Investigation of Models Used in Equating Testlet-Based Tests

Authors : Ertunç Ukşul, Hülya Kelecioğlu

Pages : 951-970

Doi:10.7822/omuefd.1419482

View : 27 | Download : 53

Publication Date : 2024-12-30

Article Type : Research Paper

Abstract :This study aims to examine the effects of testlets on test equating. For this purpose unidimensional item response theory, two-factor item response theory and testlet response theory models were applied to the testlet-based tests for the estimation of item and ability parameters. In order to equate the tests, the parameters were placed on the common scale using mean-mean, mean-sigma and Stocking-Lord scale transformation methods under the common-item non-equivalent groups design. Then, the equating errors of the models depending on the scale transformation method and the number of testlets were calculated and compared. Equating errors were compared with Root Mean Squared Error. In the study, the science test of the Trends in International Mathematics and Science Study project administered in 2019 was used as the data collection tool. As a result of the study, it was determined that the use of unidimensional item response theory model increased the equating error, while the use of two-factor and testlet response theory models decreased the equating error as the number of testlets in the test increased. In order to compare the models, the correlation between the parameters obtained from the models after scale transformation was examined and it was found that the item parameters were more affected by the model selection than the ability parameter. In addition, it was concluded that the equating errors obtained from the mean-mean and Stocking-Lord scale transformation methods were lower than the mean-sigma method.
Keywords : Test eşitleme, Ölçek dönüştürme, Madde tepki kuramı, Madde takımı tepki modeli, İki faktör model.