Introduction

Old neighborhood regeneration is the micro-scale of urban regeneration1,2, focusing on people-oriented localized organic restoration and micro-scenario regeneration3. Old neighborhood regeneration attempts to address issues such as habitat degradation, environmental pollution, continuation of historical structures, and social consistency4,5 to create livable places that serve and apply to people. Compared with large-scale redevelopment, old neighborhood regeneration emphasizes public participation in following the people-oriented philosophy and sustainable development vision6. Consensus-based participation produces higher quality and creative outcomes7, reduces costs and increases acceptance and satisfaction8, mitigates the risk of single-agent dominance9, and addresses social inequalities from regeneration10. This is a crucial step toward sustainable communities11.

Old neighborhood regeneration is becoming an essential strategy for cities to develop sustainably12,13 and is receiving increasing attention in China14. According to the National Bureau of Statistics and the Ministry of Housing and Urban–Rural Development of China, in 2022, the country completed the regeneration of 52,500 old neighborhoods, involving 8.76 million households. By 2025, it is planned to complete the regeneration of 219,000 old neighborhoods. Despite the active regeneration of old neighborhoods in China and plans to expand these programs, residents’ participation remains low due to reliance on the traditional government-dominated model. This poses challenges to the sustainability of regeneration, including higher maintenance costs15, lower satisfaction and benefits16, and discontinuous management leading to regression17. Over time, this will deplete regeneration achievements and waste national resources. Therefore, understanding and mobilizing the drivers of residents’ participation has become urgent for sustainable neighborhood regeneration.

Most old neighborhoods in China are “danwei compounds” built in the 1970s and 1980s, combining work and residence. Residents are closely connected through karmic and geographic ties18, forming a typical “acquaintance social network” that covers all aspects of their lives. According to the Embeddedness Theory, the behavioral mechanisms of residents are deeply embedded in their social networks and are impacted by other individuals in their social networks19. Therefore, exploring the mechanisms by which residents’ social network embeddedness impacts their participation intention in regeneration in the acquaintance society context has significant practical implications. From a psychological perspective, behavior arises from an intricate psychological journey. As a psychological model for predicting and explaining personal behavioral intentions, the Theory of Planned Behavior (TPB) can offer a sound theoretical framework for explaining these complexities20. However, its application tends not to directly consider the social structural variables themselves as viable determinants21 and, to some extent, ignores the fact that individuals are socially characterized22. Therefore, it is necessary to combine social network embeddedness with the TPB model to provide insights into the externalization-internalization process involved in residents’ participation in old neighborhood regeneration23.

As a unique organizational structure and system in China, the “danwei system” is essential in understanding the differences in residents’ participation in old neighborhoods. However, few scholars have considered this factor in public participation studies. The danwei system in China was established during the planned economy era when the government managed society comprehensively through danwei. Residents entered the danwei system through national planning and allocation. The danwei were not only workplaces but also provided comprehensive welfare benefits such as housing, healthcare, education, etc. Walder highlighted the uniqueness of the Chinese danwei compared to Western corporate organizations, noting that individuals not only receive resources like material goods and opportunities but also develop a dependence on the danwei24, shaping their values and behavior25. With China’s economic reforms and the rise of the market economy, the influence of the danwei system in some old neighborhoods has weakened or disappeared. However, its characteristics persist in the family compounds of government agencies, public institutions, and universities. This transitional period is termed the “post-danwei period”26. Existing research has not addressed the danwei system’s role in the relationship between social network embeddedness and residents’ participation during this period.

Therefore, to deeply explore the relationship between social network embeddedness, TPB, danwei system, and residents’ participation intentions and the processes involved, this study constructed a theoretical framework with social network embeddedness as a precursor factor, TPB-related variables as multiple mediators, and danwei system as moderators, and test the validity of this framework among residents of China’s old neighborhoods. Through this moderated multiple mediator model, this study attempts to enrich existing related research in three aspects: First, compared to exploring each mediating pathway in isolation, multiple mediating effect analysis based on TPB provides a deeper understanding of the impact and mechanisms of social network embeddedness on participation intention, as well as the differences and interactions between various mediating effects. Second, based on China’s unique organizational structure and governance system, we incorporate the variable of the danwei system into the original TPB model of the related study for the first time, aiming to explore its moderating role between social network embeddedness and residents’ participation intention. Third, despite the widespread use of TPB, our understanding of the antecedents and origins of its associated variables is still limited. Based on the acquaintance society of old neighborhoods, this study extends the theoretical framework of TPB by explaining the formation mechanism of intrinsic factors such as residents’ attitudes, subjective norms, and perceived behavioral control of participation in regeneration. Based on exploring the above issues, this study aims to identify the critical influence pathways of social network embeddedness on participation intention, providing crucial intervention information for the government and policymakers. The remainder of the paper is organized as follows: “Literature review and research hypotheses” section presents a literature review and the research hypotheses. The methodology and theoretical model are introduced in the following section. “Results” section validated the hypotheses and showed the results. “Discussion” section discusses the study’s results. The study’s conclusion is given at the end.

