Statistics Explained, 4th Edition
Visit the link below to download the full version of this book:
https://medipdf.com/product/statistics-explained-4th-edition/
Click Download Now
Statistics Explained
Fourth edition
PERRY R. HINTON
Designed cover image: Getty
Fourth edition published 5
by Routledge
Park Square Milton Park Abingdon Oxon OX RN
and by Routledge
6 5 Third Avenue New York NY 58
Routledge is an imprint of the Taylor Francis Group an informa business
© 5 Perry R Hinton
The right of Perry R Hinton to be identi ed as author of this work has been
asserted in accordance with sections 77 and 78 of the Copyright Designs
and Patents Act 988
All rights reserved No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic mechanical or other means now
known or hereafter invented including photocopying and recording or in
any information storage or retrieval system without permission in writing
from the publishers
Trademark notice: Product or corporate names may be trademarks or reg
istered trademarks and are used only for identi cation and explanation
without intent to infringe
First edition published by Routledge 995
Second edition published by Routledge
Third edition published by Routledge
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN: 978 67 66 8 hbk
ISBN: 978 67 66 5 pbk
ISBN: 978 9 5 6 ebk
DOI: 978 9 5 6
Typeset in Berkeley Stone Sans
by Apex CoVantage LLC
Access the Support Material: www routledge com 978 67 66 5
To Anna Anthony and Emma
Contents
List of figures xi
Prefacexiii
CHAPTER 1 INTRODUCTION1
CHAPTER 2 THE ELEMENTS OF STATISTICAL ANALYSIS 7
THE IMPORTANCE OF SCORES AND POPULATIONS 7
COLLECTING DATA: MEASUREMENT 8
LOCATING SCORES IN A POPULATION 10
DESCRIBING THE POPULATION 10
THE POSITION OF A SCORE IN A FREQUENCY DISTRIBUTION 15
PROPORTIONS, PROBABILITIES AND PERCENTAGES 16
THE IMPORTANCE OF SAMPLES AND SAMPLING DISTRIBUTIONS 23
SAMPLE STATISTICS 24
THE SAMPLING DISTRIBUTION 26
KEY POINTS 28
CHAPTER 3 NULL HYPOTHESIS SIGNIFICANCE TESTING 31
THE LOGIC OF HYPOTHESIS TESTING USING STATISTICS 31
HYPOTHESIS TESTING WITH A SAMPLE 38
HYPOTHESIS TESTING WITH A SAMPLE WHEN THE POPULATION
STANDARD DEVIATION IS NOT KNOWN 40
CONFIDENCE INTERVALS 44
HYPOTHESIS TESTING WITH TWO SAMPLES
(THE TWO-SAMPLE t TEST) 46
THE PAIRED (RELATED) SAMPLES t TEST 50
viii CONTENTS
THE INDEPENDENT SAMPLES t TEST 52
ENTERING DATA INTO A STATISTICS PROGRAM: GROUPING VARIABLES 56
KEY POINTS 56
CHAPTER 4 ISSUES WITH NULL HYPOTHESIS SIGNIFICANCE TESTING 59
VARIABLES 60
TYPE I AND TYPE II ERRORS 64
STATISTICAL POWER 66
EFFECT SIZE 68
NORMALLY DISTRIBUTED POPULATIONS 73
SAMPLE SKEW AND KURTOSIS 78
TESTS OF NORMALITY 82
OUTLIERS 84
SAMPLE VARIATION 86
BOOTSTRAPPING 87
ANALYZING A SAMPLE 90
KEY POINTS 91
CHAPTER 5 THE ANALYSIS OF VARIANCE 93
BEYOND THE t TEST 93
WHY DO SCORES VARY IN A SET OF DATA? 94
THE VARIANCE RATIO (THE F STATISTIC) 96
ONE-FACTOR INDEPENDENT ANOVA 98
MULTIPLE COMPARISONS 102
THE RELATIONSHIP OF F TO t 109
ONE-FACTOR REPEATED MEASURES ANOVA 110
SPHERICITY 116
MULTIPLE COMPARISONS 118
KEY POINTS 119
CHAPTER 6 THE INTERACTION OF FACTORS IN THE ANALYSIS OF
VARIANCE 121
MAIN EFFECTS AND THE INTERACTION 121
THE TWO-FACTOR INDEPENDENT ANOVA 123
THE TWO-FACTOR MIXED DESIGN ANOVA 129
THE TWO-FACTOR REPEATED MEASURES ANOVA 136
KEY POINTS 143
CHAPTER 7 NONPARAMETRIC ANALYSIS 145
PARAMETRIC AND NONPARAMETRIC TESTS 145
NOMINAL DATA, CATEGORIES AND FREQUENCY COUNTS 145
ANALYZING FREQUENCY DATA: CHI-SQUARE (χ2) 146
CONTENTS ix
CHI-SQUARE (χ2) AS A GOODNESS OF FIT TEST 148
CHI-SQUARE (χ ) AS A TEST OF INDEPENDENCE
2
151
THE CHI-SQUARE DISTRIBUTION 153
THE ASSUMPTIONS OF THE χ2 TEST 154
NONPARAMETRIC COMPARISONS WITH ORDINAL DATA 155
CALCULATING RANKS 157
THE MANN-WHITNEY U TEST (FOR INDEPENDENT SAMPLES) 159
THE WILCOXON SIGNED-RANKS TEST (FOR PAIRED SAMPLES) 164
ONE-FACTOR ANOVA FOR RANKED DATA 167
THE KRUSKAL-WALLIS TEST (FOR INDEPENDENT MEASURES) 167
THE FRIEDMAN TEST (FOR RELATED SAMPLES) 173
TREATING ORDINAL DATA AS INTERVAL DATA 178
KEY POINTS 179
CHAPTER 8 CORRELATION AND REGRESSION 181
CORRELATION 181
THE PEARSON CORRELATION COEFFICIENT r 183
THE SPEARMAN rS CORRELATION COEFFICIENT 186
