Introduction

Leisure-time physical activity (LTPA) is beneficial to human health. According to the World Health Organization (WHO), insufficient physical activity is one of the leading risk factors for global mortality, responsible for 6% of deaths globally. Approximately 27.5% of adults and 81% of adolescents do not meet the recommended levels of physical activity1. Increasing LTPA can reduce the risks associated with heart disease, hypertension, and cancer2,3,4. Studies have shown that a favorably objective built environment (OBE) can promote physical activity through features like mixed land use, ample green spaces, and favorable road accessibility and service facilities, which increase the frequency and duration of physical activities5,6,7. Meanwhile, perceived built environment (PBE), such as the perceived accessibility of service facilities including shopping centers, metros, and public spaces, has been significantly correlated with walking and cycling behaviors8,9,10. These insights underline the importance of considering both OBE and PBE in promoting LTPA.

In reviewing the literature on how the OBE and PBE influence LTPA, we find that discrepancies often exist between the two constructs7. The mechanisms driving this discrepancy are multifaceted. People rarely have adequate information and capacity to fully comprehend their surrounding built environment without bias11,12. Consequently, people often cannot make sound and economically optimal decisions regarding healthy behaviors based on perceived environmental information7,13. Another explanation suggests that perceptions of the same objective built environment can differ among individuals with varied socioeconomic, cognitive characteristics, and attitudes5,14, leading to differences in physical behavior choices.

While the impact of the built environment on physical activity is acknowledged, the complex interplay among the OBE, PBE, and LPTA merits detailed exploration5,7,14. A enlightening explanation is that the OBE can influence healthy behaviors through its impact on the PBE15,16. The Stimulus-Organism-Response (SOR) model provides the theoretical foundation for understanding the mediation role of PBE17. According to this model, environmental characteristics affect one’ s emotional response, which in turn influences their behavioral responses, such as walking or not18,19. Within the realm of environment-behavior studies, individuals’ awareness and perception of the objective built environment are shaped not only by real-world stimuli received through primary senses like sight and hearing20, but also by factors such as socioeconomic status, cognitive ability, and personal preferences5, which subsequently lead to varied behavioral responses among individuals10.

Empirical studies have explored the mediating role of PBE. Some studies confirm the path of environmental perception, showing that perceived attributes such as greenness, availability and accessibility of public facilities, residential density, and walkability can mediate the relationship between objective built environment features and physical activity7,10. For example, a study in Hong Kong using mobile phone and travel survey data found that perceived availability and accessibility of service facilities mediate the association between objective equivalents and the number of walking distance21. Similarly, a Spanish study suggested that perceived walkability significantly mediates the effect of objectively-measured walkability on walking7.

However, the mediation role of the PBE in mediating the effect of OBE on the LTPA remains poorly understood, necessitating further empirical investigations to elucidate this relationship clearly. Moreover, the existing literature originates from Western countries5,7,8, with scant evidences from developing context, especially in Asia. The considerable differences in urban morphology and socioeconomic factors between Chinese cities and their Western counterparts necessitate the context-specific investigations in future studies10,22.

To address the gaps mentioned above, the present study is to study investigate the mechanism underlying the association between the built environment and physical activity. Specifically, this work aims to answer the following questions: (1) Is there a mismatch between the OBE and PBE? (2) Do the OBE and PBE have different impacts on leisure-time physical activity? (3) Does the PBE mediate the association between OBE and leisure-time physical activity? Fig. 1 illustrates our conceptual framework, which is based on the Stimulus-Organism-Response model.

Figure 1
figure 1

Conceptual framework of the complex interactions between OBE, PBE and LITA.

Full size image

Methods

Sample

Between June and August 2017, we conducted an online survey in Fuzhou, the capital city of Fujian Province, China. As a typical medium-sized city in China, Fuzhou offers a diverse urban structure and favorable climate conducive to leisure-time physical activities. The survey covered Gulou, Taijiang, Cangshan, and parts of Jin'an Districts, which effectively represent the mixture of residential, commercial, and recreational spaces typical of such urban environments. Out of 2000 distributed questionnaires, 1712 were retrieved, and after excluding incomplete and out-of-area responses, 760 valid questionnaires remained. The statistical distributions of key variables showed consistency between the total and analytic samples. The study area and the location of respondents are shown in Fig. 2. For a detailed explanation of the survey methodology, please refer to Zhang et al.15.

