Analysis of change between two points in time with a categorical measurement scale is a very well established procedure of analysis in biological and social research literature, but the application of this procedure to analyze developmental change is extremely rare in psychological research. With categorical variables the most common reports use elementary descriptive percentages and tests based on ² distribution. In this paper we provide a survey of loglinear, association and latent variable models that can be potentially applied to test several hypotheses and specific effects on developmental change. We illustrate how to use these models with data taken from a longitudinal research on the use of morphological spelling strategies of Nunes et al. (1997). Many of models discussed require specialized software not implemented in widespread statistical packages. We also provide input examples to fit these models with an experimental version of EM program (Vermunt, 1998).