Literature review and research hypotheses

The impact of social network embeddedness on participation intention

“Embeddedness” is the fundamental theoretical position of the social network analysis paradigm, emphasizing the intrinsic connection between individual behavior and the social network it embeds in19. In social life, an individual’s behavior is not independent but “embedded” in a particular social network and influenced by others in the network27. “Embeddedness” is a multi-dimensional structure. Granovetter deconstructed “embeddedness” from two dimensions: relational and structural embeddedness19. Following Granovetter, many scholars have expanded the “embeddedness” analytic framework28,29,30. Among them, the impact of cognitive embeddedness on individual behavior and intention has received widespread attention from scholars31,32,33. Cognitive embeddedness describes a shared value or paradigm that allows people to have a common understanding of the appropriate way to behave34. Based on existing literature, this paper describes the social network embeddedness of residents from three dimensions: relational embeddedness, structural embeddedness, and cognitive embeddedness.

Firstly, relational embeddedness primarily emphasizes the network relationships between residents and neighbors during social interactions, typically including relationship frequency, strength, and relationship quality34. In old neighborhoods, residents are closely connected and frequently interact, sharing information and resources. This close-knit network fosters reciprocity and trust among residents, promoting active cooperative behaviors35. Due to aligning with social norms and public expectations, such positive behaviors are more easily spread and accepted within close-knit social networks36. In other words, in old neighborhoods with deep relational embeddedness, residents are more likely to be influenced by positive participation behaviors.

Secondly, structural embeddedness focuses on the individual network structure where residents are embedded, typically including network size, density, and position37. Granovetter’s concept of “embeddedness” indicates that an individual’s economic and social behaviors are influenced by their social network structure19. In old neighborhoods, residents in a central position or have numerous connections are more likely to receive detailed information and positive feedback about neighborhood regeneration, enhancing their intention to participate. Key nodes in this network structure can play a role in information dissemination and demonstration, encouraging more residents to actively participate in regeneration.

Finally, cognitive embeddedness refers to residents’ recognition of neighborhood values, goals, and norms. When positive participation behaviors within the neighborhood are widely recognized and promoted, residents will learn and imitate by observing others’ behaviors and outcomes, forming consistent cognitions and thus enhancing their intention to participate38. In the context of active advocacy by the government, participating in neighborhood regeneration is seen as behavior that aligns with injunctive norms39. Residents with deeper cognitive embeddedness are more likely to accept and respond to these norms, thus actively participating in regeneration.

Therefore, the hypothesis below is suggested:

H1

Social network embeddedness significantly positively affects residents’ intention to participate in old neighborhood regeneration.

The impact of TPB variables on participation intention

TPB is a social-cognitive model that posits that intention is determined by attitude, subjective norm, and perceived behavioral control. A favorable attitude and supportive subjective norm motivate behavior, with specific intentions forming when perceived behavioral control is strong40.

Attitude is an evaluative response of an individual toward an object or behavior, encompassing cognitive, affective, and behavioral components41. The cognitive mechanism involves an individual’s expectations and evaluations of behavioral outcomes. A positive attitude indicates that residents believe participating in regeneration can improve their neighborhood environment and living conditions, enhancing their intention to participate20. The affective mechanism involves an individual’s emotional reactions and feelings toward the behavior. Positive emotional reactions can improve an individual’s intrinsic motivation, strengthen their behavioral intentions, and prompt them to act accordingly42. The behavioral inclination mechanism refers to an individual’s intentions and readiness to act. A positive attitude strengthens residents’ behavioral inclination and readiness to act, thereby supporting neighborhood regeneration through actual participation. This is a critical step in transforming attitude into intention43.

Therefore, the hypothesis below is suggested:

H2

Attitude significantly positively affects residents’ intention to participate in old neighborhood regeneration.

Subjective norms refer to an individual’s perception of external pressure from others when deciding to perform a specific behavior. They can enhance behavioral intentions through normative beliefs and compliance motivation44. Normative beliefs are an individual’s perception that important others want them to engage in a particular behavior. Positive expectations from significant others increase the likelihood of developing corresponding behavioral intentions. Compliance motivation refers to the degree of an individual’s willingness to follow the expectations of significant others. Residents with high compliance motivation are more likely to actively participate in regeneration activities to meet these expectations and maintain good social relationships45. In the highly interdependent social structure of old neighborhoods, normative beliefs, and compliance motivation have a more significant impact on individual behavior46. When residents perceive that their neighbors generally expect them to participate in regeneration, their intention is enhanced.

Therefore, the hypothesis below is suggested:

H3

Subjective norm significantly positively affects residents’ intention to participate in old neighborhood regeneration.

