“…Different item selection methods have different degrees of resistance to test security breaches. For any specific item selection method, such as the weighted deviation method , the a-stratified method (Chang & Ying, 1999;Yi & Chang, 2003), and constrained CAT with shadow tests (van der Linden, 2000), a comparison of test security between the single-pool approach and the multiple-pool approach can be conducted by a simulation study. Below, we discuss a simulation study using a popular CAT item selection method, the maximum item information method (Lord, 1980) with the Sympson-Hetter item exposure control (Hetter & Sympson, 1997;Sympson & Hetter, 1985), to find out whether a two-pool design outperforms a single-pool design in terms of test security.…”