PERFORMANCE OF GENERALIZED ESTIMATING EQUATIONS IN PRACTICAL SITUATIONS

Moment methods for analyzing repeated binary responses have been propo sed by Liang and Zeger (1986, Biometrika 73, 13-22), and extended by P rentice (1988, Biometrics 44, 1033-1048). In their generalized estimat ing equations (GEE), both Liang and Zeger (1986) and Prentice (1988) e stimate the parameters associated with the expected value of an indivi dual's vector of binary responses as well as the correlations between pairs of binary responses. In this paper, we discuss one-step estimato rs, i.e., estimators obtained from one step of the generalized estimat ing equations, and compare their performance to that of the fully iter ated estimators in small samples. In simulations, we find the performa nce of the one-step estimator to be qualitatively similar to that of t he fully iterated estimator. When the sample size is small and the ass ociation between binary responses is high, we recommend using the one- step estimator to circumvent convergence problems associated with the fully iterated GEE algorithm. Furthermore, we find the GEE methods to be more efficient than ordinary logistic regression with variance corr ection for estimating the effect of a time-varying covariate.