ABSTRACT
Latent Markov (LM) models are latent variable models tailored to the analysis of longitudinal data, typically used when the response variables are categorical. These models make use of time-specific latent variables, which are assumed to be discrete, and then represent a rather sophisticated class of latent variable models. In fact, even in its simplest formulation, an LM model may be seen as a generalization of the latent class (LC) model, which is a well-known model for classifying a sample of subjects on the basis of a set of categorical responses. Another possible interpretation of an LM model is as an extension of the Markov chain (MC) model allowing for measurement errors.
