Divergent Pathways to Place Attachment: How Heterogeneous Communities Shape Human–Green Space Relationships in Beijing

Abstract

Land transition in China has led to the emergence of highly heterogeneous neighborhoods. This process challenges the social sustainability of public green spaces. This research investigates the driving mechanisms of place attachment within green space across diverse community typologies in Beijing. This study constructed a structural equation model (SEM) based on 626 valid questionnaires, using the Stimulus–Organism–Response (S-O-R) framework. The overall SEM results indicate that place identity significantly contributes to civic behavior (β = 0.439, p < 0.001). However, a persistent ‘value-action’ gap remains, with 65.81% of residents demonstrating high identity yet low participation. Furthermore, the multi-group analysis (MGA) reveals that place attachment logic diverges significantly across groups. Regarding user identity, public events promote visitors’ place identity, but this effect remains insignificant among residents (β = −0.064, p > 0.05). Regarding generational differences, the macro-spatial environment is significantly associated with place dependence for young people (β = 0.330, p < 0.001) but is insignificant for the elderly. Community heterogeneity reveals distinct failure modes. In commodity housing communities, a disconnect exists where daily usage fails to foster dependence (β = 0.026, p > 0.05). Conversely, urban–rural resettlement communities display an identity deficit where public events fail to translate into place identity (β = 0.131, p > 0.05). The study proposes differentiated renewal pathways tailored to three community types. For commercial housing communities, it advocates precise interventions that prioritize social engagement. Meanwhile, for urban–rural resettlement communities, the focus shifts to accessibility and culturally rooted activities to help reconnect displaced populations.

1. Introduction

1.1. Background and Significance

Amid social structural changes and urban renewal in China, urban communities are experiencing a significant transition. Social structures are shifting from a traditional, acquaintance-based society to contemporary stranger societies. This transformation replaces dense social ties with anonymity and social heterogeneity [1]. Urban communities now exhibit highly heterogeneous demographic structures. This results from the convergence of resettlement housing, commercial properties, and aging neighborhoods, alongside the co-residence of native inhabitants, new citizens, and migrant populations. However, current public green space development predominantly prioritizes physical enhancements, such as increasing greenery coverage and paving quality. This physical-centric approach often overlooks the diverse psychological perceptions of stakeholders [2]. This mismatch results in spaces that are environmentally upgraded yet suffer from “placelessness” and social alienation. Consequently, restoring the people-place bond and reshaping community cohesion through place attachment have become critical challenges in contemporary urban governance [3]. Geographer John K. Wright first proposed the concept of “place”, defined it as a region imbued with subjectivity. Subsequent research expanded this into a sense of place, place image, and place attachment [4]. Environmental psychologists have defined place attachment as an individual’s emotional investment in a specific setting [5]. This theory fundamentally explores the emotional bonds between people and places, encompassing how diverse individuals perceive, cognize, and process spatial information. Applying place attachment theory is particularly essential for heterogeneous communities characterized by diverse population structures and fragmented social networks. In these contexts, public green spaces serve as physical environments that provide vital settings for leisure, recreation, and physical exercise. However, beyond these tangible functions, they also act as complex arenas for the negotiation of interests and emotional investment among different groups. The central governance challenge lies in mitigating social isolation, lack of belonging, and spatial exclusion arising from this demographic complexity [6,7]. Therefore, utilizing place attachment theory to analyze these spaces and promote social integration holds significant theoretical and practical value.

Given this context, this study outlines the conceptual framework, methodological approaches, and defining characteristics of place attachment. The Stimulus–Organism–Response (S-O-R) model from environmental psychology constructs a theoretical logic for public green spaces in heterogeneous communities. Nine representative communities in Beijing serve as case studies to illustrate the specific pathways and strategies through which public green spaces can facilitate social integration. The mechanisms revealed here offer transferable insights for urban governance in other transitional societies grappling with the ‘stranger society’ dilemma.

1.2. Literature Review

(1) The mechanism of place attachment formation: Place attachment, the emotional bond between humans and place, has long been a core topic in environmental psychology and human geography. The mechanisms driving its formation remain a subject of theoretical debate. Scholars generally agree that this bond emerges from dynamic human–environment interactions. Early phenomenological perspectives laid the groundwork by distinguishing meaningful “places” from undifferentiated “spaces” through human experience. Subsequent theoretical frameworks further posit that a place’s physical characteristics, authenticity, and alignment with individual traits form the structural basis of attachment [8].

Regarding the physical dimension, the physical environment underpins social meaning; Favorable landscape features and facility accessibility directly induce functional dependence [9,10]. Conversely, regarding the social dimension, the literature highlights that physical spaces evoke unique social meanings through the ‘genius loci’ and active user engagement. The cognitive and affective processing of these environments allows them to function as remedies for loneliness and sources of relaxation, thereby solidifying the human–place bond [11,12,13]. High-quality public green spaces can enhance residents’ attachment, confirming the crucial role of social factors by serving as essential remedies for urban isolation [14,15].

Despite these parallel inquiries, a critical theoretical gap persists: the literature lacks a unified framework synthesizing how physical and social elements interact, frequently treating them as isolated variables rather than competing or synergistic forces [16,17,18]. While previous literature highlights the positive attributes of the physical environment, the diverse interpretation of these identical physical signals across cultural and socioeconomic groups remains underexplored. Consequently, in heterogeneous communities, this divergence in perception may contribute to ‘alienation’ rather than the expected attachment. As the community shifts from homogeneity to heterogeneity, whether the physical environment alone can still effectively support the production of social meaning remains to be empirically tested.

(2) Effects of place attachment: Place attachment is not merely a psychological state but a potent predictor of behavior. Contemporary research has shifted from single-dimensional satisfaction assessments to the prediction of complex, contradictory behavioral outcomes.

First, attachment significantly impacts individual well-being and psychological resilience. Research shows that highly attached residents use community spaces more effectively for psychological restoration when facing life stressors [19]. In the context of public health crises, such as the COVID-19 pandemic, this place-based bond serves as a critical resource for adaptive coping and psychological flourishing, helping individuals navigate disruption [20]. Furthermore, this sense of belonging acts as a crucial buffer against urban loneliness, thereby enhancing overall community resilience in the post-pandemic era [21]. Second, and more critically, attachment acts as a catalyst for civic and pro-environmental behaviors. It is widely identified that place attachment motivates residents to engage in “place-protective actions” [22]. However, the assumption of a linear positive effect often ignores the unintended consequences of high-density living. The “universal altruism” hypothesis is increasingly contested, particularly within high-density urban contexts. Critical urban scholars argue that assuming a uniformly positive effect fundamentally ignores the negative impact of place attachment—spatial competition. Empirical evidence from Chinese communities indicates that mere “functional dependence” may be linked to self-serving behaviors, such as the private appropriation of public green spaces. Only when attachment elevates to “emotional identification” does it transform into altruistic actions that safeguard public interests [23].

Recent studies have further elucidated the specific mechanisms driving this transformation. On one hand, attachment does not operate in isolation; it synergizes with social norms and perceived environmental responsibility to foster urban citizenship behaviors [24]. On the other hand, specific positive emotions play a mediating role; for instance, “place-based pride” has been highlighted as a key emotional tie that links attachment to active pro-environmental behaviors [25,26]. These findings highlight the urgency of guiding the transition of attachment levels—from dependence to identity and pride—for effective governance in heterogeneous communities.

(3) Place attachment in heterogeneous communities: The impact of community heterogeneity on social cohesion has been a subject of enduring debate. A dominant perspective argues that without shared lifestyles, heterogeneity often breeds social conflict rather than integration [27]. Heterogeneity typically refers to the diversification of population structures, social strata, and cultural backgrounds within a community. Theoretical models further suggest that high heterogeneity directly leads to declining community trust and diminished public participation [28]. Extensive economic and sociological studies empirically substantiated this stance, demonstrating that greater diversity correlates with reduced social capital and civic disengagement [29,30]. However, simply transplanting this conflict perspective to the Chinese context requires critical scrutiny. Unlike the racial and ethnic heterogeneity dominant in Western discourse, the heterogeneity in contemporary Chinese cities is primarily driven by institutional transformation and housing marketization. Heterogeneity in Chinese cities manifests uniquely as stratification based on housing tenure policy and rural-urban migration backgrounds. At the spatial level, this heterogeneity manifests as competing claims over usage rights and decision-making power regarding public green spaces among different groups.

