The Impact of Social Capital on Community Resilience: A Comparative Study of Seven Flood-Prone Communities in Nanjing, China

Abstract

Social capital plays a crucial role in enhancing community resilience during flood disasters. This study investigates the influence of social capital on community resilience in Nanjing, China. Social capital is composed of five aspects: cohesion, collective efficacy, sense of belonging, trust and reciprocity and informal social control. Factor analysis and multiple regression analysis are employed to analyze the dimensions of social capital and its impact on community resilience. Our results demonstrate that social cohesion and collective efficacy are the most representative factors of social capital. Reformed housing communities typically have higher cohesion than those in commercial and affordable housing. Affordable housing communities in flood-prone areas have higher collective efficacy but lower trust and reciprocity. Commercial housing communities have higher informal social control but have great internal differences in collective efficacy. We strongly urge government decision makers to enhance flood resilience by fostering social capital within local communities.

1. Introduction

The global increase in flood risk highlights the growing importance of understanding the link between social capital and community resilience. In the past decade, China’s urbanization rate exceeded 65% in 2023, with a permanent urban population of 932 million and 117,000 urban communities [1]. Socio-spatial differentiation has increased housing inequality in major cities like Nanjing due to the expanding diversity in housing provision and consumption following economic reforms. Since the 1990s, the government in China has encouraged the commercialization of urban housing, leading to a pattern of spatial segregation in housing systems [2]. Nanjing has three types of housing, namely reformed housing 1, commercial housing and affordable housing. Reformed housing is a kind of public housing made up of work units under the central planning system. Commercial housing 2 is built for higher-income groups with access to mortgage financing. Affordable housing 3 is often provided to low-income groups to meet family needs [3,4]. Studies have confirmed that social capital plays a crucial role in enhancing social resilience to natural disasters, especially for susceptible individuals and cities with spatial heterogeneity [5,6,7].

The concept of resilience has become a fundamental paradigm for examining risks and safety threats [8,9]. Most researchers have focused on the engineering dimensions of resilience and overlooked the social aspects, particularly the lack of exploration into the interactions between social capital and community resilience. Social capital refers to the relationships of responsibility, expectation, trust and power among individuals and groups [10]. Social capital refers to the characteristics of social organizations, such as networks, norms and trust that facilitate action and cooperation for mutual benefit [11,12]. Social capital is a resource embedded in social networks [13]. With the increasing occurrence of disaster losses, experts in disaster management have embraced various forms of resilience as a means to mitigate losses and recover from their impacts [5]. The Federal Emergency Management Agency recommends that local and national responders establish partnerships among emergency management, community sectors, and organizations. They also encourage empowering local action through increased social capital and civic activity [14]. The Community Resilience Development Framework in the UK argues that social capital includes the cohesiveness of communities, barriers to inclusion, access to shared resources, behaviors and social norms

Efforts to assess social capital in the context of real disasters can be carried out through various methods, including quantitative surveys, in-depth interviews, field observations, and the analysis of statistical indicators derived from publicly available data [5]. Based on evidence from various disasters, the indices used to measure disaster resilience include the social capital index. This includes participation in nonprofit, religious and civic organizations, as well as the number of registered voters and voter participation [15,16,17]. Scholars assessed the social capital within the community through both qualitative and quantitative methods. For instance, Choo and Yoon (2022) conducted a study on the impact of social capital on disaster response capacity in Seoul using factor analysis [18]. Fraser (2021) utilized publicly available data to construct new bonding, bridging and linking social capital indices through principal component analysis [19].

