The longitudinal relationship between physical activity and social skills in children with autism: Basic...

To examine the longitudinal association between physical activity and social skills in children with autism and evaluate basic psychological need satisfaction in exercise as a potential longitudinal indirect pathway.

A three-wave longitudinal study was conducted in seven special education schools in Sichuan, Hunan, and Yunnan, China. Data were collected in November 2025, February 2026, and May 2026 at three-month intervals. Primary caregivers completed all questionnaires. After excluding attrition cases and invalid responses, 584 valid cases were retained. Physical activity, basic psychological need satisfaction in exercise, and social skills were assessed using validated Chinese scales. Measurement invariance testing and fixed-effect cross-lagged panel modeling were supplemented by a longitudinal latent-variable model, demographic- and physical-activity-scaling sensitivity analyses, and random-effects cross-lagged panel modeling.

The preliminary bivariate CLPM showed inadequate global fit; therefore, its path coefficients were not used for substantive inference. In the primary three-variable longitudinal model, physical activity showed consistent positive prospective associations with social skills across both adjacent intervals (β = 0.274 and 0.169, respectively; both p < 0.001). T1 physical activity was positively associated with T2 basic psychological need satisfaction in exercise (β = 0.105, p = 0.002), which was subsequently associated with higher T3 social skills (β = 0.188, p < 0.001). This pathway yielded a small positive standardized indirect effect (effect = 0.020, 95% CI [0.007, 0.039], p = 0.015). The directly estimated association between T1 physical activity and T3 social skills also remained positive (effect = 0.137, 95% CI [0.068, 0.213], p < 0.001), indicating that basic psychological need satisfaction in exercise accounted for part, but not all, of the longitudinal association. The focal indirect effect was replicated in the longitudinal latent-variable, covariate-adjusted, and alternative physical-activity-scoring analyses, with estimates ranging from 0.022 to 0.029. The targeted random-effects model yielded a comparable point estimate (0.022), although with greater uncertainty (95% CI [-0.012, 0.056]).

These findings position physical activity as a recurring developmental context associated with later social-skill development in children with autism and indicate that basic psychological need satisfaction in exercise may modestly help explain this longitudinal relationship. They further suggest that the social-developmental relevance of physical activity may depend partly on whether participation supports autonomy, competence, and relatedness, underscoring the importance of considering the psychological quality of participation alongside activity volume.