Research Square · Oct 19, 2026
Completed education and reported fitness/leisure nonparticipation in 2022 among linked CFPS respondents reporting participation in 2020: a preregistered analysis
By Tao Z, Li R
Background Educational differences in fitness/leisure participation may reflect unequal access to time, resources, and supportive environments. Among adults reporting participation in 2020, we examined whether a 2022 report of no participation differed by completed education. The same-item responses were interpreted as a two-wave report pair, not exact cessation. Methods This preregistered observational analysis used linked 2020 and 2022 China Family Panel Studies data. Eligible respondents were aged ≥ 25 years, directly self-reported in both waves, had complete mandatory data, and reported any past-12-month fitness/leisure participation in 2020. Original, non-imputed completed education entered a prespecified linear model. An unweighted modified-Poisson model with covariance clustered by baseline primary sampling unit (PSU) standardized the probability of a 2022 “never” report at 6 and 12 education years. No population weighting or attrition correction was applied. Results Among 4,639 respondents in 176 PSUs, model-standardized probabilities of a 2022 “never” report were 47.2% (95% confidence interval [CI], 44.2%–50.1%) at 6 education years and 28.8% (95% CI, 26.9%–30.7%) at 12 years. The absolute probability difference (12 minus 6 years) was − 18.4 percentage points (95% CI, − 21.1 to − 15.6; P<.001), and the complementary probability ratio was 0.610 (95% CI, 0.569–0.654). Logistic-link and cluster/influence checks were directionally concordant. Five prespecified analyses stopped under the preregistered numerical rule; robustness to the age threshold, education functional form, baseline frequency, and four-state specification remains unestablished. Conclusions In this selected linked analytic sample, the model-standardized probability of reporting no fitness/leisure participation in 2022 was lower at 12 than at 6 completed education years. The finding supports population-representative investigation of modifiable barriers among adults with fewer completed years of education. The association is noncausal and not population-representative; the report pair does not measure total physical activity or establish exact cessation.