Urban green spaces (UGS) are increasingly framed as key infrastructures for improving population well-being. However, in metropolitan regions marked by entrenched socio-spatial inequality, well-being associations linked to urban nature may not be equitably patterned. While a substantial body of research documents positive associations between contact with green spaces and psychological health, most studies rely on average-effect models, implicitly assuming that similar levels of engagement generate comparable returns across social groups. This leaves a critical gap regarding which groups show stronger well-being associations with UGS engagement and how place-based inequalities are reflected in engagement-to-well-being associations. This study examines whether engagement with UGS is associated with socially unequal psychological well-being patterns across gender, age, and race/ethnicity groups in the Metropolitan Region of São Paulo (MRSP), a large metropolis in the Global South. Drawing on cross-sectional self-reported survey data and employing multigroup Structural Equation Modeling, we estimate statistical associations in engagement-to-well-being pathways while accounting for socioeconomic position. Engagement was specified in the SEM as a behavioral pathway connecting perceived environmental characteristics, naturalness, infrastructure, and accessibility, to psychological well-being, measured through self-reported symptoms of stress, anxiety, and depression. In the covariate-adjusted H1 model, engagement was positively associated with psychological well-being in the total sample (standardized β = 0.292, SE = 0.030, Wald-type test statistic = 10.16, p < 0.001, 95% CI 0.244, 0.361). Group-specific coefficients were positive across most strata and were largest among middle-aged adults; however, a pairwise Wald contrast supported a statistically significant H1 contrast only between young and middle-aged adults (Wald-type test statistic = -2.63, p = 0.009). In the covariate-adjusted H2 model, engagement remained positively associated with well-being (standardized β = 0.398, SE = 0.063, Wald-type test statistic = 6.25, p < 0.001, 95% CI 0.269, 0.514). Naturalness showed the strongest indirect association with well-being through engagement (standardized indirect association = 0.319, bootstrap 95% CI 0.218, 0.434), whereas infrastructure and accessibility displayed smaller positive indirect associations (0.023 and 0.028, respectively). Pairwise contrasts in H2 were statistically supported for women versus men (Wald-type test statistic = 2.13, p = 0.033) and for young versus middle-aged adults (Wald-type test statistic = -2.14, p = 0.033), whereas racial contrasts were not statistically supported in the pairwise tests. These findings indicate that engagement-centered associations between UGS and psychological well-being are socially patterned, but the strongest evidence of heterogeneity in this sample is concentrated in gender and life-course differences rather than in definitive racial contrasts. Because the analysis is based on self-reported engagement and perceived UGS characteristics, the findings should be interpreted as conditional associational evidence rather than as evidence of causal effects, objective exposure differences, or optimal spatial design for UGS provision.
Silva et al. (2026) studied this question.