The Differences in Item Difficulty Parameter Estimations for Simulated Data having (MC1- PL) Model in Item Response Theory


Abstract in English

This study aimed at finding out the differences in item difficulty parameter estimations for Simulated Data having (MC1- PL) Model in Item Response Theory (IRT) according to the differences in a test Dimensions (3D; 2D; 1D), correlations between these dimensions (0.0, 0.50, 0.86), and the statistical program used in analyzing the data (NOHARM ؛Bilog- MG3). The Monte Carlo simulations data having (MC1- PL) Model in IRT using (RESGENT) program; that fully filled in a 21-item was used to achieve the study aims. Data were analyzed using the statistical programs (NOHARM ؛Bilog- MG3). The results revealed no statistically significant differences in item difficulty parameter estimations that construct the multidimensional test within items due to differences in a test, correlations between these dimensions, and the statistical program used in analyzing the data (NOHARM ؛Bilog- MG3), as well as the estimations were consistent and high. Finally, the study recommends using these statistical programs for a data having (MC1- PL) Model, especially when similar assumptions are satisfied in a real data.

References used

Ackerman, T. Using multidimensional item response theory to understand what items and tests are measuring. Applied Measurement in Education, 7, (1994). 255- 278
(Capar, N. (2000). Analyzing multidimensional response data structure represented by unidimensional IRT models increase the precision of scoring using out-of-scale information. Paper presented at the annual meting of the Florida educational research association (45th Tallahassee, FL, November 8-10, 2000
(Dawadi, B. (1999). Robustness of the polytomous IRT to violations of the unidimensionality assumption. Paper presented at the annual meeting of the American educational research association (Montreal, Canada, April 19-23, 1999

Download