Academic Achievement and Physical Activity: A Meta-analysis

26 intervention studies, 10,205 children aged 4-13

Children aged 4-13 in school and after-school physical-activity interventions

Random-effects meta-analysis of intervention studies (not restricted to RCTs) on academic achievement and classroom behaviour, with moderator analysis of programme characteristics. Graded C — mixed intervention designs, heterogeneous outcome instruments (the authors name this as their principal limitation), and no selection-model publication-bias adjustment.

The most-cited pediatric meta-analysis supporting the achievement claim, and reading it by outcome class is instructive. Language-related skills 0.16 (95% CI -0.06 to 0.37) — CROSSES ZERO. Mathematics 0.21 (0.09 to 0.33), reading 0.13 (0.02 to 0.24), composite 0.26 (0.07 to 0.45). And time on task 0.77 (0.22 to 1.32) — four times any achievement effect, on the least standardized outcome in the set. That ordering is the signature of the field: the more researcher-proximal and behavioural the measure, the bigger the effect. The achievement estimates here (0.13-0.26) are precisely the range that the four large cluster RCTs with independent standardized national/state tests failed to reproduce, and that study-level bias correction removes.

Includes non-randomized intervention studies, which admits selection into treatment; the randomized subset is not separately reported, so the pooled figure inherits mixed confounding.