Karma Testlerin Psikometrik Özelliklerini Belirlemede Klasik Test Kuramı ve Rasch Modelinin Karşılaştırılması

Abstract

The aim of this study is to indicate which one of the Classical Test Theory and Rasch model is providing more detailed statistical information in determining psychometric properties of mixed tests through comparing their similarities and differences. While estimating item and test parameters according to Classical Test Theory and Rasch Model, Microsoft Excel and Winsteps version 4.3.2 are used respectively. The data set of the study consists of 11193 students’ answers to the ABİDE (Monitoring and Evaluating of Academic Skills) exam. Analysis are restrained to mathematics and science subtests of ABİDE exam. Each of the subtests comprises of eighteen items including nine multiple choice and nine open ended questions. The two methods showed similar results in determining item difficulties. While examining the possible problematic items regarding reliability and validity; the two methods pointed out the same items in science subtest but Rasch model signaled one more item in mathematics subtest. Both subtests’ reliability is approved by both methods. It is shown that the less common Rasch model provides more detailed statistical information about items and persons. Therefore, it is stated that the Rasch model could be a good alternative to classical test theory while achieving statistical information on large scale assessments.