Categorical Data Analysis Using The SAS® System, 2nd Edition

15.1. Introduction

The weighted least squares methodology described in Chapter 12 is a useful approach to the analysis of repeated binary and ordered categorical outcome variables. However, it can only accommodate categorical explanatory variables and can’t easily handle missing values. In addition, the WLS methodology requires sufficient sample size for the marginal response functions at each time in each subpopulation to have an approximately multivariate normal distribution. This requirement can be very restrictive.

In recent years, researchers have begun to use a new method for the repeated measurements analysis of categorical outcomes. The generalized estimating equation (GEE) approach (Liang and Zeger 1986) is an extension of generalized ...