ABSTRACT
All of the treatment to this point has assumed that the true responses are independent, or at least uncorrelated. Here we turn our attention to mixed or longitudinal models where multiple observations are associated with a main “unit,” leading to correlation among observations. A unit may be an individual person, plant, etc. or a cluster, such as a hospital or a school. For convenience, the unit is referred to generically as a subject. There is a huge literature on treating such models, which comes under various names, with connections to hierarchical, structural, multilevel and latent variable models. A few representative texts are Davidian and Giltian (1995), Demidenko (2004), Diggle et al. (1994), Fitzmaurice et al. (2004), McCulloch and Searle (2001) and Skrondal and Rabe-Hesketh (2004).
