Logistic Regression Using SAS

Paul D. Allison

Description

If you are a researcher or student with experience in multiple linear regression and want to learn about logistic regression, Paul Allison's Logistic Regression Using SAS: Theory and Application, Second Edition, is for you! Informal and nontechnical, this book both explains the theory behind logistic regression, and looks at all the practical details involved in its implementation using SAS. Several real-world examples are included in full detail. This book also explains the differences and similarities among the many generalizations of the logistic regression model. The following topics are covered: binary logistic regression, logit analysis of contingency tables, multinomial logit analysis, ordered logit analysis, discrete-choice analysis, and Poisson regression. Other highlights include discussions on how to use the GENMOD procedure to do loglinear analysis and GEE estimation for longitudinal binary data. Only basic knowledge of the SAS DATA step is assumed. The second edition describes many new features of PROC LOGISTIC, including conditional logistic regression, exact logistic regression, generalized logit models, ROC curves, the ODDSRATIO statement (for analyzing interactions), and the EFFECTPLOT statement (for graphing nonlinear effects). Also new is coverage of PROC SURVEYLOGISTIC (for complex samples), PROC GLIMMIX (for generalized linear mixed models), PROC QLIM (for selection models and heterogeneous logit models), and PROC MDC (for advanced discrete choice models).
This book is part of the SAS Press program.

Sujets

• Contingency table
• Statistical model

Informations

• Publié par: SAS Institute
• Date de parution: 30 mars 2012
• Nombre de lectures: 0
• EAN13: 9781607649953
• Langue: English
• Poids de l'ouvrage: 21 Mo

Extrait

Logistic
Regression
Using SAS ®
Theory and Application
Second Edition
Paul D. Allison
THE POWER TO KNOW ®
The correct bibliographic citation for this manual is as follows: Allison, Paul D. 2012. Logistic Regression Using SAS ® : Theory and Application, Second Edition. Cary, NC: SAS Institute Inc.
Logistic Regression Using SAS ® : Theory and Application, Second Edition
Copyright © 2012, SAS Institute Inc., Cary, NC, USA
ISBN 978-1-60764-995-3 (electronic book) ISBN 978-1-59994-641-2
All rights reserved. Produced in the United States of America.
For a hard-copy book: No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, or otherwise, without the prior written permission of the publisher, SAS Institute Inc.
For a Web download or e-book: Your use of this publication shall be governed by the terms established by the vendor at the time you acquire this publication.
The scanning, uploading, and distribution of this book via the Internet or any other means without the permission of the publisher is illegal and punishable by law. Please purchase only authorized electronic editions and do not participate in or encourage electronic piracy of copyrighted materials. Your support of others' rights is appreciated.
U.S. Government Restricted Rights Notice: Use, duplication, or disclosure of this software and related documentation by the U.S. government is subject to the Agreement with SAS Institute and the restrictions set forth in FAR 52.227-19, Commercial Computer Software-Restricted Rights (June 1987).
SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513-2414
1st printing, April 2012
SAS ® Publishing provides a complete selection of books and electronic products to help customers use SAS software to its fullest potential. For more information about our e-books, e-learning products, CDs, and hard-copy books, visit the SAS Publishing Web site at support.sas.com/publishing or call 1-800-727-3228.
SAS ® and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are registered trademarks or trademarks of their respective companies.


