- persuasive rationale to justify this study. It
duction of the research study. makes the purpose worth pursuing. The sig-
l Index/table of contents: This must include the nificance of the study answers the questions:
sequence of the contents placed in the research - Why is your study important?
report (thesis/dissertation) along with the page - To whom is it important?
numbers. This subsection facilitates the reader - What is the benefit(s) of carrying out your
to have easy access of the desired information study?
from the research report without wasting time.  Research problem/questions: This section in-

l List of tables: This includes a list of the head- cludes the statement of the research problem
ings of the tables in the sequence they are placed or research question. For example: ‘A descrip-
in the thesis/dissertation. This subsection facili- tive study on the prevalence of pin site infec-
tates the easy location of the information and tion among patients with external skeletal
data as per the need of the reader. fixation admitted in the trauma ward of
l List of figures: This includes a list of the head- Nehru Hospital, PGIMER, Chandigarh’.
ings of figures in the sequence they are placed in  Objectives: This section includes the state-

the thesis/dissertation. This subsection facili- ments of action and outcome which the re-
tates the easy location of the information and searcher wants to achieve during the research
data as per the need of the reader. activity. For example:
l List of abbreviations: This includes the descrip- - To assess the prevalence of pin site infec-
tion and details of the abbreviations used in the tion among patients with external skeletal
thesis/dissertation. This subsection provides reader fixation
information about the abbreviations in advance, so - To determine the relationship of pin site
that a smooth reading can be facilitated. infection with selected biophysiological
characteristics of the patients with external
MAIN TEXT skeletal fixation
This section includes the following:  Hypotheses or assumption: Each research is

l Chapter I: Introduction (background of the based on a hypothesis or an assumption,

study) which researchers have to test through their
 Introductory paragraphs (generally 2–5 pages): research findings.
The first chapter must include a few paragraphs  Scope and delimitation: All research studies

about the background of the study, statement of have limitations and a finite scope. Limita-
the problem and purpose of the study. The pri- tions are often imposed by time and budget
mary goal of the introductory paragraphs is to constraints. Precisely list the limitations of the
attract the attention of the readers and to get study. Describe the extent to which you be-
them interested in the subject. It sets the stage lieve the limitations degrade the quality of the
for the paper and places the topic in perspective. research.
The introduction often contains dramatic and  Operational definitions: This section involves

general statements about the need for the study. the operationally defined terms used in the
It uses dramatic illustrations or quotes to set the research study in the manner in which the
tone. When writing the introduction, the researcher is going to study the variables.
author(s) must think from the readers’ perspec- l Chapter II: Literature review
tives, i.e. would the target audience continue This chapter is important because it shows what
reading the text after the first chapter? previous researchers have explored and discov-
 Significance or need of the study: This section ered about the phenomenon under study. It is
lays the foundation on which the research usually quite long and primarily depends upon
 Writing Proposals, Reports and Publications 265

how much research has previously been done  The present recommendations should be
in the area you are planning to investigate. If based on your findings. Avoid the temptation
you are planning to explore a relatively new to present the results based on your own be-
area, the literature review should cite similar liefs or biases that are not specifically sup-
areas of study or studies that lead up to the ported by your data. Recommendations fall
current research. Never say that your area is so into two categories. The first is recommenda-
new that no research exists. It is one of the key tions to the study sponsor. What are the ac-
elements that proposal readers look at when tions that you recommend they should take
deciding whether or not to approve a proposal. based upon the data? The second is recom-
l Chapter III: Methodology (materials and mendations to other researchers. There are
methods) almost always ways that a study could be im-
This section includes the following elements: proved or refined. What would you change if
- Design of the research study you were to conduct your study over again?
- Research setting These are the recommendations to other
- Target population researchers.
- Sampling technique and sample size
- Development and description of research END MATTER
instrumentation (include copy in appendix) l References: References may be written in the
- Validity and reliability of research tool(s) Harvard, American Psychological Association
- Procedure and time frame of data collection (APA), Campbell or Vancouver style. However,
(may present as an algorithm) in health sciences, Vancouver and Harvard styles
- Pilot study of referencing are commonly used. Details of
- Feasibility of the study Harvard, APA and Vancouver style of referencing
- Ethical considerations are discussed separately in this chapter, and they
- Analysis plan (state critical alpha level and may be referred to for more information.
type of statistical tests expected to be used) l Bibliography: Bibliography is a list of all the
l Chapter IV: Analysis and interpretation of materials that have been consulted while con-
data (results) ducting a research study or writing an academic
This section presents the description of the article, paper or book based on research. Refer-
study sample, and analysis and interpretation ences, on the other hand, are those that have
of the data through descriptive and inferential been directly referred to and referenced in your
statistics; the data are usually presented research report, article or book.
through tables, graphs, etc. l Appendix/annexure: At the end of the report,
l Chapter V: Discussion appendices should be written to account for all
The discussion section includes the explanation the technical data, such as research instruments
of findings, where the researcher presents his (e.g. questionnaire, interview schedule), sample
or her findings through critical analysis along information and mathematical derivations.
with a comparison with other similar research
findings. This section also presents the verdict
WRITING REFERENCES/
on whether your findings support existing
theories. Explain the results and present pos- BIBLIOGRAPHY
sible reasons why the results might have References or bibliography may be written using
turned out the way they did. the American Psychological Association (APA),
l Chapter VI: Conclusions and recommenda- Modern Language Association (MLA), Chicago
tions Manual of Style format, Campbell, Vancouver and
 Conclusion includes the few paragraphs that Harvard styles. However, in health sciences, the
summarize what you did and found. Vancouver style of references is commonly used;
266 Research Methodology and Biostatistics

this chapter includes a detailed discussion on the A bibliography may contain all the items that have
Vancouver style of referencing. A brief description appeared in the reference list, plus some addi-
with examples of the Harvard style of referencing tional works.
is also given in Appendix XIII. Both bibliography and references are arranged
alphabetically. But a reference list can also be ar-
DIFFERENCES BETWEEN REFERENCES ranged in numeric style, which means arranging
AND BIBLIOGRAPHY the references chronologically according to the
Most people are not aware of the differences be- numbers in the text.
tween bibliography and references. They often While writing a bibliography, you should have
make the mistake that the two are the same. How- to include the authors last and first name, year of
ever, they are different and used in different con- publication, name of the book or journal, place of
texts in each essay, article or book. publication and name of publishers. In case of
Bibliography is a listing of all the materials that websites, organizations as authors or other such
have been consulted while writing an article or a sources, additional information should be pro-
book. References, on the other hand, are those that vided for the convenience of the readers. Therefore,
have been referred to or referenced in your article to sum up, the differences between bibliography
or book. and references are as follows:
You might have consulted a lot of books, essays l Bibliography is a listing of all the materials that

and websites for writing something. Though you have been consulted while writing an article, es-
might have referred to these materials while pre- say or book. References, on the other hand, are
paring the rough draft, the contents of these may those that have been directly referred to or ref-
or may not have been included in the actual text. erenced in your article or book.
A list of all such materials is known as bibliogra- l Items of a bibliography are not directly included

phy. References are a list of those materials that are in the text. References are those that are directly
directly included in your actual text. cited in your actual text.
While all the items in the references are cited di- l Both bibliography and references are arranged

rectly in the text, all the items of the bibliography alphabetically, but a reference list can also be
may not be cited directly in the text. While references arranged in numeric or Vancouver style.
can be used to support your statement or argument
or used to supplement your findings, a bibliography VANCOUVER STYLE OF REFERENCE WRITING
does not have such roles. As such, references are used The Vancouver Style, or Uniform Requirements
for establishing something in a more authoritative Style, is based on an American National Standards
way. Readers could check your references and evalu- Institute (ANSI) standard adapted by the National
ate the correctness of your statement. Meanwhile, Library of Medicine (NLM) for databases, such as
a bibliography does not support your arguments Medline. It was developed in Vancouver in 1978
directly as it is not cited, and the reader cannot by editors of medical journals, who now meet an-
clearly establish where a particular item of the bibli- nually as the International Committee of Medical
ography is referred to in your study. Journal Editors (ICMJE). Over 500 medical jour-
A bibliography will contain all research materi- nals (including prestigious journals, such as the
als, including books, magazines, periodicals, web- British Medical Journal [BMJ], Canadian Medical
sites and scientific papers, which you have re- Association Journal [CMAJ] and Journal of American
ferred. References contain sources of material like Medical Association [JAMA]) use this style.
quotes or texts, which has been actually used
when writing an essay or book. Citation within the Text
Both bibliography and references appear at the Indicating the Relevant Reference in the Text
end of the book or article, but if both are available, A number in superscript format, e.g. 6 or enclosed
the bibliography comes after the reference list. in brackets (round or square), e.g. (1) or [26],
 Writing Proposals, Reports and Publications 267

placed in the text of the essay indicates the rele- l Authors: List each author’s last name and ini-
vant reference. Citations are numbered consecu- tials; full first names are not included as per this
tively in the order in which they appear in the text reference style. List all authors of a given publi-
and each citation corresponds to a numbered ref- cation, but if the number exceeds six, give the
erence containing publication information about first six names followed by ‘et al.’ For books with
the source in the reference list at the end of the chapters written by different authors, list the
publication, essay or assignment. Once a source authors of the chapter first, then the chapter
has been cited, the same number is used for all title, followed by ‘In:’, the editors’ names, and
subsequent references. No distinction is made the book title.
between print and electronic sources of references l Book and chapter titles: Capitalize the first let-
when citing within the text. Here are some exam- ter of the first word in the title. The rest of the
ples of this kind of referencing. title is in lowercase, with the exception of proper
l Superscript format names. Do not underline the title; do not use
 The largest lesion in the first study was 10 cm.13 italics.
 The theory was first put forward in 1987.1 l Journal citations: Provide the abbreviated jour-
 Scholtz2 has argued that… nal title, period year, semicolon, volume, issue
 Several recent studies3,4,15,16 have suggested number in parentheses, colon, page range
that… (elided to the minimum) and a period. For
 For example, see.7 example: Brain Res. 2002;935(1–2):40–6.
l Bracket format l Pages: For journals, the entire page range of an
 The largest lesion in the first study was 10 cm article or chapter is given, not the specific page
[13]. on which the information was found. For books,
 The theory was first put forward in 1987 (1). no page numbers are given, with two excep-
 Scholtz (2) has argued that… tions: the page number of a dictionary entry
 Several recent studies [3,4,15,16] have sug- and the page range of a chapter within a multi-
gested that… authored title.
 For example, see (7).
FORMAT OF WRITING THE REFERENCE
Citing More Than One Reference at a Time Books
When citing more than one source at a time, the Only the first letter of the first word in the title of
preferred method is to list each reference number a book or conference should be capitalized except
separately with a comma or en dash (without for proper nouns or acronyms. Capitalize the ‘V in
spaces) between each reference. volume for a book title.
l Superscript format

 1,3,5

 1–5
Standard Format
 2–5,9,13 #. Author/editor AA. Title: subtitle. Edition (if not
l Bracket format the first). Vol. (if a multivolume work). Place of
 (1,3,5) publication: Publisher; Year. Page number(s) (if
 (1–5)
appropriate).
 [2–5,9,13]

l Single author or editor

Important Issues  Hoppert M. Microscopic techniques in bio-

l Numbering: List all references in order by num- technology. Weinheim: Wiley-VCH; 2003.
ber, not alphabetically. Each reference is listed  Storey KB, editor. Functional metabolism: reg-

once only, since the same number is used ulation and adaptation. Hoboken (NJ): J. Wiley
throughout the paper. & Sons; 2004.
268 Research Methodology and Biostatistics

l Two or more authors or editors on behavioural adaptation. Canberra: A.G.P.S.;

 Lawhead JB, Baker MC. Introduction to vet- 1977. Australian Government Department
erinary science. Clifton Park (NY): Thomson of Science, Antarctic Division. ANARE scien-
Delmar Learning; 2005. tific reports. Series B(4), Medical science
 Gilstrap LC, Cunningham FG, Van Dorsten JP, No. 0126.
editors. Operative obstetrics. 2nd ed. New York: l Thesis or dissertation
McGraw-Hill; 2002.  Borkowski MM. Infant sleep and feeding: a

l No author telephone survey of Hispanic Americans. PhD

 The Oxford concise medical dictionary. 6th ed. [dissertation]. Mount Pleasant (MI): Central
Oxford: Oxford University Press; 2003. p. 26. Michigan University; 2002.
l Later edition  Hincks CL. The detection and characterisation

 Murray PR, Rosenthal KS, Kobyashi GS, of novel papillomaviruses. Biomedical Science,
Pfaller MA. Medical microbiology. 4th ed. Honours [thesis]. Murdoch: Murdoch Univer-
St Louis: Mosby; 2002. sity; 2001.
l Multiple volumes l Patent/standard
 Lea Lee GR, Bithell TC, Foerster J, Athens J W,  Pagedas AC, inventor; Ancel Surgical R&D

Lukens JN, editors. Wintrobes clinical hema- Inc., assignee. Flexible endoscopic grasping
tology. 9th ed. Vol. 2. Philadelphia: Lea & and cutting device and positioning tool as-
Febiger; 1993. sembly. United States patent US 20020103498.
l Series 2002 Aug 1.
 Unwin E, Codde J P, Bartu A. The impact of

drugs other than alcohol and tobacco on the Parts of a Book

health of Western Australians. Perth: Depart- Note: These examples are for chapters or parts of
ment of Health; 2004. Epidemiology occasional edited works in which the chapters or parts have
papers series, No. 20. individual title and author/s, but are included in
l Corporate author (i.e. a company or organiza- collections or textbooks edited by others. If the
tion) editors of a work are also the authors of all of the
 American Veterinary Medical Association. Na- included chapters, then it should be cited as a
tional Board Examination Committee. North whole book using the examples given earlier
American Veterinary Licensing Examination: (books). Capitalize only the first word of a paper
bulletin of information for candidates. Bismarck or book chapter.
(ND): The Committee; 2001.
l Conference (complete conference proceedings) Standard Format
 Harnden P, Joffe JK, Jones WG, editors. Germ

cell tumours V. Proceedings of the 5th Germ #. Author of Part, AA. Title of chapter or part. In:
Editor A, Editor B, editors. Title: subtitle of Book.
Cell Tumour conference; 2001 Sep 13–15;
Edition (if not the first). Place of publication:
Leeds, UK. New York: Springer; 2002. Publisher; Year. Page numbers.
l Government publication
 Australia. Commonwealth Department of

Veterans’ Affairs and Defence. Australian Gulf l Article/chapter in a book

War veterans’ health study 2003.  Meltzer PS, Kallioniemi A, Trent JM. Chromo-

 Australian Bureau of Statistics. Disability, age- some alterations in human solid tumors. In:
ing and carers: summary of findings. Canberra: Vogelstein B, Kinzler KW, editors. The genetic
ABS; 1999. ABS publication 4430.0. basis of human cancer. New York: McGraw-
l Scientific/technical report Hill; 2002. p. 93–113.
 Lugg DJ. Physiological adaptation and health  Ford HL, Sclafani RA, Degregori J. Cell cycle

of an expedition in Antarctica: with comment regulatory cascades. In: Stein GS, Pardee AB,
 Writing Proposals, Reports and Publications 269

editors. Cell cycle and growth control: biomo- l Drummond PD. Triggers of motion sickness
lecular regulation and cancer. 2nd ed. Hoboken in migraine sufferers. Headache. 2005;45(6):
(NJ): Wiley-Liss; 2004. p. 42–67. 653–6.
l Conference or seminar paper l Halpern SD, Ubel PA, Caplan AL. Solid-organ
 Christensen S, Oppacher F. An analysis of transplantation in HIV-infected patients. N Engl
Koza’s computational effort statistic for ge- J Med. 2002;347(7):284–7.
netic programming. In: Foster JA, Lutton E, l Geck MJ, Yoo S, Wang JC. Assessment of cervi-
Miller J, Ryan C, Tettamanzi AG, editors. cal ligamentous injury in trauma patients using
Genetic programming. EuroGP 2002: Pro- MRI. J Spinal Disord. 2001;14(5):371–7.
ceedings of the 5th European Conference  More than six authors

on Genetic Programming; 2002, Apr 3–5; - Gillespie NC, Lewis RJ, Pearn JH, Bourke
Kinsdale, Ireland. Berlin: Springer; 2002. ATC, Holmes MJ, Bourke JB, et al. Cigua-
p. 182–91. tera in Australia: occurrence, clinical fea-
l Study guides and unit readers tures, pathophysiology and management.
Note: You should not cite from unit readers, Med J Aust. 1986;145:584–90.
study guides or lecture notes, but where pos-  Organization as author

sible you should go to the original source of - Diabetes Prevention Program Research
the information. If you do need to cite articles Group. Hypertension, insulin, and proinsu-
from a unit reader, treat the reader articles as lin in participants with impaired glucose
if they are book or journal articles. In your tolerance. Hypertension. 2002;40(5):679–86.
citations, refer to the page numbers from the  No author given

reader, if given, not the original page numbers - 21st century heart solution may have a sting
(unless you have independently consulted the in the tail. BMJ. 2002;325(7537):184.
original).  Volume with supplement

- Lynch M. God’s signature: DNA profiling, - Geraud G, Spierings EL, Keywood C. Tolera-
the new gold standard in forensic science. bility and safety of frovatriptan with short-
Endeavour. 2003;27(2): 93–7. Reprinted In: and long-term use for treatment of migraine
Forensic Investigation (BIO373) unit reader and in comparison with sumatriptan. Head-
for forensic DNA component. Murdoch: ache. 2002;42 Suppl 2:S93–9.
Murdoch University; 2005.  Issue with supplement

- Glauser TA. Integrating clinical trial data

Journal Articles into clinical practice. Neurology. 2002;58
Note: Capitalize only the first word of an article (12 Suppl 7):S6–12.
title, except for proper nouns or acronyms. List  Volume with part

the first six authors followed by ‘et al.’. The titles of - Abend SM, Kulish N. The psychoanaly tic
journals should be abbreviated as they appear in method from an epistemological viewpoint.
the MEDLINE Journals Database. Volume, issue Int J Psychoanal. 2002;83(Pt 2):491–5.
and page numbers are given but not labelled.  Issue with part

To indicate a page range, use elided ranges, e.g. - Ahrar K, Madoff DC, Gupta S, Wallace MJ,
123–9, 126–34 or 111–222. If you refer to only one Price RE, Wright KC. Development of a
page, use only 111. large animal model for lung tumors. J Vasc
Interv Radiol. 2002;13(9 Pt 1):923–8.
 Issue with no volume
Standard Format
- Banit DM, Kaufer H, Hartford JM. Intraop-
#. Author of Article AA, Author of Article BB, erative frozen section analysis in revision
Author of Article CC. Title of article. Abbreviated total joint arthroplasty. Clin Orthop. 2002;
Title of Journal. Year; vol.(issue):page number(s). (401):230–8.
270 Research Methodology and Biostatistics

 Newspaper article infection in central African nonhuman pri-

- O’Leary C. Vitamin C does little to prevent mates. J Infect Dis [abstract]. 2004 [cited 2005
winter cold. The West Australian. 2005 Jun 30];190(11):1895. Available from: Pro-
Jun 29;1. Quest. http://www.umi.com/proquest/.
Note: Abstract citations are only included in a
E-Books reference list if the abstract is substantial or if
Standard Format the full text of the article could not be accessed.
l Journal article from online full-text database:
#. Author A, Author B. Title of e-book [format].
Note: When including the internet address of
Place: Publisher; date of original publication
[cited year abbreviated month day]. Available articles retrieved from searches in full-text
from: Source. URL. databases, please use the site ‘Recommended
URLs for Full-text Databases’, which are the
l Van Belle G, Fisher LD, Heagerty PJ, Lumley TS. URLs for the main entrance to the service and
Biostatistics: a methodology for the health sci- are easier to reproduce.
ences [e-book]. 2nd ed. Somerset (NJ): Wiley  Palsson G, Hardardottir KE. For whom the

InterScience; 2003 [cited 2005 Jun 30]. Available cell tolls: debates about biomedicine (1). Curr
from: Wiley InterScience electronic collection. Anthropol [serial online]. 2002 [cited 2005
l Sommers-Flanagan J, Sommers-Flanagan R. Clin- Jun 30];43(2):2711[about 31 pages]. Avail-
ical interviewing [e-book]. 3rd ed. New York: John able from: Academic OneFile. http://find.
Wiley & Sons; 2003 [cited 2005 Jun 30]. Available galegroup.com.
from: NetLibrary.  Allen C, Crake D, Wilson H, Buchholz A. Poly-

 Article in an electronic reference book: cystic ovary syndrome and a low glycemic in-
- Widdicombe J. Respiration. In: Blakemore dex diet. Can J Diet Pract Res [serial online].
C, Jennett S, editors. The Oxford compan- 2005 [cited 2005 Jun 30];Summer:3. Available
ion to the body [e-book]. Oxford: Oxford from: ProQuest. http://il.proquest.com.
University Press; 2001 [cited 2005 Jun 30.]. l Journal article in a scholarly journal (pub-
Available from: Oxford Reference Online. lished free of charge on the internet):
http://www.oxfordreference.com.  Eisen SA, Kang HK, Murphy FM, Blanchard

- Lee HC, Pagliaro EM. Serology: blood iden- MS, Reda DJ, Henderson WG, et al. Gulf War
tification. In: Siegel J, Knupfer G, Saukko P, veterans’ health: medical evaluation of a U.S.
editors. Encyclopedia of forensic sciences cohort? Ann Intern Med [serial on the Internet].
[e-book]. San Diego: Academic Press; 2000 2005 [cited 2005 June 30];142(11):8811[about
[cited 2005]: 1331–8. Available from: Sci- 12 pages]. Available from: http://www.annals.
ence Direct Reference. www.elsevier.com. org/.
l Journal article in electronic journal sub-
E-Journals scription:
 Barton CA, McKenzie DP, Walters EH, et al.
Standard Format
Interactions between psychosocial problems
#. Author A, Author B. Title of article. Abbreviated
and management of asthma: who is at risk of
Title of Journal [format]. Year [cited year abbrevi-
ated month day];vol.(issue):page numbers [esti- dying? J Asthma [serial on the Internet]. 2005
mated if necessary]. Available from: database [cited 2005 Jun 30];42(4): 249–56. Available
name (if appropriate). URL. from: http://www.tandf.co.uk/journals/.
l Newspaper article from online database:
l Journal article abstract accessed from online  Dearne K. Dispensing with the chemist. The

database: Australian [newspaper online]. 2005 Jun 14

 Leroy EM, Telfer P, Kumulungui B, Yaba P, [cited 2005 Jun 30];[about 8 screens]. Avail-
et al. A serological survey of Ebola virus able from: Factiva. http://global.factiva.com.
 Writing Proposals, Reports and Publications 271

l Newspaper article from the internet: Internet]. Sydney: ABC Television; 2007 June 4
 Diseased organs may be used to deter smok- [cited 2008 Feb 4]. Available from http://
ers. Sydney Morning Herald [newspaper on www.abc.net.au/tv/pekingtoparis/podcast/
the Internet]. 2005 Jun 29 [cited 2005 Jun 30]; pekingtoparis.xml.
para. 4–5. Available from: http://www.smh.  Gary, S. Black hole death ray. StarStuff [pod-

com.au/. cast on the Internet]. Sydney: ABC News

Radio; 2007 Dec 23 [cited 2008 Feb 4]. Avail-
Internet Documents able from http://abc.net.au/newsradio/podcast/
STARSTUFF.xml.
Standard Format l Microform:
#. Author A, Author B. Document title. Webpage  Terry KW, Hewson GS, Rowe MB. Charac-

name [format]. Source/production information; terisation of inhaled dusts at minesites

date of internet publication [cited year month [microfiche]. Perth: Minerals and Energy
day]. Available from: URL. Research Institute of Western Australia;
1998.
l Professional internet site: l Video recording:
 Australian Institute of Health and Welfare.  Hillel J, writer. Out of sight out of mind: in-

Chronic diseases and associated risk factors digenous people’s health in Australia [video
[document on the Internet]. Canberra: The recording]. Bendigo: Video Education Aus-
Institute; 2004 [updated 2005 June 23; cited tralasia; 2003.
2005 Jun 30]. Available from: http://www. l Television programme:
aihw.gov.au/cdarf/index.cfm.  Cohen J, reporter. Messing with heads. Four

l Personal internet site: corners [television broadcast]. Australian

 Stanley F. Information page—Professor Fiona Broadcasting Corporation; 2005 Mar 21.
Stanley. Telethon Institute for Child Health
Research [homepage on the Internet]. Perth: Personal Communications
The Institute; 2005 [cited 2005 Jun 30] Avail- l Conversation:
able from: http://www ichr.uwa.edu.au/about/  In a conversation with a colleague from the

schools/. School of Population Health (Jameson LI

l General internet site: 2002, oral communication, 7th August)…
 Lavelle P. Mental state of the nation. Health l Letter:
matters [document on the Internet]. ABC  As stated in a letter from B.J. Samuels, MD, in

online; 2005 May 19 [cited 2005 Jul 1]. Avail- July 2002….
able from: http://abc.net.au/health/features/ l Email:
mentalstate/.  Smith P. New research projects in gastroenter-

ology [online]. Email to Matthew Hart (mh@

Writing Reference for Non-Book Formats hospital.wa.gov.au) 2000 Feb 5 [cited 2000
Mar 17].
Standard Format

#. Person AA, Responsibility (if appropriate). Title:

AMERICAN PSYCHOLOGICAL ASSOCIATION
subtitle [format]. Special credits (if appropriate). STYLE OF REFERENCES
Place of publication: Publisher; Year. Books
l Single-author book and part of a book:
l Podcasts:  Alverez, A. (1970). The savage god: A study of

 Brown W, Brodie K, & George P. From Lake suicide. New York: Random House.
Baikal to the halfway mark, Yekaterinburg.  Garner, B. A. (2003). Garner’s Modern Ameri-