Figure 2
figure 2

Research area.

Full size image

Measures

Objective built environment

We used ArcGIS to generate 500 m radius buffers around respondents' coordinates to assess nearby OBE. This buffer size is practical for everyday activities, representing an approximate 10-min walking distance.23,24,25 Additionally, many Chinese communities are designed with a radius of around 500 m, which aligns with the concept of community-life cycles for 10-min walking distances26,27,28. This approach aligns with methodologies used in similar studies28,29. We selected green spaces, recreational facilities, and catering services as key features of the built environment3,4. Green spaces offer natural settings for relaxation and exercise29, recreational facilities provide structured physical activity opportunities30, and catering services enhance the liveliness of public spaces31. These elements were quantified using points of interest (POIs) within each buffers. All relevant data were sourced from the Baidu Open Map Platform (http://map.baidu.com).

Perceived built environment

The assessment of the PBE was primarily conducted via questionnaires using a binary scale, where participants could respond with either 'yes' or 'no'. This study, approved by the Ethics Committee of the School of Architecture and Urban–rural Planning at Fuzhou University, adhered to the Declaration of Helsinki and all related ethical guidelines. Informed consent was secured from all individual participants and their legal guardians. The questionnaire covered the presence and quality of amenities such as green spaces, gyms, internet cafes, KTVs, bars, snack bars, and restaurants within a 500 m radius. The responses were averaged to quantify perceived green spaces, recreational facilities, and catering services, thereby aligning these measures with the objective built environment variables.

Leisure-time physical activity

LTPA was measured as a latent variable combining "weekday" and "weekend" leisure-time physical activity. Participants reported the daily duration of LTPA, characterized as purposeless activity distinct from commuting. The reliability of this measure was confirmed with a Cronbach's alpha coefficient of 0.856, exceeding the threshold of 0.8, indicating high internal consistency and suitability for assessing leisure-time physical activity.

Covariate variable

To account for personal factors influencing LTPA, we controlled for variables such as gender (male or female), age (actual numerical value), self-assessed income levels (ranging from lower to upper), educational attainment (bachelor's degree or higher), and exercise inclination (scored on a scale from 1 to 4).

Statistical analysis

Initially, we conducted descriptive statistics to elucidate the distribution characteristics of the sample. Subsequently, hotspot analysis identified spatial clustering of high and low values in both OBE and PBE, facilitating visualization of regions with built environmen32. Then, we performed Pearson correlation analysis on all endogenous and exogenous variables to evaluate potential multicollinearity concerns33. To evaluate the consistency between OBE and PBE, the Intraclass Correlation Coefficient (ICC) was calculated using a two-way fixed effects model7.

Structural Equation Modeling (SEM) was chosen for its ability to effectively analyze covariance structures and manage both latent and observed variables, making it particularly suitable for investigating the hypothesized interactions among variables in our study34. A hierarchical SEM was constructed to address the research questions previously outlined. Model 1 incorporated latent variables as independent variables. Model 2 combined OBE assessments with covariates. Model 3 incorporated PBE and covariates as independent variables. Model 4 employed PBE as a mediator between OBE and LTPA. To mitigate potential bias from residential self-selection, we analyzed samples from residents in work unit compounds and affordable housing35,36,37, where housing choices are often allocated, thus limiting self-selection effects38.

Each model controlled for PBE and LTPA using covariates, with robustness ensured by 5000 bootstrap replications. Mediating effects' significance was determined using z-values (> 1.96) and bias-corrected confidence intervals excluding zero. SEM was conducted using IBM SPSS Amos 26, with data cleaning and standardization via STATA version 17.1.

Ethics approval and consent to participate

The use of questionnaire data in this study was approved by the Ethics Committee of the School of Architecture and Urban–rural Planning at Fuzhou University. All procedures involving human participants were performed in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study and their legal guardians.

Results

Descriptive statistics

Table 1 provides the descriptive statistics for the study's variables, encompassing means, standard deviations, frequencies, and percentages. Notably, approximately 90% of respondents categorized their income as lower, relatively lower, or comparable to local income levels. Over 40% of the participants attained higher education degrees. Additionally, two-thirds of the participants demonstrated a strong affinity for sports, indicated by ratings of 'more love' or 'very love', reflecting a significant interest in physical activities. In terms of the OBE, the average respondent's 500m radius included 34.3 catering services, 2.58 green spaces, and 6.86 recreational facilities. Notably, the perceived ratings for recreational facilities (0.29) were substantially lower than those for catering services (0.67) and green spaces (0.61), suggesting a divergence from the objective measures of the OBE.

