Abstract
BACKGROUND: Rural ethnic minority children in Southwest China face unique developmental challenges, particularly regarding social-emotional competence (SEC). While environmental psychology suggests nature connectedness (NC) promotes socio-emotional development, its mechanisms within culturally distinct groups like the Yi ethnic minority remain unexplored. This study aims to explore the relationship between NC and SEC and examine evidence for the mediating role of empathy among upper-grade Yi pupils. METHODS: A cross-sectional survey was conducted with 647 Yi ethnic pupils (Grades 4-6) from rural Liangshan, China. Participants completed Chinese versions that have demonstrated acceptable psychometric properties of the Nature Connectedness Scale, the Delaware Social-Emotional Competence Scale, and the Basic Empathy Scale. Data analysis employed Pearson correlations, t-tests, ANOVA, and Hayes' PROCESS macro (Model 4, 5,000 bootstraps) with gender as a covariate. RESULTS: NC significantly positively predicted SEC (β = 0.28, p < 0.001). NC significantly positively predicted empathy (β = 0.20, p < 0.001). Empathy significantly positively predicted SEC (β = 0.27, p < 0.001). Empathy showed a partial mediating effect in the NC-SEC relationship (indirect effect = 0.04, 95% CI[0.02, 0.07]), accounting for 18.98% of the total effect. Significant gender differences emerged, with girls scoring higher on NC, empathy, and SEC (p < 0.001). No significant grade differences were found. CONCLUSION: This study provides evidence supporting empathy as a significant partial mediator linking nature connectedness to social-emotional competence among rural Yi children. Findings provide empirical support for the pathway within a culturally distinct context, highlighting the synergistic potential of leveraging ethnic ecological-cultural resources (e.g., nature-based rituals, traditional practices) for fostering socio-emotional development in marginalized populations. Future research should utilize longitudinal designs and incorporate multi-informant data.