The association between reported artificial light and reported physical activity at night: the mediating roles of environmental restorativeness and perceived safety

Artificial light at night may facilitate physical activity after sunset. To investigate this possibility, we conducted a survey with 646 Chinese adults to examine the associations between both subjectively and objectively measured artificial light at night and various intensities and domains of nighttime physical activity. We also explored the potential mediating roles of environmental restorativeness and perceived safety in these associations. Data were collected via an online questionnaire based on participants’ self-reported information. Through natural language analysis, we identified the reasons participants chose to engage in nighttime physical activity. We applied Spearman-based pairwise and network analyses to examine relationships among the variables, followed by generalized additive models (GAMs) and linear regression to explore potential patterns between artificial light at night and other variables. For associations showing near-linear trends, we conducted mediation analyses using saturated models. Our results revealed that over 70% of participants engaged in nighttime physical activity, mainly due to limited availability during the day. Across all statistical approaches, objective artificial light at night showed no significant association with nighttime physical activity. In contrast, subjective artificial light at night, when rated within a certain range (≥4), was positively associated with perceived safety, environmental restorativeness, and nighttime physical activity. Within this rating range, environmental restorativeness played a central mediating role, largely explaining the relationship between subjective artificial light at night and all forms of nighttime physical activity. The mediating role of perceived safety was limited. It is important to note that our analyses were based on cross-sectional data and guided by theoretical frameworks; thus, causal relationships among the variables cannot be inferred.