Perceived behavioral control refers to people’s subjective assessment of the difficulty of carrying out a particular action20. Perceived behavioral control can enhance an individual’s confidence and expectations of behavioral outcomes, thereby enhancing their intentions40. If residents perceive that they have sufficient resources, time, and skills to participate, their confidence will be improved. Perceived behavioral control can also boost an individual’s intrinsic motivation, making them more willing to take action47. Furthermore, perceived behavioral control is significant in the context of social support and resource availability. When residents perceive sufficient support and resources from neighbors, the community, and the government, their perceived behavioral control increases, enhancing their intention to participate48.

Therefore, the hypothesis below is suggested:

H4

Perceived behavioral control significantly positively affects residents’ intention to participate in old neighborhood regeneration.

Multiple mediating roles of TPB variables

The mediating role of attitude

Firstly, residents’ embeddedness in the neighborhood network can enhance their positive attitude towards participating in regeneration. Relational embeddedness can improve individuals’ sense of belonging and responsibility towards community affairs through frequent and close interactions, helping residents form a positive participation attitude49. Individuals in core positions are more likely to obtain positive information and feedback about neighborhood regeneration, which helps maintain a positive attitude50. Cognitive embeddedness deepens individuals’ recognition of neighborhood values and goals, further strengthening their positive evaluation of participation in regeneration51.

Secondly, a positive attitude acts as a mediator, enhancing the impact of social network embeddedness on the participation intention in regeneration. A positive attitude encompasses a favorable cognitive evaluation of the regeneration results and a positive emotional response to the participation process52. This aligns with mainstream positive values and public expectations, making spreading within the community network easier and strengthening residents’ intrinsic participation intentions.

Therefore, the hypothesis below is suggested:

H5

Attitude positively mediates the effect of social network embeddedness on residents’ intention to participate in old neighborhood regeneration.

The mediating role of subjective norm

Firstly, social network embeddedness can significantly enhance individuals’ subjective norms towards participation in regeneration. Close and deep interactions can significantly improve individuals’ perception of others’ expectations, thereby forming subjective norms53. In Chinese collectivist culture, residents emphasize caring for others, integrating into society, and harmonious interdependence in their neighborhood interactions54. When some residents actively participate in neighborhood regeneration, individuals feel social pressure and compliance motivation, believing they should also participate.

Secondly, subjective norms further enhance the impact of social network embeddedness on participation intention. Deep embeddedness in social networks makes individual behavior susceptible to the influence of those around them, with subjective norms playing a key role in this process53. This not only directly influences their behavioral intentions but also enhances their sense of social identity and belonging, making the positive impact of social network embeddedness on participation intentions more significant55.

Therefore, the hypothesis below is suggested:

H6

Subjective norm positively mediates the effect of social network embeddedness on residents’ intention to participate in old neighborhood regeneration.

The mediating role of perceived behavioral control

Firstly, social network embeddedness can enhance individuals’ perceived behavioral control by increasing their access to resources and social support. When residents have extensive social connections and close interactions in old neighborhoods, they can more easily access resources, information, and other social support related to neighborhood regeneration56, which helps enhance their belief in their control over participating in regeneration actions.

Secondly, perceived behavioral control can further enhance the impact of social network embeddedness on the intention to participate in regeneration. Through embedding in social networks, residents more easily develop communication and cooperation. In this process, residents’ perceived behavioral control over participating in regeneration is enhanced, and they are more likely to create collective efficacy57. This positive feedback can modulate and strengthen individuals’ beliefs in their abilities and outcomes58 and reduce uncertainty59, thereby indirectly promoting their behavioral intentions.

Therefore, the hypothesis below is suggested:

H7

Perceived behavioral control significantly mediates the effect of social network embeddedness on residents’ intention to participate in old neighborhood regeneration.

The moderating role of the danwei system

In the period of danwei society, Chinese society showed the characteristics of “high integration” and “low differentiation.” The state fully occupies and controls various social resources and realizes the overall social integration by following the path of “state-danwei-individual.” Thus, the danwei exists and plays a role as a complex field, demonstrating strong coverage60. The “danwei people” form a closely connected danwei community in the neighborhood and condense into the communication logic and behavioral habits that match the danwei society61.

However, since implementing China’s state-owned enterprise reform system, some state-owned enterprises and factories have been evacuated from their affiliated family compounds due to bankruptcy and closure, and their residents have gradually been transformed from “danwei people” to “non-danwei people.” With the “withdrawal” of the danwei system, the traditional close-knit danwei community gradually disintegrated, and the collective consciousness was shaken, which brought about a tendency towards the “atomization” of social ties among residents62. The impact of danwei social capital embedded in the neighborhood network on residents’ behavior is gradually diminishing63.

