Math identity, or the degree to which one considers oneself a “math person,” has become an area of interest among social scientists hoping to better understand what drives students to enter Science, Technology, Engineering, and Mathematics (STEM) fields. We leverage natural language processing (NLP), click-stream analyses, temporal features of performance, and survey data to predict students’ mathematics success and math identity (namely, self-concept, interest, and value of mathematics) in a blended learning environment. Following our limited success with purely cognitive approaches, we explore the influence of social factors on the relationship between students’ interaction with the learning system and their math identity.
Data - Survey, interaction log, language, school demographics
Methods - Time series analysis, natural language processing, correlational analysis, regression analysis
- Karumbaiah, S., Labrum, M., Ocumpaugh, J., Baker, R.S. (2019) The Influence of School Demographics on the Relationship Between Students’ Help-Seeking Behavior and Performance and Motivational Measures. Proceedings of the 11th International Conference on Educational Data Mining (EDM). [pdf]
- Crossley, S.A., Karumbaiah, S., Ocumpaugh, J., Labrum, M., Baker, R.S. (2019) Predicting Math Success in an Online Tutoring System Using Language Data and Click-stream Variables: A longitudinal analysis. Proceedings of the Conference on Language, Data and Knowledge (LDK). [pdf]
- Karumbaiah, S., Ocumpaugh, J., Labrum, M., Baker, R.S. (2019) Temporally Rich Features Capture Variable Performance Associated with Elementary Students’ Lower Math Self-concept. To appear in Proceedings of the Workshop on Online Learning and social-Emotional Learning at the 9th International Conference on Learning Analytics and Knowledge (LAK). [pdf]