Statistics Explained An Introductory Guide for Life Scientists

- 2nd Edition

Visit the link below to download the full version of this book:

https://medipdf.com/product/statistics-explained-an-introductory-guide-for-life-
scientists-2nd-edition/

Click Download Now

C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TTL.3D iii [3–3] 27.8.2011 10:33AM

Statistics Explained
An Introductory Guide for Life Scientists

SECOND EDITION

Steve McKillup
Central Queensland University,
Rockhampton
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518IMP.3D iv [4–4] 27.8.2011 10:36AM

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town,

Singapore, São Paulo, Delhi, Tokyo, Mexico City

Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org
Information on this title: www.cambridge.org/9781107005518

© S. McKillup 2012

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 2012

Printed in the United Kingdom at the University Press, Cambridge

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication data

ISBN 978-1-107-00551-8 Hardback

ISBN 978-0-521-18328-4 Paperback

Additional resources for this publication at www.cambridge.org/9781107005518

Cambridge University Press has no responsibility for the persistence or

accuracy of URLs for external or third-party internet websites referred to in
this publication, and does not guarantee that any content on such websites is,
or will remain, accurate or appropriate.
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D v [5–12] 27.8.2011 11:48AM

Contents

Preface page xiii

1 Introduction 1
1.1 Why do life scientists need to know about experimental
design and statistics? 1
1.2 What is this book designed to do? 5

2 Doing science: hypotheses, experiments and disproof 7

2.1 Introduction 7
2.2 Basic scientific method 7
2.3 Making a decision about an hypothesis 11
2.4 Why can’t an hypothesis or theory ever be proven? 11
2.5 ‘Negative’ outcomes 12
2.6 Null and alternate hypotheses 12
2.7 Conclusion 14
2.8 Questions 14

3 Collecting and displaying data 15

3.1 Introduction 15
3.2 Variables, experimental units and types of data 15
3.3 Displaying data 17
3.4 Displaying ordinal or nominal scale data 23
3.5 Bivariate data 25
3.6 Multivariate data 26
3.7 Summary and conclusion 28

4 Introductory concepts of experimental design 29

4.1 Introduction 29
4.2 Sampling – mensurative experiments 30
4.3 Manipulative experiments 34
4.4 Sometimes you can only do an unreplicated experiment 41

v
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D vi [5–12] 27.8.2011 11:48AM

vi Contents

4.5 Realism 42
4.6 A bit of common sense 43
4.7 Designing a ‘good’ experiment 44
4.8 Reporting your results 45
4.9 Summary and conclusion 46
4.10 Questions 46

5 Doing science responsibly and ethically 48

5.1 Introduction 48
5.2 Dealing fairly with other people’s work 48
5.3 Doing the experiment 50
5.4 Evaluating and reporting results 52
5.5 Quality control in science 53
5.6 Questions 54

6 Probability helps you make a decision

about your results 56
6.1 Introduction 56
6.2 Statistical tests and significance levels 57
6.3 What has this got to do with making a decision about
your results? 60
6.4 Making the wrong decision 60
6.5 Other probability levels 61
6.6 How are probability values reported? 62
6.7 All statistical tests do the same basic thing 63
6.8 A very simple example – the chi-square test
for goodness of fit 64
6.9 What if you get a statistic with a probability
of exactly 0.05? 66
6.10 Statistical significance and biological significance 67
6.11 Summary and conclusion 69
6.12 Questions 70

7 Probability explained 71
7.1 Introduction 71
7.2 Probability 71
7.3 The addition rule 71
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D vii [5–12] 27.8.2011 11:48AM

Contents vii

7.4 The multiplication rule for independent events 72

7.5 Conditional probability 75
7.6 Applications of conditional probability 77

8 Using the normal distribution to make statistical

decisions 87
8.1 Introduction 87
8.2 The normal curve 87
8.3 Two statistics describe a normal distribution 89
8.4 Samples and populations 93
8.5 The distribution of sample means is also normal 95
8.6 What do you do when you only have data from one
sample? 99
8.7 Use of the 95% confidence interval in significance testing 102
8.8 Distributions that are not normal 102
8.9 Other distributions 103
8.10 Other statistics that describe a distribution 105
8.11 Summary and conclusion 106
8.12 Questions 106

