Modern Psychometric Approaches for Diagnostic Assessment

Ray Reichenberg, Ph.D.

Research Assistant Professor, Nebraska Academy for Methodology, Analytics and Psychometrics
Download Powerpoint


This presentation will provide an overview of some modern psychometric methods for modeling diagnostic assessment data and generating attribute profiles. These methods include latent class models, diagnostic classification models and Bayesian networks.

Diagnostic assessment is the evaluation of an individual’s strengths and weaknesses with respect to the components of a particular construct for the purpose of formulating a course of instruction or treatment.

Diagnostic measures are designed to estimate attribute profiles defined by the respondent’s state or ability across a collection of related, latent characteristics. These profiles can guide next steps for the respondent, such as a particular lesson or intervention.

Recent issues, such as modeling attribute hierarchies (e.g., learning progressions) and approaches for capturing growth in an attribute over time, also will be discussed, along with software options for implementing these models.