Item response models for assessing change in dichotomous items

The paper discusses three methodological approaches to measuring change in developmental research using item response models. To illustrate application, an empirical example using data from the Vienna Developmental Study is presented: 94 kindergarten students were presented twice with a set of k dichotomous items measuring general developmental status. First, we analysed individual changes using a residual-analytic approach based on measures from the dichotomous Rasch Model. Second, in a confirmatory approach, we investigated group-specific changes in a quasiexperimental design using the Hybrid Linear Logistic Model with Relaxed Assumptions. Third, we presented an exploratory method to search for subgroups of subjects that differ systematically in developmental patterns using Mixed Rasch Models. The discussion focuses on the virtues and limitations of the three approaches.