An Introduction to Missing Data in the Context of Differential Item Functioning.

Abstract

This article introduces practitioners and researchers to the topic of missing data in the context of differential item functioning (DIF), reviews the current literature on the issue, discusses implications of the review, and offers suggestions for future research. A total of nine studies were reviewed. All of these studies determined what effect particular missing data techniques would have on the results of certain DIF detection procedures under various conditions. The most important finding of this review involved the use of zero imputation as a missing data technique. The review shows that zero imputation can lead to inflated Type I errors, especially in cases where the examinees ability level has not been taken into consideration.