Introduction to Robust Estimation and Hypothesis

Testing 5th Edition Rand R. Wilcox - eBook PDF

download

https://ebooksecure.com/download/introduction-to-robust-
estimation-and-hypothesis-testing-ebook-pdf/

Download more ebook from https://ebooksecure.com

We believe these products will be a great fit for you. Click
the link to download now, or visit ebooksecure.com
to discover even more!

Introduction to Robust Estimation and Hypothesis

Testing 4th Edition Rand Wilcox - eBook PDF

https://ebooksecure.com/download/introduction-to-robust-
estimation-and-hypothesis-testing-ebook-pdf-2/

Applications of Hypothesis Testing for Environmental

Science 1st Edition - eBook PDF

https://ebooksecure.com/download/applications-of-hypothesis-
testing-for-environmental-science-ebook-pdf/

Psychological Testing and Assessment: An Introduction

to Tests and Measurement 10th Edition Ronald Jay Cohen
- eBook PDF

https://ebooksecure.com/download/psychological-testing-and-
assessment-an-introduction-to-tests-and-measurement-ebook-pdf/

Introduction to Mechatronics and Measurement Systems

5th Edition

http://ebooksecure.com/product/introduction-to-mechatronics-and-
measurement-systems-5th-edition/
(eBook PDF) Introduction to Learning and Behavior 5th
Edition

http://ebooksecure.com/product/ebook-pdf-introduction-to-
learning-and-behavior-5th-edition/

(Original PDF) Introduction to Human Communication by

Susan R. Beauchamp

http://ebooksecure.com/product/original-pdf-introduction-to-
human-communication-by-susan-r-beauchamp/

Population: An Introduction to Concepts and Issues 13th

Edition John R. Weeks - eBook PDF

https://ebooksecure.com/download/population-an-introduction-to-
concepts-and-issues-ebook-pdf/

Introduction to Hospitality Management 4th Edition John

R Walker - eBook PDF

https://ebooksecure.com/download/introduction-to-hospitality-
management-ebook-pdf/

(eBook PDF) Load Testing of Bridges: Proof Load Testing

and the Future of Load Testing (Structures and
Infrastructures Book 13)

http://ebooksecure.com/product/ebook-pdf-load-testing-of-bridges-
proof-load-testing-and-the-future-of-load-testing-structures-and-
infrastructures-book-13/
Introduction to Robust Estimation and
Hypothesis Testing
 Introduction to Robust
Estimation and Hypothesis
Testing
Fifth Edition

Rand R. Wilcox
Professor of Psychology
University of Southern California
Los Angeles, California, United States
Academic Press is an imprint of Elsevier
125 London Wall, London EC2Y 5AS, United Kingdom
525 B Street, Suite 1650, San Diego, CA 92101, United States
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
Copyright © 2022 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, 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 permissions 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).
Notices
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.
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 products liability, negligence or otherwise, or from any use or operation of
any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

British Library Cataloguing-in-Publication Data

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

ISBN: 978-0-12-820098-8

For information on all Academic Press publications

visit our website at https://www.elsevier.com/books-and-journals

Publisher: Katey Birtcher

Editorial Project Manager: Sara Valentino
Production Project Manager: Beula Christopher
Designer: Bridget Hoette
Typeset by VTeX
Printed in the United States of America
Last digit is the print number: 9 8 7 6 5 4 3 2 1
à mon seul œuf
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Problems With Assuming Normality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 The Influence Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 The Central Limit Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Is the ANOVA F Robust? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.7 More Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.8 R Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.9 Some Data Management Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.9.1 Eliminating Missing Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.10 Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Chapter 2: A Foundation for Robust Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1 Basic Tools for Judging Robustness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.1 Qualitative Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.1.2 Infinitesimal Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.3 Quantitative Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Some Measures of Location and Their Influence Function . . . . . . . . . . . . . . . . . . 31
2.2.1 Quantiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.2 The Winsorized Mean. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.3 The Trimmed Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.2.4 M-Measures of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.2.5 R-Measures of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.3 Measures of Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.4 Scale-Equivariant M-Measures of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5 Winsorized Expected Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Chapter 3: Estimating Measures of Location and Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1 A Bootstrap Estimate of a Standard Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.1 R Function bootse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

vii
Contents

3.2 Density Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.2.1 Silverman’s Rule of Thumb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.2 Rosenblatt’s Shifted Histogram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.3 The Expected Frequency Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.4 An Adaptive Kernel Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.5 R Functions skerd, kerSORT, kerden, kdplot, rdplot, akerd, and
splot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.3 The Sample Median and Trimmed Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.3.1 R Functions mean, tmean, median, and lloc . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.3.2 Estimating the Standard Error of the Trimmed Mean . . . . . . . . . . . . . . . . 60
3.3.3 Estimating the Standard Error of the Sample Winsorized Mean . . . . . 65
3.3.4 R Functions winmean, winvar, winsd, trimse, and winse . . . . . . . . . . . . 65
3.3.5 Estimating the Standard Error of the Sample Median . . . . . . . . . . . . . . . . 66
3.3.6 R Function msmedse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.4 The Finite Sample Breakdown Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.5 Estimating Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.5.1 Estimating the Standard Error of the Sample Quantile . . . . . . . . . . . . . . . 68
3.5.2 R Function qse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.5.3 The Maritz–Jarrett Estimate of the Standard Error of x̂q . . . . . . . . . . . . . 70
3.5.4 R Function mjse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.5.5 The Harrell–Davis Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.5.6 R Functions qest and hd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.5.7 A Bootstrap Estimate of the Standard Error of θ̂q . . . . . . . . . . . . . . . . . . . . 73
3.5.8 R Function hdseb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.6 An M-Estimator of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.6.1 R Function mad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.6.2 Computing an M-Estimator of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.6.3 R Function mest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.6.4 Estimating the Standard Error of the M-Estimator . . . . . . . . . . . . . . . . . . . 82
3.6.5 R Function mestse. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.6.6 A Bootstrap Estimate of the Standard Error of μ̂m . . . . . . . . . . . . . . . . . . . 84
3.6.7 R Function mestseb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.7 One-Step M-Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.7.1 R Function onestep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.8 W-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.8.1 Tau Measure of Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.8.2 R Function tauloc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.8.3 Zuo’s Weighted Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.9 The Hodges–Lehmann Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.10 Skipped Estimators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.10.1 R Functions mom, zwe, and bmean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

