Speaker
Description
ABSTRACT
Computer-Based Examinations (CBEs) have increasingly adopted randomized question pools to enhance test security and efficiency. While this approach minimizes predictability, it raises fairness concerns as different students may encounter test versions with varying levels of difficulty. This study applied the Two-Parameter Logistic (2PL) model of Item Response Theory (IRT) to evaluate fairness in randomized question pools using a simulated dataset. The analysis focused on two key parameters: item difficulty, which reflects the level of ability required to answer an item correctly, and item discrimination, which measures how well an item differentiates between students of differing ability levels. Ability estimates of test takers were further derived to assess overall performance across the simulated cohort. The findings show that the assessment exhibits a balanced range of item difficulties, with some items being relatively more challenging, and that most items demonstrate acceptable to strong discrimination parameters. These results suggest that while the test was generally fair and reliable, variations in item characteristics highlight the importance of careful calibration in constructing randomized question pools to ensure equity in CBEs.