Abstract Does the integration of games into mathematics instruction genuinely enhance learning? While numerous studies have explored this question, their findings remain inconsistent. To address this uncertainty, the present study employed a second‐order meta‐analysis approach, synthesizing 20 existing meta‐analyses that encompass 688 primary studies. This methodological approach enabled an examination of how games influence students' cognitive and affective outcomes in mathematics education, along with the underlying moderating mechanisms. The results demonstrate that gamified learning yields a moderate yet statistically significant positive effect ( g = 0.407) compared to traditional instruction. Further analysis revealed a significant positive effect on cognitive outcomes ( g = 0.454), whereas the impact on affective outcomes, although directionally positive ( g = 0.226), did not reach statistical significance. Regarding potential moderators, the methodological quality of the included studies emerged as a significant moderating variable, with lower‐quality studies tending to overestimate the actual effects of games. Educational stage, game type, publication year and effect size independence failed to demonstrate significant moderating effects. Furthermore, robustness tests involving recursive elimination of overlapping samples confirmed that the core findings remained consistent across varying degrees of sample overlap. This study not only affirms the positive value of gamified learning in developing mathematical cognition but also underscores the crucial importance of research quality in determining the reliability of conclusions, thereby providing high‐level evidence to inform future gamified teaching practices and related research.
D. Z. Yao (Tue,) studied this question.