Abstract
BACKGROUND: Food literacy (FL) comprises the knowledge, skills, and motivation needed for food production, selection, preparation, intake, and waste management. This study examined whether socio-ecological correlates of FL differ across settlement contexts in China. METHODS: Cross-sectional survey data from Chinese adults (N = 1145) were analyzed across four settlement tiers: tier-1 metropolitan cities (R1), provincial/secondary cities (R2), smaller prefecture-level cities (R3), and county/rural areas (R4). General linear models estimated associations between socio-ecological predictors and overall FL after adjustment for sociodemographics, health behaviors, chronic disease, and BMI. Significant interactions were probed using HC3-robust simple slopes and pairwise slope contrasts. Robustness checks included domain-specific measurement invariance, variance inflation factor (VIF) diagnostics, and a regional sensitivity analysis. RESULTS: The fully adjusted model explained substantial variance in FL (R(2) = 0.629). Awareness showed the strongest association with FL, followed by family support, injunctive norms, and social norms. Moderation was modest and predictor-specific: dining preferences and family support were positively associated with FL across all regions, with the strongest effects in county/rural areas. Although the omnibus interaction for injunctive norms was statistically significant, follow-up slope contrasts were not, indicating limited substantive regional heterogeneity. Component analyses indicated that preference-related heterogeneity was concentrated in food intake and food choices/selection, whereas family-support heterogeneity was most pronounced for waste disposal. Domain-level invariance analyses supported broad cross-regional comparability of the FL structure, VIFs were all below 5, and the regional distribution of valid and invalid responses did not differ significantly. CONCLUSIONS: Socio-ecological correlates of FL were broadly robust across China, with limited context-specific variation driven mainly by stronger household-support effects in county/rural settings. These findings support region-sensitive FL strategies that strengthen household-based support while leveraging normative influences across regions.