viii
 Contents

3.11 Some Comparisons of the Location Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

3.12 More Measures of Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.12.1 The Biweight Midvariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.12.2 R Function bivar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.12.3 The Percentage Bend Midvariance and tau Measure of Variation . . . 96
3.12.4 R Functions pbvar and tauvar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.12.5 The Interquartile Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.12.6 R Functions idealf, idrange, idealfIQR, and quantile. . . . . . . . . . . . . . . . . 99
3.13 Some Outlier Detection Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.13.1 Rules Based on Means and Variances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.13.2 A Method Based on the Interquartile Range . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.13.3 Carling’s Modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.13.4 A MAD-Median Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.13.5 R Functions outbox, out, and boxplot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.13.6 R Functions adjboxout and adjbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.14 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Chapter 4: Inferences in the One-Sample Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.1 Problems When Working With Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.1.1 P-Values and Testing for Equality: Hypothesis Testing Versus
Decision Making . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.2 The g-and-h Distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.2.1 R Functions ghdist, rmul, rngh, rmul.MAR, ghtrim, and gskew . . . . . 115
4.3 Inferences About the Trimmed, Winsorized Means . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.3.1 Comments on Effect Size and Non-Normal Distributions . . . . . . . . . . . 120
4.3.2 R Functions trimci, winci, D.akp.effect.ci, and depQSci. . . . . . . . . . . . . 121
4.4 Basic Bootstrap Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.4.1 The Percentile Bootstrap Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.4.2 R Functions onesampb and hdpb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.4.3 Bootstrap-t Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.4.4 Bootstrap Methods When Using a Trimmed Mean. . . . . . . . . . . . . . . . . . . 126
4.4.5 Singh’s Modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.4.6 R Functions trimpb and trimcibt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.5 Inferences About M-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.5.1 R Functions mestci and momci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.6 Confidence Intervals for Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.6.1 Beware of Tied Values When Making Inferences About Quantiles . 137
4.6.2 A Modification of the Distribution-Free Method for the Median . . . . 138
4.6.3 R Functions qmjci, hdci, sint, sintv2, qci, qcipb, and qint . . . . . . . . . . . 140
4.7 Empirical Likelihood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.8 Inferences About the Probability of Success . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

ix
Contents

4.8.1 R Functions binom.conf.pv and cat.dat.ci. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

4.9 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
4.10 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Chapter 5: Comparing Two Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.1 The Shift Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.1.1 The Kolmogorov–Smirnov Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5.1.2 R Functions ks, kssig, kswsig, and kstiesig . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.1.3 Confidence Bands for the Shift Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.1.4 R Functions sband and wband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.1.5 Confidence Band for Specified Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.1.6 R Functions shifthd, qcomhd, qcomhdMC, and q2gci . . . . . . . . . . . . . . . 168
5.1.7 R Functions g2plot and g5plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.2 Student’s T Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.3 Comparing Medians and Other Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
5.3.1 R Functions yuen and msmed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
5.3.2 A Bootstrap-t Method for Comparing Trimmed Means . . . . . . . . . . . . . . 179
5.3.3 R Functions yuenbt and yhbt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5.3.4 Measuring Effect Size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
5.3.5 R Functions ESfun, akp.effect.ci, KMS.ci, ees.ci, med.effect, qhat,
and qshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
5.4 Inferences Based on a Percentile Bootstrap Method . . . . . . . . . . . . . . . . . . . . . . . . . 193
5.4.1 Comparing M-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
5.4.2 Comparing Trimmed Means and Medians . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
5.4.3 R Functions trimpb2, pb2gen, medpb2, and M2gbt . . . . . . . . . . . . . . . . . . 197
5.5 Comparing Measures of Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
5.5.1 Comparing Variances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
5.5.2 R Functions comvar2 and varcom.IND.MP. . . . . . . . . . . . . . . . . . . . . . . . . . . 200
5.5.3 Comparing Biweight Midvariances and Deviations From the
Median . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
5.5.4 R Functions b2ci, comvar.locdis, and g5.cen.plot . . . . . . . . . . . . . . . . . . . . 201
5.6 Permutation Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
5.6.1 R Function permg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
5.7 Methods Based on Ranks and the Typical Difference . . . . . . . . . . . . . . . . . . . . . . . . 203
5.7.1 The Cliff and Brunner–Munzel Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
5.7.2 A Quantile Shift Measure of Effect Size Based on the Typical
Difference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
5.7.3 R Functions cidv2, bmp, wmwloc, wmwpb, loc2plot, shiftQSci,
akp.effec.ci, ES.summary, ES.summary.CI, ES.sum.REL.MAG, and
loc.dif.summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
5.8 Comparing Two Independent Binomial and Multinomial Distributions . . . . 216

