ABSTRACT
In Chapter 5, we discuss two categories of models. The first category consists of latent class models. Typical of latent class models is that they classify people on the basis of their pattern of scores on a set of variables, in this case, a set of items from a test or questionnaire. The latent classes are characterized by response probabilities for each combination of an item score and an item, and these probabilities vary between classes. It is assumed that latent classes identify meaningful types of people on the basis of their score patterns, for example, by solution strategy for cognitive items or by personality types. Usually, latent classes are un-ordered, but latent class models for ordered latent classes are also available. We discuss the latent class model and several goodness-of-fit methods. The second category of models contains cognitive diagnostic models, which can be seen both as latent class models and item response models. Cognitive diagnostic models also identify classes but now on the basis of the presence and absence of an array of latent attributes that the researcher hypothesizes to be necessary for solving particular problems or that are typical of particular personality types or pathologies. We discuss several models and focus on the goodness-of-fit methods for these models.
