Perceived Spatial Environment and Outdoor Activity in Middle-Aged and Older Chinese Adults: A Cross-Sectional...

Background: Outdoor activity is essential for healthy aging, yet the affective, cognitive, and behavioral pathways linking perceived neighborhood environment to outdoor activity in middle-aged and older Chinese adults remain underexplored. Objective: This cross-sectional study examined associations between perceived spatial environment and outdoor activity, and tested five candidate mediators in a parallel-mediation framework, among community-dwelling adults aged ≥ 50 in Xiamen, China. Methods: Of 251 returned questionnaires, 99 cases with inconsistent response patterns and 12 high-leverage cases (Cook's D > 3/n) were excluded, yielding N = 140. Hierarchical regression and non-parametric bootstrap mediation (5000 resamples; bias-corrected 95% Cis) were conducted, with sensitivity analyses, Fornell-Larcker discriminant-validity assessment, and Harman's single-factor test for common-method bias. Results: Perceived spatial environment was strongly associated with outdoor activity (β = 0.96, p < 0.001). Bootstrap analyses identified four significant indirect pathways-place attachment (0.26 [0.13, 0.39]), perceived social environment (0.21 [0.11, 0.31]), service environment (0.18 [0.09, 0.28]), and attitudinal preferences (0.17 [0.05, 0.29])-whereas the a priori hypothesized behavioral mediator, social interaction, did not reach statistical significance (0.10 [-0.000, 0.205]). Sensitivity analyses confirmed coefficient stability across outlier-trimmed and untrimmed samples. Conclusions: Findings are consistent with an environment-behavior model in which affective and cognitive constructs-particularly place attachment-appear to be more proximal mediators than enacted social-interaction frequency. Given high inter-construct correlations (limited discriminant validity) and a self-selected community-active sample, results are best interpreted as hypothesis-generating and require replication with longitudinal and multi-method designs.