Applying a Cognitive Diagnosis Model to Mental Cutting Test

On the basis of the framework of the G-DINA model, the attributes (skills) required to solve the Mental Cutting Test (MCT) and the Q-matrix that describes the relationship between those attributes and items (problems) of the MCT were specified with the help of suggestions from statistical methods. Item parameters, which represent effects of attribute mastery on the probabilities of answering the item correctly, were estimated with good model fit. Estimated values of the item parameters indicate that many of the items contribute to discrimination of the examinees’ skill status. The estimated mastery probability of each attribute for each examinee provides information on the individual strength and weakness related to solving the MCT problem. The relationship between the MCT scores and estimated mastery status of attributes of each examinee suggests that the status of mastery of attributes is different between examinees, even those with similar MCT scores (especially around the mean scores). These findings mean that CDM analysis applied to response data of the MCT has the potential extracting rich information about the spatial skill status of examinees. Refining the Q-matrix to improve the accuracy of diagnostic information is an issue that requires further consideration.