Peking to Paris: Episode 3 [podcast on the can usage. New York: Oxford University Press.
272 Research Methodology and Biostatistics

l Book with two or more authors: referring to more than one volume, include all the
 Natarajan, R., & Chaturvedi, R. (2003). Geol- relevant volume numbers (as in the earlier exam-
ogy of the Indian Ocean floor. Hartford, CT: ple). The date after the author or editor’s name
Merganser University Press. should include the range of years of publication
 Ligon, M., Carpenter, K., Brown, W., & Milsop, (say, if the work appeared over a number of years).
A. (1983). Computers in the world of business
communications. Hartford, CT: Capital Press. Secondary Resources
l Edition other than first (republished book): l O’Connor, C. O. & DeLoatch, K. L. (2003).
 Creech, P. J. (2004). Radiology and the technol- Whatever happened to the humanities? In
ogy of the absurd. Boston: Houghton-Mifflin. I. Rubenzahl (Ed.), Studies in Byzantine intrigue
(Original work published 1975). (pp. 235–278). Hartford, CT: Merganser Uni-
After the author’s name, list the most recent versity Press.
date of publication you can find. In the text cita- To use material that is quoted or paraphrased
tion, give both dates: (Creech, 1975/2004). elsewhere when you do not use the original re-
l An edited volume (anthology or reference source, your reference will include the source of
book): your language (or idea).
 Stanton, D. C., & Farbman, E. (Eds.). (2003). In your text, you would quote or paraphrase
The female autograph: Theory and practice of the idea that O’Connor and others have quoted or
autobiography. Middletown, CT: Ibis Book- used, as follows:
store Press. As Villa trenchantly points out, “Perhaps the
If you are referring to an article or signed chap- conflict seems so strong because the stakes are so
ter in an edited volume, your reference would look low” (as cited in O’Connor & DeLoatch, 2003).
like this:
l Pepin, R. E. (2002). Uses of time in the political Dictionary
novels of Joseph Conrad. In C. W. Darling, Jr., l Shorter Oxford English dictionary (5th ed.).(2002).
J. Shields, & E. C. Farbman (Eds.), Chronological New York: Oxford University Press.
looping in political novels (pp. 99–135). Hartford:
Capital Press. Dissertation or Dissertation Abstract
Notice that there are no quote marks around When you have used an abstract of the disserta-
the title of the article and that only the first word tion found on microfilm in Dissertation Abstracts
of a title is capitalized. The names of editors are International (DAI). If you are using a microfilm
listed first-initials-first (not reversed). The inclu- source, include in parentheses at the end of your
sive page numbers are preceded by the abbrevia- entry, the university microfilm number.
tions for pages’: pp. l Darling, C. W. (1976). Giver of due regard: the

l Book without author or editor listed: poetry of Richard Wilbur. Dissertation Abstracts
 Webster’s New Collegiate Dictionary. (1961). International, 44, 4465. (AAD44–8794)
Springfield, MA: G. & C. Merriam. When you have used the actual dissertation (usually
To alphabetize such an entry on your References from the shelves of the University where it was writ-
page, use the first significant word of the title ten, sometimes obtained through interlibrary loan):
(ignoring the articles, ‘a’, ‘an’, and ‘the’). l Darling, C. W. (1976). Giver of due regard: the

l Multivolume work: poetry of Richard Wilbur. Unpublished doc-

 Nadeau, B. M. & Darling, J. M. (Eds.). toral dissertation, University of Connecticut,
(1994–2003). Studies in the history of cutlery Storrs, CT.
(Vols. 4–6). Utica, NY: Mohican Valley-River
Press. Magazines/Periodicals
To refer to a single volume, include only the l Wheatcroft, G. (2004, June). The Tragedy of
relevant date and volume number. When you are Tony Blair. The Atlantic, 293, 56–72. Thomas, E.
 Writing Proposals, Reports and Publications 273

& Hosenball, M. (2004, May 31). Bush’s Mr. If the text being cited is from an editorial or letter
Wrong: The Rise and Fall of Chalabi. Newsweek, to the editor indicate the nature of the source in
143, 22–32. brackets after the title and date.
Use inclusive page numbers. Do not use the  Silverman, P.H. (2004, June). Genetic Engi-

abbreviations ‘p.’ or ‘pp.’ neering [Letter to the editor]. The Atlantic,

293, 14.
Scholarly Journal
Many scholarly journals number their pages con- Nonprint Resources
secutively throughout a given volume. Thus, vol- l Film:
ume one might end on page 345 and the first page  Redford, R. (Director). (1980). Ordinary people

of volume two would be number 345. Do not use [Film]. Hollywood: Paramount.
p. or pp. to indicate page numbers. Notice that The source is identified in brackets after the
proper nouns are capitalized in an APA style title, title
but other words are not. A comma separates the - Film with limited circulation:
title of the journal from the volume number, but - Holdt, D. (Producer), & Ehlers, E.
the volume number (as well as the comma that (Director). (2002). River at High Sum-
follows) is also italicized. mer: The St. Lawrence [Film]. (Avail-
l Christie, John S. (1993) Fathers and virgins: able from Merganser Films, Inc., 61
Garcia Marquez’s Faulknerian Chronicle of a Woodland Street, Hartford, CT 06105)
Death Foretold. Latin American Literary Review, l Cassette:
13, 21–29.  Lake, F. L. (Author and speaker). (1989). Bias

If that article is found in a journal in which and organizational decision making [Cassette].
each issue has pages numbered separately (each Gainesville: Edwards.
issue begins with a page 1), the issue number l Television programme:
should be given in parentheses but not in italics  Safer, M. (Narrator). (2004). Torture at Abu

after the volume number. Ghraib [Television broadcast]. Hartford:

l Christie, John S. (1993) Fathers and virgins: WFSB.
Garcia Marquez’s Faulknerian Chronicle of a l Musical recording:
Death Foretold. Latin American Literary Review,  Barber, S. (1995). Cello Sonata. On Barber

13(3), 21–29. [CD]. New York: EMI Records Ltd.

Newspaper Articles, Editorials, Letters Personal Communication

to the Editor Because this material is not recoverable (i.e. it is
If the article is ‘signed’ (i.e. you know the au- not possible for someone else to see or hear it), it
thor’s name), begin with that author’s name. should not be listed in the list of references. It can,
(Notice how discontinuous pages are noted however, be cited parenthetically within the text. It
and separated by commas.) Do not omit the is extremely important that what is cited in this
‘The’ from the title of a newspaper (unlike the way be legitimate and have scholarly integrity.
MLA technique). l Interview:

 Poirot, C. (2004, March 17). HIV prevention  Wilbur finds himself sometimes surprised by

pill goes beyond ‘morning after’. The Hartford the claims of religiosity made by contempo-
Courant, pp. F1, F6. raries. (personal letter, March 28, 1977).
If the author’s name is not available, begin the l Phone conversation:

reference with the headline or title in the au-  According to Connie May Fowler, the sources

thor position. for her novel Sugar Cane were largely autobi-
 New exam for doctor of future. (1989, March 15). ographical (personal phone conversation, July
The New York Times, p. B-10. 22, 2003).
274 Research Methodology and Biostatistics

Classroom Lecture photocoagulation: Manual of operations [675

Like personal interviews and phone conversations, paragraphs]. Online Journal of Current Clini-
material presented in a classroom lecture is re- cal Trials [On-line serial]. Available: Doc
garded as nonretrievable data. A lecture, therefore, No. 92
should be cited within the text but not be included l Online abstract:
in your list of references. The following model You can cite an entire Website within your
could also be used for more formal lecture settings. text, but do not include it in your list of
l In an Introduction to Literature lecture at Capi- references. For instance, you could say
tal Community College on April 14, 2004, Pro- something incredibly profound you learned
fessor Charles Darling described William Carlos from this Web page, and cite the URL at
Williams’ poem as a barnyard snapshot (C.W. the end of the paragraph (http://www.ccc.
Darling, ENG 102 lecture, April 14, 2004). commnet.edu/apa/). If a line-break is nec-
If the lecturer distributed material at the lecture, essary when you cite a URL, make the break
you could cite that resource in your list of References: immediately after a slash mark and care-
l Darling, C.W. (2004, April). Images at Work in fully avoid the insertion of a hyphen where
the Poetry of William Carlos Williams. Outline none is appropriate.
presented in a classroom lecture at Capital l Electronic database:
Community College, Hartford, CT. College and public libraries subscribe to elec-
tronic databases, such as EBSCO, LexisNexis,
Government Documents OCLC, WilsonWeb and SIRS. There are also
l Report from the Government Printing Office, online databases, such as ERIC, PsychINFO
corporate author. Example: and Public Affairs Information Service (PAIS).
 National Institute of Mental Health. (1982). These databases contain full-text articles and
Television and behavior: Ten years of scientific article abstracts.
progress (DHHS Publication No. A 82–1195).
Washington, DC: U.S. Government Printing
Office.
l Reports from a Document and Deposit Service
WRITING SCIENTIFIC ARTICLES
(NTIS, ERIC) other than U.S. government. Guidelines for writing research articles to
Examples: publish in professional scientific journals gen-
 Tandy, S. (1980). Development of behavioral erally vary from journal to journal based on
techniques to control hyper-aggressiveness in the manuscript guidelines followed by the
young children (CYC Report No. 80–3562). particular journal. However, most journals
Washington, DC: Council on Young Children. follow nearly the same format as discussed as
(NTIS No. P880–14322). follows.
 Gottfredson, L. S. (1980). How valid are occu- l Title page: Full title of the paper, which should

pational reinforcer pattern scores? (Report No. be concise, informative and generally not ex-
CSOS-R-292). ceeding 10–12 words. Title of the article is
 Baltimore, MD: Johns Hopkins University. followed by the name of the authors in the
Center for Social Organization of Schools. form that publication requires should be pro-
(ERIC Document Reproduction Service No. vided. The degree of affiliation and full official
ED 182 465). address should be given on a separate sheet at
the end of the paper. One of the authors
Electronic and Online Resources should be designated to receive correspon-
l Online article: dence; he or she must take responsibility of
Central Vein Occlusion Study Group. (1993. keeping the other authors informed of the
October 2). Central vein occlusion study of progress of the paper.
 Writing Proposals, Reports and Publications 275

For example,  Objectives of the study: This section includes

the sentences of the general as well as specific
Factors Affecting Bowel Movement in Critically Ill
objectives.
Patients
 Materials and methods: This section includes
Dr. Suresh K. Sharma, a brief description of the research methodol-
Professor, Department of Nursing, ogy, which involves the research design, re-
All India Institute of Medical Sciences, Rishikesh, search setting, target population, sampling
Uttarakhand
technique, inclusion and exclusion criteria for
[email protected]
the sample, sample size, description of research
tool(s), reliability and validity of the research
l Abstract: The abstract should be brief, concise tool(s), methods of data collection, ethical
and factual, generally not exceeding 200 words in consideration and plan for data analysis.
length. It may be written in the form of a para-  Results: This section includes the presentation
graph comprising brief information about the of the analysed study results through tables,
introduction, objectives, basic methodology, main illustrations and brief descriptions.
results and conclusion. Otherwise, concise infor-  Discussion: This section presents the explana-
mation may be provided under these headings
tion of the results and findings, where the re-
keeping in mind the word limits. Abstract must be
searchers explain their research findings through
followed by 3–5 keywords for indexing purpose.
critical analysis along with comparison with
For example,
other similar research findings.
 Conclusion and recommendations: This sec-
Abstract: This prospective study was conducted on 50
tion of the article presents the summary of what
consecutive purposively selected patients (age 12–55
years), who were admitted in selected ICUs of DMCH, you did and found. It also presents the recom-
Ludhiana. Daily follow-up of subjects ranged between 7 mendations based on factual research findings.
and 21 days with mean of 13.54 6 4.31 days and a total  References: In most health science journals,
of 720 observations were made. Critically ill patients the references are presented by using the Van-
who were NPO, consumed small amount of liquid diet couver style of references.
(,0.5 L/24 h), consumed selected drugs (opioids, anal-
gesics, anticonvulsants, antidepressants, sedatives, iron
supplements and calcium preparation), the uncon- CRITICAL REVIEW OF RESEARCH
scious (had no movement) had higher observations of WORK
absence of daily bowel movements and it was found
highly statistical significant (p , 0.001). Patients who Research critique is a planned, careful critical
were on mechanical ventilator had more number of evaluation of a piece of research work against the
absences of daily bowel movements and it was found prespecified criteria to judge the strengths and
statistically significant (p , 0.05). In addition, patients weaknesses of the research study. Critique should
with 24 h negative fluid balance had higher percentage be balanced, where alternative suggestions must
of absence of daily bowel movement but it was not
be provided to further enforce the strengths and
found statistical significant (p . 0.05).
Keywords: Factors, bowel movements, critically ill eliminate the weaknesses of the study to improve
patients overall quality of the research project.

l Main text: Generally, a manuscript contains PURPOSES

2500–3000 words, including the references. The l To provide inputs regarding the strength and
main text of the article should be divided into weakness of a study to the researchers
sections, with the following headings: l To provide suggestions to the students regard-
 Introduction: This section includes a brief ing the methodological flaws in their research
description of the background information, project and also to evaluate the understanding
need of the study and literature review. of research by the students
276 Research Methodology and Biostatistics

l To judge the scientific merits of the study addition, we must not just let go after framing
l To take a decision whether to publish the study negative comments; alternative suggestions must
in a journal or not be provided for further improvement in the re-
search project.
GUIDELINES FOR WRITING A RESEARCH l Positive and negative comments must be sup-
CRITIQUE ported with examples to make a clear stand
l Read and understand the research report carefully. about the strengths and weaknesses of the re-
l Carry out critical appraisal of all the aspects of search report.
the research report before writing the critique.
l Avoid general vague statements; be objective GUIDELINES FOR CRITIQUING A RESEARCH
and sensitive while framing negative comments REPORT
and be practical by considering all the limita- Some of the positive points of the study can be
tions of the research. gathered from the questions listed in the follow-
l Keep a balanced approach in the research cri- ing section; the more the number of questions
tique by presenting both strong and weak answered with ‘yes’, the better the study is con-
points of the report, because certainly no re- sidered. A rating scale as presented in Table 11.3
port can be without any negative aspects. In may also be used for critiquing a research report.

Table 11.3. Rating scale for research critique

Title
Does the title of the research report identify the major study variables and the population un- 1 2 3
der investigation
Abstract
Is the research abstract clearly written, highlighting the major features of the investigation 1 2 3
Style of Organization
Does the researcher present the research report in a clear, understandable and exciting way 1 2 3
Objectives of the Study
• Do the objectives describe why the study was designed 1 2 3
• Do the objectives reflect how results will be used 1 2 3
• Are the objectives clearly specified 1 2 3
Problem/Hypothesis Under Study
• Is the hypothesis clearly stated 1 2 3
• Has the hypothesis specified the relationship between the variables 1 2 3
Theoretical Framework
• Is a clear conceptual model described in the study 1 2 3
• Are the major concepts of the models identified and defined 1 2 3
• Are the relationships among concepts within the model described 1 2 3
• Does the selected model best explain the phenomena under study 1 2 3
• Is the theorem to be tested clearly stated 1 2 3
Literature Review
• Has an in-depth review been conducted 1 2 3
• How recent are the identified publications 1 2 3
• Have any primary sources been used 1 2 3
• Is the methodology of the studies present in the description 1 2 3
 Writing Proposals, Reports and Publications 277

Table 11.3. Rating scale for research critique—cont’d

Research Design
• Is this the best-suited design for the study 1 2 3
• Are controls applied to the study of the greatest extent possible 1 2 3
• Are the extraneous variables identified or taken into account in the design of the study 1 2 3
• Is there any evidence of bias in relation to the methods used 1 2 3
Data Collection Measures
• Is the level of data (nominal, ordinal, interval, ratio) appropriate given the focus of the study 1 2 3
• How valid and reliable are the instruments used in the study 1 2 3
Sample
• Are there any factors related to the process that may have caused the sample to be biased 1 2 3
• If random sampling was used, has the process been described 1 2 3
• Are generalizations made that are not warranted given the sampling procedure used 1 2 3
Analysis of Data
• Have ratios, rates, proportions and measures of central tendency been correctly calculated 1 2 3
• Do the calculations provide useful information 1 2 3
• Is the confidence interval interpreted as the frequency with which the population mean lies 1 2 3
between the stated limits
• Are the results of the study expressed in terms of probability 1 2 3
• Is the level of probability specified 1 2 3
• Is the level of significance reported 1 2 3
• Is the correct form of inferential test used 1 2 3
• If randomization is not used, are results interpreted with that factor in mind 1 2 3
Findings of Study
• Are the tables clearly labelled, and convey the desired information in a clear simple fashion 1 2 3
• Is each hypothesis addressed 1 2 3
• Are the study limitations discussed 1 2 3
• Does the interpretation of the results fit with other components of the study 1 2 3
• Is adequate information presented relating to the statistical test used 1 2 3
Discussion
• Are the results of the study discussed in the light of previous works with insight to the 1 2 3
current study results
• Does the researcher highlight major findings 1 2 3
Suggestion for Use of Findings
Does the researcher clearly identify how the study results relate to the field of related health 1 2 3
care practice, education and administration
References
Are the references relevant and completely follow the recommended style 1 2 3
Notes: Rate the afore-mentioned statements on a three-point rating scale: 1 5 weak or absent; 2 5 appropriate; 3 5 strong.
Please tick mark on the box you feel appropriate for the given statements.

CRITIQUE OF THE STUDY BACKGROUND l Is the problem significant for the particular
Research Problem
health care profession and has the researcher
provided good arguments for the significance?
l Is the problem statement clearly and concisely l Is the research problem within the professional
articulated? domain?
278 Research Methodology and Biostatistics

l Is the problem not complex to study and ap- l Are all the concepts in the conceptual frame-
propriately delimited in its scope? work clearly, adequately and logically defined
l Does the problem statement clearly depict the and articulated in a way that they help either in
variables, population and place of the study? the testing or generation of a theory?
l Is the research problem feasible to study in ref- Note: Qualitative research, descriptive studies
erence to time, resources, researcher’s ability and biophysiological studies typically will not
and availability of respondents? have a theoretical framework.
l Is the research problem directed to develop or
test the health care theories? Literature Review
l Is the literature completely in accordance with
Objectives
the research problem and covers majority of
l Are the objectives in accordance with research available studies on the topic under study?
problem? l Is the literature reviewed from current and pri-
l Are the objectives stated in clear and concise form? mary sources?
l Are the objectives stated using accepted action l Is the literature review well organized under
verbs in a logical manner? subheadings in a chronological order according
to the year of the study?
Operational Definitions l Is the literature review presented in an analytical
l Are all the study variables operationally defined? way to draw meaningful inferences?
l Are operational definitions stated in a logical
and measurable form? CRITIQUE OF THE METHODOLOGICAL ASPECTS
Hypotheses (Hypotheses are Not Used in Research Design
Qualitative Studies and Some Descriptive l Is the design specified correctly? What was the
Studies) design used?
l Are the hypotheses stated? If not, does the l Is the design appropriate to answer the research
researcher provide sufficient information for question?
not stating the hypotheses? l Did the researcher attempt to control for threats
l Are hypotheses logically related with the research to internal and external validity?
problem and objectives?
l If stated, are the hypotheses clear? Specific? Sample and Sampling Technique
Testable? l Is the sample size adequate?
l Are hypotheses stated in measurable terms by l Is the sample likely to be similar to members of
stating the level of significance (e.g. p 5 0.05) to the population overall?
accept or reject the hypotheses? l Are the criteria for including and/or excluding
l Are hypotheses stated in a way that they express people or items from the sample clear and
a predicted relationship between two or more appropriate?
variables? l Is an appropriate sampling technique used to
l Are hypotheses based on the conceptual frame- draw a sample from the population?
work used in the research study?
Data Collection Method and Tools
Conceptual Framework l Are the instruments or other means for data
l Is the conceptual framework appropriate and in collection described sufficiently?
accordance with the research problem? l Are the reliability and validity of instruments
l Is the conceptual framework based on a health care addressed? Are these adequate?
theory or a theory from other related disciplines? l Are data collection methods described clearly?
 Writing Proposals, Reports and Publications 279

l Are the data collection methods appropriate? Interpretation and Discussion of Data
Could the researcher mention the specific ben- l Does the discussion ‘fit’ with the data? Is it logi-
efits and problems faced by the research with cal based on the data and results presented?
particular methods of data collection? l Does the researcher discuss the findings in
regard to previous research?
Ethical Considerations l Does the researcher discuss the findings in
l Are the participants safe from any physical regard to the theoretical framework?
harms, risks, psychological and social distress l Does the researcher identify limitations of
and discomfort? the study? How do these affect the quality of
l Were confidentiality of information and ano- the study?
nymity and privacy of subjects maintained? l Does the researcher discuss implications for
l Was an appropriate written or verbal consent practice? Are these appropriate?
taken from the study participants?
l Was a written permission obtained from the Conclusion and Recommendations
competent authority to conduct the research l Has the study concluded concisely and precisely
study? depicting most significant findings?
l If vulnerable populations were involved, was l Are the conclusions drawn in accordance with
special consideration given to informed consent study objectives, hypotheses/assumptions and
and study procedure? conceptual framework?
l Did the benefits of the study outweigh the risk l Are practical recommendations made towards
for individual subjects and was a risk/benefit improvement in the health care practices?
assessment considered? l Are the recommendations consistent with the
l Were subjects recruited, selected and assigned to research findings and results?
groups in an equitable way? l Are recommendations made with suggested
changes and improvements in the methodolo-
gies for further researcher studies on similar
CRITIQUE OF THE RESULTS, CONCLUSION AND
topics?
MISCELLANEOUS ASPECTS OF THE STUDY
Analysis of data References
l Is the process used to analyse data clear? l Are the in-text references cited as per the rec-
l Were the processes for data analysis appropriate ommended guidelines (e.g. using superscript
to answer the research question. format)?
l Do the results provide an answer to the research l Are the references relevant and completely fol-
question? low the recommended style (e.g. Vancouver
l If tables are provided, are they clear and under- style)?
standable?
Note: For qualitative studies, some additional DEVELOPING AND PRESENTING A RESEARCH
items for critique include the following: PROPOSAL
l Does the researcher provide sufficient exam- A research proposal is an outline of a research idea,
ples of the data (e.g. passages or quotations which helps in communicating this idea to faculty
from interviews) to support the identified or a funding agency, so that an approved project
results? can be implemented. Development of the research
l Does the researcher describe processes that were proposal starts when the research problem is suf-
used to avoid biasing or influencing the data ficiently specified to begin work and the researcher
obtained and the analysis procedures used (e.g. is satisfied that the problem is feasible. The re-
peer review, logs, memos, member check)? search proposal serves as a strong foundation for
280 Research Methodology and Biostatistics

the actual research study, which is conducted after l The research proposal also serves as a contract
due approval by experts from the faculty or fund- between researcher, guide and university.
ing agency. l It also can be used to seek ethical approval from
A well-written research proposal should con- the institutional, regional or national-level re-
vince members of the scientific community that search ethical committee.
the research is significant. The clearer and more l Research proposals are generally submitted to a
precise the research proposal, the smoother is the scholarship committee or other funding agency
actual conduct of a research project. In academic to seek financial grants for implementations of
institutions, the research proposal is commonly the research project.
known as synopsis.
Developing the Proposal
Meaning of Research Proposal Development of the research proposal involves:
A research proposal is a written document speci- l Choosing a topic

fying what the investigator proposes to study, and l Narrowing down and focusing on the topic

is, therefore, written before the project has com- l Formulating research objectives or questions

menced. Proposals serve to communicate the and ideas for analysis

research problem, its significance and planned l Outlining the key literature in the topic area

procedure for solving the problem to the inter- l Deciding on research methodology, research

ested party. That party may be a funding agency, a design and methods
faculty advisor or institutional officers depending l Proposing an approach to data analysis

upon the circumstances. l Proposing a format, e.g. how many chapters and

A research proposal is a comprehensive sum- suggested chapter headings

mary of what you intend to do, how it will be done l Developing a timeline

and why it is important. Some agencies view writ- l Developing a budget and resources required for

ten research proposal as an actual commitment on the project

the part of the researcher that the study will pro- l Organizing references and bibliography

ceed as originally outlined. Writing a research proposal can be a demand-

ing, frustrating, challenging and time-consuming
Purposes of Proposal process. A research proposal does not perma-
l A research proposal provides an opportunity to nently state what the researcher will do. It is a
the researcher and the experts to think through starting point and throughout the research, it will
a project carefully, and clarify and define what probably be adjusted. The researcher will be able
exactly to study. to trace the development of ideas and measure
l It serves as a blueprint and guiding path for the the progress made by referring back to the pro-
researcher to carry out the research project. posal.
l A research proposal helps the researcher to
communicate to the supervisor, faculty, depart- Format of Research Proposal
ment and funding agency what he or she is go- Different funding agencies or colleges may have
ing to study, so that the desired suggestions and different guidelines/format requirements. When
support to carry out the research project can be this is the case, the principal investigator should
obtained. prepare the proposal in the required format,
l The proposal also helps the department to make provided that the format used gives all or most
the right decision about allotment of guide for of the details as contained in the guidelines of
the candidate pursuing a particular project. the particular institute. Before an attempt is
l It also gives an opportunity to receive feedback made to start with a research project, a research
from the supervisor and others in the academic proposal should be compiled. For the beginner
community as well as possible funders. researcher, this is usually the most difficult part.
 Writing Proposals, Reports and Publications 281

It is, however, the most important aspect of the Institution(s) under Whose Umbrella the Research
research project and should be considered care- Project Will Be Conducted
fully by the researcher. This requires not only For example:
subject knowledge but also insight into the l World Health Organization

problem that is going to be investigated, so as to l All India Institute of Medical Sciences, Rishikesh

give logical structure to the research study. The or

research proposal can be envisaged as the pro- l Indian Council Medical Research

cess (step-by-step guidelines) to plan and to give

structure to the prospective research with the Background Information and Introduction
final aim of increasing the validity of the re- This should include the following:
search. It is, therefore, a written submission to l A review of the relevant literature: It should

spell out in a logical format the nature of the be most recent. (Majority being in the past
design and the means and strategies that are 5–10 years at most.)
going to be used. l Locally available information published or un-