Simultaneously, some old neighborhoods remain where the original danwei has not yet retired. The danwei is still the primary unit of community governance. Under such circumstances, residents’ “danwei attributes” have even been strengthened in some places64. The daily interactions between residents have a distinctly danwei character. This close comradeship based on danwei social networks can inspire a strong sense of collective honor and belonging among residents, encouraging a proactive attitude towards participation65,66. Meanwhile, the authoritative capital of the danwei, through its core position and influence in the network, can effectively deliver positive mobilization messages, promote positive energy, and mobilize neighborhood elites to guide residents to participate in regeneration. It can also enhance residents’ perception of subjective norms regarding neighborhood regeneration through internal social networks. Under this mechanism, residents are more likely to adopt “compliant participation” to gain recognition from the danwei63.

Additionally, as a “functional unity” organization, the danwei carries out many social and political functions in addition to those based on professional specialization67, such as organizing learning, training, and promoting regeneration actions. These activities can enhance residents’ consensus on value and cognition, giving them more information, channels, and opportunities to participate. Therefore, compared to non-danwei people, danwei people have stronger cognitive and control beliefs about neighborhood participation. Thus, we suggest that the danwei system plays a significant positive moderating role in the first stage of the relationship between social network embeddedness and participation intention, as hypothesized below:

H8

The danwei system positively moderates the effect of social network embeddedness on attitude. That is, in the positive impact of social network embeddedness on attitude, the danwei people are more robust than the non-danwei people.

H9

The danwei system positively moderates the effect of social network embeddedness on subjective norm. That is, in the positive impact of social network embeddedness on subjective norm, the danwei people are more robust than the non-danwei people.

H10

The danwei system positively moderates the effect of social network embeddedness on perceived behavioral control. That is, in the positive impact of social network embeddedness on perceived behavioral control, the danwei people are more robust than the non-danwei people.

Methodology

Data sources

This study focuses on Xi’an, China, a key city from the planned economy era with many old neighborhoods. As an essential central city in western China, Xi’an was chosen as an initial pilot for urban regeneration. In 2021, Xi’an initiated 988 new regeneration projects and completed 1,003, benefiting about 141,000 households. This high volume of projects provides a strong research foundation. The ongoing governmental focus on regeneration in Xi’an facilitates data collection. Moreover, Xi’an’s similarities with other new first-tier cities in China make it a valuable case for replicating and improving regeneration efforts elsewhere (Fig. 1).

Figure 1
figure 1

Overview of Xi’an: (a) the location of Xi’an in China; (b) the distribution of administrative divisions in Xi’an.

Full size image

Three academics and two community practice experts were invited to assess the questionnaire to ensure the appropriateness of the scale questions. The pilot survey was conducted from May 16 to May 31, 2022, in the Xi’an University of Architecture and Technology family compound. We distributed 50 pilot survey questionnaires and collected 43 valid ones, which exceeds the minimum relevant sample size suggested by Gay68 and is no less than 3–5 times (or 5–10 times) the maximum number of item subscales in the questionnaire69,70. After the pilot survey, we refined certain items on the scale, such as simplifying and popularizing the language, improving question clarity, adjusting the number and content of response options, optimizing the question order, and splitting overly long or complex questions, making them more suitable for the current situation. Pilot survey data is not included in the final data analysis.

The formal survey was conducted from June to August 2022. To ensure sample representativeness, we used multi-stage sampling. First, we randomly selected 30–35 old neighborhoods from each of Xi’an’s six central districts (Xincheng, Beilin, Yanta, Lianhu, Baqiao, and Weiyang), totaling 194 sample neighborhoods (Fig. 2). With community managers’ consent, the research team entered the neighborhoods and randomly selected 4–6 residents from each for interviews and distributed questionnaires. We distributed 1053 questionnaires and collected 975 valid ones. Among the respondents, 43.1% were males and 56.9% were females. The proportion of residents aged between 61 and 70 is the highest, at 31%. 71.7% of the interviewed residents were homeowners, and 28.3% were tenants. In addition, 52.21% of the residents’ workplaces (danwei) have not yet “retired” from the affiliated neighborhoods.

Figure 2
figure 2

Distribution of sample neighborhoods.

Full size image

In addition, we must make the following declaration:

We confirm that our methods were carried out in accordance with relevant guidelines and regulations.

We confirm that all experimental protocols were approved by the Institutional Review Board of Xi’an University of Architecture and Technology.

We confirm that informed consent was obtained from all participants for our study, that participant anonymity and confidentiality were assured, and that participation was entirely voluntary, in accordance with the ethical principles outlined in the Declaration of Helsinki.

Measures

The survey scale contains four sections: demographic characteristics, social network embeddedness, TPB, and the danwei system. The demographic information consisted of question items such as gender, age, education, income, etc., and was measured using a nominal scale. We referenced scales from Burt71, Moran72, Capioto et al.73, and Kramer74 to develop nine relational, structural, and cognitive embeddedness items to evaluate residents’ social network embeddedness. In the TPB section, the measurements of attitude, subjective norm, and perceived behavioral control were adapted from the classic scale proposed by Ajzen20 and Taylor and Todd75 and modified to fit the context related to residents’ participation in neighborhood regeneration. The social network embeddedness question items and the TPB were based on a 5-point Likert scale.

