Identification of Differential Item Functioning Using Item Response Theory and the Likelihood-Based Model Comparison Approach

Abstract

IRT and the likelihood-based model comparison approach comprise a powerful tool for DIF detection that can aid in the development, refinement, and evaluation of measures for use in ethnically diverse populations.

Testing parent dyad interchangeability in the parent proxy-report of PedsQL™ 4.0: a differential item functioning analysis

“…This process is an iterative backward anchor-class procedure, since the method includes the majority of items in the anchor at the beginning and then successively excludes the items from the anchor. After finding a common set of anchor items, in the second stage each of the candidate items is reassessed for DIF with the use of purified anchor items (Interested readers can refer to Orlando [30] for more detailed discussions of this DIF detection procedure). This study used IRTPRO2.1 to detect DIF across parent dyads.…”

Evaluation of MIMIC-Model Methods for DIF Testing With Comparison to Two-Group Analysis

“…However, Type I error is inflated when the anchor set is contaminated by DIF (Stark et al, 2006;Wang & Yeh, 2003) so it is better to designate anchors. Various strategies for empirically identifying anchors have been applied or suggested (Bolt, Hare, Vitale, & Newman, 2004;Edelen, Thissen, Teresi, Kleinman, & Ocepek-Welikson, 2006;Kim & Cohen, 1995;Rensvold & Cheung, 2001;Stark et al, 2006;Wang, 2004;Woods, in press). Many of them are iterative purification procedures wherein the analysis is repeated several times with apparently differentially functioning items removed one by one.…”

Development of the PROMIS(R) Positive Emotional and Sensory Expectancies of Smoking Item Banks

“…These evaluations were conducted using the full daily (N = 4,201) and nondaily (N = 1,183) smoker samples with IRTPRO . DIF was evaluated for significance according to gender, race/ethnicity (White, Black, Hispanic), and age (18-30, 31-50, 51+) using established procedures (Edelen, Thissen, Teresi, Kleinman, & Ocepek-Welikson, 2006;Hansen et al, 2014;Orlando & Marshall, 2002). Items with significant DIF were further evaluated for "impact" by considering the weighted area between the expected score curves ("wABC") and the expected difference in expected a posteriori score ("dEAP") indices described in more detail in Hansen et al Items with wABC values greater than 0.30 were screened for potential removal by evaluating graphical illustrations of the subgroups' expected scores curves, along with the values of the wABC and dEAP indices.…”