Log-Linear Models

420 Pages·1990·9.915 MB·English

by  Ronald Christensen (auth.)

Springer Texts in Statistics
Advisors:
Stephen Fienberg Ingram Olkin
Springer Texts in Statistics
Alfred Elements of Statistics
for the Life and Social Seiences
Biom Probability and Statistics:
Theory and Applications
Chow and Teicher Probability Theory: Independence,
Interchangeability, Martingales.
Second Edition
Christensen Plane Answers to Complex Questions:
The Theory of Linear Models
Christensen Linear Models for Multivariate, Time
Series, and Spatial Data
Christensen Log-Linear Models
du Toit, Steyn and Graphical Exploratory Data Analysis
Strumpf
Finkeistein and Levin Statistics for Lawyers
Kalbfleisch Probability and Statistical Inference:
Volume 1: Probability.
Second Edition
Kalbfleisch Probability and Statistical Inference:
Volume 2: Statistical Inference.
Second Edition
Keyfitz Applied Mathematical Demography.
Second Edition
Kiefer Introduction to Statistical Inference
Kokoska and Nevison Statistical Tables and Formulae
Madansky Prescriptions for Working Statisticians
McPherson Statistics in Scientific Investigation:
Its Basis, Application, and Interpretation
Nguyen and Rogers Fundamentals of Mathematical Statistics:
Volume 1: Probability for Statistics
(continued afterindex)
Ronald Christensen
Log-Linear Models
Springer Science+Business Media, LLC
Ronald Christensen
Department of Mathematics and Statistics
University of New Mexico
Albuquerque, NM 87131
USA
Editorial Board
Stephen Fienberg Irrgram Olkin
Department of Statistics Department of Statistics
Carnegie-Mellon U niversity Stanford University
Pittsburgh, PA 15213 Stanford, CA 94305
USA USA
Mathematics Subject Classification: 62H17
Library of Congress Cataloging-in-Publication Data
Christensen, Ronald R.
Log-linear models / Ronald Christensen.
p. cm. - (Springer texts in statistics)
Includes bibliographical references and index.
1. Log-linear models. I. title. II. Series.
QA278.C49 1990 90-42812
519.5'35-dc20 CIP
Printed on acid-free paper.
© 1990 by Springer Science+B usiness Media N ew York
Originally published by Springer-Verlag New York Inc. in 1990.
Softcoverreprint ofthe hardcover1st edition 1990
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, 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 for
bidden.
The use of general descriptive names, trade names, trademarks, etc., in this pub
lication, even if the former are not especially identified, is not to be taken as a
sign that such names, as understood by the Trade Marks and Merchandise Marks
Act, may accordingly be used freely by anyone.
Photocomposed copy prepared from the author's lffiTRXfile.
9 8 7 6 5 4 3 2 1
ISBN 978-1-4757-4113-1 ISBN 978-1-4757-4111-7 (eBook)
DOI 10.1007/978-1-4757-4111-7
To Sharon and Fletch
Preface
This book examines log-linear models for contingency tables. Logistic re
gression and logistic discrimination are treated as special cases and gener
alized linear models (in the GLIM sense) are also discussed. The book is
designed to fill a niche between basic introductory books such as Fienberg
(1980) and Everitt (1977) and advanced books such as Bishop, Fienberg,
and Holland (1975), Haberman (1974), and Santner and Duffy (1989). lt
is primarily directed at advanced Masters degree students in Statistics but
it can be used at both higher and lower levels. The primary theme of the
book is using previous knowledge of analysis of variance and regression to
motivate and explicate the use of log-linear models. Of course, both the
analogies and the distinctions between the different methods must be kept
in mind.
The book is written at several levels. A basic introductory course would
take material from Chapters I, II (deemphasizing Section II.4), III, Sec
tions IV.1 through IV.5 (eliminating the material on graphical models),
Section IV.lü, Chapter VII, and Chapter IX. The advanced modeling ma
terial at the end of Sections VII.1, VII.2, and possibly the material in
Section IX.2 should be deleted in a basic introductory course. For Mas
ters degree students in Statistics, all the material in Chapters I through
V, VII, IX, and X should be accessible. For an applied Ph.D. course or for
advanced Masters students, the material in Chapters VI and VIII can be
incorporated. Chapter VI recapitulates material from the first five chapters
using matrix notation. Chapter VIII recapitulates Chapter VII. This ma
terial is necessary (a) to get standard errors of estimates in anything other
than the saturated model, (b) to explain the Newton-Raphson (iteratively
reweighted least squares) algorithm, and ( c) to discuss the weighted least
squares approach of Grizzle, Starmer, and Koch (1969). I also think that
the more general approach used in these chapters provides a deeper un
derstanding of the subject. Most of the material in Chapters VI and VIII
requires no more sophistication than matrix arithmetic and being able to
understand the definition of a column space. All of the material should be
accessible to people who have had a course in linear models. Throughout
the book, Chapter XV of Christensen (1987) is referenced for technical de
tails. For completeness, and to allow the book to be used in nonapplied
Ph.D. courses, Chapter XV has been reprinted in this volume under the
