An Entropy-Based Measure for Person Fit in Item Response Theory

• In Event: Roundtable Session 6
  In Roundtable Session: Item Response Theory: Studies on Estimation Methods and Model Fit

Abstract

In this study we expand entropy from a measure of model-fit in LCA to a measure of person-fit in IRT models. We also introduce the idea of weighting entropy by the amount of misfit. We (1) provide a generalized conceptual overview of data-model fit and person-fit indices, (2) introduce a new variant of entropy and a ratio statistic, Emf, entropic misfit adapted to account for misfit of persons in dichotomous IRT models via conceptual, and applied examples, and (3) consider extensions into other IRT models. Our scenarios demonstrate several cases where Emf is superior to lzin differentiating misfit. In the final paper we expand comparisons with Emf to include multiple forms of person-fit (e.g. group-based and likelihood methods)

Authors

• William R. Dardick, The George Washington University

• Brandi A. Weiss, The George Washington University