This paper adopted theDIF-free-then-DIF (DFTD) strategy to form scale purification with therank-based (RB) method, denoted as RB-S method, together with the computersoftware IRTLRDIF for assessing differential item functioning (DIF). Two simulationstudies were conducted to evaluate the performance of the RB-S method and thefeasibility of the DFTD strategy for DIF assessment of dichotomous items. The resultsshow that, first, the RB-S method outperforms the RN method on selection ofanchors (4 items). Second, the RB-S method can avoid to yield an inflated TypeI error rate. Third, the power rate from the RB-S method was higher than thatfrom the RB method. Fourth, the RB-S method outperforms the traditionalall-other-items (AOI) and AOI with scale purification (AOI-S) methods incontrolling Type I error rates when the average singed area (ASA) is larger. Butthe AOI-S method yields a power rate higher than the RB-S method when the ASAis not large. Finally, the RB-S method for the selection of DIF-free items andthe DFTD strategy are supported.
Comment:
This paper introduced the commonlyused methods for establishing the matching variable, and described the relatedresearches for DIF assessment. It is very helpful for me to have a general understandingin DIF topic and help me to find the related literatures for future study.