Quantifying Item Invariance for the Selection of the Least Biased Assessment

• In Event: 14.046 - Poster Session 1
  In Poster Session: 14.046-6 - Contemporary Models and Their Applications: A Poster Session

Abstract

The decision to select a measure that contains the least amount of differential item functioning (DIF) is paramount for practitioners to ensure they are making accurate decisions. There exists no user-friendly indices that capture the amount of DIF to allow comparisons across assessments or subtests from different tests. We introduce four effect sizes measures based on well-known DIF methods for such use. A simulation study evaluates the four indices in terms of quantifying item invariance across measures. Results show that these measures show promise in assisting practitioners in making such a decision. Increasing accessibility and use of such indices to inform instrument selection should add positively to the process of test score use and better decisions about individuals.

Authors

• William Holmes Finch, Ball State University

• Brian F. French, Washington State University

• Maria E. Hernandez Finch, Ball State University