Interindividual differences in intraindividual change in categorical variables

In this article, we proceed from the assumption that constancy and change in development are not necessarily universal. This deviation from the general assumption of universal developmental patterns is embedded in the theory of person-oriented research. In addition, we propose that constancy and change can reflect local associations instead of associations that cover the entire range of admissible scores. Models of Configural Frequency Analysis are proposed to explore and test hypotheses concerning person-specific local associations in repeated observation data. Three models are considered for lagged data. These models differ in the reasons that are assumed as causes for local associations. The first model reflects variable associations of any kind. The second model reflects case-specific variable associations. The third reflects differences between cases. In an example, data from a study on the development of alcoholics are used. The data in this example reflect case-specific associations in the development of drinking behavior over a span of two years. In the discussion, the person- and the variable-oriented elements of longitudinal research are addressed. In addition, assumptions concerning the independence of longitudinal data are made explicit.