How Does the Built Environment Influence Social Capital in the Community Context: The Mediating Role of Subjective Residential Satisfaction

Abstract

Despite the growing body of literature on the built environment and social capital, there remains a significant gap in understanding the mediating role of subjective residential satisfaction. Examining how residents perceive and experience their environment offers a fresh angle for understanding the intricate relationship between physical spaces and social dynamics. Our study assessed social capital and subjective residential satisfaction through an extensive questionnaire survey conducted across 60 communities, involving 1684 participants in Tianjin’s metropolitan area, China. We evaluated the elements of the built environment using the ‘5D’ framework, and the pathways of influence were examined using a multilevel structural equation model. Our results reveal a notable mediating effect, with subjective residential satisfaction being a key factor in the intricate process by which the built environment affects social capital. The results also indicate that land use diversity negatively impacts social capital, while population density, access to facilities and services, and public transport density have positive effects. These insights offer practical guidance for fostering social capital and community development by considering subjective perceptions. The results enhance our understanding of effective strategies for building social capital through improved socio-spatial interventions.

1. Introduction

The scope of community development spans from tangible, physical aspects to intangible, social components, addressing a diverse array of interdisciplinary concerns. The existing literature has realized the impact of the built environment on social capital, which is a crucial element of community dynamics that supports various economic and social endeavors. While much of the existing literature has focused on the direct effects of the built environment on social capital, there is growing recognition of the importance of subjective factors. However, scholars still have a relatively limited understanding of how these elements influence each other through intermediary relationships and the mediating role of residents’ subjective satisfaction between them. Subjective residential satisfaction refers to residents’ perceptions and evaluations of their living environment, encompassing both physical and social dimensions. These subjective perceptions can significantly influence how residents interact with their environment and with each other, thereby shaping social capital. The absence of a comprehensive understanding of how residents’ views shape social capital outcomes hinders the development of interventions that effectively address both physical and social aspects of community. This intricate issue calls for a multidisciplinary approach, integrating insights from urban design, sociology, and psychology to thoroughly examine and grasp the complex interplay between the built environment and community social dynamics.

This paper seeks to address the existing knowledge gap by exploring how subjective residential satisfaction influences the impact of the built environment on social capital. Our objective is to deepen the understanding of the pathways and mechanisms through which these factors interact and affect each other. By comprehending the role of subjective residential satisfaction, policymakers and urban planners can implement more effective interventions to enhance social capital and cultivate vibrant, cohesive communities. Comprehensive research on this topic will enhance our ability to develop effective strategies for fostering social capital through optimized design and environmental interventions. This further underscores the significance of humanism and civic participation in community planning and governance processes and emphasizes a more human-centered approach to development in the future.

2. The Relationship Between the Built Environment and Social Capital

2.1. The Built Environment

The built environment at the community scale plays a crucial role in shaping residents’ daily experiences and social dynamics. It encompasses various physical elements such as buildings, streets, parks, and public spaces that collectively form the backdrop for human activities and interactions. Stedman’s [1] incorporation of physical attributes in his studies underscores the significance of spatial factors in community environments, emphasizing how the tangible aspects of a place can influence social relationships, community cohesion, and overall quality of life.

Despite the recognized importance of the built environment, measuring its impact presents significant challenges due to its multifaceted nature. The complexity arises from the diverse range of elements that constitute the built environment, including architectural design, urban planning, infrastructure, and land use patterns. The lack of standardized measures for evaluating the built environment hinders comparative studies across different communities and limits the ability to draw generalizable conclusions about the relationship between spatial characteristics and community outcomes. Developing robust and universally applicable metrics for assessing the built environment remains an important area for future research in urban studies and community development. Based on the ‘5D’ elements proposed by scholars such as Ewing and Cervero [2] and Cervero and Kockelman [3], density, diversity, design, destination accessibility, and distance to transit are five key indicators.

• Density: Many studies have shown a significant relationship between density and social capital, but the nature of this relationship remains controversial. Some studies suggest that increased community density generally has a positive impact on social capital. For example, Lau et al. [4] (pp. 163–176) found that more social contacts help alleviate loneliness and enhance trust in the community. Guo et al. [5] explained that high-density areas may provide more opportunities for social activities. However, Brueckner et al. [6] argued that excessive density could lead to urban problems such as congestion, crime, pollution, and public health issues. Wirth [7] pointed out that increased population density might lead residents to withdraw from social interactions, thereby weakening social capital.
• Diversity: Mixed land use is one of the basic land planning principles for maintaining urban vitality [8]. However, some studies suggest that mixed land use may attract too many external visitors, inhibiting the formation of local social relationships and reducing community identity [9]. This could negatively impact residents’ daily lives, as a higher density of points of interest may reduce residents’ sense of security. Some scholars argue that mixed land use inevitably includes facilities that may harm social interactions, while others believe that the high presence of commercial uses may attract too many external visitors, thereby inhibiting the formation of social capital [9].
• Design: Road network design is generally considered to have a significant impact on the built environment. For example, better connectivity in road networks increases walkability, encouraging residents to walk more and providing more opportunities for informal interactions among neighbors. Bonaiuto et al. [10] found that street spaces are important for fostering emotional bonds between residents and their neighbors. Moudon et al. [11] discovered that smaller block sizes increase walking activities and opportunities for residents to interact.
• Destination Accessibility: Van et al. [12] found that residents in communities with higher accessibility to public spaces tend to have higher levels of trust and more opportunities to establish common norms and mutually beneficial relationships. Adlakha et al. [13] studied community environments in India and found that community participation provides opportunities for the elderly to interact with others, while various facilities offer social spaces. Cultural and commercial amenities such as cafes, restaurants, and shopping malls also provide venues for residents to meet and interact, serving as important places for forming and maintaining social relationships.
• Distance to Transit: It is generally believed that the closer residents are to public transport stations, the more likely they are to walk, increasing the chances of interaction. Mouratidis et al. [14] found that accessibility to public transport is positively correlated with community social cohesion. Conversely, inconvenient transport conditions can hinder residents from achieving their travel goals, such as visiting friends and relatives [15].

