Matthew Madison, University of Georgia

Diagnostic classification models (DCMs), also known as cognitive diagnosis models (CDMs), are psychometric models designed to classify respondents into interpretable categories. These categories, for example, can be interpreted as proficiency levels (e.g., educational assessment) or behavior diagnoses (e.g., psychological assessment). Despite their potential in many applications where classification is the primary objective, DCMs are not widely used in large scale assessment or research applications. In this talk, I conjecture some reasons why this may be the case. More specifically, I discuss some logistical, methodological, and interpretational challenges, and explore potential solutions to overcome some of these challenges.

about the speaker

Matthew Madison is Assistant Professor of Quantitative Methodology at the University of Georgia, where he studies and teaches courses in educational measurement and statistics. His research centers around the advancement of multivariate psychometric models. More specifically, Dr. Madison has focused his research on a class of contemporary item response models called diagnostic classification models (DCMs) and their application in educational contexts. Motivated by his background in mathematics education, his research seeks to use DCMs to improve the utility of educational assessments. Recently, he has investigated longitudinal DCMs as alternative to traditional methods for modeling learning over time. His research has been supported by Institute of Education Sciences and National Science Foundation, and he has published research in journal such as Psychometrika, Journal of Educational Measurement, and Journal of Educational and Behavioral Statistics.