How effective is gamification in education? A second-order meta-analysis of effect sizes, methodological quality, and small-study effects

Meta-analyses of gamification in education have multiplied, but their overlapping evidence, uneven quality, and divergent estimates make the size and robustness of gamification effects difficult to judge. This second-order meta-analysis included 55 first-order meta-analyses published through October 2026, identified through Crossref, Semantic Scholar, ERIC, and citation searching. One estimate per meta-analysis and outcome domain was pooled in random-effects models with restricted maximum likelihood and the Hartung-Knapp-Sidik-Jonkman adjustment; methodological quality was appraised with AMSTAR 2, and overlap with the corrected covered area. Gamification had moderate positive average effects on cognitive outcomes (g = 0.68, 95% CI [0.55, 0.81], k = 28), motivational and affective outcomes (g = 0.58 [0.50, 0.67], k = 21), and behavioral outcomes (g = 0.56 [0.42, 0.71], k = 9). Cognitive estimates varied far beyond second-order sampling error (95% prediction interval [0.08, 1.28]), whereas motivational and behavioral estimates were more consistent. Meta-analyses with larger standard errors reported larger cognitive effects, and precision-effect estimates reduced the cognitive effect to between 0.29 and 0.48. Of the 28 meta-analyses with accessible full texts, 24 were rated critically low in confidence, mainly because they lacked protocols, risk-of-bias assessment of primary studies, and justified exclusions. Educational level, subject, publication period, and focus on specific tools did not reliably moderate effects, and moderator findings reported within meta-analyses were inconsistent. Gamification appears beneficial on average, but the meta-analytic literature probably overstates its cognitive effects and offers little guidance on when it works. Future syntheses should be preregistered, appraise primary studies, and model small-study effects.