This study makes several contributions to the existing literature. Firstly, social capital has been investigated, but few studies have focused on the differences in community type, especially in the context of transition periods in a developing country. Previous research has primarily focused on examining the relationship between social capital and the process of disaster recovery; for example, the Chicago Heat Wave Study found that isolated, elderly and economically disadvantaged communities have high mortality rates, with less public organization and limited social capital present in these communities. Communities with strong trust, norms and participation can recover from disasters more quickly [20]. Mutual trust and interdependence enhance awareness of disaster management, volunteer opportunities and responsibilities, thereby supporting disaster preparedness [21]. A case study in Sri Lanka confirmed the important role of social capital in rural communities’ ability to respond to disasters like landslides and floods [22]. Scholars assessed social capital before, during and after disasters. They also studied the types of social capital—bonding, bridging and linking—in relocation and reconstruction efforts [23]. The current study contributes to the existing literature by empirically analyzing the differences in community-based social capital in China. Secondly, while there have been studies on community resilience, few have specifically focused on the relationship between social capital and disaster recovery or preparedness. Furthermore, there has been a lack of examination regarding the impact of social capital on community resilience. Previous research has shown that the reconstruction of communities is significantly influenced by social networks, such as churches, and connections with relatives and friends [9,24]. Lo et al., (2015) proposes that the development of social capacity and trust within a community will be more effective in strengthening resilience than simply increasing awareness of potential hazard risks [25]. Liang et al. (2017) suggest that government support, community activities and local Hukou are key factors influencing residents’ adaptation capacity [26]. Cui and Li (2020) assess community resilience from the perspective of social capital at different stages of flood disaster. Chinese communities commonly face challenges such as limited pre-disaster awareness, passive resistance during disasters and rapid decline post-disasters, leading to the recurrence of similar situations [25,27]. Our study involves an analysis on the impact of social capital on community resilience and the simultaneous consideration of the impacts of individual characteristics and risk perception on community resilience.

Thirdly, previous research on the social capital of communities has primarily focused on developed countries and a few developing countries. Aldrich and Meyer (2015) investigated the relationship between social capital and community resilience [5]. Cutter, Ash and Emrich (2014) examined the landscape of disaster resilience indicators in the USA [28]. Fraser (2021) assessed the social capital and social vulnerability of specific communities in Japan, modeling the impact of these factors to better prepare for future disasters [19]. Shahid and Bohara (2022) proposed a social capital index (SCI) to assess formal and informal settlements in Pakistan, finding a strong positive correlation between actual and perceived social capital [29]. However, these studies on social capital in China focused on farmers’ climate-related disaster adaptation behavior, perceived collective efficacy and self-efficacy, social capital and evacuation behavior in rainstorm disasters [30,31]. Our studies focused on the different types of communities and the impact of social capital on community resilience.

Taking seven communities in Nanjing as a case study, this research examines social capital from the following perspectives: (1) developing an evaluation system for social capital; (2) assessing the comprehensive scores of social capital in different types of communities; and (3) analyzing the factors that impact community resilience, including social capital, individual characteristics and risk perception.

2. Methods

2.1. Study Area and Data

Nanjing is the capital city of Jiangsu Province and the center of the Yangtze River Economic Belt. In 2023, the land area was 6587 km², the population was 9.547 million and there was an urbanization rate of 87.2%. The city has average annual precipitation of more than 1100 mm, and its flooding season usually ranges from May to September. The Yuhuatai District is located in the south of urban center and the Yangtze River in the west, with an area of 132.39 km² (Figure 1). In 2021, the Yuhuatai District, consisting of 7 subdistricts and 64 communities, with a total population of 608,800 residents, has been severely impacted by rainstorms and floods. Our surveys conducted from April to June 2021 focused on seven flood-prone communities: Mingchengshijia community, Kangsheng Garden community, Blue Valley community, Shuiwen Garden community, Xishan apartment community, Golden Leaf Garden community and Banqiao Xincun community (Figure 2). These seven communities were selected based on their low-lying terrains and history of waterlogging. They are under the jurisdiction of five subdistricts, including Sanhongqiao, Yuhua, Tiexinqiao, Xishanqiao, Guxiong and Yuhuatai EDZ.

This research was conducted at two levels. We used the term ‘communities’ to refer to neighborhoods, and ‘residents’ to refer to individual respondents. At the community level, we designed a basic information form to collect detailed data on communities, including the type of community, year of construction, floor area ratio, green ratio and number of building stories.

Table 1. Basic information of the communities.

Street	Community	Type	Construction Year	Area (ha)	Floor Area Ratio	Green Ratio (%)	Building
Sanhongqiao	Mingchengshijia	Commercial	2009	5.32	2	36	High-story
Yuhua	Kangsheng Garden	Commercial	1997	10.34	1.5	45	Multiple-story
Tiexinqiao	Blue Valley	Commercial	2007	2.31	1.78	40	Multiple-story
Tiexinqiao	Shuiwen Garden	Reform	1982	1.3	2	35	Small high-story
Xishanqiao	Xishan apartments	Reform	1996	1.86	1.82	45	Small high-story
Yuhuatai EDZ	Golden Leaf Garden	Affordable	1990	7.04	1.6	20	Multiple-story
Guxiong	Banqiao Xincun	Reform	1990	2.86	2	28	Small high-story

