Assessment and Development of Differential Item Functioning

Abstract

Assessment and Development of Differential Item Functioning Ya-Hui Su Department of Psychology, National Chung Cheng University, Taiwan ?: This talk will start with an introduction to differential item functioning (DIF) assessment. DIF assessment refers to an analysis that identifies items for which examinees with the same overall ability level have different probabilities of answering correctly because they are from different sub-populations. In order to eliminate bias against sub-populations, DIF detection has become an essential routine item analysis procedure in test development to maintain test fairness and validity. There are two major kinds of DIF detection: a parametric approach that assumes a specific item response model and a nonparametric approach that does not. Of these, parametric methods suffer from model misspecification because even a small amount of misfit may result in serious Type I error inflation. In contrast, nonparametric methods have the advantage that they do not require specific forms of item response functions, large same sizes, or intensive computation. This talk will include my previous researches with parametric and onparametric DIF detection methods for dichotomous and polytomous items. Since it is important to locate a set of DIF-free items to serve as anchors in DIF analysis and scale linking, the DIF-free-then-DIF strategy is proposed and found to achieve better DIF detection in terms of Type I error and power. Finally, a real data of math basic achievement test from the BCTEST center will be analyzed and discussed. Some future research avenues will be discussed as well.