Errors Are Also Important: One Learning Analytics for Mathematics Multiple-Choice Questions (Poster 1)

• In Event: AERA Poster Session Wednesday 10:50 a.m.
  In Poster Session: Technology-Mediated Learning and Social Interactions

Abstract

Knowledge tracing is a method used to evaluate students' knowledge states by predicting the binary correctness of their next answer based on historical problem-solving records. It is widely applied in closed-ended questions to effectively identify students' future option selection tendencies. This study is based on millions of multiple-choice questions from a mathematics learning platform, which includes two types of Knowledge Components (KC): skills and misconceptions. We extracted features from students' problem-solving sequences and learning gains which were evaluated by the KT model, to analyze learning patterns. The results indicate that for beginners in middle school mathematics, making repeated errors on the same skill may lead to greater learning gains. Our work contributes to the field of learning analytics for knowledge assessment.

Authors

• Hai Li, University of Florida

• Wanli Xing, University of Florida