Adjusting for confounding by cluster using generalized linear mixed models

We show how to use generalized linear mixed models to adjust for confounding by cluster of the effect of a within-cluster covariate. We derive estimators for both a cluster-specific causal effect and a population-averaged causal effect.