Comparing Two Multigroup Differential Item Functioning Methods in Standardized Assessments

• In Event: Differential Item Functioning and Measurement Invariance

Abstract

In the emerging literature about multi-group DIF methods, two multi-group DIF detection methods were found outstanding for they are straightforward in detecting both uniform and nonuniform DIF. These two methods are the two-step improved Wald test and the generalized logistic regression procedure. No prior research had demonstrated the relative performance of these two methods when they were compared with each other. The present study evaluated these two methods by assessing Type-I and Type-II error rates of the two methods across 80 simulated conditions. Our results suggested the improved Wald test is preferable when (1) detecting nonuniform DIF with two or three groups, or (2) detecting uniform DIF when the proportion of DIF items is moderate (21%) to zero.

Authors

• Dandan Chen, Pearson

• Jinming Zhang, University of Illinois at Urbana-Champaign