ABSTRACT
Exposed test items are a major threat to the validity of computer-based testing. Interestingly and importantly, the findings about the item exposure detection statistics and how they functioned are applicable to all forms of computer-based testing from linear or linear-on-the-fly to multistage, to fully adaptive tests. A different approach for addressing the problem is to focus attention on the generation and investigation of item statistics that can reveal whether test items have become known to examinees prior to seeing the items in the test they were administered. As for the item exposure detection statistics, research showed a strong advantage to the two item response theory-based statistics. The results from the study were revealing for all of the variables studied: ability shifts, item exposure models, item exposure detection statistics, and item statistics.