The ‘5D’ framework is widely recognized and used in urban planning and social science research. It is based on the idea that the physical characteristics of a neighborhood can significantly impact social dynamics. The ‘5D’ framework is applicable across different types of communities and urban settings, making it a flexible tool for analysis. Each dimension of the ‘5D’ framework has been reported to influence social interactions in different ways. For example, higher residential density can increase the likelihood of social encounters and interactions among residents. The ‘5D’ framework thus provides a robust and well-established approach to evaluating the elements of the built environment that are likely to influence social capital.

2.2. Community’s Social Capital

Social capital is a crucial concept for understanding the well-being and development of communities. Scholars generally concur that trust, reciprocity, and social networks constitute the primary components of social capital. Putnam [16] defined community capital as an enabler of collective action, specifically the features of social life—networks, norms, and trust—that enable participants to act together more effectively in pursuit of shared objectives. Dekke and Uslaner [17] conceptualized social capital as the value of social networks and reciprocity. Liu et al. [18] and Li et al. [19] categorized community social capital into trust and social network structure. Zhu [20] operationalized social capital into three dimensions: trust, reciprocity, and social support, thereby constructing a social capital measurement system. Existing studies have employed various methods to measure it, each with its own strengths and limitations (

Table 1. Measurement of Community Social Capital in Existing Studies.

Author(s)	Dimensions	Measuring Indicators/Questionnaire Items
Putnam [16], Liu et al. [18], Li et al. [19]	Social Trust	Informal social capital on which we happen to have reasonably reliable time-series data involves neighborliness, measured by social time spent with neighbors. The number of neighbors that residents frequently acknowledged with a greeting and the number of neighbors considered as friends within a community.
Leyden [21], Zhu [20]	Social Network	Mutual trust among neighbors in the community, shared values, willingness to help each other, and residents’ trust in community organizations.
Liu et al. [18], Zhu [20], Putnam [16]	Community Trust	Resident relationships, children’s safety, and night-time safety, a vibrant civic life in soils traditionally inhospitable to self-government.
Dekke & Uslaner [17], Zhu [20]	Neighborhood Reciprocity	Accessibility of medical facilities, mutual trust in helping, sense of responsibility, neighborhood connection, and community identity.
Zhu [20], Adlakh [13], Mouratidis & Poortinga [14]	Social Support	Family support and community support; social participation in community spaces; feel that neighbors help one another.

). One of the most common approaches is through surveys. Researchers use questionnaires to gather data on aspects such as trust, reciprocity, and participation.

As the built environment encompasses various elements, all of these elements can shape social capital and influence a community’s social dynamics in nuanced ways. Moreover, the impact of the built environment on social capital is a complex and multifaceted phenomenon that extends beyond direct effects. While research has established connections between physical neighborhood characteristics and social outcomes, the intricate pathways through which these influences occur remain largely unexplored. While many studies have examined the direct effects of the built environment on social capital, few have explored the mediating role of other factors. Sun et al. [22] indicated that the impact of the built environment on social capital and related influencing pathways remains vague. This gap limits our understanding of the underlying mechanisms through which the built environment influences social capital outcomes. To address this gap, future research should focus on uncovering the mediating factors that link the built environment to social capital outcomes.

2.3. The Mediating Role of Subjective Residential Satisfaction

The dynamic interplay between physical spaces and social processes in communities is often compounded by the subjective experiences of residents. While the existing literature provides valuable insights into how the built environment influences social capital, there is a growing recognition of what factors play a critical mediating role in this complex process. Subjective residential satisfaction can be a factor worth assessing in the intricate interplay between subjective and objective assessments. Moreover, according to Yoo and Lee [23], objective measures of the neighborhood built environment do not necessarily correspond to residential perceptions of these qualities. Some studies have linked the subjective satisfaction of residents with social capital as part of a valuable evaluation of their living experience, while others do not, and there is no consensus in academia [24,25]. Some studies discussed residential satisfaction on the individual level, household level, and neighborhood level. According to Bonaiuto et al. [10] the assessment of perceived residential environment quality often includes features of urban space, social relation and place attachment, services and facilities. Wang et al. [26] developed three distinct but interconnected components of residential satisfaction: individual housing, neighborhood amenities, and social environment. Previous studies on community social capital, though broadly linked with the built environment, rarely notice the role of subjective residential satisfaction, or perceptions of the environment, as crucial independent variables in the complex socio-spatial relationship.

Researchers have begun exploring how social capital is formed under the influence of the built environment, introducing subjective residential satisfaction as intermediary variables. Guo et al. [27] discovered that subjective impressions act as a mediator between the built environment and community attachment in a study focusing on elderly individuals in Hong Kong. The extent of mixed land use was found to have an overall positive impact, partly influenced by subjective perception of the community environment. Du et al. [28] enriched the literature by confirming the mediation effect of community satisfaction as part of the chained pathways between community environment and sense of belonging. A study conducted by Yoo and Lee [23] revealed that the perceived characteristics of the local community environment have a more statistically significant impact than the objective features themselves. Hidalgo and Hernandez [29] noted that while previous research concentrated on social elements, individuals also form unique subjective impressions of their communities’ spatial elements. There is a need for a more nuanced understanding of how subjective perceptions of the environment interact with objective measures. Current research often treats these as separate constructs, neglecting their interplay.

