In contrast, computerized adaptive testing (CAT), grounded in modern test theory, uses an item selection algorithm so that each subsequent item is selected based on the respondent's previous responses, allowing the item order to vary across individuals. In theory, CAT is more efficient and, within the capacity of the item bank, should achieve a given measurement precision level with the same or a shorter test length than progressive testing. However, CAT requires an item response theory (IRT) model and computerized administration, which leads to additional complexity.
This project aims to: 1) provide a transparent toolkit that guides the procedure of setting appropriate stopping rules within the existing progressive testing structure, and 2) explore the psychometric promises and practical concerns of the transition from progressive tests to CATs in applied settings.
The PhD project is part of the SEALS project.