Drawing on Wang’s61 research, we measured the danwei system through two question items, “Are you an employee of the danwei affiliated with your neighborhood?” and “Is the danwei affiliated with your neighborhood still present?” Respondents chose between “yes” and “no” options. We assigned a value of 1 to those who answered “yes” to both questions and 0 to the other respondents, which allowed us to categorize the respondents into two groups: danwei people and non-danwei people. The danwei people are more hostage to the danwei system. For non-danwei people, the influence of the danwei system on them has gradually dissipated. Table 1 lists the measurement questions for each construct.

Table 1 The measurement questions for each construct.
Full size table

Analytical method

Model selection

We tested the theoretical hypotheses using Structural Equation Modeling (SEM). SEM allows simultaneous analysis of multiple factors and their relationships within a unified model. Given that relational, structural, and cognitive embeddedness are second-order factors of social network embeddedness, the choice of theoretical models is involved before the specific analysis. According to Marsh and Hocevar76, with three second-order factors, the second-order model is statistically equivalent to the first-order model, and they have the same goodness of fit. The second-order model simplifies the first-order model, effectively explaining factor covariation. Thus, this study chooses the second-order model for analysis. The research model is shown in Fig. 3.

Figure 3
figure 3

The theoretical model of this study.

Full size image

Analytical strategy

We used the two-step strategy of Anderson and Gerbing for model analysis77. Firstly, confirmatory factor analysis (CFA) was utilized to validate the measurement model. In the second step, research hypotheses were validated, and path coefficients were analyzed using SEM.

Results

Measurement model

We carried out the CFA on each construct to test the reliability and validity of the model, and the results are shown in Tables 2 and 3. According to Table 2, Cronbach’s alpha of each construct ranges from 0.775 to 0.853, higher than the 0.70 recommended by Nunnally and Bernstein78, indicating adequate reliability. Each construct’s composite reliability (CR) ranges from 0.782 to 0.853, higher than the CR threshold of 0.6079, suggesting that the internal consistency test was passed. Each construct’s average variance extracted (AVE) ranges from 0.556 to 0.663, higher than the 0.5 recommended by Fornell and Larcker80, indicating sufficient convergent validity between constructs. According to Table 3, the square root of AVE for each variable is greater than the inter-construct Pearson correlation coefficient, indicating discriminant validity between the variables80.

Table 2 The reliability and validity analysis of each measurable variable.
Full size table
Table 3 Correlations and square roots of AVE of each construct.
Full size table

Since the data in this study were obtained from respondents’ self-reports, there may be common method variance (CMV) issues81. Therefore, we adopted the methods suggested by Podsakoff et al. to evaluate the CMV issue in this study. First, we conducted the Harmon one-factor test on the variables of each construct in the model81. The results showed that the maximum covariance explained by one factor was 33.407%, less than 50%82, indicating that the study results are unlikely to be significantly affected by CMV. Then, according to Podsakoff et al.83,84, we constructed a one-factor model (Model 1) which loaded on all observed variables. We also constructed a multi-factor model (Model 2) that included attitude, subjective norm, perceived behavioral control, participation intention, relational embeddedness, structural embeddedness, and cognitive embeddedness, with each factor loading its original observed variables and all factors correlated with each other. By comparing the differences in model fit, we further assessed the CMV issue in this study, with the results shown in Table 4. The logic behind the one-factor model is that if CMV largely causes covariation among observed variables, confirmatory factor analysis should show that the one-factor model fits the data better. The test results showed that the fit indices of Model 2 were significantly better than those of Model 1 (Δ ({upchi }^{2})=4122.844, ΔDF = 21, P < 0.05), indicating no serious CMV issue in this study.

Table 4 Results of Common-Method Variance Test.
Full size table

Structural model

Model goodness-of-fit

An important assumption in SEM covariance and mean structure analysis is that the data must follow a multivariate normal distribution. Although the normality test shows univariate normality with skewness and kurtosis within acceptable ranges (absolute values below 1 and 7, respectively85), the Multivariate C.R. of 28.238 exceeds Bentler’s recommended 586, indicating a deviation from multivariate normality. This can inflate chi-square statistics, worsen model fit, and underestimate standard errors, affecting significance estimation87. To address this, we used the Bollen-Stine bootstrap p procedure to adjust for multivariate non-normality88,89, recalculating the goodness of fit during iteration with standardized residual covariance and modification index. The adjusted fit in Table 5 indicates a good model fit.

Table 5 Modified goodness-of-fit indices.
Full size table

Hypothesis testing

The structural model was tested using AMOS26.0. According to the standardized path coefficients and P-values in Fig. 4, social network embeddedness significantly impacts intention ((upbeta =0.14), (text{P}<0>), and H1 is validated. Attitude has a significant positive impact on intention ((upbeta =0.10), (text{P}<0>), subjective norm has a significant positive impact on intention ((upbeta =0.33), (text{P}<0>), and perceived behavioral control has a significant positive impact on intention ((upbeta =0.43), (text{P}<0>). H2, H3, and H4 are supported.