x
 Contents

5.8.1 R Functions binom2g and bi2CR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219

5.8.2 Comparing Discrete (Multinomial) Distributions . . . . . . . . . . . . . . . . . . . . 220
5.8.3 R Functions binband, splotg5, and cumrelf . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
5.9 Comparing Dependent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
5.9.1 A Shift Function for Dependent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
5.9.2 R Function lband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
5.9.3 Comparing Specified Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
5.9.4 R Functions shiftdhd, Dqcomhd, qdec2ci, Dqdif, and difQpci . . . . . . 228
5.9.5 Comparing Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
5.9.6 R Function yuend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
5.9.7 A Bootstrap-t Method for Marginal Trimmed Means . . . . . . . . . . . . . . . . 234
5.9.8 R Function ydbt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
5.9.9 Inferences About the Typical Difference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
5.9.10 R Functions loc2dif, l2drmci, and dep.dif.fun . . . . . . . . . . . . . . . . . . . . . . . . 235
5.9.11 Percentile Bootstrap: Comparing Medians, M-Estimators, and Other
Measures of Location and Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
5.9.12 R Functions two.dep.pb and bootdpci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
5.9.13 Handling Missing Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
5.9.14 R Functions rm2miss and rmmismcp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
5.9.15 Comparing Variances and Robust Measures of Scale . . . . . . . . . . . . . . . . 244
5.9.16 R Functions comdvar, rmVARcom, and RMcomvar.locdis . . . . . . . . . . 246
5.9.17 The Sign Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
5.9.18 R Function signt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
5.9.19 Effect Size for Dependent Groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
5.9.20 R Function dep.ES.summary.CI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
5.10 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Chapter 6: Some Multivariate Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
6.1 Generalized Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
6.2 Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
6.2.1 Mahalanobis Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
6.2.2 Halfspace Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
6.2.3 Computing Halfspace Depth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
6.2.4 R Functions depth2, depth, fdepth, fdepthv2, and unidepth. . . . . . . . . . 259
6.2.5 Projection Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
6.2.6 R Functions pdis, pdisMC, and pdepth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
6.2.7 More Measures of Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
6.2.8 R Functions zdist, zoudepth prodepth, Bagdist, bagdepth, and
zonoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
6.3 Some Affine Equivariant Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
6.3.1 Minimum Volume Ellipsoid Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265

xi
Contents

6.3.2 The Minimum Covariance Determinant Estimator . . . . . . . . . . . . . . . . . . . 266

6.3.3 S-Estimators and Constrained M-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . 267
6.3.4 R Functions tbs, DETS, and DETMCD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
6.3.5 Donoho–Gasko Generalization of a Trimmed Mean . . . . . . . . . . . . . . . . . 269
6.3.6 R Functions dmean and dcov . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
6.3.7 The Stahel–Donoho W-Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
6.3.8 R Function sdwe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
6.3.9 Median Ball Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
6.3.10 R Function rmba . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
6.3.11 OGK Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
6.3.12 R Function ogk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
6.3.13 An M-Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
6.3.14 R Functions MARest and dmedian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
6.4 Multivariate Outlier Detection Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
6.4.1 The Relplot and Bagplot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
6.4.2 R Functions relplot and Bagplot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
6.4.3 The MVE Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
6.4.4 Methods MCD, DETMCD, and DDC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
6.4.5 R Functions covmve and covmcd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
6.4.6 R Functions out and outogk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
6.4.7 The MGV Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
6.4.8 R Function outmgv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
6.4.9 A Projection Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
6.4.10 R Functions outpro and out3d . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
6.4.11 Outlier Identification in High Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
6.4.12 R Functions outproad and outmgvad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
6.4.13 Methods Designed for Functional Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
6.4.14 R Functions FBplot, Flplot, medcurve, func.out, spag.plot, funloc,
and funlocpb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
6.4.15 Comments on Choosing an Outlier Detection Method . . . . . . . . . . . . . . . 295
6.5 A Skipped Estimator of Location and Scatter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
6.5.1 R Functions smean, mgvmean, L1medcen, spat, mgvcov, skip, and
skipcov . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
6.6 Robust Generalized Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
6.6.1 R Function gvarg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
6.7 Multivariate Location: Inference in the One-Sample Case . . . . . . . . . . . . . . . . . . . 300
6.7.1 Inferences Based on the OP Measure of Location . . . . . . . . . . . . . . . . . . . . 300
6.7.2 Extension of Hotelling’s T 2 to Trimmed Means . . . . . . . . . . . . . . . . . . . . . 302
6.7.3 R Functions smeancrv2 and hotel1.tr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
6.7.4 Inferences Based on the MGV Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
6.7.5 R Function smgvcr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

xii
 Contents

6.8 The Two-Sample Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

6.8.1 Independent Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
6.8.2 Comparing Dependent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
6.8.3 R Functions smean2, mul.loc2g, MUL.ES.sum, Dmul.loc2g,
matsplit, and mat2grp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
6.8.4 Comparing Robust Generalized Variances . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
6.8.5 R Function gvar2g. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
6.8.6 Rank-Based Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
6.8.7 R Functions mulrank, cmanova, and cidMULT . . . . . . . . . . . . . . . . . . . . . . 310
6.9 Multivariate Density Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
6.10 A Two-Sample, Projection-Type Extension of the Wilcoxon–
Mann–Whitney Test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
6.10.1 R Functions mulwmw and mulwmwv2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
6.11 A Relative Depth Analog of the Wilcoxon–Mann–Whitney Test . . . . . . . . . . . 315
6.11.1 R Function mwmw . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
6.12 Comparisons Based on Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
6.12.1 R Functions lsqs3 and depthg2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
6.13 Comparing Dependent Groups Based on All Pairwise Differences . . . . . . . . . 323
6.13.1 R Function dfried. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
6.14 Robust Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
6.14.1 R Functions prcomp and regpca . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
6.14.2 Maronna’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
6.14.3 The SPCA Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
6.14.4 Methods HRVB and MacroPCA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
6.14.5 Method OP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
6.14.6 Method PPCA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
6.14.7 R Functions outpca, robpca, robpcaS, SPCA, Ppca, and
Ppca.summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
6.14.8 Comments on Choosing the Number of Components . . . . . . . . . . . . . . . . 332
6.15 Cluster Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
6.15.1 R Functions Kmeans, kmeans.grp, TKmeans, and TKmeans.grp . . . 337
6.16 Classification Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
6.16.1 Some Issues Related to Error Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
6.16.2 R Functions CLASS.fun, CLASS.bag, class.error.com, and menES 344
6.17 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Chapter 7: One-Way and Higher Designs for Independent Groups . . . . . . . . . . . . . . . . 351
7.1 Trimmed Means and a One-Way Design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
7.1.1 A Welch-Type Procedure and a Robust Measure of Effect Size . . . . . 353
7.1.2 R Functions t1way, t1wayv2, t1way.EXES.ci, KS.ANOVA.ES,
fac2list, t1wayF, and ESprodis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356

