Cognitive diagnosis modeling of clinical data: An example and step by step illustration

Abstract

This is a two-part seminar that showcases the application of cognitive diagnosis models (CDMs), specifically the generalized deterministic inputs, noisy “and” gate (G-DINA) model, in the clinical setting. Although primarily developed in education, CDMs are general tools that can also be used to identify the presence or absence of clinical disorders. In the first part of the seminar, we provide a brief review of the various models and procedures within the G-DINA model framework. This is followed by a CDM analysis of MCMI-III data. Using empirically based simulated data, we illustrate in the second part how CDM analysis can be performed using the GDINA R package. This is a hands-on exercise where participants will conduct a number of CDM procedures, which include parameter estimation, model fit evaluation, Q-matrix validation, differential item functioning analysis, and classification accuracy estimation.