User-Friendly and Effective Item Purification for Differential Item Functioning Analysis in IRTPRO

• In Event: Roundtable Session 6
  In Roundtable Session: Differential Item Functioning: Methods and Applications

Abstract

IRTPRO can calibrate models and test for differential item functioning (DIF) using the Wald test, an improved Lord’s χ2 test (Lord, 1980). If anchors are not known, then the Wald-2 test can be used (Cai, Thissen, & du Toit, 2015). Wald-2 benefits from anchor purification, but it is an involved and sometimes overcomplicated process for the end user. Therefore, the purpose of this study is to develop a user-friendly Web-based utility for full purification and to compare three purification methods (none, partial, and full) in terms of their performance in DIF analysis in a simulation study. Full purification was found to improve Type 1 error without sacrificing power, an improvement from partial purification made possible through the newly-developed utilities.

Authors

• David R J Fikis, Georgia State University

• T. C. Oshima, Georgia State University