A research proposal usually consists of the published: It may include clinical or laboratory
following elements. observations (e.g. increasing number of adult
males presenting with head injuries at the
Title CMCH during national holidays).
This should be specific and precise. It should not
be more than two to three lines long, and should Rationale/Justification for the Research Project
indicate what one intends to do or find out, e.g. There should be a statement explaining why the
validation of an algorithm on the management of researcher(s) feel the research project is important
urethral discharge in CMCH, Ludhiana, Punjab. and, therefore, should be carried out (i.e. the poten-
The title is usually only formulated after the tial significance to health care delivery, or otherwise).
research problem and subproblems have been It should not be more than a paragraph or two.
stated in a more-or-less final format. The research
project title should demarcate the following: Objectives of the Study
l Who or/and what is researched l Broad objectives: The main issues that are be-
l Where ing looked at/for are laid down in the broad
l When objectives (e.g. to survey the sociodemographic
l How and reproductive profiles of patients with acute
l An indication of the envisaged solution or pos- gonococcal urethritis.
sible new product l Specific objectives: The specific issues that are
being looked at/for. These must be measurable,
About Investigators either qualitatively or quantitatively, and form a
l Full names, qualifications, academic titles of all guide to the research methodology, data analysis
the investigators, including trainees (assistant and presentation of results. For instance:
lecturers or postgraduate students), and their  To survey the age distribution of patients pre-

institutional/departmental affiliation(s). senting with acute gonococcal urethritis.

l The principal/main investigator (responsible  To assess the sexual behaviour of such patients.

for the work) should be the first one. If there are

co-investigators, these should be indicated as Formulating Hypotheses
appropriate with their qualifications, academic Because hypotheses give structure and direction
titles and institutional affiliations. to research, the following aspects should be kept
l Brief up-to-date curriculum vitae of each of the in mind when formulating a hypothesis:
investigators and co-investigators should be l Hypotheses can only be formulated after

provided. the researcher has gained enough knowledge

282 Research Methodology and Biostatistics

regarding the nature, extent and intensity of the  Its justification

problem.  The selection/inclusion/exclusion criteria, e.g.:
l Hypotheses should figure throughout the research - Every 10th client
process in order to give structure to the research. - Randomized (and how random)
l Hypotheses are tentative statements/solutions - Every patient who consents
or explanations of the formulated problem.  The need for, and type of, consent must be

Care should be taken not to oversimplify and specified, and how it will be obtained, e.g.
generalize the formulation of hypotheses. written or verbal consent
l The research problem does not have to consist l Data collection: The following details should be
of one hypothesis only. The type of problem provided:
area investigated and the extent to which it en-  What information is going to be looked for/

hances the research field are the determining collected?

factors on how many hypotheses will be in-  How that is going to be done (e.g. labora-

cluded in the research proposal. tory test or with provision of appropriate

details)?
Methodology  Requirements for the study should be listed

This should be very detailed as it is your guide on (e.g. reagents, culture media or blood samples
how the study will be done as well as the data and their relevant tests). It should include the
analysis. It should provide relevant information names of the people who will take responsi-
about the following: bility for each of the aspects of data collection
l The type of research study: For example, ran- (e.g. who will draw blood, perform the tests,
domized clinical trial, descriptive and cross- do the physical examination and interview
sectional. This should reflect exactly what the the study group).
researchers intend to do. l Data management and analysis:
l Location of study/study setting: It is the area  Details should be provided on how the collected

where the study is going to be conducted (e.g. data are going to be managed (e.g. coding).
10 secondary schools in Ludhiana district and/or  Details on data analysis, such as the computer

the CMC, Ludhiana). All the areas in/at which the package to be used in data entry and analysis,
survey/study will be carried out must be indicated. e.g. SPSS and EPI-INFO 6.0.
l Study population: The subjects who are to be  The type of statistical tests to be used, e.g.

included in the study or from which group(s) of regression analysis or Student’s ‘t’-test of
people is the study group going to be drawn, e.g. significance.
patients presenting with multiple pregnancy at l Results presentation: A brief explanation of
the CMCH, Ludhiana, Punjab. the format of the results as they will be pre-
l Study period: sented, e.g.:
 The entire period of the study including prep-  Pie charts

aration of the proposal, submission and ap-  Histograms

proval, training (where necessary), pretesting  Line graphs

(of the questionnaire), data collection, data  Tables

analysis, report preparation and dissemina- l Dissemination of the results:

tion of the findings.  Indicate the person or institution to whom

 If the study is in phases, each must be speci- the report is going to be submitted and why.
fied and the time for each given.  How does/do the investigator(s) propose to

l Sample size: disseminate their research findings, such as:

 Details on the sample size and how it has been - Attendance at local, regional or interna-
arrived at/worked out tional conferences/seminars, workshops
 Writing Proposals, Reports and Publications 283

- Holding of a dissemination seminar/work- l The investigator should indicate the amount

shop to be asked for and for what the institution(s)
- Publications in peer-reviewed journals, etc. under which the research project will be con-
l Ethical considerations: ducted will contribute.
 What ethical issues need to be addressed? l Each item should be quantified, if possible. If it
 How are they going to be addressed? is not possible (e.g. premises or space), it should
 This should include protection of human just be mentioned as an institutional contribu-
subjects (clients’ rights). tion. However small the institutional contribu-
l Possible constraints: tion is, it should be indicated.
 Any envisaged problems in undertaking the l Subtotals for each group should be indicated
study. and then the grand total shown.
 How these will be addressed, by whom and

when? Justification of the Budget

l Requirements: Details should be provided on the The investigators must indicate how they ar-
requirements of the research project, for example: rived at the amount of money being asked
 Personnel: All the people who will be in- for, and how it is going to be disbursed. For
volved and their individual roles example, a budget proposal for two principal
 Training: Of whom, why, when, where, by whom investigators for overall supervision of the proj-
 Paper: For the questionnaires, reports, etc. ect, data analysis, report production and dis-
 Transport: What form and for what semination: 10% of their time for 24 months
 Reagents: Which, how much of each and for @ Rs 30,000 3 2 (investigator) 5 Rs 1,440,000.
what A detailed budget planning for a research pro-
 Drugs: Which, how much of each, for what, posal is presented in Table 11.4.
their source(s)
 Space: How that will be obtained, where, when References
l The cited literature should be as current as pos-
Budgetary Estimates sible and include locally available information.
l Each line item should be quantified in mone- l These should be in the internationally accepted
tary terms. format (e.g. Vancouver style).

Table 11.4. An example of budget planning for a research proposal

1st 2nd 3rd Total
1. Staff
Junior research fellow (1) 120,000 120,000 120,000 360,000
Research assistant 84,000 84,000 84,000 252,000
(a) Total 204,000 204,000 204,000 612,000
2. Contingency
Recurring 500,000 300,000 200,000 100,000
Nonrecurring (equipment) 100,000 100,000 200,000 400,000
Travel (national) 20,000 20,000 20,000    60,000
(a) Total 620,000 420,000 420,000 1,460,000
(b) Total yearly 724,000 724,000 624,000 2,072,000
3. Overhead charges @10% Rs 2,07,200
(a) Grand total Rs 22,79,200
284 Research Methodology and Biostatistics

(c) to illustrate the conceptual framework that

USES OF COMPUTER IN MEDICAL
emerges from your data, especially in qualitative
RESEARCH research.
Use of computer in research is so extensive that it l Research design: Selection of appropriate re-
is difficult to conceive today a scientific research search designs may be achieved with the help of
project without a computer. In-depth and exten- computer. Computer programmes, new tech-
sive studies cannot be carried out without the use nologies and the internet can be used both to
of computer, particularly those involving complex facilitate research design planning and data col-
computations and data analysis. Computer in lection and to provide the context for a research
medical research is used at all stages of study from study. Word processing, spreadsheets and data-
the proposal stage to the presentation of findings. base programmes can all be applied to plan the
A brief description about the major uses of com- study.
puter in medical research is given here: l Sampling and sample size estimation: In ran-
l Research problem, purpose, main question or dom sampling, cryptographically secure ran-
hypothesis: The internet and electronic research dom number generators, such as Fortuna or
databases provide access to a large portion of the Yarrow can be used to generate true random
existing quantitative and qualitative literature. number selection to facilitate an unbiased ran-
This provides foundational data, theory and re- dom sample selection for experimental and
search findings to help shape your research pur- quasi-experimental studies. Computers can also
pose and investigate your selected problem in be used for the estimation of the sample size.
depth. By examining the literature, you can be- l Research instruments and data collection:
gin to identify the gaps in the literature, which Computer programmes, the internet and other
can help you to formulate your primary research technologies can also be used to both create
question(s) and/or hypotheses. and/or implement a research instrument in
l Literature review: This phase of the research both quantitative and qualitative researches.
process is also facilitated by this initial search, For example, emails or online surveys can be
since a large amount of full-text research papers conducted online using web forms or other
and articles are available through these elec- software so that participants can complete
tronic research databases. Access to these data- them and send them to the researcher instantly.
bases is usually achieved through academic and This often elicits a much higher return rate
health organizations or through personal or than mailed-out surveys. Software can also be
group subscriptions. Examples include Aca- used to create surveys that are completed in
demic Search Premier, EBSCO, MEDLINE, person or printed for mail-out. Biophysical
PsychINFO and JSTOR. There are also some measurements are monitored through com-
open-access databases available on the internet, puter only; these monitored physiological mea-
such as the Directory of Open Access Journals sures provide data for research. In addition, the
(DOAJ, available at http://www.doaj.org/). use of computers to facilitate interviews is be-
l Conceptual framework: Visual display software coming quite common, through the use of chat
can be used to create a visual representation in rooms, forum-like settings, ‘talking’ computers
the form of concept maps, flow charts or visual that use Flash-based, narrative-led interviews
models of your research conceptual framework and web video-recorded sessions. Researchers
(either qualitative or quantitative). This helps find these types of interviews very suitable
you to accomplish three results: (a) to visually when exploring sensitive issues, such as vio-
represent your main concepts and subconcepts, lence, sexual behaviour, maltreatment and in-
to organize and guide the research process plan- terpersonal interactions.
ning; (b) to visually explain your research pro- l Legal and ethical considerations: Computer
posal to reviewers and funding agencies; and may help in taking informed consent. All
 Writing Proposals, Reports and Publications 285

research, including computer-assisted or -focused products/productoverview/product_over-

research must include the means to provide view.htm.
enough information to research participants to  WEFT QDA (http://www.pressure.to/qda):

agree to give informed consent before being This open-source qualitative analysis tool
involved in the study. The actual consent form works much like N6 explained earlier, but in
can be created on a computer using a word- a simpler, more straightforward manner. It is
processing programme. It can also be included free to use, and allows the researcher to orga-
as a downloadable file through email or internet. nize narrative and interview data into themes,
A key consideration of all research studies is the matrices, frequency counts and to apply
issue of participant confidentiality and security Boolean queries.
of all collected data, both demographic and study  AtlasTi (http://www.atlasti.com): This ‘knowl-

related. When computers are used to collect the edge workbench’ software is excellent for ana-
data, this issue becomes even more critical, since lysing text, video, audio and other multimedia
potential illegal access to the information be- qualitative data.
comes more likely. Therefore, this must be han-  HyperResearch 2.6 (http://www.researchware.

dled sensitively. com/): This software also analyses both text

l Data analysis methods: One of the most pro- and multimedia data, enabling the researcher
found influences that technology has had on to code, retrieve, theorize and conduct data
research is in data analysis applications. Liter- analyses.
ally, hundreds of hours can be saved by using l Discussion of findings, recommendations, im-
the appropriate analysis software in both quan- plications, limitations and summary: Word-
titative and qualitative researches. More rigor- processing software like Microsoft Office or the
ous testing can be achieved through a huge open-source software Open Office can be used
reduction in analyst error. In quantities studies, to create text documents, spreadsheets, tables,
SPSS (http://www.spss.com/spss) has become graphs and charts as well as PowerPoint presen-
a very popular tool for data analysis. This soft- tations for preparing tables and visual displays
ware has been considered the premier paramet- of the research study findings.
ric and nonparametric data analysis software l Budget: An important part of any research pro-
for several decades. Once available only in posal is an accurate and easy-to-follow budget,
mainframe computers, it is now available in PC especially when the researcher is seeking funding.
versions also. Examples of statistics that can be Various spreadsheet software applications can
easily calculated: descriptive statistics, chi- help in this, including the spreadsheet available
square, correlation, t-tests and analysis of vari- in Open Office (mentioned earlier) or Microsoft
ance (ANOVA). In qualitative studies, data can Excel that comes with MS Office. In addition,
also be analysed using some selected software budget software, such as Quicken, can also be
packages, such as: useful for this part of the research process.
 QSR’s N6 and NVitro: QSR, an Australian l References and bibliography: Researchers often
company, offers two elite qualitative analysis use style and database software to help them
software packages particularly suited to ana- organize their reference resources and to format
lysing phenomenological, action, grounded their reference and bibliography lists correctly
theory and mixed methods data. They may be using the common citation styles, such as
accessed at http://www.qsrinternational.com/ Vancouver, APA or Chicago styles.
286 Research Methodology and Biostatistics

MULTIPLE CHOICE QUESTIONS

1. Which of the following is not one of the seven major 5. Ideally the abstract should be of approximately how
parts to the research report? many words?
(a) Results (a) 50–75
(b) Abstract (b) 75–100
(c) Method (c) 150–175
(d) Footnotes (d) 250–300

2. It is in this section that you fully interpret and evaluate 6. What should research study recommendations
your results: include?
(a) Introduction (a) No further need for research
(b) Method (b) No benefit for use in practice
(c) Results (c) Ways to change practice based on results
(d) Discussion (d) Ways to avoid using the results in other studies

3. Which is the factor that should determine whether 7. The definition of a research critique is understood
you decide to prepare a research report of your study to imply which of the following?
for a conference or for publication? (a) Analytical examination or commentary of a
(a) Whether the study is free from flaws research report
(b) Whether the study is important enough to jus- (b) A negative assessment related to the weakness of
tify presentation or publication a research report
(c) Whether others would be interested in the work (c) An analytical evaluation of the literature review
(d) All of the above (d) A positive assessment of the research design

4. As per Vancouver style of reference writing, while 8. Through which of the following communication
citing reference, after how many authors should the channel for dissemination of research finding can
et al. be used? one reach the largest percentages of the target group?
(a) 2 (a) Oral presentation in conference
(b) 3 (b) Poster presentation in conference
(c) 6 (c) Thesis/dissertation
(d) 9 (d) Journal articles

Answers of Multiple Choice Questions

1. d; 2. d; 3. d; 4. c; 5. c; 6. c; 7. a; 8. d.

REFERENCES
Adamsen, L., Larsen, K., Bjerregaard, L., & Madsen, J. K. (2003). tions to promote the implementation of research findings.
Danish research-active clinical nurses overcome barriers in re- British Medical Journal, 317, 465–468.
search utilization. Scandinavian Journal of Caring Science, Bjorkstrom M.E., & Hamrin E. K. F. (2001). Swedish nurses atti-
17(1), 57–65. tudes towards research and development within nursing.
Akinsaya, J. A. (1993). Melting research useful to the piecing Journal of Advanced Nursing, 34(5), 706–714.
nurse. Journal of Advanced Nursing, 19, 174–179. Bostrom, J., & Newton Suter, W. (1993). Research utilisation: the link
Berggren, A. C. (1996). Swedish midwives awareness of, atti- to practice. Journal of Nursing Staff Development, 9(11), 28–34.
tudes to and use of selected research findings. Journal of Brett, J. (1987). Use of nursing practice research findings. Nursing
Advanced Nursing, 23, 462–470. Research, 36(6), 344–349.
Bero, L. A., Grilli, R., Grimshaw, J. M., Harvey, E., Oxman, A. D., Bucknal, T., Copnell, B., Shannon, K., & McKinely, D. (2001).
& Thomson, M. A. (1998). Closing the gap between research Evidence-based practice: are critical care nurses ready for it?
and practice: an overview of systematic reviews of interven- Australian Critical Care, 14(3), 92–98.
 Writing Proposals, Reports and Publications 287

Canadian Nursing Informatics Association. (2003). Educating Manninen, E. (1999). Longitudinal study of Finnish nursing
tomorrow’s nurses: where is nursing informatics? ,http://cnia. students’ preferences for knowledge in nursing practice.
ca/OHIHfi naltoc.htm. Accessed 18.07.05. Scandinavian Journal and Caring Sciences, 13, 83–90.
Carroll, D. L., Greenwood, R., Lynch, K. E., Sullivan J. K., Ready, C. Marquis, B., & Huston, C. (1996). Leadership roles and manage-
H., & Fitzmaurice, J. B. (1997). Barriers and facilitators to the ment functions in nursing. Philadelphia: Lippincott-Raven.
utilisation of Nursing Research. Clinical Nurse Specialist, McKenna, H., Ashton, S., & Keeney, S. (2004). Barriers to
111(5), 207–212. evidence-based practice in primary care. Journal of Advanced
Closs, S. J., & Bryar, R. M. (2001). The BARRIERS scale: does it fit Nursing, 45, 178–189.
the current NHS research culture? NT Research, 6, 853–865. McKibbon, A. (1999). PDQ: evidence-based principles and practice.
Crane, J. (1995). The future of research utilization. Nursing Clinics Hamilton, BC: Decker Inc.
of North America, 30(3), 565–577. Melnyk, B. M., Fineout-Overholt, E., Stone, P., & Ackerman, M.
Cruickshank, D. (1996). The experience of action research in (2000). Evidence-based practice: the past, the present, and
practice. Contemporary Nurse, 5(3), 127–132. recommendations for the millennium. Pediatric Nursing,
Dick, B. (1997). A naive philosophy of action research. ,http:// 26(1), 77–80.
www.scu.edu.au/schools/gcm/ar/arp/naive.html.. Accessed Miller, J., & Messenger, S. (1978). Obstacles to applying nursing
12 March 2015. research findings. American Journal of Nursing, 78(4),
Druding, M. C. (1997). Re-examining the practice of normal sa- 632–634.
line instillation prior to suctioning. MEDSURG Nursing, 6(4), Nightingale, F. (1969). Notes on nursing. New York: Dover Publica-
209–212. tions, Inc.
Evans, D., & Pearson, A. (2001). Systematic reviews: gatekeepers of Nilsson, K., Nordstrom, G., Krusebrant, A., & Bjorvell, H. (1998).
nursing knowledge. Journal of Clinical Nursing, 10(5), 593–599. Barriers to and facilitators of research utilization, as perceived
Fink, R., Thompson, C. J., & Bonnes, D. (2005). Overcoming bar- by a group of registered nurses in Sweden. Journal of Advanced
riers and promoting the use of research in practice. Journal of Nursing, 27, 798–807.
Nursing Administration, 35(3), 121–129. Oranta, O., Routasalo, P., & Hupli, M. (2002). Barriers to and
Funk, S., Champagne, M., Wiese, R., & Tornquist, E. (1991). facilitators of research utilization among Finnish registered
Barriers: the barriers to research utilization scale. Applied Nurs- nurses. Journal of Clinical Nursing, 11, 205–221.
ing Research, 4(1), 39–45. Parahoo, K. (1998). Research utilization and research related ac-
Funk, S., Tornquist, E., & Champagne, M. (1995). Barriers and tivities of nurses in Northern Ireland. International Journal of
facilitators of research utilization. Nursing Clinics of North Nursing Studies, 35, 283–291.
America, 30(3), 395–407. Parahoo, K. (2000). Barriers to, and facilitators of, research utiliza-
Greenhalgh, T., & Douglas H. R. (1999). Experiences of general tion among nurses in Northern Ireland. Journal of Advanced
practitioners and practice nurses of training courses in Nursing, 31(1), 89–98.
evidence-based practice: a qualitative study. British Journal of Pearcy, P. A. (1995). Achieving research-based nursing practice.
General Practitioners, 49, 536–540. Journal of Advanced Nursing, 22, 33–39.
Hatcher, S., & Tranmer, J. (1997). A survey of variables related to Pearson, A., Puntin, S., & Durant, I. (1992). Nursing beds: an
research utilization in nursing practice in the acute care setting. evaluation of the effects of therapeutic nursing. St. Louis, MD:
Canadian Journal of Nursing Administration, 10(3), 31–53. Scutari Press.
Hicks, C. (1995). The shortfall in published research: a study of Retsas, A. (2000). Barriers to using research evidence in nursing
nurses’ research and publication activities. Journal of Advanced practice. Journal of Advanced Nursing, 32(3), 599–606.
Nursing, 21, 594–604. Rizzuto, C., Bostrom, J., Suter, W. N., & Chenitz, W. C. (1994).
Hicks, C. (1996). A study of nurse’s attitude towards research: a Predictors of nurses’ involvement in nursing research find-
factor analytic approach. Journal of Advanced Nursing, 23(2), ings. Image: Journal of Nursing Scholarship, 26, 143–147.
373–379. Rutledge, D., Royske, M., Greene, P., Nail, L., & Money, K. (1998).
Hunt, J. (1996). Barriers to research utilization: guest editorial. Barriers to research utilization for oncology staff nurses and
Journal of Advanced Nursing, 23(3), 423–425. nurse managers/clinical nurse specialist. Oncology Nursing
Kirkevold, M. (1997). Integrative nursing research: an important Forum, 25(3), 497–506.
strategy to further the development of nursing science and Schwenker, D. M., Ferrin, M., & Gift, A.G. (1998). A survey of
nursing practice. Journal of Advanced Nursing, 25(5), 977–984. endotracheal suctioning with installation of normal saline.
Lacey, E. A. (1994). Research utilization in nursing practice – a American Journal of Critical Care, 7(4), 255–260.
pilot study. Journal of Advanced Nursing, 19, 987–995. Silagy, C., & Haines, A. (1998). Evidence based practice in primary
Lekander, B. J., Tracy, M. F., & Lindquist, R. (1994). Overcoming care. London: BMJ Books.
the obstacles to research-based clinical practice. AACN Clinical Staggers, N., & Bagley Thompson, C. (2002). The evolution of
Issues, 5, 115–123. definitions for nursing informatics: a critical analysis and re-
Lindeman, C. (1975). Priorities in clinical nursing research. Nurs- vised definition. Journal of the American Medical Informatics
ing Outlook, 23(11), 693–696. Association, 9(3), 255–262.
Lofmark, A., & Thorell-Ekstrand, I. (2000). Evaluation by nurses Streubert Speziale, H., & Carpenter D. (2003). Qualitative research in
and students of a new assessment form for clinical nursing nursing: advancing the human imperative. Philadelphia, PA:
education. Scandinavian Journal of Caring Sciences, 14, 89–96. Lippincott.
288 Research Methodology and Biostatistics

Voda, A. M., Butts, S. V., & Gress, L. D. (1971). On the process Watson, B., Clarke, C., Swallow, V, & Forster, S. (2005). Explor-
of involving nurses in research. Nursing Research, 20(4), atory factor analysis of the research and development culture
303–308. index among qualified nurses. Journal of Clinical Nursing,
Walsh, M. (1997). Perceptions of barriers to implementing re- 14(9), 1042–1047.
search. Nursing Standard, 11, 34–37.
CHAPTER OUTLINE
Ethical Issues
in Biomedical
Research in India 12
INTRODUCTION COMPENSATION FOR PARTICIPATION IN RESEARCH

PRINCIPLES OF ETHICS IN BIOMEDICAL RESEARCH CONFLICT OF INTEREST IN RESEARCH

ETHICS AND REGULATION IN BIOMEDICAL RESEARCH GENERAL AND SPECIFIC PRINCIPLES FOR CLINICAL
INTERNATIONAL GUIDELINES EVALUATION OF DRUGS/CLINICAL TRIALS GUIDELINES
IN INDIA
ETHICS AND REGULATION IN BIOMEDICAL RESEARCH
NATIONAL GUIDELINES INSTITUTE ETHICAL COMMITTEE

PATIENT INFORMATION DOCUMENT AND INFORMED

CONSENT

LEARNING OBJECTIVES

This chapter is designed to enable the reader to: • Identify the rules of compensation for participation in
• Understand the concept of ethics in biomedical research. research and conflict of interest in research.
• Discuss the principles of ethics in biomedical research. • Appraise the general and specific principles for clinical
• Explain the national and international guidelines for eth- evaluation of drug/clinical trial guidelines in India.
ics and regulations of biomedical research.
• Describe the components and brief guidelines for pre-
paring patient information documents, informed con-
sent and institute ethical committee.

