Online problem-based learning interaction networks and medical student performance

A study of 135 students and 15 tutors across 15 problem-based learning groups analyzed 2,620 online interactions, 89 percent of them student-to-student. Out-degree and tutor interaction had small positive correlations with performance, showing useful relational signals without supporting causal ranking. The results support examining who exchanges help with whom inside PBL groups rather than reducing collaboration to total logins, while keeping intervention decisions proportional to modest effects.

Student-to-student exchanges formed 89 percent of 2,620 interactions; out-degree correlated with performance at about 0.27 and tutor interaction at about 0.22, both small observational associations. This exploratory single-course case used platform activity as a proxy for educational exchange and did not model content quality, prior achievement, nested dependence, or unobserved offline interaction comprehensively. Its small correlations do not justify causal claims or automated high-stakes decisions.