Journal of Pedagogical Research (Aug 2021)

Differential item functioning analysis of a high stake test in terms of statistical regions of Turkey

  • Yeşim Özer Özkan,
  • Meltem Acar Güvendir

DOI
https://doi.org/10.33902/JPR.2021371303
Journal volume & issue
Vol. 5, no. 3
pp. 122 – 134

Abstract

Read online

Large scale assessment is conducted at different class levels for various purposes such as identifying student success in education, observing the impacts of educational reforms on student achievement, assessment, selection, and placement. It is expected that these tests and their items are used in education do not display different traits with regard to the responses of individuals at the same ability level but from different groups. The purpose of this study was to identify whether the test includes Differential Item Functioning (DIF), and if it does, to identify whether the items are biased or not. The Rasch model was performed using the Winsteps package software to determine if the items contain DIF. DIF was identified in eight items during the analysis.  Expert opinion was sought to determine if the difference between the regions in item is due to DIF or item effect. Based on the feedback from the experts, no bias was observed in the items with regard to regions.

Keywords