Table 1 Descriptive statistics.
Full size table

Figure 3 illustrates the spatial distribution of high and low values for both OBE and PBE. The analysis reveals a distinct ring-shaped configuration of objective catering and recreational facilities, concentrated around the core of Fuzhou city. The density of these facilities decreases with increasing distance from the city center. Additionally, the central area of Fuzhou exhibits significant clusters of objective green spaces, which are sparse in other regions. In the southeast region of Fuzhou, there is a perceived scarcity of functional green spaces, which different from the results of objective review. Our field investigations revealed that many parks in this area are either too small, poorly maintained, or have been repurposed. These conditions not only contribute to the parks being less valued by the local residents but also lead to their underutilization. Moreover, the clusters representing high perceived values for catering and recreational facilities are significantly smaller than those indicated by objective data, implying that residents might frequently undervalue the actual amenities available in their vicinity. Overall, PBE seems to be generally lower than OBE.

Figure 3
figure 3

Hot spot analysis.

Full size image

Figure 4 elaborates on the correlations between covariates and exogenous variables. The correlation coefficients among the OBE, PBE, and the five covariates are all below 0.7, indicating no significant collinearity and confirming the appropriateness of the model estimation parameters. The correlation analysis reveals significant positive relationships for the three pairs of OBE and PBE variables, suggesting a potential alignment between the objective and perceived built environments. Moreover, a robust correlation exists between the attributes of catering and recreational facilities across both objective and perceived dimensions, likely reflecting the common spatial clustering of these facilities near densely populated commercial centers39.

Figure 4
figure 4

Pearson correlation analysis.

Full size image

However, while the Pearson correlation coefficient quantifies the degree of linear relationship, it may not capture the full extent of consistency among variables. Therefore, the study further conducted Intraclass Correlation Coefficient analyses on the same pairs of variables. These analyses indicated that the ICC values for all pairs were below 0.3, demonstrating only modest consistency between the objective and perceived built environments. Detailed statistical results are provided in Table 2.

Table 2 Intraclass correlation coefficient analyses.
Full size table

Structural equation analysis with the full sample

Table 3 presents the estimated outcomes from four distinct models. Model 1b, which achieves a Goodness of Fit Index (GFI) of 0.988, surpasses Model 1c (GFI = 0.927) in predictive accuracy, suggesting that objective built environments more effectively predict residents' leisure-time physical activity than perceived built environments. Model 1d, which utilizes the complete dataset, establishes a robust final model (GFI = 0.938).

Table 3 Structural equation analysis with the full sample.
Full size table

Figure 5 summarizes the aggregate effects from Model 1d, illustrating that various elements of the neighborhood's objective built environments are significantly associated with LTPA. Interestingly, the number of catering services is inversely related to LTPA, whereas other objective built environment factors show positive influences, aligning with urban planning objectives. Furthermore, both perceived green spaces and recreational facilities significantly enhance LTPA, with perceived recreational facilities serving a crucial mediating role between objective recreational facilities and LTPA.

Figure 5
figure 5

Structural equation analysis with the full sample. Note: *Relationship is significant at the 0.05 level **Relationship is significant at the 0.01 level ***Relationship is significant at the 0.001 level. The dashed lines indicate insignificant hypothetical relationships.

Full size image

Specifically, the count of objective catering services has a negative impact on residents' leisure-time physical activity, as reflected by a regression coefficient of − 0.225, highlighting a detrimental influence. Notably, residents' perceptions of catering services do not significantly correlate with LTPA, nor do they mediate the relationship between objective catering services and LTPA.

Conversely, objective green spaces significantly predict a positive influence on LTPA, as evidenced by a regression coefficient of 0.954. Similarly, perceived green spaces independently predict a beneficial effect on LTPA. However, the mediating role of perceived green spaces in translating objective green space attributes to LTPA is not significant.

