An Examination of Sensitivity to Measurement Error of Rasch Residual-Based Fit Statistics

• In Event: 66.077 - Methodological Considerations of Rasch Models

Abstract

The purpose of this paper is to examine the sensitivity of commonly used Rasch fit measures to different distributions of error in item responses. Using Monte Carlo methods, we generated five different measurement error distributions within the Rasch Partial Credit model, and we recorded the distribution of INFIT and OUTFIT for each error distribution or combination of error distributions. INFIT and OUTFIT were not very sensitive to different error distributions, but the separation reliability was sensitive, as expected. The findings have implications for the use fit measures in diagnosing model misfit.

Authors

• R. Noah Padgett, Baylor University

• Grant B. Morgan, Baylor University