Testing proportionality in the proportional odds model fitted with GEE

Generalized estimating equations (GEE) methodology as proposed by Liang and Zeger has received widespread use in the analysis of correlated binary dat a. Miller et al, and Lipsitz er al. extended GEE to correlated nominal and ordinal categorical data; in particular, they used GEE for fitting McCullag h's proportional odds model. In this paper, we consider robust (that is, em pirically corrected) and model-based versions of both a score test and a Wa ld test for assessing the assumption of proportional odds in the proportion al odds model fitted with GEE. The Wald test is based on fitting separate m ultiple logistic regression models for each dichotomization of the response variable, whereas the score test requires fitting just the proportional od ds model. We evaluate the proposed tests in small to moderate samples by si mulating data from a series of simple models. We illustrate the use of the tests on three data sets from medical studies. (C) 1999 John Wiley & Sons, Ltd.