Hurdles and solutions for cognitive diagnosis

Abstract

Cognitive diagnostic modeling has become an important field of psychometric research. The models are special because they provide examinees with diagnostic information regarding whether or not they have mastered individual skills in certain area. They provide as meaningful sources of information for examinees to understand their learning status and the criteria that will be used to evaluate their performance. Furthermore, in a school context, they enable teachers to know the strengths and weaknesses of their students and thus help them to select and structure teaching materials. Cognitive diagnosis models help administrators support their decision-making concerning learning and teaching objectives and arrangements. Despite their obvious advantages, researchers and practitioners face three fundamental hurdles in the implementation of cognitive diagnosis models. The purpose of this dissertation is to provide the solutions to these problems: the estimation of the Q-Matrix, the estimation of model parameters, and the assessment of model-fit.