Small Sample Cognitive Diagnosis Using Bayesian Catalytic Priors

• In Event: Cognitive Diagnostic Models II

Abstract

Cognitive Diagnosis Models (CDMs) are powerful tools but can be underutilized in classroom settings due to large sample size requirements. Through simulation studies and using Tatsuoka’s fraction subtraction dataset, we demonstrate how Bayesian catalytic priors can lead to improved estimation and classification accuracy of CDMs in small sample sizes.

Authors

• David Arthur, University of Washington Tacoma

• Arman Sabbaghi, Purdue University Department of Statistics

• Hua-Hua Chang, Purdue University