Interactions Between Objective and Subjective Built Environments in Promoting Leisure Physical Activities: A Case Study of Urban Regeneration Streets in Beijing

Abstract

The built environment plays a critical role in promoting residents’ physical activity, yet the interactive mechanisms between objective environmental factors and subjective perceptions remain insufficiently understood. This study examines three functionally distinct neighborhoods in Beijing’s Xicheng District—Xinjiekou (historic), Financial Street (administrative), and Baizhifang (residential)—representing typical urban regeneration contexts. Using an ordered logit model based on 1072 valid questionnaires, we analyze how objective and subjective built environment factors jointly influence residents’ leisure physical activities. Results reveal that socioeconomic attributes (income, age, education) are primary determinants of activity engagement. Among objective factors, facility accessibility and land-use mix exert the strongest direct effects, while subjective perceptions—particularly satisfaction with environmental attractiveness—significantly moderate these relationships. Based on these findings, we propose differentiated spatial renewal strategies tailored to each street type. This research provides empirical evidence for implementing health-oriented urban regeneration policies in high-density metropolitan areas.

1. Introduction

The built environment profoundly shapes human health behaviors, serving as both a facilitator and barrier to physical activity engagement. As urbanization accelerates globally, the relationship between spatial configurations and health outcomes has emerged as a critical concern for both urban planning and public health disciplines. With over 5 million annual deaths attributed to physical inactivity worldwide [1], understanding how urban spaces influence leisure physical activities has become imperative for evidence-based interventions.

In China, this urgency is manifested in multiple national strategic initiatives. The Healthy China 2030 plan explicitly calls for “integrating health into all policies” [2], recognizing that health promotion extends far beyond traditional medical interventions to encompass the fundamental restructuring of urban environments. More recently, the 2025 “Weight Management Year” initiative has placed particular emphasis on neighborhood-level interventions to enhance physical activity [3], acknowledging that sedentary lifestyles and rising obesity rates pose significant threats to public health. These policy frameworks reflect a paradigm shift toward preventive health strategies that leverage urban design as a tool for behavior modification. However, translating these ambitious goals into actionable design principles requires robust empirical evidence on how specific environmental features influence physical activity patterns. This need is particularly acute in China’s historic urban cores, where ongoing urban regeneration creates both opportunities and constraints for health-promoting environmental interventions.

China’s historic urban cores are currently undergoing profound transformation through urban regeneration programs, a context that fundamentally shapes the possibilities for health-promoting environmental interventions. Over the past two decades, China’s approach to urban renewal has undergone a significant paradigm shift—from large-scale demolition and redevelopment toward small-scale, community-oriented micro-regeneration that emphasizes heritage conservation and public participation [4]. This transition reflects evolving policy priorities articulated in national strategic documents, including the integration of heritage conservation planning into territorial spatial planning frameworks and the updated conservation principle of “prioritizing protection, strengthening management, exploiting values, effective utilization while bringing heritage to life” [4]. These policy frameworks position historic neighborhoods as primary sites for micro-regeneration initiatives, yet these same neighborhoods present distinctive challenges for health-oriented spatial interventions.

Historic neighborhoods in Chinese cities face distinctive functional constraints that complicate health-oriented interventions. Traditional spatial configurations—characterized by narrow hutong alleyways, high building density, and limited public open spaces—restrict opportunities for physical activity infrastructure development. Heritage protection regulations further constrain physical modifications, as conservation requirements for listed buildings limit flexibility to introduce modern fitness facilities or reconfigure existing spaces [4]. Additionally, these neighborhoods often exhibit demographic imbalances, with aging original residents remaining while younger populations relocate to peripheral areas, creating communities with elevated health needs but constrained mobility capacities. These conditions generate inherent tensions among three competing objectives: heritage conservation, high-density development, and healthy living environments. Heritage preservation may conflict with density optimization when protection requirements limit development intensity within historic cores while surrounding areas undergo intensive construction. Conservation priorities can also compromise health-promoting interventions when traditional spatial scales prove incompatible with contemporary fitness requirements. Meanwhile, high-density development presents paradoxical effects on physical activity—potentially suppressing outdoor activity through spatial compression and safety concerns while simultaneously enhancing facility accessibility through concentrated service provision [4]. Despite growing scholarly attention to each dimension individually, integrated frameworks examining how these objectives can be synergistically optimized remain underdeveloped. Addressing this gap requires first clarifying what aspects of the “built environment” are most relevant to residents’ physical activity within these complex neighborhood contexts.

The term “built environment” encompasses multiple scales and disciplines—from metropolitan-level urban form to building-interior design. Given the neighborhood-scale focus necessitated by the micro-regeneration context discussed above, this study specifically examines built environment features directly relevant to residents’ daily leisure physical activities within their immediate living contexts. The objective dimension includes quantifiable spatial characteristics such as facility accessibility, land-use mix, development intensity (urban density and floor area ratio), and public transit connectivity, measured through GIS analysis and point-of-interest databases. The subjective dimension captures residents’ perceptual evaluations of their neighborhood environments, including perceived accessibility, safety, and attractiveness. This focused definition excludes metropolitan-scale urban morphology and building-interior design considerations, concentrating instead on the street and community-level environmental attributes that shape spontaneous health behaviors.

The relationship between the built environment and leisure-time physical activity (LTPA) has been extensively examined from two complementary perspectives. From the objective dimension, research has established that higher residential density, mixed land use [5,6,7,8], and accessibility to recreational facilities [9] consistently promote physical activity engagement. These findings have informed planning guidelines emphasizing compact, mixed-use development as health-promoting urban forms. From the subjective dimension, perception-based studies have revealed that residents’ psychological responses—including perceived safety, environmental aesthetics, and social cohesion—play equally important roles in determining activity levels [10,11]. Neighborhoods perceived as attractive and safe encourage spontaneous outdoor activities, while environments perceived as threatening suppress physical activity regardless of objective characteristics.

Despite these advances, significant gaps remain in our understanding of environment-behavior relationships. First, most existing studies examine objective and subjective factors in isolation, failing to capture the complex interactions between physical features and human perceptions. Whether and how subjective perceptions modify the influence of objective environmental attributes, and whether perceived environmental quality can amplify or diminish the effects of physical infrastructure, remains unclear. Second, the mechanisms through which different types of built environment factors jointly influence spontaneous health behaviors such as walking, fitness exercises, and sports participation remain poorly understood. Third, existing research predominantly focuses on Western contexts, with limited attention to high-density Asian cities where spatial constraints, cultural norms, and socioeconomic stratification create distinct environment-health dynamics. Finally, the translation of research findings into context-specific design strategies for different neighborhood types remains underdeveloped, limiting the practical utility of academic insights for urban regeneration practice.

This study addresses these gaps by pioneering an integrated analytical approach that simultaneously examines objective environmental features, subjective perceptions, and their interactive effects on LTPA patterns. Using Beijing’s Xicheng District as the research site, we construct a multi-level ordered logit model based on 1072 resident surveys integrated with multi-source geographic data, including land use records, facility databases, and street network analysis. The following discussion elaborates on the rationale for site selection and establishes Xicheng District’s representativeness for this investigation.

The selection of Xicheng District is strategically justified for several reasons. First, regarding policy significance, Xicheng District constitutes part of the Capital Core Functional Area as designated in the Control Detailed Planning for the Core Functional Area of the Capital (2018–2035), which explicitly assigns three intertwined objectives to this area: historical and cultural preservation, optimization of administrative and commercial functions, and enhancement of livable environments [12]. This triple mandate—balancing heritage protection, functional upgrading, and quality-of-life improvement—epitomizes the multidimensional challenges confronting urban regeneration initiatives across Chinese megacities. Second, regarding spatial characteristics, with a population density of 22,000 persons per square kilometer [13], Xicheng District represents one of the most densely populated urban areas in China, where the tension between spatial resource constraints and residents’ health needs reaches its most acute expression. Third, regarding typological coverage, the district encompasses diverse neighborhood typologies—from traditional hutong areas with heritage buildings to modern residential clusters and contemporary commercial zones—enabling comparative analysis across functionally distinct urban contexts. The three selected study areas (Xinjiekou, Financial Street, and Baizhifang), respectively, represent historical preservation, administrative-commercial, and residential regeneration scenarios, corresponding to the primary urban renewal typologies encountered throughout China’s major cities. Fourth, regarding data infrastructure, Xicheng District benefits from comprehensive demographic, facility, and land-use databases maintained by municipal authorities, which provide robust quantitative foundations for empirical investigation.

