An IRT modeling of change over time for repeated measures item response data using a random weights linear logistic test model approach

摘要

In this study, repeated measures item response data from an English language proficiency (ELP) test administered over 2 years were analyzed by item response theory-based approaches to model change over time. First, change over time was modeled by an item side change approach where change was parameterized at the item level. Second, more importantly, this study introduced and demonstrated an application of an interaction model for change, which overcomes a limitation of the item side change approach by allowing interaction effects between persons and item side change parameters. The results showed that student changes in the ELP data used in this study should be modeled for the three sub-content areas and that modeling interaction for change was necessary.