same title, Chapter XV.
The prerequisites differ for the various courses described above. At a
minimum, readers should have had a traditional course in statistical meth
ods. To understand the vast majority of the book, courses in regression,
analysis of variance, and basic statistical theory are recommended. To fully
appreciate the book, it would help to already know linear model theory.
lt is difficult for me to understand but many of my acquaintance view
viii Preface
me a.s quite opinionated. While I admit that I have not tried to keep my
opinions to myself, I have tried to clearly acknowledge them a.s my opinions.
There are many people I would like to thank in connection with this
work. My family, Sharon and Fletch, were supportive throughout. Jackie
Damrau did an exceptional job of typing the first draft. The folks at BMDP
provided me with copies of 4F, LR, and 9R. MINITAB provided me with
Versions 6.1 and 6.2. Dick Lund gave me a copy of MSUSTAT. All of the
computations were performed with this software or GLIM. Several people
made valuable comments on the manuscript; these include Rahman Azari,
Larry Blackwood, Ron Schrader, and Elizabeth Slate. Joe Hill introduced
me to statistical applications of graph theory and convinced me of their
importance and elegance. He also commented on part of the book. My
editors, Steve Fienberg and Ingram Olkin, were, as always, very helpful.
Like many people, I originally learned about log-linear models from Steve's
book. Two people deserve special mention for how much they contributed
to this effort. I would not be the author of this book were it not for the
amount of support provided in its development by Ed Bedrick and Wes
Johnson. Wes provided much of the data used in the examples. I suppose
that I should also thank the legislature of the state of Montana. It was
their penury, while I worked at MontanaState University, that motivated
me to begin the project in the spring of 1987. If you don't like the book,
blame them!
Ronald Christensen
Albuquerque, New Mexico
April 5, 1990
(Happy Birthday Dad)
Contents
Preface vii
I Introduction 1
I.1 Conditional Probability and Independence . 1
I.2 Random Variablesand Expectations 11
I.3 The Binomial Distribution . . 12
I.4 The Multinomial Distribution 14
I.5 The Poisson Distribution 18
I.6 Exercises 20
II Two-Dimensional Tables 23
Il.1 Two Independent Binomials . . . . . . 23
Il.2 Testing Independence in a 2 x 2 Table . 29
Il.3 I x J Tables . . . . . . . . . . . . . . . 33
II.4 Maximum Likelihood Theory forTwo-Dimensional Tables 41
Il.5 Log-Linear ModelsforTwo-Dimensional Tables 46
Il.6 Exercises . . . . . . . . . . . . . . . . . . . . . . 53
III Three-Dimensional Tables 61
III.1 Simpson's Paradox and the Need for Higher Dimensional
Tables . . . . . . . . . . . . . . . . . . 62
III.2 Independence and Odds Ratio Models
for Three-Dimensional Tables ........... . 63
III.3 Iterative Computation of Estimates . . . . . . . . 78
III.4 Log-Linear Models for Three-Dimensional Tables . 81
III.5 Product-Multinomial and Other Sampling Plans 91
III.6 Exercises ..................... . 95
IV Higher Dimensional Tables 99
IV.1 Model Interpretations and Graphical Models 100
IV.2 Collapsing Tables ............ . 113
IV.3 Stepwise Procedures for Model Selection 115
IV.4 Initial Models for Selection Methods ... 118
x Contents
IV.5 Example of Stepwise Methods . . . . . 128
IV.6 Aitkin's Method of Backward Selection 137
IV. 7 Model Selection Among Decomposable
and Graphical Models . . . . . . . . . . 143
IV.8 Model Selection Criteria . . . . . . . . 149
IV.9 Residuals and Influential Observations. 154
IV.lü Drawing Conclusions 161
IV.ll Exercises . . . . . . . . . . . . . . . . . 162
V Models for Factors with Quantitative Levels 167
V.1 Models for Two-Factor Tables 168
V.2 Higher Dimensional Tables 175
V.3 Unknown Factor Scores 177
V.4 Exercises . . . . . . . . . . 183
VI The Matrix Approach to Log-Linear Models 185
Vl.1 Maximum Likelihood Theory for Multinomial Sampling 189
Vl.2 Asymptotic Results . . . . . . . 193
Vl.3 Product-Multinomial Sampling . . . . . . . . . . . . . 210
VI.4 Inference for Model Parameters . . . . . . . . . . . . 213
Vl.5 Methods for Finding Maximum Likelihood Estimates 216
Vl.6 Regression Analysis of Categorical Data . 218
VI. 7 Residual Analysis and Outliers 224
Vl.8 Exercises . . . . . . . . . . . . . . . . . . 230
VII Response Factors and Logistic Discrimination 233
VII.1 Logit Models . . . . . . . . . . . . . . . . . . . . 236
VII.2 Logit Modelsfora Multinomial Response Factor 246
VII.3 Logistic Regression . . . . . . . . . . . 255
VI1.4 Logistic Discrimination and Allocation 269
VII.5 Recursive Causal Models 279
VII.6 Exercises . . . . . . . . . . . . . . . . . 294
VIII The Matrix Approach to Logit Models 300
VIII.1 Estimation and Testing for Logistic Models 300
VIII.2 Asymptotic Results . . . . . . . . . . . . . 308
VIII.3 Model Selection Criteria for Log;istic Regression. 316
VIII.4 Likelihood Equations and Newton-Raphson 318
VIII.5 Weighted Least Squares for Logit Models . . . . 320
VIII.6 Multinomial Response Models . . . . . . . . . . 322
VIII. 7 Discrimination, Allocation, and Retrospective Data 323
VIII.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . 330