The survey was conducted from November 1 to December 20, 2025, at five higher education institutions in different regions of Province A, China. The researchers used stratified sampling, randomly selecting one institution from each of five regions. Of 996 submitted questionnaires, 249 were excluded because of completion times under three minutes, incomplete responses or patterned responses, leaving 749 valid responses and a valid response rate of 75.2%. The final sample included 301 male students, or 40.2%, and 448 female students, or 59.8%. An a priori power analysis for the chained mediation model estimated a minimum sample of 99 participants under an effect size of Cohen’s f² = 0.15, significance level of 0.05 and statistical power of 0.90.

Participants completed four self-administered instruments. The 17-item Social Support Scale assessed subjective support, objective support and use of support. Physical activity was measured with the three-item Physical Activity Rating Scale-3, which combined exercise intensity, duration per session and frequency on five-point scales; scores of 19 or less indicated low activity, 20 to 42 moderate activity and 43 or more high activity. A 25-item Peer Attachment Scale assessed relationships with peers, and a nine-item Exercise Self-Efficacy scale assessed perceived ability to exercise in challenging conditions such as bad weather or fatigue. Cronbach’s alpha values were 0.948 for social support, 0.906 for peer attachment, 0.944 for self-efficacy and 0.705 for physical activity.

All four constructs showed statistically significant positive associations. Social support correlated with physical activity at r = 0.341, peer relationships at r = 0.621 and exercise self-efficacy at r = 0.335. Physical activity also correlated with peer relationships at r = 0.403 and self-efficacy at r = 0.366, while peer relationships correlated with self-efficacy at r = 0.330; all reported correlations had p < 0.01. Mean scores were 4.031 for social support, 2.882 for physical activity, 3.676 for peer attachment and 6.401 for exercise self-efficacy. The authors regarded social support and peer attachment as relatively strong, while physical activity and self-efficacy remained lower, but the PARS-3 scores were not direct measurements of weekly exercise minutes or days.

Hierarchical regression examined how the variables related to physical activity after accounting for gender and year of study. A model containing only those two characteristics explained 12.2% of the variance in physical activity. Adding social support increased the explained variance to 20.8%, adding peer relationships raised it to 26.3%, and adding exercise self-efficacy brought it to 28.4%. In the final model, social support retained a positive standardized coefficient of 0.088 after adjustment for peer relationships and self-efficacy, p < 0.05. Peer relationships had the largest corresponding coefficient, 0.268, followed by exercise self-efficacy at 0.163; both were significant at p < 0.01. Year of study was not significant, whereas male students had higher physical activity scores than female students, with a final standardized coefficient of 0.245.

The chained mediation analysis used Hayes PROCESS Model 6 with 5,000 bootstrap repetitions and controlled for gender and academic year. The estimated total effect of social support on physical activity was B = 0.351, with a fully standardized coefficient of 0.301 and a bootstrapped 95% confidence interval from 0.265 to 0.436. After peer relationships and exercise self-efficacy were included, the estimated direct effect was B = 0.103, with a standardized coefficient of 0.088 and a 95% confidence interval from 0.004 to 0.204. The total indirect effect was B = 0.248, with a standardized coefficient of 0.212 and a 95% confidence interval from 0.182 to 0.317. The confidence intervals for all three specific indirect paths excluded zero, supporting their presence in the statistical model.

Peer relationships alone formed the largest indirect pathway. The association from social support to physical activity through peer attachment was B = 0.194, with a fully standardized coefficient of 0.166 and a bootstrapped 95% confidence interval from 0.132 to 0.257. The path through exercise self-efficacy alone was smaller at B = 0.030, with a standardized coefficient of 0.026 and a 95% confidence interval from 0.011 to 0.055. The sequential path from social support through peer relationships and exercise self-efficacy to physical activity yielded B = 0.023, a standardized coefficient of 0.020 and a 95% confidence interval from 0.011 to 0.038. In this model, the immediate relational context of support was more strongly associated with physical activity than either self-efficacy alone or the complete sequence.

The observed pattern is compatible with the social cognitive framework used by the researchers, in which environmental resources can relate to behavior both directly and through cognition and interpersonal interaction. Encouragement, information and practical assistance may create opportunities to exercise, while peers can provide invitations, role models and feedback. Joint participation may also supply experiences that help students judge their ability to exercise despite fatigue, time constraints or unfavorable conditions. The researchers described this as an “interpersonal interaction, cognitive restructuring, behavioral practice” mechanism. However, the survey did not directly observe joint exercise, role modeling or changes in confidence, and it did not demonstrate that social support caused students to become more active or more self-efficacious.

Demographic and methodological findings further constrain the interpretation. Men reported higher scores than women for every main variable, with moderate differences in physical activity and exercise self-efficacy and smaller differences in social support and peer attachment. Academic year was associated with small differences in social support, peer relationships and self-efficacy, but not physical activity; the authors reported that first-year students scored higher than second-year students on the first three measures, although the effect sizes were small. Statistical checks identified heteroscedasticity in the peer-relationship and physical-activity equations and some evidence of nonlinearity in the self-efficacy equation. The researchers used bootstrapped confidence intervals to address uncertainty in the indirect-effect estimates, but these procedures cannot eliminate confounding or establish temporal order.

The central limitation is that the cross-sectional design cannot determine whether social support preceded peer relationships, self-efficacy or physical activity. Recall bias, social desirability and common-method bias are possible because all four constructs were reported through self-administered questionnaires at a single point in time. Harman’s single-factor test extracted eight factors with eigenvalues above one; the first explained 35.673% of the variance, below the authors’ 40% threshold, but that test cannot rule out shared-method effects. PARS-3 also measured exercise through intensity, duration and frequency rather than total daily movement, excluding domains such as work, commuting and housework. The sample came from only five institutions in one anonymized province, and information on institution type, field of study, urban or rural background and socioeconomic status was not systematically collected. The findings therefore identify an interconnected network of associations among social environments, peer ties, confidence and student activity, rather than proving a causal program for increasing exercise.

Subject of Research: Cross-sectional mediation analysis of social support, peer relationships, exercise self-efficacy and physical activity among university students in China.

Article Title: The relationship between social support and physical activity among college students: the chain mediation of peer relationships and exercise self-efficacy

Article References: Chen, J., & Zhou, W. (2026). The relationship between social support and physical activity among college students: the chain mediation of peer relationships and exercise self-efficacy. Frontiers in Psychology, 17, Article 1873443. https://doi.org/10.3389/fpsyg.2026.1873443

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1873443

Keywords: physical activity, university students, social support, peer relationships, exercise self-efficacy, mediation analysis, cross-sectional survey, China, social cognitive theory, relationship, between, social