A Comparison of Estimation Methods for a Multi-Unidimensional Graded Response Item Response Theory Model

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

Abstract

This study compared several parameter estimation methods for multi-unidimensional graded response models using their corresponding statistical software programs and packages. Specifically, we compared two marginal maximum likelihood (MML) approaches (Bock-Aitkin expectation-maximum algorithm, adaptive quadrature approach), four fully Bayesian algorithms (Gibbs sampling, Metropolis-Hastings, Hastings-within-Gibbs, blocked Metropolis), and the Metropolis-Hastings Robbins-Monro (MHRM) algorithm via the use of IRTPRO, BMIRT, and MATLAB. Preliminary results suggested that, with 1,000 persons and 20 items, the two MML approaches, together with blocked Metropolis and MHRM, as implemented in IRTPRO had an overall better parameter recovery than the others when multiple latent traits measured by the instrument had a low or high correlation.

Authors

• Tzu Chun Kuo, Southern Illinois University - Carbondale

• Yanyan Sheng, Southern Illinois University - Carbondale