Estimating and Assessing Multidimensional Item Response Theory Models for Polytomous Response Under the Bayesian Framework

• In Event: Developments in Item Response Theory Modeling

Abstract

The Bayesian full-information approach via the Markov chain Monte Carlo (MCMC) method is found increasingly useful across various MIRT settings. But the development of the Bayesian model assessment for MIRT models lags behind. This research first targets to estimate different MIRT models for polytomous responses via MCMC. The multidimensional versions of two widely-used polytomous models are selected for evaluation and comparison: GRM and GPCM. The second target of this research is to find discrepancy measures via PPMC that are sensitive to model misfit for polytomous MIRT models. Several PPMC discrepancy measures will be discussed and evaluated using the above two models. Simulation study will be the main vehicle to achieve both targets.

Author

• Jinsong Chen, Sun Yat-Sen University