While the existing literature acknowledges the governance challenges posed by heterogeneity, it primarily analyzes causes through macro-level lenses such as social capital [31], demographic composition [32], and institutional management [33]. Place attachment theory remains underexplored within the context of social heterogeneity [34,35]. In the context of contemporary urban renewal, scholars further elucidated the complexity of this dynamic. Advanced machine learning model techniques reveal the nonlinear, spatially differentiated mechanisms that link spatial quality and resident satisfaction [36,37]. Recent scholarship has shifted focus toward place attachment as a critical psychological mechanism for navigating such complexity. Current research has begun to explore how attachment functions within diverse demographic structures. In the context of ethnic heterogeneity, investigations into diasporic communities reveal that attachment is not singular but characterized by plurilocality and simultaneity, serving as a source of psychological grounding for marginalized groups [38]. Regarding the transition from homogeneous to heterogeneous communities in China, evidence from transitional work-unit (Danwei) communities confirms that the physical neighborhood environment remains a pivotal predictor of community attachment, even as traditional social bonds loosen [39]. Furthermore, in mixed contexts involving residents and external visitors, resident-centric attachment strategies effectively translate into support behaviors and involvement, thereby bridging the gap between diverse stakeholders [40].

Despite these advancements, a critical gap remains. While existing literature acknowledges the importance of attachment in heterogeneous communities, few scholars have examined micro-level “spatial–psychological” mechanisms. Consequently, there is a lack of research investigating how differentiated spatial intervention strategies can bridge group divides and foster consensus in fragmented social structures [41,42,43]. More importantly, the current debate largely assumes a homogeneous user base. While the universally positive attributes of the physical environment are well-documented, there is a distinct lack of critical engagement regarding how identical physical signals are decoded differently across diverse socioeconomic groups.

1.3. Research Gaps

Existing scholarship on place attachment predominantly focuses on tourist destinations or traditional, homogeneous communities, often overlooking high-density, highly heterogeneous residential neighborhoods. Moreover, micro-level investigations of public green spaces remain scarce. A disciplinary disconnect persists: urban planning tends to prioritize physical spatial interventions, while place attachment research focuses on psychological and emotional analysis. Consequently, there is insufficient dialogue between these two perspectives.

Specifically, current research on the “perception–psychology–behavior” chain remains largely at the level of static evaluation. It lacks quantitative empirical evidence grounded in causal mechanisms. Consequently, it cannot explain why standardized spatial interventions often fail to foster a deep-seated sense of identity among residents. Furthermore, sociological studies have analyzed the governance challenges arising from heterogeneity. However, few explore how spatial interventions can reshape consensus within such fragmented social structures. These limitations underscore the urgent need to examine the cultivation of place attachment in public green spaces.

To bridge these gaps, the SOR framework adopts environmental psychology to construct a place attachment model for heterogeneous community public green spaces. The empirical data reveal how physical environments and social interactions build place attachment, which, in turn, promotes civic behavior. These findings offer evidence-based strategic guidance for spatial governance in heterogeneous societies.

2. Methodology

2.1. Stimulus–Organism–Response Model Framework

Mehrabian and Russell proposed the S-O-R model, which serves as a foundational paradigm in environmental psychology. It posits that environmental attributes (stimulus) elicit internal cognitive and emotional changes (organism), which, in turn, trigger approach or avoidance behaviors (response). Given the complexity of human–place interactions within heterogeneous communities, simple linear models fail to capture the underlying mechanisms. Therefore, this study adopts the SOR framework to map the driving pathways of place attachment in these settings.

The rationale for adopting the SOR framework lies in its unique “process-oriented” perspective. For urban residents, the relationship with green spaces is fluid. Attachment is not a pre-existing condition but a state constructed through continuous sensory and social engagement. The SOR theory offers distinct advantages over traditional static models. In heterogeneous communities, the people–place relationship is often fractured or undergoing reconstruction, meaning attachment is not a fixed state but a dynamic formation process. Unlike static frameworks that merely correlate variables, the SOR model explicitly maps how residents internalize external environmental attributes into internal emotional states. This capability makes the model particularly suitable for elucidating the formation mechanism of place attachment in transitional urban settings. It reveals how physical and social cues effectively repair severed emotional bonds. Furthermore, the S-O-R framework categorizes ‘stimulus’ into physical and social dimensions. The former includes elements like green space quality and accessibility, while the latter encompasses factors such as community atmosphere. The ‘organism’ component captures the internal states of individuals by integrating variables including community heterogeneity, cultural background, socioeconomic status, and length of residence. This framework examines both the formation of place attachment and its subsequent outcomes, specifically pro-environmental behavior and mental health. This comprehensive approach proves highly practical for analyzing community governance and policy interventions.

In this framework, stimulus refers to the objective attributes and social context of the public green space. To reflect the dual nature of heterogeneous communities, the conceptual model operationalizes the ‘stimulus’ factor into physical and social dimensions. The former focuses on the physical environment, including macro-spatial character and micro-scale elements; the latter addresses social interaction, encompassing daily routines and public events. Organism represents the internal psychological processing induced by these stimuli, which the framework defines as place attachment. Acting as the mediating link between the external environment and behavioral outcomes, it encompasses two sub-dimensions: cognitive evaluation (place dependence) and emotional connection (place identity). Finally, response denotes the behavioral intentions arising from these emotional states (Figure 1). Within the context of community governance, this manifests specifically as civic behavior, including civic engagement and active spatial practices.

2.2. Research Hypotheses

(1) Stimulus refers to the objective attributes of the public green space. The physical environment serves as the fundamental basis for place perception [44,45]. Chang et al. argue that physical attributes directly influence psychological states [46]. At the macro-level, the overall spatial character and layout determine functional appropriateness and the capacity for self-regulation, thereby fostering functional dependence. At the micro-level, detailed design features enhance physical comfort. Simultaneously, these features evoke memories through cultural symbols to establish intimate human–place bonds. Based on these premises, we hypothesize:

H1–H2. 

The Macro-Spatial Character positively influences place attachment (place dependence/place identity).

H3–H4. 

Micro-scale Elements positively influence place attachment (place dependence/place identity).

(2) Organism represents the internal psychological processing induced by these green stimuli, serving as the core concept of place attachment. Public green spaces function as vital vessels for social life [47]. High-frequency daily activities increase residents’ functional reliance on a space and imbue it with social meaning [48,49]. Public events enrich functional offerings. Furthermore, drawing on psychological ownership theory, empower residents by enhancing their sense of efficacy and ownership [50]. Based on this, the framework suggests:

H5–H6. 

Social interaction (daily routine) positively influences place attachment (place dependence/place identity).

H7–H8. 

Public events positively influence place attachment (place dependence/place identity).

(3) The relationship between dependence and identity. For everyday spaces like neighborhoods, functional logic often precedes emotional logic. Individuals typically develop functional dependence first as spaces meet their specific needs. Through frequent interaction, this utilitarian connection gradually accumulates into familiarity. Eventually, this familiarity internalizes into deep emotional bonds and self-identity. Based on this, we hypothesize:

H9. 

Place dependence acts as the foundation of place identity and positively influences it.

(4) Based on the SOR theory, positive emotional states are key antecedents to behavioral intentions [51,52,53]. Models of place-protective behavior confirm that high levels of place attachment motivate individuals to engage in community citizenship behaviors (CCB). Such behaviors include voluntary maintenance, environmental protection, and active participation in park management [54,55,56,57]. Deep emotional connections are critical variables that trigger altruistic actions. Therefore, we make the following hypotheses:

H10–H11. 

Place attachment (place dependence/place identity) positively influences civic behavior (Figure 2).

3. Research Methods

3.1. Study Area

To ensure the validity of the social structural analysis, the sampling strategy was strictly targeted at communities with distinct binary or multi-dimensional population mixing. Conventional research measures heterogeneity through socio-cultural dimensions such as race and immigration status [58,59,60,61,62,63,64]. These Western-centric metrics lack explanatory power in China’s reality. For instance, the Han ethnic group accounts for approximately 95.2% (roughly 20.85 million) of Beijing’s permanent resident population, rendering racial or ethnic metrics completely ineffective for capturing social stratification. Instead, socio-spatial stratification in China is fundamentally driven by population mobility and the profound evolution of the housing system. Specifically, the transition from welfare public housing to a highly marketized commodity housing system [65].

