ABSTRACT
In previous chapters, we have mostly focused on likelihood methods for estimation and inference. The likelihood methods are based on distributional assumptions for the data. For example, given the random effects, in GLMMs we assume that the responses in the models follow distributions in the exponential family, and in LME and NLME models we assume that the responses in the models follow normal distributions. If the distributional assumptions hold, the likelihood methods are very attractive since the MLEs are asymptotically most efficient and asymptotically normal under some regularity conditions (see Chapter 12).