Regarding recreational facilities, both objective and perceived measures positively correlate with LTPA, with regression coefficients of 0.598 and 16.025, respectively. Perceived recreational facilities effectively mediate the relationship between objective facilities and LTPA-related behaviors, enhancing the positive impacts of objective attributes. The analysis of direct, indirect, and total effects of objective recreational facilities on LTPA shows that the indirect effect's coefficient is 0.105, constituting about one-sixth of the total effect.

Review of Models 1a-1d reveals that incorporating both objective and perceived environmental variables leads to minor changes in the significance and coefficients of the covariates. Notably, socio-economic factors such as educational attainment and sports inclinations significantly enhance residents' participation in LTPA. For instance, higher education levels correlate with increased LTPA participation, and a strong interest in sports is associated with longer engagement in LTPA. However, other covariates including gender, age, and income level, exhibit minimal impact on LTPA.

Structural equation analysis excluding the residential self-selection sample

After excluding samples affected by residential self-selection, the analysis included 152 individuals residing in work unit compounds and affordable housing, detailed in Table 4. The assessment revealed that Model 2d attained a Goodness of Fit Index of 0.908, suggestive of a moderately satisfactory model congruence. Subsequent to the exclusion, the investigation discerned that the interrelations amongst the majority of variables and leisure-time physical activity preserved their stability, notwithstanding slight modifications in certain variables. Notably, the influence of green spaces, measured through both objective assessments and subjective perceptions, lost statistical significance in influencing LTPA. This change may be attributed to the peripheral urban settings of work unit compounds and affordable housing, often characterized by limited green spaces. Furthermore, the negative correlation between the quantity of objectively assessed catering services and LTPA intensified, with the regression coefficient changing from -0.227 to -0.718. Additionally, the perceived recreational facilities continued to play a crucial mediating role in linking the objective recreational amenities to LTPA.

Table 4 Structural equation analysis excluding the residential self-selection sample.
Full size table

Discussion

Functional impact of built environment features on LTPA

Our study emphasizes the critical role of strategic urban design in enhancing LTPA through well-planned built environments. Specifically, our findings indicate that the presence of catering services negatively impacts LTPA. This adverse effect can be attributed to the primary function of catering services, which is to satisfy basic physiological needs like hunger. Locations dominated by dining establishments tend to promote sedentary lifestyles. Despite positive perceptions of such environments, their association with inactive behaviors means they do not substantially boost physical activity31,40. In contrast, green spaces and recreational facilities have a positive influence on LTPA, catering to both physical needs and psychological well-being41. These spaces, characterized by their recreational potential, naturally encourage more active uses, such as walking, playing, or exercising, which are intrinsically motivated and less constrained by physiological needs42,43,44. To maximize the effectiveness of urban planning in fostering LTPA, it is crucial that planners incorporate functional urban features aligned with active lifestyles. Urban planners should ensure that areas intended to promote physical activity are equipped with appropriate amenities like parks and recreational centers, which not only serve leisure purposes but are also accessible and appealing to the community.

Mismatch between the objective and perceived built environments

Initial ICC tests showed poor homogeneity between OBE and PBE. This discrepancy likely stems from PBE being based on respondents' memories, which often focus more on frequently visited areas rather than the entire 500-m radius considered in objective audits45. However, although some studies have prompted respondents to describe and assess the built environment along commonly traversed routes, notable differences remain when compared to objective audits46. This persistent mismatch highlights the crucial need for urban planners to integrate both objective measurements and personal perceptions in the design of public health interventions and urban environments47,48,49. By adopting this comprehensive strategy, urban planners not only support the development of environments that encourage physical activity, but also enhance our understanding of the environment-behavior relationship.

Our regression models demonstrate that OBE predictors forecast LTPA outcomes more accurately across the full sample compared to PBE measures. This superior predictive power of OBE likely stems from its temporal stability. Unlike perceptions, which may fluctuate with changes in personal health, mood, or weather, OBE characteristics remain constant over time, enhancing their reliability in physical activity studies5,47. Although this study shows that OBE more effectively predicts LTPA than PBE, other research suggests that PBE may be a stronger predictor of overall physical activity50. This variance underscores the need for further research to explore these differing results and to enhance the scientific robustness of our understanding of how environments influence physical activity.