The research examines three functionally distinct neighborhoods within Xicheng District, each representing different urban regeneration scenarios. Xinjiekou Street exemplifies historical preservation contexts, characterized by traditional hutong alleyways, the highest land-use mix among the three areas, and predominantly low-rise courtyard housing with a floor area ratio of 1.23. The area maintains a relatively balanced demographic composition with 8.7% of residents aged 66 and above. Financial Street represents Beijing’s premier administrative and financial center, generating a regional GDP of 464.09 billion CNY—over thirteen times that of Xinjiekou. The built environment is dominated by modern high-rise office towers with the highest floor area ratio (1.89), while accessibility to public parks remains the lowest among the three streets. Baizhifang Street typifies aging residential clusters developed during the 1980s–1990s work-unit (danwei) housing era, featuring mid-rise buildings (floor area ratio: 1.44) and the lowest land-use mix. The demographic profile is notably older, with 16.0% of residents aged 66 and above and only 5.8% earning above 12,001 RMB monthly, yet the area demonstrates the highest accessibility to parks and open spaces. By comparing these diverse contexts, the study generates differentiated design strategies tailored to specific neighborhood characteristics rather than generic prescriptions.

This research aims to achieve three primary objectives, each corresponding to distinct scientific contributions. First, we quantify the relative contributions of objective built environment factors (density, land-use mix, facility accessibility, transit connectivity) and subjective perceptions (safety, aesthetics, attractiveness) to LTPA engagement, controlling for individual socioeconomic attributes. This objective advances environment-behavior theory by proposing an “objective constraints—subjective modification—socioeconomic determination” framework that captures multi-layered drivers of health behaviors in complex urban systems. Second, we identify the moderating effects of subjective perceptions on objective environmental influences, revealing whether and how psychological responses mediate the impact of physical features. This integrated perspective moves beyond conventional approaches that examine objective and subjective dimensions in isolation. Third, we develop evidence-based, context-sensitive urban renewal strategies for distinct neighborhood types, translating statistical findings into actionable design recommendations aligned with China’s Healthy China 2030 agenda. Methodologically, the study integrates multi-source geographic data (POI databases, Baidu heatmaps, land-use records, street network analysis) with large-scale questionnaire surveys, employing the Dynamic Two-Step Floating Catchment Area (D2SFCA) method to quantify facility accessibility and ordered logit modeling to examine environment-behavior associations—offering a replicable template for health-oriented urban research in data-rich Chinese urban contexts.

2. Materials and Methods

2.1. Sample Selection

Xicheng District is part of the core functional area of the capital. In 2022, the population density reached 22,000 people per square kilometer [13]. The tension between spatial resource constraints and residents’ health needs is characteristic of urban renewal areas in large cities across the country. According to the Control Detailed Planning for the Core Functional Area of the Capital (Block Level 2018–2035), Xicheng District is planned to be a core functional area of the capital with outstanding administrative functions, rich cultural heritage, and a high-quality living environment [12]. The three core objectives of Xicheng District are historical and cultural preservation, optimization of commercial and administrative functions, and the enhancement of livable environments. This study divides Xicheng District into functional zones, forming three types of neighborhoods: historical, administrative, and residential. Typical representative streets, such as Xinjiekou Street (historical preservation), Financial Street (administrative center), and Baizhifang Street (residential cluster), are selected as the research objects (Figure 1). A comparison of key characteristics across the three study streets is presented in

Table 1. Comparison of Study Streets. Source: Drawn by the authors.

	Xinjiekou Street	Financial Street	Baizhifang Street
Positioning [5]	Historical Type	Administrative Type	Residential Type
Spatial Form
Economic Data [12]	34.92 billion CNY	464.09 billion CNY	0.17 billion CNY

.

2.2. Selection of Objective and Subjective Built Environment Variables

The selection of subjective built environment dimensions in this study was guided by the Stimulus-Organism-Response (S-O-R) framework, which posits that objective environmental features act as stimuli, subjective perceptions represent organism states, and behavioral outcomes constitute responses [14]. This theoretical perspective suggests that physical environment characteristics must be mediated by individual psychological assessments before they can translate into behavioral responses such as physical activity engagement. Drawing on this framework and systematic reviews of environment-perception-behavior relationships in neighborhood contexts [15,16,17,18,19,20,21], this study identified three core perceptual dimensions as particularly relevant to leisure physical activity in urban regeneration areas: perceived accessibility (convenience of reaching recreational facilities), perceived safety (sense of security during outdoor activities), and perceived attractiveness (aesthetic appeal of the surrounding environment). These dimensions align with established constructs in environmental psychology research and correspond to validated indicators in recent built environment studies examining the dual influence of objective features and subjective perceptions on health behaviors [14].

This study, based on previous relevant literature (

Table 2. Selection and Explanation of Built Environment Variables. Source: Drawn by the authors.

Variable Selection			Variable Explanation	Basis for Indicator Selection
Objective Built Environment	Urban Function within the Area (Facility Accessibility)	Leisure Facility Accessibility	Accessibility of Parks, Green Spaces, Plazas, and Open Spaces within the Three Different Streets	[15,20]
			Accessibility of Gyms within the Three Different Streets
	Basic Attributes within the Area (Development Intensity)	Urban Density	Urban Density of Each Street	[17,24]
		Floor Area Ratio	Floor Area Ratio of Each Street	[10,17]
		Land Use Mix	Land Use Mix of Each Street	[15,24]
	Transportation within the Area (Public Transportation)	Connectivity with Public Transportation	Number of Bus Stops within Each Street	[10,16,18]
Subjective Built Environment	Accessibility Satisfaction	Is it convenient for you to access facilities (parks, plazas, sports venues) from your residence?		[20,21]
	Safety Satisfaction	Do you feel safe walking around your neighborhood?		[22]
	Attractiveness Satisfaction	Do you find the surrounding environment pleasant?		[23]

), collects objective and subjective built environment factors that significantly influence physical activity. Objective built environment factors include density, mixed-use, accessibility, and street connectivity [15,16,17,18,19,20]. This study refines the objective built environment variables into three categories. First, accessibility to recreational facilities is classified as an urban function indicator, as it is closely related to leisure open spaces. Second, urban density, floor area ratio, and land-use mix are classified as regional attribute indicators. Third, public transportation is classified under regional transportation indicators. The specific measurement methods for each indicator are presented in Table 2. The impact of the subjective built environment on physical activity primarily focuses on three dimensions [21,22]. Satisfaction with accessibility refers to residents’ subjective perception of whether various facilities around their residential area are easily accessible and usable. Satisfaction with safety refers to residents’ subjective perception of security when accessing various facilities around their residential area [23]. Satisfaction with attractiveness refers to residents’ subjective perception of the aesthetic appeal of the open spaces around their residential area [23].

2.3. Data Collection

For leisure physical activity, the level of residents’ leisure physical activity in this study was measured using the leisure section of the International Physical Activity Questionnaire (IPAQ), which assesses the frequency and duration of leisure-time physical activities. IPAQ classifies leisure physical activities into three types: walking, moderate-intensity activities, and vigorous-intensity activities. Moderate-intensity activities include Tai Chi, square dancing, table tennis, and badminton, while vigorous-intensity activities include running, swimming, and basketball. The average intensity of these three types of leisure physical activities is 3.3, 4, and 8 METs, respectively. By combining the frequency and duration of these activities, activity levels were calculated. The estimates were further aligned with the Guidelines for Weight Management (2024 edition) [3], which specifies differentiated physical activity standards for children, adolescents, and adults (as shown in

Table 3. Intervention Program for Physical Activity Health Standards for Different Populations (2024 Edition of the “Guidelines for Weight Management”).

Indicator Population	Children (Under 7 Years)	Adolescents (7–18 Years)	Adults
Frequency	≥6 days/week	≥5 days/week	≥5 days/week
Intensity	Encourage free movement for children under 1 year; engage in at least 180 min of varied physical activity daily	It is recommended to accumulate at least 60 min of moderate- to vigorous-intensity physical activity daily, both inside and outside school.	Start at moderate intensity and gradually increase to higher intensity
Time/Recommended Amount			30 min/day, totaling 150 min/week; gradually increase to 60 min/day or at least 250–300 min/week

). Since the leisure physical activity levels are ordinal variables, the data were fitted using an Ordered Logit Model (OLM). The survey questionnaire was designed to collect information on residents’ basic characteristics, health status, and environmental evaluations, aiming to gather residents’ ratings of the built environment factors in Xicheng District.

The survey was administered through face-to-face interviews using printed questionnaires. The target population comprised permanent residents of the study areas. During the survey, respondents were screened by asking whether they were local residents; tourists and temporary visitors were excluded to ensure that respondents had sufficient familiarity with their neighborhood environment. The survey covered residents of all age groups to reflect the demographic diversity of the communities.