presents the fundamental construction status of the surveyed communities. At the individual level, we conducted a survey to assess residents’ experiences, knowledge, and attitudes regarding flooding. The questionnaire consisted of four parts. The first part gathered basic demographic information, including gender, age, education level, occupation, and income. The second part focused on community resilience and historical flood events, such as the severity and duration of past floods. The third part examined six aspects of social capital. The fourth part addressed flood countermeasures. A total of 328 questionnaires were collected from seven typical communities using accidental and snowball sampling methods. Of these, 289 questionnaires were deemed valid, resulting in an effective response rate of 88% (

Table 2. The attributes of the samples.

Item	Classification	Proportion (%)	Item	Classification	Proportion (%)
Sex	Male	43	Marriage	Married	59
	Female	57		Single	27
Age	0–20	3		Divorced	2
	21–40	29		Widow	12
	41–80	62	Hukou Location	Local	77
	Over 81	6		Migrant	23
Education	Primary school	37	Hu Kou	Town	76
	Junior high school	22		Countryside	24
	High school	9	Housing	Self-owned	86
	Undergraduate	26		Rental	14
	Postgraduate	6

). We compared the demographic characteristics (sex, age, educational background) of our sample with those reported in the city’s latest statistical yearbook and found high consistency between the two datasets. This indicates that our sample is representative of the citywide population.

2.2. Variables

The questionnaires were divided into two parts: community resilience and social capital (

Table 3. The indicators of social capital and community resilience to flood.

Variables	Criterion	Indicator
Dependent variable: Community resilience	Social Resilience	Resident participation
		Shelter
		Purchase insurance
		Rainstorm information
	Economic resilience	Reduced expenses
	Organizational resilience	Neighborhood help
		Committee help
		Government help
Independent variable: Social capital	Cohesion	Close interaction
		Resident cooperation
		Resident participation
		Number of greetings
		Neighborhood communication
		Mutual assistance
	Collective Efficiency	Property concern
		Committees’ concern
		Government concern
		Trust of committees
	Sense of Belonging	Residence intention
		Livability Comfort
		Community members
	Trust and Reciprocity	Problem-solving cooperation
		Residents’ trust
		Residents’ mutual assistance
		Trust level
	Informal Social Control	Protection of public health
		Acceptance of unethical behavior
		Stopping unethical behavior

). Social capital is a collection of actual or potential resources with a structural nature, which can produce corresponding social benefits whilst reducing action costs. The independent variable comprised 22 indicators of social capital, representing the five dimensions of collective efficacy, sense of belonging, trust and reciprocity, cohesion and informal social control; meanwhile, the dependent variable comprised 8 indicators of community resilience, representing the three dimensions of economic resilience, social resilience and organizational resilience.

Cohesion refers to the closeness and degree of relationship between community residents. Collective efficacy refers to the flood fighting efficiency of residents, property management and government when flood disaster comes. Sense of belonging refers to the degree of community identity of residents. Trust and reciprocity refer to mutual trust and mutual assistance between residents. Informal social control refers to the degree to which residents consciously abide by the code of conduct. Likert scales were used to assess the community’s evaluation and attitude towards social capital in 21 aspects.

In this study, social capital and community resilience are comprehensive concepts with multiple dimensions and indexes. To assess the impact of social capital on community resilience, factor analysis and multiple regression methods were employed to analyze the influence of social capital factors on community resilience.

2.3. Models

(1)

Factor analysis

Principal component analysis (PCA) is a well-known method for dimension reduction, and plays a crucial role in presenting data in a more concentrated manner. Factor analysis utilizes the PCA method for factor extraction, aiming to simplify the factor structure of a set of original variables. Generally, the determination of factors through linear combination is expressed as

$$F_{ik}={\displaystyle \sum _j{W}_{jk}{X}_{ij}}$$

(1)

F_ik = score of community i social capital on factor k;

X_ij = value of original variable j for community i social capital, which is standardized in PCA;

W_jk = factor loading of variable j on factor k representing the proportion of variance of variable j explained by factor k.

The comprehensive score of social capital or community resilience in community i is then calculated using a weighted sum method.

$$S_i={\displaystyle \sum _k\sqrt{{\lambda }_k}}·F_{ik}/{\displaystyle \sum _k\sqrt{{\lambda }_k}}$$

(2)

S_i = comprehensive factor scores of communities i representing the social capital or community resilience of the community;

λ_k = the eigenvalue of factor k.