Mediating role of the perceived built environment

Our findings affirm the crucial role of the OBE in influencing LTPA by shaping perceived equivalents, as posited by the Stimulus-Organism-Response theory. Specifically, our study demonstrates that OBE not only acts as a direct stimulus but also significantly molds individuals' perceptions and cognitions of their surroundings, which in turn dictate their activity choices. This influence is supported by analogous studies5,7,16, such as research conducted across three American states which demonstrated that objective factors like population density and road intersection density can influence both leisure walking through perceived land use mix51.

Given the positive mediating role of the PBE, urban planners and policymakers are encouraged to adopt a dual-focus strategy in environmental design and policy formulation. First, enhancing the objective quality of urban spaces, particularly recreational facilities, to ensure they are not only functionally adequate but also perceived as beneficial and inviting. Second, fostering a richer understanding of environmental perceptions through community engagement and feedback loops can inform more effective urban interventions. By considering both the objective and perceived built environments, strategies can be crafted that more effectively catalyze LTPA, contributing to public health goals and creating more liveable urban spaces. This nuanced approach underscores the importance of a comprehensive urban planning paradigm that integrates empirical insights with perceptual realities to optimize the health and well-being of urban populations.

However, our analysis did not reveal significant mediation effects of perceived availability of green spaces and catering services on LTPA. The reason might lie in the direct relevance of recreational facilities to physical activity. These facilities explicitly support exercise, such as gyms and sports courts, making their perceived availability more impactful on LTPA decisions52. Green spaces, while beneficial for general wellness, often facilitate more passive uses, such as relaxation and social gatherings, which may not directly encourage physical activity29,53. Similarly, catering services, despite their perceived availability, primarily cater to dietary needs and do not inherently promote physical activity54. This underscores the necessity for urban planners and public health professionals to focus on the functional aspects of built environments when aiming to enhance physical activity. Aligning facility provision with the active lifestyles they intend to promote can lead to more effective urban health interventions, ensuring that the built environment supports the public's physical and psychological well-being.

Strengths and limitations

To our knowledge, this study is among the few that examine the mediation effect of PBE in China. Our contributions are significant in three aspects. Firstly, this study broadens the understanding of how the built environment influences human health by focusing on PBE, extending beyond psychological perceptions such as perceived safety, which have been the focus of previous research55. Secondly, we enhance the body of knowledge on the mediation effect of PBE within a developing context, where urban development patterns, urban form, and socioeconomic characteristics distinctly differ from those in Western countries10,56. Finally, our findings highlight a critical mismatch between OBE and PBE, along with their inconsistent effects on physical activity. This discrepancy underscores the importance of integrating both environments to more accurately interpret human behavior.

However, this study is subject to several limitations. Firstly, similar to many other investigations5,7,21, this study is cross-sectional, which restricts our analysis to correlations between OBE and PBE and LTPA, rather than causal relationships. Future research would benefit from longitudinal data to explore these dynamics more comprehensively2,57. Secondly, the data on LTPA were collected via residents' recall, which might introduce recall bias regarding the duration of physical activity. We recommend the use of more advanced technologies for future studies, such as mobile phone signaling data and GPS-based fitness trackers, to objectively record and analyze physical activity behaviors21. Additionally, the use of a 500 m buffer area as neighborhoods, as in many studies28,29, suffers from the uncertain geographic context problem58. While this buffer size is practical and aligns with Chinese community-life cycles for 10-min walking distances28, future research should explore the impact of using different buffer sizes.

Conclusions

Main findings

This study leverages the Stimulus-Organism-Response framework to examine the interplay among the objective built environment, its perceived built environment, and their effects on LTPA. Our findings highlight that OBE better predicts residents' LTPA than PBE. Additionally, people significantly underestimated the OBE around them. We also found that the perceived availability of recreational facilities mediates the effect of the objective availability of recreational facilities on LTPA, whereas perceived availability of green spaces and catering services do not show significant mediation effects.

Implications

Drawing on these insights, we offer several recommendations for urban planners and policymakers to optimize environments for enhancing physical activity. First, increasing public awareness and appreciation of supportive and accessible environments could notably boost community activity levels. Second, urban design should prioritize not only functional support for physical activity but also perceptual engagement for residents. Third, improving the quality and maintenance of recreational facilities and green spaces is essential to making these areas more attractive and conducive to physical activity. By integrating a deep understanding of human–environment interactions into urban design, planners can craft spaces that effectively foster physical activity and enhance overall well-being.