A spatial grid sampling approach combined with convenience sampling was employed. Each of the three neighborhoods was divided into 200 m × 200 m grid cells, consistent with the spatial resolution used for objective built environment analysis. Within each grid cell, investigators visited locations where leisure physical activities typically occur, such as parks, plazas, and community open spaces. Respondents were recruited through a combination of random intercept and active invitation, with priority given to residents who were currently engaging in or had just completed physical activities, ensuring that respondents had direct experience with leisure physical activity in their neighborhoods.

During the survey process, attention was paid to the gender and age distribution of respondents, with investigators making efforts to balance the representation of different demographic groups. However, since respondents were recruited at leisure activity locations, the sample may exhibit selection bias toward more physically active individuals, a limitation that is discussed in Section 4.

The survey was conducted from 1 August to 3 October 2025, covering both weekdays and weekends. The survey team consisted of 17 groups, with 3–4 trained investigators per group. Daily surveys were conducted during three time periods: morning (7:00–11:00), midday to afternoon (11:00–15:00), and afternoon to evening (15:00–20:00), capturing residents who engaged in leisure physical activities at different times of day. Specific survey locations included: Xinjiekou Street—Huguo Temple area, Jishuitan Park surroundings, and the commercial district near Xinjiekou intersection; Financial Street—Yuetan Park vicinity, Financial Street Plaza, and Xidan commercial area; Baizhifang Street—Taoranting Park entrances, community squares within residential compounds, and neighborhood commercial streets.

Participation was voluntary, and no material compensation was provided to participants. Investigators explained the research purpose and estimated time commitment when inviting potential respondents; respondents then decided whether to participate at their own discretion. Each face-to-face interview lasted approximately 15 min. Investigators asked questions following the questionnaire sequence and recorded responses, providing clarification when necessary while avoiding leading questions.

Completed questionnaires underwent a two-stage quality control process. In the first stage, investigators reviewed questionnaires before data entry to identify and flag those containing clearly unreasonable responses, such as self-contradictory information or implausible physical activity durations. In the second stage, data were entered using Microsoft Excel and verified by another researcher to ensure accuracy. Questionnaires that did not meet the inclusion criteria (e.g., those completed by tourists) or had serious data quality issues were excluded. Of 1106 questionnaires collected, 1072 were retained for analysis (validity rate: 96.9%), comprising 391 from Xinjiekou Street, 355 from Financial Street, and 326 from Baizhifang Street. The statistical data are shown in

Table 4. Statistics of the Socioeconomic Attributes Questionnaire.

Attribute	Grouping	Percentage (%)
		Xinjiekou Street	Financial Street	Baizhifang Street
Gender	Male	46.4	46.2	57.8
	Female	53.6	53.7	42.1
Residential Status	Living Alone	16.1	11.8	12.3
	Living with Children	41.2	43.1	39.0
	Co-renting with Others	18.2	24.6	13.8
	School or Work Accommodation	16.6	17.8	30.1
	Other	7.6	2.4	4.6
Health Status	Very Poor	2.3	1.3	0.9
	Poor	8.4	10.5	4.0
	Average	32.5	26.8	43.6
	Good	32.0	37.4	32.6
	Healthy	24.6	23.7	18.7
Monthly Income	≤3000 RMB	23.8	12.1	36.0
	3001–6000 RMB	19.7	23.5	20.6
	6001–9000 RMB	26.9	29.2	28.6
	9001–12,000 RMB	17.6	23.1	8.9
	≥12,001 RMB	11.7	11.8	5.8
Age	0–6 years	2.8	0.2	5.3
	7–17 years	12.8	8.8	10.2
	18–40 years	51.0	50.8	47.5
	41–65 years	24.6	27.7	20.9
	66 years and above	8.7	12.3	16.0
Days of Moderate-or-Above Intensity Physical Activity in the Past 7 Days	0 days	18.2	16.9	33.0
	1 day	20.2	57.4	12.7
	2–3 days	19.2	24.4	21.8
	4–5 days	20.0	0.6	7.2
	6–7 days	22.3	0.4	25.0
Average Daily Duration of Moderate-or-Above Intensity Physical Activity	0–10 min	31.0	28.8	36.9
	11–60 min	21.5	42.5	26.4
	61–120 min	14.1	18.0	13.5
	121–180 min	14.6	6.6	9.2
	≥181 min	18.7	3.9	13.8

.

The leisure physical activity section of the questionnaire adopted the International Physical Activity Questionnaire (IPAQ), a well-established instrument with demonstrated validity and reliability across diverse populations and cultural contexts. The subjective built environment items were adapted from validated scales used in prior research examining perceived accessibility, safety, and attractiveness in neighborhood environments [15,16,17,18,19,20,21], with minor wording modifications to reflect the specific urban context of Beijing’s historic neighborhoods. For instance, general items regarding facility accessibility were contextualized to include locally relevant facility types such as parks, plazas, and sports venues that are characteristic of the study area.

The environmental evaluation questionnaire adopted the Likert five-point scale, with each scale point accompanied by clear semantic labels to ensure consistent interpretation across respondents. The scale anchors ranged from “very dissatisfied” (−2), indicating environmental conditions perceived as severely hindering outdoor activity engagement, to “very satisfied” (+2), indicating conditions perceived as strongly encouraging such activities. These semantic anchors were printed directly on the questionnaire, providing respondents with behavioral reference points for their evaluations rather than requiring abstract satisfaction judgments. This operationalization approach follows methodological recommendations for linking subjective assessments to behavioral implications in built environment research [14]. The design of the environmental evaluation questionnaire is shown in

Table 5. Questionnaire Design for Subjective Built Environment.

Attribute	Grouping	Score
How satisfied are you with the land use in your neighborhood?	How satisfied are you with the mixed land use in your neighborhood?	The options “Very Dissatisfied, Dissatisfied, Neutral, Satisfied, Very Satisfied” correspond to the values (−2, −1, 0, 1, 2), respectively. Respondents were instructed to evaluate each attribute based on the extent to which the environmental condition facilitates or hinders their willingness to engage in outdoor physical activities.
	How satisfied are you with the floor area ratio (FAR) of buildings in your neighborhood?
	How satisfied are you with the building density in your neighborhood?
	How satisfied are you with the green space area in your neighborhood?
	How satisfied are you with the road length in your neighborhood?
	How satisfied are you with the number of intersections in your neighborhood?
How satisfied are you with the economic conditions of your neighborhood?	How satisfied are you with the housing prices in your neighborhood?
	How satisfied are you with the age of buildings in your neighborhood?
	How satisfied are you with the fitness costs at gyms in your neighborhood?
	How satisfied are you with the diversity of facility types in your neighborhood?
	How satisfied are you with the number of commercial facilities in your neighborhood?
	How satisfied are you with the number of educational facilities in your neighborhood?
	How satisfied are you with the number of medical facilities in your neighborhood?
How satisfied are you with the accessibility of facilities in your neighborhood?	How satisfied are you with the number of fitness facilities in your neighborhood?
	How satisfied are you with the number of public fitness facilities in your neighborhood?
	How satisfied are you with the number of commercial fitness facilities in your neighborhood?
	How satisfied are you with the number of fitness advertising facilities in your neighborhood?
How satisfied are you with the reachability of facilities in your neighborhood?	How satisfied are you with the number of bus stops in your neighborhood?
	How satisfied are you with the number of subway stations in your neighborhood?
	How satisfied are you with the walkability to the nearest fitness facility from your residence?
How satisfied are you with the comfort of your neighborhood?	How satisfied are you with the sunlight exposure in your neighborhood?
	How satisfied are you with the wind environment in your neighborhood?
	How satisfied are you with the thermal (temperature) environment in your neighborhood?
	How satisfied are you with the noise environment in your neighborhood?
	How satisfied are you with the greenery coverage in your neighborhood?
	How satisfied are you with the pavement quality in your neighborhood?
	How satisfied are you with the architectural design of buildings in your neighborhood?
How satisfied are you with the spatial safety of your neighborhood?	How satisfied are you with the number of lighting facilities in your neighborhood?
	How satisfied are you with the number of surveillance facilities in your neighborhood?
	How satisfied are you with the number of police kiosks in your neighborhood?
	How satisfied are you with the situation of vehicles occupying road space in your neighborhood?
Overall, how satisfied are you with engaging in sports and fitness activities in your neighborhood?

.

