Researchers from the Skoltech Engineering Center have conducted an umbrella review of two decades of research on Project‑Based Learning (PjBL). The review, published in Educational Research Review — one of the most prestigious journals in education and learning sciences, ranking in the top 1% of Elsevier publications — offers a comprehensive analysis of how effective this approach is and under what conditions it works best.
Project-Based Learning is an instructional method in which students acquire knowledge by solving real-world practical problems. Today, it is actively implemented in education worldwide — from primary schools to universities, across disciplines ranging from science and engineering to the humanities. However, the academic community has accumulated a substantial body of conflicting evidence: some meta-analyses reported modest effects, while others found impressive gains. To make sense of this diversity, the authors conducted an umbrella review — a study that synthesizes findings not from individual experiments but from entire meta-analyses, and which sits at the highest level of the scientific evidence hierarchy.
Their review brought together 15 meta-analyses spanning 351 primary studies and roughly 50,000 students.
“The analysis revealed a consistently positive effect of project-based learning compared to traditional methods,” noted Sabah Farshad, a research scientist at the Skoltech Engineering Center. “The study confirmed that these positive results are not coincidental — primary-study overlap across different meta-analyses was only 0.28 %, meaning the findings draw on independent sources of evidence rather than repeated re-analysis of the same studies.”
Of particular interest is the analysis of conditions under which project-based learning delivers the best results. The study showed that two factors consistently influence effectiveness: working in small groups (3–5 students) and using technology in project activities. By contrast, the common assumption that project-based learning is particularly effective in STEM disciplines did not find clear support — the data here are contradictory.
The authors also highlight an important methodological finding: the quality of the meta-analytic studies themselves remains insufficient. All 15 reviews analyzed received a “critically low” rating on the AMSTAR 2 standard — primarily due to the absence of pre-registered protocols and risk-of-bias assessment in primary studies. The authors frame this not as a verdict against PjBL, but as a clear agenda for strengthening the evidence further.
Professor Clement Fortin, the project’s principal investigator, adds: “A real strength of this review is that it cleanly separates Project-Based Learning from Problem-Based Learning, which the literature often blends together. Bringing two decades of meta-analyses together at the highest level of evidence gives educators a clearer, more trustworthy foundation for using PjBL well — and a sharp agenda for the research that will strengthen it further.”
This work is the outcome of a long-term research program at the Skoltech Engineering Center, aimed at understanding how engineering students learn and engage in the educational process. Building on these rigorous standards, Skoltech is already developing new educational technologies — the StayTeam.ai platform (2024–2025) and SEQA (Smart Engagement Quiz Assistant), an interactive AI-powered lecture game with a built-in personal tutor for each student, currently under development.