Figure 4
figure 4

Structural model standardized path coefficients.

Full size image

To examine the indirect effects of dependent variables through mediations, percentile bootstrapping, and bias-corrected percentile bootstrapping were performed using 10,000 bootstrap samples within a 95% confidence interval90. Percentile bootstrapping involves repeatedly sampling with replacement from the original dataset (10,000 times), calculating the mediation effect estimate for each resample, forming an empirical distribution, and extracting the 95% confidence interval. Bias-corrected percentile bootstrapping further adjusts for bias-correction and acceleration parameters, providing a more accurate 95% confidence interval. The significance of the indirect effect is tested by calculating the confidence interval’s lower and upper bounds91. As Table 6 indicates, the results of the bootstrap test confirmed the existence of a significant positive mediating effect between social network embeddedness and participation intention (total effect = 0.689, (text{P}<0>; direct effect = 0.164, (text{P}<0>; indirect effect = 0.525, (text{P}<0>). The standardized regression coefficients show that the direct effect coefficient of social network embeddedness on intention is 0.140, accounting for 23.81% of the total effect, while the standardized indirect effect coefficient is 0.448, accounting for 76.19% of the total effect.

Table 6 Total effect, direct effect, total indirect effect.
Full size table

To validate H5-H7, we examined the specific indirect effects of attitude, subjective norm, and perceived behavioral control. According to Preacher and Hayes91, the indirect effect of social network embeddedness on participation intention through attitude is calculated by multiplying the path coefficients from social network embeddedness to attitude and from attitude to participation intention. The same method is used for indirect effects through subjective norms and perceived behavioral control. The significance of these indirect effects is then tested using percentile bootstrapping and bias-corrected percentile bootstrapping methods, as shown in Table 7. Attitude, subjective norm, and perceived behavioral control all play a significant positive mediating role between social network embeddedness and intention ((0.053), (text{P}<0>; (0.269), (text{P}<0>), and H5, H6, and H7 are supported.

Table 7 Specific indirect effects and their contrasts.
Full size table

In addition, we also compared the differences between specific indirect effects. Table 7 shows significant differences between the specific indirect effects of attitude and those of subjective norm and perceived behavioral control (Attitude-Subjective norm: –(0.149), (text{P}<0>; Perceived behavioral control-Attitude: (0.216), (text{P}<0>). The results show that the mediating effect of attitude between social network embeddedness and intention is much lower than that of subjective norm and perceived behavioral control. No significant difference exists between the specific indirect effects of subjective norm and perceived behavioral control (Subjective norm-Perceived behavioral control: −(0.066), (text{P}>0.05)).

To investigate the moderating role of the danwei system, we used a multi-group structural model to test H8-H10. Multi-group structural modeling compares structural equation models across different groups to test for moderating effects by examining whether path coefficients are equal92. For example, to examine the moderating effect of the danwei system on the impact of social network embeddedness on attitude, the data sample is divided into danwei and non-danwei subgroups, and the models are fitted separately. In the constrained model, the path coefficient from social network embeddedness to attitude is equal across groups, while other coefficients are freely estimated. In the unconstrained model, all coefficients are freely estimated. The significance of the (Delta {upchi }^{2}) between the constrained and unconstrained models is used to test hypothesis H8. According to Table 8, the p-value of the (Delta {upchi }^{2}) is significant ((Delta {upchi }^{2}=7.253), (text{P}<0>). In the danwei group, the standardized estimation coefficient of Social network embeddedness → Attitude is 0.379 ((text{P}<0>), but 0.553 ((text{P}<0>) in the non-danwei group. The positive relationship between social network embeddedness and attitude is stronger in danwei people than in non-danwei people, and H8 is validated.

Table 8 Invariance test of the two-group structural model and testing of moderating effects.
Full size table

Then, following the same way, we tested the moderating role of the danwei system between social network embeddedness and subjective norm, as well as between social network embeddedness and perceived behavioral control. According to Table 8, the (Delta {upchi }^{2}) between the constrained and unconstrained models of Social network embeddedness → Subjective norm is not significant ((Delta {upchi }^{2}=2.937), (text{P}>0.05)), indicating that the positive relationship between social network embeddedness and subjective norm is not significantly different between danwei and non-danwei people, and H9 is not supported. However, the (Delta {upchi }^{2}) of the constrained and unconstrained models of Social network embeddedness → Perceived behavioral control is significant ((Delta {upchi }^{2}=14.404), (text{P}<0>), and the positive correlation between social network embeddedness and perceived behavioral control is more robust in the danwei group than in the non-danwei group. Therefore, the result supports H10.

Finally, we verified the moderating effect of the danwei system in the second stage of the relationship between social network embeddedness and intention (that is, Attitude → Intention, Subjective norm → Intention, and Perceived behavioral control → Intention). According to Table 8, as predicted, none of the moderating effects of the danwei system in the second stage is significant.