2.4. Leisure Physical Activity Data

According to the survey data, the study found that the largest proportion of residents in Xicheng District engage in low-level leisure physical activities (as shown in Figure 2). Overall, 64% of respondents reported low activity levels, 31% reported high levels, and only 5% reported moderate levels. However, notable variations were observed across the three streets: Financial Street exhibited the highest proportion of low-level activity (73.7%) and the lowest proportion of high-level activity (20.3%), while Xinjiekou Street showed the most favorable distribution with the highest proportion of high-level activity (38.7%) and the lowest proportion of low-level activity (57.8%). Baizhifang Street demonstrated an intermediate pattern with 60.9% low-level and 33.5% high-level activity. These spatial differences suggest that neighborhood-specific built environment characteristics may influence residents’ physical activity engagement.

Regarding age-specific patterns, for children aged 0–6 and seniors aged 66 and above, more than 50% engage in high levels of activity, displaying a typical “active at both ends” pattern. Among adolescents aged 7–17, 55% engage in low levels of activity. For the 18–40 age group, 66% engage in low-level activities, which is the highest proportion across all age groups. Meanwhile, 39% of middle-aged individuals aged 41–65 engage in high levels of activity, indicating that they are more active compared to younger age groups (as shown in Figure 3).

2.5. Subjective Built Environment Data

Based on the 1072 valid questionnaires collected, the average score for each attribute was calculated by averaging the values obtained from the different rating groups for each evaluation attribute (as shown in

Table 6. Statistics of the Subjective Built Environment Questionnaire.

Attribute	Grouping	Percentage (%)
		Xinjiekou Street	Financial Street	Baizhifang Street
Satisfaction with Land Use in the Neighborhood	−2	2.8	6.8	1.8
	−1	17.1	14.0	6.4
	0	44.1	35.6	42.4
	1	26.9	31.7	35.3
	2	8.9	11.6	13.8
Satisfaction with Economic Conditions of the Neighborhood	−2	4.8	2.2	2.4
	−1	25.3	17.4	10.4
	0	43.3	39.4	54.1
	1	20.7	28.6	23.0
	2	5.6	12.3	9.8
Satisfaction with Accessibility of Facilities in the Neighborhood	−2	2.5	1.1	1.5
	−1	22.3	14.9	3.6
	0	43.3	34.5	43.0
	1	22.5	35.0	36.0
	2	9.2	14.3	15.6
Satisfaction with Reachability of Facilities in the Neighborhood	−2	1.0	2.4	1.2
	−1	14.6	9.4	5.5
	0	37.6	37.6	33.5
	1	31.7	35.2	43.3
	2	14.8	15.1	16.3
Satisfaction with Comfort of the Neighborhood	−2	1.5	3.7	1.8
	−1	10.7	8.8	5.8
	0	9.2	46.4	41.8
	1	39.2	30.3	38.1
	2	39.2	10.5	12.3
Satisfaction with Spatial Safety of the Neighborhood	−2	2.3	3.0	1.5
	−1	23.3	10.5	5.8
	0	36.9	30.8	27.3
	1	30.5	34.5	43.3
	2	6.9	20.9	21.8
Overall Satisfaction with Engaging in Sports and Fitness Activities in the Neighborhood	−2	1.0	3.0	0.6
	−1	7.4	12.1	4.6
	0	12.3	32.8	41.2
	1	47.4	32.3	38.7
	2	31.7	19.6	14.7

Note: The options “Very Dissatisfied, Dissatisfied, Neutral, Satisfied, Very Satisfied” correspond to the values (−2, −1, 0, 1, 2), respectively. Respondents were instructed to evaluate each attribute based on the extent to which the environmental condition facilitates or hinders their willingness to engage in outdoor physical activities.

). In this study, the percentage data were converted into an intuitive score (ranging from −2 to 2) by calculating the weighted average score for each item. The calculation formula is shown in Equation (1):

$$\overline{X}={\displaystyle \frac{{\sum }_{i=1}^n(w_i\cdot f_i)}{N}},$$

(1)

In the equation, $w_i$ represents the weight value of the $i$-th option, which corresponds to the score for that option. $f_i$ represents the frequency of selecting the $i$-th option, $n$ represents the total number of options, and $N$ represents the total number of valid samples.

The statistical results are shown in Table 6: Baizhifang Street has the highest overall satisfaction with a composite score of 0.56, demonstrating the most balanced performance. It leads the other two streets in terms of facility accessibility, reachability, and spatial safety. Xinjiekou Street has a composite score of 0.41, with the lowest overall satisfaction, but its advantages are very distinct. It significantly outperforms the others in comfort and satisfaction with sports and fitness facilities. However, it shows clear weaknesses in areas such as affordability and land use. Financial Street has a composite score of 0.43, with an overall satisfaction level in the middle. It excels in affordability and spatial safety but is relatively weak in terms of comfort.

To facilitate cross-street comparison of subjective built environment satisfaction, Figure 4 presents a radar chart visualization of the weighted average scores from

Table 7. Weighted Average Scores of Street-Level Items.

Evaluation Dimension Street	Xinjiekou Street	Financial Street	Baizhifang Street	Best Performer
Land Use	0.10	0.28	0.47	Baizhifang Street
Economic Conditions	−0.02	0.32	0.27	Financial Street
Facility Accessibility	0.14	0.47	0.60	Baizhifang Street
Facility Reachability	0.45	0.42	0.68	Baizhifang Street
Comfort	1.03	0.36	0.52	Xinjiekou Street
Spatial Safety	0.18	0.60	0.78	Baizhifang Street
Sports and Fitness	1.01	0.54	0.62	Xinjiekou Street
Overall Average Score	0.41	0.43	0.56	Baizhifang Street

. The chart reveals distinct satisfaction profiles across the three streets: Xinjiekou Street demonstrates notably higher scores in Comfort (1.03) and Sports & Fitness (1.01), while Baizhifang Street shows more balanced performance across all dimensions with the highest overall average score (0.56). Financial Street exhibits moderate satisfaction levels with particular strength in Spatial Safety (0.60).

2.6. Objective Built Environment Data—Facility Accessibility

Data Collection Method

The main steps include extracting dynamic population data from Baidu Heatmaps and Points of Interest (POI) of various leisure facilities, using D2SFCA for accessibility evaluation, and performing result analysis. First, dynamic population data for the study area was obtained using Baidu Heatmap extraction methods and then integrated and processed into dynamic population density. Second, utilizing Baidu Map’s urban leisure facility data, grid data, average dynamic population data, and road network data, the accessibility of the study area was calculated using D2SFCA.

• Dynamic Population Density Data Collection;

The dynamic population distribution data was extracted from Baidu Heatmaps, which reflect the dynamic activity trajectories of users based on real-time location data from Baidu software (Version V21) and specific websites, providing insights into the active users in the study area. These heatmaps offer a new source of geographic spatial data with low acquisition costs, high spatial resolution, and strong real-time capabilities. Python (Version 3.12) was used to access the Baidu Map Open Platform (https://lbsyun.baidu.com/index.php?title=webapi (accessed on 4 August 2025)) and a total of 56 heatmaps were obtained for the study area, covering the week from 4 August to 10 August 2025. The sampling time was from 07:00 to 21:00, with a 2 h interval. For data preprocessing, the heatmaps were imported for clipping, georeferencing, and projection transformation. The results are shown in

Table 8. Example of Preprocessed Population Heat Map in the Study Area on 4 August 2025, 07:00–21:00 (Author’s Illustration).

4 August, 07:00	4 August, 09:00	4 August, 11:00	4 August, 13:00
4 August, 15:00	4 August, 17:00	4 August, 19:00	4 August, 21:00

.

The population heatmaps reflect the activity trajectories of active users accessing the application, representing only a sample of the total population in the study area. Given the limitations in accuracy of Baidu Heatmaps, the data was reorganized to create heatmaps with a 200 m resolution. The average number of people and the population activity counts over the entire time span were calculated to extract more accurate data on crowd activities.

In this study, we reclassified the data based on the correspondence between the colors of Baidu Heatmaps and population density. We identified the population density for each grid in the daily average heatmap. Next, we created a fishing net (grid) with the same resolution as the heatmap (200 m resolution) to divide the average dynamic population heatmap. We then accurately extracted the weekly average grid activity level using the pixel depth as a representation of activity, which corresponds to the dynamic active population. Finally, the Extract Values to Points function in ArcGIS (Version 10.8) was used to calculate the average population activity for each grid, and the data was processed and overlaid to generate an average population density map (Figure 5).

2.

Service Capacity of Leisure Facilities;

The method for determining the service area of facilities primarily relies on the network analysis functionality in ArcGIS. Network analysis is used to calculate the area within the service range centered around each facility, with the range distance depending on the facility type. Finally, the coverage areas are merged based on the names of the leisure facilities.

