A Polytomous Multilevel Measurement Model for Assessing Differential Functioning on Testlet-Based Polytomous Items

• In Event: Advancements in Detecting Differential Item Functioning and Measurement Invariance

Abstract

In this study, a two-level polytomous multilevel measurement (PMMMT-2) model was specified to assess differential functioning on polytomous testlet items using Bayesian Markov chain Monte Carlo (MCMC) approach. The PMMMT-2 model’s assessment of differential functioning was compared with that of IRT-Based DIF detection model for investigating different sources of differential functioning at step and testlet level. Simulation results supported the PMMMT-2 model is a useful and effective tool to assess differential functioning on testlet-based polytomous items. This research finding has important implications for psychometric research and practice.

Authors

• Wei Xu, The American Board of Anesthesiology

• M. David Miller, University of Florida