xiii
Contents

7.1.3 A Generalization of Box’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360

7.1.4 R Function box1way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
7.1.5 Comparing Medians and Other Quantiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
7.1.6 R Functions med1way and Qanova . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
7.1.7 Bootstrap-t Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
7.1.8 R Functions t1waybt, btrim, and t1waybtsqrk . . . . . . . . . . . . . . . . . . . . . . . . 365
7.2 Two-Way Designs and Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
7.2.1 R Functions t2way and bb.es.main . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
7.2.2 Comparing Medians. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
7.2.3 R Functions med2way and Q2anova . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
7.3 Three-Way Designs, Trimmed Means, and Medians . . . . . . . . . . . . . . . . . . . . . . . . . 375
7.3.1 R Functions t3way, fac2list, and Q3anova . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
7.4 Multiple Comparisons Based on Medians and Other Trimmed Means. . . . . . 380
7.4.1 Basic Methods Based on Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
7.4.2 R Functions lincon, conCON, IND.PAIR.ES, and stepmcp. . . . . . . . . . 383
7.4.3 Multiple Comparisons for Two-Way and Three-Way Designs. . . . . . . 390
7.4.4 R Functions bbmcp, RCmcp, mcp2med, twoway.pool, bbbmcp,
mcp3med, con2way, and con3way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
7.4.5 A Bootstrap-t Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
7.4.6 R Functions linconbt, bbtrim, and bbbtrim . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
7.4.7 Controlling the Familywise Error Rate: Improvements on the
Bonferroni Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
7.4.8 R Functions p.adjust and mcpKadjp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
7.4.9 Percentile Bootstrap Methods for Comparing Medians, Other
Trimmed Means, and Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
7.4.10 R Functions linconpb, bbmcppb, bbbmcppb, medpb, Qmcp,
med2mcp, med3mcp, and q2by2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
7.4.11 Deciding Which Group Has the Largest Measure of Location . . . . . . 405
7.4.12 R Functions anc.best.PV, anc.bestpb, PMD.PCD, RS.LOC.IZ,
best.DO, and bestPB.DO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410
7.4.13 Determining the Order of the Population Trimmed Means . . . . . . . . . . 411
7.4.14 R Function ord.loc.PV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
7.4.15 Judging Sample Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
7.4.16 R Function hochberg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
7.4.17 Measures of Effect Size: Two-Way and Higher Designs . . . . . . . . . . . . . 414
7.4.18 R Functions twowayESM, RCES, interES.2by2, interJK.ESmul, and
IND.INT.DIF.ES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
7.4.19 Comparing Curves (Functional Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
7.4.20 R Functions funyuenpb and Flplot2g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
7.4.21 Comparing Variances and Robust Measures of Scale . . . . . . . . . . . . . . . . 418
7.4.22 R Functions comvar.mcp and robVARcom.mcp . . . . . . . . . . . . . . . . . . . . . . 418

xiv
 Contents

7.5 A Random Effects Model for Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418

7.5.1 A Winsorized Intraclass Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
7.5.2 R Function rananova . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
7.6 Bootstrap Global Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
7.6.1 R Functions b1way, pbadepth, and boot.TM . . . . . . . . . . . . . . . . . . . . . . . . . 426
7.6.2 M-Estimators and Multiple Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
7.6.3 R Functions linconm and pbmcp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
7.6.4 M-Estimators and the Random Effects Model . . . . . . . . . . . . . . . . . . . . . . . . 431
7.6.5 Other Methods for One-Way Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
7.7 M-Measures of Location and a Two-Way Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
7.7.1 R Functions pbad2way and mcp2a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
7.8 Ranked-Based Methods for a One-Way Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
7.8.1 The Rust–Fligner Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
7.8.2 R Function rfanova . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
7.8.3 A Rank-Based Method That Allows Tied Values. . . . . . . . . . . . . . . . . . . . . 438
7.8.4 R Function bdm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
7.8.5 Inferences About a Probabilistic Measure of Effect Size . . . . . . . . . . . . 439
7.8.6 R Functions cidmulv2, wmwaov, and cidM . . . . . . . . . . . . . . . . . . . . . . . . . . 441
7.9 A Rank-Based Method for a Two-Way Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
7.9.1 R Function bdm2way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
7.9.2 The Patel–Hoel, De Neve–Thas, and Related Approaches to
Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
7.9.3 R Functions rimul, inter.TDES, LCES, linplot, and plot.inter . . . . . . . 447
7.10 MANOVA Based on Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
7.10.1 R Functions MULtr.anova, MULAOVp, fac2Mlist, and YYmanova 451
7.10.2 Linear Contrasts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
7.10.3 R Functions linconMpb, linconSpb, YYmcp, and fac2BBMlist . . . . . 455
7.11 Nested Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
7.11.1 R Functions anova.nestA, mcp.nestAP, and anova.nestAP. . . . . . . . . . . 460
7.12 Methods for Binary Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
7.12.1 R Functions lincon.bin and binpair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
7.12.2 Identifying the Group With the Highest Probability of Success . . . . . 462
7.12.3 R Functions bin.best, bin.best.DO, and bin.PMD.PCD . . . . . . . . . . . . . . 463
7.13 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Chapter 8: Comparing Multiple Dependent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
8.1 Comparing Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
8.1.1 Omnibus Test Based on the Trimmed Means of the Marginal
Distributions Plus a Measure of Effect Size . . . . . . . . . . . . . . . . . . . . . . . . . . 467
8.1.2 R Functions rmanova and rmES.pro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468