KEY TERMS

• Ethical principles 290 • National ethical guidelines • Conflict of interest in research 302
• International ethical for biomedical research 297 • General ethical principles 303
guidelines for biomedical • Indian Council of Medical • Specific ethical principles 303
research 291 Research 307 • Clinical trials 303
• Declaration of Helsinki 291 • Patient information • Drug trials 303
• Code of Nuremberg 292 document 299 • Adverse reactions 307
• Declaration of Geneva 293 • Informed consent 300 • Institute ethical committee 307
• Declaration of Helsinki • Compensation for participation
Amendment 293 in research 302
290 Research Methodology and Biostatistics

INTRODUCTION PRINCIPLES OF ETHICS IN

The Greek word ‘ethos’ is the root of the term BIOMEDICAL RESEARCH
‘ethics’, which means custom or character. The The major ethical principles which should be con-
research conducted by the health care profession- sidered in planning or reviewing research studies
als involves human beings. Thus, it becomes es- are principles of beneficence, justice and respect
sential for them to know what is right and wrong of human dignity.
in conducting research on human subjects. Ethics
is the systematic study of values and guides the PRINCIPLES OF BENEFICENCE
health care professional researchers about what is This is the most important ethical principle in
right and what is wrong in the conduction of the medical research, where every researcher must
research. ensure the following:
In initial periods of the health sciences re- l Establishing the positive risk–benefit ratio,

search, it involved no or very little awareness of where the risk of the research should never
ethical principles of research. However, progres- exceed expected benefits for people from
sively, it was observed that there is inhuman the knowledge generated by the research
handling of vulnerable human subjects used in activity.
research by health care researchers, which has led l The potential risk of the research study must

to formulation of ethical guidelines in various be carefully assessed and participants pro-

parts of the world. The cruellest example of in- tected from any harmful effect of the research
human handling of human subjects in research is activity.
the experiments carried out by the Nazi doctors l In addition to physical harm, study subjects

on prisoners, where they kept the vulnerable hu- should also be protected from expected adverse
man subjects in compressed chambers and freez- psychological consequences caused by the re-
ing water. They created gunshot wounds and search study, e.g. psychological or emotional
even transplanted grafts among twins to see the distress caused from self-disclosure, introspec-
body’s response in such adverse situations. More- tion, fear of the unknown or interacting with a
over, they have also conducted ante-mortum stranger.
dissection on these human subjects to assess the l Research must be conducted by a scientifically

effect of the experimental interventions on their qualified expert to avoid undue discomfort or
body. The observations of these experiments distress to study participants.
stunned the whole world and it triggered the idea l Participants must be provided with maximum

for formulation of Nuremberg Code in Germany physical, psychological, social and religious
to prevent such cruelty on human subjects in comfort, and undue disturbance and time utili-
future. zation of the subjects should be avoided.
This was the first international code of ethics
for biomedical research, which gave guidelines PRINCIPLE OF JUSTICE
for research on human subjects. It included ten This ethical principle directs researchers
principles to be followed by biomedical re- to abide by the participant’s right of fair treat-
searchers e.g. obtaining a voluntary consent ment and maintenance of privacy and confi-
from human subjects, freedom for the subjects dentiality. Therefore, researchers must ensure
to withdraw from experiment, availability of the following:
preclinical data before human subjects are ex- l The fair and nondiscriminatory selection of

posed to the experiments and not to conduct participants, such that any risk and benefits, will
any experiment on human subjects which could be equally shared by study participants. Partici-
lead to major harm/injury or death among pant’s selection should be based on research re-
them. quirement and not on convenience, gullibility
 Ethical Issues in Biomedical Research in India 291

or compromised position of certain type of illegal medical human experiments dates back to
people. the World War II Nuremberg trial (1946–1947),
l The nonprejudicial treatment of individuals also known as the Doctor’s trial, where the World
who decline to participate or who withdraw War participants alleged that incriminated experi-
from the study after agreeing to participate ments were similar to that in prewar testing. The
l Anonymity of participants and confidentiality comprehensive definition of legal and illegal ex-
of information must be maintained. periments was published in the Code of Nurem-
l No information collected from study participants berg (CoN) in 1947, later incorporated in medical
can be used for other than research purpose. research with the key statement that voluntary
l Vulnerable subjects, such as children, pregnant contribution of any participant must be based on
women, mentally ill patients, physically dis- informed consent. CoN is still considered as the
abled, terminally ill and institutionalized (pris- cornerstone of human research ethics. Later in
oners) people, who are conveniently and easily 1948, World Medical Association (WMA) formu-
accessible, must be protected from overuse and lated the well-known Declaration of Geneva
undue use for research purpose. (DoG), which emphasized the efforts of the physi-
cian towards humanitarian aims of medicine
PRINCIPLE OF RESPECT OF HUMAN DIGNITY which is a revised formulation of the Oath of
This principle of ethics emphasizes freedom of Hippocrates.
choice, where participants have the right to accept Another unethical Tuskegee Syphilis Study
or reject being part of the research study. In addi- (1932–1972) was convicted with abuse of vulner-
tion, volunteers must be provided full informa- able subjects in medicine and was seen as the
tion about the research study and an informed symbol of racism. This resulted in the identifica-
consent must be obtained, preferably a written tion of principles for research which was
one. Therefore, under this principle, the researcher mentioned in the Belmont report released in
must ensure the following: 1979 by the National Commission for Protection
l Participants have full right to question the re- of Human Subjects of Biomedical and Behav-
searcher for any additional information or clar- ioral Research. The report identified respect for
ification of doubts. person, beneficence and justice as guidelines for
l Participants have the right to quit the study at formulating any research. Accurate knowledge
any stage of the research study. Participants and awareness of the research problem does not
should be allowed to leave the study without necessarily imply willingness to participate in the
assigning any cause. biomedical research; rather, several structural
l A fully informed consent must be taken from and sociocultural factors by themselves influence
the participants. In case of fetus, infant, toddler, the willingness to participate.
younger child or psychological, neurological or
physical inability to give informed consent, this DECLARATION OF HELSINKI
can be obtained from parents or legal guardians. The most influential and respected document in
In case of a child aged between 7 and 18 years, research ethics was developed in the year 1964 by
an assent may be obtained. WMA which consisted of a set of ethical princi-
ples for conduct of medical experiments and
ETHICS AND REGULATION was called the Declaration of Helsinki (DoH).
This document enlightened the formulation of
IN BIOMEDICAL RESEARCH –
guidelines for physicians committed to medical
INTERNATIONAL GUIDELINES research projects. Focussing on the rights of
Ethics is the central dogma of medical research. research subjects, strict rules of voluntary contri-
Participating human subjects need to be particu- bution as described by CoN and personal in-
larly alert about the research. Issue of legal and formed consent of each participant were
292 Research Methodology and Biostatistics

modified ensuring subject’s protection as the experiment; the method and means by which it is
researcher’s responsibility. Moreover, under cir- to be conducted; all inconveniences and hazards
cumstances of inability of the subject to provide reasonably to be expected; and the effects upon
autonomous consent, a legal representative or a his health or person which may possibly come
responsible relative could provide consent on his/ from his participation in the experiment. The
her behalf. This made the research on children duty and responsibility for ascertaining the qual-
(especially for vaccines) and incompetent/captive ity of the consent rest upon each individual who
population permissible. Since its approval, DoH initiates, directs or engages in the experiment. It is
has undergone six revisions which improved and a personal duty and responsibility which may not
incorporated solutions to the problems faced dur- be delegated to another with impunity.
ing the study. The first revision in 1975 incorpo- 2. The experiment should be such that fruitful
rated the concept of supervision and control of the results could be achieved for the good of soci-
study which lead to defining and positioning of ety, unprocurable by other methods or means
independent committees such as institutional re- of study, and not random and unnecessary in
view boards and research ethics committees (ECs). nature.
Later, comparator participants’ right to get 3. The experiment should be so designed and
best available care during the therapy studies was based on the results of animal experimentation
emphasized. Additionally, each research proposal and knowledge of the natural history of the
needs to clearly state that DoH had been ac- disease or other problems under study that the
cepted and fulfilled. Significant extensions were anticipated results will justify the performance
implemented in the fourth revision of 1996 which of the experiment.
included the addition of the phrase ‘This does 4. The experiment should be conducted so as to
not exclude the use of inert placebo in studies avoid all unnecessary physical and mental
when no proven diagnostic or therapeutic method suffering and injury.
exists’ in the placebo-controlled trial under 5. No experiment should be conducted when
Article II.3. The fifth revision issued in 2000 there is an a priori reason to believe that death
provided additional modifications apart from or disabling injury will occur; except, perhaps,
changes in the structure and assembly of DoH. in those experiments where the experimental
physicians also serve as subjects.
THE CODE OF NUREMBERG 6. The degree of risk to be taken should never
(Source: The Nuremberg Code [1947] BMJ 1996; 313: 1448. exceed that determined by the humanitarian
Available at http://www.bmj.com/content/313/7070/1448.1) importance of the problem to be solved by the
1. The voluntary consent of the human subject is experiment.
absolutely essential. This means that the person 7. Proper preparations should be made and ade-
involved should have legal capacity to give con- quate facilities provided to protect the experi-
sent; should be so situated as to be able to exercise mental subject against even remote possibilities
free power of choice, without the intervention of of injury, disability or death.
any element of force, fraud, deceit, duress, over- 8. The experiment should be conducted only by
reaching or other ulterior form of constraint or scientifically qualified persons. The highest
coercion; and should have sufficient knowledge degree of skill and care should be required
and comprehension of the elements of the sub- through all stages of the experiment of those
ject matter involved as to enable him/her to make who conduct or engage in the experiment.
an understanding and enlightened decision. This 9. During the course of the experiment the human
latter element requires that before the acceptance subject should be at liberty to bring the experi-
of an affirmative decision by the experimental ment to an end if he/she has reached the physical
subject, the following should be made known to or mental state where continuation of the experi-
him/her: the nature, duration and purpose of the ment seems impossible for him/her to be possible.
 Ethical Issues in Biomedical Research in India 293

10. During the course of the experiment the scien- including research on identifiable human
tist in charge must be prepared to terminate material and data. The Declaration is in-
the experiment at any stage, if he/she has tended to be read as a whole and each of its
probable cause to believe in the exercise of the constituent paragraphs should not be ap-
good faith, superior skill and careful judgment plied without consideration of all other
required of him/her that a continuation of the relevant paragraphs.
experiment is likely to result in injury, disabil- 2. Although the Declaration is addressed pri-
ity or death to the experimental subject. marily to physicians, the WMA encourages
other participants in medical research involv-
DECLARATION OF GENEVA ing human subjects to adopt these principles.
At The Time Of Being Admitted As A Member Of 3. It is the duty of the physician to promote
The Medical Profession: and safeguard the health of patients, includ-
 I Solemnly Pledge To Consecrate My Life To ing those who are involved in medical
The Service Of Humanity; research. The physician’s knowledge and
 I Will Give To My Teachers The Respect And conscience are dedicated to the fulfilment of
Gratitude That Is Their Due; this duty.
 I Will Practise My Profession With Con- 4. The Declaration of Geneva of the WMA
science And Dignity; binds the physician with the words, ‘The
 The Health Of My Patient Will Be My First health of my patient will be my first consid-
Consideration; eration’, and the International Code of Med-
 I Will Respect The Secrets That Are Confided ical Ethics declares that ‘A physician shall act
In Me, Even After The Patient Has Died; in the patient’s best interest when providing
 I Will Maintain By All The Means In My medical care’.
Power, The Honor And The Noble Traditions 5. Medical progress is based on research that
Of The Medical Profession; ultimately must include studies involving
 My Colleagues Will Be My Sisters And Brothers; human subjects. Populations that are un-
 I Will Not Permit Considerations Of Age, derrepresented in medical research should
Disease Or Disability, Creed, Ethnic Origin, be provided appropriate access to participa-
Gender, Nationality, Political Affiliation, Race, tion in research.
Sexual Orientation, Social Standing Or Any 6. In medical research involving human sub-
Other Factor To Intervene Between My Duty jects, the well-being of the individual re-
And My Patient; search subject must take precedence over all
 I Will Maintain The Utmost Respect For other interests.
Human Life; 7. The primary purpose of medical research
 I Will Not Use My Medical Knowledge Con- involving human subjects is to understand
trary To The Laws Of Humanity, Even Under the causes, development and effects of dis-
Threat; eases and improve preventive diagnostic and
 I Make These Promises Solemnly, Freely And therapeutic interventions (methods, proce-
Upon My Honour dures and treatments). Even the best current
interventions must be evaluated continually
THE DECLARATION OF HELSINKI AMENDMENT through research for their safety, effective-
2008 ness, efficiency, accessibility and quality.
A. Introduction 8. In medical practice and in medical research,
1. The World Medical Association (WMA) most interventions involve risks and burdens.
has developed the Declaration of Helsinki 9. Medical research is subject to ethical stan-
as a statement of ethical principles for med- dards that promote respect for all human
ical research involving human subjects, subjects and protect their health and rights.
294 Research Methodology and Biostatistics

Some research populations are particularly arrangements for post-study access by

vulnerable and need special protection. study subjects to interventions identified
These include those who cannot give or as beneficial in the study or access to other
refuse consent for themselves and those appropriate care or benefits.
who may be vulnerable to coercion or 15. The research protocol must be submitted
undue influence. for consideration, comment, guidance and
10. Physicians should consider the ethical, legal approval to a research ethics committee
and regulatory norms and standards for re- before the study begins. This committee
search involving human subjects in their must be independent of the researcher, the
own countries as well as the applicable inter- sponsor and any other undue influence. It
national norms and standards. No national must take into consideration the laws and
or international ethical, legal or regulatory regulations of the country or countries in
requirement should reduce or eliminate any which the research is to be performed as
of the protections for research subjects set well as applicable international norms and
forth in this Declaration. standards, but these must not be allowed to
B. Basic principles for all medical research reduce or eliminate any of the protections
11. It is the duty of physicians who participate for research subjects set forth in this Dec-
in medical research to protect the life, health, laration. The committee must have the
dignity, integrity, right to self-determination, right to monitor ongoing studies. The re-
privacy and confidentiality of personal searcher must provide monitoring infor-
information of research subjects. mation to the committee, especially infor-
12. Medical research involving human subjects mation about any serious adverse events.
must conform to generally accepted scien- No change to the protocol may be made
tific principles, be based on a thorough without consideration and approval by the
knowledge of the scientific literature, other committee.
relevant sources of information and ade- 16. Medical research involving human subjects
quate laboratory and, as appropriate, animal must be conducted only by individuals
experimentation. The welfare of animals with the appropriate scientific training and
used for research must be respected. qualifications. Research on patients or
13. Appropriate caution must be exercised in healthy volunteers requires the supervision
the conduct of medical research that may of a competent and appropriately qualified
harm the environment. physician or other health care professional.
14. The design and performance of each re- The responsibility for the protection of
search study involving human subjects research subjects must always rest with the
must be clearly described in a research physician or other health care professional
protocol. The protocol should contain a and never the research subjects, even
statement of the ethical considerations though they have given consent.
involved and should indicate how the prin- 17. Medical research involving a disadvan-
ciples in this Declaration have been ad- taged or vulnerable population or commu-
dressed. The protocol should include nity is only justified if the research is re-
information regarding funding, sponsors, sponsive to the health needs and priorities
institutional affiliations, other potential of this population or community and if
conflicts of interest, incentives for subjects there is a reasonable likelihood that this
and provisions for treating and/or com- population or community stands to bene-
pensating subjects who are harmed as a fit from the results of the research.
consequence of participation in the re- 18. Every medical research study involving hu-
search study. The protocol should describe man subjects must be preceded by careful
 Ethical Issues in Biomedical Research in India 295

assessment of predictable risks and bur- time without reprisal. Special attention
dens to the individuals and communities should be given to the specific information
involved in the research in comparison needs of individual potential subjects as
with foreseeable benefits to them and to well as to the methods used to deliver the
other individuals or communities affected information. After ensuring that the
by the condition under investigation. potential subject has understood the infor-
19. Every clinical trial must be registered in a mation, the physician or another appropri-
publicly accessible database before recruit- ately qualified individual must then seek
ment of the first subject. the potential subject’s freely given in-
20. Physicians may not participate in a re- formed consent, preferably in writing. If
search study involving human subjects the consent cannot be expressed in writing,
unless they are confident that the risks in- the nonwritten consent must be formally
volved have been adequately assessed and documented and witnessed.
can be satisfactorily managed. Physicians 25. For medical research using identifiable
must immediately stop a study when the human material or data, physicians must
risks are found to outweigh the potential normally seek consent for the collection,
benefits or when there is conclusive proof analysis, storage and/or reuse. There may be
of positive and beneficial results. situations where consent would be impossi-
21. Medical research involving human subjects ble or impractical to obtain for such research
may only be conducted if the importance or would pose a threat to the validity of the
of the objective outweighs the inherent research. In such situations the research may
risks and burdens to the research subjects. be done only after consideration and ap-
22. Participation by competent individuals as proval of a research ethics committee.
subjects in medical research must be vol- 26. When seeking informed consent for partici-
untary. Although it may be appropriate to pation in a research study the physician
consult family members or community should be particularly cautious if the poten-
leaders, no competent individual may be tial subject is in a dependent relationship
enrolled in a research study unless he or with the physician or may consent under
she freely agrees. duress. In such situations the informed con-
23. Every precaution must be taken to protect sent should be sought by an appropriately
the privacy of research subjects and the qualified individual who is completely inde-
confidentiality of their personal informa- pendent of this relationship.
tion and to minimize the impact of the 27. For a potential research subject who is
study on their physical, mental and social incompetent, the physician must seek
integrity. informed consent from the legally autho-
24. In medical research involving competent rized representative. These individuals
human subjects, each potential subject must not be included in a research study
must be adequately informed of the aims, that has no likelihood of benefit for them
methods, sources of funding, any possible unless it is intended to promote the health
conflicts of interest, institutional affilia- of the population represented by the po-
tions of the researcher, the anticipated tential subject, the research cannot instead
benefits and potential risks of the study be performed with competent persons,
and the discomfort it may entail, and any and the research entails only minimal risk
other relevant aspects of the study. The and minimal burden.
potential subject must be informed of the 28. When a potential research subject who is
right to refuse to participate in the study or deemed incompetent is able to give assent
to withdraw consent to participate at any to decisions about participation in research,
296 Research Methodology and Biostatistics

the physician must seek that assent in addi- preventive, diagnostic or therapeutic value
tion to the consent of the legally authorized and if the physician has good reason to
representative. The potential subject’s dis- believe that participation in the research
sent should be respected. study will not adversely affect the health of
29. Research involving subjects who are physi- the patients who serve as research subjects.
cally or mentally incapable of giving con- 32. The benefits, risks, burdens and effective-
sent, for example, unconscious patients, ness of a new intervention must be tested
may be done only if the physical or mental against those of the best current proven
condition that prevents giving informed intervention, except in the following
consent is a necessary characteristic of the circumstances:
research population. In such circumstances - The use of placebo, or no treatment, is
the physician should seek informed con- acceptable in studies where no current
sent from the legally authorized represen- proven intervention exists; or
tative. If no such representative is available - Where for compelling and scientifically
and if the research cannot be delayed, the sound methodological reasons the use of
study may proceed without informed con- placebo is necessary to determine the ef-
sent provided that the specific reasons for ficacy or safety of an intervention and
involving subjects with a condition that the patients who receive placebo or no
renders them unable to give informed con- treatment will not be subject to any risk
sent have been stated in the research proto- of serious or irreversible harm. Extreme
col and the study has been approved by a care must be taken to avoid abuse of this
research ethics committee. Consent to re- option.
main in the research should be obtained as 33. At the conclusion of the study, patients
soon as possible from the subject or a le- entered into the study are entitled to be
gally authorized representative. informed about the outcome of the study
30. Authors, editors and publishers all have and to share any benefits that result from
ethical obligations with regard to the pub- it, for example, access to interventions
lication of the results of research. Authors identified as beneficial in the study or to
have a duty to make publicly available the other appropriate care or benefits.
results of their research on human subjects 34. The physician must fully inform the pa-
and are accountable for the completeness tient which aspects of the care are related
and accuracy of their reports. They should to the research. The refusal of a patient to
adhere to accepted guidelines for ethical participate in a study or the patient’s deci-
reporting. Negative and inconclusive as sion to withdraw from the study must
well as positive results should be published never interfere with the patient–physician
or otherwise made publicly available. relationship.
Sources of funding, institutional affilia- 35. In the treatment of a patient, where proven
tions and conflicts of interest should be interventions do not exist or have been
declared in the publication. Reports of ineffective, the physician, after seeking ex-
research not in accordance with the prin- pert advice, with informed consent from
ciples of this Declaration should not be the patient or a legally authorized repre-
accepted for publication. sentative, may use an unproven interven-
C. Additional principles for medical research tion if in the physician’s judgment it offers
combined with medical care hope of saving life, re-establishing health
31. The physician may combine medical re- or alleviating suffering. Where possible,
search with medical care only to the extent this intervention should be made the ob-
that the research is justified by its potential ject of research, designed to evaluate its
 Ethical Issues in Biomedical Research in India 297

safety and efficacy. In all cases, new infor- the subject and to benefit all members of the
mation should be recorded and, where human species as well as to further the eco-
appropriate, made publicly available. logical and environmental well-being of the
planet.
ETHICS AND REGULATIONS IN II. Principles of voluntariness, informed con-
sent and community agreement by which
MEDICAL RESEARCH – NATIONAL research participants are fully informed about
GUIDELINES the research and its impact and risk on the
The Indian Council of Medical Research (ICMR) research participant and others; and whereby,
brought out the ‘Policy Statement on Ethical Con- the research participants retain the right to
siderations Involved in Research on Human Sub- refrain from further participation in the re-
jects’ in 1980 and revised these guidelines in 2000 search regardless of any legal or other obliga-
as the ‘Ethical Guidelines for Biomedical Research tion that might have been agreed to by such
on Human Subjects’. The rapid developments in human participants or someone on their be-
science and technology in India and the multicul- half, subjecting only to minimal recitative
tural Indian society with a multiplicity of health obligations of any advance consideration re-
care systems of considerably varying standards ceived and outstanding. Whenever any such
have raised the need of revision and in 2006, the research involves treating any community or
Ethical Guidelines for Biomedical Research on Hu- group of individuals as a research participant,
man Participants was developed, which is known these principles of voluntariness and in-
as the ICMR Code of Ethics for Biomedical Re- formed consent will apply, mutatis mutandis,
search. It shall consist of two main sections: (a) to the community as a whole and to each in-
Statement of General Principles on Research Using dividual member who is a participant of the
Human Participants in Biomedical Research, research or experiment. If the human partici-
which is applicable to all areas of biomedical re- pant is unqualified for giving consent, and the
search and (b) Statement of Specific Principles on research or experimentation to be conducted
Research Using Human Participants in Specific on such a person who is incapable to give as-
Areas of Biomedical Research, such as drug/herb sent is deemed essential, the principle of
trials, human genetics studies, transplant research, voluntariness and informed consent shall con-
research in assisted reproductive techniques and tinue to apply, and such consent and
epidemiological studies (detailed document may voluntariness shall be obtained and effected
be seen at the official website of ICMR, www.icmr. on behalf of such research participants by
nic.in). The 12 general principles that are common someone empowered and under a duty to act
to all areas of biomedical research and any research on their behalf. The principles of informed
using the human beings are as follows: consent and voluntariness are fundamental
I. Principles of essentiality by which the re- principles to be abided throughout the re-
search involving the use of human participants search and experiment, including its after-
is considered to be completely essential after math and applied use so that research partici-
duly examining all options by taking into ac- pants are continually updated of any and all
count the present knowledge in the proposed developments to the degree or extent that they
subject of research and after the proposed sub- affect them and others. However, without, in
ject has been duly examined and reckoned by any way, undermining the prime importance
an appropriate and responsible body of indi- of obtaining informed consent from any hu-
viduals who are external to the particular re- man participant involved in any research, the
search and who, after careful consideration, nature and form of the consent and the evi-
conclude that the proposed research is required dential requirements to prove that such con-
to advance the present knowledge pertaining to sent was obtained shall depend on the degree
298 Research Methodology and Biostatistics

and seriousness of the invasiveness into the concerned, or someone authorized on their
concerned human participant’s person and behalf; and after ensuring that the said hu-
privacy, health and life in general as well as man participants do not suffer from any
the overall objective and the significance of form of illness, discrimination, or stigmati-
the research. The ethics committee shall de- zation as a consequence of having partici-
cide on the form of consent to be taken or its pated in the research or experiment.
waiver on the basis of the degree of risk that V. Principles of precaution and risk minimi-
might be involved. zation by which caution and proper care is
III. Principles of nonexploitation through which taken at all levels of the research and experi-
as a general rule, research participants are ment (from its initiation as a research idea,
compensated for their involvement in the re- its subsequent research design, the conduct
search or experiment; and, irrespective of the of the research or experiment, and its appli-
social and economic condition or status or cative use) to ensure that the research par-
literacy or educational levels attained by the ticipant and those affected by it including the
research participants, they are completely up- community are put to the minimum risk,
dated of all the hazards occurring in and out suffer from no known irreversible adverse
of the research so that they can understand all effects, and generally, benefit from and by the
the physical and psychological risks as well as research or experiment; and that necessary
moral repercussions of the research whether steps are affected to ensure that both profes-
to themselves or others, including those yet to sional and ethical reviews of the research are
be born. Such human participants should be undertaken at appropriate levels so that fur-
selected so that the weight and benefits of the ther and specific guidelines are established,
research are distributed without arbitrariness, and necessary directions are stated, in respect
bias or whim. Each research shall include an to the conduct of the research or experiment.
in-built process for compensation to the par- VI. Principles of professional competence by
ticipants either through insurance cover or which the research is conducted at all times
any other suitable means to insure all foresee- by capable and qualified persons who act
able and unforeseeable risks by providing for with complete integrity and neutrality and
corrective action and all-inclusive aftercare, who have been made aware of, and are heed-
including treatment during and after the re- ful of, preferably through training, the ethi-
search or experiment, in respect to any effect cal considerations to be kept in mind with
that the conduct of research or experimenta- respect to such research or experiment.
tion might have on the human participant VII. Principles of accountability and transpar-
and to ascertain that immediate compensa- ency by which the research or experiment
tory and rehabilitative actions are effected to will be conducted in a fair, honest, unbiased
all affected, if and when necessary. and transparent manner after complete dis-
IV. Principles of privacy and confidentiality by closure is made by all those who are associ-
which the identity and records of the human ated with the research or experiment of each
participants of the research or experiment are facet of their interest in the research, and any
to the highest degree possible kept confiden- conflict of interest that might exist; and,
tial, and that no particulars of said human whereby, subject to the principles of privacy
participants that would result in the disclo- and confidentiality and the rights of the re-
sure of their identity are disclosed without searcher, full and complete records of the
valid scientific and legal reasons that might be research inclusive of data and notes are re-
crucial for the objectives of therapeutics or tained for such reasonable period as may
other interventions, without the specific con- be prescribed or considered necessary for
sent in writing of the human participants the purposes of post-research monitoring,
 Ethical Issues in Biomedical Research in India 299

