Advanced Log-Linear Models Using SAS (Zelterman)

Contents

Preface

Acknowledgments

1 Discrete Distributions

1.1 Introduction

1.2 The Binomial Distribution

1.3 The Poisson Distribution

1.4 The Multinomial Distribution

1.5 Negative Binomial and Negative Multinomial Distributions

2 Basic Log-Linear Models and the GENMOD Procedure

2.1 Introduction

2.2 Log-Linear Models for a 2x2 Table

2.3 Log-Linear Models in Higher Dimensions

2.4 Residuals for Log-Linear Models

2.5 Tests of Statistical Significance

2.6 The Likelihood Function

3 Ordered Categorical Variables

3.1 Introduction

3.2 Log-Linear Models with One Ordered Category

3.3 Two Cross-Classified Ordered Categories

4 Non-Rectangular Tables

4.1 Introduction

4.2 Independence in a Triangular Table

4.3 Interactions in a Circular Table

4.4 Bradley-Terry Model for Pairwise Comparisons

5 Poisson Regression

5.1 Introduction

5.2 Poisson Regression for Mortality Data

5.3 Poisson Regression with Overdispersion

6 Finite Population Size Estimation

6.1 Introduction

6.2 A Small Example

6.3 A Larger Number of Lists

7 Truncated Poisson Regression

7.1 Introduction

7.2 Mathematical Background

7.3 Truncated Poisson Models with Covariates

7.4 An Example with Overdispersion

7.5 Diagnostics and Options

8 The Hypergeometric Distribution

8.1 Introduction

8.2 Derivation of the Distribution

8.3 Extended Hypergoemetric Distribution

8.4 Hypergeometric Regression

8.5 Comparing Several 2x2 Tables

9 Sample Size Estimation and Power for Log-Linear Models

9.1 Introduction

9.2 Background Theory

9.3 Power for a 2x2 Table

9.4 Sample Size for an Interaction

9.5 Power for a Known Sample Size

A The Output Delivery System

B Programming Statements for Generalized Linear Models

C Additional Readings

References

Index