The aim of this study is to develop a cognitive diagnostic model to analyze tests with Constructed Response (CR) items. The problem solving process of CR item transferred to categorical data can provide more information than dichotomized data. However, most cognitive diagnostic models modeling categorical or dichotomized data are proposed for tests with Multiple Choice (MC) items. In this study, an extended DINA model using multiple category scoring is proposed for modeling categorical data obtained from CR items. Each response category of categorical data is a combination of multiple skills measured by the item. The Expectation-Maximization (EM) algorithm is applied to estimate the parameters in the proposed model. Finally, results of simulation studies and analyses of real data are presented and discussed.
|Publication status||Published - Jul 2015|