Madde Düzeyinde Boyutluluk Modellerinin Bilgisayar Ortamında Bireyselleştirilmiş Test Yöntemleri Üzerindeki Etkisinin İncelenmesi

Abstract

The purpose of this study is to measure students’ language abilities with Compensatory Multidimensional Computerized Adaptive Testing (MCAT) designs using different ability estimation, item selection methods and stopping rules; and to examine the effect of item-level dimensionality models on MCAT. For this purpose, real data set from English Proficiency Test (EPT) administered by Hacettepe University was used to conduct post-hoc simulation, in which each test consist of three dimensions listening, reading and grammar, respectively. In this study, 10 EPT data sets administered between 2009 and 2013, were used to conduct analysis. Item parameters were estimated with compensatory multidimensional 2 parameter logistic model (CM-2PLM) and item pool for with-in item dimensionality model consisted of 565 items, while item pool for between item dimensionality consisted of 559 items.