The theory of socio-spatial segregation in transitional economies provided the conceptual foundation for adapting “community heterogeneity” from a Western context to the Chinese urban reality. As previously noted, traditional socio-cultural metrics like race, language, and religion are highly homogeneous in Beijing. Therefore, these metrics lack explanatory power. Consequently, this study argues that housing prices and residential diversity serve as much more sensitive indicators of socio-spatial stratification. Since racial and ethnic differences are less prominent in Beijing, social stratification is primarily driven by institutional history, housing marketization, and urban regeneration policy. Therefore, the customized indicators: framework prioritizes the following dimensions:

1. Social heterogeneity.

China’s Hukou system functions as a significant institutional boundary [66]. Consequently, the indicator framework prioritized population mobility over traditional immigration metrics. The mix of locals and non-locals represents varying levels of access to urban welfare and public resources. Furthermore, demographic factors such as age structure capture generational differences and varying lifestyle needs within public spaces, which constitute a practical dimension of social heterogeneity in Chinese communities.

2. Economic stratification.

In marketized megacities like Beijing, differences in housing prices act as a spatial filtering mechanism. Because housing serves as a key proxy for socioeconomic status (SES) and wealth in China, the variance in housing prices within a sub-district indirectly reflects the spatial coexistence of divergent economic groups [67]. Consequently, housing price differentiation serves as a highly relevant and practical indicator to measure economic heterogeneity.

3. Residential diversity.

Following the perspective of the “social production of space”, building typologies in Chinese cities serve as spatial manifestations of institutional shifts. The coexistence of traditional work-unit compounds (Danwei), resettlement housing, and modern gated commodity estates within the same area indicates a mixing of different property rights and social groups [68]. Therefore, physical spatial diversity can mirror structural social complexity.

To adapt this theory to the Chinese context, this study developed a context-specific “Sociology–Economic–Spatial” assessment framework. This model replaces implicit cultural metrics with quantifiable indicators of physical and economic factors. Leveraging data from the Seventh National Population Census and real estate platforms [69,70,71], this research established the comprehensive community heterogeneity Index (Appendix A 

Table A1. Community heterogeneity indicators.

Definitions			Equation 1	Notation
population heterogeneity	age structure deviation index	measures deviation from the citywide age structure; higher values indicate greater demographic imbalance [93]	$ID ={\displaystyle \frac{\left|{\mathit{P}}_{\mathit{c}} - {\mathit{P}}_{\mathit{e}}\right|}{\left|{\mathit{P}}_{\mathit{c}}+{ \mathit{P}}_{\mathit{e}}\right|}} - {ID}_{city}$	ID represents the age structure imbalance index; values closer to 1 indicate a more severe imbalance. Pc represents the population aged 0–14; Pe represents the community’s population aged 60 and above
	aging stress index	higher values reflect a heavier elderly burden and significant inter-generational demand disparities [94]	$\mathit{H} =1 -{\displaystyle \sum _{\mathit{i}=1}^{\mathit{k}}}\left({{\mathit{P}}_{\mathit{i}}}^2\right)$	Pi represents the proportion of the population belonging to category i out of the total population; k denotes the total number of categories.
	population mobility index	higher ratios of migrant residents indicate a more complex and unfamiliar social composition [95]	$\mathit{H} =1-{\displaystyle \sum _{\mathit{i}=1}^{\mathit{k}}}\left({{\mathit{P}}_{\mathit{i}}}^2\right)$	${\mathit{P}}_{\mathit{i}}$ represents the proportion of non-local registered residents within the community
building heterogeneity	high-density agglomeration index 1	represents the density of residents; higher values imply greater spatial resource intensity	$\mathit{D} ={\displaystyle \frac{\mathit{P}}{\mathit{A}}}$	D represents building density; P represents total building footprint area; A represents total street land area.
	residential diversity index	measures the richness of mixed building types within a community, such as bungalows, slab buildings, and tower blocks [96]	$SHDI =-{\displaystyle \sum _{\mathit{j}=1}^{\mathit{m}}}({\mathit{P}}_{\mathit{i}}\times \mathit{ln}{\mathit{P}}_{\mathit{j}})$	SHDI measures building typology diversity. Pj represents the proportion of buildings of a given type relative to the total number of buildings in the community; m denotes the total number of building types.
economic heterogeneity	housing price differentiation index	measures the inequality of housing prices among neighborhoods. A value closer to 1 indicates a larger gap in housing prices [97]	$CV ={\displaystyle \frac{\sqrt{\frac{{\sum }_{\mathit{i}=1}^{\mathit{n}}{{\mathit{w}}_{\mathit{i}}\times ({\mathit{v}}_{\mathit{i}} -{\mathit{v}}_{\mathit{w}})}^2}{{\sum }_{\mathit{i}=1}^{\mathit{n}}{\mathit{w}}_{\mathit{i}}}}}{\frac{{\sum }_{\mathit{i}=1}^{\mathit{n}}({\mathit{v}}_{ii}\times {\mathit{w}}_{\mathit{i}})}{{\sum }_{\mathit{i}=1}^{\mathit{n}}{\mathit{w}}_{\mathit{i}}}}}$	wi represents the number of buildings in the neighborhood. vi denotes the average housing price in the neighborhood; i is the neighborhood index; n is the total number of neighborhoods

¹ All indicators are dimensionless.

). Figure 3 displays the district-level classification results for Beijing, derived after Z-score standardization and weighted calculation.

The initial sampling phase isolated the 50 most heterogeneous sub-districts, representing the top 15% of the statistical distribution. The step ensured statistically significant divergence in objective indicators, including population structure, building typology, and economic differentiation. A K-means clustering analysis (k = 3) classified these areas into objective typologies. The analysis identified three typologies of community heterogeneity: work-units declining communities (cluster 1), commodity housing communities (cluster 2), and urban–rural resettlement communities. To maximize feature distinctiveness, the sampling procedure excluded intermediate samples and selected the three archetypal cases closest to each cluster center, yielding nine representative communities (Figure 4,

Table 1. Research site selection statistics.

Neighborhood Typology	Key Characteristics	Location	Heterogeneity Index 1	Name	Site Code
transitional work-unit decline communities	pronounced demographic mixing	Anzhen Subdistrict	3.91	Anzhen Community Park	T1-A
		Fangzhuang Subdistrict	3.78	Fangzhuang Sports Park	T1-B
		Donggaodi Subdistrict	3.31	Donggaodi Cultural Park	T1-C
commodity housing communities	significant socio-economic stratification	Shuangjing Subdistrict	3.98	Qingfeng Park	T2-A
		Jianwai Subdistrict	4.67	CBD Forest Park	T2-B
		Wangjing Subdistrict	4.14	Wangjing Park	T2-C
urban–rural resettlement communities	distinct spatial morphology variation	Liulitun Township	4.05	Erdaogou Park	T3-A
		Xiaohongmen Township	3.85	Hongbo Park	T3-B
		Cuigezhuang Township	4.81	Heli Habitat Park	T3-C

¹ The heterogeneity index is a standardized composite score calculated from census data.

).

Within these nine communities, strict criteria guided the selection of the final public green space sample points based on three core principles. First, the principle of importance required the selected sites to hold central spatial positions or possess significant cultural and natural resources. Such high-value locations naturally attract widespread attention and use, establishing a fundamental condition to effectively observe human–environment interactions. Second, the principle of consensus demanded that residents widely recognize and utilize the selected spaces. Strategic targeting focused on spaces located at the intersections of different residential types. This placement ensures the spaces serve diverse communities and serve as key places for daily neighborhood interactions. Finally, the principle of operability mandated that sufficient research data be available for the selected sites. The verification process confirmed the availability of historical records, physical spatial data, and online public reviews for each location. Selecting spaces with complete data profiles enabled robust cross-case comparisons, despite their varying current conditions. The continuous application of these rigorous criteria successfully identified nine core empirical sample points. These public green spaces encompass both formal and informal venues situated within the 15 min community living circle.

Based on these criteria, this research identified nine core empirical sample points. These public green spaces encompass both formal and informal venues situated within the community living circle. They serve as vital “contact zones” for the daily lives of diverse residents and act as crucial bonds for fostering community sentiment. Importantly, although the selected sites are physically situated within residential neighborhoods, this research conceptually and functionally defines them as public green spaces. Unlike gated residential gardens, these venues are open, public-access spaces integrated into the city’s green network.

3.2. Screening Green Space Elements and Scale Construction in Heterogeneous Communities

A hybrid approach drove the scale development to ensure both theoretical rigor and contextual validity. First, we adapted the core theoretical dimensions from well-established scales in the existing literature [72,73]. These core dimensions specifically include place attachment, social interaction, and community civic behavior [74]. However, heterogeneous communities present high socio-spatial complexity. Directly applying generalized Western measurement items might fail to capture local cultural nuances. Consequently, this research employed grounded theory. This inductive method helped to refine and supplement the specific measurement items for community green spaces.