xv
Contents

8.1.3 Pairwise Comparisons and Linear Contrasts Based on Trimmed

Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
8.1.4 Linear Contrasts Based on the Marginal Random Variables . . . . . . . . . 473
8.1.5 R Functions rmmcp, rmmismcp, trimcimul, wwlin.es,
deplin.ES.summary.CI, and boxdif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
8.1.6 Judging the Sample Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
8.1.7 R Functions stein1.tr and stein2.tr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
8.1.8 Identifying the Group With the Largest Population Measure of
Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
8.1.9 Identifying the Variable With the Smallest Robust Measure of
Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
8.1.10 R Functions comdvar.mcp, ID.sm.varPB, rmbestVAR.DO,
rmanc.best.PV, RM.PMD.PCD, rmanc.best.PB, RMPB.PMD.PCD,
and rmanc.best.DO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
8.2 Bootstrap Methods Based on Marginal Distributions . . . . . . . . . . . . . . . . . . . . . . . . 481
8.2.1 Comparing Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
8.2.2 R Function rmanovab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
8.2.3 Multiple Comparisons Based on Trimmed Means . . . . . . . . . . . . . . . . . . . 482
8.2.4 R Functions pairdepb and bptd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
8.2.5 Percentile Bootstrap Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
8.2.6 R Functions bd1way and ddep. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
8.2.7 Multiple Comparisons Using M-Estimators or Skipped Estimators . 490
8.2.8 R Functions lindm and mcpOV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
8.2.9 Comparing Robust Measures of Scatter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
8.2.10 R Function rmrvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
8.3 Bootstrap Methods Based on Difference Scores and a Measure of Effect
Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
8.3.1 R Functions rmdzero and rmES.dif.pro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
8.3.2 Multiple Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
8.3.3 R Functions rmmcppb, wmcppb, dmedpb, lindepbt, and qdmcpdif . 498
8.3.4 Measuring Effect Size: R Function DEP.PAIR.ES . . . . . . . . . . . . . . . . . . . 500
8.3.5 Comparing Multinomial Cell Probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
8.3.6 R Functions cell.com, cell.com.pv, and best.cell.DO. . . . . . . . . . . . . . . . . 501
8.4 Comments on Which Method to Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502
8.5 Some Rank-Based Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
8.5.1 R Functions apanova and bprm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506
8.6 Between-by-Within and Within-by-Within Designs. . . . . . . . . . . . . . . . . . . . . . . . . . 506
8.6.1 Analyzing a Between-by-Within Design Based on Trimmed Means 507
8.6.2 R Functions bwtrim, bw.es.main, and tsplit. . . . . . . . . . . . . . . . . . . . . . . . . . . 509
8.6.3 Data Management: R Function bw2list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
8.6.4 Bootstrap-t Method for a Between-by-Within Design . . . . . . . . . . . . . . . 512

xvi
 Contents

8.6.5 R Functions bwtrimbt and tsplitbt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513

8.6.6 Percentile Bootstrap Methods for a Between-by-Within Design . . . . 514
8.6.7 R Functions sppba, sppbb, and sppbi. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
8.6.8 Multiple Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
8.6.9 R Functions bwmcp, bwmcppb, bwmcppb.adj, bwamcp, bw.es.A,
bw.es.B, bwbmcp, bw.es.I, bwimcp, bwimcpES, spmcpa, spmcpb,
spmcpbA, and spmcpi. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 520
8.6.10 Within-by-Within Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
8.6.11 R Functions wwtrim, wwtrimbt, wwmcp, wwmcppb, wwmcpbt,
ww.es, wwmed, and dlinplot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
8.6.12 A Rank-Based Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
8.6.13 R Function bwrank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
8.6.14 Rank-Based Multiple Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
8.6.15 R Function bwrmcp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
8.6.16 Multiple Comparisons When Using a Patel–Hoel Approach to
Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
8.6.17 R Function BWPHmcp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
8.7 Three-Way Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
8.7.1 Global Tests Based on Trimmed Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
8.7.2 R Functions bbwtrim, bwwtrim, wwwtrim, bbwtrimbt, bwwtrimbt,
wwwtrimbt, and wwwmed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534
8.7.3 Data Management: R Functions bw2list and bbw2list . . . . . . . . . . . . . . . 535
8.7.4 Multiple Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536
8.7.5 R Functions wwwmcp, bbwmcp, bwwmcp, bbwmcppb, bwwmcppb,
wwwmcppb, and wwwmcppbtr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
8.8 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538
Chapter 9: Correlation and Tests of Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
9.1 Problems With Pearson’s Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
9.1.1 Features of Data That Affect r and T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
9.1.2 Heteroscedasticity and the Classic Test That ρ = 0 . . . . . . . . . . . . . . . . . . 546
9.2 Two Types of Robust Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
9.3 Some Type M Measures of Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
9.3.1 The Percentage Bend Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
9.3.2 A Test of Independence Based on ρpb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
9.3.3 R Function pbcor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
9.3.4 A Test of Zero Correlation Among p Random Variables. . . . . . . . . . . . . 550
9.3.5 R Function pball. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
9.3.6 The Winsorized Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553
9.3.7 R Function wincor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554
9.3.8 The Biweight Midcovariance and Correlation . . . . . . . . . . . . . . . . . . . . . . . . 555

