Consistent maximum likelihood estimation using subsets with applications to multivariate mixed models (Q2196200)

The authors extend to a general class of multivariate generalized linear mixed models (MGLMMs) the subset argument for the consistency of maximum likelihood estimators (MLEs). The subset argument relates the likelihood for the full data to that of a subset consisting of independent and identically distributed random variables. This argument is particularly useful in the framework of mixed models, where random effect designs may induce dependence among the response variables. Indeed, the authors discuss with full details two examples where the theoretical results of the paper allow us to check the consistency of the MLE, while it was not possible to prove such a consistency with other techniques from previous literature.