evaluation of the research, conducting fur- contributing to the funding of the research,
ther research (whether by the initial re- the institution(s) where the research is
searcher or otherwise), and to make such conducted and various individuals, groups,
records available for scrutiny by the appro- or undertakings who fund, use, or derive
priate legal and administrative authority, if benefit from the research, market the prod-
necessary. uct (if any), or prescribe its use so that, inter
VIII. Principles of the maximization of the pub- alia, the effect of the research or experiment
lic interest and of distributive justice by is properly monitored and constantly sub-
which the research or experiment and its ject to review and corrective action at all
subsequent applicative use are conducted levels of the research and experiment and its
and used to benefit humanity and not only future use.
those who are socially better off but also XII. Principles of compliance by which there is
those who are the least advantaged; and in a general and positive duty on all individu-
particular, the research participants them- als conducting, associated or connected
selves and/or the community from which with any research entailing the use of a hu-
they are drawn. man participant to ensure that both the let-
IX. Principles of institutional arrangements ter and the spirit of these guidelines, as well
by which there shall be a duty on all indi- as any other norms, directions and guide-
viduals connected with the research to make lines that have been specifically established
sure that all procedures required to be or prescribed and are applicable for that
abided with and all institutional arrange- area of research or experimentation, are
ments needed to be made with respect to scrupulously observed and duly complied
the research and its subsequent use or ap- with.
plication are duly made in a bona fide and
transparent manner; and to take all suitable
PATIENT INFORMATION DOCUMENT
steps to guarantee that research reports,
materials and data connected with the re- AND INFORMED CONSENT
search are suitably preserved and archived. It is important for the researcher to provide ample
X. Principles of public domain by which the knowledge to the prospective participants about
research and any further research, experi- the study objectives, its methodology, details about
mentation or evaluation in response to, and the kind of involvement expected from the partici-
originating from, such research is brought pants and pros and cons of the study. Once the
into the public domain so that its results are document is read/understood by the participant,
generally made known through scientific only then should the informed consent be sought.
and other publications subject to such rights A brief guideline is given in the following to pre-
as are available to the researcher and those pare the patient information document (PID).
associated with the research under the law
in force at that time. Brief Guidelines for Preparing Patient Information
XI. Principles of totality of responsibility by Document
which the professional and moral responsi- 1. Study title and name/s of institution/s con-
bility, for the due adherence of all the prin- ducting the study
ciples, guidelines, or prescriptions estab- 2. Invitation paragraph – In the initial para-
lished generally or in respect to the research graph, it should be made clear to the partici-
or experiment in question, devolve on all pant that it would be his/her voluntary
those directly or indirectly connected with participation in the study, he/she should read
the research or experiment including the and/understand the information given in the
researchers, those responsible for funding or document and should participate in the study
300 Research Methodology and Biostatistics

if fully satisfied, and if there is any doubt, it surgery, treatment of all other ailments prefer-
can be clarified from the researcher. Name of ence in treatment, etc.) will be given to him/
the sponsoring agency can also be mentioned her if he/she takes part in the study.
in the initial paragraph. PID should also describe, in case of any side
3. Objectives and usefulness of the study – Aim effect/any untoward situation, what would
and objectives of the study can be described in be the compensation and how it can be re-
simple, nontechnical language. This paragraph ceived by him/her or family members.
should also define how this study can be ben- 8. Status of participants after trial/study com-
eficial for the larger population. pletes or stops in the middle – PID should
4. Brief research methodology – This paragraph describe whether patient’s treatment/procedure
should define the total number of participants will be continued after the stoppage of study or
in the study, criteria for selecting the partici- not, and whether this can affect him/her.
pants, total duration of study, etc. It should also 9. Procedure of complaints if participants
be defined how many times he/she would be faces anything wrong – It should be clearly
required to come to hospital or how many time defined where complaint can be registered
he/she may be enquired/asked to provide in- and who is the authority for hearing such
formation in reference to the study. What kind complaints. For example, a patient may com-
of intervention (drug/device/procedure) will plain against doctor/nurses regarding their
be performed on him/her (in case of interven- behaviour.
tional trials), etc. In the PID, research design 10. Maintenance of confidentiality of the data –
can also be briefly described in simple language PID should describe what kind of patient data
especially if it is an interventional study or will be collected, who all can see that data
blinding is done. (medical officer, researcher, sponsoring agency,
5. Benefits and risks of the participation – It etc.). PID should also put a statement that the
should be clearly described in the PID about patient’s identity should not be disclosed at
the potential benefits and risks involved. any point of publication of data.
He/she should also be informed about the 11. Contact details of the researcher – PID
time requirement, e.g. time for conducting should mention the name of the PI, address,
interview/time for conducting procedure. telephone numbers and name of the Member
There should also be clear-cut description of Secretary of the Ethics Committee and ad-
potential side effects. It should also be men- dress with telephone numbers
tioned that participant will have full right to At the end, the researcher may express his
leave the study at any time without assigning gratefulness to the participant for his or her coop-
the cause. eration and for sparing his/her valuable time and
6. Responsibility of participants – Study partici- should sign the document with date.
pants should be clearly informed about the
instructions they are required to follow while INFORMED CONSENT
they are under study, e.g. for checking fasting Informed consent is more than a form or a signa-
blood sugar level, for how many hours they are ture; it is a process of information exchange that
expected to fast and can they take water or not? includes recruitment material, verbal dialogues,
If the patient has to follow certain restriction presentation of written materials, questions and
like abstinence, avoid certain food items, cer- answers, and an agreement that is documented by
tain life style changes, etc., then these should be a signature. In biomedical research involving hu-
clearly defined. man participants, the investigator must obtain
7. Compensation for participation – Participants the informed consent of the prospective partici-
should be clearly informed what kind of mon- pant or in the case of an individual who is not
etary or any other benefit (free treatment, capable of giving informed consent; the consent
 Ethical Issues in Biomedical Research in India 301

of a legal guardian is must. Informed consent state, or when he or she is mentally incompe-
protects the individual’s freedom of choice and tent. As soon as the person regains conscious-
respects the individual’s autonomy; it is given ness or become mentally competent, a fresh
voluntarily to participate in research or not. Ad- informed consent should be taken; however,
equate information about the research is given in this procedure should be described in the in-
a simple and easily understandable unambiguous formed consent form beforehand.
language in a document known as the informed 3. Researcher decides to make changes in the du-
consent form. Generally, in the consent form, only ration of follow-up, methods of intervention,
the essential content of undertaking is provided; procedure, number of visits, etc.
detailed information should be provided as PID 4. Researcher must take fresh consent if there are
to all the study participants. any chances of revelation of identity through
The contents of an informed consent must in- data presentation or photographs. Researcher
clude the following: should take utmost effort to ensure confidenti-
l Title of study and name(s) of investigator(s), ality; however, in such situations, the photo-
name of the institution conducting the study graph should be adequately covered, names
l Introduction and invitation to subjects should be changed.
l Purpose of the research study

l Basis of subject selection Waiver of Informed Consent

l Procedural process and any intervention in- It is mandatory to obtain a voluntary informed
volved in the study consent from each research participant. However,
l Expected time and nature of participants’ in- it may be waived off in the situation where risk on
volvement in the study the participant is very minimal, emergency condi-
l Type of data/information to be obtained for tion with involved risk and participants are duly
respondents protected for both privacy and confidentiality and
l Details of the potential discomfort or risk for do not violate the rights of participants. Thus, IEC
participants in the study may waive off informed consent in the following
l Details of the support available in case any situations:
harm occurs to participants l An emergency situation, where it is practically

l Mechanisms to protect privacy and anonymity not possible to obtain surrogate consent
of participants l In a situation where it is not practically possible

l Contact detail to collect information and clari- to conduct research since confidentiality of per-
fication about doubts sonally identifiable information has to be main-
l Statement about voluntary participation of sub- tained throughout research as may be required
jects and right of termination without any by the sensitivity of the research objective, e.g.
penalty study on disease burden of sexually transmitted
l Description of the expected benefits of the re- diseases
search study l In a situation where research study is to be con-

l Signature lines (to obtain the signature of the ducted on anonymized biological samples from
subject and counter-signature by investigator deceased individuals, left over samples after
who obtained the consent along with the date) clinical investigation, cell lines or cell-free de-
Situations for taking fresh consent or reconsent rivatives, like viral isolates, DNA or RNA from
are as follows: recognized institutions or qualified investiga-
1. When research is bound to make changes in tors, samples or data from repositories or regis-
protocol in the light of new information. tries
2. In a few cases, someone may have given the in- l In a research study where data to be obtained

formed consent on behalf of the study partici- from publicly available information, documents,
pants when he or she is in the unconscious records, works, performances, reviews, quality
302 Research Methodology and Biostatistics

assurance studies, archival materials or third l In case of an abandoned institutionalized indi-

party interviews, service programmes for benefit vidual or person under judicial custody, sur-
of public having a bearing on public health pro- rogate consent may be obtained from autho-
grammes, and consumer acceptance studies rized relative or legal custodian or head of the
There are ethical dilemmas where taking in- institute.
formed consent may distort the actual situation and l A surrogate consent also may be obtained, if the
research finding. Given here are a few examples: participant loses consciousness or competence
l A study on nurse’s time utilization in produc- of consent during research study.
tive and nonproductive activities during work- l Investigator must ensure that none of the
ing period patient–clinician relationship is affected and
l Behaviour of medical officer with patients and treatment/care is influenced by the decision
their family in surgical wards of the individual to participate or not in a
l Performance observation of the medical intern particular research study.
in medical ward
Such ethical dilemma must be dealt with care- COMPENSATION FOR PARTICIPATION
fully, where surrogate consent may be obtained
through competent authority of the institute until
IN RESEARCH
research does not actually harm the participants or Each study participant should be adequately com-
breach their privacy. However, voluntary consent pensated for various reasons as follows:
from each participant is always a requirement for 1. The most important issue is expenditure of
every research proposal. time because of his involvement in the study.
This can lead to loss of wages, doing extra
Responsibilities of an Investigator in Obtaining arrangements for his absence or taking leave
Informed Consent from the job, etc. so he/she should be given
The investigator carries out the following responsi- adequate compensation for that.
bilities adequately in obtaining the informed consent: 2. He/she should also be compensated for the
l The investigator must provide all the necessary expenditure incurred for travelling to the place
and desired information to the participant and of study.
it is the right of participants to know about each 3. He/she should be given free medical services
aspect of the research; this knowledge must be for any side effect/complaint arising out of the
adequately provided. disease under study; however, for any other
l It must be ensured by the investigator that none complaint he/she may be given primary care
of the participants is unduly influenced or with appropriate referral.
intimidated. 4. If participants withdraw from the study
l The investigator should ensure that preferably because of medical reasons, he/she should be
informed written consent is obtained only after given full compensation; otherwise, partici-
adequate information is provided to the partici- pants withdrawing because of any other rea-
pant; the participant must not be given any false son should be paid an amount as per the days
assurance or unjustifiable influence to affect his/ of participation.
her decision; the consent should preferably be 5. The researcher should built in all these expendi-
witnessed by a person not related to the trial. If a ture in his budgetary plan and get them approved
participant cannot provide consent, a legal by the respective committees including EC.
guardian or duly authorized representative may
give consent on behalf of the participant.
l A verbal informed consent may be obtained, CONFLICT OF INTEREST IN RESEARCH
when participant refuses to sign or give thumb ‘Conflict of interest’ refers to situations in which
impression or cannot do so. However, it is better financial or other personal benefits may influence,
to document the consent through audio or video. or have the appearance of influencing the
 Ethical Issues in Biomedical Research in India 303

researcher’s professional judgment in conducting guidelines will help protect the interest of the re-
or reporting research, e.g. not ensuring patient’s search participant by protecting him/her against
welfare and skewed interpretation of data in fa- any avoidable risk, guide the researchers in prepara-
vour of sponsoring agency. tion of research proposals/protocols and facilitate
Therefore, it is important for the researcher to the ethical committee to review and finally approve
disclose the source of research funding while sub- such kind of studies. Guidelines for clinical trials
mitting the proposal. They must also disclose any are as follows:
personal relationships with people or organizations
which can manipulate (bias) work or its result or its GENERAL PRINCIPLES
reporting. Examples of potential conflicts of inter- Conducting research involving human participants
est can be promotion, employment, consultancies should be based on the four basic ethical principles of
and monetary gain. All institutes undertaking re- autonomy, beneficence, nonmalfeasance and justice.
search in collaboration with commercial agencies/ The researcher responsible for the human trial should
industries should strongly review such proposals take responsibility not only of the trial but also of the
and try to resolve these conflicts. protection of rights, and health and welfare of the
The researcher should also avoid undue encour- participants recruited for the study. He/she should be
agement of individual participants, families and qualified and competent in conducting clinical
populations through compensation. However, any research trials, capable of using different methods of
benefit as an outcome or in the process of conducting conducting such trial and aware of and able to comply
the study may be shared with them like technology with the steps and requirement of the protocol enu-
transfer, training, provision of health care services, merated in the general principles and general issues.
and reimbursement of travel and loss of wages.
SPECIFIC PRINCIPLES
Drug Trials
GENERAL AND SPECIFIC PRINCIPLES
Schedule ‘Y’ provides regulation for ‘requirements
FOR CLINICAL EVALUATION OF DRUGS/
and guidelines for permission to import and/or
CLINICAL TRIAL GUIDELINES IN INDIA manufacture of new drugs for sale or undertake
Increasing disease patterns and evolving complex clinical trials’. As per the revised Schedule ‘Y’ of
syndromes in the human race needs changing the Drugs and Cosmetics Act (2005), ‘a clinical
intervention modes. Concern about the safety, ef- trial is a systematic study of new drug(s) in hu-
fectiveness and usefulness of any kind of interven- man subject to generate data for discovering and/
tion needs to be addressed by research on therapies, or verifying the clinical, pharmacological (includ-
diagnostic procedures and preventive measures in- ing pharmacodynamic, and pharmacokinetics), of
cluding primordial and secondary prevention. the investigational product(s) and/or identify any
Therefore, the experimental procedure has to be adverse effects with the objective of ascertaining
more complex, varied and multifaceted. It is also its safety and/or efficacy’.
accepted that it is essential to carry out research on Clinical trial of drugs is a randomized single-
human volunteers to discover better medical and or double-blind controlled study in human par-
therapeutic modalities for the betterment of ticipants, designed to evaluate prospectively the
humanity. Working on today`s evolved human par- safety and effectiveness of new drugs/new formu-
ticipants is associated with some degree of risk to lations before the product can be put in to the
the concerned individual. With this frame of refer- market. As per the Drug and Cosmetic Rules 1945
ence, guidelines have been developed to carry out (DCR) and subsequent amendments, a new drug
evaluation of drugs, vaccines, devices and other includes the following:
diagnostic materials on human participants includ- 1. A new chemical entity (NCE).
ing stem cells therapy and nanotechnology in 2. Drug which is approved for certain indication
accordance with basic ethical principles. These in a particular dose regime and by a particular
304 Research Methodology and Biostatistics

route and is being investigated for another ii. Clinical trials of drugs without approval of
indication, by another route in a different dose the Indian regulatory authority and appro-
regime. priate agencies should be condemned and
3. A combination of two or more drugs which, dealt strongly and punished under the court
although approved individually, is proposed to of law.
be combined for the first time in a fixed dose iii. If the drug is found to be effective, it should
combination (FDC). be mandatory for the sponsoring agency to
The clinical trial can be treatment, preventive, provide the drug to the participants till it is
diagnostic, screening trial or a quality-of-life trial. marketed in the country and thereafter at a
reduced rate, whenever possible. A suitable
Approval a priori agreement should be reached on
The proposed clinical trial can be initiated only post-trial benefits.
after approval from the Drug Controller General iv. The criteria for termination of a trial must
of India (DCGI) which is mandatory as per the be defined a priori in the proposal of the
Schedule ‘Y’ of the Drugs and Cosmetics Act 1940. trial and plan of interim analysis must be
The researcher needs to get the proposal approved clearly presented. With the interim analysis
by the institutional ethical committee at the site showing that the trial drug is more effective/
level. If the services of a laboratory or facilities or less effective than the standard drug, such
outside India are to be availed, its name, address agreement will ensure better drug to the
and specific services to be used should be stated in patients receiving less effective drug follow-
the protocol to avail DCGI approval to send the ing discontinuation of trial.
sample outside the country. Also information of v. Issues of partner notification and discor-
the services availed from other than the investiga- dant couples should be taken care of before
tional site must be furnished to the DCGI prior to initiating any HIV-/AIDS-related trial.
initiation of trial. vi. Following preclinical testing (laboratory
Compliance to the guiding principles should screening and animal testing) and discovery
be maintained irrespective of whether the drug of new drug formulation in India, clinical
has been developed in this country or abroad or trials are required to be carried out in India
whether clinical trials have been carried out out- from phase I through phase III and data
side India or not. should be submitted as required under
Distinction between the therapy and research items 5, 6 and 7 respectively of Appendix I
should be maintained through the drug trial. of the revised Schedule ‘Y’ of Drugs and
Therefore, the investigator who participates in the Cosmetics Act 1940.
research by administering the new drug to Permission to carry out these trials shall
consenting participants should ensure that generally be given in stages, considering
participant understand and remember that the the data emerging from earlier phase(s).
drug is experimental and that its benefits for the vii. For new drug substances discovered in
condition under study are yet to be proven. countries other than India, phase I data,
as required under item 5 of Appendix I,
Special Considerations from other country(ies) should be sub-
i. Protocols using the placebo in drug trial or mitted along with the application.
sham surgery require careful consideration as it After phase I data generated outside India
is unethical to deny available treatment to the have been submitted to the Licensing Au-
control (placebo) group of patients. It has been thority, permission may be granted to
intensely questioned and even the Helsinki conduct phase II and phase III trials con-
Declaration (fifth version) has not defined and currently with other global trials for that
given any clear verdict to such a query. drug.
 Ethical Issues in Biomedical Research in India 305

viii. In case of amendment or deviation in the pharmacodynamics of a new drug. It helps determine
protocol, not only the approval of IEC has the maximum tolerable dose and nature of adverse
to be obtained but also the licensing au- drug reactions (ADRs) that can be expected in healthy
thority has to be notified of the same. In adults of both sexes. Female volunteers should be in-
order to optimize and expedite the develop- cluded provided they have completed their family or
ment of drugs indicated in life-threatening/ do not intend to bear children in future.
serious diseases or specific diseases of rele- These studies include single and multiple dose
vance to India, the toxicological and clinical administration and should be carried out at a
data requirements shall be decided on a well-equipped site. Initiating of trial requires the
case-by-case basis. In such cases, data of following:
particular studies may be abbreviated, de- 1. Each dose should be administered to at least
ferred or omitted, as deemed appropriate two participants to establish the safe dose range
by the licensing authority and not by ECs. using maximum dose tolerated, and ADRs if
The Indian Good Clinical Practices (GCP) based any, recording their intensity and nature.
on the international guidelines issued by WHO and 2. Since it involves testing on human subjects for
the International Committee on Harmonization the first time, it is safer to use a cohort of vol-
(ICH) provide operative guidelines for ethical and unteers. It should begin with the minimum
scientific standards for designing, conduct, perfor- dose which should be increased only after its
mance, monitoring, auditing, recording, analysis safety is established.
and reporting of clinical trials. It also provides 3. With a readily measurable short duration of
tassurance that the data and reported results are drug exposure, study of preliminary outcome of
credible and accurate and that the integrity, rights therapeutic benefit of the drug may be con-
and confidentiality of trial subjects are protected. ducted in phase I as a secondary objective. It can
(http://cdsco.nic.in/html/GCP.htm.) also be carried out on patients if the drug is cy-
The clinical trials usually are of three types: totoxic or if a quicker result is needed, e.g. HIV.
i. Specific demarcated research on effect of inter- 4. Pharmacokinetics, i.e. drug absorption distribu-
vention, e.g. phase I trial of a new compound tion, metabolism and excretion (ADME), should
ii. Clinical trials with mixed objectives of stan- be performed to support and develop formula-
dard medical practice and specific research tion of drug across different age groups and
question, e.g. trials of two competing antinau- their dosing recommendations. Obtaining phar-
sea drugs following standard chemotherapy macokinetic information in subpopulations
iii. Studies involving research on therapeutic such as patients with impaired elimination (re-
practices, such as the trial of two already ap- nal or hepatic failure), the elderly, children and
proved antidiabetic drugs for withdrawal and ethnic subgroups should also be considered.
restriction 5. Pharmacodynamics: Considering the drug and
the end point studied, pharmacodynamic stud-
ies which relate to blood levels of drug to re-
PHASES OF CLINICAL TRIALS sponse may be conducted in healthy volunteers
All four phases of clinical trials require approval or in patients with the concerned disease. It
from the ethical committee – the first three phases may help to formulate dosage and dose regi-
requiring DCGI clearance in addition to EC men in later studies.
approval. 6. An investigator trained in clinical pharmacol-
ogy should preferably carry out these studies.
PHASE I (HUMAN PHARMACOLOGY) 7. Conducting two trials on the same volunteers
This is the first stage of testing in human subjects and should be done with a time lapse of a mini-
is a nontherapeutic trial with an objective to deter- mum of 3 months and the volunteer should be
mine the safety–tolerability–pharmacokinetics and covered under an insurance scheme.
306 Research Methodology and Biostatistics

8. Compensation is given by the sponsors of PHASE III (THERAPEUTIC CONFIRMATORY

newly developed drugs. The amount may vary TRIALS)
depending upon the discomfort experienced This large-scale testing is done to acquire adequate
by the participant and the number of samples data about the efficacy and safety of drug for pa-
taken or being subjected to procedures. EC tients of either sex. It is conducted at multiple
should ensure that there is no unnecessary in- centres usually in comparison to a standard drug
ducement or misconduct on the part of both and/or a placebo if the standard drug does not ex-
the sponsors and the participants. ist for the disease under study. It reconfirms and
9. There should be adequate safeguards for man- validates the efficacy and safety of the drug found
agement of ADRs, including resuscitative mea- in phase II. Following a successful phase III trial,
sures as in intensive care. approval for marketing of the drug is granted.
Since this phase of the study further explores
COMBINED PHASE II AND PHASE III the dose–response relationship to drug concentra-
Such trials are conducted on large population tion in blood and clinical response in a wider
(,100–200) for whom the therapeutic options population in different stages of disease, safety
are exhausted, e.g. case of HIV/AIDS and cancer. and efficacy of the drug per se and in combination
Also toxic drugs are not tested in healthy volun- with other drugs can also be established. Drugs
teers as in phase I studies because their risk to given for longer duration and their exposure can
benefit ratio is more and therefore they are planned be studied in such trials although they may be
only in patients suffering from the targeted disease. initiated in phase II. The results of this trial helps
Since these patients are vulnerable, the study has to complete the prescribing information needed to
be planned carefully. EC has an important role in support adequate instruction for the use of drugs.
these studies as the risk to benefit ratio may influ- These trials may be carried out by clinicians in the
ence the decision and participation of terminally ill concerned therapeutic areas having facilities ap-
patients. The researcher also has to consider very propriate to the protocol. If the drug is already
carefully the risks involved. approved/marketed in other countries, phase III
data should generally be obtained in sufficient
PHASE II (THERAPEUTIC EXPLORATORY number of patients distributed over adequate
TRIALS) number of centres, primarily to confirm the effi-
These controlled studies are done on a limited cacy and safety of the drug in Indian patients
number of patients of both sexes to determine the when used as recommended in the prescribing
therapeutic effect and the effective dose range of information. Open noncomparative trials do not
a drug. The study population selection is there- generate any generalized data and, therefore, are
fore very precise with narrow inclusion criteria unethical.
and is nearly homogenous. The dose used is less
than the highest dose used in phase I. It also PHASE IV
evaluates the potential end points, therapeutic Phase IV studies should have valid scientific objec-
regimes (concomitant medications and target tives. Following market approval of the new drug
population) and mild versus severe disease, for during the phase IV trial, postmarketing surveil-
further studies in phase II or III. It can be brought lance is undertaken to obtain additional informa-
about by detailed analysis of the subsets of data tion about the risks and benefits of long-term usage
and by including multiple end points in the trials. of drug. Its major benefit is that the ADRs induced
Normally, 20–25 patients should be studied for by drugs, if any, should be brought to the notice of
assessment of each dosage. These studies are usu- the ethics committee. Correlating the adverse effect
ally limited to three to four centres. It is advisable reported during phase IV trial with the adverse
to include a clinical pharmacologist as a effect during the animal trial will help to draw
co-investigator in such studies. markers for future warning of adverse events (AEs)
 Ethical Issues in Biomedical Research in India 307

that is likely to occur with other drugs. This phase of All other serious unexpected reactions (ADRs)
trial is required by the licensing authority for opti- that are not fatal or life-threatening must be filed
mizing its use of the studied drug. This phase of trial as soon as possible but not later than 14 calendar
also includes studies on specific pharmacological days. At the end of the trial, all AEs whether
effect, drug interactions, dose–response studies, related to trial or not are to be listed, evaluated
which are neither studied on patient population nor and discussed in detail in the final report.
adequately studied in premarketing phase. Bio- The investigator owns the responsibility
equivalence and bioavailability study also falls un- of medical management of the AE and protocol
der this category. In addition there are phase IV of the management of these events and allocation
studies that are designed to evaluate the marketed of responsibilities must be predefined in the pro-
drug in specifically designed studies, which have tocol submitted to the ethical committee. There
inclusion/exclusion criteria, objectives and end must also be a financial plan (including, if neces-
points. The drug is used for the labelled indication sary, insurance) to manage AEs and compensation
in these studies and hence do not require approval for trial-related injury. The EC reviewing the
from the licensing authority; however, EC permis- protocol must review these aspects as well before
sion is needed. Another postmarketing study giving approval.
involves evaluation of the drug for a new indication
of a marketed drug; here, DCGI permission and EC
approval are needed. INSTITUTE ETHICAL COMMITTEE
In 1980, for the first time Indian Council of Medi-
cal Research (ICMR) policy statement mentioned
MONITORING AND REPORTING
about the need of an ethical committee. The insti-
OF ADRs OR AEs tutional ethical committee (IEC) is also referred to
According to ICH GCP sec 1.2, an ADR is any as know as institutional review board (IRB), ethics
untoward medical occurrence in a patient or clin- review board (IRB) and research ethics board
ical investigation subject administered a pharma- (REB) in other countries. It is mandatory that each
ceutical product and which does not necessarily research study in health care sciences involving hu-
have causal relationship with this treatment. man subjects should be evaluated and approved by
Any unexpected/expected AE/ADR should be the IEC before they are conducted. The primary
specified in the concerned SOP. Medically, they can responsibility of the IEC is to ensure the welfare
be mild, moderate or severe/serious and hence and safety of the participants in the research study.
causality relationship to any of the above should be The main responsibilities of the IEC are as follows:
examined. An AE or unexpected ADR requires l To protect the dignity, rights and well-being of

express review by the EC. Any unexpected AE/ the potential research participants
ADRs and all SAE (serious AE) should be reported l To ensure that universal ethical values and in-

by the investigator to the sponsor within 24 h and ternational scientific standards are expressed in
to the EC that accorded approval to the study pro- terms of local community values and customs
tocol within 7 days. In the event of death the EC l To assist in the development and the education

should also be informed within 24 h. Any unex- of a research community responsive to local
pected SAE as defined in the Indian GCP Guide- health care requirements
lines occurring during a clinical trial should be Ideally, the IEC should be kept of fairly small
communicated by the sponsor to the licensing size (8–12 members), where minimum 5 mem-
authority within 14 calendar days and also com- bers are required to complete the quorum. Other-
municated to the investigator(s) of other trial sites wise, the decision made by committee stands in-
participating in the study. This will help to stop the valid. The committee should be a mix of medical,
clinical trial and withdraw the market-approved nonmedical and even lay persons, considering the
drug on report of phase IV studies. age, gender, etc., so that the committee can have
308 Research Methodology and Biostatistics

representation from all the sections. The composi- l Guidelines on extension of the term of the
tion of the committee may be as given below: members of the committee and policy on
l Chairperson: A person from outside the institute
percentage of members should be changed on
l Members from basic medical sciences: 1/2
completion of a term.
l Members from clinical specialties: 1/2 l Policy on the replacement of members who are
l Legal expert/retired judge: 1
missing the meetings frequently due to illness or
l Social scientists: 1
unforeseen reasons. The criteria for number of
l Lay person: 1
missed meetings may be defined to replace the
l Member secretary: Should be from institute to
person.
conduct business of the committee The research proposal should be thoroughly
Note: The subject matter experts may be invited for reviewed by an appropriate scientific review
a particular meeting based on the need of the committee before it is placed to IEC for the
research project. For example, a research project review, so that scientific soundness of the
from oncology discipline may impose need to invite proposal is evaluated before ethical review pro-
a subject matter expert from the oncology speciality. cess is carried out. The committee should evalu-
A subcommittee of main IEC may be constituted ate the possible risks to the participants with
for the review of undergraduate and postgraduate re- proper justification, the expected benefits and
search proposals. There could be more than one IEC adequacy of documentation for ensuring pri-
in an institute based on the number of research pro- vacy, confidentiality and the justice issues. Be-
posals that need review in a particular time period. fore the ethical review, the member secretary of
The ethical committee members must be facili- the IEC should screen the proposals for their
tated for training to make them oriented with newer completeness and categorize them in three basic
changes in national and international ethical guide- categories depending on the risk involved,
lines. For drug trial reviews, the institutional ethical namely, exemption from review, expedited
committee (IEC) must be trained in good clinical review and full review. The noninterventional
practices. The IEC must have the standard operating studies are generally given exemption from
procedures (SOPs) to define the following: review, but it has to be decided by the IEC. The
l The term of appointment of the members.
researcher himself or herself cannot decide
l Policy on removal, replacement and resignation
about it. The research studies involving routine
of member of the committee. defined risk such as known surgeries or inter-
l Frequency of meetings, honorarium/remunera-
ventions are categorized under the expedited
tions for the members or invited experts of the review and new drug trials are placed under the
committee. full review.