9 Comparing the means of one and two samples

of normally distributed data 108
9.1 Introduction 108
9.2 The 95% confidence interval and 95% confidence limits 108
9.3 Using the Z statistic to compare a sample mean and
population mean when population statistics are known 108
9.4 Comparing a sample mean to an expected value when
population statistics are not known 112
9.5 Comparing the means of two related samples 116
9.6 Comparing the means of two independent samples 118
9.7 One-tailed and two-tailed tests 121
9.8 Are your data appropriate for a t test? 124
9.9 Distinguishing between data that should be analysed by a
paired sample test and a test for two independent samples 125
9.10 Reporting the results of t tests 126
9.11 Conclusion 127
9.12 Questions 128
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D viii [5–12] 27.8.2011 11:48AM

viii Contents

10 Type 1 error and Type 2 error, power and sample size 130
10.1 Introduction 130
10.2 Type 1 error 130
10.3 Type 2 error 131
10.4 The power of a test 135
10.5 What sample size do you need to ensure the risk of Type 2
error is not too high? 135
10.6 Type 1 error, Type 2 error and the concept
of biological risk 136
10.7 Conclusion 138
10.8 Questions 139

11 Single-factor analysis of variance 140

11.1 Introduction 140
11.2 The concept behind analysis of variance 141
11.3 More detail and an arithmetic example 147
11.4 Unequal sample sizes (unbalanced designs) 152
11.5 An ANOVA does not tell you which particular treatments
appear to be from different populations 153
11.6 Fixed or random effects 153
11.7 Reporting the results of a single-factor ANOVA 154
11.8 Summary 154
11.9 Questions 155

12 Multiple comparisons after ANOVA 157

12.1 Introduction 157
12.2 Multiple comparison tests after a Model I ANOVA 157
12.3 An a posteriori Tukey comparison following a significant
result for a single-factor Model I ANOVA 160
12.4 Other a posteriori multiple comparison tests 162
12.5 Planned comparisons 162
12.6 Reporting the results of a posteriori comparisons 164
12.7 Questions 166

13 Two-factor analysis of variance 168

13.1 Introduction 168
13.2 What does a two-factor ANOVA do? 170
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D ix [5–12] 27.8.2011 11:48AM

Contents ix

13.3 A pictorial example 174

13.4 How does a two-factor ANOVA separate out the effects of
each factor and interaction? 176
13.5 An example of a two-factor analysis of variance 180
13.6 Some essential cautions and important complications 181
13.7 Unbalanced designs 192
13.8 More complex designs 192
13.9 Reporting the results of a two-factor ANOVA 193
13.10 Questions 194

14 Important assumptions of analysis of variance,

transformations, and a test for equality of variances 196
14.1 Introduction 196
14.2 Homogeneity of variances 196
14.3 Normally distributed data 197
14.4 Independence 201
14.5 Transformations 201
14.6 Are transformations legitimate? 203
14.7 Tests for heteroscedasticity 204
14.8 Reporting the results of transformations and the
Levene test 205
14.9 Questions 207

15 More complex ANOVA 209

15.1 Introduction 209
15.2 Two-factor ANOVA without replication 209
15.3 A posteriori comparison of means after a two-factor
ANOVA without replication 214
15.4 Randomised blocks 214
15.5 Repeated-measures ANOVA 216
15.6 Nested ANOVA as a special case of a single-factor ANOVA 222
15.7 A final comment on ANOVA – this book is only an
introduction 229
15.8 Reporting the results of two-factor ANOVA without
replication, randomised blocks design, repeated-measures
ANOVA and nested ANOVA 229
15.9 Questions 230
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D x [5–12] 27.8.2011 11:48AM

x Contents

16 Relationships between variables: correlation

and regression 233
16.1 Introduction 233
16.2 Correlation contrasted with regression 234
16.3 Linear correlation 234
16.4 Calculation of the Pearson r statistic 235
16.5 Is the value of r statistically significant? 241
16.6 Assumptions of linear correlation 241
16.7 Summary and conclusion 242
16.8 Questions 242

17 Regression 244
17.1 Introduction 244
17.2 Simple linear regression 244
17.3 Calculation of the slope of the regression line 246
17.4 Calculation of the intercept with the Y axis 249
17.5 Testing the significance of the slope and the intercept 250
17.6 An example – mites that live in the hair follicles 258
17.7 Predicting a value of Y from a value of X 260
17.8 Predicting a value of X from a value of Y 260
17.9 The danger of extrapolation 262
17.10 Assumptions of linear regression analysis 263
17.11 Curvilinear regression 266
17.12 Multiple linear regression 273
17.13 Questions 281

