Extended generalized estimating equations for clustered data

Typically, analysis of data consisting of multiple observations on a cluste r is complicated by within-cluster correlation. Estimating equations for ge neralized linear modeling of clustered data have recently received much att ention. This article proposes an extension to the generalized estimating eq uation method proposed by Liang and Zeger, which treats within-cluster corr elations as nuisance parameters. Using ideas from extended quasi-likelihood , estimating equations for regression and association parameters are provid ed simultaneously. The resulting estimators are proven to be asymptotically normal and consistent under certain conditions. The consistency of regress ion estimators allows incorrect modeling of the correlation among repeated responses. The method is illustrated with an analysis of data from a develo pmental toxicity study.