In this study, the root mean square error values of item parameters’ estimation in a two-dimensional structure condition were examined under different conditions, considering three and five categories with different algorithms (Expectation–Maximization, Metropolis–Hastings Robbins–Monro, Quasi-Monte Carlo Expec tation–Maximization). The simulation conditions included two different sample sizes (1500 and 3000) in a two-dimensional structure, three test lengths (12, 24, and 36), three different interdimensional correlations (0.20, 0.50, and 0.80), and two different category numbers (three and five). Analyses were conducted with three algorithms and the graded response model from the multidimensional item response theory in 36 different conditions with 100 replications. When the errors were examined in terms of the root mean square error, an increase in the number of categories resulted in a partial decrease in most item parameters under the condition of 1500 sample size. For researchers conducting analyses in the polytomous multidimensional item response theory, it is recommended to use as large a sample as possible, at least 24 items, five categories, and the Quasi-Monte Carlo Expectation–Maximization algorithm.
Cite this artcle as: Büyükkıdık, S., & Atar, H. Y. (2023). Investgatng item parameter estmaton accuracy in multdimensional polytomous data under various conditons. HAYEF: Journal of Educaton, 20(3), 221-230.