MULTIPLE CHOICE QUESTIONS

1. The first international guideline pertaining to 2. Informed consent refers to
research involving human participants in the history (a) Principle of autonomy
was the (b) Voluntary but uninformed decision making
(a) Declaration of Helsinki (c) A voluntary decision to participate in research,
(b) Declaration of Japan by a competent individual who has received and
(c) Nuremberg code understood the necessary information
(d) WHO ethical guidelines (d) Permission to participate in research
 Ethical Issues in Biomedical Research in India 309

3. Informed consent is important because 7. Which of the following activities would result in a
(a) It enables the participant to understand vital conflict of interest?
information in the proposed trial. (a) Not ensuring patient’s welfare
(b) It provides the participant with all the informa- (b) Skewed interpretation of data in favour of
tion regarding remote risk. sponsoring agency
(c) It enables the investigator to recruit participants (c) Undue encouragement of participants, fami-
of his choice. lies and population through compensation
(d) It promotes clinical research. (d) All the above

4. Which of the following is necessary in obtaining 8. In case of adverse event or adverse drug reaction:
informed consent? (a) It should be reported to the sponsor within
(a) A description of the statistical analysis that will 24 h and to the ethics committee within 7 days.
be carried out (b) It should be reported to the sponsor within a
(b) A description of the purpose of the research month.
(c) A description of the reliability and validity of (c) The researcher can simply quit the study with-
test instruments out reporting.
(d) A list of publications that the researcher has (d) The researcher can continue the study until
done in the past the study is over without reporting to the
concerned authority.
5. The World Medical Association ethical guidelines
were first formulated under 9. What must be done if a researcher withholds infor-
(a) The CIOMS International ethical guidelines for mation, use deception and/or there is a potential
biomedical research for harmful effects on participation?
(b) The Declaration of Helsinki (a) Inquiry
(c) The Council of Trent (b) Debriefing
(d) The Tuskegee syphilis study (c) Detoxing
(d) Dehoaxing
6. Can the requirement of informed consent be waived?
(a) Yes, at the discretion of the investigator. 10. Which of the following is not ethical for conducting
(b) No, unless a designated ethics review committee research with humans?
approves. (a) Getting informed consent of the participant
(c) Never. (b) Telling the participants they must continue
(d) Yes, with the consent of the participant. until the study has been completed
(c) Keeping participants identity anonymous
(d) Telling participants they are free to withdraw
anytime

Answers of Multiple Choice Questions

1. c; 2. c; 3. a; 4. b; 5. b; 6. a; 7. d; 8. a; 9. b; 10. b.

REFERENCES
Charles, W., & Paul, S. (2002). The therapeutic misconception: and methodology (1st ed.). New Delhi: Jaypee Brothers; 2007.
problems and solutions. Medical Care 40, 55–63. pp. 93–105.
Council for International Organizations of Medical Sciences. Krugman, S. (1986). The Willowbrook hepatitis studies revisited:
(1993). International ethical guidelines for biomedical research ethical aspects. Reviews of Infectious Diseases, 8, 157–162.
involving human subjects. Geneva: Council for International Melo-Martín, I., & Ho, A. (2008). Beyond informed consent: the
Organization of Medical Sciences therapeutic misconception and trust. Journal of Medical Ethics,
Ethical guidelines for biomedical research on human subjects. New 34, 202–205.
Delhi: Indian Council of Medical Research; 2006. National Commission for the Protection of Human Subjects of
Kher, S. (2007). Informed consent process: protecting subjects Biomedical and Behavioural Research. (1979). The Belmont
rights. In: S. K. Gupta, (Ed.). Basic principles of clinical research Report. Washington DC: US Government Printing Office.
310 Research Methodology and Biostatistics

Nuremberg Military Tribunal. (1996). The Nuremberg code. Jour- World Health Organization. (2000). Operational guidelines for eth-
nal of the American Medical Association, 276, 1691. ics committees that review biomedical research. Geneva: World
Sanmukhani, J., & Tripathi, C. B. (2011). Ethics in clinical re- Health Organization.
search: the Indian perspective. Indian Journal of Pharmaceuti- World Medical Association. Declaration of Helsinki – Ethical Prin-
cal Sciences, 73(2), 125–130. ciples for Medical Research Involving Human Subjects. (2008).
Srinivasan, S. Some questionable drug trials. 2005. ,http:// ,http://www.wma.net/en/30publications/10policies/b3/index.
infochangeindia.org/20051112276/Health/Features/Some- html. Last cited: 13-04-10.
questionable-drug-trials.html. Last cited: April, 2010. Zong, Z. (2008). Should post-trial provision of beneficial experi-
Thatte, U. (2007) Ethical issues in clinical research. In: Gupta, S. K. mental interventions be mandatory in developing countries?
(ed.). Basic principles of clinical research and methodology Journal of Medical Ethics34, 188–192.
(1st ed.) pp. 58–73. New Delhi: Jaypee Brothers.
 Appendices

Appendix I
Table of t
Probability of larger value of t:
df 0.10 0.05 0.02 0.01 0.001
1 6.31 12.71 3.82 63.66 636.62
2 2.92 4.30 6.97 9.93 31.60
3 2.35 3.18 4.54 5.84 12.92
4 2.13 2.78 3.75 4.60 8.61
5 2.02 2.57 3.37 4.03 6.87
6 1.94 2.45 3.14 3.71 5.96
7 1.90 2.37 3.00 3.50 5.41
8 1.86 2.31 2.90 3.36 5.04
9 1.83 2.26 2.82 3.25 4.78
10 1.81 2.23 2.76 3.17 4.59
11 1.80 2.20 2.72 3.11 4.44
12 1.78 2.18 2.68 3.06 4.32
13 1.77 2.16 2.65 3.01 4.22
14 1.76 2.15 2.65 2.98 4.14
15 1.75 2.13 2.60 2.95 4.07
16 1.75 2.12 2.58 2.92 4.02
17 1.74 2.11 2.57 2.90 3.97
18 1.73 2.10 2.55 2.88 3.92
19 1.73 2.09 2.54 2.86 3.88
20 1.73 2.09 2.53 2.85 3.85
21 1.72 2.08 2.52 2.83 3.82
22 1.72 2.07 2.51 2.82 3.79
23 1.71 2.07 2.50 2.81 3.77
24 1.71 2.06 2.49 2.80 3.75
25 0.17 2.06 2.49 2.79 3.73
26 0.17 2.06 2.48 2.78 3.71
27 1.70 2.05 2.47 2.77 3.69
Continued
312 Research Methodology and Biostatistics

df 0.10 0.05 0.02 0.01 0.001

28 1.70 2.05 2.47 2.76 3.67
29 1.70 2.05 2.46 2.76 3.66
30 1.69 2.04 2.46 2.75 3.65
40 1.68 2.02 2.42 2.70 3.55
60 1.67 2.00 2.39 2.66 3.46
120 1.37 1.98 2.36 2.62 3.37
∞ 1.65 1.96 2.33 2.58 3.29
The table gives the probability of observing the highest t value by the chance at particular degrees of freedom. The probability of
observing the value of t . 2.95 at 15 degrees of freedom is 0.01 or 1%.

Appendix II
Variance Ratio
5%, point of e22 (0.05):
n1/n2 1 2 3 4 5 6 8 12 24 ∞
1 161.40 199.50 215.70 224.60 230.20 234.00 238.90 243.90 249.00 254.30
2 18.51 19.00 19.16 19.25 19.30 19.33 19.37 19.41 19.45 19.50
3 10.13 9.55 9.28 9.12 9.01 8.94 8.84 8.74 8.64 8.53
4 7.71 6.94 6.59 6.39 6.26 6.16 6.04 5.91 5.77 5.63
5 6.61 5.79 5.41 5.19 5.05 4.95 4.82 4.69 4.53 4.36
6 5.99 5.14 4.76 4.53 4.39 4.28 4.15 4.00 3.84 3.67
7 5.59 4.74 4.35 4.12 3.97 3.87 3.73 3.57 3.41 3.23
8 5.32 4.46 4.07 3.84 3.69 3.58 3.44 3.20 3.12 2.93
9 5.12 4.26 3.86 3.63 3.48 3.37 3.23 3.07 2.90 2.71
10 4.96 4.10 3.71 3.48 3.33 3.22 3.07 2.91 2.74 2.54
11 4.84 3.98 3.59 3.36 3.20 3.09 2.95 2.79 2.61 2.40
12 4.75 3.88 3.49 3.26 3.11 3.00 2.85 2.69 2.50 3.30
13 4.67 3.80 3.41 3.18 3.02 2.92 2.77 2.60 2.42 2.21
14 4.60 3.74 3.34 3.11 2.96 2.85 2.70 2.53 2.35 2.13
15 4.54 3.68 3.29 3.06 2.90 2.79 2.64 2.48 2.29 2.07
16 4.49 3.63 3.24 3.04 2.85 2.74 2.59 2.42 2.24 2.01
17 4.45 3.59 3.20 2.96 2.81 2.70 2.55 2.38 2.19 1.96
18 4.41 3.55 3.16 2.93 2.77 2.66 2.51 2.34 2.15 1.92
19 4.38 3.53 3.13 2.90 2.74 2.63 2.48 2.31 2.11 1.88
20 4.35 3.49 3.10 2.87 2.71 2.60 2.45 2.28 2.08 1.84
21 4.32 3.47 3.07 2.84 2.68 2.57 2.42 2.25 2.05 1.81
22 4.30 3.44 3.05 2.83 2.66 2.55 2.40 2.23 2.03 1.78
23 4.28 3.42 3.03 2.80 2.64 2.53 2.24 2.20 2.00 1.76
24 4.26 3.40 3.01 2.78 2.62 2.51 2.36 2.18 1.98 1.73
Continued
 Appendices 313

n1/n2 1 2 3 4 5 6 8 12 24 ∞
25 4.24 3.38 2.97 2.76 2.60 2.49 2.34 2.16 1.96 1.71
26 4.22 3.37 2.98 2.74 2.59 2.47 2.36 2.15 1.95 1.69
27 4.21 3.35 2.96 2.73 2.57 2.46 2.30 2.13 1.93 1.67
28 4.20 3.34 2.95 2.71 2.56 2.44 2.29 2.12 1.91 1.65
29 4.18 3.33 2.93 2.70 2.54 2.43 2.28 2.10 1.90 1.64
30 4.17 3.32 2.92 2.69 2.53 2.42 2.27 2.09 1.89 1.62
40 4.08 3.23 2.84 2.61 2.45 2.34 2.18 2.00 1.79 1.51
60 4.00 3.15 2.76 2.52 2.37 2.25 2.10 1.92 1.70 1.39
120 3.92 3.07 2.68 2.45 2.29 2.17 2.02 2.83 1.61 1.25
∞ 3.84 2.99 2.60 2.37 2.21 2.10 1.94 1.75 1.52 0.00
Lower 5% points are found by interchange of n1 and n2 , i.e. n1 must always correspond with greater mean square.

1%, point of e3z(0.01):

n1/n2 1 2 3 4 5 6 8 12 24 ∞
1 4052.00 4999.00 5403.00 5625.00 5764.00 5859.00 5982.00 6106.00 6234.00 6366.00
2 98.50 99.00 99.17 99.25 99.30 99.33 99.37 99.42 99.46 99.50
3 34.12 30.82 29.46 28.71 28.24 27.91 27.49 27.05 26.60 26.12
4 21.20 18.00 16.69 15.98 15.52 15.21 14.80 14.37 13.93 13.46
5 16.26 13.27 12.06 11.39 10.97 10.67 10.29 9.89 9.47 9.02
6 13.74 10.92 9.78 9.15 8.75 8.47 8.10 7.72 7.31 6.88
7 12.25 9.55 8.45 7.85 7.46 7.19 6.84 6.47 6.07 5.65
8 11.26 8.65 7.59 7.01 6.63 6.37 6.03 5.67 5.28 4.86
9 10.56 8.02 6.99 6.42 6.06 5.80 5.47 5.11 4.73 4.31
10 10.04 7.56 6.55 5.99 5.64 5.39 5.06 4.71 4.33 3.91
11 9.65 7.20 6.22 5.67 5.32 5.07 4.74 4.40 4.02 3.60
12 9.30 6.93 5.95 5.41 5.06 4.82 4.50 4.16 3.78 3.36
13 9.07 6.70 5.74 5.20 4.86 4.62 4.30 3.96 3.59 3.16
14 8.86 6.51 5.56 5.03 4.69 4.46 4.14 3.80 3.43 3.00
15 8.68 6.36 5.42 4.89 4.56 4.32 4.00 3.67 3.29 2.87
16 8.53 6.23 5.29 4.77 4.44 4.20 3.89 3.55 3.18 2.75
17 8.40 6.11 5.18 4.67 4.34 4.10 3.79 3.45 3.08 2.65
18 8.28 6.01 5.09 4.58 4.25 4.01 3.71 3.37 3.00 2.57
19 8.18 5.93 5.01 4.50 4.17 3.94 3.63 3.30 2.92 2.49
20 8.10 5.85 4.94 4.43 4.10 3.87 3.56 3.23 2.86 2.42
21 8.02 5.78 4.87 4.37 4.04 3.81 3.51 3.17 2.80 2.36
22 7.94 5.72 4.82 4.31 3.99 3.76 3.45 3.12 2.75 2.31
23 7.88 5.66 4.76 4.26 3.94 3.71 3.41 3.07 2.70 2.26
24 7.82 5.61 4.72 4.22 3.90 3.67 3.36 3.03 2.66 2.21
Continued
314 Research Methodology and Biostatistics

n1/n2 1 2 3 4 5 6 8 12 24 ∞
25 7.77 5.57 4.68 4.18 3.86 3.63 3.32 2.99 2.62 2.17
26 7.72 5.53 4.64 4.14 3.82 3.59 3.29 2.96 2.58 2.13
27 7.68 5.49 4.60 4.11 3.78 3.56 3.26 2.93 2.55 2.10
28 7.64 5.45 4.57 4.07 3.75 3.53 3.23 2.90 2.52 2.06
29 7.60 5.42 4.54 4.04 3.73 3.50 3.20 2.87 2.49 2.03
30 7.56 5.39 4.31 4.02 3.70 3.17 3.47 2.84 2.47 2.01
40 7.31 5.18 4.31 3.83 3.51 3.29 2.99 2.66 2.29 1.80
60 7.08 4.98 4.13 3.65 3.34 3.12 2.82 2.50 2.12 1.60
120 6.85 4.79 3.95 3.48 3.17 2.96 2.66 2.34 1.95 1.38
∞ 6.64 4.60 3.78 3.32 3.02 2.80 2.51 2.18 1.79 1.00
Lower 1% points are found by interchange of n1 and n2 , i.e. n1 must always correspond with greater mean square.

Appendix III
Table of x2
Probability (p)
df 0.50 0.10 0.05 0.02 0.01 0.001
1 0.46 2.71 3.84 5.41 6.64 10.83
2 1.39 4.61 5.99 7.82 9.21 13.82
3 2.37 6.25 7.82 9.84 11.34 16.27
4 3.36 7.78 9.49 11.87 13.28 18.47
5 4.35 9.24 11.07 13.39 15.09 20.52
6 5.35 10.65 12.59 15.03 16.81 22.46
7 6.35 12.02 14.07 16.62 18.48 24.32
8 7.34 13.36 15.51 18.17 20.09 16.13
9 8.34 14.68 16.92 19.68 21.67 27.88
10 9.34 15.89 18.31 21.16 23.21 29.59
11 10.34 17.28 19.68 22.62 24.73 31.26
12 11.34 18.55 21.03 24.05 26.22 32.91
13 12.34 19.81 22.36 25.47 27.69 34.53
14 13.34 21.06 23.69 26.87 29.14 36.12
15 14.34 22.31 24.99 28.26 30.58 37.70
16 15.34 23.54 26.30 29.63 32.00 39.25
17 16.34 24.77 27.59 30.99 33.41 40.79
18 17.34 25.99 28.87 32.35 34.81 42.31
19 18.34 27.20 30.14 33.69 36.19 43.82
20 19.34 28.41 31.41 35.02 37.57 45.32
21 20.34 29.62 32.67 36.34 3.93 46.80
22 21.34 30.81 33.92 37.66 40.29 48.27
Continued
 Appendices 315

df 0.50 0.10 0.05 0.02 0.01 0.001

23 22.34 32.01 35.17 38.97 41.64 49.73
24 23.34 33.20 36.42 40.27 42.98 51.18
25 24.34 34.38 37.65 41.57 44.31 52.62
26 25.34 35.56 38.89 42.86 45.64 54.05
27 26.34 36.74 40.11 44.14 46.96 55.48
28 27.34 37.92 41.34 45.42 48.28 56.89
29 28.34 39.09 42.56 46.69 49.59 58.30
30 29.34 40.26 43.77 47.96 50.89 59.70
2
The table gives the highest values of x , at particular degrees of freedom, corresponding to probability p of occurrence by
chance in nature; e.g. at 10 degree of freedom, x2 value larger than 18.31 will occur less than five times in 100 (p , 0.05) and is
interpreted as significant at 5% level.

Appendix IV
The Correlation Coefficient Table
Probability (p)
df 0.10 0.05 0.02 0.01 0.001
1 0.9877 0.9969 0.9995 0.99988 0.99999
2 0.9000 0.9500 0.9800 0.9900 0.9990
3 0.805 0.878 0.9343 0.9587 0.9912
4 0.729 0.811 0.882 0.9172 0.9741
5 0.669 0.755 0.833 0.875 0.9507
6 0.621 0.707 0.789 0.834 0.925
7 0.582 0.666 0.750 0.798 0.898
8 0.549 0.632 0.716 0.765 0.872
9 0.521 0.602 0.685 0.735 0.847
10 0.497 0.576 0.658 0.708 0.823
11 0.476 0.553 0.634 0.684 0.801
12 0.457 0.632 0.612 0.661 0.780
13 0.441 0.514 0.592 0.641 0.760
14 0.426 0.497 0.574 0.623 0.742
15 0.412 0.482 0.558 0.606 0.725
16 0.400 0.468 0.543 0.590 0.708
17 0.389 0.456 0.529 0.575 0.693
18 0.378 0.444 0.516 0.561 0.679
19 0.369 0.433 0.503 0.549 0.665
20 0.360 0.423 0.492 0.537 0.652
25 0.323 0.381 0.445 0.487 0.597
30 0.296 0.349 0.409 0.449 0.554
Continued
316 Research Methodology and Biostatistics

df 0.10 0.05 0.02 0.01 0.001

35 0.275 0.325 0.381 0.418 0.519
40 0.257 0.304 0.358 0.393 0.490
45 0.243 0.288 0.338 0.372 0.465
50 0.231 0.273 0.322 0.354 0.443
60 0.211 0.250 0.295 0.325 0.408
70 0.195 0.232 0.274 0.302 0.380
80 0.183 0.217 0.257 0.283 0.357
90 0.173 0.205 0.242 0.267 0.338
100 0.164 0.195 0.230 0.254 0.321
The table gives percentage points for the distribution of the estimated correlation coefficient when true value is zero. Thus, when
there are 10 degrees of freedom (i.e. in a sample of 12), the probability of observing r greater in absolute value than 0.576 by
chance is 0.05 or 5%.
In this case, df 5 degree of freedom 5 n 2 2.

Appendix V
Table of Probability from the Standard Gaussian Distribution
The table helps to find percentage of area or observations in the normal distribution, which lie
beyond various standard deviation units (Z 5 X 2 X/s) from the mean, Z is at the mean.
Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
0.0 0.5000 0.4960 0.4920 0.4880 0.4840 0.4801 0.4761 0.4721 0.4681 0.4641
0.1 0.4602 0.4562 0.4522 0.4483 0.4443 0.4404 0.4364 0.4325 0.4286 0.4247
0.2 0.4207 0.4168 0.4129 0.4090 0.4052 0.4013 0.3974 0.3936 0.3897 0.3859
0.3 0.3821 0.3783 0.3745 0.3707 0.3669 0.3632 0.3594 0.3557 0.3520 0.3483
0.4 0.3446 0.3409 0.3372 0.3336 0.3300 0.3264 0.3228 0.3192 0.3156 0.3121
0.5 0.3085 0.3050 0.3015 0.2981 0.2946 0.2912 0.2877 0.2843 0.2810 0.2776
0.6 0.2743 0.2709 0.2676 0.2643 0.2611 0.2578 0.2546 0.2514 0.2483 0.3451
0.7 0.2420 0.2389 0.2358 0.2327 0.2296 0.2266 0.2236 0.2205 0.2177 0.2148
0.8 0.2119 0.2090 0.2061 0.2033 0.2005 0.1977 0.1949 0.1920 0.1894 0.1867
0.9 0.1841 0.1814 0.1788 0.1762 0.1736 0.1711 0.1685 0.1660 0.1635 0.1611
1.0 0.1587 0.1562 0.1539 0.1515 0.1492 0.1469 0.1446 0.1423 0.1401 0.1379
1.1 0.1357 0.1335 0.1314 0.1292 0.1271 0.1251 0.1230 0.1210 0.1190 0.1170
1.2 0.1151 0.1131 0.1112 0.1093 0.1070 0.1056 0.1038 0.1020 0.1003 0.0985
1.3 0.0968 0.0951 0.0934 0.0918 0.0901 0.0855 0.0869 0.0853 0.0838 0.0823
1.4 0.0808 0.0793 0.0778 0.0764 0.0749 0.0735 0.0721 0.0708 0.0694 0.0681
1.5 0.0668 0.0655 0.0643 0.0630 0.0618 0.0606 0.0594 0.5820 0.0571 0.0559
1.6 0.0548 0.0537 0.0526 0.0516 0.0505 0.0495 0.0485 0.0475 0.0465 0.0455
1.7 0.0446 0.0436 0.0427 0.0418 0.0409 0.0401 0.0392 0.0384 0.0375 0.0367
1.8 0.0359 0.0351 0.0344 0.0336 0.0329 0.0322 0.0314 0.0307 0.0301 0.0294
1.9 0.0287 0.0281 0.0274 0.0268 0.0262 0.0256 0.0250 0.0244 0.0239 0.0233
Continued
 Appendices 317

Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
2.0 0.0228 0.0222 0.0217 0.0212 0.0207 0.0202 0.0197 0.0192 0.0188 0.0183
2.1 0.0179 0.0174 0.0170 0.0166 0.0162 0.0158 0.0154 0.0150 0.0146 0.0143
2.2 0.0139 0.0136 0.0132 0.0129 0.0125 0.0122 0.0119 0.0116 0.0113 0.0110
2.3 0.0107 0.0104 0.0102 0.0099 0.0096 0.0094 0.0091 0.0089 0.0087 0.0084
2.4 0.0082 0.0080 0.0078 0.0075 0.0073 0.0071 0.0069 0.0068 0.0066 0.0064
2.5 0.0062 0.0060 0.0059 0.0057 0.0055 0.0054 0.0052 0.0051 0.0049 0.0048
2.6 0.0047 0.0045 0.0044 0.0043 0.0041 0.0040 0.0039 0.0038 0.0037 0.0036
2.7 0.0035 0.0034 0.0033 0.0032 0.0031 0.0030 0.0029 0.0028 0.0027 0.0026
2.8 0.0026 0.0025 0.0024 0.0023 0.0023 0.0022 0.0021 0.0021 0.0020 0.0019
2.9 0.0019 0.0018 0.0018 0.0017 0.0016 0.0016 0.0015 0.0015 0.0014 0.0014
3.0 0.0013 0.0013 0.0013 0.0012 0.0012 0.0011 0.0011 0.0011 0.0010 0.0010
Vertical column headed Z gives the Z-value up to one decimal point such as 0.1 and 0.2, while horizontal values to the right
of Z-values give up to two decimal points such as 0.00, 0.001 and 0.02.
To know the area or percentage of observation lying beyond the Z-value of 1.96, find 1.9 in the Z column at the extreme left
and then read across to the column headed 0.06. There, the mentioned score is 0.0250; hence, the percentage of observations
will be 2.5%.

