Evaluating undergraduate education through multiple choice tests

Abstract

Multiple choice tests have been recently introduced in Italian Universities as an instrument for evaluating students’ performance. In our paper we appraise the objectiveness of such a classification, adopting a Dichotomous Rasch Model (RM, Rasch (1960)) that transforms, via a Simple Logistic Model (SLM), test scores into the general structure of correct and incorrect answers. As widely understood, measurement requires that the quantification assigned to the construct be independent of both the observer and the sample of respondents. Subsequently, we extend the analysis to the polytomous answers, by means of a Partial Credit Model (PCM, Masters (1982)), in order to further distinguish different levels of incorrectness, between the classification of the errors above.