xvii
Contents

9.3.9 R Functions bicov and bicovm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556

9.3.10 Kendall’s tau. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
9.3.11 Spearman’s rho . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
9.3.12 R Functions tau, spear, cor, taureg, COR.ROB, and COR.PAIR. . . . . 558
9.3.13 Heteroscedastic Tests of Zero Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
9.3.14 R Functions corb, corregci, pcorb, pcorhc4, and rhohc4bt . . . . . . . . . . . 562
9.4 Some Type O Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
9.4.1 MVE and MCD Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
9.4.2 Skipped Measures of Correlation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
9.4.3 The OP Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
9.4.4 Inferences Based on Multiple Skipped Correlations . . . . . . . . . . . . . . . . . 565
9.4.5 R Functions scor, scorall, scorci, mscorpb, mscorci, mscorciH,
scorreg, scorregci, and scorregciH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
9.5 A Test of Independence Sensitive to Curvature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
9.5.1 R Functions indt, indtall, and medind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
9.6 Comparing Correlations: Independent Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
9.6.1 Comparing Pearson Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573
9.6.2 Comparing Robust Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573
9.6.3 R Functions twopcor, tworhobt, and twocor . . . . . . . . . . . . . . . . . . . . . . . . . . 574
9.7 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 574
Chapter 10: Robust Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577
10.1 Problems With Ordinary Least Squares . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579
10.1.1 Computing Confidence Intervals Under Heteroscedasticity . . . . . . . . . 581
10.1.2 An Omnibus Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586
10.1.3 R Functions lsfitci, olshc4, hc4test, and hc4wtest . . . . . . . . . . . . . . . . . . . . 587
10.1.4 Comments on Comparing Means via Dummy Coding . . . . . . . . . . . . . . . 590
10.1.5 Salvaging the Homoscedasticity Assumption. . . . . . . . . . . . . . . . . . . . . . . . . 590
10.2 The Theil–Sen Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
10.2.1 R Functions tsreg, tshdreg, correg, regplot, and regp2plot . . . . . . . . . . . 594
10.3 Least Median of Squares. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
10.3.1 R Function lmsreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
10.4 Least Trimmed Squares Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
10.4.1 R Function ltsreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
10.5 Least Trimmed Absolute Value Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
10.5.1 R Function ltareg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
10.6 M-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
10.7 The Hat Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
10.8 Generalized M-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
10.8.1 R Function bmreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
10.9 The Coakley–Hettmansperger and Yohai Estimators. . . . . . . . . . . . . . . . . . . . . . . . . 605

xviii
 Contents

10.9.1 MM-Estimator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607

10.9.2 R Functions chreg and MMreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 608
10.10 Skipped Estimators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
10.10.1 R Functions mgvreg and opreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
10.11 Deepest Regression Line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
10.11.1 R Functions rdepth.orig, Rdepth, mdepreg.orig, and mdepreg. . . . . . . 611
10.12 A Criticism of Methods With a High Breakdown Point. . . . . . . . . . . . . . . . . . . . . . 611
10.13 Some Additional Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
10.13.1 S-Estimators and τ -Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
10.13.2 R Functions snmreg and stsreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
10.13.3 E-Type Skipped Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614
10.13.4 R Functions mbmreg, tstsreg, tssnmreg, and gyreg. . . . . . . . . . . . . . . . . . . 615
10.13.5 Methods Based on Robust Covariances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616
10.13.6 R Functions bireg, winreg, and COVreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
10.13.7 L-Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
10.13.8 L1 and Quantile Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
10.13.9 R Functions qreg, rqfit, and qplotreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 621
10.13.10 Methods Based on Estimates of the Optimal Weights. . . . . . . . . . . . . . . . 622
10.13.11 Projection Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 622
10.13.12 Methods Based on Ranks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623
10.13.13 R Functions Rfit and Rfit.est. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
10.13.14 Empirical Likelihood Type and Distance-Constrained Maximum
Likelihood Estimators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625
10.13.15 Ridge Estimators: Dealing With Multicollinearity . . . . . . . . . . . . . . . . . . . 625
10.13.16 R Functions ols.ridge, rob.ridge, and rob.ridge.liu . . . . . . . . . . . . . . . . . . . 627
10.13.17 Robust Elastic Net and Lasso Estimators: Reducing the Number of
Independent Variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
10.13.18 R Functions lasso.est, lasso.rep, RA.lasso, LAD.lasso, H.lasso,
HQreg, and LTS.EN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
10.14 Comments About Various Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
10.14.1 Contamination Bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
10.15 Outlier Detection Based on a Robust Fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
10.15.1 Detecting Regression Outliers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
10.15.2 R Functions reglev and rmblo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
10.16 Logistic Regression and the General Linear Model . . . . . . . . . . . . . . . . . . . . . . . . . . 640
10.16.1 R Functions glm, logreg, logreg.pred, wlogreg, logreg.plot,
logreg.P.ci, and logistic.lasso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
10.16.2 The General Linear Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
10.16.3 R Function glmrob . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643
10.17 Multivariate Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644
10.17.1 The RADA Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645