The service capacity of leisure facilities is generally determined by the ratio of facility area to the service population. The area of parks, green spaces, and open spaces was obtained using the area calculation function in ArcGIS, while the area of gyms was obtained through merchant registration information on Baidu Maps. In this study, the service population refers to the total dynamic population within the coverage area of the leisure facilities, which can be extracted using the raster extraction function in ArcGIS to obtain the total population count from the dynamic population raster map. The park service capacity is calculated as the ratio of the park area to the total dynamic population within the coverage grid. The calculation formula is shown in Equation (2). The results are presented in

Table 9. Comparison of Objective Built Environment among Study Streets. (Note: The red areas in the figure indicate the administrative boundaries of the streets studied in this paper).

Elements Legend	Facility Service Area Map (15 min Walking Radius)	Facility Coverage Population Map	Per Capita Facility Service Capacity Map
Parks, Green Spaces, and Open Spaces
Gyms

.

$$R_j={\displaystyle \frac{S_j}{P_j}}$$

(2)

In the equation, $R_j$ represents the service capacity of the leisure facility $j$; $S_j$ is the area of facility j; and $P_j$ is the total dynamic population within the coverage area of the facility $j$.

3.

Accessibility Calculation;

Building on the traditional 2SFCA method, this study combines Baidu Heatmaps and network analysis to construct the D2SFCA, which is used to measure accessibility at the street scale within the study area. The overall process is as follows:

• The first step is to calculate the supply-demand ratio of facilities, that is, the service capacity of the facilities. This step has been explained in the previous section.
• The second step is to calculate the accessibility of each grid’s centroid. This calculation still requires network analysis. The study employs a point-based method, where the accessibility of a point is expressed by calculating the total per capita service capacity of leisure facilities within the range of the centroid. The calculation formula is shown in Equation (3):

  $$A_i=\sum _{in{d}_{ij}\le d_0}{R}_j G\left(d_{ij},d_0\right).$$

  (3)
• In the equation, $A_i$ represents the accessibility of leisure facilities at centroid $i$. The larger the value of $A_i$, the better the accessibility of leisure facilities at that point. $d_{ij}$ represents the path distance between centroid i and facility $j$, while $d_0$ represents the maximum walking distance threshold. Parks beyond this threshold do not influence the accessibility of the service point. $G\left(d_{ij},d_0\right)$ is the Gaussian decay function, which models the decreasing service capacity of leisure facilities as the distance from the service point increases.
• The calculation method of the Gaussian function is shown in Equation (4):

  $$G\left(d_{ij},d_0\right)=\left\{\begin{array}{lll}{\displaystyle \frac{e^{-\frac{1}{2}\times {\left(\frac{d_{ij}}{d_0}\right)}^2}-e^{-\frac{1}{2}}}{1-e^{-\frac{1}{2}}}},& & \mathrm{i}\mathrm{f}{d}_{ij}\le d_0\\ & & \\ 0,& \mathrm{i}\mathrm{f}{d}_{ij}>d_0& \end{array}\right.$$

  (4)
• Subsequently, the Jenks natural breaks classification method was applied to categorize the accessibility of each sampling point, thereby dividing the accessibility levels of public facilities across the study area into five classes. The accessibility values were visualized on a 200 m fishnet grid, where color gradients were used to represent different levels of accessibility.
• To more intuitively illustrate spatial variations within the study area, the grid map was further processed using kernel density estimation (KDE) to generate a heatmap depicting the spatial distribution of accessibility. Based on administrative boundaries, the mean accessibility values derived from the kernel density map were then calculated for each sub-district. These mean values were classified into five categories—very low, low, moderate, high, and very high—using the natural breaks method. This approach minimizes the influence of differences in population size or land area between the two types of recreational facilities, ensuring that the results are comparable across the study area.
• As shown in

  Table 10. Evaluation of Walking Accessibility of Study Sub-districts Calculated Using the D2SFCA Method. (Note: The red areas in the figure indicate the administrative boundaries of the streets studied in this paper).

  Element Legend	Facility Accessibility Represented on 200 m Fishnet Grid	Street Accessibility Calculated Using the D2SFCA Method
  Parks, Green Spaces, and Open Spaces
  Gyms

  , the results reveal that within the study area, Baizhifang Sub-district demonstrates the highest level of accessibility to parks and open spaces, whereas Financial Street Sub-district exhibits the lowest. In contrast, the accessibility to gyms shows an almost opposite pattern, indicating a complementary spatial relationship in the demand for these two types of recreational facilities.
• The computed accessibility results correspond well with the actual conditions of the three sub-districts, particularly regarding the number of parks, green spaces, and gyms, as well as their per capita service capacity. This consistency suggests that the accessibility evaluation framework effectively reflects the real-world distribution and utilization of recreational resources within the urban area.

2.7. Objective Built Environment Data—Development Intensity

2.7.1. Method of Land Use Mix Analysis

The land use mix (LUM) reflects the degree of functional diversity within an urban area, serving as an important indicator for assessing spatial structure and urban vitality. In this study, the entropy index method was adopted to quantify the degree of land use mix within the study area. The entropy index is widely used in urban studies due to its capacity to objectively measure the heterogeneity of land use categories within a given spatial unit. The calculation formula is shown in Equation (5):

$$H={\displaystyle \frac{-{\sum }_{i=1}^n{P}_i\ln P_i}{\ln n}},$$

(5)

where $H$ represents the land use mix index, $P_i$ denotes the proportion of the $i^{th}$ land use type within the total area of the spatial unit, and n is the total number of land use categories. The value of $H$ ranges from 0 to 1, where a higher value indicates a greater degree of land use diversity, while a lower value suggests land use homogeneity.

To ensure comparability and spatial consistency, all land use data were classified into five major categories: residential, commercial, public service, green space, and transportation. The calculation was conducted at the block level using GIS-based spatial analysis. Subsequently, the results were visualized through spatial interpolation and natural breaks classification, enabling the identification of areas with varying degrees of land use mix across the study area. This approach provides a quantitative foundation for analyzing the relationship between land use composition and the accessibility of recreational facilities, thereby contributing to a more comprehensive understanding of urban spatial equity and functional integration. As shown in Figure 6, Xinjiekou Sub-district exhibits the highest value, while Baizhifang Sub-district shows the lowest.

2.7.2. Other Indicators of Development Intensity

The floor area ratio, Urban Density, and the number of bus stops were calculated using GIS-based spatial data, as shown in

Table 11. Comparison of Development Intensity Attributes Across Study Sub-districts. (Note: The red areas in the figure indicate the administrative boundaries of the streets studied in this paper.)

Data Item Sub-District	Xinjiekou Street	Financial Street	Baizhifang Street
Urban Density	36.1%	31.4%	26.6%
Floor Area Ratio (FAR)	1.23	1.89	1.44
Number of Bus Stops (POI Map)

.

3. Results

In the regression models, this study classifies leisure physical activity levels into three ordered categories: low, medium, and high, as the dependent variable. Independent variables are constructed according to Models 1, 2, and 3 (as shown in

Table 12. Regression Models for Leisure-Time Physical Activity.

			Model 1		Model 2		Model 3
			B	Exp(β)	B	Exp(β)	B	Exp(β)
Independent Variables			Socioeconomic Attributes		Socioeconomic Attributes + Objective Built Environment		Socioeconomic Attributes + Objective Built Environment + Subjective Built Environment
Socioeconomic Attributes	Gender (Reference = Female)		−0.147	0.863	−0.166	0.847	−0.193	0.825
	Living with Children (Reference = Co-residing)		−0.044	0.957	−1.868	0.154	−2.112	0.121
	Education Level		0.505 **	1.657	0.16	1.173	0.034	1.035
	Health Status (Reference = Healthy)		−0.481 **	0.618	−0.380 **	0.683	−0.201	0.818
	Monthly Income (Reference = ≤3000 RMB)
	12,001 RMB		2.44 ***	11.574	1.74 ***	5.706	1.412 **	4.104
	9001–12,000 RMB		−3.014 ***	0.049	−3.64 ***	0.026	−3.342 ***	0.035
	6001–9000 RMB		0.137 *	1.147	0.713	2.04	0.745	2.106
	3001–6000 RMB		−0.287	0.751	−0.896	0.408	−1.165	0.312
	Age (Reference = ≤6 years)
	≥66 years		3.163 ***	23.547	4.116 ***	61.377	3.809 ***	45.138
	41–65 years		−2.251 ***	0.105	−1.647 ***	0.193	−2.55 ***	0.078
	18–40 years		0.464	1.591	1.574 **	4.825	1.573 **	4.819
	7–17 years		0.671	1.956	1.173	3.231	0.586	1.796
Objective Built Environment	Accessibility of Leisure Facilities	Accessibility of Parks, Green Spaces, Plazas, and Open Spaces			1.163 **	3.2	2.896 ***	18.102
		Accessibility of Gyms			2.312 *	10.093	1.447 **	4.253
	Public Transportation Convenience	Number of Bus Stops			0.454 ***	1.575	0.449 ***	1.567
	Development Intensity	Land Use Mix			0.820 **	2.27	1.078 **	2.939
		Urban Density			−11.435 **	0.001	−16.918 ***	0.001
		Floor Area Ratio			0.525	1.689	−0.035	0.966
Subjective Built Environment	Accessibility Satisfaction						0.801 **	2.229
	Safety Satisfaction						0.602 **	1.826
	Attractiveness Satisfaction						0.962 **	2.617
	Nagelkerke R2		0.695		0.735		0.748
Sample Size			1072		1072		1072

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

). By controlling for respondents’ socioeconomic characteristics, the study analyzes the impact of both subjective and objective built environment factors on physical activity.

