Poster #26 - Many Ways for Item Response Theory: Comparing the Performance of Different Software Packages in Parameterization

• In Event: 79.052 - MTCC Poster Session 19
  In Poster Session: 79.052-2 - Educational Measurement, Psychometrics, and Assessment Poster Session

Abstract

This study compares the performance of different software, include lme4, ltm, and psych in R, irt, and meqrlogit in STATA and flexMIRT, in estimating 1PL and 2PL item response theory (IRT) models. Both simulated and collected data are used, and the performances of different software are compared by their mathematical parameter definition, accuracy of estimation and the effects of multidimensionality. Results indicates different packages vary in parameter definitions, especially for the signal of parameters. Although using different parameter definitions, the packages reported similar accuracy in parameter estimation once they have enough number of quadrature points and using logit link. The psych package provided the least accuracy. Suggestions are provided to researchers according to the results.

Author

• Yixi Wang, The Ohio State University