18 Analysis of covariance 284

18.1 Introduction 284
18.2 Adjusting data to remove the effect of a confounding
factor 285
18.3 An arithmetic example 288
18.4 Assumptions of ANCOVA and an extremely important
caution about parallelism 289
18.5 Reporting the results of ANCOVA 295
18.6 More complex models 296
18.7 Questions 296
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D xi [5–12] 27.8.2011 11:48AM

Contents xi

19 Non-parametric statistics 298

19.1 Introduction 298
19.2 The danger of assuming normality when a population is
grossly non-normal 298
19.3 The advantage of making a preliminary inspection
of the data 300

20 Non-parametric tests for nominal scale data 301

20.1 Introduction 301
20.2 Comparing observed and expected frequencies: the
chi-square test for goodness of fit 302
20.3 Comparing proportions among two or more independent
samples 305
20.4 Bias when there is one degree of freedom 308
20.5 Three-dimensional contingency tables 312
20.6 Inappropriate use of tests for goodness of fit and
heterogeneity 312
20.7 Comparing proportions among two or more related
samples of nominal scale data 314
20.8 Recommended tests for categorical data 316
20.9 Reporting the results of tests for categorical data 316
20.10 Questions 318

21 Non-parametric tests for ratio, interval or

ordinal scale data 319
21.1 Introduction 319
21.2 A non-parametric comparison between one sample and
an expected distribution 320
21.3 Non-parametric comparisons between two independent
samples 325
21.4 Non-parametric comparisons among three or more
independent samples 331
21.5 Non-parametric comparisons of two related samples 335
21.6 Non-parametric comparisons among three or more
related samples 338
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518TOC.3D xii [5–12] 27.8.2011 11:48AM

xii Contents

21.7 Analysing ratio, interval or ordinal data that show gross

differences in variance among treatments and cannot be
satisfactorily transformed 341
21.8 Non-parametric correlation analysis 342
21.9 Other non-parametric tests 344
21.10 Questions 344

22 Introductory concepts of multivariate analysis 346

22.1 Introduction 346
22.2 Simplifying and summarising multivariate data 347
22.3 An R-mode analysis: principal components analysis 348
22.4 Q-mode analyses: multidimensional scaling 361
22.5 Q-mode analyses: cluster analysis 368
22.6 Which multivariate analysis should you use? 372
22.7 Questions 374

23 Choosing a test 375

23.1 Introduction 375

Appendix: Critical values of chi-square, t and F 388

References 394
Index 396
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518PRF.3D xiii [13–14] 27.8.2011 10:50AM

Preface

If you mention ‘statistics’ or ‘biostatistics’ to life scientists, they often look

nervous. Many fear or dislike mathematics, but an understanding of sta-
tistics and experimental design is essential for graduates, postgraduates and
researchers in the biological, biochemical, health and human movement
sciences.
Since this understanding is so important, life science students are usually
made to take some compulsory undergraduate statistics courses. Nevertheless,
I found that a lot of graduates (and postgraduates) were unsure about design-
ing experiments and had difficulty knowing which statistical test to use (and
which ones not to!) when analysing their results. Some even told me they had
found statistics courses ‘boring, irrelevant and hard to understand’.
It seemed there was a problem with the way many introductory biosta-
tistics courses were presented, which was making students disinterested and
preventing them from understanding the concepts needed to progress to
higher-level courses and more complex statistical applications. There
seemed to be two major reasons for this problem and as a student I
encountered both.
First, a lot of statistics textbooks take a mathematical approach and often
launch into considerable detail and pages of daunting looking formulae
without any straightforward explanation about what statistical testing
really does.
Second, introductory biostatistics courses are often taught in a way that
does not cater for life science students, who may lack a strong mathematical
background.
When I started teaching at Central Queensland University, I thought
there had to be a better way of introducing essential concepts of biostatistics
and experimental design. It had to start from first principles and develop an
understanding that could be applied to all statistical tests. It had to demys-
tify what these tests actually did and explain them with a minimum of
formulae and terminology. It had to relate statistical concepts to exper-
imental design. And, finally, it had to build a strong understanding to help
the student progress to more complex material. I tried this approach with