xix
Contents

10.17.2 The Least Distance Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646

10.17.3 R Functions MULMreg, mlrreg, and Mreglde . . . . . . . . . . . . . . . . . . . . . . . . 647
10.17.4 Multivariate Least Trimmed Squares Estimator . . . . . . . . . . . . . . . . . . . . . . 648
10.17.5 R Function MULtsreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
10.17.6 Other Robust Estimators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
10.18 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
Chapter 11: More Regression Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
11.1 Inferences About Robust Regression Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
11.1.1 Omnibus Tests for Regression Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
11.1.2 R Functions regtest, ridge.test and ridge.Gtest . . . . . . . . . . . . . . . . . . . . . . . 660
11.1.3 Inferences About Individual Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
11.1.4 R Functions regci, regciMC and wlogregci . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
11.1.5 Methods Based on the Quantile Regression Estimator . . . . . . . . . . . . . . . 665
11.1.6 R Functions rqtest, qregci, and qrchk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 667
11.1.7 Inferences Based on the OP-Estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
11.1.8 R Functions opregpb and opregpbMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670
11.1.9 Hypothesis Testing When Using a Multivariate Regression
Estimator RADA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670
11.1.10 R Function mlrGtest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
11.1.11 Robust ANOVA via Dummy Coding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 672
11.1.12 Confidence Bands for the Typical Value of y, Given x . . . . . . . . . . . . . . . 672
11.1.13 R Functions regYhat, regYci, and regYband . . . . . . . . . . . . . . . . . . . . . . . . . 674
11.1.14 R Function regse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676
11.2 Comparing the Regression Parameters of J ≥ 2 Groups . . . . . . . . . . . . . . . . . . . . . 676
11.2.1 Methods for Comparing Independent Groups . . . . . . . . . . . . . . . . . . . . . . . . 676
11.2.2 R Functions reg2ci, reg1way, reg1wayISO, ancGpar, ols1way,
ols1wayISO, olsJmcp, olsJ2, reg1mcp, and olsWmcp . . . . . . . . . . . . . . . 682
11.2.3 Methods for Comparing Two Dependent Groups . . . . . . . . . . . . . . . . . . . . 687
11.2.4 R Functions DregG, difreg, and DregGOLS . . . . . . . . . . . . . . . . . . . . . . . . . . 689
11.3 Detecting Heteroscedasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
11.3.1 A Quantile Regression Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 690
11.3.2 The Koenker–Bassett Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
11.3.3 R Functions qhomt and khomreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 691
11.4 Curvature and Half-Slope Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 692
11.4.1 R Function hratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
11.5 Curvature and Non-Parametric Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
11.5.1 Smoothers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696
11.5.2 Kernel Estimators and Cleveland’s LOWESS . . . . . . . . . . . . . . . . . . . . . . . . 696
11.5.3 R Functions lplot, lplot.pred, and kerreg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 698
11.5.4 The Running-Interval Smoother . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 700

xx
 Contents

11.5.5 R Functions rplot, runYhat, rplotCI, and rplotCIv2 . . . . . . . . . . . . . . . . . . 705

11.5.6 Smoothers for Estimating Quantiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708
11.5.7 R Function qhdsm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 709
11.5.8 Special Methods for Binary Outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 710
11.5.9 R Functions logSM, logSM2g, logSMpred, rplot.bin, runbin.CI,
rplot.binCI, and multsm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
11.5.10 Smoothing With More Than One Predictor . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
11.5.11 R Functions rplot, runYhat, rplotsm, runpd, and RFreg . . . . . . . . . . . . . . 715
11.5.12 LOESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719
11.5.13 Other Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721
11.5.14 R Functions adrun, adrunl, gamplot, and gamplotINT . . . . . . . . . . . . . . . 723
11.5.15 Detecting and Describing Associations via Quantile Grids . . . . . . . . . . 724
11.5.16 R Functions smgridAB, smgridLC, smgrid, smtest, and smbinAB . . 725
11.6 Checking the Specification of a Regression Model. . . . . . . . . . . . . . . . . . . . . . . . . . . 728
11.6.1 Testing the Hypothesis of a Linear Association . . . . . . . . . . . . . . . . . . . . . . 729
11.6.2 R Function lintest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 730
11.6.3 Testing the Hypothesis of a Generalized Additive Model . . . . . . . . . . . . 731
11.6.4 R Function adtest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732
11.6.5 Inferences About the Components of a Generalized Additive Model 732
11.6.6 R Function adcom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733
11.6.7 Detecting Heteroscedasticity Based on Residuals . . . . . . . . . . . . . . . . . . . . 733
11.6.8 R Function rhom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734
11.7 Regression Interactions and Moderator Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734
11.7.1 R Functions kercon, riplot, runsm2g, ols.plot.inter, olshc4.inter,
reg.plot.inter, and regci.inter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736
11.7.2 Mediation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740
11.7.3 R Functions ZYmediate, regmed2, and regmediate . . . . . . . . . . . . . . . . . . 743
11.8 Comparing Parametric, Additive, and Non-Parametric Fits. . . . . . . . . . . . . . . . . . 743
11.8.1 R Functions reg.vs.rplot, reg.vs.lplot, and logrchk . . . . . . . . . . . . . . . . . . . 744
11.9 Measuring the Strength of an Association Given a Fit to the Data . . . . . . . . . . 745
11.9.1 R Functions RobRsq, qcorp1, and qcor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 748
11.9.2 Comparing Two Independent Groups via the LOWESS Version of
Explanatory Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
11.9.3 R Functions smcorcom and smstrcom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
11.10 Comparing Predictors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 750
11.10.1 Comparing Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751
11.10.2 R Functions TWOpov, TWOpNOV, corCOMmcp, twoDcorR, and
twoDNOV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754
11.10.3 Methods for Characterizing the Relative Importance of Independent
Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756

xxi
Contents

11.10.4
R Functions regpre, regpreCV, corREG.best.DO, and
PcorREG.best.DO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 759
11.10.5 Inferences About Which Predictors Are Best. . . . . . . . . . . . . . . . . . . . . . . . . 762
11.10.6 R Functions regIVcom, regIVcommcp, logIVcom, ts2str, and
sm2strv7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767
11.11 Marginal Longitudinal Data Analysis: Comments on Comparing Groups . . 768
11.11.1 R Functions long2g, longreg, longreg.plot, and xyplot. . . . . . . . . . . . . . . 770
11.12 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 771
Chapter 12: ANCOVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773
12.1 Methods Based on Specific Design Points and a Linear Model . . . . . . . . . . . . . 775
12.1.1 Method S1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
12.1.2 Method S2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776
12.1.3 Linear Contrasts for a One-Way or Higher Design . . . . . . . . . . . . . . . . . . . 779
12.1.4 Dealing With Two or More Covariates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 780
12.1.5 R Functions ancJN, ancJNmp, ancJNmpcp, anclin, reg2plot,
reg2g.p2plot, and ancJN.LC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 781
12.2 Methods When There Is Curvature and a Single Covariate . . . . . . . . . . . . . . . . . . 784
12.2.1 Method Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 784
12.2.2 Method BB: Bootstrap Bagging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
12.2.3 Method UB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 788
12.2.4 Method TAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789
12.2.5 Method G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 790
12.2.6 A Method Based on Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 792
12.2.7 R Functions ancova, anc.ES.sum, anc.grid, anc.grid.bin,
anc.grid.cat, ancovaWMW, ancpb, rplot2g, runmean2g, lplot2g,
ancdifplot, ancboot, ancbbpb, qhdsm2g, ancovaUB, ancovaUB.pv,
ancdet, ancmg1, and ancGLOB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 792
12.3 Dealing With Two or More Covariates When There Is Curvature . . . . . . . . . . . 802
12.3.1 Method MC1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
12.3.2 Method MC2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
12.3.3 Methods MC3 and MC4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
12.3.4 R Functions ancovamp, ancovampG, ancmppb, ancmg, ancov2COV,
ancdes, ancdet2C, ancdetM4, and ancM.COV.ES . . . . . . . . . . . . . . . . . . . . 807
12.4 Some Global Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815
12.4.1 Method TG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815
12.4.2 R Functions ancsm and Qancsm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 818
12.5 Methods for Dependent Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819
12.5.1 Methods Based on a Linear Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819
12.5.2 R Functions Dancts and Dancols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 820