Data collection followed a dual-source approach. The data collection process involved scraping public reviews from Dianping, a dominant consumer platform. Python 3.12.5 scripts captured authentic resident feedback. Simultaneously, NVivo 12 software processed 86 policy documents to extract objective elements regarding community green spaces. Grounded theory principles guided the coding phase. Open coding filtered the semantic noise and isolated high-frequency tags. Axial coding synthesized these initial tags into broader subcategories. This analytical progression yielded four aggregate categories containing 24 specific measurement indicators (Appendix A 

Table A2. Grounded theory coding.

Selective Coding	Axial Coding	Initial Code	Source 1
Objective composition dimension from policy	macro-spatial character	green space accessibility	Reflect the degree of integration between green nodes and multi-functional urban zones [98]; focus on the realization of 15 min community life circles [99].
		landscape	Evaluate the aesthetic value of greenway services. emphasize the promotion of healthy lifestyles through landscape quality [100].
		spatial boundary permeability	Measures the openness of green space boundaries. emphasize the control of development intensity in high-density old districts [101].
		spatial scale and enclosure	Evaluate the effectiveness of urban acupuncture; focus on the spatial layout of micro-green spaces and pocket parks [102,103,104].
	micro-scale elements	Ecological natural features	Assess the utilization of local vegetation. Highlight the interaction between ecological services and social values [104].
		landscape landmarks	Identify landscape nodes with distinct cultural characteristics [105]; evaluate the balance between practical functionality and aesthetic appeal [106].
		garden ornaments	Evaluate the integration of artistic elements into park furniture [107]; focus on the display of local heritage through paving and street facilities [108].
		Recreational infrastructure	Evaluates the completeness of community-embedded support facilities; it focuses on the accessibility of essential service points [109].
	daily routines	daily outdoor activities	Capture the distinctiveness of local living environments. reflect the embodiment of tradition in daily use [110].
		social inclusivity	Measure the inclusivity for vulnerable groups [111]; focus on child-friendly and aging-friendly spatial upgrades [112].
		social bonding	Assess the preservation of existing social networks; emphasize the impact of in situ retention on social capital [102].
	public events	green space engagement	Reflect the degree of resident participation in space governance. Implement the co-construction and co-sharing model [113]
		organized activity	Evaluate the density of cultural facilities such as intangible heritage centers [114]; focus on the provision of immersive cultural exhibitions [115]
		management	Focus on the long-term maintenance mechanisms involving residents [116].
Subjective composition dimension from user reviews	place dependence	functional-spatial fit	Assess the degree to which green space layout meets diverse needs for evening strolls and seasonal recreation; emphasize the functional irreplaceability.
		outdoor lifestyle adaptation	Evaluate the convenience of park access and the familiarity of the environment; focus on the adaptation to reservation-free open spaces.
		physical activity support	Measure the diversity of fitness trails and ball game facilities; assess the capacity of green space to support multi-type sports.
	place identity	place-based collective memory	Capture the historical memory evoked by the oasis environment. reflect the sense of nostalgia and childhood resonance triggered by the space.
		perceived openness and inclusiveness	Evaluate the friendliness toward pets and barriers; focus on the sense of ease experienced by children and seniors.
		cultural identity	Assess the resonance with old Beijing’s charm and heritage; focus on the sense of cultural legacy, such as the golden capital history.
		place attachment and resonance	Measure the psychological healing effect of the landscape; reflect the emotional relaxation and self-rediscovery in the urban environment.
	civic behavior	neighborhood mutual support	Assess the spontaneous prosocial behaviors among visitors; focus on mutual assistance during games or recreational activities.
		green space co-governance participation	Measure the involvement in traditional festival events; reflect the enthusiasm for interactive community activities.
		public property maintenance willingness	Assess the self-discipline in protecting the green environment; reflect the willingness to discourage uncivilized behaviors.

¹ Due to space constraints, only representative excerpts are listed in the source column to illustrate the link between raw data and theoretical categories.

). The resulting indicators directly informed the questionnaire design, capturing both resident perceptions and practical planning requirements. A mapping procedure aligned these indicators with the core dimensions of the SOR framework.

3.3. Pilot Testing and Reliability Verification in Heterogeneous Communities

The study followed strict ethical protocols for consent and privacy. The survey instrument captures the key dimensions: demographic background, perceptions of the socio-physical environment, place attachment, and civic behavior (

Table 2. Question items.

Latent Variables	Observed Variables	Item Code	Measurement Items 1
macro-spatial character (MC)	green space accessibility	MC1	The green space is well-integrated into the neighborhood and is easily accessible.
	landscape visual quality	MC2	The landscape design effectively reflects the unique geographical and ecological features.
	spatial boundary permeability	MC3	The boundaries of the green space are open and permeable, ensuring natural transitions.
	spatial scale and enclosure	MC4	The spatial scale of the site provides a comfortable sense of enclosure without feeling confined.
micro-scale elements (ME)	ecological and natural features	ME1	The ecological design details demonstrate a high degree of adaptation to the local environment.
	landscape landmarks	ME2	The space features iconic landmarks or structures that are highly representative of the area.
	garden ornaments	ME3	The decorative elements and garden ornaments exhibit distinct cultural characteristics.
	recreational infrastructure	ME4	The recreational infrastructure and facilities align perfectly with residents’ daily lifestyles.
daily routines (DR)	daily outdoor activities	DR1	The daily outdoor activities in this green space are diverse, vibrant, and well-supported.
	social inclusive	DR2	People from diverse age groups and backgrounds can coexist harmoniously in this space.
	social interaction and bonding	DR3	This green space serves as an essential physical platform for maintaining neighborly relationships.
public events (PE)	cultural engagement in green space	PE1	The space provides high levels of interactivity that encourage active cultural participation.
	organized activity provision	PE2	Public events and organized activities are frequently held here to enhance social vitality.
	management and maintenance response	PE3	The management team provides timely and effective responses to maintenance complaints.
place dependence (PD)	functional-spatial fit	PD1	The functional layout and facilities of this space meet my daily activity needs perfectly.
	outdoor lifestyle adaptation	PD2	Utilizing this green space for outdoor activities has become a part of my daily routine.
	physical activity support	PD3	I significantly prefer this specific green space over other alternative locations for my activities.
place identity (PI)	place-based collective memory	PI1	This environment successfully evokes collective historical memories or personal life experiences.
	perceived openness and inclusiveness	PI2	I feel psychologically at ease and comfortable being myself within this inclusive environment.
	cultural identity	PI3	The specific socio-cultural atmosphere of this place provides a strong sense of belonging.
	place attachment and resonance	PI4	I have developed a deep emotional resonance with this space that transcends its physical functions.
civic behavior (CB)	neighborhood mutual support	CB1	I am proactively inclined to offer assistance to neighbors when interacting in this green space.
	green space co-governance participation	CB2	I am willing to participate in public discussions or decision-making regarding space renovation.
	public property maintenance willingness	CB3	I feel a responsibility to discourage any behaviors that might damage the green environment.

¹ Respondents rated all items on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree).

).

Since online reviews and policy texts informed the measurement dimensions, potential limitations regarding context and sampling bias existed. To validate construct validity and optimize the questionnaire design, an expert review refined the items. To ensure methodological reliability before the formal survey, an offline pilot study with 100 valid samples provided further empirical testing.

Scale purification and reliability testing. To evaluate the internal consistency and discriminatory power of the initial scale, the validation phase employed corrected item-total correlations (CITCs). CITC represents the Pearson correlation between an individual item and the aggregate score of the remaining items, serving as a critical filter to “purify” the measurement construct. The analysis excluded items with a CITC value below 0.30 [75].

The pilot testing results indicated that the item recreational infrastructure (ME4) fell below this threshold. The low correlation suggests a conceptual overlap with the subsequent public events dimension. Residents likely perceive this attribute as a part of their social activities rather than a specific micro-design element. Therefore, the questionnaire excluded item ME4 to avoid redundancy and optimize the structural validity of the scale. All remaining items met the strict retention criteria and ensured a robust instrument for the final survey.

Following this exclusion, the final scale retained 23 items. Reliability and validity tests on the pilot data confirmed the robustness of the optimized instrument: the overall Cronbach’s Alpha reached 0.941 (>0.80), the Kaiser–Meyer–Olkin (KMO) measure scored 0.783 (>0.60), and Bartlett’s Test of Sphericity was statistically significant (p < 0.05). These metrics demonstrate that the scale possesses excellent internal consistency and discriminatory power. The item justifying its use for the main study.

3.4. Formal Survey and Sampling Strategy of Green Spaces in Heterogeneous Communities

The formal data collection utilized a combined stratified and quota sampling strategy. Standard sample size formula for a 95% confidence level. The maximum variability (p = 0.5), the theoretical minimum sample size was calculated at 384 (with a 5% margin of error) [76]. Structural equation model (SEM) and multi-group analysis (MGA) analysis typically require a minimum of 100 samples per subgroup to ensure statistical robustness [77]. To meet this threshold and accommodate an estimated 20% invalid response rate, we distributed 50 to 100 questionnaires at each location across the 9 communities. Projections estimated a total sample size exceeding 600, ensuring data representativeness and analytical precision.

3.5. Workflow

Figure 5 illustrates the sequential methodology used to systematically investigate the mechanisms and renewal pathways for green space in heterogeneous communities.

(1) Selection of study areas and subjects.

First, this study identified heterogeneous communities based on three dimensions: population demographics, the built environment, and economic indicators. Subsequently, we selected core green spaces within these communities as specific study sites, guided by principles of representativeness, accessibility, and social consensus.

(2) Research preparation and scale development.

Employing grounded theory, the study derived evaluation dimensions and designed questionnaire items through an in-depth analysis of online public reviews and relevant policy texts. A pilot survey (N = 100) tested the items. CITC analysis then screened and purified these preliminary variables.

(3) Formal survey and data processing.

Adopting a stratified sampling strategy, the formal survey distributed 50 to 100 questionnaires in each selected location. To ensure data quality, the analysis implemented a rigorous cleaning protocol. This process excluded responses that failed two embedded attention-check questions aimed at detecting contradictory answers. Questionnaires completed in under 120 s were also removed to filter out data with insufficient engagement. The final valid dataset underwent reliability and validity testing to ensure quality (N = 626).

(4) Statistical analysis.

The study employs Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) to validate the scale structure. The analysis further uses SEM to test the hypotheses. and perform multi-group analysis (MGA) to uncover the underlying mechanisms of variable interactions.

(5) Results interpretation and implications.

Based on the SEM path diagrams, the study analyzed differences in path coefficients and variable weights to understand how place attachment translates into civic behaviors. These findings elucidated issues such as supply-demand mismatches and spatial differentiation. Ultimately, findings guided the formulation of green space renewal strategies in heterogeneous communities.

4. Results

4.1. Descriptive Statistics and Preliminary Data Analysis

The final dataset comprises 626 valid responses (366 residents, 260 visitors) via face-to-face interviews, representing a 96.7% validity rate. The descriptive statistics indicate a balanced gender distribution. The age structure was primarily concentrated in the 16–30 cohort (47.0%), followed by the 46–60 cohort (27.8%).

To address potential survivorship bias, we conducted a detailed demographic comparison between our on-site survey sample and the Seventh National Population Census data (Figure 6). A comparison with the Seventh National Population Census indicates that our on-site survey introduces survivorship bias, inherently skewing the findings toward active users. Specifically, local registered residents (58.5% compared to 41.5% in the census) and working-age adults (86.3% compared to 63.6%) are significantly over-represented. While this demographic structure differs from the broader census distribution, it inherently captures the profile of the most active users within these public spaces. Given the study’s focus on human–environment interactions, the perspectives of this engaged group provide highly relevant insights into actual spatial utilization and place attachment.

Consistent with established methodological protocols [78], a random partitioning procedure was used to partition the dataset into two sub-samples. We used Sub-sample 1 for EFA to identify the underlying factor structure and reserved Sub-sample 2 for CFA to validate the model fit and ensure analytical reliability.

To assess potential common-method bias arising from self-reported questionnaire data, Harman’s single-factor test analyzed the full sample (N = 626) using principal component analysis (unrotated solution, fixed to 1 factor). The results showed that the first factor accounted for 35.378% of the total variance, which is well below the commonly accepted threshold of 50%. This indicates that no single factor dominates the data, providing evidence that standard method bias is not a significant concern in this study.

4.2. Validation of the Model for Public Green Spaces in Heterogeneous Communities

(1) exploratory factor analysis (EFA)

A KMO value of 0.877 (>0.7) and a significant Bartlett’s Test of Sphericity (p < 0.001) confirmed the suitability of the datasets for factor analysis. Principal component analysis (PCA) with varimax rotation extracted the underlying structure. The factor analysis yielded seven factors with eigenvalues exceeding 1.0, accounting for 76.18% of the cumulative variance. All item factor loadings exceeded 0.5, demonstrating a clean and stable structure. Reliability testing indicated that Cronbach’s coefficients for the seven factors ranged from 0.837 to 0.893, all exceeding the 0.8 threshold, suggesting excellent internal consistency (

Table 3. Exploratory factor analysis results and factor loadings (sample 1, n = 313).

Observed Variables	Item Code	Factor Loadings						Communality 1
		1	2	3	4	5	6
MC1	0.840							0.780
MC2	0.866							0.806
MC3	0.787							0.697
MC4	0.797							0.740
ME1				0.803				0.764
ME2				0.809				0.752
ME3				0.849				0.783
DR1			0.828					0.798
DR2			0.875					0.814
DR3			0.828					0.801
PE1					0.798			0.717
PE2					0.815			0.766
PE3					0.812			0.768
PD1						0.812		0.794
PD2						0.739		0.715
PD3						0.832		0.8
PI1		0.790						0.738
PI2		0.764						0.714
PI3		0.785						0.753
PI4		0.747						0.709
CB1							0.750	0.759
CB2							0.783	0.746
CB3							0.833	0.807

¹ Rotation method: varimax rotation.

). Consequently, these rigorous steps established a final measurement model comprising 23 items across seven distinct dimensions.

(2) Confirmatory factor analysis (CFA)

CFA using AMOS 26.0 verifies the stability of the measurement model on Sample 2 (n = 313). The model fit indices demonstrated excellent performance (CMIN/DF = 1.565 < 3, RMSEA = 0.043 < 0.08, GFI, CFI, NFI all > 0.9), suggesting a high degree of alignment between the hypothesized model and the observed data. All critical ratio (C.R.) values were substantially greater than 1.96, confirming that every indicator significantly measures its corresponding latent construct at the p < 0.001 level. Regarding convergent validity, the composite reliability (CR) for all latent variables exceeded 0.7, and the average variance extracted (AVE) surpassed 0.5, demonstrating robust internal consistency and low measurement error. The Fornell–Larcker criterion assessed discriminant validity. The square roots of the AVEs for each latent variable (

Table 4. Confirmatory factor analysis results and factor loadings (sample 1, n = 313).

Path	Indicators	Estimate	S.E.	C.R.	P 1	Std. Estimate	AVE	CR
MC	MC1	1				0.843	0.690	0.899
	MC2	1.046	0.059	17.677	***	0.842
	MC3	0.904	0.056	16.068	***	0.788
	MC4	0.977	0.055	17.825	***	0.847
ME	ME1	1				0.807	0.624	0.833
	ME2	0.931	0.069	13.439	***	0.780
	ME3	0.893	0.068	13.181	***	0.784
DR	DR1	1				0.801	0.703	0.876
	DR2	1.116	0.071	15.738	***	0.839
	DR3	1.145	0.071	16.175	***	0.874
PE	PE1	1				0.812	0.678	0.864
	PE2	1.035	0.069	15.04	***	0.82
	PE3	1.088	0.071	15.296	***	0.839
PD	PD1	1				0.839	0.686	0.868
	PD2	0.985	0.061	16.066	***	0.826
	PD3	0.940	0.059	15.917	***	0.819
PI	PI1	1				0.805	0.686	0.897
	PI2	1.091	0.063	17.242	***	0.863
	PI3	0.995	0.062	16.099	***	0.818
	PI4	0.960	0.059	16.278	***	0.825
CB	CB1	1				0.811	0.637	0.84
	CB2	1.005	0.072	13.915	***	0.769
	CB3	1.027	0.070	14.702	***	0.814

¹ *** p < 0.001.

) were greater than their correlations with other variables (

Table 5. Discriminant validity analysis (sample 2, n = 313).

	MC	ME	DR	PE	PI	PD	CB
MC	0.83 1
ME	0.393 ***1	0.79
DR	0.239 ***	0.332 ***	0.838
PE	0.374 ***	0.368 ***	0.239 ***	0.824
PI	0.362 ***	0.506 ***	0.428 ***	0.424 ***	0.828
PD	0.313 ***	0.451 ***	0.371 ***	0.249 ***	0.599 ***	0.828
CB	0.409 ***	0.527 ***	0.416 ***	0.288 ***	0.659 ***	0.569 ***	0.798

¹ *** p < 0.001.

). This confirms significant statistical distinctiveness among the factors and the absence of severe multicollinearity.

Finally, Pearson correlation analysis revealed that all seven core variables exhibited significant positive correlations (p < 0.001). These results validate the covariation trends among variables and provide essential empirical support for the subsequent SEM, laying the foundation for testing the causal pathways within the SOR framework (Figure 7).

(3) Structural Equation Model (SEM)

The overall fit indices indicate a strong alignment between the theoretical model and the observed data (N = 626). All key metrics meet the recommended thresholds (

Table 6. Model fit indices (Full Sample, N = 626).

Indicator	CMIN/DF	GFI	NFI	RFI	IFI	TLI	CFI	RMSEA
General Indicator 1	<3	>0.9	>0.9	>0.9	>0.9	>0.9	>0.9	<0.08
Measurement Model Fitting Results	2.194	0.940	0.944	0.934	0.969	0.963	0.969	0.044

¹ Model fit indices and established criteria for evaluation.

). It confirms that the theoretical model aligns robustly with the observed data. The path analysis results indicate significant relationships among variables (

Table 7. Path analysis within SEM (full sample, N = 626).

Path Test			Estimate	S.E.	C.R.	P 1	STD. Estimate
PD	← 2	MC	0.157	0.045	3.462	***	0.161
	←	ME	0.304	0.056	5.400	***	0.276
	←	DR	0.253	0.048	5.261	***	0.239
	←	PE	0.113	0.056	2.017	0.044 *	0.096
PI	←	MC	0.080	0.038	2.117	0.034 *	0.087
	←	ME	0.130	0.048	2.727	0.006	0.125
	←	DR	0.181	0.041	4.407	***	0.181
	←	PE	0.277	0.048	5.788	***	0.248
	←	PD	0.324	0.044	7.336	***	0.343
CB	←	PD	0.298	0.049	6.092	***	0.313
	←	PI	0.442	0.053	8.377	***	0.439

¹ * p < 0.05 *** p < 0.001. ² → represents the directional path from predictor to outcome.

). All hypothesized paths reached statistically significant (p < 0.05), supporting hypotheses H1-H11 (Figure 8). For place dependence, macro-spatial character (β = 0.276) and daily routine (β = 0.239) emerged as the primary factors. In contrast, place identity was most strongly predicted by place dependence (β = 0.343) and public events (β = 0.248), indicating that functional dependence and specific environmental attributes are key antecedents of emotional identification. Regarding behavioral outcomes, both attachment dimensions significantly fostered civic behavior. Notably, place identity exerted a substantially stronger influence (β = 0.439, p < 0.001) than place dependence (β = 0.313, p < 0.001), identifying emotional bonding as the primary correlate of pro-social actions. This establishes a baseline model for the subsequent multi-group analysis.

4.3. Multi-Group Analysis of Heterogeneity: Identity and Generation

MGA tested the dataset for heterogeneity. The structural model introduced respondent identity and age as moderating variables to examine differences in key path relationships.

Given the mixed demographic structure, our classification scheme categorized respondents based on their human–place interaction modes. Existing studies suggest that residential stability fosters deep emotional bonds, whereas mobility tends to be associated with functional dependence or distinct attachment profiles [79,80,81]. Therefore, the demographic classification scheme divided respondents into residents (permanent inhabitants) and visitors (temporary entrants for tourism, transit, or consumption). To account for generational differences in spatial perception, the demographic classification divided the sample into two cohorts: individuals under 45 and those aged 45 and older. This classification references the standards of the 1980 Asia-Pacific Gerontology Conference and accounts for the requirement of balanced sample sizes [82]. This cutoff serves as a proxy for distinct life-cycle stages and aesthetic preferences, while ensuring statistically balanced sample sizes for the MGA.

(1) Before the MGA testing, analysis of variance (ANOVA) tests evaluated mean differences across diverse groups (

Table 8. Mean differences across groups (ANOVA).

Variable Dimensions	Grouping Category	Sub-Groups	Mean ± SD	F	P 1
place dependence	identity	resident	4.79 ± 1.30	8.513	0.004 **
		visitor	4.47 ± 1.36
	age	<45	4.60 ± 1.31	1.541	0.215
		≥45	4.74 ± 1.37
	community type	type 1	4.70 ± 1.34	1.487	0.227
		type 2	4.74 ± 1.39
		type 3	4.53 ± 1.26
place identity	identity	resident	4.68 ± 1.33	5.922	0.015 *
		visitor	4.43 ± 1.24
	age	<45	4.52 ± 1.28	1.83	0.177
		≥45	4.66 ± 1.33
	community type	type 1	4.67 ± 1.27	4.831	0.008 **
		type 2	4.71 ± 1.36
		type 3	4.35 ± 1.25
civic behavior	identity	resident	4.61 ± 1.35	9.301	0.002 **
		visitor	4.29 ± 1.29
	age	<45	4.44 ± 1.34	0.805	0.37
		≥45	4.54 ± 1.32
	community type	type 1	4.57 ± 1.28	6.743	0.001 **
		type 2	4.65 ± 1.38
		type 3	4.21 ± 1.30

¹ Total sample (N = 626), residents (n = 366), visitors (n = 260), <45 (n = 376), ≥45 (n = 250). community type1 (n = 209), type2 (n = 211), type3 (n = 206). * p < 0.05, ** p < 0.01.

). Regarding respondent identity, permanent residents recorded significantly higher scores than visitors in all three dimensions: place dependence (p < 0.01), place identity (p < 0.05), and civic behavior (p < 0.01). In terms of age groups, however, no statistically significant differences were observed between the younger (<45) and older (≥45) cohorts across any dimension (p > 0.05). Furthermore, the analysis of community types revealed a distinct gap in social-emotional outcomes. These preliminary findings provide a baseline for the subsequent multi-group SEM analysis. While physical provision (place dependence) is relatively equitable, the shift to emotional belonging (place identity) remains geographically and socially stratified.

(2) The model comparison tests indicated significant differences in the attachment mechanisms to green spaces for respondent identity (p = 0.03) and community type (p = 0.048). The result justifies the need for separate group analyses (

Table 9. Results of structural invariance tests.

Difference	Model	DF	CMIN	P 1	Result
identity	structural weights	27	58.315	***	significant difference
age	structural weights	27	39.744	0.054	marginally significant
community type	structural weights	54	72.401	*	significant difference

¹ Significant difference (CMIN, p < 0.05) indicates that the structural relationships in the model are not equivalent across the compared groups. * p < 0.05 *** p < 0.001.

).

Specific path differences emerge regarding green spaces utilization (

Table 10. Path coefficients comparison and critical ratios.

Path	Resident β	Visitor β	|C.R.| 1
DR--->PD	0.404 *** 2	0.040 (NS 3)	4.042
PE--->PD	−0.064 (NS)	0.343 ***	4.127
PD--->CB	0.420 ***	0.168 *	2.612
Path	<45 β	≥45 β	|C.R.|
MC--->PD	0.330 ***	0.021 (NS)	3.134
PD--->PI	0.207 **	0.473 ***	3.086
Path	Type1 β	Type2 β	|C.R.|
DR--->PD	0.416 ***	0.026 (NS)	3.544
Path	Type2 β	Type3 β	|C.R.|
MC--->PD	0.046 (NS)	0.293 ***	2.095
DR--->PD	0.026 (NS)	0.302 ***	2.394
PE--->PI	0.390 ***	0.131 (NS)	2.235

¹ Critical ratio (CR). A path difference is significant if $\left|CR\right|$ > 1.96. ² * p < 0.05 ** p < 0.01 *** p < 0.001. ³ Not significant (NS).

). Daily routines significantly predicted place dependence among residents (β = 0.404, p < 0.001) but showed a nonsignificant association for visitors (β = 0.040, ns). Conversely, public events significantly influenced place dependence only for visitors (β = 0.343, p < 0.001). The path from place dependence to civic behavior was significantly stronger for residents (β = 0.420) than for visitors (β = 0.168).

For the younger cohort, macro-spatial character significantly drove place dependence (β = 0.330, p < 0.001), whereas this effect was not significant for the older cohort. However, the path from place dependence to place identity was significantly stronger for the older cohort (β = 0.473) than for the younger group (β = 0.207).

The path coefficients revealed significant disparities in how green space usage relates to attachment. The path from daily routine to place dependence was significant in work-unit communities (type 1, β = 0.416) and urban–rural resettlement communities (type 3, β = 0.302), contrasting with a nonsignificant path in commodity housing communities (type 2, β = 0.026). Furthermore, public events significantly fostered place identity in type 2 communities (β = 0.390, p < 0.001), but this path was not significant in type 3 communities (β = 0.131).

5. Discussion

5.1. Distinct Driving Mechanisms of Place Attachment in Heterogeneous Communities

(1) Comparison of path coefficients of public green spaces. The path linking attributes to place dependence and identity reveals several key conclusions (Figure 9). Based on these results, several key conclusions emerge.

The pragmatic logic of place dependence: SEM results reveal that practical needs drive place attachment formation in heterogeneous communities. Micro-scale design (β = 0.276) and everyday social atmosphere (β = 0.239) emerge as the dominant predictors of attachment. This finding suggests that, in socially complex environments, macro-spatial character (β = 0.161) is less effective, likely because reconciling divergent aesthetic preferences across heterogeneous groups is difficult. Conversely, within the context of park construction, user-friendly features such as barrier-free greenways and ergonomic outdoor seating meet residents’ practical needs, laying a solid foundation for functional dependence. Unlike tourism contexts where visual aesthetics dominate experiential attachment [83,84], this study confirms that in everyday residential settings, functional comfort and tangible design details outweigh purely visual stimuli.

The meaning construction of place identity: Distinct from place dependence, the formation of place identity hinges on meaning construction. Organized public events (β = 0.314) surpass static landscape attributes, vegetation coverage, and paving factors as the strongest exogenous variable driving place identity. Social interaction plays a pivotal role in deepening emotional bonds. Thus, social activation within urban parks plays a more pivotal role in fostering emotional bonding than physical landscape attributes alone.

The dual role of daily routine: Notably, the daily routine serves a dual function within the model. It serves as both the cornerstone of place dependence (β = 0.239) and a catalyst for place identity (β = 0.181). Daily routines act as the primary explanatory variable in the model. This latent variable is defined by three observable indicators: daily outdoor activities, social inclusivity, and social bonding. The data show that daily routines exert the strongest influence on both place dependence and place identity. They effectively translate functional green space usage into deep emotional attachment [85].

(2) The importance-performance analysis (IPA) model further evaluates the provision of these key factors. The diagnostic assessment shifts its focus to the specific observed variables (Figure 10). The intersection of the mean factor loadings (0.815) and mean satisfaction ratings (4.517) established the coordinate origin. This intersection delineates four distinct quadrants.

The results highlight a misalignment in the top priority quadrant (high importance, low performance). Indicators falling into this area include green space accessibility (0.839, 4.433), landscape (0.851, 4.222), enclosure spatial scale (0.834, 4.377), and social bonding (0.866, 4.433). This indicates that heterogeneous communities currently suffer from scale imbalances and circulation barriers. Notably, social bonding exhibited the highest factor loading across the entire model. This underscores that for residents, the capacity of a public green space to facilitate stable social networks is paramount, surpassing the need for mere physical gathering. However, its correspondingly low performance score reveals a deficiency: while current spaces provide the physical locus for co-presence, they lack the necessary “social catalysts” to foster deep interpersonal connections. This discrepancy suggests that neighborhood interactions remain superficial and fail to translate into a robust sense of belonging.

To address the deficit in social bonding, community interventions should integrate spatial, cultural, and institutional strategies. At the spatial and cultural level, creating tangible platforms for sustained interaction is essential. The co-management of shared community gardens, alongside the organization of culturally rooted local festivals, provides critical venues that bridge demographic divides. These physical and cultural anchors transform residents from passive users into active spatial stewards. By generating shared experiences, these collaborative practices deeply embed residents’ place attachment to the local environment.

Although micro-scale elements and public events do not fall into the most critical improvement zone, their internal weighting offers strategic direction for future micro-renewal. Within the micro-environment, the weight assigned to ecological and natural features (0.809) significantly outweighs that of garden ornaments (0.784). This indicates a resident preference for function over form. Regarding public events, management maintenance (0.839) is prioritized over organized activity or green space engagement. The analysis reveals a strategic shift toward pragmatic functionalism. Future micro-renewal should prioritize substantive ecological restoration and fundamental maintenance. These elements outweigh superficial ornamentation or sporadic social activities. This approach directly addresses residents’ core demands for high-quality living environments.

5.2. Disconnecting Emotion from Behavior

The structural equation modeling results reveal a distinct hierarchy in the factors associated with civic behavior. Place identity exerts a dominant influence (β = 0.439, p < 0.001), surpassing the effect of place dependence (β = 0.313, p < 0.001). This empirical evidence corroborates the theoretical assertion of Vaske [86], who posits that proactive stewardship is rooted more deeply in symbolic emotional bonds than in functional reliance. However, research indicates a significant gap between residents’ high PI and lower levels of CB [87]. Specifically, 65.81% of respondents reported that their place identity scores exceeded their civic behavior scores (Figure 11). High-quality public green spaces successfully foster deep emotional bonds and a sense of belonging. However, this discrepancy suggests that psychological states do not automatically translate into active stewardship or voluntary maintenance behaviors [88].

Expanding beyond psychological factors like spatial disturbances, this “value-action gap” is heavily constrained by external response barriers. First, demographic constraints play a critical role, because working-age adults (15–59 years) dominate the dataset. These younger residents often face severe “time poverty” due to demanding work schedules. This time, poverty strictly limits their capacity to engage in civic behaviors, regardless of their emotional attachment. Second, at the institutional level, there is a pervasive lack of formal, accessible, and transparent channels for community governance. Particularly in certain community types, this lack of formal institutional channels directly prevents deep emotional attachment from becoming tangible civic action.

Third, the traditional top-down governance model of urban public spaces structurally positions residents as passive beneficiaries rather than active co-producers. In China, municipal departments or commissioned agencies typically maintain open public green spaces. This provision cultivates a strong reliance on formal authorities. It inadvertently diminishes residents’ perceived responsibility and awareness for voluntary spatial stewardship. Furthermore, ambiguous spatial property rights and a lack of legal or financial incentive mechanisms for voluntary micro-transformations often deter residents from taking action.

To bridge this gap, future green space management could move beyond simply increasing green coverage and focus on targeted institutional innovations. Planners should develop flexible, digital engagement platforms to overcome the strict time constraints of younger, working-age residents. Establishing accessible co-governance frameworks is equally essential to address the lack of formal participation channels. This structural shift can effectively transition residents from passive beneficiaries to active co-producers. Furthermore, local governments should clarify maintenance boundaries and introduce practical incentives to resolve spatial ambiguity. These combined efforts will genuinely empower residents to engage in daily green space stewardship and voluntary horticultural micro-transformations.

5.3. Mechanistic Divergence in Heterogeneous Community Green Spaces

(1) Residents consistently exhibit higher attachment and civic behavior scores than transient visitors (Figure 12a). The MGA identifies distinct behavioral patterns between these two groups [89]. For residents, the prediction of daily routines on place dependence suggests that their place attachment relies heavily on consistent functional activities. Furthermore, this dependence smoothly translates into civic behavior. Conversely, transient users lack this functional foundation; their interaction with the space tends to be temporary and instrumental.

In contrast, visitors engage in episodic spatial consumption [90]. Their attachment relies heavily on the stimulation of public events. This behavioral divergence explains the functional conflict within the data. High-intensity public events attract visitors but disrupt environmental stability. Residents view community green spaces as a semi-private extension of their domestic environment. High-intensity external events are often perceived as disturbances that exceed the community’s social carrying capacity, disrupting their sense of environmental stability [91,92]. Mitigating this conflict requires integrated spatial-temporal strategies. Planners must use spatial zoning to direct high-intensity activities toward the periphery or transit nodes. Simultaneously, staggered scheduling can restrict large-scale events during residents’ core resting and commuting hours. This approach accommodates the spatial vitality desired by transient users while safeguarding the environmental stability required by residents.

(2) Generational divergence and green infrastructure quality.

