The comparison of classification accuracy of two tests measuring the same feature obtained by different methods

In this study, it was aimed to compare three different statistical methods presented as an alternative to traditional standard setting methods. Accordingly, two different forms of the mathematics achievement test developed by the researcher in accordance with the 7th grade level were used as data collection tools. These forms consist of multiple choice and short answer items that measure the same features. The data were obtained from a total of 326 students, 159 of which were girls and 167 boys, in two different state schools in Gaziantep in the 2020-2021 academic year. 45 points determined as the passing score by the Ministry of National Education were accepted as the criterion, and the classification accuracies for each method and each form were obtained by examining the pass / fail situations from applied tests with logistic regression, cluster and ROC analysis methods. In the findings of the study, when considering the methods, the most accurate classification percentage for MCF (95.9%) and for SAF (93.2%) was obtained by ROC analysis. Considering the forms, the classification accuracies obtained separately for each of the three methods did not show a statistically significant difference. Keywords: Standart Setting, Classification Accuracy, Logistic Regression Analysis, Cluster Analysis, ROC Analysis