Socioeconomic differentiation in sport engagement: frequent exercise, live sport attendance, and their overlap in the 2023 Chinese General Social Survey.
Lu C, Dai H, Xu J • Frontiers in public health • 2026
In the CGSS2023 analytical sample, socioeconomic patterning differed across combinations of frequent leisure-time exercise and live sport attendance: education was positively associated with active-only and dual engagement across weighting specifications, income associations varied with model specification, and evidence for urban-rural differences depended on weighting.
Key Findings
Results
Education showed the most consistent positive association with sport engagement categories, with the largest relative risk ratio for dual engagement (frequent exercise plus live attendance).
In the primary unweighted multinomial model, the relative risk ratio for dual engagement was 1.185 (95% CI 1.148-1.224).
Education was positively associated with active-only and dual engagement across the weighting specifications examined.
The sample comprised 5,585 adults from the 2023 Chinese General Social Survey.
Results
Urban residence was positively associated with dual engagement in the primary unweighted model, but this was not robust to weighting or selection adjustment.
Primary unweighted model: relative risk ratio 2.287 (95% CI 1.753-2.985).
Corresponding estimates were 1.412 in the combined-weight model and 1.458 in the selection-adjusted model.
None of the three urban-rural contrasts was statistically significant in either the combined-weight or selection-adjusted model.
Evidence for urban-rural differences therefore depended on weighting.
Results
Income associations with exercise and attendance categories were not stable and varied with analytic choices.
Associations varied with income coding, work-status adjustment, weighting, and the attendance threshold used in the outcome definition.
Sensitivity analyses varied weighting, sample composition and outcome definitions.
Results
Subjective social status showed only modest associations with the sport engagement categories.
The abstract describes associations with subjective social status as 'modest'.
Status was modeled alongside education, income and residence in covariate-adjusted multinomial models.
Results
The education-by-residence interaction was statistically significant after multiplicity adjustment in the unweighted linear-score model but not in the categorical interaction under combined weighting.
Interaction tests were exploratory and adjusted for multiple comparisons.
The categorical interaction was not statistically significant under combined weighting.
Methods
Frequent exercise and live sport attendance were cross-classified into four categories (low, active-only, spectator-only, dual engagement), and socioeconomic patterning differed across these combinations.
Frequent leisure-time exercise and any past-year live attendance defined the categories.
Weighted percentages described category distribution; covariate-adjusted associations were from an unweighted multinomial logistic model with robust standard errors.
The cross-sectional design limits causal inference.
Conclusions
The authors conclude that monitoring of sport engagement should distinguish exercise from live attendance and examine their overlap.
This approach can identify the forms of engagement in which socioeconomic differences are concentrated.
What This Means
This study used a national Chinese survey from 2023 covering 5,585 adults to look at who exercises often, who goes to live sporting events, who does both, and who does neither. It asked how education, income, self-rated social standing, and living in a city versus the countryside related to these four patterns.
Education stood out. People with more education were consistently more likely to exercise frequently and to both exercise and attend live sport, and this held across different ways of weighting the data. City residents appeared much more likely to do both in the main analysis, but that difference shrank and was no longer statistically significant when the data were weighted or adjusted for who was included in the sample. Income results changed depending on how income was measured and which other factors were considered, and self-rated social status had only small associations.
This research suggests that exercising and watching live sport are not the same thing, and that social differences may be concentrated in particular combinations of the two. Because the data come from a single point in time, the study cannot show that education causes more sport engagement. The authors suggest that tracking sport participation should look at exercise and attendance separately and together.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Lu C, Dai H, Xu J. (2026). Socioeconomic differentiation in sport engagement: frequent exercise, live sport attendance, and their overlap in the 2023 Chinese General Social Survey.. Frontiers in public health. https://doi.org/10.3389/fpubh.2026.1965653