xxii
 Contents

12.5.3
Dealing With Curvature: Methods DY, DUB, and DTAP and
a Method Based on Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821
12.5.4 R Functions Dancova, Dancova.ES.sum, Dancovapb, DancovaUB,
Dancdet, Dancovamp, Danc.grid, and ancDEP.MULC.ES . . . . . . . . . . 822
12.6 Exercises. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 825
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 827
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885

xxiii
Preface

There are many new and improved methods in this 5th edition. The R package written for
this book provides a crude indication of how much has been added. When the 4th edition was
published, it contained a little over 1200 R functions. With this 5th edition, there are now over
1700 R functions.
All of the chapters have been updated. For the first three chapters the changes deal with soft-
ware issues and some technical issues. Major changes begin in Chapter 4. One of the major
additions has to do with measuring effect size. Coverage of this topic has been expanded con-
siderably. For example, Chapter 5 now describes six robust measures of effect size when
comparing two independent groups. Confidence intervals for all six are returned by the R
function ES.summary.CI. Each provides a different perspective on how the groups compare.
When there is little or no difference between the groups, all six tend to give very similar re-
sults. But when the groups differ substantially there is a possibility that measures of effect
size can differ substantially. In effect, multiple measures of effect size can help provide a
deeper and more nuanced understanding of data, as illustrated in Section 5.3.5. Easy-to-use
R functions have been added for measuring effect size when dealing with multiple groups.
There have been advances even at the most basic level. Consider, for example, the goal of
computing a confidence interval for the probability of success when dealing with a binomial
distribution. There are now clearer guidelines on when it is advantageous to use the Schilling–
Doi method versus the Agresti–Coull method. When comparing two independent binomial
distributions, more recent results point to using a method derived by Kulinskaya et al. (2010).
A related issue is computing a confidence interval for the probability of success given the
value of a covariate. It is known that when this is done via a logistic regression model, a slight
departure from this model can yield inaccurate results. A method for dealing with this concern
is covered in Chapter 11.
New functions related to regression have been added that include robust methods for dealing
with multicollinearity and robust analogs of the lasso method. The coverage of classification
(machine learning) methods has been expanded. Included is a function making it a simple
matter to compare the false positive and false negative rates of a collection of techniques.

xxv
Preface

There are new results on comparing measures of association as well as measures of varia-
tion. New techniques related to ANOVA-type methods are covered in Chapters 7 and 8 and
new ANCOVA techniques are covered in Chapter 12. A variety of new plots have been added
as well.

As was the case in previous editions, this book focuses on the practical aspects of modern,
robust statistical methods. The increased accuracy and power of modern methods, versus con-
ventional approaches to ANOVA and regression, is remarkable. Through a combination of
theoretical developments, improved and more flexible statistical methods, and the power of
the computer, it is now possible to address problems with standard methods that seemed in-
surmountable only a few years ago.

Consider classic, routinely taught and used methods for comparing groups based on means. A
basic issue is how well these methods perform when dealing with heteroscedasticity and non-
normal distributions. Based on a vast body of literature published over the last 60 years, the
situation can be briefly summarized as follows. When comparing groups that have identical
distributions, classic methods work well in terms of controlling the Type I error probability.
When distributions differ, they might continue to perform well, but under general conditions
they can yield inaccurate confidence intervals and have relatively poor power. Even a small
departure from a normal distribution can destroy power and yield measures of effect size, sug-
gesting a small effect when in fact for the bulk of the population there is a large effect. The
result is that important differences between groups are often missed, and the magnitude of
the difference is poorly characterized. Put another way, groups probably differ when null hy-
potheses are rejected with standard methods, but in many situations, standard methods are the
least likely to find a difference, and they offer a poor summary of how groups differ and the
magnitude of the difference. A related concern is that the population mean and variance are
not robust, roughly meaning that an arbitrarily small change in a distribution can have an in-
ordinate impact on their values. In particular, under arbitrarily small shifts from normality,
their values can be substantially altered and potentially misleading. For example, a small shift
toward a heavy-tailed distribution, roughly meaning a distribution that tends to generate out-
liers, can inflate the variance, resulting in low power. Thus, even with arbitrarily large sample
sizes, the sample mean and variance might provide an unsatisfactory summary of the data.

When dealing with regression, the situation is even worse. That is, there are even more ways
in which analyses, based on conventional assumptions, can be misleading. The very founda-
tion of standard regression methods, namely estimation via the least squares principle, leads
to practical problems, as do violations of other standard assumptions. For example, if the error
term in the standard linear model has a normal distribution, but is heteroscedastic, the least
squares estimator can be highly inefficient, and the conventional confidence interval for the

xxvi
 Another Random Document on
Scribd Without Any Related Topics
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
back
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about testbank and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebooksecure.com