The data for this study were derived from the 2020 National Fitness Survey (22), which covered 31 provinces (autonomous regions and municipalities) across China and included survey responses from adults aged 20 to 59 years. The survey was conducted from January 1 to March 31, 2020. A three-stage probability proportional to size (PPS) sampling method was employed. In the first stage, a total of 471 county-level units were selected across the 31 provinces, with the number of sampled counties in each province determined by provincial population size. In the second stage, 13 village or neighbourhood committees were randomly selected within each sampled county, resulting in 5760 sampled village units. In the third stage, 12 adults aged 20 to 59 years were randomly selected from each village or neighbourhood committee using local household registration lists. A total of 59,747 samples were collected.

Prior to data collection, a standardized training session was conducted for technical supervisors from all 31 provinces (autonomous regions and municipalities). These supervisors then provided secondary training to the interviewers in their respective regions to ensure a consistent understanding of the survey protocol and research objectives, thereby minimizing interviewer bias. During the formal survey, participants were sampled based on information provided by local statistical bureaus. Face-to-face interviews were conducted in households. Before each interview, the purpose and significance of the study were explained to the participants, and written informed consent was obtained. To protect privacy, all personal identifiers were anonymized using a three-level coding system. The study was approved by the Ethics Committee of the China Institute of Sport Science (Approval No. CISSLA-20191029).

Normality of variables was assessed using the Kolmogorov–Smirnov test. Continuous variables are presented as mean ± standard deviation (X ± SD), and categorical variables as frequencies and percentages (n, %). Sample weights were calculated and applied based on the age-specific and urban–rural distribution data of the 20–59 years old population from the 2020 Seventh National Population Census to enhance the representativeness of the sample. Group differences were examined using one-way ANOVA or Chi-square test. Harman’s single-factor test was used to check for common method bias. Pearson correlation analysis was used to examine bivariate relationships, and multiple linear regression was employed to analyze influencing factors. Model assumptions including linearity, normality of residuals, and homoscedasticity were examined and met. Mediation effects were tested using the SPSS macro-PROCESS 4.0 (29) with bias-corrected bootstrapping (5,000 resamples). A 95% confidence interval not including zero indicated a significant mediation effect. Subgroup regression analyses were conducted by gender, residence, and age group to explore potential differences in the direct effect of family SES on MVPA. For all regression analyses, 95 percent confidence intervals were reported and used together with p-values to assess the statistical significance of effects. All analyses were performed using SPSS version 30.0 (IBM Corp, Armonk, NY, USA), with a p-value < 0.05 considered statistically significant.

The final analytical sample consisted of 55,804 valid respondents. Among them, 28,427 (50.9%) were male and 27,377 (49.1%) were female. Regarding residence, 36,256 (65.0%) lived in urban areas and 19,548 (35.0%) in rural areas. Significant differences were observed across age groups for all variables except gender (all p < 0.001) (Table 1).

Correlation analyses (Table 2) showed that family SES was positively correlated with EE (r = 0.314, p < 0.01), EM (r = 0.104, p < 0.01), and MVPA (r = 0.053, p < 0.01), indicating that adults with higher family SES reported better exercise environments, stronger motivation, and slightly higher MVPA. EE was positively correlated with both EM (r = 0.168, p < 0.01) and MVPA (r = 0.115, p < 0.01). A positive correlation was also found between EM and MVPA (r = 0.190, p < 0.01), suggesting that individuals with stronger motivation tend to engage in higher MVPA. With the exception of the non-significant correlation between EM and Residence, all four main variables were significantly correlated with the control variables (p < 0.01). Age was positively correlated with MVPA (r = 0.076, p < 0.01), while all other correlations between the main variables and control variables were negative (p < 0.01).

A series of multiple linear regression models using a stepwise approach were conducted to examine the relationships between family SES (independent variable), MVPA (dependent variable), EE and EM (mediating variables), while controlling for gender, age, and residence. To indicate model explanatory power, R² values for model 1 ~ 4 ranged from 0.009 ~ 0.052 (Table 3). In Model 1, age was positively associated with MVPA (β = 0.074, 95%CI = 0.146 ~ 0.182, p < 0.01), whereas gender and residence (gender, β = −0.023, 95%CI = −1.491 ~ −0.691, p < 0.01, residence, β = −0.050, 95%CI = −2.964 ~ −2.126, p < 0.01) showed negative associations. In Model 2, after adding family SES, age remained positively associated with MVPA (β = 0.089, 95%CI = p < 0.01), while gender and residence (gender, β = −0.018, 95%CI = −1.291 ~ −0.491, p < 0.01, residence, β = −0.031, 95%CI = −1.988 ~ −1.104, p < 0.01) remained negatively associated. Family SES was positively associated with MVPA (β = 0.063, 95%CI = 0.910 ~ 1.213, p < 0.01), indicating that higher family SES was associated with higher MVPA levels. In Model 3, after further adding EM, the relationships for the control variables remained consistent. Both family SES (β = 0.033, 95%CI = 0.405 ~ 0.717, p < 0.01) and EM (β = 0.108, 95%CI = 0.871 ~ 1.023, p < 0.01) were positively associated with MVPA, suggesting that adults with stronger motivation accumulated more MVPA. In Model 4, after adding EE, the control variables again showed consistent relationships. Family SES (β = 0.022, 95%CI = 0.215 ~ 0.522, p < 0.01), EE (β = 0.081, 95%CI = 0.631 ~ 0.783, p < 0.01), and EM (β = 0.176, 95%CI = 0.576 ~ 0.633, p < 0.01) were all positively associated with MVPA, showing that better facility availability and higher motivation each contributed to higher MVPA.

