The goal of this study is to evaluate the performance of six commonly used item exposure control methods—Randomesque, Sympson-Hetter (SH), Unconditional and Conditional Multinomial Method (UMM, CMM), Fade-Away (FA), and Progressive-Restricted (PR), alongside a no-exposure control condition. The evaluation was conducted through a Monte Carlo item simulation process within a fixed-length Computerized Adaptive Testing (CAT) framework. For this purpose, an item bank of 160 abstract reasoning items was employed, calibrated using the two-parameter logistic model (2PL) and equated using the Haebara Conversion method. Item difficulty ranged between −4.38 and 2.71, and discrimination parameters ranged from 0.51 to 3.29. All methods performed comparably in measurement precision as indicated by various statistical indices, including the bias statistic, mean absolute error (MAE), and root mean square error (RMSE). However, the PR method outperformed the others by fully utilizing the item bank while achieving the lowest exposure rates among administered items without compromising precision. Overall, PR emerged as the most efficient method, offering the best balance between item bank security and measurement precision.
Karagianni et al. (Wed,) studied this question.