Group-based learning in engineering education fosters technical competence and teamwork skills; however, assessment practices often struggle to ensure individual accountability, leading to grading bias and uneven participation. This study implements a multi-source peer assessment model in a Collaborative Online International Learning (COIL) program jointly conducted by Shibaura Institute of Technology (Japan) and Institut Teknologi Sepuluh Nopember (Indonesia). The whole 14-week hybrid online–onsite physics course (Magnetism and Magnetic Materials) was embedded within an interdisciplinary COIL module that integrated STEM content. Additionally, the humanities perspectives through a two-week collaboration with a History of Japan module, broadening the notion of “contribution” beyond technical problem-solving. To promote fairness, a weighted scoring model was developed, integrating within-group peer evaluations (20%), cross-group peer evaluations (10%), Indonesia-based instructor evaluations (30%), and Japan-based physics lead instructor evaluations (40%). The model was compared with a conventional instructor-only assessment approach using data from two cohorts: FY2024 (n=21) and FY2025 (n=30). Instructor-only averages exceeded peer-inclusive adjusted scores by approximately 2–5 points across cohorts, suggesting grade inflation under single-source evaluation. Statistical analysis showed significant differences (p < 0.01) across both academic years, while maintaining strong correlation (r > 0.87) and rank-order stability (ρ > 0.82). These findings indicate meaningful score correction without substantially altering performance ranking. The proposed multi-source model reduces reliance on a single evaluator, enhances transparency of individual contributions in collaborative contexts, and offers an empirically validated assessment framework for interdisciplinary, cross-institutional STEM COIL environments—an area that remains underexplored in engineering education research.
The full paper will be available to logged in and registered conference attendees once the conference starts on June 21, 2026, and to all visitors after the conference ends on June 24, 2026