Table 12 presents the analysis results of the Leisure Physical Activity Ordered Logit Model: the Nagelkerke R² value for Model 1 is 0.695, indicating that the selected socioeconomic attributes explain 69.5% of the variance in leisure physical activity. From Model 1 to Model 3, with the inclusion of subjective and objective built environment variables, the relative increase in the Nagelkerke R² value is (0.748 − 0.695)/0.695 = 7.6%, indicating that built environment variables have a limited independent impact on leisure physical activity compared to socioeconomic attributes. Meanwhile, socioeconomic attributes have a more significant influence on leisure physical activity compared to built environment variables. This supports Giles’ research conclusion that the impact of built environment factors on physical exercise behavior is weaker than that of individual and social factors [21].

The three models reveal a hierarchical structure of factors influencing leisure physical activity, with progressively enhanced explanatory power as environmental variables are incorporated. Model 1, containing only socioeconomic attributes, achieves a Nagelkerke R² of 0.695, indicating that individual characteristics alone explain approximately 70% of the variance in LTPA levels. The addition of objective built environment variables in Model 2 increases R² to 0.735, representing a relative improvement of (0.735 − 0.695)/0.695 = 5.8%. The subsequent inclusion of subjective built environment variables in Model 3 further raises R² to 0.748, a relative increase in (0.748 − 0.735)/0.735 = 1.8%. These incremental improvements suggest that while built environment factors contribute meaningfully to explaining LTPA patterns, socioeconomic attributes remain the dominant determinants.

The parameter estimates can be interpreted within the context of the three study neighborhoods. The significant positive coefficient for park accessibility (β = 2.896, p < 0.01 in Model 3) aligns with observed patterns in Baizhifang Street, where the highest park accessibility corresponds with relatively favorable LTPA outcomes despite the area’s lower socioeconomic profile. Conversely, the significant negative coefficient for urban density (β = −16.918, p < 0.01) helps explain the relatively lower high-level activity rates in Financial Street (20.3%), where high-density development dominated by office towers may create spatial compression that discourages outdoor recreational activities.

The land-use mix variable shows a significant positive effect (β = 1.078, p < 0.05), which is consistent with Xinjiekou Street’s performance—the area with the highest land-use mix also demonstrates the highest proportion of high-level physical activity (38.7%). This suggests that the diverse functional composition of traditional hutong neighborhoods, despite their spatial constraints, may facilitate physical activity by integrating residential, commercial, and recreational uses within walking distance.

Notably, the moderating role of subjective perceptions becomes evident when comparing Models 2 and 3. After controlling for subjective variables, the coefficient for park accessibility increases from 1.163 to 2.896, and urban density’s negative effect intensifies from −11.435 to −16.918. This pattern indicates that subjective perceptions partially mediate the relationship between objective environmental features and physical activity outcomes. In Baizhifang Street, for instance, despite favorable objective accessibility, residents’ relatively moderate satisfaction scores suggest that perceived environmental quality may attenuate the full potential of physical infrastructure to promote activity.

3.1. Correlation Between Subjective Built Environment and Leisure Physical Activity

The three subjective built environment variables—accessibility satisfaction, safety satisfaction, and attractiveness satisfaction—have a significant impact on leisure physical activity, indicating that leisure physical activity is more strongly influenced by residents’ subjective perceptions of the built environment. Specifically, residents living in environments perceived to have higher accessibility, safety, and attractiveness are more likely to engage in leisure physical activities. This finding supports the research of Brownson and Lin, which discovered that the aesthetic characteristics of neighborhood blocks, streets, and public spaces can enhance residents’ willingness to participate in leisure physical activities [16,22].

The regression coefficients reveal differential effects among the three subjective variables. Attractiveness satisfaction exerts the strongest influence (β = 0.962, Exp(β) = 2.616, p < 0.05), indicating that residents perceiving their neighborhood as aesthetically appealing are 2.62 times more likely to engage in higher levels of physical activity. Accessibility satisfaction shows a moderate effect (β = 0.801, Exp(β) = 2.229, p < 0.05), while safety satisfaction demonstrates a relatively smaller but still significant effect (β = 0.602, Exp(β) = 1.826, p < 0.05).

These findings can be contextualized through the street-level satisfaction data presented in Table 6. Xinjiekou Street, despite having the lowest scores in facility accessibility satisfaction (0.14) and spatial safety (0.18), achieves the highest proportion of high-level physical activity (38.7%). This apparent paradox is explained by its exceptionally high scores in comfort (1.03) and sports & fitness satisfaction (1.01)—both substantially higher than the other two streets. This pattern suggests that perceived environmental attractiveness and comfort can compensate for deficiencies in accessibility and safety perceptions, supporting the regression finding that attractiveness satisfaction has the strongest effect on LTPA. Conversely, Financial Street demonstrates moderate-to-high satisfaction across most dimensions (accessibility: 0.47, safety: 0.60) but achieves the lowest high-level activity rate (20.3%). The relatively lower comfort score (0.36) in this high-density commercial district may suppress physical activity engagement despite favorable accessibility perceptions. Baizhifang Street shows balanced satisfaction scores across all dimensions with the highest overall average (0.56), corresponding to an intermediate activity level (33.5% high-level), suggesting that consistent moderate satisfaction across multiple dimensions supports sustained physical activity engagement.

From Model 2 to Model 3, when controlling for subjective built environment variables, the significance of some objective built environment variables (such as park accessibility and urban density) increased. However, not all objective variables became significant (e.g., land use mix). This suggests that the subjective built environment acts as a mediator in the relationship between the objective built environment and physical activity: the objective built environment has both direct effects on physical activity and indirect effects through the subjective built environment. In other words, physical environment characteristics must be mediated by individual psychological assessments before they can translate into behavioral responses.

3.2. Correlation Between Objective Built Environment and Leisure Physical Activity

The objective built environment has a significant impact on leisure physical activity, particularly in terms of facility accessibility, development intensity, and public transportation.

3.2.1. Accessibility of Leisure Facilities

Table 12 shows that the accessibility of leisure facilities is positively correlated with the level of leisure physical activity. That is, residents are more likely to use open spaces (such as parks and gyms) within walking distance. This supports the core concept of Newman’s “Defensible Space Theory”: when fitness facilities are deeply integrated into residents’ daily activity circles, the sense of spatial ownership effectively translates into motivation for physical activity [23].

3.2.2. Development Intensity

Table 12 shows that urban density has a significant negative inhibitory effect on leisure physical activity. After adding subjective environmental variables, the negative effect of urban density is further strengthened, indicating that the deterioration of subjective environmental perceptions (such as a lower sense of safety) in high-density areas exacerbates the inhibition of leisure physical activity. The impact of floor area ratio (FAR) on leisure physical activity is not significant, likely because high FAR areas are often accompanied by dense bus stops and commercial facilities, which dilute the explanatory power of FAR. Land use mix has a significant positive effect on leisure physical activity, similar to Frank’s conclusion [24]: the higher the level of mix, the more it can stimulate residents’ willingness to engage in physical activity.

3.2.3. Public Transportation

This study uses the number of bus stops to represent the convenience of public transportation for residents. Table 12 shows that the convenience of public transportation has a significant positive impact on leisure physical activity. This confirms that an increased number of bus stops provides residents with a more convenient green transportation environment, thereby promoting an increase in physical activity levels.