Mediation analysis was performed using the SPSS macro-PROCESS 4.0. The results (Table 4) indicated significant indirect effects of SES on MVPA, as the bias-corrected confidence intervals for all paths did not include zero. This confirms the significant mediating roles of the two proposed mediators. The total indirect effect was comprised of three specific paths: (1) SES → EE → MVPA, with an effect value of 0.439 (95% CI = 0.387 ~ 0.494); (2) SES → EM → MVPA, with an effect value of 0.168 (95% CI = 0.141 ~ 0.195); and (3) SES → EE → EM → MVPA, with an effect value of 0.170 (95% CI = 0.128 ~ 0.151). All three pathways indicated positive associations, meaning that higher SES was linked to better EE, stronger EM, and subsequently higher MVPA. The contribution rates of these three paths were 48.67, 18.63, and 15.52%, respectively (Table 4; Figure 2).

To further explore the relationship between family SES and MVPA, stratified regression analyses were conducted by age group, gender, and residence, adjusting for confounding factors (Table 5). The results showed that family SES was significantly and positively associated with MVPA across all age groups. The strongest association was observed in the 20–29 age group (β = 0.090, 95%CI = 1.139 ~ 1.726, p < 0.01), followed by the 50–59 group (β = 0.081, 95%CI = 1.410 ~ 2.101, p < 0.01), the 30–39 group (β = 0.076, 95%CI = 0.854 ~ 1.311, p < 0.01), and the 40–49 group (β = 0.059, 95%CI = 0.731 ~ 1.282, p < 0.01). In the gender stratification, the association between family SES and MVPA was more pronounced in males (β = 0.089, 95%CI = 1.315 ~ 1.709, p < 0.01) compared to females (β = 0.014, 95%CI = 0.033 ~ 0.430, p < 0.05). Regarding residence, SES was positively associated with MVPA in both urban (β = 0.038, 95%CI = 0.469 ~ 0.814, p < 0.01) and rural (β = 0.043, 95%CI = 0.592 ~ 1.160, p < 0.01) populations. These findings suggest that the promotive effect of family SES on physical activity is not homogeneous but is moderated by age, gender, and place of residence. These findings suggest that the association between family SES and MVPA varies by age, gender, and residence.

In this study, family SES demonstrated a significant positive association with MVPA (accounting for 82.82% of the total effect). This pattern may reflect several interrelated mechanisms. Adults with higher family SES generally possess higher levels of education, income, and occupational prestige (32). (1) Economic advantages can provide greater access to activity-supportive environments and opportunities, which may facilitate stronger autonomous motivation and lower participation barriers. Adults with higher family SES also tend to have better access to health information and engage more in health-promoting behaviors, which are associated with higher MVPA levels. Households with higher family SES often reside in neighborhoods with more accessible exercise facilities (33), a factor shown to support greater physical activity participation (34). Conversely, residents with lower family SES may face constraints that limit their use of available facilities (35). (2) Higher educational levels may enhance individuals’ understanding of health-related information (36) and support competence related beliefs that facilitate the internalization of MVPA values, consistent with SDT (37). This may help individuals appreciate the long-term value of MVPA and support more stable patterns of participation. These psychosocial mechanisms are associated with more consistent engagement in MVPA (38). (3) Higher occupational level may also be linked to broader social capital and health-supportive norms (39), which can influence adults’ opportunities and readiness to engage in MVPA. Adults with higher family SES may also have more stable careers and greater flexibility in leisure time allocation, which has been associated with higher MVPA participation (40). Overall, these factors interact through improved health literacy, access to supportive environments and motivational processes, contributing to higher MVPA.