xiii
C:/ITOOLS/WMS/CUP-NEW/2647705/WORKINGFOLDER/MCKI/9781107005518PRF.3D xiv [13–14] 27.8.2011 10:50AM

xiv Preface

my undergraduate classes and the response from a lot of students, including

some postgraduates who sat in on the course, was ‘Hey Steve, you should
write an introductory stats book!’
Ward Cooper suggested I submit a proposal for this sort of book to
Cambridge University Press. The reviewers of the initial proposal and the
subsequent manuscript made most appropriate suggestions for improve-
ment. Ruth McKillup read, commented on and reread several drafts, pro-
vided constant encouragement and tolerated my absent mindedness. My
students, especially Steve Dunbar, Kevin Strychar and Glenn Druery
encouraged me to start writing and my friends and colleagues, especially
Dearne Mayer and Sandy Dalton, encouraged me to finish.
I sincerely thank the users and reviewers of the first edition for their
comments and encouragement. Katrina Halliday from CUP suggested an
expanded second edition. Ruth McKillup remained a tolerant, pragmatic,
constructive and encouraging critic, despite having read many drafts many
times. The students in my 2010 undergraduate statistics class, especially
Deborah Fisher, Michael Rose and Tara Monks, gave feedback on many of
the explanations developed for this edition; their company and cynical
humour were a refreshing antidote.
C:/ITOOLS/WMS/CUP-NEW/2647576/WORKINGFOLDER/MCKI/9781107005518C01.3D 1 [1–6] 27.8.2011 11:17AM

1 Introduction

1.1 Why do life scientists need to know about experimental

design and statistics?

If you work on living things, it is usually impossible to get data from every
individual of the group or species in question. Imagine trying to measure
the length of every anchovy in the Pacific Ocean, the haemoglobin count
of every adult in the USA, the diameter of every pine tree in a plantation
of 200 000 or the individual protein content of 10 000 prawns in a large
aquaculture pond.
The total number of individuals of a particular species present in a
defined area is often called the population. But because a researcher usually
cannot measure every individual in the population (unless they are studying
the few remaining members of an endangered species), they have to work
with a very carefully selected subset containing several individuals (often
called sampling units or experimental units) that they hope is a represen-
tative sample from which they can infer the characteristics of the popula-
tion. You can also think of a population as the total number of artificial
sampling units possible (e.g. the total number of 1m2 plots that would cover
a whole coral reef) and your sample being the subset (e.g. 20 plots) you have
to work upon.
The best way to get a representative sample is usually to choose a number
of individuals from the population at random – without bias, with every
possible individual (or sampling unit) within the population having an
equal chance of being selected.
The unavoidable problem with this approach is that there are often
great differences among sampling units from the same population.
Think of the people you have seen today – unless you have met some
identical twins (or triplets etc.), no two would have been the same. This

1
C:/ITOOLS/WMS/CUP-NEW/2647576/WORKINGFOLDER/MCKI/9781107005518C01.3D 2 [1–6] 27.8.2011 11:17AM

2 Introduction

Figure 1.1 Even a random sample may not necessarily be a good

representative of the population from which it has been taken. Two samples,
each of five individuals, have been taken at random from the same population.
By chance sample 1 contains a group of relatively large fish, while those in
sample 2 are relatively small.

can even apply to species made up of similar looking individuals (like

flies or cockroaches or snails) and causes problems when you work with
samples.
First, even a random sample may not be a good representative of the
population from which it has been taken (Figure 1.1). For example, you
may choose students for an exercise experiment who are, by chance, far less
(or far more) physically fit than the student population of the college they
represent. A batch of seed chosen at random may not represent the varia-
bility present in all seed of that species, and a sample of mosquitoes from a
particular place may have very different insecticide resistance than the same
species occurring elsewhere.
C:/ITOOLS/WMS/CUP-NEW/2647576/WORKINGFOLDER/MCKI/9781107005518C01.3D 3 [1–6] 27.8.2011 11:17AM

1.1 Why do life scientists need to know about design and statistics? 3

Figure 1.2 Samples selected at random from very different populations may
not necessarily be different. Simply by chance the samples from populations
1 and 2 are similar, so you might mistakenly conclude the two populations are
also similar.

Therefore, if you take a random sample from each of two similar pop-
ulations, the samples may be different to each other simply by chance. On
the basis of your samples, you might mistakenly conclude that the two
populations are very different. You need some way of knowing if a differ-
ence between samples is one you would expect by chance or whether the
populations they have been taken from really do seem to be different.
Second, even if two populations are very different, randomly chosen
samples from each may be similar and give the misleading impression
the populations are also similar (Figure 1.2).