Appendix VI
Table of Critical Values of Spearman Correlation Coefficient
Minimum value of rs
Sample size (n) a 5 0.10 a 5 0.05 a 5 0.02 a 5 0.01
1, 2 or 3 None None None None
4 1.000 None None None
5 0.900 1.000 1.000 None
6 0.771 0.886 0.943 1.000
7 0.714 0.786 0.892 0.929
8 0.643 0.738 0.810 0.857
9 0.600 0.683 0.783 0.817
10 0.564 0.648 0.733 0.781
Note: Minimum value of rs for different n (#10) that is statistically significant for different a levels (two-tailed); for n  11, use
Student’s t-test.

Appendix VII
Table of Critical Values of Correlation Coefficient
df 0.05 0.01 df 0.05 0.01 df 0.05 0.01
1 0.996917 0.9998766 11 0.5529 0.6835 25 0.3809 0.4869
2 0.95000 0.990000 12 0.5324 0.6614 30 0.3494 0.4487
3 0.8783 0.95873 13 0.5139 0.6411 35 0.3246 0.4182
Continued
318 Research Methodology and Biostatistics

df 0.05 0.01 df 0.05 0.01 df 0.05 0.01

4 0.8114 0.91720 14 0.4973 0.6226 40 0.3044 0.3932
5 0.7545 0.8745 15 0.4821 0.6055 45 0.2875 0.3721
6 0.7067 0.8343 16 0.4683 0.5897 50 0.2732 0.3541
7 0.6664 0.7977 17 0.4555 0.5751 60 0.2500 0.3248
8 0.6319 0.7646 18 0.4438 0.5614 70 0.2319 0.3017
9 0.6021 0.7348 19 0.4329 0.5487 80 0.2172 0.2830
10 0.5760 0.7079 20 0.4227 0.5368 90 0.2050 0.2673
100 0.1946 0.2540
Note: The degrees of freedom (df ) 5 the number of pairs in the sample 2 2.

Appendix VIII
Table of Percentage Points for the Wilcoxon Signed Rank Sum Test
The function tabulated is the critical value for the smaller of the signed rank sums, T1 and T2.
An observed value equal to or less than the tabulated value is significant at the two-sided significance
level shown.

Two-Sided Significance Level

Sample Size, n (Nonzero Differences) 0.05 0.01
6 0 –
7 2 –
8 3 0
9 5 1
10 8 3
11 10 5
12 13 7
13 17 9
14 21 12
15 25 15
16 29 19
17 34 23
18 40 27
19 46 32
20 52 37
21 58 42
22 66 48
23 73 54
24 81 61
25 89 68
 Appendices 319

Appendix IX
Statistical Formulas
1. Arithmetic mean:
Sum of the value(X )
X
Number of the value (n)

n 1 
2. Mode: M  
 2 
3. Range (R) 5 H – L
( )
2
 X X
4. Standard deviation: SD 
n

5. Z-score: Z  X  

6. Karl Pearson correlation coefficient:

r
(
 X  X Y Y)( )
( ) ( )
2 2
 X  X  Y Y

7. Spearman rank correlation coefficient:

6d 2
  1
(
n n 2 1 )
8. Standard error:

Standard error ( SE ) 
(SD1 )2 
(SD2 )2
n1 n2

9. t-Test:
Observed difference ( X1  X 2 )
t -Test 
SE

10. Chi-square test:

Column or vertical total × Row or horizontal tota l
Calculation of expected frequency E 
Sample total
( O  E )2
 df2  
E
320 Research Methodology and Biostatistics

Appendix X
Random Numbers
03 47 43 73 86 36 96 47 36 61 46 98 63 71 62 33 26 16 80 45
97 74 24 67 62 42 81 14 57 20 42 53 32 37 32 27 07 36 07 51
16 76 63 27 66 56 50 26 71 7 31 90 79 78 53 13 55 38 58 59
12 56 85 99 26 96 96 68 27 31 05 03 72 93 15 57 12 10 14 21
55 59 56 35 64 38 54 82 46 22 31 62 43 09 90 06 18 44 32 53
16 22 77 94 39 49 54 43 54 82 17 37 93 23 78 87 35 20 96 43
84 42 17 53 31 57 24 55 06 88 77 04 74 47 67 21 76 33 50 25
63 01 63 78 59 16 95 55 67 19 98 10 50 71 75 12 86 73 58 7
33 21 12 34 29 78 64 56 07 82 52 42 7 44 38 15 51 0 13 42
57 60 86 32 44 09 47 27 96 54 49 17 46 09 62 90 52 84 77 27
18 18 07 92 46 44 17 16 58 9 79 83 86 19 62 06 76 50 03 10
26 62 38 97 75 81 16 07 44 99 83 11 46 32 24 20 14 85 88 45
23 42 40 64 74 82 97 77 77 81 07 45 32 14 08 32 98 94 07 72
62 36 28 19 95 50 92 26 11 97 00 56 76 31 38 80 22 02 53 53
37 85 94 35 12 39 50 08 30 42 34 7 96 88 54 42 06 87 98
07 29 17 53 13 33 20 38 26 13 89 03 74 17 76 37 13 4
56 62 18 37 35 96 83 50 87 75 97 12 25 93 47 70 33 24 03 54
99 49 57 22 77 88 42 95 45 72 16 64 36 16 00 04 43 18 66 79
16 08 15 04 72 33 27 14 34 9 45 59 34 68 49 12 72 07 34 45
31 16 93 32 43 50 27 89 87 19 20 15 37 00 49 52 85 66 90 11
68 34 30 13 70 55 74 30 77 40 44 22 78 84 26 04 33 46 09 52
74 57 25 65 76 59 29 97 68 60 71 91 38 67 54 13 58 18 24 76
27 42 37 86 53 48 55 90 65 72 96 57 69 36 10 96 46 92 42 45
00 39 68 29 61 66 37 32 20 30 77 84 57 03 29 10 45 65 04 26
29 94 98 94 24 68 49 69 10 82 53 75 91 93 30 34 25 20 57 27
16 90 82 66 59 83 62 64 11 12 67 19 00 71 74 60 47 21 29 68
11 27 94 75 6 06 09 19 74 66 02 94 37 34 02 76 70 90 30 86
35 24 10 16 20 33 32 51 26 38 79 78 45 04 91 16 92 53 56 16
38 23 16 86 38 42 38 97 01 50 87 75 66 81 41 40 01 74 91 62
31 96 25 91 47 96 44 33 49 13 34 86 82 53 91 00 52 43 48 85
66 67 40 67 14 64 05 71 95 86 11 05 65 09 68 76 83 20 37 90
14 90 84 45 11 75 73 88 05 90 52 27 41 14 86 26 98 12 22 8
68 05 51 18 0 33 96 02 75 19 07 60 62 93 55 59 33 82 43 90
20 46 78 73 90 97 51 40 14 02 04 02 33 31 08 39 54 16 49 36
64 19 58 97 79 15 06 15 93 20 01 90 10 75 06 40 78 78 89 62
05 26 93 70 60 22 35 85 15 13 92 03 51 59 77 59 56 78 06 83
Continued
 Appendices 321

07 97 10 88 23 09 98 42 99 64 61 01 62 99 15 06 51 29 16 93
68 91 86 85 85 54 87 66 47 54 73 32 8 11 12 44 95 92 63 16
26 99 61 65 53 58 37 78 80 70 42 10 50 67 42 32 17 55 85 74
14 65 52 68 75 87 59 36 22 41 26 78 63 06 55 13 08 26 01 5

Appendix XI
An Example of Probability Proportion to Size (PPS) Cluster Sampling
Study title:
A population-based survey of the prevalence of hepatitis C virus infection in Punjab, North India.
Sample design:
This is a cross-sectional survey of people 5 years and older residing in Punjab. The survey was planned
to provide state-level estimates of the prevalence of HCV infection in the general population.
The sampling frame is based on the 2011 Punjab Census data. If Census 2011 data are not available
at the time of sampling, we would base our study on the 2001 Census data.
The sample from rural areas will be selected in the following stages:
 1st stage: Selection of districts using probability proportional to size (PPS) method

 2nd stage: Selection of talukas using PPS

 3rd stage: Selection of villages using PPS

 4th stage: Systematic selection of households within each selected village

The sample from urban areas will be selected in the following stages:
 1st stage: Selection of districts using PPS

 2nd stage: Selection of urban wards using PPS

 3rd stage: Selection of one Census enumeration block (CEB) from each selected ward using PPS

 4th stage: Systematic selection of households within each selected CEB

The sample will be explicitly stratified according to urban/rural area. Since the variability in socioeco-
nomic status and literacy are not as pronounced in Punjab as it is in other parts of India, stratification
along these factors was considered unnecessary.

Method of Sampling
Based on the aforementioned discussion in the protocol, the following steps were taken.
*For this study, the Census data of 2001 will be used to obtain cluster.
Stage 1: Selection of district:
Step 1: In total, there are 17 districts in Punjab in the 2001 Census data. Based on the discussion with
the investigator and sponsor, 50% (;9) districts were required to be selected by the PPS method.
*The districts selected will be common for both urban and rural areas.
Step 2: The total population will be taken for each district considered, and cumulative population will
be obtained.

District Total Population Cumulative Population

Amritsar 3,096,077 3,096,077
Bathinda 1,183,295 4,279,372
Faridkot 550,892 4,830,264
Fatehgarh Sahib 538,041 5,368,305
Firozpur 1,746,107 7,114,412
Continued
322 Research Methodology and Biostatistics

District Total Population Cumulative Population

Gurdaspur 2,104,011 9,218,423
Hoshiarpur 1,480,736 10,699,159
Jalandhar 1,962,700 12,661,859
Kapurthala 754,521 13,416,380
Ludhiana 3,032,831 16,449,211
Mansa 688,758 17,137,969
Moga 894,854 18,032,823
Muktsar 777,493 18,810,316
Nawanshahr 587,468 19,397,784
Patiala 1,844,934 21,242,718
Rupnagar 1,116,108 22,358,826
Sangrur 2,000,173 24,358,999
Total 24,358,999

For Amritsar (being first in the district column): The cumulative population 5 Total population
For Bathinda: Cumulative population 5 Total population of Bathinda 1 Total population of
Amritsar
For Faridkot: Cumulative population 5 Total population of Faridkot 1 Total population of
Amritsar 1 Total population of Bathinda
… and so on, until the last district.
Step 3: Obtain sampling interval by the following formula:
Sampling interval (SI) 5 Cumulative population/Number of districts
5 24,358,999/9
5 2,706,555
Step 4: Obtain random start using RANDBETWEEN function in Excel. By using this function, we can
obtain any random number between 1 and the SI (5 2,706,555). Assign random start for each
district using the following method:
Random start (RS) 5 1,444,042
RS1 5 1,444,042
RS2 5 1,444,042 1 2,706,555 5 4,150,597
RS3 5 4,150,597 1 2,706,555 5 6,857,153
… and so on, until we get RS greater than total population (5 24,358,999) for the first time.
District Total Population Cumulative Population Sampling Interval Random Start
Amritsar 3,096,077 3,096,077 2,706,555 1,444,042
Bathinda 1,183,295 4,279,372 4,150,597
Faridkot 550,892 4,830,264 6,857,153
Fatehgarh Sahib 538,041 5,368,305 9,563,708
Firozpur 1,746,107 7,114,412 12,270,264
Gurdaspur 2,104,011 9,218,423 14,976,819
Hoshiarpur 1,480,736 10,699,159 17,683,375
 Appendices 323

District Total Population Cumulative Population Sampling Interval Random Start

Jalandhar 1,962,700 12,661,859 20,389,930
Kapurthala 754,521 13,416,380 23,096,486
Ludhiana 3,032,831 16,449,211 25,803,041
Mansa 688,758 17,137,969
Moga 894,854 18,032,823
Muktsar 777,493 18,810,316
Nawanshahr 587,468 19,397,784
Patiala 1,844,934 21,242,718
Rupnagar 1,116,108 22,358,826
Sangrur 2,000,173 24,358,999
Total 24,358,999

Step 5: To select the district, each random start is picked and obtained from the cumulative interval
in which it falls and the interval of cumulative population and corresponding district will be
selected later.
*The districts highlighted in bold were selected by this procedure.
Stage 2: Selection of subdistrict for rural and urban areas:
Step 1: All available subdistricts within each selected district and corresponding total rural population
was obtained.
Step 2: Approximately 50% of the total number of subdistricts in each district will be selected.
Steps 2–5 will be repeated for the selection of subdistricts within each district.
*Twenty-three subdistricts for each rural and urban area will be selected.
Stage 3: Selection of villages from the selected subdistrict for rural area:
Step 1: In total, 23 subdistricts were selected for further sampling of villages. The selection of the
villages will be based on the total populations in rural and urban areas.
Step 2: As per the Census 2001 data, the total households in Punjab are 4,348,580 (rural 5 2,796,284 and
urban 5 1,552,296). Based on these data, 65% of the total number of households are in rural areas
and 35% are in urban areas. To select overall 1100 households, 715 households are selected from rural
and 385 from urban. Hence, we have selected 143 villages from rural and 77 wards from urban areas.
Step 3: Selection of 143 villages was done by using PPS. The total population of villages in each
subdistrict will be considered and cumulative population will be obtained.
Steps 4 and 5 of selection of districts are repeated and allocation of 143 villages is performed.
Stage 4: Systematic selection of households within each selected village:
Step 1: The number of individuals per village is obtained by the following formula:
No. of individuals 5 Total no. of individuals required in rural area* (total population in respective
village/total number of population in selected villages of rural area).
Step 2: Consecutive five households will be selected in a given village to get individuals in each cluster.
Note: Asterisks (*) represent the important points considered while sampling.

Sample Selection in Rural Areas

In rural areas, list of villages (as per Census 2001) will serve as the sampling frame. The Census data will
provide information on households, persons, males and females for all the villages in each district of
Punjab. Villages with fewer than five households will not be considered for the survey. Furthermore, to
ensure the selection of a sufficient number of households in the selected PSUs, all smaller villages with
324 Research Methodology and Biostatistics

fewer than 50 households will be clubbed with the nearby village. Thus, a minimum size of 50 house-
holds will be ensured in all the villages in the sampling frame.
The selection of the required number of households will be done using systematic sampling.
Considering refusals by the households and resource considerations, runs of five consecutive
households rather than single households will be selected systematically.
The number of households selected in a rural village will be the product of the number of households
listed in the village and the probability of selection of a household in the selected rural village.

Sample Selection in Urban Areas

In urban areas, the sampling frame for the first stage of selection will be the list of all wards in a selected
subdistrict. A sample of wards will be drawn from the list of wards by PPS sampling.
The sample of households from a selected ward will be drawn by adopting two-stage sampling. Each
ward comprises several CEBs. A list of all CEBs in a selected ward formed the sampling frame at the
second stage. Each CEB may be composed of about 150–200 households. One CEB from each selected
ward will be selected at first stage.
In each selected CEB, the selection of the required number of households from the list of households
will be done using systematic sampling. Considering refusals by households and resource consider-
ations, runs of five consecutive households rather than single households will be selected systematically.
The number of households selected in an urban ward will be the probability of selection of a person in
an urban ward to the total number of households in the selected urban ward.

Appendix XII
Important institutes for health science research funding in India:
1. Indian Medical Council of India (ICMER):
Ministry of Health & Family Welfare, Govt. of India
Website: http://www.icmr.nic.in
2. Department of Biotechnology (DBT):
Ministry of Science & Technology, Govt. of India
Website: http://dbtindia.nic.in
3. Department of Science & Technology (DST):
Ministry of Science & Technology, Govt. of India
Website: http://www.dst.gov.in

Appendix XIII
Format of Writing the References Using Different Styles with Relevant Examples
Vancouver style of reference writing
Type of Resource Standard Format and Bibliography Examples
Books SF: Author/editor AA. Title: subtitle. Edition (if not the first). Vol. (if a multivolume
work). Place of publication: Publisher; Year. page number(s) (if appropriate).
Single author Hoppert M. Microscopic techniques in biotechnology. Weinheim: Wiley-VCH; 2003.
Two or more authors Gilstrap LC, Cunningham FG, Van Dorsten JP, editors. Operative obstetrics. 2nd ed.
New York: McGraw-Hill; 2002.
No author The Oxford concise medical dictionary. 6th ed. Oxford: Oxford University Press;
2003. p. 26.
 Appendices 325

Type of Resource Standard Format and Bibliography Examples

Later edition Murray PR, Rosenthal KS, Kobyashi GS, Pfaller MA. Medical microbiology. 4th ed.
St Louis: Mosby; 2002.
Multiple volumes Lee GR, Bithell TC, Foerster J, Athens JW, Lukens JN, editors. Wintrobes clinical hema-
tology. 9th ed. Vol. 2. Philadelphia: Lea & Febiger; 1993.
Series Unwin E, Codde JP, Bartu A. The impact of drugs other than alcohol and tobacco on
the health of Western Australians. Perth: Department of Health; 2004. Epidemiol-
ogy occasional papers series, No. 20.
Corporate author American Veterinary Medical Association. National Board Examination Committee.
(i.e. organization) North American Veterinary Licensing Examination: bulletin of information for
candidates. Bismarck (ND): The Committee; 2001.
Conference proceedings Harnden P, Joffe JK, Jones WG, editors. Germ cell tumours V. Proceedings of the 5th
Germ Cell Tumour conference; 2001 Sep 13–15; Leeds, UK. New York: Springer;
2002.
Government publication Australia. Commonwealth Department of Veterans’ Affairs and Defence. Australian
Gulf War veterans’ health study 2003. Canberra: Commonwealth of Australia; 2003.
Scientific/technical Lugg DJ. Physiological adaptation and health of an expedition in Antarctica:
report with comment on behavioural adaptation. Canberra: A.G.P.S.; 1977. Australian
Government Department of Science, Antarctic Division. ANARE scientific reports.
Series B(4), Medical science No. 0126.
Chapter in a book SF: Author of Part, AA. Title of chapter or part. In: Editor A, Editor B, editors.
Title: subtitle of Book. Edition (if not the first). Place of publication: Publisher;
Year. Page numbers.
Meltzer PS, Kallioniemi A, Trent JM. Chromosome alterations in human solid tumors.
In: Vogelstein B, Kinzler KW, editors. The genetic basis of human cancer. New York:
McGraw-Hill; 2002. p. 93–113.
Thesis or dissertation Sharma SK. Nursing Care Administration: Patient satisfaction with nursing care in
selected public and private hospitals of Punjab. PhD [Thesis]. Chandigarh: Panjab
University; 2009.
Journal articles SF: Author of article AA, Author of article BB, Author of article CC. Title of
article. Abbreviated Title of Journal. Year; vol. (issue):page number(s).
Less than six authors Drummond PD. Triggers of motion sickness in migraine sufferers. Headache.
2005;45(6):653–6.
Halpern SD, Ubel PA, Caplan AL. Solid-organ transplantation in HIV-infected patients.
N Engl J Med. 2002;347(7):284–7.
More than six authors Gillespie NC, Lewis RJ, Pearn JH, Bourke ATC, Holmes MJ, Bourke JB, et al. Ciguatera
in Australia: occurrence, clinical features, pathophysiology and management.
Med J Aust. 1986;145:584–90.
Organization as author Diabetes Prevention Program Research Group. Hypertension, insulin, and proinsulin
in participants with impaired glucose tolerance. Hypertension. 2002;40(5):679–86.
No author given 21st century heart solution may have a sting in the tail. BMJ. 2002;325(7537):184.
Volume with Geraud G, Spierings EL, Keywood C. Tolerability and safety of frovatriptan with short-
supplement and long-term use for treatment of migraine and in comparison with sumatriptan.
Headache. 2002;42 Suppl 2:S93–9.
Issue with supplement Glauser TA. Integrating clinical trial data into clinical practice. Neurology.
2002;58(12 Suppl 7):S6–12.
Continued
326 Research Methodology and Biostatistics

Type of Resource Standard Format and Bibliography Examples

Volume with part Abend SM, Kulish N. The psychoanalytic method from an epistemological viewpoint.
Int J Psychoanal. 2002;83(Pt 2):491–5.
Issue with part Ahrar K, Madoff DC, Gupta S, Wallace MJ, Price RE, Wright KC. Development of a large
animal model for lung tumors. J Vasc Interv Radiol. 2002;13(9 Pt 1):923–8.
Issue with no volume Banit DM, Kaufer H, Hartford JM. Intraoperative frozen section analysis in revision
total joint arthroplasty. Clin Orthop. 2002;(401):230–8.
Newspaper article O’Leary C. Vitamin C does little to prevent winter cold. The West Australian.
2005 Jun 29;1.
E-Books SF: Author A, Author B. Title of e-book [format ]. Place: Publisher; Date of original
publication [cited year abbreviated month day]. Available from: Source. URL.
Van Belle G, Fisher LD, Heagerty PJ, Lumley TS. Biostatistics: a methodology for the
health sciences [e-book]. 2nd ed. Somerset (NJ): Wiley InterScience; 2003 [cited
2005 Jun 30]. Available from: Wiley InterScience electronic collection.
Sommers-Flanagan J, Sommers-Flanagan R. Clinical interviewing [e-book]. 3rd ed.
New York: John Wiley & Sons; 2003 [cited 2005 Jun 30]. Available from: NetLibrary.
Article in an Elec- Widdicombe J. Respiration. In: Blakemore C, Jennett S, editors. The Oxford compan-
tronic Reference ion to the body [e-book]. Oxford: Oxford University Press; 2001 [cited 2005 Jun 30].
Book Available from: Oxford Reference Online. http://www.oxfordreference.com.
Lee HC, Pagliaro EM. Serology: blood identification. In: Siegel J, Knupfer G, Saukko P,
editors. Encyclopedia of forensic sciences [e-book]. San Diego: Academic Press;
2000 [cited 2005 Jun 30]: 1331–8. Available from: Science Direct Reference. www.
elsevier.com.
E-Journals SF: Author A, Author B. Title of article. Abbreviated Title of Journal [format].
Year [cited year abbreviated month day];vol.(no):page numbers [estimated if
necessary]. Available from: database name (if appropriate). URL.
Journal article abstract accessed from online database
Leroy EM, Telfer P, Kumulungui B, Yaba P, et al. A serological survey of Ebola virus infection in central African
nonhuman primates. J Infect Dis [abstract]. 2004 [cited 2005 Jun 30];190(11):1895. Available from: ProQuest.
http://www.umi.com/proquest/.
Journal article from online full-text database
Palsson G, Hardardottir KE. For whom the cell tolls: debates about biomedicine (1). Curr Anthropol [serial
online]. 2002 [cited 2005 Jun 30]; 43(2):2711[about 31 pages]. Available from: Academic OneFile. http://find.
galegroup.com.
Journal article in a scholarly journal (published free of charge on the internet)
Eisen SA, Kang HK, Murphy FM, Blanchard MS, Reda DJ, Henderson WG, et al. Gulf War veterans’ health: medical
evaluation of a U.S. cohort? Ann Intern Med [serial on the Internet]. 2005 [cited 2005 June
30];142(11):8811[about 12 pages]. Available from: http://www.annals.org/.
Journal article in electronic journal subscription
Barton CA, McKenzie DP, Walters EH, et al. Interactions between psychosocial problems and management
of asthma: who is at risk of dying? J Asthma [serial on the Internet]. 2005 [cited 2005 Jun 30];42(4): 249–56.
Available from: http://www.tandf.co.uk/journals/.
Newspaper article from online database
Dearne K. Dispensing with the chemist. The Australian [newspaper online]. 2005 Jun 14 [cited 2005
Jun 30];[about 8 screens]. Available from: Factiva. http://global.factiva.com.
Newspaper article from the Internet
Diseased organs may be used to deter smokers. Sydney Morning Herald [newspaper on the Internet].
2005 Jun 29 [cited 2005 Jun 30]; para. 4–5. Available from: http://www.smh.com.au/.
 Appendices 327

Type of Resource Standard Format and Bibliography Examples

Internet documents SF: Author A, Author B. Document title. Webpage name [format]. Source/
production information; Date of internet publication [cited year month day].
Available from: URL.
Professional internet site
Australian Institute of Health and Welfare. Chronic diseases and associated risk factors [document on the
Internet]. Canberra: The Institute; 2004 [updated 2005 June 23; cited 2005 Jun 30]. Available from:
http://www.aihw.gov.au/cdarf/index.cfm.
Personal internet site
Stanley F. Information page—Professor Fiona Stanley. Telethon Institute for Child Health Research [homepage
on the Internet]. Perth: The Institute; 2005 [cited 2005 Jun 30] Available from: http://www.ichr.uwa.edu.au/
about/schools/.
General internet site
Lavelle P. Mental state of the nation. Health matters [document on the Internet]. ABC online; 2005 May 19
[cited 2005 Jul 1]. Available from: http://abc.net.au/health/features/mentalstate/.
Patent/standard Pagedas AC, inventor; Ancel Surgical R&D Inc., assignee. Flexible endoscopic
grasping and cutting device and positioning tool assembly. United States patent
US 20020103498. 2002 Aug 1.
Conference or Christensen S, Oppacher F. An analysis of Koza’s computational effort statistic for ge-
seminar paper netic programming. In: Foster JA, Lutton E, Miller J, Ryan C, Tettamanzi AG, editors.
Genetic programming. EuroGP 2002: Proceedings of the 5th European Conference
on Genetic Programming; 2002, Apr 3–5; Kinsdale, Ireland. Berlin: Springer; 2002.
p. 182–91.
Television Cohen J, reporter. Messing with heads. Four corners [television broadcast]. Australian
programme Broadcasting Corporation; 2005 Mar 21.
Video recording Hillel J, writer. Out of sight out of mind: indigenous people’s health in Australia
[video recording]. Bendigo: Video Education Australasia; 2003.
Personal conversation In a conversation with a colleague from the School of Population Health (Jameson LI
2002, oral communication, 7th August)…
Letter As stated in a letter from B.J. Samuels, MD, in July 2002…
Email Smith P. New research projects in gastroenterology [online]. E-mail to Matthew Hart
([email protected]) 2000 Feb 5 [Cited 2000 Mar 17].
Notes: SF, Standard format.
If the number of authors is six or more, then et al. is used after the surname with initials of the first six authors in reference list/bibliogra-
phy. In text, however, et al. is used after family initials of the first author.

