A Primer on REGRESSION ARTIFACTS DONALD T. CAMPBELL DAVID A. KENNY Foreword by Charles S. Reichardt THE GUILFORD PRESS New York London © 1999 The Guilford Press A Division of Guilford Publications, Inc. 72 Spring Street, New York, NY 10012 http://www. guilford.com All rights reserved No part of this book may be reproduced, translated, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written permission from the Publisher. Printed in the United States of America This book is printed on acid-free paper. Last digit is printnumber: 9 8 7 6 5 4 3 2 1 Library of Congress Cataloging-in-Publication Data Campbell, Donald Thomas, 1916-1996 A primer on regression artifacts / Donald T. Campbell and David A. Kenny. p. cm. — (Methodology in the social sciences) Includes bibliographical references and index. ISBN 1-57230-482-0 1. Regression analysis. 2. Social sciences—Statistical methods. I. Kenny, David A., 1946- . IL. Title. III. Series. HA31.3.C35 1999 519.5'36—DC21 99-23003 CIP A PRIMER ON REGRESSION ARTIFACTS METHODOLOGY IN THE SOCIAL SCIENCES David A. Kenny, Series Editor PRINCIPLES AND PRACTICE OF STRUCTURAL EQUATION MODELING Rex B. Kline SPECTRAL ANALYSIS OF TIME-SERIES DATA Rebecca M. Warner A PRIMER ON REGRESSION ARTIFACTS Donald T. Campbell and David A. Kenny To those who worked with Donald T. Campbell; may they continue to make science timeless! About the Authors Donald T. Campbell, PhD, before his death in 1996 was Universi- ty Professor of Social Relations, Psychology, and Education at Lehigh University. He had previously taught at Ohio State Univer- sity, the University of Chicago, Northwestern University, and Syra- cuse University. He was a member of the National Academy of Sci- ences and a President of the American Psychological Association. He was the recipient of nine honorary doctorates. David A. Kenny, PhD, is Professor of Social Psychology at the University of Connecticut. He has been a visiting professor at Ox- ford University and Arizona State University. He was a Fellow at the Center for Advanced Study in the Behavioral Sciences. Vii Foreword Regression toward the mean is as inevitable as death and taxes. Academic performance, emotional well-being, medical diagnosis, investment return, athletic feats, motion picture sales, and any other variables you can think of all exhibit regression toward the mean. But even more remarkable than the ubiquitousness of regres- sion toward the mean is how commonly the phenomenon is misun- derstood, usually with undesirable consequences. Social scientists incorrectly estimate the effects of ameliorative interventions, sports writers misguidedly attribute poor performance to jinxes, and snake-oil peddlers earn a healthy living all because our intuition fails when trying to comprehend regression toward the mean. Even intellectual prowess is not an antidote to this shortcoming. Sir Francis Galton is one of those rare geniuses whose name is still renowned 90 years after his death. This is, in part, because Galton was the first to recognize and provide a label for “regression toward mediocrity.” Galton even demonstrated convincingly that individ- ual height regresses to the mean across generations. But he got it wrong when he tried to explain how regression toward the mean operates. If someone of Galton’s immense intellectual abilities cannot understand regression toward the mean even when looking it square in the eye, how can we mere mortals expect to figure it out? The answer is, in more ways than one, within your grasp. The ele- gant and concise volume you hold in your hands provides a simple and comprehensive explanation. This is not to say that under- standing the volume’s content will always be easy. The mystery of regression toward the mean is not going to be revealed without ef- fort on your part. You must be willing to grapple at times with no- ix x Foreword tions that appear to violate common sense. You must be prepared to stop on occasion to puzzle over apparent paradoxes. But if you are willing to extend yourself just a little, your efforts will be well re- warded. If you are a novice in the topic, you will become an expert by reading A Primer on Regression Artifacts. If you are already an ex- pert, you will learn things you will be surprised you did not already know. In either case, you will find that the authors meet you more than halfway; they guide your inquiry with ample encouragement, engaging illustrations, and good humor. As a result, your reading will prove to be both eye-opening and enjoyable. Of course, those in the know would expect no less from Don Campbell and Dave Kenny. They are two of the preeminent social science methodologists of our generation. It is hard to imagine a duo that is more capable of making comprehensible a challenging methodological topic. Under ordinary circumstances, no more need be said about ei- ther of the authors because of their stellar reputations. But circum- stances are not ordinary and a few more words must be said about Campbell in particular. There are few, if any, scholars who have had as sustained and profound an influence on research methods in the social sciences as Campbell has had. For example, Campbell’s separate volumes on quasi-experimentation with Julian Stanley and with Tom Cook are the equivalent of the old and the new tes- taments of research design. Campbell’s article with Don Fiske on the multitrait-multimethod matrix is the single most widely cited article in the Psychological Bulletin in the last 50 years. Based on such work, Campbell has received virtually every honor that is available to a research psychologist. If the Nobel prize were award- ed in psychology, he would likely have won that as well. As a re- sult, Campbell’s name, like Galton’s, will be revered long after his death. Sadly, that time period has already begun, for Don Campbell died in the spring of 1996. Campbell and Kenny planned the present volume together, and they wrote much of it together before Don’s untimely death. But Kenny had to finish it alone. The result is a tribute to Kenny’s dedication both to Don and to the topic of regression toward the mean. Kenny has done both of them proud. While Campbell was passionate about all of his work, there was no topic he embraced with more fervor than regression toward the mean. He never tired of talking about it, and he very much wanted others to come to understand it as he did. As a result, Don is now smiling. He has seen the publication of the last, and one of the most beloved, of his works.