A1487
Title: A cognitive diagnostic model for matching format tests
Authors: Rinhi Higashiguchi - The University of Tokyo (Japan) [presenting]
Kentaro Fukushima - The University of Tokyo (Japan)
Kensuke Okada - The University of Tokyo (Japan)
Abstract: Matching format tests are widely used in achievement assessments and psychological evaluations, where a respondent is presented with a list of test items and response alternatives and asked to match each response alternative with a test item. However, the development of cognitive diagnostic models (CDMs) tailored to matching format tests remains unexplored. CDMs are models designed to identify the presence or absence of multiple fine-grained attributes, enabling more precise and informative assessments. When matching format tests are analyzed using conventional CDMs, there is a risk of overestimating respondent parameters due to the unique structure of the test format. To address this issue, a deterministic inputs are proposed, noisy, and gate (DINA) model for matching test formats. Amongst CDMs, the DINA model assumes that all attributes are required to correctly answer an item (in exam settings) and is parsimonious and easily interpretable. A comparison between the conventional DINA model and the proposed model was conducted through a simulation study. The results suggest that the proposed DINA model for matching format items offers a promising approach to improving the diagnostic accuracy of CDMs when applied to this unique test format.
