Discrete Multivariate Analysis Yvonne M. M. Bishop, Stephen E. Fienberg, and Paul W. Holland with the collaboration of Richard J. Light and Frederick Mosteller Discrete Multivariate Analysis Theory and Practice 4 i Springer Yvonne M. Bishop Washington, DC 20015-2956
[email protected] Stephen E. Fienberg Department of statistics Carnegie-Mellon University Pittsburgh, PA 15213
[email protected] Paul W. Holland Educational Testing Service Princeton, NJ 08541
[email protected] ISBN 978-0-387-72805-6 Library of Congress Control Number: 2007928365 © 2007 Springer Science+Business Media, LLC, This Springer edition is a reprint of the 1975 edition published by MIT Press. All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper. Printed in the United States of America. 987654321 springer.com Preface The analysis of discrete multivariate data, especially in the form of cross-classifications, has occupied a prominent place in the statistical literature since the days of Karl Pearson and Sir R. A. Fisher. Although Maurice Bartlett's pioneering paper on testing for absence of second-order interaction in 2 x 2 x 2 tables was published in 1935, the widespread development and use of methods for the analysis of multidimensional cross-classified data had to await the general availability of high-speed computers. As a result, in the last ten years statistical journals, as well as those in the biological, social, and medical sciences, have devoted increasing space to papers dealing with the analysis of discrete multivariate data. Many statisticians have contributed to this progress, as a glance at the reference list will quickly reveal. We point, especially, to the sustained and outstanding contributions of Joseph Berkson, M. W. Birch, I. J. Good, Leo A. Goodman, James E. Grizzle, Marvin Kastenbaum, Gary G. Koch, Solomon Kullback, H. O. Lancaster, Nathan Mantel, and R. L. Plackett. The one person most responsible for our interest in and continued work on the analysis of cross-classified data is Frederick Mosteller. It is not an overstatement to say that without his encouragement and support in all phases of our effort, this book would not exist. Our interest in the analysis of cross-classified data goes back to 1964 and the problems which arose during and af