The interaction effects in this study revealed a non-linear trend where the influence of family SES on MVPA first weakened and then strengthened with age. During young and middle adulthood, increased work and caregiving demands may limit discretionary time for PA. Even with high family SES, its translation into physical activity behavior is hindered, thus weakening its positive effect (41). From age 50 onwards, as family responsibilities diminish and health risks become more apparent, individuals with high family SES are more likely to translate health cognition into active behavior, thereby strengthening the influence of family SES (42). The SES-MVPA association was stronger among men, who may have more leisure time available for structured PA (43). The association was also stronger in rural areas, where more accessible outdoor spaces and lower participation costs may support PA participation (44).

This study found that family SES was positively associated with EE among Chinese adults, a result consistent with findings from other countries (7). Adults with higher family SES often reside in neighbourhoods with greater access to sports and recreational resources. Higher family SES groups may also have greater access to health information and resources that support perceived competence and autonomy, thereby facilitating more autonomous motivation under SDT (45). In contrast, adults with lower family SES may experience greater barriers to accessing health information and PA opportunities, which can make it more challenging to develop autonomous motivation and consistent activity habits. These differences may contribute to the socioeconomic gradients observed in motivation and MVPA (46). Furthermore, this study also confirmed a positive correlation between EE and adult MVPA, consistent with prior research showing that proximity and accessibility of facilities support regular physical activity (47). Supportive physical environments can foster more enjoyable and competence enhancing experiences, which are fundamental to autonomous motivation in SDT (14). When adults perceive their environment as safe, convenient, and supportive, they are more likely to engage in positive activity experiences (48). Positive experiences within supportive environments may reinforce perceived competence and enjoyment, further strengthening autonomous motivation (49). This process may support more stable and sustained engagement in PA. Particularly in the context of the widespread sedentary lifestyle in contemporary China (50), supporting autonomous motivation remains an important psychological factor associated with adults’ willingness to engage in physical activity.

This study found that family SES was positively associated with EM among Chinese adults, consistent with results from other countries (51, 52). Adults with higher family SES often have greater access to health information and activity supportive environments, which may help foster stronger autonomous forms of motivation. In the SDT, autonomous forms of motivation such as engaging in MVPA because it is enjoyable or personally meaningful reflects higher perceived autonomy and competence. EM was also positively associated with adult MVPA in this study, consistent with evidence that individuals who experience great enjoyment and value in MVPA are more likely to maintain regular participation (53). These findings suggest that family SES difference in access to supportive conditions may contribute to variation in autonomous motivation, which is in turn associated with MVPA.

The chain mediation model constructed in this study (family SES → EE → EM → MVPA) illustrates the roles of EE and EM in the association between family SES and adult MVPA. Specifically, EE represents an environmental pathway through which socioeconomic resources may relate to variation in MVPA. Supportive environments may foster greater perceived autonomy, competence, and enjoyment, which are central components of autonomous motivation in SDT. These motivational processes are in turn associated with higher MVPA. This suggests that family SES may relate to physical activity both through access to supportive environments and through psychological mechanisms linked to autonomous motivation. As mentioned, higher family SES is often associated with more supportive physical environments (18). Research indicates that characteristics such as accessibility, convenience, and perceived safety can support more positive activity experiences, which are linked to higher motivation (54). From an SDT perspective, environments that support autonomy and competence can facilitate the internalization of motivation, which is associated with more consistent MVPA participation (55). Supportive social environments, such as positive neighborhoods interactions and available community resources, have also been associated with higher perceived support for physical activity, which may relate to stronger motivation (56–58). Adults in such environments may find it easier to engage in activity that aligns with their motivational preferences, thereby supporting more regular participation. These combined processes are associated with more regular participation in physical activity.

These findings offer practical directions for MVPA promotion. Improving access to safe and affordable facilities may help reduce family SES-related disparities in MVPA, and combining such environmental improvements with strategies that strengthen autonomous motivation—especially among young adults—may enhance effectiveness. The mediation proportions observed here are broadly consistent with prior work showing that environmental and motivational pathways explain part of socioeconomic differences in PA (11, 14, 18). From a socio-ecological and self-determination theory perspective, these results suggest that interventions should address both opportunity and motivation. Future studies using longitudinal or objective PA measures are needed to verify these pathways.

This study has several limitations. First, MVPA and EM were measured using self-reported questionnaires, which may introduce recall and social desirability biases. Second, the cross-sectional design precludes any causal inference, and potential endogeneity cannot be ruled out. Third, the sample included only adults aged 20–59 years, which limits the generalisability of the findings to youth and older adults. Fourth, although Harman’s test indicated no substantial method bias, this approach is limited in its ability to detect subtle forms of bias. Finally, the survey was conducted before the COVID-19 pandemic, and subsequent shifts in mobility patterns and use of public spaces may affect the applicability of the findings to current conditions.