Cognitive Diagnostic Computerized Adaptive Testing: Recent Developments and Future Directions
Ying (Alison) Cheng – University of Notre Dame, USA
Abstract: This talk will provide a comprehensive and up-to-date overview of cognitive diagnostic computerized adaptive testing (CD-CAT). Compared to the well-known traditional CAT, a key distinction of CD-CAT is that its goal is to obtain the latent mastery profile for each respondent in an efficient manner, typically based on the cognitive diagnostic models (CDMs), also known as the diagnostic classification models (DCMs). In contrast, the goal of the traditional CAT is to reach an accurate latent ability estimate or multiple latent ability estimates quickly, and in some cases make a classification decision on that basis. In this talk, the connections and differences will be discussed between CD-CAT, unidimensional CAT, multidimensional CAT, and classification CAT. Under CD-CAT, this talk will cover both the single-purpose CD-CAT, which focuses on the latent mastery profile itself, and dual-purpose CD-CAT, which intends to estimate both the latent mastery profile and the latent ability simultaneously. In addition, emerging topics such as extension of CD-CAT to incorporate response times, cognitive diagnostic multi-stage testing, and automated test assembly under CDM/DCM will also be explored.