Equating Test Forms Composed of Testlets Using Dichotomous and Polytomous IRT Models.

Item response models can be applied in many test equating situations by making strong statistical assumptions. Thus, studying the robustness of the models to violations of the assumptions and investigating model-data fit are essential in all item response theory (IRT) equating applications (M. Kolen and R. Brennan, 1995). Previous studies dealing with tests composed of testlets have indicated that the assumptions of the dichotomous IRT models are frequently violated; however, passage scores can be used instead of item scores to eliminate the effect of the dependence among within-passage items (G. Lee and D. Frisbie, 1997; H. Wainer and D. Thissen, 1996; S. Sireci, D. Thissen, and H. Wainer, 1991). The purpose of this study was to compare the performance of polytomous IRT models to that of the dichotomous three-parameter logistic model in the context of equating test forms composed of testlets, using traditional equating methods as criteria for both. For equating test forms composed of testlets, equating methods based on polytomous IRT models were found to produce results that more closely agreed with the results from traditional methods than did equating methods based on the dichotomous three-parameter logistic IRT model. (Contains 7 tables, 10 figures, and 32 references.) (Author/SLD) ******************************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. ******************************************************************************** Equating Test Forms Composed of Testlets Using Dichotomous and Polytomous IRT Models PERMISSION TO REPRODUCE AND DISSEMINATE THIS MATERIAL HAS BEEN GRANTED BY