Psychological outcomes of primary-school physical education interventions: a systematic review and multilevel meta-analysis

Abstract

Physical education (PE) offers opportunities for psychological learning, but evidence of benefit depends on the outcome, design, and instructional context. This systematic review and multilevel meta-analysis included 74 reports from 70 trial families of primary-school PE interventions. Four databases covered English-language reports published from January 2000 through 31 May 2026. Randomized and nonrandomized controlled studies were distinguished from uncontrolled pre -post and short-term within-child designs. Multilevel models addressed dependent effects, with trial-family CR2 inference. Source-informed controlled analyses contained five core social-emotional effects from three families (g = 0.947, 95% CI [−0.774, 2.667]; Holm-adjusted p = 0.424) and 13 adjacent psychological effects from five families (g = 0.215, 95% CI [−0.386, 0.817]; Holm-adjusted p = 0.424). Model-based prediction intervals were [−2.387, 4.280] and [−1.368, 1.799], respectively. The broader controlled analysis, requiring additional variance or clustering assumptions, included 126 effects from 28 families and yielded g = 0.381, 95% CI [0.135, 0.627]. The broader and combined source-informed estimates were similar (g = 0.381 and 0.421, respectively), although the latter was less precise; because the analyses overlap, they do not provide independent confirmation of benefit. The primary estimates remain too uncertain to define a dependable average benefit, rather than demonstrating that benefits are absent. Positive broader evidence warrants attention alongside substantial heterogeneity, source-reporting limitations, and an inconclusive exploratory small-study diagnostic. Future trials should align psychological targets, instructional content, implementation measures, and validated outcomes, and report allocation-aware uncertainty.