3.2.4. Integrated Analysis of Objective Built Environment Variables

Synthesizing the above findings, a clear hierarchy of objective environmental influences emerges. Facility accessibility variables demonstrate the strongest effects, with park accessibility (β = 2.896, Exp(β) = 18.102) and gym accessibility (β = 1.447, Exp(β) = 9.187) showing substantial positive associations with LTPA. Development intensity variables show mixed effects: land-use mix positively influences activity (β = 1.078), while urban density exerts a strong negative effect (β = −16.918), and floor area ratio shows no significant relationship.

The street-level data illustrate these relationships concretely. Baizhifang Street combines the highest park accessibility with the lowest urban density (26.6%), creating favorable conditions for outdoor physical activity despite its aging infrastructure and lower-income demographic profile. In contrast, Financial Street’s combination of lowest park accessibility, highest floor area ratio (1.89), and moderate urban density (31.4%) creates an environment that objectively constrains leisure physical activity opportunities, reflected in its lowest high-level activity rate (20.3%).

Xinjiekou Street presents an instructive intermediate case. Despite having the highest urban density (36.1%)—typically associated with suppressed physical activity—the area achieves the highest activity rates. This outcome can be attributed to compensating factors: the highest land-use mix enabling diverse activity opportunities within walking distance, and moderate park accessibility. This pattern suggests that negative density effects can be mitigated through strategic land-use planning that integrates recreational opportunities into the urban fabric.

The interaction between objective and subjective variables merits particular attention. Comparing coefficient changes from Model 2 to Model 3 reveals that controlling for subjective perceptions amplifies the effects of objective variables: park accessibility coefficients increase from 1.163 to 2.896, and urban density effects intensify from −11.435 to −16.918. This pattern indicates that subjective perceptions partially suppress the apparent effects of objective features in Model 2, and that the true magnitude of objective environmental influences becomes visible only when perceptual mediators are accounted for.

3.3. Correlation Between Socioeconomic Attributes and Leisure Physical Activity

Socioeconomic attributes have a significant impact on leisure physical activity, particularly in terms of education level, income, and age.

3.3.1. Education Level

The level of education shows significance in Model 1, but after adding objective and subjective environmental variables in Models 2 and 3, its effect is diluted or mediated. This suggests that individuals with higher education levels are more likely to choose residential areas with higher environmental quality, and it is the high-quality living environment that enhances their interest in leisure physical activity, rather than education level itself directly driving physical activity behavior.

3.3.2. Income Level

In Models 1–3, high-income groups (above 12,001 yuan) consistently show significance, but the effect gradually decreases, with subjective environmental variables partially mediating the impact of income. This suggests that the promotion of physical activity by high-income individuals is not merely a direct result of economic advantages, but is indirectly realized through the choice of higher-quality living environments (optimizing the objective environment and enhancing subjective satisfaction). Additionally, the physical activity participation rate of the moderately high-income group (9001–12,000 yuan) is significantly lower than that of the reference group (below 3000 yuan). This is because individuals in the moderately high-income group, often employed in high-intensity professions such as finance and IT, face significant constraints on their leisure time. Their income level not only fails to support the demand for living in high-end communities but also requires them to bear high commuting costs, leading to a reduced willingness to engage in physical activity [15].

3.3.3. Age

In Models 1–3, the proportion of individuals aged 66 and above participating in high-level leisure physical activity is the highest within this group. This group shows a positive effect in their perception of the objective environment, but this effect weakens when subjective perceptions are introduced. This suggests that the elderly actively engage in physical activity based on health needs (such as chronic disease management), and further optimizing subjective environmental perceptions (such as enhancing walking safety) can unlock their potential for more physical activity.

Among the 41–65 age group, the highest proportion falls into the low-level leisure physical activity category. This is due to the negative impact of work and family responsibilities occupying leisure time. Convenient facilities in high-density areas (such as community gyms) help reduce time costs, and this convenience gradually diminishes the negative impact. However, this group generally has low satisfaction with safety and attractiveness, as evidenced by field surveys where issues such as inadequate nighttime lighting and severe equipment aging continue to suppress their willingness to engage in physical activity.

After gradually incorporating both objective and subjective built environment variables, the 18–40 age group exhibits a positive moderating effect on environmental perception. This reflects the interest of the younger group in high-quality fitness services (such as personal training courses) and their willingness to pay a premium, which indirectly increases their level of leisure physical activity.

4. Discussion

Based on the above analysis of the regression model, the following results are obtained:

• Residents’ subjective perception of the environment modifies their responses to spontaneous physical activity through cognitive evaluation. Satisfaction with the perceived built environment significantly influences willingness to engage in physical activity and can reinforce the facilitating effect of the objective environment. Key dimensions include safety (e.g., nighttime lighting, public security), accessibility (e.g., whether a park is reachable within a 15 min walk), and attractiveness (e.g., quality of greenery, street cleanliness). Among these factors, improvement in satisfaction with attractiveness exerts the most direct effect on promoting exercise. Residents’ subjective perceptions serve as a key mediating factor in the development of healthy cities. Safety and attractiveness enhance psychological comfort, thereby encouraging spontaneous physical activity. These findings suggest that street-level physical environment improvements should be implemented in tandem with perception optimization.
• The objective environment shapes opportunities for basic physical activity. Accessibility to open spaces, land use mix, and the number of bus stops—all components of the objective built environment—positively facilitate leisure physical activity. However, excessively high urban density (e.g., areas with dense high-rise residential buildings) may inhibit residents’ motivation to go outdoors due to feelings of spatial compression and reduced sense of safety. Among these factors, the positive effect of accessibility to recreational facilities is the most pronounced, indicating that residents are more likely to engage in exercise using conveniently located facilities along their daily activity routes.
• Socioeconomic characteristics play a fundamental role in shaping residents’ physical activity. Indicators such as income, age, and educational level predominantly influence behavioral choices. High-income individuals can indirectly increase their activity frequency by selecting higher-quality communities. Older adults exhibit the highest willingness to engage in physical activity due to health needs, whereas middle-aged individuals show the lowest willingness, largely due to time constraints.
• The promotion of physical activity among residents in urban neighborhoods exhibits a three-dimensional driving pattern: “constraints imposed by the objective environment—modifications by subjective perception—determination by socioeconomic resources.” The study further indicates that, after controlling for subjective environmental variables, the effects of the objective built environment are largely more pronounced. Both factors show partial correlation yet independently influence leisure physical activity. From Model 1 to Model 2, the inclusion of objective built environment variables increases the relative explanatory power for leisure physical activity by (0.735 − 0.695)/0.695= 5.8%. From Model 2 to Model 3, the addition of subjective built environment variables increases the relative explanatory power by (0.748 − 0.735)/0.735=1.8%. The slight change in Nagelkerke R² indicates that, compared to the subjective built environment, the objective built environment more directly affects opportunities for activity (1.8% < 5.8%). In summary, residents’ socioeconomic attributes determine the baseline of their leisure physical activity habits; the objective environment can directly influence the level of their activity, whereas the subjective built environment serves to modify the effects of objective variables.
• Several limitations of this study should be acknowledged. First, the cross-sectional design limits causal inference regarding environment-behavior relationships. Second, while the spatial grid sampling approach ensured geographic coverage, the recruitment of respondents at leisure activity locations—with priority given to residents currently engaging in or having just completed physical activities—may have introduced selection bias toward more physically active individuals. This recruitment strategy was necessary to ensure respondents had direct experience with neighborhood physical activity environments, but may limit generalizability to less active population segments. Third, although the IPAQ is a well-validated instrument and the subjective built environment items were adapted from established scales, this study did not conduct a formal pilot study or reliability assessment (e.g., Cronbach’s alpha) for the current sample. Future research should consider longitudinal designs, probability-based sampling methods, and comprehensive psychometric testing to strengthen the robustness of findings.

4.1. Parallels Between Objective and Subjective Indicators Across Neighborhoods

A systematic comparison of objective measurements and subjective perceptions across the three study neighborhoods reveals instructive convergences and divergences that illuminate the complex environment-behavior relationship.

In Baizhifang Street, objective and subjective indicators demonstrate strong alignment. The area’s objectively measured highest park accessibility corresponds with the highest subjective satisfaction scores for facility accessibility (0.60) and facility reachability (0.68). Similarly, the lowest urban density (26.6%) aligns with the highest spatial safety satisfaction (0.78). This convergence suggests that in residential neighborhoods with relatively homogeneous populations, subjective perceptions tend to accurately reflect objective environmental conditions, and both dimensions consistently support physical activity engagement—explaining the area’s favorable activity outcomes (33.5% high-level) despite its lower socioeconomic profile.