Harvard style of reference writing

Type of Resource Standard Format Bibliography Example In-Text Example
Book Family/surname, initials. (Publica- Neville, C. (2010) The Neville (2010) argues that
(single author) tion year in brackets) Book title – Complete Guide to … “Quotation” (Neville,
italicized or underlined. Series Referencing and 2010, p.76)
title and volume if applicable. Avoiding Plagiarism.
Edition – if not the first. Place of 2nd Ed. Maidenhead:
publication: publisher. Open University
Press.
Continued
328 Research Methodology and Biostatistics

Type of Resource Standard Format Bibliography Example In-Text Example

Book Family/surname, initials., family/ Bradbury, I., Boyle, J. Bradbury, Boyle and
(two to three surname, initials. and family/ and Morse, A. (2002) Morse (2002) …
authors) surname, initials. (Publication Scientific Principles As noted by Bradbury,
year in brackets) Book title – for Physical Geogra- Boyle and Morse
italicized or underlined. Series phers. Harlow: (2002) … “Quotation”
title and volume if applicable. Prentice Hall. (Bradbury, Boyle and
Edition – if not the first. Place Note: Use either ‘and’ or Morse, 2002, p.51)
of publication: Publisher ‘&’ between authors’
names as dictated by
the book’s own pre-
sentation.
Book It is discretionary as to whether you Campbell, N. A. et al. (Campbell et al., 2008) …
(four or more list all authors and also whether (2008) Biology. 8th “Quotation” (Campbell
authors) you use ‘et al.’ or ‘and others’ as Ed. London: Pearson. et al., p.76)
below:
Family/surname, initials. et al. or
and others. (Publication year in
brackets) Book title – italicized or
underlined. Series title and vol-
ume if applicable. Edition – if not
the first. Place of publication:
Publisher.
Book (editor/s) Family/surname, initials. (ed.) or Fontana-Giusti, G. (ed.) (Fontana-Giusti, 2008)
(eds.) – in brackets for editor(s). (2008) Designing
(Publication year in brackets) Cities for People:
Book title – italicized or Social, Environmental
underlined. Series title and and Psychological
volume if applicable. Edition – if Sustainability.
not the first. Place of publication: London: Earthscan.
Publisher.
Chapter in an Family/surname, initials of the Marshall, W. A. (1975) As noted by Marshall
edited book author writing the chapter. The Child as a Mirror (1975) … “Quotation”
(Publication year in brackets) Title of his Brain’s Devel- (Marshall, 1975, p.76)
of chapter. In: family/surname, opment. In: Sants, J.
initials of author or editor of & Butcher, H. J. (eds.).
book (ed.) or (eds.). Book title – Development Psy-
italicized or underlined. Series chology. Aylesbury,
title and volume if applicable. Bucks: Hazell Watson
Edition – if not the first. Place & Viney Ltd.
of publication: Publisher.
Organization as Includes publications by Govern- Great Britain. Depart- The Great Britain Depart-
author ment departments, Committees: ment of Energy. ment of Energy (1977)
Country. Name of issuing body. (1977) Tidal Power concluded that …
(Year of publication in brackets) Barrages in the Severn “Quotation” (Great
Title of publication – in italics or Trent Estuary: Recent Britain, Department
underlined. Place of publication: Evidence on their of Energy, 1977, p.12)
Publisher. (Report Number – if Feasibility. London:
applicable in brackets). H. M.S. O. (Energy
Papers 23)
 Appendices 329

Type of Resource Standard Format Bibliography Example In-Text Example

E-Book Family/surname, initials. (Publica- Sadler, P. (2003) Strategic Sadler (2003) argues that
tion year in brackets) Book title – Management. [On- … “Quotation” (Sadler,
italicized or underlined. [Online] line] Sterling. VA Ko- 2003, p.18)
Series title and volume if gan Page. Available
applicable. from: http://www.
Edition – if not the first. Place of netlibrary.com/
publication: Publisher. Available reader/. [Accessed:
from – URL. [Accessed: date]. 6th May 2012].
Journal article Author(s) of article’s family/surname, Trefts, K. & Blacksee, S. Trefts and Blacksee
(printed) initials. (Publication year in brack- (2000) Did you hear (2000) argue that …
ets) Title of article. Title of journal – the one about Bool- “Quotation” (Trefts and
italicized or underlined. Volume ean Operators? Incor- Blacksee, 2000, p.376)
number (Part number/month in porating comedy into
brackets). p. followed by the page the library induction.
numbers of the article. Reference Services Re-
view. 28 (4). p.369–378.
Journal article If you are referencing a journal from Wilson, J. (1995) Enter Wilson (1995) argues that
(electronic/ an online database service which the Cyberpunk librar- … “Quotation” (Wilson,
online) is password accessible only, i.e. ian: future directions 1995, p.66)
EBSCO, you can shorten the URL in cyberspace. Library
to the home page of the data- Review. [Online] Em-
base service. If you are accessing erald Database 44
a journal article directly and for (8). p.63–72. Avail-
free from the internet, you will able from: http://
need the entire URL. www.emeraldinsight.
Author(s) of article’s family/sur- com. [Accessed: 30th
name, initials. (publication year in January 2012].
brackets) Title of article. Title of
journal – italicized or underlined.
[Online in square brackets] Name of
database the article is from if ap-
propriate. Volume number (Part
number/month in brackets). p.
followed by the page numbers of
the article. Available from: URL.
[Accessed: followed by the date
viewed in square brackets].
Newspaper Author(s) of article’s family/surname, Randerson, J. (2008) Randerson (2008) argues
(online) initials. (Publication year in brackets) Researchers find fish that … “Quotation”
Title of article. Title of Newspaper – that can count up to (Randerson, 2008,
italicized or underlined. [Online in four. The Guardian. p.14)
square brackets] Name of database [Online] 26th Febru-
the article is from if appropriate. ary. p.14. Available
Day and month of the article. Page from: http://the-
number of the article if applicable. guardian.co.uk.
Available from: URL. [Accessed: [Accessed: 22nd May
followed by the date viewed in 2012].
square brackets].
Continued
330 Research Methodology and Biostatistics

Type of Resource Standard Format Bibliography Example In-Text Example

Website Author of website Family/surname, BBC News. (2008) Factory … as reported by the
initials or Website name if no author gloom worst since BBC (2008) … “Quota-
is available. (Year – in brackets) Title 1980. [Online] tion” (BBC, 2008)
of website in italics or underlined. Available from:
Any number if necessary or http://news.bbc.
available if website is part of a co.uk/1/hi/business/
series. [Online in square brackets] 7681569.st m.
Available from: URL. [Accessed: [Accessed: 19th June
followed by date in square brackets]. 2012].
Films [DVD], Title – in italics or underlined. (Year Chicken Run. (2000) Ani- If you refer to a film in
[VHS], [Blu-ray of distribution in brackets) Mate- mated Film. Directed the body of your work,
Disc] rial type. Directed by – name of by Peter Lord and the title will need to
director(s). [Format of source in Nick Park. [VHS] UK: be underlined or
square brackets] Place of distri- Pathe Distribution. placed in italics: …
bution: Distribution company. Requiem for a Dream. the animation movement
(2000) Film. Directed (Chicken Run, 2000)
by Darren Aronofsky.
[DVD] UK: Momentum
Pictures.
Note: If the number of authors is four or more, then et al. is used after the surname with initials of the first author both in the text as well
as in the reference list/bibliography.
 Index

A Bipolar questions, 155

Bivariate study, 30
Abstract, 3 Blueprint. See Research design, 13
Academic literature, 38 Bracket format, 267
Accessible population, 4, 117 Brainstorming sessions, 24
Accidental sampling, 130
Acknowledgement, 263–264
Action research, 102 C
data collection methods, 102 Cafeteria questions, 154
importance of, 102 Case–control studies, 82–83
Action verbs, 35t Casual hypothesis, 39
Adaptation, 11b Chain referral sampling, 132
Additional research designs, 87–90 Chance error, 144
Administration, 156 Checklist, 170–171
Administrative support, 25 Chi-square test, 243, 250–251
Agent, defined, 53 calculation of, 250
Analysis, 190, 211, 235 prerequisites of, 250
of data, 190, 195 CINAHL. See Cumulative Index to Nursing and Allied Health
interpretation of data, 15 Literature (CINAHL), 53
qualitative data, 200–211 Closed format questions, 154–155
qualitative data analysis, 200–210 Cluster random sampling, 126
Analysis of Variance (ANOVA) test, 247–250 demerits, 126
Annotated bibliography, 47 merits, 127
ANOVA. See Analysis of Variance (ANOVA) test, 247 Coding, 97–98, 190
APA references, 271–274, 285 axial, 98
Applied research, 10–11 open, 97
Arithmetic mean, 224–225 selective, 98
Association technique, 174–175 Cohort study, 83–84
Associative hypothesis, 38–39 Communicating the research, 258–259
Attitude scales, 161–165 criteria for, 259
Authoritative sampling, 128 methods of, 259–261
demerits, 130 Comparative design, 82
merits, 130 Comparative rating scale, 169
Compilation, 190
B Completion projective techniques, 175
Bar diagram, 219 Complex hypothesis, 38
Basic research, 10–11 Composite scales, 161
Beneficence, 290 Comprehending, 197
principle of, 290 Computer, 284–285
Beta error, 195 Computer analysis, 198–200, 211
Bibliography, 265 Computer-assisted literature search, 50–53
Binomial test, 243 Conceptual framework, 4
Biophysiologic methods, 172–174 Conceptual theoretical framework, 22
advantages of, 174 Concurrent validity, 180
definition, 172 Confidence interval, 138–139, 195
disadvantages of, 174 Constitutive phenomenological research, 94
purposes of, 172 Construct validity, 180
332 Index

Construction technique, 175 Design, 34

Contemporaneous corroboration, 101 action research, 17
Content validity, 180 ethnographical design, 17
Contingency questions, 154 phenomenological, 16
Contingency tables, 190 Diagrams, 205–206, 218
Continuous frequency table, 224, 226 Dichotomous questions, 154
calculation of median, 224 Directional hypothesis, 39
calculation of mode, 226 Directive interview, 158–159
Control, 69 Discrete data, 225
Convenience sampling, 130 calculation of median, 225–226
Coping Resources Inventory for Stress (CRIS), 79 calculation of mode, 226–227
Correlation coefficient, 195, 216, 233 Dispersion, 227–228
Correlation diagram, 222 Disproportionate stratified random sampling, 125
Counterbalancing, 70 Dotted diagrams, 222
Cover page, 263 Drug Controller General of India (DCGI), 7
CRIS. See Coping Resources Inventory for Stress (CRIS), 79
Criterion validity, 180 E
Criterion variables, 3 EBP. See Evidence-Based Practices (EBP), 5
Critical review, 275–283 Educators, 259
Criticism, 99–100 E-Journals, 270–271
Crossover design, 74–75 Electronic sources, 50–53
Cross-sectional design, 80 literature review, 50–54
Cumulative frequency curve, 222 ERIC database, 51
Cumulative Index to Nursing and Allied Health Literature Ethical consideration, establishing of, 14
(CINAHL), 53 Ethical dimension, 12
Ethical problem, 166
D Ethics, 290–291
Data, 3, 188–194 code of ethics, 292–293
analysis of, 188–194 Ethnographic research, 94–95
criticism, 99–100 characteristics, 95
graphical presentation of, 218–223 Ethnographical design, 16
measures to condense, 195 Ethnography, 94–95
table, 216 defined, 94–95
diagrams and graphs, 218–223 Evaluation research studies, 89–90
parts of, 217 Evidence-Based Practices (EBP), 5–6
Data analysis, 94 Exclusion bias, 144
Data collection, 14, 96–97, 149 Existential phenomenological research, 94
composite scale, 161–165 Experimental designs, 72–75
concept of, 150–153 types of, 72–75
criteria of evaluation, 153 Experimental effect, 66
interview, 157–161 Experimental research designs, 66–69
Likert scale, 162–163 Experimentation, 66–69
methods of, 151t Exploratory design, 82
observation, 165–168 Exponential discriminative snowball sampling, 132
questionnaire, 153–157 Exponential nondiscriminative snowball sampling, 132
selection of methods of, 151–153 External criticism, 99–100
semantic differential scale is, 163–164 External validity, 65–66
Data sources, 150–151 Extraneous variable, 3, 5
DCGI. See Drug Controller General of India (DCGI), 7
Declarative format, 28 F
Degree of freedom, 195, 243 Face validity, 179–180
Demographic variables, 3, 29 Factorial design, 73–74
Dependent variables, 3, 6 Feasibility, 25–26
Descriptive rating scales, 169 Female feticide, 22
Descriptive research designs, 80–82 Finite population correction, 137
features, 80 Fisher’s Exact Probability Test, 244
types of, 80–82 Flow charts, 206–208
Descriptive statistics, 216 Focused group interview, 159–160
 Index 333

Folklores, 24 I
Follow-up studies, 86
ICH. See International Committee on Harmonization (ICH), 305
Formal theory, 98
IEC. See Institutional Ethical Committee (IEC), 304
Format, 27–28
IGCP. See Indian Good Clinical Practices (IGCP), 305
declarative, 28
IGNOU. See Indira Gandhi National Open University (IGNOU)
interrogative, 28
IMRAD format, 261–262
Framework, 4
Infant mortality rate, 233–234
conceptual, 4
In vivo biophysiologic methods, 172t, 173–174
Frequency distribution table, 190
Independent variables, 3, 29
Frequency polygon, 220–221, 221f
In-depth interview, 159
India, 6–7
Indian Council of Medical Research (1980), 297–299
G
Indian Good Clinical Practices, The (IGCP), 305
Gaussian curve, 232 Inferential statistics, 191–194
Good research. See also Research Institutional Ethical Committee (IEC), 304
characteristics, 7–8 Instrumentation, 65
Graphic rating scale, 169 Internal consistency, 182
Graphical presentation of, 218–223 Internal criticism, 100
Graphs, 218, 223 Internal validity, 71
limitations of, 223 International Committee on Harmonization (ICH), 305
Grounded theory, 95–98 Internet documents, 271
characteristics, 96–98 Interrelated concepts, 4
data collection issues in, 97f Interrogative format, 28
types of, 98 Interval-level measurement, 149–150
Grounded-theory approach, 16 Interview
advantages, 161
benefits, 158
H
characteristics, 158
Hawthorne effect, 66 closing, 161
Health Research: A Community-Based Approach, 271 concept, 160
Healthy user bias, 144 definitions, 157–158
Hermeneutial phenomenological research, 94 disadvantages, 161
Histogram, 219–220 preinterview introduction, 160–161
Historical research design, 98–102 preparation for, 160
areas of historical study, 101 recording, 161
characteristics, 98–99 types, 158–160
methods and tools used in, 101–102 Interviewing process, 160–161
Historical research Introduction, writing of, 57–58
steps of, 99–101 In vitro biophysiologic methods, 173–174, 174f
Historical revisionism, 101
Holtzman Inkblot test (HIT), 175 J
Homogeneity, 70–71, 182
Human dignity, 291 Journal articles, 269–270
principle of, 291 Journals, 53
Hypotheses, 38 Justice, 290–291
academic literature, 38
K
associative, 38–39
casual, 39 Karl Pearson’s correlation coefficient, 234
characteristics of, 37 Key informants, 16
complex, 38 Knowing, 259
directional, 39 Kolmogorov–Smirnov Test, 243
nondirectional, 39
null, 39 L
previous research, 38 Likert questions, 155
real-life experiences, 38 Likert scale, 162–163
simple, 38 advantages, 163
sources of, 38 characteristics, 163
theoretical or conceptual frameworks, 38 definitions, 162
334 Index

Likert scale (Continued) Multiple bar diagram, 219f

disadvantages, 163 Multiple-choice questions, 154
scoring, 163 Multistage sampling, 126
uses, 164 Multivariate study, 30
Likert, Rensis, 162–163
Delimitations, 41–42 N
definition, 42
Narrative analysis, 208
methodological, 160
Nominal-level measurement, 149
theoretical, 160
Nondirectional hypothesis, 39
Line graph, 221–222
Nondirective interview, 159
Linear snowball sampling, 132
Nonequivalent control group design, 77
Linear Relationship, 237
Nonexperimental research design, 79
List of tables, 264
advantages, 86–87
List of figures, 264
disadvantages, 87
Literature review, 12–13, 15–16, 47
Nonexperimental research designs, 79
importance of, 47–48
need of, 79–80
meaning of, 47
types of, 80
primary sources, 50
Nonparametric tests, 196–198, 243
purpose of, 48
advantages, 244
secondary sources, 50–54
disadvantages, 244
steps of, 54–58
Nonparticipant observation, 167
Longitudinal research design, 86
Nonprobability sampling, 127–133
Lottery method, 122
defined, 127–128
features of, 128
M types of, 128–133
Macroethnography, 94–95 uses of, 128
Main text, 264–265 Nonrandomized control group design, 77
Maintaining belief, 14 Nonstandardized interviews, 159
Manipulated variables, 3 Normal probability curve (Z-score), 232–233
Manipulation, 4, 69 Normal probability curve, 232–233
Mann–Whitney U–Test, 244, 253–254 importance of, 232
Map diagram, 223 properties of, 232–233
Matching, 70 Note taking, 96
Matrices, 205 Novelty effect, 66
Matrix questions, 155 Null hypothesis, 39
Maternal mortality rate, 28 Numerical rating scale, 169
McNemar Test, 244 Nurse administrators, 259
Measures of central tendency, 223–227
Measures of dispersion, 227–228 O
Median, 225–226
Observation, 197
Median Test, 244
as method of data collection, 165–168
Medical Literature Analysis and Retrieved System Online
planning for, 167–168
(MEDLINE), 51
Ogive, 222
MedlinePlus, 51
One-stage cluster sampling, 126
Memoing, 96f
Online databases, 270
Meta-analysis, defined, 87–88
Online journals, 274
Methodological dimension, 12
Open format questions, 154
Methodological delimitation, 42
Operational research, 90
Methodological studies, 87
Oral reports, 260
Microethnography, 94–95
Ordinal-level measurement, 149
Microsoft Excel, 204
Outcome research, 89
Minitab, 199–200
Miscellaneous tables, 218
Mode, 226–227 P
Moderately negative correlation, 234 Paired t-test, 245
Moderately positive correlation, 233–234 Panel studies, 86
Morning sickness, 11–12 Parametric tests, 243
Mortality, 65 Participant observation, 167
 Index 335

Pearson’s correlation coefficient formula, 182 Qualitative data collection, 168–169

Pearson’s product-moment correlation coefficient (PMCC), 237 types of, 168–169
Peer support, 25 Qualitative research design, 90–92, 93t
Perfect negative correlation, 233 phases of, 91–92
Perfect positive correlation, 233 types of, 92, 93t
Phenomenological design, 16 Qualitative research, 8
Phenomenological research, 92, 94b example of statement of research problems and variables in, 30t
characteristics, 92–94 features, 9–10
Phenomenology,defined, 92–94 vs quantitative research, 10t
Photography, 101 types, 10
Pictograms, 222–223 Qualitative research process, 15–18
Picture diagram, 222–223, 222f Qualitative studies, 133
Pie diagram, 219 sample size in, 133
Pilot study, 5, 183–184 Quantitative research, 8–9
Population, 4 example of, 9b
accessible, 4 vs qualitative research, 10t
target, 4 steps of, 12f, 15–17
Population specification, 14 Quantitative research designs, 67–68t, 90–92
Positional average, 240 Quasi-experimental research design, 76–78
Posttest-only control design, 72 advantages, 78
Power analysis, 134 characteristics, 80
Practical dimension, 12 disadvantages, 78
Predictive validity, 180 types, 78–79
Pre-experimental research design, 78–79 Quasi-experimental study, 13, 30
advantages, 79 Questionnaire, 153–157
disadvantages, 79 advantages of, 157
types, 80 definitions, 154
Preliminary pages, 263–264 designing of, 155–157
Pretest-posttest-only design, 72 disadvantages of, 157
Primary sources types, 154–155
literature review, 54–58 Questionnaire administration, 156
Principle of justice, 290–291 Quintuplets, 25
Principle of respect of human dignity, 291 Quota sampling, 131
Printed sources, 53–54
Probability level, 137–138 R
Probability sampling, 122–127
Random assignment of subjects, 77
features of, 122
Randomization, 71
types of, 120f
Randomized block design, 74
Problem statement, 22
Range, 227
components, 22
Rank order questions, 154
Product moment correlation, 237–239
Rating questions, 155
Projective techniques, 174–176
Rating scale, 168–170
definition, 174
Ratio-level measurement, 150
Proportion bar diagram, 220f
Reactive effect of the pretest, 66
Proportionate stratified random sampling, 124–125
Realistic phenomenological research, 94
Proposal. See Research proposal
Real-life experiences, 38
Proposition, 4
Recontextualizing, 200
PsycINFO database, 51
Regression analysis, 239–241
PubMed, 51
Regression Line, 240
Purposive sampling, 129–130
Reliability, 4, 180–183
uses of, 128
Representative sample, 4
types of, 128–133
Requests for proposals (RFPs), 23
Research
Q applied, 10–11
Q-sorts, 176–177 basic, 10–11
Qualitative data analysis, 200, 208 characteristics of, 14
establishing the validity of, 210–211 operational, 90
steps of, 200–210 outcome, 90
336 Index

Research (Continued) Scientific methods, 7–8

qualitative, 8 characteristics of, 7–8
quantitative, 8 Secondary data analysis, 88
uses of assumption, 103 Secondary sources
Research article, 274–275 literature review, 50b
writing, 274–275 Sector diagram, 219
factors affecting selection of, 63–64 Selection bias, 65
types of, 66–76 Selection
validity of, 64–66 participants, 290
Research hypothesis, 39 of research design, 92–103
Research objectives, 64–66 Self-selection bias, 144
action verbs used to formulate, 35t Semantic differential scale, 163–164, 165f
characteristics of, 33 advantages of, 164
meaning of, 33 definition, 164
method of stating, 34 designing, 164
need of, 33–34 disadvantages of, 164
types of, 34 uses of, 165
Research problem, 11–12, 22–28 Semistructured interview, 159
criteria for selecting, 24–26 Sentence completion test, 175
formulation of, 26–28 Sequential sampling, 127
guidelines for evaluation of, 27t Sigma Theta Tau International Honor Society of Nursing, 53
Research proposal, 279–283 Sign test, 244
developing, 279–283 Simple bar diagram, 219f
format of, 280–281 Simple hypothesis, 38–39
purpose of, 240 Simple random sampling, 122
Research report, 261–263, 264 demerits, 123
characteristics, 261 merits, 123
format of, 261–262, 263–265 Snowball sampling, 132
steps involved in writing, 262–263 Solomon four-group design, 72–73
Research study setting, 4 Spearman’s rank correlation coefficient, 234–235
Research variables, 29 Specific objectives, 34
Review, 4 Split-half reliability of research instrument, 182
Rorschach test, 175 Spot map, 293
Rosenzweig test, 175 SPSS. See Statistical Package for Social Sciences (SPSS)
Roy’s Adaptation Model, 38 Statistics
defined, 216
S descriptive, 216
inferential, 242
Sample selection, 119
scale of measurement, 149–150
Sample size, 120
Story completion test, 175
factors affecting, 142–143
Stratified random sampling
sampling errors and, 144
demerits, 125
statistical determining the, 136
merits, 125
Sampling, 116–117
Structured interview, 158–159
factors influencing, 119–120
Structured observation, 167
nonprobability sampling, 127–133
Substantive dimension, 12
presentation of, 117f
Substantive theory, 98
probability sampling, 120
Superscript format, 79, 89
process, 116–117
Symbolic presentations, 71
purposes of, 118
Systematic random sampling, 125
reasons of errors, 143–145
demerits, 126
stages of, 119f
merits, 126
types of techniques, 120
Sampling bias, 144
Sampling errors, 144, 145 T
sample size and, 144 Table of contents, 264
Scattered diagram, 222f Tables, 282
Scatter diagram method, 235 Tabulation, 189
Scattered or dotted diagrams, 222 principles of, 216–217
 Index 337

Target population, 4 Variables, 28–30

TAT. See Thematic Apperception Test (TAT), 175 demographic, 29
Techniques of data collection, 151 dependent, 29
Telephone interview, 160 extraneous, 29
Test of significance, 194 independent, 28–29
Test–retest reliability (stability), 181 operational definition, 30–32
Thematic Apperception Test (TAT), 175 research, 29
Theoretical framework, 4 Variance, 138
Theoretical delimitations, 42 Ventilator-associated pneumonia (VAP), 26f
Thesis, 263–265 Vignettes, 177–179
Time-series design, 77–78 Designing, 178–179
Tomkins–Horn picture arrangement test, 175 Visual analogous scale, 164–165
Total enumerative sampling, 131 Definitions, 164–165
Trend studies, 86 uses of, 165
True experimental design, 69–76
advantages, 75–76
characteristics, 70t W
disadvantages, 75–76 Wilcoxon Signed Rank Test, 252–253
t-Tests, 245–246 Word association technique, 174–175
Two-stage cluster sample, 126 Writing, 32–34
Type-I error, 242 Vancouver style of references, 265–266
Type-II error, 242 Writing the Reference, 267–271
Format of Writing the Reference, 267–271
U references/bibliography, 265–274
research articles, 274–275
Univariant descriptive designs, 81
research report, 50f
Univariate study, 29
Written literature, 46–61
Unpaired t-test, 245
Written report, 260
Unstructured interview, 159
Written review literature, 58f
Unstructured observation, 167
elements of, 58f
Unwarranted assumptions, 238
Written survey, 92t
Utilization of research findings, 15

V Z
Validity, 5 Z score, 232–233
VAP. See Ventilator-associated pneumonia (VAP), 29 Z-test, 246–247
This page intentionally left blank