Statistics indicate only marginal differences in overall attachment scores between age groups (Figure 12b). However, the underlying structural pathways reveal a different reality. Age significantly influences the perception of green space attributes. Younger demographics view green spaces as integrated nodes within a broader urban network. Their place attachment correlates with accessibility, macro-spatial quality, and systemic connectivity. Younger users’ preference suggests a demand for spatial quality and systemic connectivity. Poor accessibility or rigid boundaries reduce their willingness to use the space.

In contrast, older adults experience reduced mobility. They utilize parks primarily for localized, static socialization. It makes their attachment more reliant on specific, everyday facility usage. This generational dichotomy suggests that standardized, uniform planning fails to address the heterogeneous demands of different age cohorts, highlighting the need for age-responsive landscape strategies.

(3) The analysis reveals distinct disparities across community types (Figure 12c). Urban–rural resettlement communities (type 3) exhibit the lowest scores. This disconnect suggests that conventional, consumption-oriented urban landscape design is misaligned with the agrarian lifestyles and kinship-based networks retained by resettled populations. Therefore, planning for green space in these contexts should move beyond uniform, ornamental approaches and prioritize functions tied to daily life. Parks can convert portions of purely visual greenery into participatory gardens or community plots. Sufficient paved areas should also be reserved to accommodate traditional uses of public space, such as laundry drying, local festivals, and ceremonial events. This functional integration encourages informal social interaction and fosters spontaneous stewardship. Transforming these landscapes into multifunctional venues significantly strengthens emotional connections.

Similarly, transitional work-unit decline communities (type 1) exhibit a profound dependence on daily routines to build place attachment. This reflects the legacy of the “danwei” system. Strong collective social ties persist despite physical spatial decay. Residents rely on community green spaces as vital extensions of their aging domestic environments. Their attachment is deeply rooted in daily, localized interactions. Therefore, urban renewal in these communities should focus on “micro-regeneration.” Park should prioritize the upgrading of basic functional elements such as seating, paved areas, and accessible ramps.

In contrast, residents of commodity housing communities rarely derive place attachment from basic daily routines or the static physical morphology of the green space. Instead, their place identity is powerfully catalyzed by public events. This divergence highlights a mechanism of spatial consumption typical of urban middle-class lifestyles. They perceive them as experiential venues for social vitality and lifestyle expression. Consequently, their attachment stems from the dynamic atmosphere generated by organized activities. To support this socio-spatial demand, park planning in these areas should prioritize programmatic flexibility. Designers should prioritize adaptable, open spaces to support diverse public events, such as accommodating temporary exhibitions, markets, and cultural festivals. Here, the curation and management of social activities become more critical than the mere provision of daily facilities.

5.4. Theoretical Contributions

This study offers three distinct theoretical contributions to the literature on place attachment and urban governance:

(1) This research extends place attachment theory to typologically heterogeneous communities. It challenges the traditional linear assumption that usage inherently generates attachment. The study empirically identifies the phenomenon of functional decoupling in commodity housing communities where daily routines fail to drive place dependence. Furthermore, the identity deficit in urban–rural resettlement communities is where public events fail to foster place identity.

(2) The data empirically validate the S-O-R framework in community studies. The MGA reveals that the younger cohort (< 45) is significantly driven by macro-spatial accessibility, whereas the older cohort (≥45) relies on functional conversion to form identity. Furthermore, it validates the stewardship effect among residents, who show a stronger translation from place attachment to civic behavior compared to migrants. This study provides a validated tool for analyzing the pathway from space to action in non-Western urban settings.

(3) Previous studies often emphasize choice-based belonging or positive social contact. In contrast, this research identifies different mechanisms inherent to the heterogeneous communities’ transitional context. High-quality spatial provision does not automatically translate into place identity. Public events can occasionally trigger territorial defensiveness rather than social integration.

(4) The findings establish a theoretical foundation for differentiated urban governance. The SEM and IPA analyses identify a significant impact of a low performance gap. Social bonding exerts the most decisive influence on public green space, yet its current satisfaction scores are low. Recognizing that physical space provision alone cannot trigger meaningful interaction, this research advocates a theoretical shift from uniform material interventions to a comprehensive paradigm of socio-spatial empowerment. This framework operates across three interconnected dimensions: spatially, facilitating physical co-presence through shared communal nodes; culturally, providing platforms for collective experiences; and institutionally, establishing formal channels for community-led initiatives to deliberate public affairs. Ultimately, the synergy of spatial, cultural, and institutional empowerment effectively fosters mutual trust, addressing the root causes of social disassociation in heterogeneous environments.

5.5. Limitations

Constrained by research duration and objective conditions, the current analysis has several limitations that offer avenues for future research:

First, cross-regional validation of sample breadth. Our findings are grounded in the specific context of Beijing. However, urban fabrics and cultural contexts vary significantly across the globe. Therefore, cross-regional validation is required to capture these potential variations and enhance theoretical universality.

Second, methodological refinements to address data biases. Questionnaire surveys efficiently capture self-reported perceptions but may overlook inactive residents due to mobility or scheduling issues, skewing results toward active users. Furthermore, uncontrolled confounding factors such as community management and income levels remained outside the scope of this analysis. Future studies could employ mixed-methods approaches, including interviews and observations, to develop a more comprehensive representation.

Third, the cross-sectional design of this study inherently imposes temporal constraints. Consequently, the findings show the significant spatial and statistical correlations between community characteristics and place attachment at a specific point in time. To track evolving social dynamics and further validate these relationships, future urban renewal projects should incorporate post-occupancy evaluations (POEs) as a longitudinal data collection mechanism. By conducting repeated surveys over time, researchers can better reveal the long-term emotional impacts of spatial interventions and assess their sustained effectiveness.

6. Conclusions

Place attachment in heterogeneous community public green spaces emerges as a complex, multi-dimensional construct. Micro-scale elements and daily routine serve as the foundational pillars. These factors are crucial in heterogeneous communities, where social cohesion relies on physical details and functional support. The SEM confirms that functional comfort outweighs macro-spatial aesthetics in these residential settings. Among all observed variables, social bonding exhibits the highest importance yet records the lowest performance score.

Attachment logic varies significantly across different demographics. Younger demographics are more sensitive to macro-spatial characteristics. Conversely, public events failed to generate significant attachment among original residents. The data suggests that current event strategies fail to resonate with the local population. Such interventions may inadvertently alienate long-term residents rather than foster integration.

The empirical scope focuses primarily on high-density contexts like Beijing. The derived typological framework demonstrates strong applicability to similar metropolitan areas experiencing rapid urbanization, housing commodification, and socio-spatial segregation. Rather than uniform renewal, this research suggests typological interventions as a potential framework. This tailored approach may better address the distinct socio-spatial needs of heterogeneous communities. However, urban fabrics and cultural contexts vary significantly across the globe. Generalizing these specific micro-indicators to low-density, rural, or entirely different cultural contexts requires caution.

For commodity housing (type 2), neighborhood park governance heavily benefits from “active social curation”. Strategies might consider emphasizing high-quality, thematic public events to leverage place identity. These events, acting as “social catalysts”, could specifically involve parent–child festivals and community markets. Park layouts can be designed to offer flexible event containers rather than relying solely on fixed ornamental landscapes, thereby accommodating diverse social programming. Planners are encouraged to position high-intensity event containers at the park edges or near transit nodes where feasible. Market squares and festival zones serve as excellent models for these high-intensity areas. Concurrently, quieter, internally sheltered buffer zones could be preserved primarily for daily residential use. This approach seeks to accommodate experiential spatial consumption while protecting the community’s psychological boundaries.

For work-unit declined communities (type 1) and resettlement communities (type 3), green space interventions need to extend beyond standard planting to offer specific, replicable interventions. Planners should prioritize barrier-free accessibility and functional facilities. Installing accessible ramps represents a crucial step in this direction. Furthermore, specific physical details can be thoughtfully integrated into the landscape to support residents’ daily routines better. These details primarily include well-designed seating, appropriate paving, and fitness equipment. To address the place identity deficit, event strategies might pivot from generic activities toward culturally rooted local festivals. Incorporating elements of collective memory and local culture into green space programming can help bridge the emotional gap caused by displacement. Micro-retrofitting interventions can also gently formalize residents’ spontaneous, localized uses of space. Spatial interventions could transform underutilized paved corners into community-managed gardens or designated, sunlit zones for traditional domestic extensions. Ultimately, these targeted strategies provide low-pressure, institutionalized opportunities for heterogeneous groups to interact, transforming abstract emotional bonds into actionable civic integration.