2.4. Moderating Variables

The relationship between the built environment and subjective residential satisfaction may not be easily discernible. Residents’ perceptions are influenced by a complex interplay of factors, including personal preferences, cultural background, and social context [24]. For instance, while some residents may value high-density urban environments for their convenience and vibrancy, others may prefer low-density suburban areas for their tranquility and space [27]. This variability in perceptions underscores the importance of considering subjective factors in understanding the impact of the built environment on social capital. Socio-demographic factors, such as age, gender, income, education, and residency status, play a significant role in shaping individuals’ perceptions and experiences of their living environment. These factors can moderate the relationship between the built environment, subjective residential satisfaction, and social capital. Understanding how socio-demographic factors influence these relationships is crucial for residential satisfaction and foster social cohesion. For example, older adults may place a higher value on accessibility and safety, while younger residents may prioritize amenities and recreational facilities [30]. It is expected that well-educated residents may have adapted to their environment and developed higher satisfaction [31]. However, residents may become habituated to their environment and develop a sense of complacency over time. This can lead to a decrease in satisfaction levels as they become less sensitive to improvements or deterioration in the built environment [24]. Studies have shown that older residents tend to have higher levels of satisfaction with their neighborhoods, possibly due to a stronger sense of place attachment and familiarity [32]. In contrast, younger residents may experience lower satisfaction due to higher expectations and more dynamic lifestyle needs, and this has less contribution to the development of social capital. Individuals with a higher education background, particularly those in urban areas, often prioritize career and personal development over community engagement, leading to lower levels of social capital formation [33]. The roles of age and education in the complex relationship remain vague.

We constructed a new conceptual model by incorporating subjective residential satisfaction as a mediating variable and adding age and education as moderating variables (Figure 1). Based on the theoretical framework, the following hypotheses are proposed in this study:

Hypothesis 1. 

The built environment (BE) has a significant effect on social capital (SC). The impact is complex and varies depending on specific built environment variables, with both positive and negative influences.

Hypothesis 2. 

Subjective residential satisfaction (SRS) has a significant effect on social capital (SC). Communities with greater perceived residential satisfaction tend to exhibit stronger social capital.

Hypothesis 3. 

Subjective residential satisfaction (SRS) mediates the relationship between the built environment (BE) and social capital (SC). The way the built environment objectively influences residents’ subjective views plays a role in developing social capital.

Hypothesis 4. 

Socio-demographic factors, namely age and education, have significant moderate effects on social capital. Elderly individuals exhibit higher levels of satisfaction and have a stronger predictive impact compared to younger people. Additionally, those who have received higher education tend to have a more positive influence on social outcomes.

The built environment encompasses various physical attributes such as density, diversity, design, destination accessibility, and distance to transit. These attributes shape residents’ daily experiences and interactions within their communities. Subjective residential satisfaction, which reflects residents’ perceptions and contentment with their living environment, has been influenced by these physical attributes but may alter the original results.

3. Research Methods

3.1. Data Collection and Sources

A questionnaire survey was conducted within the metropolitan core of Tianjin (Figure 2) between 2018 and 2020. The dataset’s comprehensive local coverage across 60 communities and multi-year collection period provide a robust foundation for statistical inference. The selection of communities was intended to provide a representative sample of the diverse urban environments within this region. The sample communities were chosen based on two principles: having physical environmental characteristics that are rich and diverse and having locations that are geographically dispersed. This also increases the likelihood of capturing the diversity of urban conditions, including variations in population density, land use, and socioeconomic factors. We confirm that these communities represent diversified locations, construction periods, and development models, which can avoid a small-scale sample’s limited representativeness. We used a stratified sampling approach to ensure representation of different types of communities. Each survey area had a sufficient sample size. The survey included questions regarding respondents’ personal and family information, subjective residential satisfaction, and community social capital. A total of 1684 valid questionnaires were collected by the end of 2020.

3.1.1. Local Built Environment: Indicators and Measurements

• Density: Micro-level population density data were sourced from Worldpop, which provides global population grid maps based on publicly available population data (grid size: 100 m × 100 m).
• Diversity: Land use mix entropy (Entropy Index) was used to measure diversity, calculated based on the proportion of different POI types within a 500 m buffer zone. It is calculated as in Equation (1)

  $$\mathrm{E}\mathrm{I}=\sum {\mathrm{S}}_{\mathrm{i}}\times \ln (1/{\mathrm{S}}_{\mathrm{i}})$$

  (1)

  Si—the proportion of the number of POIs of category i to the total number of POIs within each 500 m buffer. The higher the entropy value, the closer the proportion of each type of POI in the study unit, and the more balanced the land use, that is, the higher the land use diversity.
• Design: Intersection density was used as an indicator of design, with higher intersection density indicating better street connectivity. The intersection density is calculated as in Equation (2)

  $$\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{d}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}=\mathrm{c}/\mathrm{A}$$

  (2)

  c—number of roadway intersections in each study unit (n); A—area of each study unit (m²). High intersection density characterizes better street connectivity, and it is generally believed that better street connectivity better meets the needs of people’s daily activity destinations, and improves the convenience and willingness of residents to travel.
• Destination Accessibility: Measurement of the respective total number of categories including parks and green spaces, commercial facilities (restaurants, cafes, supermarkets, markets, shopping centers, convenience stores, etc.), and public service facilities (hospitals, schools, community service facilities, etc.) within each community’s 15 min walkable range. Accessibility is the sum of the accessibility of each type.
• Distance to Transit: Number of accessible bus stops within walkable distance, measured by the number of bus stations within the study unit. It is calculated as in Equation (3)

  $$\mathrm{B}\mathrm{u}\mathrm{s} \mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{d}\mathrm{e}\mathrm{n}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{y}=\mathrm{b}/\mathrm{A}$$

  (3)

  b—number of transit stops in each study unit (pcs); A—area of each study unit (km²).
• Enclosed pattern: The study’s measurement system for the built environment incorporates the characteristics of gated communities, which are prevalent in Asia, particularly in China. The city is now a mixture of new gated housing types and old lanes, alleys, and open blocks. The enclosed degree was measured based on physical barriers and control measures, such as walls, gates, and security checks (

  Table 2. Indicators for Community Pattern’s Enclosed Degree.

  Dimension	Observation Indicator	Scoring Method
  Physical Measures	The community has walls or fences at its boundaries.	Score 1 if present in the community, 0 if not
  	There are signs at the entrance restricting strangers from entering.
  	There are electronic surveillance devices inside and outside the community.
  Control Measures	Visitors need to be questioned and registered when entering the community.
  	There are security guards at the entrance.

  ).

3.1.2. Social Capital: Indicators and Measurements

Measuring social capital in this article was through indicators. The indicators were divided into two groups, namely community interaction and community trust, and each group contained four items and was tested among all the residents (

Table 3. Indicators for Social Capital.

Dimension	Item	Measure
Community Interaction	Interact with your neighbors, such as chatting and having casual conversations in daily life.	1 = Never interact, 2 = Rarely interact, 3 = Just know each other but don’t interact, 4 = Interact quite a lot, 5 = Interact a great deal
	Neighbors often take care of each other.	1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree
	There is a lot of interaction among neighbors.
Community Trust	Most residents can be trusted.	1= Strongly distrust, 2 = Distrust, 3 = Neutral, 4 = Trust, 5 = Strongly trust
	How much do you trust the neighborhood committee/village committee staff?
	How much do you trust the property service staff?
	How much do you trust the members of the homeowners’ committee?
	How much do you trust the community workers?

Sample Size: N = 1684.

). All the indicators were measured based on a 5-point Likert scale, with all items weighted equally. The mean score of the items was to represent the level of social capital in the sample communities.

3.1.3. Subjective Residential Satisfaction: Indicators and Measurements

The mediating variable subjective residential satisfaction (SRS) was evaluated across three aspects: satisfaction with the environment, satisfaction with services, and satisfaction with livability. This assessment comprised a total of nine questions (

Table 4. Indicators for Subjective Residential Satisfaction.

Evaluation Dimension	Evaluation Indicator	Evaluation Method
satisfaction with the environment	The community has excellent sanitation.	Likert Scale, scored from 1–5
	The community has excellent greening.
satisfaction with services	The schools near the community are very good.
	The medical facilities near the community are very convenient.
	There are good vegetable markets/convenience stores in and around the community.
	There are good restaurants in and around the community.
	The public transportation around the community is very convenient.
satisfaction with livability	The community is very suitable for children to grow up in.
	The community is very suitable for the elderly to live in.

Sample Size: N = 1684.

). All the questions were measured by using the method of a 5-point Likert scale. Each item was given a score from 1 to 5. The SRS measure result was the mean score of all the nice items.

3.2. Structural Equation Modeling

Structural equation modeling (SEM) is a widely used technique for evaluating mediation effects, capable of addressing both nested data structures and mediation effect estimation simultaneously. It has extensive applications in community-related research in psychology and sociology. We utilized the data gathered from the survey and employed the SEM approach to examine the relationships among these three variables, testing whether subjective residential satisfaction serves as a mediator.

The effects of socio-demographic variables were also taken into account, and a small number of variables were selectively added to the structural equation model to verify their effects. To gain a deeper understanding of the built environment’s impact pathway, reduce bias from missing variables, and more precisely estimate relationships, this study also integrated multilevel modeling to investigate the influence and pathways of both community and individual-level variables on social capital. Our SEM test was conducted in four phases: (1) null model evaluation; (2) examining the direct effect c of the independent variable (BE) on the dependent variable (SC); (3) assessing the direct effect a of the independent variable on the mediating variable subjective residential satisfaction (SRS); and (4) analyzing the effects c and b of both the independent and mediating variables on the dependent variable.

The null (a.k.a. intercept) model of multilevel modeling was used to test whether the data have a nested structure, using the following test formula:

$$\mathrm{A}={\mathsf{\alpha }}_{\begin{array}{c}00\\ \end{array}}+{\mathsf{\mu }}_{0\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(4)

α₀₀ denotes the overall community social capital, μ_0t signifies the random effect associated with the t th community in the sampled community groups, and ε represents the error term.

The variance of community sentiment data was split into two components—within-group and between-group—by calculating the variance on both sides of the equation. To determine if the study data exhibited a nested structure, we employed the intragroup correlation coefficient (ICC) as a metric.

$$\mathrm{I}\mathrm{C}\mathrm{C}={\displaystyle \frac{{{\mathsf{\sigma }}_{\mathrm{u}}}^2}{{{\mathsf{\sigma }}_{\mathrm{r}}}^2+{{\mathsf{\sigma }}_{\mathrm{u}}}^2}}$$

(5)

The ICC ranges from 0 to 1, with values closer to 1 indicating stronger correlations. Generally, an ICC below 0.06 suggests minimal correlation within the dataset. This implies that the research sample maintains individual independence and is suitable for the direct application of single-level regression. Conversely, higher ICC values indicate that variations in the dependent variable are partially attributable to differences between communities. In such cases, employing multilevel modeling or utilizing clustering robust standard errors for estimation has proven more effective and necessary.

The research process was divided into four steps: (1) the null model test; (2) the test of the direct effect c of the independent variable (BE) on the dependent variable (SC); (3) the test of the direct effect a of the independent variable on the mediating variable (SRS); and (4) the test of the effects c and b to examine the role of the independent and mediating variables on the dependent variable.

Since the direct effect c of BE on SC and the direct effect a of the BE on SRS both involve cross-level effects of the stratum 2 variable on the stratum 1 variable, the null model test for the mediating effect needs to be divided into two parts, the first of which is to conduct the null model test with CM as the dependent variable and to calculate the corresponding within-group correlation coefficients, and the second is to conduct the null model test with SRS as the dependent variable. The second part is the null model test with SRS as the dependent variable.

Hierarchical linear analysis can be conducted using mixed modules on software like SPSS, Mplus, and MLwiN. Using the Mplus 8.3 platform, we performed our analysis employing multilevel path analysis. This approach allowed us to simultaneously address the nested structure of the data and the estimation of mediated effects. Maximum likelihood estimation with robust standard errors (MLR) was used to estimate the model parameters because it was robust to non-normality in continuous outcomes.

The results of the null model test for social capital and subjective residential satisfaction were 0.360 and 0.021 (

Table 5. Results of the Null Model Test.

	Community-Level Variance	Individual-Level Variance	ICC
Social Capital	0.023	0.041	0.360
Subjective Residential Satisfaction	0.196	9.219	0.021

), respectively, meaning that 36.0% and 2.1% of the respective differences in social capital and subjective residential satisfaction can be explained by differences between communities. In other words, compared with the single-level model, the multilevel model can explore the influencing factors and path of social capital more effectively. Subjective residential satisfaction is not only affected by the community-level variable of the built environment but also by the individual-level variable of socioeconomic attributes, so it is important to disentangle the variables at different levels. The independent variable BE is at the higher community level, meanwhile, the mediator variable SRS and the dependent variable SC are at the lower individual level; therefore, this paper considers its multilevel mediation effect and adopts the 2-1-1 multilevel mediation model (level-2 independent, level-1 mediator, level-1 dependent).

A multilevel path analysis model combining a multilevel linear regression model and mediation effects was used to measure the impact and path of community-scale and individual-scale variables on residents’ social capital.

The regression model was divided into two layers: the first layer was the individual resident layer (N = 1684) and the second layer was the community layer (N = 60). The individual layer file includes 1684 cases, and testing variables determined at this layer were age, length of residence, social capital, and subjective residential satisfaction. The community layer file includes 60 cases and the following objective variables: population density, land use mixing entropy, intersection density, accessibility to commercial amenities, accessibility to services, accessibility to parks and greenspaces, density of transit stops, and community enclosed pattern.

Firstly, the regression model between the dependent variable (SC) and the independent variable (BE) was established in order to obtain the total effect (coefficient β) of the independent variable on the dependent variable: model 1 considers the total effect of the built environment of the community on the capital of the community only, controlling for the socio-economic attributes of the individual and the household.

$$\begin{gathered} {\mathrm{S}\mathrm{C}}_{\mathrm{i}\mathrm{j}}={\mathsf{\beta }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\beta }}_{1\mathrm{j}}{\mathrm{B}\mathrm{E}}_{1\mathrm{j}}+{\mathsf{\beta }}_{2\mathrm{j}}{\mathrm{S}\mathrm{E}}_{2\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{i}\mathrm{j}} \\ {\mathsf{\beta }}_{\mathrm{i}\mathrm{c}}={\mathsf{\gamma }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\mu }}_{\mathrm{i}\mathrm{c}} \end{gathered}$$

(6)

i and j represent community i and individual j, respectively; SC, BE, and SE represent social capital, built environment, and residents’ socioeconomic attributes, respectively.

In the second step, a regression model is constructed between the mediating variable (SRS) and the independent variable (BE) to obtain the effect (coefficients) of the independent variables on the mediating variables;

$$\begin{gathered} {\mathrm{S}\mathrm{C}}_{\mathrm{i}\mathrm{j}}={\mathsf{\beta }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\beta }}_{1\mathrm{j}}\mathrm{\prime }{\mathrm{B}\mathrm{E}\mathrm{\prime }}_{1\mathrm{j}}+{\mathsf{\beta }}_{2\mathrm{j}}{\mathrm{\prime }\mathrm{S}\mathrm{E}\mathrm{\prime }}_{2\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{i}\mathrm{j}}\mathrm{\prime } \\ {\mathsf{\beta }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}={\mathsf{\gamma }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\mu }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}} \end{gathered}$$

(7)

In the third step, a regression model between the independent variable, the mediator variable, and the dependent variable is developed to obtain the effect of the mediator variable (SRS) on the dependent variable (SC) (${\beta }_{3j}\mathrm{\prime }\mathrm{\prime }$) and the direct effect of the independent variable (BE) on the dependent variable (${\beta }_{1j}\mathrm{\prime }\mathrm{\prime }$).

$$\begin{gathered} \mathrm{S}\mathrm{C}={\mathsf{\beta }\mathrm{\prime }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\beta }}_{1\mathrm{j}}\mathrm{\prime }{\mathrm{\prime }\mathrm{B}\mathrm{E}\mathrm{\prime }\mathrm{\prime }}_{1\mathrm{j}}+{\mathsf{\beta }}_{2\mathrm{j}}{\mathrm{\prime }\mathrm{\prime }\mathrm{S}\mathrm{E}\mathrm{\prime }\mathrm{\prime }}_{2\mathrm{j}}+{\mathsf{\beta }}_{3\mathrm{j}}\mathrm{\prime }\mathrm{\prime }{\mathrm{S}\mathrm{R}\mathrm{S}\mathrm{\prime }\mathrm{\prime }}_{3\mathrm{j}}+{\mathsf{\gamma }}_{\mathrm{i}\mathrm{j}}\mathrm{\prime }\mathrm{\prime } \\ {\mathsf{\beta }\mathrm{\prime }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}={\mathsf{\gamma }\mathrm{\prime }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}}+{\mathsf{\mu }\mathrm{\prime }\mathrm{\prime }}_{\mathrm{i}\mathrm{c}} \end{gathered}$$

(8)

where i and j are community i and individual j, respectively; SC, BE, SE, and SRS represent social capital, built environment, socioeconomic attributes, and subjective residential satisfaction, respectively; γ is the random error term for stratum one; μ is the random error term for stratum two; and ${\gamma }_{ic}$ denotes the random intercept of the i th community. The mediation effect is namely equal to the coefficient product of $\mathsf{\beta }\mathrm{\prime }\ast {\mathsf{\beta }}_{3\mathrm{j}}\mathrm{\prime }\mathrm{\prime }$.

Broadly speaking, researchers employ three primary approaches for analyzing mediation effects: the causal step method, the coefficient product and difference technique, and the bootstrap method [34]. In this paper, we employed the bootstrap method, which operates by treating the initial sample as a complete unit, generating numerous new sub-samples, and conducting statistical analyses through resampling. The bootstrap method’s wide confidence intervals are derived from the actual distribution of the ab product, eliminating the need for distributional assumptions and thus avoiding potential violations associated with the coefficient product test. We operated the test on the Mplus platform was utilized to generate 2000 estimates of the coefficient product by the bootstrap method, with the sample size set at 2000. These estimates were arranged in ascending order, and the values at the 2.5% and 97.5% percentiles formed a 95% confidence interval. The significance of the coefficient was determined by examining this interval: if it contained zero, the coefficient was deemed not significant; if it excluded zero, the coefficient was considered significant.

4. Results

In our sample areas, the average value of population density is 29,800 people per square kilometer. The range spans from a minimum of 300.38 people/km² to a maximum of 87,900 people/km². The mean diversity value of land use entropy in the study areas was 1.86. The entropy measurements ranged from a low of 1.42 to a high of 2.46. Regarding the road intersection density design variable, the highest observed value is 104.96/km², while the lowest density among communities is 7.52/km². The destination accessibility result was cumulative opportunity accessibility, while kernel density was applied for analyzing three types of facilities. Figure 3 displays the measurement results, indicating that overall accessibility to shopping and public service facilities is superior. We also identified five communities that lack parks and green spaces within a 500 m radius. Distance to transport link was covered by the number of accessible bus stops, with an average of 10.3 stops per community in the sample areas. The highest number of accessible bus stops found in the sample communities was 24, while the lowest was 2.

Table 6. Variable Types, Definitions, and Descriptive Analysis of Built Environment.

Variable	Variable Type	Definition	Mean	Standard Deviation	Minimum	Maximum
Population Density	Continuous Variable	Population/Research Unit Area	29,800 people/km2	24,700	300 people/km2	87,900 people/km2
Plot Ratio	Variable Type	Total Building Area/Research Unit Area	1.60	0.57	0.80	3.87
Land Use Mix	Continuous Variable	Mixture degree	1.86	0.22	1.42	2.46
Road Intersection Density	Continuous Variable	Number of Intersections/Research Unit Area	53.48	30.70	7.52 intersections/km2	104.96 intersections/km2
Bus Stop Density	Continuous Variable	Number of Bus Stops/Research Unit Area	10.3	8.78	2	24
Accessibility of Commercial Amenities	Continuous Variable	Distance to amenities	14.54	10.45	3	45
Accessibility of Green Space and Parks	Continuous Variable	Distance to spaces	6.19	8.91	0	43
Accessibility of Public Service	Continuous Variable	Distance to services	26.64	84.96	19	327
Community Enclosed pattern	Continuous Variable	Enclosed Degree	4.60	1.76	1	7

Sample Size: N = 60.

shows the summarized statistics for all the built environment variables, which were all numeric and treated as continuous variables.

The reliability and validity of the survey data were tested by Bartlett’s test of sphericity and the Kaiser–Meyer–Olkin (KMO) measure. The analysis yielded a KMO statistic of 0.857 and a significant Bartlett’s test result (Sig. < 0.05), demonstrating the high validity of the survey’s outcome (

Table 7. KMO and Bartlett’s Test Results.

Kaiser–Meyer–Olkin Measure of Sampling Adequacy	Kaiser–Meyer–Olkin Measure of Sampling Adequacy	0.857
Bartlett’s Test of Sphericity	Approximate Chi-Square	2245.717
	df	190
	Sig.	0.000

). The structural equation model showed good fit characteristics, with our model falling within an acceptable range when evaluated against established standards (

Table 8. Overall Model Fit Test Result.

Fit Index	CMIN/DF	RMSEA	SRMR	SRMR	CFI	TLI
Standard	≤3 for good fit; ≤5 for reasonable fit	<0.08	≤0.05 for good fit; ≤0.08 for reasonable fit	≤0.05 for good fit; ≤0.08 for reasonable fit	>0.90	>0.90
Calculation Results	3.213	0.072	0.021	0.067	0.942	0.913

).

The socioeconomic features of the survey respondents are shown in

Table 9. The socioeconomic features of survey respondents.

Item	Categories	Frequency	Percentage
Gender	Male	923	54.8
	Female	761	45.2
Age	18–29	145	8.6
	30–39	254	15.1
	40–49	381	22.6
	50–59	334	19.8
	60 and above	570	33.8
Educational Level	Never attended school	15	0.9
	Elementary School	69	4.1
	Middle School	394	23.4
	High School/Vocational school	484	28.7
	Associate Degree	303	18.0
	Bachelor’s degree	339	20.1
	Graduate student	80	4.8
Occupation	General staff	510	30.3
	Middle and senior managers	198	11.8
	Retired	893	53.0
	Individual business and freelancers	34	2.0
	Students	45	2.7
	Others	4	0.2

Sample Size: N = 1684.

