Bias in predicting learning productivity, measured as accuracy differences across gender and parental educational background, was highly dependent on the ML method used¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Predictive accuracy differed between student groups defined by gender and parental educational background, with the extent of this separation varying across machine learning models. → Kubsch 2025

Evidence¶

Kubsch 2025¶

Kubsch, M., Strauß, S., Grimm, A., Gombert, S., Drachsler, H., Neumann, K., & Rummel, N. (2025). Self-regulated Learning in the Digitally Enhanced Science Classroom: Toward an Early Warning System. Educational Psychology Review. https://doi.org/10.1007/s10648-025-10011-9

q2 · i? · associational · r2

Bias analysis (research question 2) evaluated separation—whether models make correct and incorrect predictions at similar rates for different groups—on the best-performing models from research question 1, focusing on the technical aspect of bias.

"To investigate to what extent the predictions of students' learning outcomes exhibited bias with respect to gender or parental educational background, we focused on the best performing models from research question 1. For those models, we evaluated the predictive accuracy for each Gender and parental educational background separately."

Discussion¶

Related Claims¶

• Standard ML models predicting student pass/fail from VLE data exhibit biased true-positive rates across demographic subgroups — related
• Preprocessing bias mitigation (DIR, RW, SUP) reduced subgroup disparities in TPR on OULAD, with effectiveness varying by demographic attribute — related
• Post-hoc fairness interventions on a vendor-controlled EWS redistributed disparities across demographic groups without consistently reducing them — related