LINEAR REGRESSION 188
THE INTERPRETATION OF CORRELATION AND REGRESSION 192
THE SIGNIFICANCE OF THE REGRESSION 194
HOMOSCEDASTICITY 197
BINARY LOGISTIC REGRESSION 198
MULTIPLE CORRELATION 201
MULTIPLE REGRESSION 206
KEY POINTS 213
CHAPTER 9 MODELLING RELATIONSHIPS
BETWEEN VARIABLES 215
FACTOR MODELS 215
RELIABILITY 216
FACTOR ANALYSIS 218
MULTIVARIATE ANALYSIS OF VARIANCE (MANOVA) 222
DISCRIMINANT FUNCTION ANALYSIS 225
STRUCTURAL EQUATION MODELLING 227
MODERATION AND MEDIATION 232
MODERATION 233
MEDIATION 238
LATENT VARIABLES: REFLECTIVE AND FORMATIVE CONSTRUCTS 239
CONCLUSION TO STRUCTURAL EQUATION MODELLING 241
KEY POINTS 242
x CONTENTS
CHAPTER 10 AN INTRODUCTION TO THE GENERAL LINEAR MODEL 245
MODELS 245
A LINEAR MODEL 246
MODELLING DATA 248
THE MODEL AS A REGRESSION EQUATION 251
SELECTING A GOOD MODEL 253
THE VARIATION IN THE DATA EXPLAINED BY THE MODEL 255
THE LINEAR MODEL AND CORRELATION 256
THE LINEAR MODEL AND THE ANALYSIS OF VARIANCE 256
COMPARING SAMPLES (THE ANALYSIS OF VARIANCE ONCE AGAIN) 258
EXPLAINING VARIATIONS IN THE DATA BY A MODEL 260
THE GENERAL LINEAR MODEL (GLM) 261
THE GENERAL LINEAR MODEL AND MULTIVARIATE ANALYSIS 264
KEY POINTS 265
CHAPTER 11 BAYESIAN STATISTICS 267
FREQUENTIST AND BAYESIAN APPROACHES TO STATISTICS 267
AN ILLUSTRATIVE EXAMPLE: IS THAT A GORILLA? 269
CONDITIONAL PROBABILITY AND BAYES’ THEOREM 271
THE BAYESIAN DILEMMA 273
A BAYESIAN APPROACH TO STATISTICAL INFERENCE 274
CREATING BAYESIAN EQUIVALENT OF NULL HYPOTHESIS
SIGNIFICANCE TESTS 276
COMPARISON OF FREQUENTIST AND BAYESIAN TESTS 279
THE WIDER IMPORTANCE OF BAYESIAN METHODS 281
KEY POINTS 284
CHAPTER 12 ‘LIES, DAMNED LIES, AND STATISTICS’ 287
EXPERIMENTER EFFECTS 288
DEMAND CHARACTERISTICS 289
PUBLICATION BIAS 289
DATA MANIPULATION 292
THE REPLICATION PROBLEM 295
BAYES AND BEYOND 296
References 299
Appendix 303
Glossary 325
Index 337
Choosing a statistical test 342
Figures
The frequency distribution of the social science examination marks
this year
A normal distribution
The normal distribution curve superimposed on the frequency
distribution of the social science examination scores
The frequency distribution for the Sparcote students’ social science
examination marks
5 The sampling distribution of the social science scores for samples of
size 6 7
A hothousing e ect of plus IQ points
A one tailed prediction and the signi cance level 6
A two tailed prediction and the signi cance level 7
Hypothesis testing with a sample of Cyadmine a ected babies
5 Examples of the t distribution
6 The distribution of the di erence between sample means 7
The risk of a Type I and Type II error 6
The e ect of increasing the sample size on the overlap of the
distributions 7
Q Q Plot of the social science examination marks to see if they are
normally distributed 7
Skewness in frequency distributions left to right: positive skew
symmetrical and negative skew 75
5 Distributions and kurtosis 77
5 An example of an F distribution degrees of freedom = 8 97
5 The group means for the anagram study
5 The group means for the three controllers 8
6 Examples of an interaction
6 No interaction but a a main e ect of specialism and b a main
e ect of school
6 Language improvement after a student exchange 8
xii FIGURES
6 The mean number of errors over a three week period for the two
types of operator
65 The interaction of vehicle and shift on delivery times
7 The chi square distribution 5
8 Scatterplot of study time by examination mark 8
8 Finding the regression line by minimizing the error values E 9
8 Regression of Y on X and the regression of X on Y 9
8 The coe cient of determination r 9
85 The scatterplot of smiling time by items sold 98
86 The variability of the scores on three variables
9 Scree plot of the eigenvalues
9 A simple path model 7
9 A path model showing the proposal of the relationship between
the human population and stork population 9
9 The path model with the path weights
95 The path from X to Y
96 The path from X to Y moderated by M
97 Moderation shown as main e ects and interaction
98 The path from X to Y mediated by M
99 Scatter plot of the results showing the regression lines of the two
