Completion of prior problem slides is the most critical predictor of completing a module's final exercise, consistent with the doer effect¶

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? Problem slides made up 75% of features selected by GR on average across modules; students completing later problem slides were more likely to pass the final problem slide. → Vincent Zhang 2025

Evidence¶

Vincent Zhang 2025¶

Vincent Zhang, Bryn Jeffries, Irena Koprinska. (2025). A Machine Learning Approach for Predicting Student Progress in Online Programming Education. Int J Artif Intell Educ. https://doi.org/10.1007/s40593-025-00510-9

q2 · i? · associational · r2

Feature-selection analysis across modules in the Discussion (Q3, Observation 1). The article states these results are consistent with the doer effect (Koedinger et al., 2016), which posits interactive practice is more beneficial than passive reading and video watching.

"On average, problem slides make up 75% of features selected by the GR feature selection algorithm across modules. Students who complete problem slides, particularly later ones, are more likely to pass the last problem slide in the module, and students who do not make a submission or fail the problem slide are likely to make no submission on the final problem slide."

Discussion¶

Related Claims¶

• Decision tree classifiers trained on slide interaction features predict end-of-module outcomes with high accuracy across four K-12 programming courses — a broader claim this one bears on
• CFS and GR feature selection shrink decision trees to 3.2-4.7 leaf nodes without degrading predictive performance — related
• Course-module views, assignment submissions, and video-access days are the most stable predictors, all lower in the high-risk group (Cliff's delta values from 0.50 to 0.60) — related
• The approach offers a large real-time intervention window, with a median of 176.2-196.1 hours between the prediction point and the final problem deadline — related