Fitting a Structural Equation Model to Community-Based Data to Explain the Mechanism Driving the Effects of Urban Green Spaces on Human Health

Abstract

A theoretical meta-model was recently proposed to explain the mechanism driving the effects of Urban Green Space (UGS) on human health (Zhang et al. 2017: Journal of Urban Ecology, 1–13). On one hand, this meta-model suggests an indirect effect of UGS Provision (measured as quantity, quality or accessibility of UGS) through UGS Exposure (measured as visit frequency to UGS, duration of visit or intensity of activities taking place during the visit). On the other hand, UGS Provision may have an indirect effect on Exposure mediated by people’s perception of UGS, and UGS Exposure may influence also indirectly human Health but mediated by human motivation to use UGS. We tested these different expectations by fitting 12 structural equation models (SEMs) to data collected on these variables through semi-structured interviews. These 12 SEMs correspond to four different scenarios depending on how Provision was approximated. We found that any of the 12 SEMs can be used to explain the relationships between UGS and health condition (for all 12 SEMs, P>0.05). The SEMs analysis further show that only model 2 (where Provision is approximated as UGS quantity, and Exposure as duration), model 3 (where Provision is approximated as again quantity, and Exposure as intensity) and model 12 (where Provision is approximated as distance of the closest UGS from people’s house, and Exposure as intensity) are equally the best of all 12 models tested [Δ(models2,3,12) < 3]. However, apart from model 2 that has no significant path, models 3 and 12 have the same significant path (motivation ~ intensity; β=7.86±2.03, P=0.0002), suggesting that people who visit UGS may be motivated by opportunities of physical activities offered by UGS. When looking at scenario per scenario, this significant path still remains the same except for scenario 2 where the significant path is motivation ~ quality (β=-2.47±0.82, P=0.003), suggesting that the quality of UGS may also be an important motivation for a visit to UGS. In all scenarios, the best model is always the model where Exposure is measured as intensity irrespective of how provision is approximated. This suggests that it is not only UGS provision that matters most, but rather intensity, i.e. the type of activities people engage in when they visit UGSs. Overall, this study supports the theoretical meta-model tested, and we call for more questionnaire-based data collections at different scales to further clarify the mechanism driving UGS-health relationship in different contexts.