508870bp08072012
Contents
Preface
Chapter 1 Introduction
1.1   What This Book Is About
1.2   What This Book Is Not About
1.3   What You Need to Know
1.4   Computing
1.5   References
Chapter 2 Binary Logistic Regression with PROC LOGISTIC: Basics
2.1   Introduction
2.2   Dichotomous Dependent Variables: Example
2.3   Problems with Ordinary Linear Regression
2.4   Odds and Odds Ratios
2.5   The Logistic Regression Model
2.6   Estimation of the Logistic Model: General Principles
2.7   Maximum Likelihood Estimation with PROC LOGISTIC
2.8   Interpreting Coefficients
2.9   CLASS Variables
2.10 Multiplicative Terms in the MODEL Statement
Chapter 3 Binary Logistic Regression: Details and Options
3.1   Introduction
3.2   Confidence Intervals
3.3   Details of Maximum Likelihood Estimation
3.4   Convergence Problems
3.5   Multicollinearity
3.6   Goodness-of-Fit Statistics
3.7   Statistics Measuring Predictive Power
3.8   ROC Curves
3.9   Predicted Values, Residuals, and Influence Statistics
3.10 Latent Variables and Standardized Coefficients
3.11 Probit and Complementary Log-Log Models
3.12 Unobserved Heterogeneity
3.13 Sampling on the Dependent Variable
3.14 Plotting Effects of Predictor Variables
Chapter 4 Logit Analysis of Contingency Tables
4.1   Introduction
4.2   A Logit Model for a 2 × 2 Table
4.3   A Three-Way Table
4.4   A Four-Way Table
4.5   A Four-Way Table with Ordinal Explanatory Variables
4.6   Overdispersion
Chapter 5 Multinomial Logit Analysis
5.1   Introduction
5.2   Example
5.3   A Model for Three Categories
5.4   Estimation with PROC LOGISTIC
5.5   Estimation with a Binary Logit Procedure
5.6   General Form of the Model
5.7   Contingency Table Analysis
5.8   Problems of Interpretation
Chapter 6 Logistic Regression for Ordered Categories
6.1   Introduction
6.2   Cumulative Logit Model: Example
6.3   Cumulative Logit Model: Explanation
6.4   Cumulative Logit Model: Practical Considerations
6.5   Cumulative Logit Model: Contingency Tables
6.6   Adjacent Categories Model
6.7   Continuation Ratio Model
Chapter 7 Discrete Choice Analysis
7.1   Introduction
7.2   Chocolate Example
7.3   Model and Estimation
7.4   Travel Example
7.5   Other Applications
7.6   Ranked Data
7.7   More Advanced Models with PROC MDC
Chapter 8 Logit Analysis of Longitudinal and Other Clustered Data
8.1   Introduction
8.2   Longitudinal Example
8.3   Robust Standard Errors
8.4   GEE Estimation with PROC GENMOD
8.5   Mixed Models with GLIMMIX
8.6   Fixed-Effects with Conditional Logistic Regression
8.7   Postdoctoral Training Example
8.8   Matching
8.9   Comparison of Methods
8.10 A Hybrid Method
Chapter 9 Regression for Count Data
9.1   Introduction
9.2   The Poisson Regression Model
9.3   Scientific Productivity Example
9.4   Overdispersion
9.5   Negative Binomial Regression
9.6   Adjustment for Varying Time Spans
9.7   Zero-Inflated Models
Chapter 10 Loglinear Analysis of Contingency Tables
10.1   Introduction
10.2   A Loglinear Model for a 2 × 2 Table
10.3   Loglinear Models for a Four-Way Table
10.4   Fitting the Adjacent Categories Model as a Loglinear Model
10.5   Loglinear Models for Square, Ordered Tables
10.6   Marginal Tables
10.7   The Problem of Zeros
10.8   GENMOD versus CATMOD
References
Index
Preface
It's about time! The first edition of Logistic Regression Using SAS had become so outdated that I was embarrassed to see people still using it. In the 13 years since the initial publication, there have been an enormous number of changes and enhancements to the SAS procedures for doing logistic regression and related methods. In fact, I think that the rate of change has accelerated in recent years.
So, as of April 2012, this book is up to date with the latest syntax, features, and options in SAS 9.3. All the output displays use the HTML format, which is now the default. Perhaps the biggest change is that PROC LOGISTIC plays an even larger role than before. That's because it now has a CLASS statement, and it can now estimate the multinomial logit model. Here are some chapter-by-chapter details.
Chapter 2 , “Binary Logistic Regression with PROC LOGISTIC: Basics.” In the first edition, PROC GENMOD got major attention in this chapter. Now it's barely mentioned. The reason for this change is that PROC LOGISTIC has a CLASS statement, so GENMOD isn't needed until Chapter 8 . The CLASS statement in PROC LOGISTIC works a little differently than in other PROCs, so I spend some time explaining the differences. In addition, I now show how to get robust standard errors when estimating the linear probability model with PROC REG. And I demonstrate how to estimate marginal effects for the logistic model with PROC QLIM.
Chapter 3 , “Binary Logistic Regression: Details and Options.” There are lots of changes and additions in this chapter. As in Chapter 2 , GENMOD is out, LOGISTIC is in. For dealing with small samples and separation, I explain two major new features of LOGISTIC: exact logistic regression and penalized likelihood. I show how to use the new ODDSRATIO statement to get interpretable odds ratios when a model contains interactions. And I demonstrate the new EFFECTPLOT statement for getting useful graphs of predicted values, especially when there are non-linearities and interactions. Other additions include: graphs of odds ratios, ROC curves with tests for model differences, a new R-squared measure, new diagnostic graphs, and the use of PROC QLIM to model group comparisons with heteroscedastic errors.
Chapter 4 , “Logit Analysis of Contingency Tables.” The major change in this chapter is that PROC LOGISTIC i