Discussion

Discussion of findings

Previous studies have explored the link between social networks and public participation93 but have not fully considered internal, external, subjective, and objective factors. Consequently, the process by which social network embeddedness influences residents’ intentions to engage in regeneration remains unclear. Additionally, the danwei system is crucial in China’s old neighborhoods, yet it is often overlooked in research on neighborhood regeneration, which lacks quantitative analysis. This study addresses these gaps by developing a moderated multiple mediation model that integrates social network embeddedness, TPB, and the danwei system. The study aims to: (1) examine the direct and indirect effects of social network embeddedness on participation intention; (2) test the specific mediating effects of the TPB-related variables and their differences between social network embeddedness and participation intention; and (3) examine how the danwei system moderates the relation of social network embeddedness to participation intention. Combined with the findings of previous studies, we discuss our empirical findings as follows.

First, previous studies have shown that social network embeddedness can promote public participation in environmental protection94, and this study demonstrates this positive correlation in the field of old neighborhood regeneration. Relative to the focus of existing studies on the direct relationship between social network embeddedness and intention95,96, we further validate that, in addition to the direct effect, social network embeddedness indirectly affects residents’ participation intention through their internal cognitive factors (attitude, subjective norm, and perceived behavioral control). Combined with standardized regression results, social network embeddedness directly promotes residents’ participation through the herd or peer effects in acquaintance communities97,98, but this effect is smaller than its indirect impact. Social network embeddedness, an external social factor, primarily influences participation intentions by internalizing residents’ subjective cognitive factors.

The significant positive effects of attitude, subjective norm, and perceived behavioral control on intention validate the TPB framework’s applicability to Chinese residents’ behavioral intention to participate in old neighborhood regeneration, consistent with previous research results99. However, compared to the literature on public engagement, the results show that attitude has a smaller effect on intention100,101,102. This reflects the inertia of old neighborhood residents’ dependence on the government in China61. Although the government promotes old neighborhood regeneration, residents lack initiative and rely on government arrangements for public affairs, even if they believe the regeneration is beneficial.

Second, and more specifically, we examined the specific indirect effects of attitude, subjective norm, and perceived behavioral control, respectively. Studies have verified the mediating effect of attitude between social capital and intention to sort waste103. Our analysis shows that attitude, subjective norm, and perceived behavioral control collectively mediate the relationship between social network embeddedness and participation intention. Deep socialization within close neighborhood relations improves residents’ attitudes, strengthens neighborhood norms, and enhances perceived control, promoting participation intention.

Comparative analysis of specific indirect effects shows that the mediating effect of attitude differs significantly from that of subjective norm and perceived behavioral control. Specifically, the specific mediating effect of attitude between social network embeddedness and participation intention is much smaller than that of subjective norm and perceived behavioral control, respectively. These results further confirm the passive psychological characteristics of Chinese residents’ participation in public affairs104. Although close neighborhood interaction improves residents’ attitudes toward participating in old neighborhood regeneration, long-term collectivist welfare dependence hinders attitude translation into participation behavior. Additionally, there is no significant difference between the mediating effects of subjective norm and perceived behavioral control in positively mediating the impact of social network embeddedness on participation intention. This may be due to the social structure of acquaintances in China’s old neighborhoods. The old neighborhoods in China have a highly coupled network structure, as described by Granovetter105. The highly coupled structure fosters close ties and cooperation among residents106, enhancing their control beliefs and confidence in participating in neighborhood regeneration. Additionally, the “face-saving” mentality in this closed network structure “amplifies” normative pressure, leading to “convergent” behavior107.

Third, our findings further highlight the significance of the danwei system in Chinese citizens’ involvement in old neighborhood regeneration. The social system reform in China has triggered extensive discussions among scholars on the danwei system26,108,109. In these studies, there has been controversy over the changes in the danwei system and its mechanism of action during the post-danwei period. Some scholars argue that, despite some danweis withdrawing, they continue to limit residents’ autonomy by exerting an “invisible presence” in neighborhoods63. Conversely, Wang highlighted the positive effect of the danwei system on environmental participation, such as successful garbage sorting in neighborhoods with danweis61. Our research examines the impact mechanisms of the danwei system in neighborhood participation through a field perspective, which can be seen as a response to these controversies. Our study shows a significant moderating effect of the danwei system in the first stage of the relationship between social network embeddedness and participation intention. Specifically, the relationship between social network embeddedness and attitude is stronger in danwei people than in non-danwei people, as is the relationship between social network embeddedness and perceived behavioral control. This implies that the impact of the danwei system on residents’ behavior varies across different fields during China’s social transition. Although the danwei as a social institution is dissolving110, it remains a key organizational model for state and government management111. The “state-danwei-individual” model persists in neighborhoods with danweis, giving danwei people more cognitive superiority and confidence in social interactions and emphasizing collective participation112. However, this positive effect diminishes as danweis close and retire. Our results indicate that the “invisible presence” of the danwei is much less influential than its “substantial presence.”

