Empirical Selection of Anchors for Tests of Differential Item Functioning  

Oleh:

Woods, Carol M.

Jenis: Article from Journal - ilmiah internasional
Dalam koleksi: Applied Psychological Measurement vol. 33 no. 1 (Jan. 2009), page 42-57.
Topik: differential item functioning; item bias; measurement invariance; item response theory; likelihood ratio
Fulltext: 42.pdf (139.68KB)

Isi artikel

Differential item functioning (DIF) occurs when items on a test or questionnaire have different measurement properties for one group of people versus another, irrespective of groupmean differences on the construct. Methods for testing DIF require matching members of different groups on an estimate of the construct. Preferably, the estimate is based on a subset of group-invariant items called designated anchors. In this research, a quick and easy strategy for empirically selecting designated anchors is proposed and evaluated in simulations. Although the proposed rank-based approach is applicable to any method for DIF testing, this article focuses on likelihood-ratio (LR) comparisons between nested two-group item response models. The rank-based strategy frequently identified a groupinvariant designated anchor set that produced more accurate LR test results than those using all other items as anchors. Group-invariant anchors were more difficult to identify as the percentage of differentially functioning items increased. Advice for practitioners is offered.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali