The Impact of the Built Environment on Resident Well-Being: The Mediating Role of Multidimensional Life Satisfaction

Abstract

Well-being is an important goal pursued by humans, and the living environment has a profound impact on various aspects of human health. The objective of this study is to explore the mechanism by which the built environment affects the well-being of residents, specifically how multiple, distinct domains of life satisfaction mediate the effects of diverse built environment features on well-being—a nuanced pathway not yet comprehensively examined. Based on questionnaire data collected from 22 statistical districts in Macau, with a sample size of 1313 individuals, a multilevel linear regression model and mediation analysis were applied (model R² ≈ 47%). When leisure satisfaction is used as a mediator variable alone, the explanatory power of the original model increases the most (from 7.6% to 32%). Complete Mediation via Specific Domains: Health satisfaction fully mediated the effects of intersection density (p < 0.05) and bus stop accessibility (p < 0.05). All four satisfaction domains collectively fully mediated income diversity (Shannon index, p < 0.01). The 14 built environment metrics (5 socioeconomic, 9 morphological) exhibited differential mediation mechanisms: while transportation-related metrics (intersection density, bus stops) primarily operated through health/social satisfaction, diversity indices (income, education, land use) and unemployment rate engaged all satisfaction domains. Some variables showed partial mediation through various satisfaction pathways (p < 0.01–0.05). These findings underscore the necessity of considering multidimensional life satisfaction as critical pathways in urban well-being research and policy.

1. Introduction

There is a significant association between the built environment (artificial environment, including buildings, facilities, and the space between them) of communities and mental health. Existing research has explored various spatial scales of the built environment, ranging from regional and urban scales to community and street scales, as well as smaller and more personalized spaces such as homes, workplaces, and other individualized contexts. Increasingly, studies on urban built environments focus on residents’ mental health [1,2].

Previous research has widely examined important external environmental indicators at the community scale. Cao et al. [3] investigated residents’ well-being from the built environment perspective, considering density, diversity, and design dimensions. They found that the land-use mix and community population density were significantly negatively correlated with well-being. Research in developed Western countries has found that residents living in rural or urban areas of large cities have lower levels of well-being [4]. However, a study in China has shown that the smaller a city is, the happier it is, and there are certain scope limitations [5]. However, studies in China suggest a certain range of urban scales beyond which smaller cities do not necessarily have higher levels of well-being [5]. The built environment significantly impacts well-being at both the urban and neighborhood levels, with varying degrees of influence depending on the level of urbanization [6].

Researchers have conducted a series of discussions on the impact of urban density and diversity on well-being. Studies conducted in the United States indicate that residents’ well-being significantly decreases in cities with higher population density and larger urban scales [7,8]. High-density and fast-paced urban environments can stimulate residents’ senses and increase their psychological burden [9]. A study in Singapore showed that road guidance and connectivity, both in terms of physical and mental health, play secondary roles [10].

The distribution of urban facilities is also an important factor determining the well-being of residents. Researchers have examined compact and sprawling urban forms and analyzed the impact of urban public facilities on residents’ well-being by collecting data from ten cities worldwide [11]. Ting Zhang et al. [12] assessed exposure to the built environment based on the density of four types of public recreational facilities (fitness centers, parks, leisure facilities, and sports facilities) and street centrality. A study on elderly residents in Shanghai revealed that nearby healthcare facilities and other beneficial amenities significantly improved well-being, whereas living in areas dominated by leisure facilities harmed well-being [13]. Using evaluative measures, Sirgy and Cornwell [14] found a positive relationship between residents’ satisfaction with the surrounding physical environment and their well-being. Moreover, scholars have conducted empirical investigations on urbanization and health and proposed plans for promoting health through sustainable urban development, such as comprehensive planning, evidence-based decision-making, and policy implementation monitoring [15].

Life satisfaction is an individual’s comprehensive understanding and judgement of life and is often determined by social activities and the physical environment [16]. Life satisfaction can be subdivided into different dimensions; for example, Mouratidis [17] reviewed the subdivision of life satisfaction, which includes aspects such as travel, leisure, health, residential well-being, social relationships, and emotional experiences, all of which may affect subjective well-being. Existing research has also demonstrated that the built environment impacts different dimensions of life satisfaction. Hershfield’s research revealed that an increase in life satisfaction can increase people’s well-being [18].

The relationship between individual life satisfaction and the built environment in some cases stems from the influence of the built environment on individual behavioral activities [19,20]. Scholars have found that the potential improvement in residents’ self-selection effect is of great help in improving their life satisfaction [21]. Some researchers have conducted studies on residents of cities worldwide, such as Beijing and New York, and found a positive correlation between cultural and leisure facilities, transportation convenience, and residents’ well-being. The reason may be the social connections between residents and society [22].

The above studies have summarized the variables of the built environment and related research topics, and mature research methods have been developed. However, such studies often focus on only one or a few built environment variables. These studies target only a specific group, and the generalizability of the research conclusions is insufficient. The definition and differentiation of built environmental elements remain in the field of morphology, without considering the social impact of artificial environments. Existing research on the impact of the built environment on life satisfaction is insufficient, and the mechanism of action needs further exploration by the academic community. Many studies revolve around the relationship between the built environment and life satisfaction or discuss the impact of the built environment on well-being, with little research exploring the mediating effect of life satisfaction on the impact of the built environment on well-being.

Based on the literature mentioned above, the purpose of this article is as follows:

(1) To explore the mechanism of the effect of the built environment on well-being;

(2) To examine whether leisure, work, social, and health satisfaction mediate the relationship between the built environment and residents’ well-being.

This study included four types of life satisfaction, namely, entertainment satisfaction, social satisfaction, job satisfaction, and health satisfaction, as mediating variables and assumed that the constituent variables of the built environment affect resident well-being through these four types of life satisfaction. Cross-sectional data from questionnaire surveys and geographic analysis data from the study area obtained via ArcGIS 10.7 were utilized to explore the impact of built environment factors on residents’ well-being. To draw research conclusions, the correlation between the built environment, life satisfaction, and well-being was assessed through eight theoretical models using multiple linear regression and mediation effect testing methods.

2. Materials and Methods

2.1. Study Area

The Macau Special Administrative Region (SAR) is located in the southeastern coastal region of China, adjacent to Guangdong Province and west of the Pearl River Delta. The Macau SAR includes 22 statistical zones (Figure 1), covering an area of 33.3 square kilometers with a population of approximately 682,000 [23]. Macau has a total population density of 29,800 people per square kilometer [23], making it one of the most densely populated areas in the world. The sampling was based on population size in every statistical zone, and regions with larger populations were allocated larger sample sizes (Figure 1).

Historically, Macau was involved in the colonial period of Portugal [24]. Its architectural layout blends the Guangdong style and Portuguese style, and due to the application of Portuguese land rights laws and regulations, it cannot form the community concept commonly used in existing research. The study area includes all 22 statistical districts in the Macau SAR, ranging in size from 0.46 square kilometers to 7.2 square kilometers. The area, population size, and facility configuration of each statistical zone are equivalent to the community size, enabling a rough comparison of the size of each statistical district with the community scale studied in previous research [25]. Therefore, statistical zones were selected as the analysis units.

2.2. Data

This study is based on a cross-sectional questionnaire survey conducted between April and June 2022. The total sample size was 1313, representing approximately 0.2% of the total population. The collection of questionnaire data was divided into three stages. The first stage of testing ended in April 2022, and the second stage ended in April and June 2022. The third stage was adjusted based on the proportion of the sample to the actual population and the sample size in the second stage. A questionnaire comprising 81 questions was designed to encompass residents’ personal and family conditions alongside socioeconomic attributes. Well-being alongside entertainment, job, social, and health satisfaction were measured for each participant using a Likert scale. The questionnaire was distributed to residents across 22 statistical zones. These statistical districts cover the entire Macau SAR, ensuring that the survey sampling encompasses the whole region’s population.

The study adopted a stratified sampling method, ensuring that the proportion of the population in each statistical partition to the total population of Macau was basically equal while allowing for small errors. Furthermore, the age group and gender ratio in each statistical partition were allocated according to the age and gender distributions within the region. A tool provided by Raosoft proposed an effective way to calculate sample size [26]. According to the population of Macau (approx. 650k), at a 99% confidence level, at least 667 samples need to be extracted. Considering the large number of statistical divisions in Macau, sample extraction was maximized to enhance representativeness. The proportion of 22 statistical zones to the total population is recorded and then divided into 1000 initial samples according to this proportion. Afterwards, based on the age and gender distribution of the Macau population, the sample was supplemented to N = 1313 to ensure proportional representation across all descriptive characteristics.

The variables can be divided into the following three categories. The first category was collected through questionnaires, which examined the personal and family conditions of the respondents and their socioeconomic attributes, such as gender, age, and education level. The second type was collected through questionnaires and measured the well-being and life satisfaction scores of the respondents. The third type was the built environment variables, which were calculated using GIS software.

2.2.1. Individual Family Attributes and Socioeconomic Attributes

This study refers to the individual and household attribute characteristics commonly collected in previous research to distinguish each surveyed individual.

Table 1. Meaning of the samples’ correlated variables.

Variable	Variable Definition and Encoding Logic
Gender	Male = 1, Female = 2
Age	Age of the respondent
Education	Total years of formal education received by the respondent
Born place	Place of birth of the respondent. Macau = 1, Other = 2
Native language	Cantonese = 0, Mandarin = 1, English or Portuguese = 2, Other = 3
Residence in Macau	Number of years the respondent has lived in Macau
Family member	Number of household members of the respondent
Car	Number of cars owned by the respondent’s household
Family income	Monthly income of the respondent’s household (in MOP)
Housing cost	Below 30% = 1, Above 30% = 2

lists these variables and their coding logic. These individual attributes include gender (male, female), marital status (single, unmarried cohabitation, married, separated, divorced, and other), native language (Cantonese, Mandarin, Portuguese, English, other), age (categorized into six groups: under 20 years old, 20–29, 30–39, 40–49, 50–59, and above 60), years of education, place of birth (Macau SAR or non-Macau), duration of residence in Macau (categorized into five groups: less than 1 year, 1–3, 3–5, 5–10, and 10 years and above), and the residential statistical district. Household attributes included the number of household members, the number of cars owned, monthly household income, and the proportion of housing in the household.

2.2.2. Well-Being and Life Satisfaction

Table 2. Well-being and life satisfaction variables.

Variable	Variable Definitions and Coding Logic
Well-being	The average score of individual respondents’ well-being: 1 represents the lowest level, while 7 represents the highest level.
Leisure	Individual respondents’ satisfaction with leisure: 1 = extremely poor, 7 = extremely good.
Social	Individual respondents’ satisfaction with social interactions: 1 = extremely poor, 7 = extremely good.
Work	Individual respondents’ satisfaction with work: 1 = extremely poor, 7 = extremely good.
Health	Individual respondents’ satisfaction with health: 1 = extremely poor, 7 = extremely good.

presents the mediating variables and dependent variables used in this study. The measurement of well-being study primarily utilized well-established dimensions commonly employed in mainstream research. The OHQ (Oxford Happiness Questionnaire) developed by the University of Oxford’s Department of Psychology was used [27]. Respondents rated their agreement level on a 7-point scale (1 = extremely disagree, 2 = disagree, 3 = slightly disagree, 4 = neutral, 5 = slightly agree, 6 = agree, 7 = extremely agree) in response to a set of questions. These questions comprised 12 items objectively describing the respondents’ well-being. The average score of the 12 items was used as a measure of the respondents’ well-being.

Based on a previous review, the relationship between built environmental factors and psychological well-being was hypothesized to be mediated by satisfaction across four life domains: leisure, work, society, and health. Similarly, respondents were rated based on their level of agreement with the listed questions (1 = extremely disagree, 2 = disagree, 3 = slightly disagree, 4 = neutral, 5 = slightly agree, 6 = agree, 7 = extremely agree).

Leisure Satisfaction

Leisure satisfaction is a component of life satisfaction, and existing theories suggest that leisure activities can influence individuals’ physiological and psychological well-being [28]. Leisure satisfaction was measured using the Leisure Satisfaction Scale (LSS) [28], which includes items such as “My leisure activities are delightful,” “My leisure activities give me confidence,” “My leisure activities give me a sense of accomplishment,” and “I have many different abilities in leisure activities.” Respondents rated their agreement with these statements on a 7-point Likert scale (1 = extremely disagree, 2 = disagree, 3 = slightly disagree, 4 = neutral, 5 = slightly agree, 6 = agree, 7 = extremely agree).

Social Satisfaction

Social satisfaction is defined as individuals’ satisfaction with interpersonal relationships. Social satisfaction was measured using items from Cottrell’s validated scale [29]. The items include “I have many friends,” “I have many opportunities to meet new friends,” “I frequently meet with my friends and relatives,” and “I receive help from people close to me.” Respondents also rated their agreement with these statements on a 7-point Likert scale (1 = extremely disagree, 2 = disagree, 3 = slightly disagree, 4 = neutral, 5 = slightly agree, 6 = agree, 7 = extremely agree).

Job Satisfaction

Some researchers consider work satisfaction to be one of the most important factors influencing well-being [14,25]. Work satisfaction was measured using Hart’s Scale [30]. The items include “In most cases, my work (or studies) is ideal,” “My work (or studies) is excellent,” “I am satisfied with my work (or studies) life,” and “So far, my work (or studies) has provided me with important things I want in life.” Respondents rated their agreement with these statements using a 7-point Likert scale (1 = extremely disagree, 2 = disagree, 3 = slightly disagree, 4 = neutral, 5 = slightly agree, 6 = agree, 7 = extremely agree).

Health Satisfaction

Finally, health satisfaction was assessed using COOP charts [31]. The chart asked respondents to describe the extent to which they could engage in the most vigorous exercise for two minutes continuously, the number of negative emotional problems they faced, the impact of physical and mental health issues on their daily life, the extent to which physical or mental health limited their social activities, and the experience of bodily pain. The severity of these issues was rated on a scale from 1 (no impact) to 5 (very severe impact).

2.2.3. Objective Built Environment Factors

On the basis of GIS, this study considered two major dimensions to measure the built environment: socioeconomic indicators and morphology indicators. These two large dimensions can be subdivided into multiple different small dimensions, and the specific variables and their meanings are shown in

Table 3. Built environment variables and their definitions.

	Variable	Variable Definitions and Coding Logic
Socioeconomic Built Environment variables	Shannon Index of Educational Background	The diversity of the distribution of educational background in each statistical area. In formula 1, Pi refers to the proportion of residents within the research unit who have the type i level of education.
	Shannon Index of Income Background	The diversity of the distribution of income background diversity in each statistical area. In formula 1, Pi refers to the proportion of residents within the research unit who have the i-th income level.
	Shannon Index of Land Use	The diversity of the land use in each statistical area. In formula 1, Pi represents the proportion of the i-th type of land area within the research unit to the total area.
	Unemployment Rate	The percentage of unemployed individuals in each statistical area.
	Population Density	The ratio of population in each statistical area
Morphological Built Environment Variables	Total Parcel Area	The sum of the parcel areas in each statistical area (in square kilometers).
	Network Density	The ratio of the total length of roads to the area of the statistical area.
	Service Facilities	The number of service facilities in the corresponding statistical area where respondents are located. Bus stations are not included.
	Leisure Facilities	The number of leisure facilities in each statistical area, including casinos and amusement parks.
	Dining Facilities	The number of dining facilities in each statistical area, including restaurants, cafes, and hotels.
	Office Facilities	The number of office facilities in each statistical area, including office buildings, government institutions, social organizations, and schools.
	Intersection Count	The number of intersections within the boundaries of each statistical area.
	Building Density	The ratio of the total built-up area multiplied by the floor height to the area of the statistical area.
	Bus Stop Count	The number of bus stops in the corresponding statistical area where respondents are located.

.

This study measures the built environment using two major dimensions: socioeconomic indicators and morphological indicators. Among them, socioeconomic indicators include three subcategories that measure the three dimensions of the built environment: density, diversity, and centrality. These subdimensions are further defined by detailed variables that describe the built environment. The number of bus stops was collected from official data released by two bus companies in Macau (New Era and Transmac). The service and office facility counts were collected from 2021 Baidu Map data. The average monthly housing expenses, unemployment rate, and population density were obtained from Macau’s 2021 mid-year population census data. The building density, total parcel area, intersection count, and road network density were calculated based on official map data published by the Macau Land Registry.

The Shannon index (ShI) is derived from information entropy theory and is widely used to measure diversity. In this study, variables that reflect diversity include the Shannon indices of educational background categories, income background categories, and land use types. These variables share the same formula. The only difference between them is the specific definition of the category (Table 3) being measured and the dimension it belongs to. The formula is as follows:

$$\mathrm{Sh}I=-{\displaystyle {\sum }_{i=1}^m{P}_i}\log P_i$$

(1)

In this formula, $m$ represents the number of categories under study, and $P_i$ represents the proportion of the $i$-th category in the total area, indicating the uncertainty of predicting that a randomly selected point in the study area belongs to a certain category. The data for educational background and income categories were sourced from the questionnaire introduced in Section 3.2. Together with land use data, the number of types for each of the three corresponds to the “m” in Equation (1), and each type can be converted into $P_i$. Generally, the higher the index value $\mathrm{Sh}I$, the higher the diversity of the corresponding system, and vice versa.

2.3. Multiple Linear Regression

In order to evaluate the possible multicollinearity issues between the independent variables in the model and ensure the stability of regression coefficient estimation as well as the reliability of interpretation, the VIF (Variance Inflation Factor) [32] was introduced to perform diagnostic checks on the independent variables included in the model. There are multiple criteria for determining VIF values, among which the rule of 10 is most commonly used in the field of social sciences, which assumes that a VIF less than 10 will not lead to severe multicollinearity [32].

This study is divided into two parts. The first part involves using a multiple linear regression model to examine the impact of the built environment on residents’ well-being. Due to the nested individual data within community units, the random intercept model can effectively distinguish the variance components at the individual and environmental levels. The model is specified as follows:

$$Y_{ij}=\alpha +\beta X_{ij}+\gamma E_j+{\epsilon }_{ij}$$

(2)

In the equation, individual $i$ and statistical zone $j$ are nested. $X_{ij}$ represents the personal and household attributes of individual $i$ residing in statistical zone $j$. $E_j$ represents the built environment attributes of statistical zone $j$, including morphological and socioeconomic attributes. ${\epsilon }_{ij}$ represents the residual, and $\alpha$ represents the intercept.

The external environment influences well-being through individuals’ subjective perceptions [14]. Based on the research framework of Mouratidis [17], subjective perception is further refined into four different aspects of life satisfaction. These aspects of life satisfaction mediate the relationship between the external environment at the statistical level and subjective well-being. This study employs mediation analysis to examine the mediating role of life satisfaction in the relationship between the external environment and residents’ well-being. The stepwise method was used to analyze the mediation impact model. The model specification for the mediation impact is as follows:

$$M_{\mathit{ij}}=\alpha +\beta X_{\mathit{ij}}+\gamma E_j+\epsilon$$

(3)

$$Y_{\mathit{ij}}=\alpha +\beta X_{\mathit{ij}}+\gamma E_j+\delta M_{\mathit{ij}}+\epsilon$$

(4)

In the equations, $M_{\mathit{ij}}$ represents the mediating variable, which refers to the four aspects of life satisfaction among residents: leisure satisfaction, social satisfaction, work satisfaction, and health satisfaction.

2.4. Assumption Model Setting

To achieve the research objectives, a total of 8 models, which were classified into 3 categories, were established in this study.

First, well-being is taken as the dependent variable, and built environment indicators and individual family attributes are included as independent variables in Formula (2). As shown in Figure 2, Model 1 shows the direct impact of built environment variables on well-being (correspondingly, the term reflecting individual attributes in Formula (2) is 0), while Model 2 calculates the impact of the built environment on well-being after calculating individual characteristics (IC). Model 1 and Model 2 attempt to verify the hypothetical relationships between the built environment variables (BE) and well-being (WB), as shown in the following figure. The direct impact in Model 1 is denoted as c1, the direct impact of the built environment on well-being in Model 2 is denoted as c2, and the impact of personal and family attributes as control variables on well-being is denoted as c2’.

Second, the next four models aim to verify whether the built environment significantly affects the mediating variable, which is a prerequisite for testing the mediating effect. Four different life satisfaction factors are included as mediating variables in Formula (3) to verify whether there is a significant correlation between built environment factors and the mediating variables. As shown in Figure 3, Model 3 shows the impact of various built environment variables on the mediating variables when entertainment satisfaction is used as a mediator variable. Model 4, Model 5, and Model 6 demonstrate the impact of various built environment variables on resident well-being when social satisfaction, job satisfaction, and health satisfaction are used as dependent variables, respectively. Different from Model 1 and Model 2, a1, a2, a3, and a4 represent the relationships between the built environment (BE) and the four types of life satisfaction.

Finally, the four mediating variables are added to formula (2) as mediating variables to verify the mediating effect, i.e., Formula (4). The results of Model 7 demonstrate direct effects, while the results of Model 8 focus on the influence of mediating variables on well-being. In Figure 4, a5 represents the impact of the built environment on life satisfaction, b1 represents the impact of life satisfaction on well-being, the product of a5 and b1 represents the indirect effects of the built environment variables (Model 8), and c3 represents the direct effect of the built environment on well-being (Model 7). The difference between Model 7 and Model 1 is the mediating variables: Model 7 contains 4-type satisfaction as a control variable, while Model 1 does not contain any control variable.

Table 4. Summary of Models 1–8.

Model	Independent Variable	Control Variable	Dependent Variable
1	Built environment variables	N/A	Well-being
2	Built environment variables	Individual characteristics	Well-being
3	Built environment variables	Leisure Satisfaction	Well-being
4	Built environment variables	Social Satisfaction	Well-being
5	Built environment variables	Work Satisfaction	Well-being
6	Built environment variables	Health Satisfaction	Well-being
7	Built environment variables	4-type Satisfactions	Well-being
8	Built environment variables	4-type Satisfactions	Well-being

lists all eight patterns of the eight models. These eight models adopt different variable designs to achieve the research objectives.

IBM SPSS 22 for Windows and the Process3.4 plugin were used in this study. Before incorporating the data into the model, a collinearity test was conducted among the independent variables, and the tolerance level of all the independent variables was greater than 0.3, indicating acceptable collinearity.

3. Results

3.1. Descriptive Statistics

Table 5. Collinearity statistics.

Mode		Statistics
1		Tolerance	VIF
Independent Variables	Shannon Index of Income Background	0.251	3.977
	Shannon Index of Land Use	0.427	2.341
	Shannon Index of Educational Background	0.309	3.236
	Unemployment Rate	0.424	2.36
	Population Density	0.345	2.896
	Total Parcel Area	0.432	2.313
	Network Density	0.292	3.428
	Service Facilities	0.166	6.024
	Leisure Facilities	0.218	4.597
	Dining Facilities	0.129	7.763
	Office Facilities	0.163	6.117
	Intersection Count	0.231	4.327
	Building Density	0.236	4.239
	Bus Stop Count	0.417	2.399
Mean		0.288	4.001
2		Tolerance	VIF
Independent Variables	Shannon Index of Income Background	0.245	4.071
	Shannon Index of Land Use	0.409	2.446
	Shannon Index of Educational Background	0.300	3.334
	Unemployment Rate	0.416	2.404
	Population Density	0.247	4.050
	Total Parcel Area	0.417	2.397
	Network Density	0.267	3.751
	Service Facilities	0.149	6.694
	Leisure Facilities	0.217	4.614
	Dining Facilities	0.115	8.724
	Office Facilities	0.151	6.605
	Intersection Count	0.230	4.345
	Building Density	0.233	4.286
	Bus Stop Count	0.370	2.705
Control Variables	Gender	0.871	1.148
	Age	0.593	1.686
	Education	0.839	1.192
	Born place	0.710	1.408
	Native language	0.783	1.278
	Residence in Macau	0.740	1.351
	Family member	0.618	1.618
	Car	0.639	1.565
	Family income	0.748	1.337
	Housing cost	0.888	1.126
Mean		0.466	3.089

shows the collinearity statistics of the independent variables, where Mode 1 represents the results when only the independent variables are present, and Mode 2 includes personal and family attributes as dependent variables. The results indicate that all variables did not exceed the severe multicollinearity threshold (VIF > 10), and the average VIF value was much lower than 5. It is worth noting that the VIF values of service facilities, catering facilities, and office facilities are between 5 and 10, which may indicate moderate multicollinearity. But this collinearity did not substantially distort the estimation results of other core independent variables in the model. From an objective perspective, commercial clustering is a common phenomenon, as services, catering, and office facilities tend to gather more in commercial areas.

Table 6. Descriptive statistics.

	Variable	Mean	Std.
Morphology Built Environment Variables	Total parcel area	1.112	1.797
	Network Density	312.183	69.913
	Service Facilities	119.53	75.47
	Retail Facilities	18.20	15.067
	Dining Facilities	238.91	144.550
	Office Facilities	84.51	78.56
	Intersection count	407.27	212.308
	Building Density	1.879	1.164
	Bus Stop Count	16.33	5.754
Socioeconomic Built Environment Variables	Shannon Index of Educational Background	0.5378	0.774
	Shannon Index of Income Background	0.7238	0.079
	Shannon Index of Land Use	0.3531	0.039
	Unemployment Rate	0.0185	0.004
	Population Density	801.39	495.43
Well-being and life satisfaction	Well-being	4.431	0.701
	Leisure Satisfaction	4.755	0.895
	Social Satisfaction	4.860	0.831
	Job Satisfaction	4.610	1.005
	Health Satisfaction	3.397	0.545
Individual and Family Factors	Gender	1.48	0.500
	Age	38.81	14.19
	Education	13.4	3.326
	Born place	1.50	0.500
	Native language	0.75	0.436
	Residence in Macau	7.99	2.656
	Family member	3.75	1.157
	Car	1.3	0.894
	Family income	43,116.91	21,964.326
	Housing cost	1.37	0.482

presents descriptive statistical data for all observable variables. The table shows that the built environment indicators can be divided into two parts: morphological and socioeconomic. All built environment indicators are continuous variables. The morphology section includes building density, total land area, number of intersections, number of public transportation stations, road network density, office facilities, service facilities, entertainment facilities, and catering facilities. The socioeconomics section includes population density, unemployment rate, land use Shannon index, Shannon index for education background, and Shannon index for income background.

Among the morphology variables, the average total parcel area is 1112.77 (Std. = 1797.55), and the average road network density is 312.183 (Std. = 69.913). On average, there are 119.53 service facilities (Std. = 75.47) in each partition, 18.20 entertainment facilities (Std. = 15.067), 238.91 catering facilities (Std. = 144.550), and 84.51 office facilities (Std. = 78.56). The average number of bus stops and intersection density are 16.33 (Std. = 5.754) and 407.27 (Std. = 212.308), respectively.

For the socioeconomic variables, the distribution of population density has a high degree of variability (Std. = 495.43), while due to the Shannon index and unemployment rate values being between 0 and 1, their standard deviation is relatively low. The average building density of each statistical partition is 1.879 (Std. = 1.164).

The average value of well-being is 4.43 (Std. = 0.701), indicating that the average level of life satisfaction is between “neutral” and “relatively satisfied”. Entertainment satisfaction, job satisfaction, and social satisfaction have averages of 4.7 (Std. = 0.895), 4.8 (Std. = 0.831), and 4.6 (Std. = 1.005), respectively, indicating that residents’ satisfaction levels are between “relatively satisfied” and “very satisfied”. The average health satisfaction rate is 3.397 (Std. = 0.545), which is lower than 3.5, indicating that the average health satisfaction rate of residents is below the average level.

Regarding individual and household attributes, the respondents’ average age is 38.81 years, with 39.1% of residents falling in the age range of 30–60 years. Males account for 48.1% of the sample, which is slightly lower than that of females but close to the overall gender ratio in Macau [23]. Approximately half of the respondents are local-born residents or immigrants, and the proportion of Cantonese speakers is 60.2%. Approximately 55.5% of the respondents have resided in Macau for more than ten years. Apart from undergraduate education, the number of residents with different educational backgrounds is similar, with 47.3% holding a bachelor’s degree or higher. The proportion of married and unmarried cohabiting populations is 62.3%. The average monthly household income is MOP 43,116.91 (USD 5332.95), with a significant income disparity among respondents (standard deviation = 21,964.326).

3.2. Multiple Linear Regression Results of the Impact of Built Environment Indicators on Well-Being

Table 7. Multilevel linear regression results of residents’ well-being.

		Model 1		Model 2
		COEF.	t	COEF.	t
Morphological Built Environment Variables	Intersection Density	0.191 *	1.688	0.146	1.332
	Bus Stop Count	0.137 *	1.878	0.125 *	1.779
	Building Density	−0.039	−0.353	−0.061	−0.573
	Road network Density	−0.228	−1.602	−0.142	−1.024
	Service Facilities	−0.209	−1.619	−0.252 **	−2.016
	Retail Facilities	0.290 **	2.167	0.223 *	1.723
	Dining Facilities	0.187 *	1.843	0.194 **	1.974
	Office Facilities	0.403 **	2.069	0.382 **	2.036
	Leisure Facilities	−0.670 ***	1.250	−0.542 ***	−2.329
	Total Parcel Area	−0.024	−0.479	−0.032	−1.718
Socioeconomic Built Environment Variables	Population Density	0.113	1.250	0.107	1.223
	Shannon Index of Income Background	−0.191 **	−1.784	−0.140 *	−1.347
	Shannon Index of Land Use	−0.200 ***	−2.804	−0.178 ***	−2.576
	Shannon Index of Educational Background	0.074	0.909	0.053	0.668
	Unemployment Rate	−0.186 **	−2.386	−0.130 **	−1.718
Individual and Family Factors	Gender			−0.056	−1.935
	Age			−0.125 ***	−3.781
	Place of Birth			−0.006	−0.400
	Years of Education			0.064 **	−0.303
	Native Language			−0.010	2.250
	Marital Status			0.027	0.930
	Private Car Ownership			0.060 *	1.737
	Family Members			0.074 **	2.167
	Years of Residence in Macau			0.048 *	1.959
	Household Income			0.113 ***	3.677
	Housing Expenses			−0.160 ***	−5.704

Note: *, **, and *** indicate significance levels of p < 0.10, p < 0.05, and p < 0.01, respectively. Robust standard errors were used for all models. The same applies to the following.

displays the results of the multiple linear regression. The results of the null model indicate that the internal correlation coefficient is 0.056, which is less than 0.4, and multiple linear regression is needed. Model 1 represents the direct impact of various indicators of the built environment based on statistical partitioning on residents’ well-being, while Model 2 represents the impact of various indicators of the built environment on residents’ well-being after adding personal and family attributes as control variables.

Regarding morphology, in both Model 1 and Model 2, building density and total parcel area show no significant relationship with residents’ well-being. After incorporating individual and household attributes, intersection density and road network density do not exhibit a significant relationship with residents’ well-being, which is consistent with the findings of Liu [33]. In contrast, the number of bus stops and service, retail, dining, office, and leisure facilities (POI points) are all significantly correlated with well-being.

Regarding socioeconomic factors, population density and the Shannon index of income background show no significant relationship with well-being. In contrast, income background, unemployment rate, and the Shannon index of land use significantly correlate with well-being. This suggests that the overall socioeconomic conditions of the statistical districts have a considerable impact on individuals’ well-being, with variations in social diversity leading to differences in well-being.

Factors such as gender, place of birth, and native language, which are inherent characteristics, do not significantly correlate with well-being. However, age significantly affects well-being, showing a decreasing trend with increasing age. There is no significant difference in well-being between single and married individuals in Macau, possibly due to social welfare and inclusiveness factors, but further policy analysis is required.

A higher education level and income are associated with greater well-being, while higher housing expenses are linked to lower well-being. This aligns with previous research on the relationship between education level, personal income-generating capacity [34], and residents’ well-being, indicating a significant connection between financial autonomy and well-being.

The greater the density and accessibility of various facilities are, the greater the level of well-being. In terms of socioeconomic diversity, the higher the unemployment rate is, the greater the income gap and the lower the level of well-being. The building density and population density do not have significant impacts on well-being. This may be due to the high building density and population density in each statistical partition. Furthermore, the average age of residence in the sample is 7.7 years, indicating that people have become accustomed to high-density living environments, so the importance of accessibility has increased while the importance of density has decreased.

3.3. Impact of Built Environment Indicators on Mediating Variables

As shown in

Table 8. Impact of the built environment on mediating variables.

		Model 3		Model 4		Model 5		Model 6
	Dependent Variable	Leisure		Social		Work		Health
	Variables	Coef.	t	Coef.	t	Coef.	t	Coef.	t
Morphology Built Environment Variables	Intersection Density	0.0068	0.7659	0.0121	1.311	0.0083	0.8342	0.0111 **	2.0553
	Bus Stop Count	−0.001	−0.2292	−0.0068 *	−1.6971	−0.0004	−0.0732	−0.0053 **	−2.037
	Building Density	0.014	0.6574	0.0608 ***	3.0945	0.0217	0.9114	−0.001	−0.0751
	Road Network Density	0.0003	0.7929	0.0006 *	1.7866	0.0003	0.6618	0.0005 **	2.2463
	Service Facilities	0.0006 *	1.727	0.0008 ***	2.6101	0.0008 **	2.0769	0.0001	0.4142
	Retail Facilities	0.0026 ***	2.7463	0.0022 **	2.4711	0.0032 ***	3.0322	0.0009 *	1.6217
	Dining Facilities	0.001 ***	5.844	0.0008 ***	2.6101	0.0011 ***	5.9616	0.0003 **	2.4279
	Office Facilities	0.0011 ***	3.4544	0.0009 ***	3.2313	0.0013 ***	3.6946	0.0003	1.5094
	Leisure Facilities	0.0059 ***	3.6	0.0027 *	1.744	0.0067 ***	3.6328	0.0017 *	1.7292
Socioeconomic Built Environment Variables	Shannon Index of Income Background	−0.2107 ***	−7.8039	−0.161 ***	−5.9053	−0.1974 ***	−7.2927	−0.0685 **	−2.4856
	Shannon Index of Land Use	0.0423 *	1.5337	0.0394	1.4273	0.0374	1.3567	0.0179	0.6488
	Shannon Index of Educational Background	−0.119 ***	−4.3378	−0.1457 ***	−5.3334	−0.0741 ***	−2.6892	0.031	1.1242
	Unemployment Rate	−0.1715 ***	−6.3014	−0.0614 **	−2.2266	−0.1697 ***	−6.2354	−0.067 **	−2.4331
	Population Density	−0.021	−0.7601	0.0767 ***	2.785	−0.0403	−1.4602	0.0048	0.1732

Note: *, **, and *** indicate significance levels of p < 0.10, p < 0.05, and p < 0.01, respectively. Robust standard errors were used for all models. The same applies to the following.

, mediation effects were tested for all age and gender groups. Models 3, 4, 5, and 6 show the correlation between the built environment and the mediating variables when using different levels of life satisfaction as an explanatory variable.

The results of Model 3 indicate that the Shannon indices based on income (p < 0.01), land use (p < 0.1), education category (p < 0.01), and unemployment rate (p < 0.01) influence residents’ well-being through leisure satisfaction. The accessibility of various facilities, and population diversity have significant impacts on leisure satisfaction. This indicates that people’s leisure activities are closely related to the urban facilities in Macau. The greater the number of urban facilities and accessibility there are, the greater the satisfaction with life, indicating the importance of accessibility.

The results of Model 4 indicate that the mediating variables of building density (p < 0.01) and road network density (p < 0.1) have a significant impact on social satisfaction. Building density, network density, and other density building environments, as well as various facility accessibility and diversity indicators, indicate a strong correlation with social satisfaction. The theme of social satisfaction, which focuses on people’s social interaction and interpersonal relationships, indicates that a compact, diverse, and accessible urban environment can promote interpersonal interaction among residents. However, differences in income and education levels weaken communication between people.

The results of Model 5 indicate that job satisfaction is closely related to the distribution of various facilities. Moreover, income, educational diversity, and unemployment rates are closely related to job satisfaction. Model 5 shows similar results to Model 3, but the mediating impact of land use Shannon index is not as significant as that in Model 3.

Model 6 shows that health satisfaction is closely related to intersection density, the number of bus stops, and network density. The greater the intersection density is, the greater the health satisfaction; the greater the network density is, the greater the satisfaction with health; and the fewer bus stops there are, the greater the satisfaction with health. For Macau residents, as the number of bus stops decreases, the frequency of walking and cycling increases, the opportunities for physical exercise increase, and health satisfaction correspondingly increases. The accessibility of restaurants also significantly affects health satisfaction. Due to the lack of surveys on whether the restaurants sold healthy foods, as well as a lack of understanding of residents’ dietary preferences, it is impossible to explain why the greater the number of restaurants is, the greater the health satisfaction; further research is required.

Although some indicators of Model 6 are not significant for some facility distributions, they are closely related to intersection density (p < 0.05), the number of bus stops (p < 0.05), and road network density (p < 0.05). This result is consistent with the morphology index. Because morphological indicators are inherent attributes of residents in a sense, in addition to their subjective choice of place of residence, they are not related to their subjective feelings.

By comparing Model 7 and Model 8 (

Table 9. Direct impacts on well-being.

		Model 7		Model 8
		Direct Effect		Total Effect
	Variables	Coef.	t	Coef.	t
Morphology Built Environment Variables	Intersection Density	0.0175	3.3964	0.0223	3.2107
	Bus Stop Count	0.0084	3.3705	0.0074	2.2086
	Building Density	−0.0063 *	−0.5042	0.0106	0.6385
	Building Density (Projected)	−0.0508	−0.5042	0.1553	−0.5667
	Road Network Density	0.0005	2.3622	0.0007	2.4143
	Service Facilities	0.0005 ***	2.7275	0.0009 ***	3.693
	Retail Facilities	0.0024 ***	4.2958	0.004 ***	5.4482
	Dining Facilities	0.0004 ***	4.0516	0.001 ***	7.5837
	Office Facilities	0.0004 **	2.4012	0.0011 ***	4.5886
	Leisure Facilities	0.0026 ***	2.6823	0.0058 ***	4.5647
Socioeconomic Built Environment Variables	Population Density	0.0263	1.2689	0.0152	0.5506
	Shannon Index of Income Background	0.0544	2.5853	−0.102 ***	−3.7129
	Shannon Index of Land Use	−0.0294 *	−1.4337	0.0014	0.0518
	Shannon Index of Educational Background	0.0946 *	4.5845	0.0116 *	0.4204
	Unemployment Rate	0.0207 **	0.9836	−0.0982	−3.5726
Life Satisfaction Factors	Leisure Satisfaction			0.216 ***	7.539
	Social Satisfaction			0.145 ***	5.199
	Work Satisfaction			0.447 ***	16.685
	Health Satisfaction			−0.041 **	−1.793

Note: *, **, and *** indicate significance levels of p < 0.10, p < 0.05, and p < 0.01, respectively. Robust standard errors were used for all models. The same applies to the following.

), it is not difficult to see that, after adding mediating variables, there is no significant correlation between intersection density, number of bus stops, building density, road network density, population density, and the income background Shannon index and subjective well-being. A comparison of Model 7 and Model 4 reveals that the Shannon index of building density, population density, and income background has a complete mediating effect on social satisfaction. Additionally, there is a complete mediating effect between intersection density, the number of bus stops, and road network density on health satisfaction. Moreover, the Shannon Index, a measure of income background, is fully mediated by four types of life satisfaction.

The interpretability of the model is shown in

Table 10. Model interpretability.

Model	Variable Type	R2	Adjusted R2	p
1	BE	0.076	0.069	<0.001
2	BE + IC	0.152	0.134	<0.001
3	BE + Leisure	0.320	0.311	<0.001
4	BE + Social	0.274	0.207	<0.001
5	BE + work	0.302	0.295	<0.001
6	BE + health	0.112	0.101	<0.001
7&8	BE + 4-type	0.476	0.468	<0.001

. The comparison of R² and adjusted R² across models reveals significant differences in explanatory power when incorporating specific covariates.

Model 1 alone explains only 7.6% variance (adjusted R² = 0.069), indicating the limited direct impact of basic environmental factors. With leisure satisfaction variable added, the adjusted R² substantially increases to 0.311, which confirms leisure satisfaction as a critical explanatory dimension. Notably, the comprehensive model with four variable types achieves the highest adjusted R² = 0.468, demonstrating that multidimensional joint effects can explain nearly 47% variance, significantly outperforming any single-dimension extended models (p < 0.001). Model 6 shows the lowest contribution (adjusted R² = 0.101), suggesting potential needs for refined measurement in this dimension. All models reached statistical significance (p < 0.001), validating the theoretical relevance of variable selection.

4. Discussion

4.1. Main Findings

The main findings are based on a new theoretical framework that explores the impact of the social and morphological significance of the architectural environment on subjective well-being. This directly addresses Research Objective (1) by revealing the mechanisms through which the built environment influences well-being. Determinants of well-being have been identified. In addition, the intrinsic pathway between environmental factors and well-being was also considered. Different models were established using statistical dimension partitioning methods. By comparing eight models, the impact path of the built environment on well-being is revealed.

The results indicate that in terms of morphology, the most intuitive aspect is that the number of facilities within the statistical partition directly and significantly affects well-being. In addition, land use diversity, income background diversity, and education background diversity, which are socioeconomic indicators, also have a profound impact on well-being. The intersection density, road network density, and number of bus stops, which are three indicators related to walking accessibility, have a complete mediating effect on health satisfaction.

First, morphological variables require discussion. The following observations were derived from cross-sectional survey data and objective built environment data obtained via GIS:

(1) There is a significant correlation between intersection density and residents’ well-being, but after adding individual family attributes as control covariates, this correlation disappears. After adding health satisfaction as a mediator variable, this association reappears. This indicates that the variable of intersection density has a deep correlation with health satisfaction and affects residents’ well-being through health satisfaction.

(2) Road network density does not show any correlation with well-being in Models 1 and 2, but it becomes significant after incorporating social satisfaction and health satisfaction. This result indicates that its association with well-being is actually strongly correlated with social satisfaction and health satisfaction, and it should not be assumed that there is a connection between road network density and well-being.

(3) The regression results for service facilities, entertainment facilities, catering facilities, and office facilities are similar and can be discussed together. Service facilities do not have a direct correlation with well-being, and after adding personal family attributes to Model 2, the impact on well-being becomes significant. The significance of service-oriented facilities is also highlighted in Guo et al.’s article [35]. This is a normal phenomenon because service facilities include various venues, gas stations, and other facilities that are closely related to individual choices. Entertainment facilities, catering facilities, and office facilities all demonstrate a strong correlation with well-being. Lin et al. point out that access to a hospital, supermarket, and gym enhanced people’s well-being, corresponding to service facilities, retail facilities, and leisure facilities [36]. The distribution of these four facilities also affects well-being through entertainment satisfaction, social satisfaction, and job satisfaction. Although there is no correlation between service facilities and office facilities and health satisfaction, catering and entertainment facilities show a significant correlation with health satisfaction. Work facilities and service facilities have a positive impact on well-being, indicating that the rationality of urban spatial facility configuration has a considerable impact on well-being, which is similar to the conclusions of articles studying the walkability of urban facilities [37].

(4) The number of public transportation stops has a positive impact on well-being, and the impact is significant when social satisfaction and health satisfaction are used as mediating variables, indicating that the rationality of urban spatial facility configuration has a significant impact on well-being. A study in Singapore also noted that safe and convenient transportation infrastructure is often linked to people’s health [10].

(5) There is no evidence to suggest that building density and population density are related to resident well-being, which is consistent with some previous research findings [36], while some studies targeting elderly individuals have found a strong correlation between building density and well-being [37]. Possible reasons for these different research conclusions may be the selection of research units (statistical zones) or differences in research subjects (all age groups and only elderly people).

For the socioeconomic factors, the following observations can be made:

(1) The Shannon index of educational background does not show a direct correlation with well-being in Models 1 and 2 but instead shows a negative correlation with entertainment satisfaction, job satisfaction, and social satisfaction in Models 3–5. This result indicates that as the difference in educational background increases, life satisfaction tends to decrease, and the improvement in well-being often occurs under the premise of a close educational background. Pei [38] indicates that the level of education has a significant positive impact on subjective well-being, and the widening gap in educational background may weaken the sense of social equity, which aligns research findings.

(2) The Shannon index of income background shows a significant correlation with well-being in all models. Similar to the Shannon index of educational background, the coefficients are also negative, indicating an increase in income differentiation and a decreasing trend in life satisfaction and well-being. For territories, a higher index signifies greater socioeconomic fragmentation—where neighborhoods exhibit extreme heterogeneity in household incomes. Moreover, the Shannon index of income background is fully mediated by four types of life satisfaction, indicating that the diversification of income levels will affect the well-being of regional residents. Filip Fors Connolly et al. [39] finds that in wealthier countries, the correlation between income and life satisfaction decreases as per capita GDP increases, but income disparities may affect well-being through relative deprivation mechanisms. Pei’s research also suggests that the weakening effect of family economic status on happiness after exceeding the average level further supports the negative impact of income inequality [38].

(3) The Shannon index of land use is significant in Models 1 and 2. After adding the four types of life satisfaction as covariates, the significance decreases. Only when entertainment satisfaction is used as a mediator variable does it remain significant. This indicates that land use diversity directly affects the well-being of residents and is not influenced by life satisfaction. However, there is no significant relationship between population density and building density and well-being. The above results show that the overall socioeconomic development of statistical zones has a significant impact on individual residents, and differences in social diversity lead to certain differences in well-being levels. Guo demonstrates that land use change may indirectly enhance residents’ well-being through policies, rather than solely through economic intermediaries [39]. Similarly, land use diversity may directly affect residents’ experiences through resource availability and spatial functional differences; the distribution of entertainment facilities reflects this point [13].

Concerning Research Objective (2) on mediation pathways, the role of mediating variables within the causal pathway was further elucidated. Model 1 revealed the direct effect of independent variables on the dependent variable, while Models 3–6 demonstrated the impact of independent variables on mediating variables. Models 7–8 showed the effects of independent variables on the dependent variable when controlling for four mediators. Five scenarios emerged:

• Variables exhibiting significance in both Model 1 (significant c path, Figure 2) and Model 7 (significant c’ path, Figure 2) without meeting Model 1 criteria demonstrated partial mediation;
• If a built environment variable showed non-significance in Models 3/4/5/6 (non-significant a path, Figure 3) or Model 8 (non-significant b path, Figure 4), this suggests insufficient evidence for its association with residents’ well-being;
• Complete mediation occurred when variables were non-significant in Model 1 but significant in Model 7, provided they avoided Scenario 1 conditions;
• Suppression effects were identified when variables transitioned from non-significant in Model 1 to significant in Model 7, with opposing signs between the product of a coefficients (Models 3–6) and c’ coefficients (Model 7);
• Non-significant variables in both Model 1 and Model 7 with consistent coefficient signs indicated no substantial association.

The key findings are revealed below:

• Complete mediation: Intersection density (health satisfaction), bus stop count (health and social satisfaction), and income diversity Shannon index (all outcomes).
• Partial mediation: Service facilities (excluding health satisfaction), retail facilities (all outcomes), catering facilities (all outcomes), office facilities (excluding health satisfaction), recreational facilities (all outcomes), land use Shannon index (all outcomes), education diversity Shannon index (excluding health satisfaction), and unemployment rate (all outcomes).
• Non-significant associations: Building density, road network density, and population density.

Therefore, a novel classification system for environmental effects was developed. The summary of these scenarios is presented in

Table 11. Summary of mediating types of environmental variables.

Mediation Type	Key Indicators	Policy Relevance
Complete	Intersection density, Bus stops, income diversity	Require health infrastructure co-design
Partial	Various facilities, Land use mix, education diversity	Need spatial-social policy integration
Null	Building density, Road networks, population density	Demand paradigm shift in planning standards

.

4.2. Life Satisfactions: The Core Bridge Between Built Environment and Happiness

1. Health satisfaction: the core link between environmental stress and health perception

Health satisfaction is the most widely mediated pathway through which the built environment affects happiness. For example, intersection density, through its complete mediating effect on health satisfaction (rather than direct impact), suggests that high-density traffic environments may weaken residents’ perception of health through chronic stress (such as noise and safety hazards), thereby reducing their sense of well-being. Similarly, the number of bus stops serves as a dual mediator between health and social satisfaction, suggesting that the accessibility of public transportation may affect both physical health and social connectivity, but further validation of its interactive effects is needed. It is worth noticing that some variables (such as service facilities and educational Shannon index) were not mediated by health satisfaction and may be related to the differential impact of facility types on health behavior.

2. Social satisfaction: a key channel for spatial design to promote social capital

Social satisfaction is prominent in the partial mediation of the number of bus stops and service facilities, supporting the theoretical hypothesis of “space promotes social interaction”. The high accessibility of bus stops may strengthen social networks by increasing chance encounters (such as neighborhood interactions), while service facilities (such as community centers) provide physical carriers for collective activities. However, the partial mediating effect of retail and catering facilities on all satisfaction levels (including society) suggests that the business environment may affect happiness through a dual pathway: both directly improving quality of life through convenience and indirectly enhancing satisfaction through promoting socialization. This is consistent with the view of commercial spaces as social incubators in the “third place” theory.

3. Job and Entertainment Satisfaction: Environmental Diversity Supports Pathways

Job and entertainment satisfaction are not common traditional intermediary pathways, as they highlight the widespread impact of the built environment on psychological needs. The Shannon index of income diversity, through the complete mediation of all four satisfaction levels, indicates that an economic mixed environment may enhance happiness in multiple dimensions through employment opportunities (job satisfaction), leisure choices, access to health resources (health satisfaction), and community inclusiveness (social satisfaction). In contrast, the partial mediating effect of land use diversity (Shannon index) points to its direct support for work entertainment balance: such as the integration of work and residence to reduce commuting pressure. These findings expand the existing research’s understanding of the “environment happiness” pathway, suggesting the need to go beyond the health framework and incorporate the perspectives of career development and leisure equity.

4. Implications for non-significant variables: threshold effects and situational dependence of environmental indicators

The non-significance of variables such as building density and road density contradicts some research [25], possibly due to a non-linear relationship: moderate density promotes convenience, while excessive density leads to congestion pressure. In addition, these variables may indirectly act through unmeasured mediators such as sense of security and aesthetic experience, and further testing is needed in conjunction with geographical contexts such as city level.

This systematic mediation analysis clarifies the complex pathways through which built environment characteristics influence residents’ well-being, highlighting both direct and mediated relationships. The current research has several limitations that can be addressed in future studies. First, empirical analysis based on questionnaire data may suffer from self-assessment bias and issues of non-dependence. Like many other statistical models, the ordinary least squares (OLS) model used in the current study can reveal only statistical relationships within sample data. The results based on statistical models test the credibility of causal relationships in the empirical world, which may differ from causal relationships in the real world. Due to limitations in research conditions, cross-sectional data were used in this study, which may lead to limitations in the research conclusions. Future research may strengthen the causality of research through empirical sampling methods (ESMs). Also, a limitation of our modeling strategy is that the more focused models (3–8) do not include the full set of controls examined in Model 2. While this allows for cleaner tests of specific hypotheses and avoids over-controlling, it means that the estimates in these models could potentially be influenced by unobserved or omitted confounders. This study considers only limited influences on well-being. Other factors, such as tourism and psychological satisfaction, may also play a role [17]. Including more elements in the model could provide better insights into the factors influencing well-being.

5. Conclusions

This study proposes eight models by integrating relevant theories from psychology, sociology, and morphology, using four kinds of mediating life satisfaction simultaneously. By supplementing existing theories with the mediating mechanism of the impact of the built environment on happiness, it promotes a research model that combines internal and external factors in the field. The research area of this study is a high-density urban area, and the conclusions should provide corresponding inspiration for high-density and populous areas. Through empirical research, it has been confirmed that there is a close correlation between residents’ well-being and the external environment of statistical zones. It is particularly noteworthy that the greater the number of facilities in the surrounding environment is, the greater the well-being; the greater the road accessibility is, the greater the well-being.

This study classified the built environment into socioeconomic factors and morphological factors. The research findings suggest that the rational allocation of facilities is most likely related to mental health—for example, higher compact density, moderate street network design, more parks providing entertainment services, and more diverse land use may be important for residents. These findings contribute to the development of evidence-based planning strategies and provide guidance for designing community landscapes and facilities, thereby further promoting policy implementation by authorities.

This study further elucidated the important role of subjective levels of entertainment satisfaction, social satisfaction, job satisfaction, and health satisfaction in the influence of the external environment on well-being. This study revealed that the number of facilities, such as service facilities, retail facilities, catering facilities, and office facilities, is closely related to well-being and social satisfaction, producing a complete mediator. Indicators that reflect diversity, such as the Shannon index of income background, are fully mediated by all four types of life satisfaction.

Currently, well-being is highly valued by governments and people worldwide, and how to enhance the well-being of residents has become an important issue in society. The following implications are derived from the study’s findings. First, the government should play a proactive role in improving road connectivity and the accessibility of facilities to avoid issues such as inadequate resource supply and a decline in residents’ quality of life. Second, while ensuring the fair distribution of public resources and services, it is essential to establish many high-quality public service facilities to enhance Macau’s public service system, allocating necessary facilities and service functions in residential areas. Third, at the social level, there should be increased humanistic care for residents, making their work, daily life, leisure, and sports activities more convenient while also aiming to improve residents’ subjective impressions and objective experiences within the statistical districts. Fourth, efforts should be made to enhance social equity; reduce income, education, and employment disparities; increase the average living standards of the population; and decrease wealth disparities.