The Comparison of Two Input Statistics for Heuristic Cognitive Diagnosis

Abstract

Cognitive diagnosis models of educational test performance decompose ability in a domain into a set of specific binary skills called attributes. (Non-)mastery of attributes documents an examinee’s strengths and weaknesses in the domain as a profile of mental aptitude. Distinct attribute profiles define classes of intellectual proficiency to which examinees can be assigned. Nonparametric, model-free classification methods have been proposed as heuristic or approximate alternatives to maximum likelihood estimation procedures for assigning examinees to proficiency classes. These classification techniques use as input a statistic obtained by aggregating each examinee’s test item scores into a profile of attribute sum-scores. This study demonstrates that clustering examinees into proficiency classes based on their item scores rather than on their attribute sum-score profiles results in a more accurate classification of examinees.