Conditional inference about generalized linear mixed models

Abstract

We propose a method of inference for generalized linear mixed models\n(GLMM) that in many ways resembles the method of least squares. We also show\nthat adequate inference about GLMM can be made based on the conditional\nlikelihood on a subset of the random effects. One of the important features of\nour methods is that they rely on weak distributional assumptions about the\nrandom effects. The methods proposed are also computationally feasible.\nAsymptotic behavior of the estimates is investigated. In particular,\nconsistency is proved under reasonable conditions.