opinions 6
9 A re ective latent variable
9 A formative latent variable
A graph of card sorting times 7
A plot of the children’s general knowledge scores by age 9
A proposed linear relationship between general knowledge and age 9
Separating each score into a predicted score plus residual 5
5 Di erent linear models 5
6 The same linear model for two sets of data 55
7 Plot of general knowledge scores for each class of students 59
8 A linear model for the class data 59
The three distributions in Bayesian analysis 75
A noninformative prior 76
Preface
This book is written for a wide range of students and researchers primarily in the
behavioural human and social sciences who wish to learn the logic and method of
statistical analysis that underlies much of the numerical data analysis of research in
their area of study and will also employ statistical analysis in their own research work
More generally it should be useful for anyone who wishes to understand the analysis
of quantitative research data as presented in the academic literature Over a number
of years I have taught statistical analysis to undergraduate students of psychology
communication and media linguistics nursing education early childhood studies
health social work and sociology; and supervised postgraduate students in a number
of these subject areas I am grateful to the many hundreds of students who have
been in my statistics classes and commented favourably mostly on the experience
which was not necessarily something that they had looked forward to with delight
The feedback questions and criticisms of these students have helped me understand
the problems and pleasures of learning statistics I have learned a lot And these
experiences have fed back into the writing of this book which is now in its fourth
edition I found it particularly gratifying when one of my students said that she could
hear me speaking when she read the book I hope other readers feel like her that the
book reads like someone talking – hopefully a friendly helpful voice – as I believe that
there is no better learning experience than having someone explain something clearly
This fourth edition has been fully revised and reorganized The book now
begins with a focus on the population as the starting point for statistical analysis
and the location of a score in the distribution of scores in a population to explain
how statistical tests have been constructed It emphasizes the importance of the null
hypothesis signi cance testing method more explicitly as underlying these tests It
still retains the explanation of the important statistical tests employed by researchers
in the behavioural human and social sciences However this has been enhanced to
include logistic regression; and mediation and moderation in path analysis The book
also now includes a new chapter on Bayesian statistics as an alternative to the null
hypothesis signi cance testing method and a new nal chapter on ‘lies damn lies
and statistics’ reviewing the potential for bad practice in the statistical analysis of
research data and in the reporting of that data particularly in the elds that are
xiv PREFACE
associated with the social sciences and the ways of reducing it It is hoped that the
reader will not only learn how and why a range of common statistical tests have been
devised and how to interpret their outcomes but will also be able to critically assess
the value of the di erent tests in di erent circumstances
Although I am happy to welcome all readers the book is written for students
and researchers with little previous experience of or background in mathematical
statistical analysis but want to get a job done – analyze their research data and
understand the data analysis of other researchers Also rather than simply following
the instructions of a statistical analysis computer program they would like to know
how a statistical test is constructed and how it works explained in ways that a non
statistician can understand My goal has been to provide clear explanations with as
little jargon as possible I have certainly found that my students have appreciated this