. People aged 60 and above accounted for 33.8%, presenting an aging society feature. Generally speaking, the activity range of the elderly group is often limited to their residence, and the frequency of use of their community space is relatively higher. In terms of length of residence, the surveyed individuals had lived in the community for an average of 11.55 years, indicating a relatively stable residing status.

According to the regression results (

Table 10. Multilevel Linear Regression Results: The Impacts from BE on SC, with and without the Mediating Variable SRS.

		Model 1 (Social Capital)		Model 2-a (Subjective Residential Satisfaction)		Model 2-b (with Mediating Variable—Social Capital)
		Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error
		B	Standard Error	B	Standard Error	B	Standard Error
Intercept		−0.836 **	0.087	0.148 **	0.193	−0.504 **	0.102
Variables
Community-level Variables	Population Density	0.112 **	0.026	0.061 **	0.064	0.12 **	0.324
	Land Use Mix	−0.272 **	0.103	0.012**	0.008	−0.271 **	0.204
	Road Intersection Density	0.029 **	0.063	0.083 **	0.062	0.03 **	0.101
	Accessibility of Commercial Amenities	0.097 **	0.313	0.255 **	0.386	0.027 **	0.068
	Accessibility of Public Service	0.061	0.211	0.264	0.061	0.176	0.188
	Accessibility of Park Green Space	0.183 **	0.051	0.139	0.084	0.184	0.06
	Bus Stop Density	0.196 **	0.053	0.121 *	0.121	0.128 *	0.088
	Community Enclosed Pattern	−0.018	0.051	−0.042	0.155	−0.018	0.176
Individual-level Variables	age	0.132	0.142	0.014	0.033	0.054	0.039
	education	−0.07	0.04	-0.057	0.012	−0.052	0.193
Mediating Variable	Subjective Residential Satisfaction					0.219 **	0.512
Model Fit	CMIN/DF	3.801		2.231
	CFI	0.901		0.974
	RMSEA	0.071		0.062
	SRMR(within)	0.035		0.021

Sample Size: N = 1684 ** Significance at 0.05 level * Significance at 0.01 level.

), Model 1 suggests that population density, land use mix entropy, intersection density, park accessibility, commercial facility accessibility, and public transport station density were significant BE factors correlated with SC (p  <  0.05).

Model 2-a was a test of the relationship between BE and SRS. The results show that intersection density, park accessibility, public transport station density, and community enclosed pattern all had significant positive effects on SRS (p  <  0.05).

Model 2-b was a test of the relationship between the variables BE and SC after introducing the mediating variable of SRS. SRS has largely contributed to the in-depth process of how the variable BE impacts SC. We noticed that the original effects of the built environment on community social capital were not altered in direction by residents’ subjective satisfaction; however, residential satisfaction could largely intensify or significantly lessen these impacts. As an example, the effects of commercial amenity accessibility are more noticeable among individuals (B = 0.255, p  <  0.05) compared to the community (B = 0.027, p  <  0.05). While population density has not been particularly significant in residents’ minds (B = 0.061, p < 0.05), it does influence the community as a whole by increasing opportunities for interaction (B = 0.12, p < 0.05).

For detailed built environment factors, we found the major impacts to be as follows: (1) Major positive impacts on social capital are from population density (B = 0.12, p < 0.05) and bus stop density (B = 0.128, p < 0.01). (2) Among the factors influencing social capital, spatial diversity (land use mix) stands out as the sole negative contributor (B = −0.271, p < 0.05). Despite its positive association with perceived residential satisfaction, this relationship did not counteract its adverse effect on social capital. (3) Other spatial indicators all have a positive but weak effect on social capital. At the personal level, demographic factors such as age and education did not exhibit notable associations with SC.

The influencing path of the built environment and subjective residential satisfaction on social capital is shown in Figure 4. The empirical evidence gathered allows us to address the proposed hypotheses as follows:

Hypothesis 1 (affirmative). 

The built environment exerts a significant influence on social capital.

Hypothesis 2 (affirmative). 

The subjective satisfaction of residents regarding their living environment has a substantial impact on social capital.

Hypothesis 3 (affirmative). 

The effect of the built environment on social capital is mediated by subjective residential satisfaction, and no significant directional change has been observed on the mediation process. The impact of the built environment on resident’s satisfaction with the living environment is forwarded positively to social capital.

The moderating effects of age and education were found to be weak, leading to the rejection of Hypothesis 4.

5. Discussions

Our approach considers elements such as land use patterns, transportation networks, and public space to understand their collective influence on social interactions, economic activities, and overall quality of life. Although some research has argued that the spatial accessibility of bus stops and commercial centers has no indirect effect on social capital, our study identifies a positive impact, particularly regarding bus stop distribution. This tends to increase trip frequency, especially public trips, and possibilities for social encounters. The influence of most other environmental variables is minimal, which aligns with the majority of existing research findings. Multiple studies have shown the significant impact of population density [35,36]. The effects of higher population densities on social capital can be either beneficial or detrimental [37,38]. While current research has emphasized density’s positive role in social interactions, our results indicate a more complex relationship. Beyond a threshold, increased density can lead to overcrowding and reduced residential satisfaction, negatively impacting social capital. This challenges the assumption that higher density is always beneficial and emphasizes the importance of quality in high-density environments. Our research continues to support the notion that in Chinese urban settings, denser neighborhoods provide more opportunities for residents to interact, thus boosting levels of social capital.

