Scalable Skill Tracing Using Cognitive Diagnosis Model With Covariates

• In Event: AERA Research in Progress Roundtable Session Two
  In Invited Roundtable: Research-in-Progress RT002 (Thursday)

Abstract

Cognitive diagnosis models (CDMs) provide fine-grained feedback on students’ skill mastery but are typically limited to cross-sectional data. Longitudinal CDMs can capture learning progress across multiple time points, yet existing frameworks are computationally intensive and require complete data across all administrations, limiting their practical use for real-time feedback. This study proposes to use CDMs with students’ skill mastery profiles at earlier time points as covariates to improve the attribute diagnosis. A simulation study and real data analysis will evaluate the performance of the proposed method under varying conditions. Anticipated results will demonstrate efficiency gains while maintaining accuracy and precision comparable to existing models.

Authors

• Scarlett Escudero, University of Minnesota

• Wenchao Ma, University of Minnesota

• Nidhi Kohli, University of Minnesota