Financial Street presents a contrasting pattern of objective-subjective misalignment. Despite having the highest gym accessibility and moderate objective infrastructure quality, the area achieves the lowest high-level activity rate (20.3%). The subjective data reveal the explanation: comfort satisfaction (0.36) is substantially lower than the other two streets, and overall sports & fitness satisfaction (0.54) ranks second-lowest. This divergence indicates that in high-density commercial districts, objective facility provision alone is insufficient—the perceived quality of the activity environment, particularly thermal comfort and spatial experience in canyon-like streetscapes dominated by high-rise towers, critically mediates the translation of infrastructure into actual use.

Xinjiekou Street exhibits the most complex objective-subjective dynamics. Objectively, the area faces challenging conditions: the highest urban density (36.1%) typically suppresses activity, and facility accessibility ranks only moderately. Subjectively, accessibility satisfaction (0.14) and spatial safety (0.18) are the lowest among the three streets. Yet paradoxically, Xinjiekou achieves the highest proportion of high-level physical activity (38.7%). This apparent contradiction is resolved by examining specific subjective dimensions: comfort satisfaction (1.03) and sports & fitness satisfaction (1.01) are dramatically higher than other areas, suggesting that the traditional hutong environment’s human-scale spatial experience, mature tree canopy, and established community exercise culture create perceived attractiveness that overrides accessibility and safety concerns. This finding powerfully illustrates the regression result that attractiveness satisfaction (β = 0.962) exerts the strongest effect among subjective variables.

4.2. Implications for Context-Sensitive Urban Regeneration

These cross-neighborhood patterns yield differentiated implications for health-oriented urban regeneration strategies:

For historical neighborhoods like Xinjiekou, the priority should not be aggressive infrastructure expansion that might compromise the very environmental qualities (human scale, greenery, community atmosphere) driving high activity rates. Instead, interventions should focus on targeted improvements in perceived accessibility and safety—enhanced wayfinding, improved nighttime lighting, and barrier-free pathways—while preserving the comfort and attractiveness characteristics that distinguish traditional urban fabric.

For administrative-commercial districts like Financial Street, the evidence suggests that simply providing more facilities will not substantially increase activity levels. The critical gap lies in environmental comfort and perceived attractiveness. Interventions should prioritize microclimate improvement (street trees, wind barriers, shading structures), activation of ground-floor interfaces, and creation of pocket parks that humanize the high-rise environment.

For aging residential neighborhoods like Baizhifang, the strong objective-subjective alignment suggests that continued investment in objective infrastructure—particularly maintaining and expanding the park system that residents clearly value—will likely yield proportional activity benefits. The area’s demographic profile (16.0% aged ≥ 66) further emphasizes the need for age-friendly facility design and programming.

These differentiated strategies reflect the core finding that effective health promotion through built environment intervention requires understanding not only what objective features exist, but also how residents perceive and respond to those features within their specific neighborhood context.

5. Conclusions

This study reveals the complex mechanisms through which built environments influence leisure physical activity patterns in Beijing’s urban regeneration contexts, advancing our understanding of environment-behavior relationships through an integrated analysis of objective features and subjective perceptions. Four critical insights emerge from our investigation of 1072 residents across three functionally distinct neighborhoods in Xicheng District. First, socioeconomic attributes—particularly income, age, and education levels—emerge as primary drivers of LTPA engagement, fundamentally determining participation patterns, with middle-to-high income groups showing significantly higher activity rates. This finding underscores that environmental interventions alone cannot overcome socioeconomic barriers, necessitating equity-focused policies that address structural inequalities in health behavior opportunities. Second, among environmental variables, facility accessibility and land-use mix demonstrate the strongest direct effects on LTPA, while transit accessibility shows context-dependent influences that vary across neighborhood types. Third, and most significantly, subjective perceptions play a crucial moderating role: environmental attractiveness satisfaction significantly amplifies the positive effects of objective factors, suggesting that perception-based interventions can enhance the effectiveness of physical improvements. This interactive effect indicates that successful urban regeneration must address both the material quality of spaces and residents’ psychological experiences of those spaces. Fourth, a three-dimensional framework governs LTPA patterns, wherein objective environmental features establish baseline constraints, subjective perceptions modify these influences, and socioeconomic factors ultimately determine engagement capacity. This “objective constraints—subjective modification—socioeconomic determination” model effectively explains how environmental and personal factors interact to shape physical activity behaviors in complex urban systems.

These empirical findings advance theoretical understanding of environment-behavior relationships in three important ways. By empirically demonstrating the interactive effects between objective and subjective built environments, this research moves beyond traditional approaches that analyze these dimensions separately, revealing that analyzing these dimensions jointly provides additional explanatory power beyond separate analyses. The quantitative evidence for the moderating role of subjective perceptions challenges deterministic assumptions in environmental design, showing that satisfaction can amplify or diminish objective environmental influences depending on context. Furthermore, the integrated analytical framework developed here bridges urban planning, environmental psychology, and public health disciplines, offering a comprehensive model for understanding how spatial configurations, psychological responses, and social structures jointly shape health behaviors. This interdisciplinary synthesis provides a more nuanced understanding of the pathways through which urban environments influence population health, with implications extending beyond physical activity to other health-related behaviors.

Translating these insights into practice requires differentiated strategies for health-oriented urban regeneration. Our findings suggest three core approaches that should be adapted to specific neighborhood contexts. First, perception-oriented environmental optimization should enhance nighttime lighting coverage, green landscape quality, and street cleanliness while establishing community-based security mechanisms and dynamic safety monitoring systems. Integrating environmental attractiveness indicators into healthy city evaluation frameworks ensures that subjective quality receives systematic attention in planning processes. Second, spatial provision strategies must be differentiated by neighborhood type: implementing “15 min fitness circle” standards with mixed-use facilities in commercial areas, optimizing density control in residential zones while increasing open space provision, and ensuring continuity of pedestrian networks and vitality of street interfaces throughout. Third, equity-focused interventions should prioritize age-friendly facilities for middle-aged and elderly populations, implement quality improvement initiatives in low-income communities with annual renovation rates of at least 5%, and incorporate fitness facility accessibility into basic public service frameworks to reduce socioeconomic disparities in health opportunities.

Context-specific recommendations emerge from our comparative analysis of three neighborhood types. In historical districts like Xinjiekou, the priority should be enhancing public transport accessibility by increasing bus stops without compromising heritage features, directly addressing the mobility needs of the predominantly middle-aged resident population while respecting architectural conservation requirements. In administrative districts such as Financial Street, focus should shift toward subjective satisfaction improvements through enhanced safety measures, clearer wayfinding systems, and aesthetic upgrades, while adding public recreational facilities to activate diverse population groups beyond the concentrated high-income residents and daytime workers. In residential districts like Baizhifang, embedded open space development with age-friendly facilities becomes critical: establishing community squares and pocket parks within walking distance while optimizing land-use mix can stimulate physical activity among elderly residents who constitute the demographic majority in these aging neighborhoods.

Several limitations of this study suggest directions for future research. The cross-sectional design limits causal inference about environment-behavior relationships, though the statistical associations provide valuable insights for intervention design. Focusing on one district in Beijing may restrict generalizability to other Chinese cities with different densities, climates, and socioeconomic compositions, though Xicheng District’s diversity offers reasonable internal validity. Future investigations should employ longitudinal designs tracking behavioral changes following environmental modifications, expand sampling to multiple cities representing varied urbanization stages and geographic conditions, and explore seasonal variations in LTPA patterns that may interact with built environment effects. The COVID-19 pandemic has potentially altered physical activity preferences and spatial use patterns; investigating these post-pandemic shifts could provide valuable insights for designing resilient urban environments that support health under diverse conditions. Additionally, examining the cost-effectiveness of different intervention strategies would help prioritize resource allocation in resource-constrained regeneration projects.

This research provides critical evidence for implementing health-oriented urban regeneration in high-density metropolitan areas undergoing rapid transformation. By revealing the complex interplay between objective environmental features, subjective perceptions, and socioeconomic determinants, it offers a comprehensive roadmap for creating cities that actively promote physical activity rather than merely accommodating it. As China advances its Healthy China 2030 agenda and Weight Management Year initiatives, these findings can guide evidence-based interventions that transform urban spaces into catalysts for public health. The integrated analytical framework and differentiated strategies developed here demonstrate that effective health promotion through urban design requires moving beyond generic prescriptions to context-sensitive approaches that recognize the diverse needs of different populations and neighborhoods. Ultimately, achieving health equity in urban environments demands coordinated efforts addressing both the material conditions of spaces and the perceptual experiences of residents, supported by policies that reduce socioeconomic barriers to active living. This study contributes to that vision by providing empirical foundations for urban regeneration practices that prioritize population health as a fundamental design objective.