Our findings revealed a unique trend where social interaction and community trust diminished as land use diversity increased. This moderate negative effect was statistically significant, as previously mentioned. It is possible that diverse land uses create fragmented spaces that hinder social interactions or lead to conflicting interests among different user groups. Of all the factors related to the built environment, land use diversity has the most substantial impact. In Western settings, Leyden [21] found that neighborhoods designed for walking and featuring mixed uses fostered better social interactions and stronger community bonds than those reliant on cars and characterized by single-use residential areas. By contrast, our study was rooted in the Chinese environment. In many Chinese cities like our case Tianjin, there is already a significant level of mixed-use development. Commercial and business sectors are often closely distributed along with residential neighborhoods. We suggest that an excessive degree of mixed-use in urban areas might lead to spatial fragmentation and trigger resident complaints regarding noise and safety concerns. Such issues could potentially impede the growth of social capital.

In this study, we found that social capital remains strong in communities in the metropolitan region of Tianjin in China. Only a few residents claimed the cultivation of social capital was not as important to them, or they were indifferent to staying or leaving their communities. Our findings suggest that the built environment influences social capital largely through residents’ subjective perceptions. This underscores the importance of incorporating community members’ psychological expectations and individual experiences into future community development strategies. This result is consistent with the ‘community persistence theory’ proposed by Gans [39]. The community enclosed pattern assessment is a response to the large-scale gated communities that characterize the country and a theoretical extension of the Western built environment measurement system. In our study, no direct evidence was found that gated communities negatively impact the formation of social capital. This is distinct from some existing studies [40,41]. The cornerstone of contemporary urban administration is rooted in the community, which functions as a comparatively autonomous and enduring spatial entity. It also serves as the most basic unit through which governmental bodies at various levels exert their administrative authority in managing urban areas. Social capital creation at the community level fosters an improved social environment and lays a positive groundwork for community growth. Governance systems must acknowledge that building sustainable communities is a multifaceted process, encompassing not only physical infrastructure but also social structures, personal experiences, and emotional well-being.

Our explorations on the socio-spatial nexus aimed at supplementing and advancing the theoretical perspective of community research. Historically, China’s approach to community development has been characterized by a top-down administrative structure, which has often neglected the fundamental aspects of community life, such as communication and social bonds. This is evident in the creation of community living spaces, which frequently relies on inflexible spatial metrics, demonstrating a lack of comprehensive evaluation and action plans rooted in ‘social capital’. The recent urban development plan emphasizes placing residents in a positive social support network and creating harmonious and friendly community emotions. Urban planners should investigate methods to foster community engagement through the design of physical public areas and strengthen residents’ feelings of responsibility and connection to their neighborhoods. This will enhance our ability to develop effective strategies for fostering social capital through optimized environmental design methods.

The research results outline clear strategic guidance for all involved parties. Policymakers can improve residents’ perceptions and satisfaction, thereby fostering the growth of community capital. Specific guidance in neighborhood regeneration plans can include effectively managing population density, thoughtfully designing transportation networks and land use functions, and enhancing the arrangement and quality of public service facilities. Developers can assist by wisely choosing commercial locations. At the same time, residents can actively engage in the process of community planning and governance. These combined efforts can significantly boost residents’ perceptions and satisfaction, which in turn supports the growth of community capital. Future policies should also focus on new development strategies for improving social capital, such as adopting open-block layouts, high-intersection road systems, and open and shared public spaces. For existing communities, attention should be given to the renovation and establishment of community-centered approaches, with attention to optimizing some micro-scale built environmental factors such as road and greening design to improve the perception of residential experiences.

6. Conclusions

By incorporating subjective residential satisfaction as a mediating variable, this study offers a more nuanced understanding of the relationship between the built environment and social capital. Our multilevel structured equation model with an enhanced reliability design approach allows for a more rigorous examination of the pathways through which the built environment influences social capital, providing valuable insights for both researchers and practitioners. Findings from our empirical study confirm that built environment factors such as density, land use diversity, and accessibility of amenities and services have significant impacts on social capital. By examining the impact coefficients, it becomes evident that land use diversity has a more prominent influence compared to other factors, and it also presents a distinct negative trend that could impede social capital in the Chinese context. Subjective residential satisfaction with the community environment demonstrated an obvious positive mediating effect. By uncovering the novel mechanism through which the built environment influences social capital via subjective residential satisfaction, we contribute to a more comprehensive understanding of the interplay between physical spaces and social dynamics within communities. Acknowledging this mechanism is crucial for urban planners, policymakers, and community developers who aim to foster vibrant, cohesive, and resilient communities.

On a broader societal level, our findings highlight the importance of creating inclusive urban spaces. As cities grow, understanding how the built environment strengthens social bonds becomes vital for sustainable communities. This research informs the development of harmonious communities, enhancing residents’ quality of life and social cohesion. Approaching the issue from a psychological perspective to comprehend the subjective experiences of residents can better address the intangible needs of community development. Comprehensive research on this topic will enhance our ability to develop effective strategies for fostering social capital through optimized design and environmental interventions. This further underscores the significance of humanism and civic participation in community planning and governance processes and emphasizes a more human-centered approach to development in the future.

7. Limitations and Future Scope

The study acknowledges limitations such as sample constraints, as reaching marginalized or hard-to-access communities to collect data is a significant hurdle, potentially leading to an unrepresentative sample. Our discovered relationship between land use diversity and social capital is context-dependent, and urban planners should consider local conditions and resident needs when designing mixed-use neighborhoods. In subsequent research, we could investigate the evolving interplay between the constructed environment and community bonds by examining various types of neighborhoods. Future studies might delve deeper into additional factors, such as community empowerment, and examine how cultural and socioeconomic aspects might affect this relationship. This research focuses on Tianjin, a major metropolitan area in China. In the future, we plan to select various large, medium, and small cities to include comparative cases, which will help verify the robustness of our findings and allow us to examine the differences between these cities.