Surprisingly, the relationship between social network embeddedness and subjective norm does not differ significantly between danwei people and non-danwei people. This confirms the deep-rooted influence of traditional collectivist culture and the “face-saving” mentality on residents of old neighborhoods113. Once a mainstream behavior or consciousness is established in these social networks, inconsistent behaviors are significantly restrained, even without the danwei’s organizational influence.

Implications

First, from the perspective of community autonomous organizations, it is necessary to emphasize the vital role of community “capable persons” or “elites” in informal neighborhood networks. Our study highlights the crucial mediating role of subjective norm in promoting participation intention by social network embeddedness. Community elites, driven by social status, prestige, and honor rather than personal interests, typically have high reputations114. They foster identity and reciprocity norms, promote general trust, and enhance the effectiveness of neighborhood norms and constraints115, pivotal for community self-organization.

Second, our findings suggest that single community promotion may not be sufficient to stimulate residents’ enthusiasm for old neighborhood regeneration participation. Publicity activities may help to improve residents’ attitudes toward old neighborhood regeneration. However, residents’ attitudes play a much smaller role in mediating social network embeddedness and participation intention than perceived behavioral control. Therefore, from the perspective of government support, the government should shift its strategy from focusing solely on publicity to providing convenient participation platforms and professional guidance, such as neighborhood regeneration planners. This approach bridges residents’ needs with government expectations, breaking down top-down participation barriers116 and enhancing residents’ control beliefs about participation. Simultaneously, the government should leverage acquaintance networks and social learning mechanisms in neighborhood activities to promote the accumulation of direct and alternative experiences, fostering active participation among residents. Additionally, appropriate subsidy policies can be formulated to enhance residents’ willingness to participate, including subsidies for housing renovation, infrastructure improvement, and public service facility construction. These aim to increase residents’ enthusiasm and initiative in participating in old neighborhood regeneration through economic incentives and community mobilization. For example, implementing a housing renovation incentive policy that allows applications for housing renovation projects and corresponding subsidies if approved by a majority of homeowners (e.g., more than two-thirds). This leverages residents’ close social networks within the neighborhood to mobilize their autonomy in regeneration, encouraging bottom-up participation.

Third, different promotional strategies should be adopted for danwei-present neighborhoods and danwei-exit neighborhoods. In danwei-present neighborhoods, the government should leverage the positive role of the danwei system to lead old neighborhood regeneration efforts. The authority of the danwei network encourages consistent collective action among individuals and their families. In danwei-exit neighborhoods, community workers can harness the social capital of the original danwei by integrating its “power capital” and “relationship capital” to form a mobilization network. This key group, often central in the neighborhood network, can more easily influence other residents to participate in regeneration through personal relationships.

Limitations and future research directions

First, the data is cross-sectional, which may limit our ability to infer causality. Second, due to the varying progress of each old neighborhood regeneration project and the large sample size of data collected in this study, it isn’t easy to arrange a second wave of interviews with each respondent to report their actual participation behavior. Therefore, this study only focuses on residents’ intention to participate in the regeneration without analyzing the relationship between their behavioral intention and actual participation. However, intentions do not always produce corresponding behaviors, and the transformation process between residents’ participation intentions and their actual behaviors is our focus for future research.

Conclusion

Residents’ participation is crucial for sustainable old neighborhood regeneration. Focusing on the typical acquaintance social structure characteristics of old neighborhoods in China, this study analyzes how social network embeddedness and TPB affect residents’ intention to participate in regeneration and, for the first time, incorporates the danwei system, a China-specific governance structure, as a variable in the original TPB model. By constructing a moderated multiple mediation model, this study confirms the validity of social network embeddedness, TPB, and the danwei system as a theoretical framework for understanding residents’ intention to participate in regeneration and draws the following conclusions:

First, social network embeddedness enhances residents’ intention to participate in regeneration through direct and indirect pathways, with the indirect effect mediated by internal cognitive factors being larger. Second, residents’ attitudes, subjective norms, and perceived behavioral control in the indirect pathway have significant mediating effects. However, the mediating effect of attitude is significantly smaller than that of subjective norms and perceived behavioral control, which do not differ significantly. Third, the danwei system has a significant moderating effect in the first stage of the relationship between social network embeddedness and participation intention, and its impact on residents’ intention to participate in the regeneration varies with the field. That is, the relationship between social network embeddedness and attitude and perceived behavioral control is stronger in danwei residents than non-danwei residents.

These findings imply that social network embeddedness, an extrinsic social factor, drives residents’ participation intentions mainly through normative pressure and experience exchange among acquaintances. The danwei system reinforces this process through its “physical presence.” These insights clarify the relationship between social network embeddedness and participation intentions and offer strategies to actively motivate residents to engage in neighborhood regeneration at the intervention level.