Does Physical Fitness Affect Academic Achievement among Japanese Adolescents? A Hybrid Approach for Decomposing Within-Person and Between-Persons Effects

567 students (303 boys) from five Japanese junior high schools, three time points over two years, aged 12-13 at baseline

Japanese junior-high-school students aged 12-15

Hybrid (Mundlak) regression decomposing the fitness-achievement association into a WITHIN-PERSON component (does a student's own change in fitness track their own change in grades?) and a BETWEEN-PERSON component (do fitter students have better grades than less fit students?). Multiple imputation for attrition. This is the closest thing in the literature to a confound-breaking observational design for fitness and achievement: the within-person estimate absorbs every time-invariant characteristic of the child, INCLUDING genotype and stable family environment. Graded C, not B, because the outcome is teacher-assigned grade point average rather than an independent standardized test, the fitness measure is an eight-test composite dominated by strength and speed items rather than aerobic capacity, and time-varying confounders (puberty, effort, health) are unaddressed. It is not a within-family design and does not identify a causal effect.

The nearest available genetically-attenuated test of fitness as an input, and it is sex-specific and confined to teacher grades. Within-person change in the fitness composite tracked within-person change in GPA for BOYS only; for GIRLS no association appeared on either the within- or between-person component. A real biological input to learning that operates through fitness should not be absent in half the population. The result is the pattern one expects if a shared time-varying driver (adolescent effort, engagement, health) moves fitness and grades together in some students. Note also that this design searched for what the whole observational fitness-achievement literature claims and recovered only a fragment of it.

This is the reason to record the study. Within-person fixed effects remove all time-invariant genetic confounding by construction. When that adjustment is applied, the large, uniform fitness-achievement association reported by cross-sectional reviews (Marques et al. 2018) survives only in boys and only on teacher grades. NO within-family, discordant-twin or sibling-fixed-effects study of fitness or physical activity on academic achievement in children was located in this survey. That absence is itself a finding: the single most-confounded claim in this literature has never been tested with the design that would settle it.