approach and that not only do they understand what the tests involve and what they
do but they have also gained an ability to be aware of the limitations of such tests and
not accept the output from computer data analysis uncritically
I would like to thank a number of colleagues who have helped me over the years:
Sue Wilkinson for encouraging me to write the book in the rst place; Margaret
Manning David French Paul Hartmann and the students of my Analysis of Experiments
class for their support and encouragement in producing the rst edition; the students
of Research and Experimentation II who for seven years could interrogate the author
of their set text; Victoria West and Alfredo Gaitan my colleagues on that course; Ian
Robertson and Bob Cozens I would like to thank my co authors of SPSS Explained
also published by Routledge and now in its third edition: Peter Terry for joining the
team for the new edition and Charlotte Brownlow and Isabella McMurray who have
been there from the beginning when we all worked together helping students analyze
their data with SPSS I am particularly grateful to Isabella who has jointly written
a book on SPSS with me when together we made the whole process even more
enjoyable by our co ee break conversations about everything from Japanese culture
to British canal boats Finally I would like to thank all my current colleagues in the
Department of Applied Linguistics at the University of Warwick for a stimulating
and friendly academic environment and speci cally Clay Beckner whose excellent
teaching of statistical methods I have enjoyed moderating
Introduction
This book provides a guide for readers primarily students and researchers in the
behavioural human and social sciences who want to understand the logic of
statistical analysis and the common statistical procedures that are used in the analysis
of quantitative research data It does not assume any mathematical knowledge
beyond that learnt by most people in high school and dips into the statistics only as
far as is necessary to understand what is going on The book explains how and why
the statistical tests are performed using the walkthrough method with illustrative
examples to explain the logic procedure and results of the tests It is assumed that
anyone analyzing research data will use one of the increasing number of excellent
software packages such as SPSS the R software or JASP speci cally designed for
data analysis or indeed a spreadsheet program such as Microsoft Excel which
can be used to perform a surprising number of the tests described in this book
However the current book provides a useful and explanatory account of statistical
analysis regardless of which statistical computer package you use With any statistics
application ‘app’ it often appears that we input the data and ‘magically’ a statistical
conclusion pops up telling us whether our result is ‘signi cant’ or not It is relatively
easy to perform the tests once you know how to use an app but the point of the
current book is to explain why and how the tests have been devised to do what they
do – how they work An app does not explain what the outcome of the test means
which is the aim of this book The reader will learn that statistical analysis is about
examining the di erences found in a study typically between the average scores of
di erent groups on a performance measure and comparing these di erences with the
di erences that would be expected by chance alone If the former exceeds the latter
by a certain amount then the result is claimed to be signi cant and the idea that it
could have arisen by chance alone is rejected And all this is done statistically – in a
logical and formal way – giving numerical values to the outcomes Throughout the
book a range of di erent statistical tests along with their underlying logic will be
explained and illustrated But rst we will begin by examining why we want to do
statistical analysis in the rst place
The word quantitative refers to research involving quantities which we measure
by numbers: typically involving the how much the how many the how long type of
DOI: 10.4324/9780429353260-1