The following 21 indicators are categorized into cohesion, collective efficacy, sense of belonging, trust and reciprocity, and informal social control. It is important to note that these five aspects are crucial for enhancing our understanding of social capital in relation to urban flooding, and they are not entirely distinct from each other. Factor analysis is first conducted on all variables and then on each set of the five in order to assess social capital as a whole. The following 8 indicators are categorized into organizational resilience, social resilience and economic resilience. Factor analysis is first conducted on all variables to evaluate community resilience as a whole in three sets.

(2)

Multiple regression

The explanatory variable in this study is social capital, which has an influence on community resilience. Therefore, the multiple regression model is selected for analysis. Community resilience is used as the dependent variable and social capital as the independent variable to identify the impact of socio-economic attributes, risk perception and social capital. The dependent variable follows a continuous and approximately normal distribution; therefore, the multiple regression model is utilized for regression analysis. The model can be expressed as

$$Y={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+\cdots +{\beta }_p{x}_p+\epsilon$$

(3)

$Y$ is the dependent variable (community resilience index); ${\beta }_0$ is constant; ${\beta }_1,{\beta }_2\dots {\beta }_p$ represent the regression coefficients; $x_1$,$x_{2 \dots }{x}_p$ represent independent variables (social capital factor), and $\epsilon$ is the random error.

3. Results and Discussion

3.1. Social Capital of Different Communities

To differentiate the social capital of various types of communities, the factor analysis method was employed to categorize 21 indicators of social capital. These indicators were primarily derived from previous literature on social capital and urban flooding [5,32]. The test results revealed a KMO statistic of 0.840. Additionally, Bartlett’s sphericity test (p = 0.000 < 0.05) was conducted, indicating that the correlation coefficient matrix significantly differed from the identity matrix, thus demonstrating the necessity and suitability of factor analysis for this study. The result of the factor analysis identified five principal components, which accounted for a cumulative variance contribution rate of 59.77%, encompassing most of the original variables (

Table 4. Factor analysis of urban community social capital.

Factor	Evaluation Index System	Load Factor	Eigenvalue	Variance Contribution Rate (%)	Cumulative Variance Contribution Rate (%)
F1 Cohesion	Close interaction	0.698	3.313	15.060	16.060
	Resident cooperation	0.722
	Resident participation	0.700
	Number of greetings	0.461
	Neighborhood communication	0.723
	Mutual assistance	0.699
F2 Collective efficacy	Property concern	0.846	2.976	13.060	28.589
	Committees’ concern	0.866
	Government concern	0.663
	Trust of committees	0.761
F3 Sense of Belonging	Residence intention	0.583	2.647	12.030	40.619
	Livability	0.842
	Comfort	0.843
	Community members	0.575
F4 Trust and Reciprocity	Problem-solving cooperation	0.578	2.646	12.025	52.769
	Residents’ trust	0.646
	Residents’ mutual assistance	0.805
	Trust level	0.623
F5 Informal Social Control	Protection of public health	0.686	1.567	7.124	59.769
	Acceptance of unethical behavior	0.696
	Stopping unethical behavior	0.492

). Consequently, the 21 indicators were categorized into five groups. The first principal component (F1), named the cohesion factor, included close interaction, resident cooperation, resident participation, number of greetings, neighborhood communication and mutual assistance. The second principal component (F2), named the collective efficacy factor, comprised property concern, committee concern, government concern and trust in committees. The third principal component (F3), named the sense of belonging factor, encompassed residence intention, livability, comfortable living conditions and community members’ satisfaction. The fourth principal component (F4), named trust and reciprocity, included problem-solving cooperation, residents’ trust, residents’ mutual assistance and level of trust within the community. Finally, the fifth principal component (F5), named informal social control, encompassed protecting public health as well as the acceptance or prevention of unethical behavior (Table 4).

The comprehensive factor score of each community can be calculated using the weighted sum method from Equation (3). This score reflects the level of social capital in the various indicators of the community, and a score below zero indicates that it is below average.

Z = 0.1506F₁ + 0.1306F₂ + 0.1203F₃ + 0.12025F₄ + 0.7124F₅

(4)

The highest scores for social capital belong to commercial housing communities, with great internal differences, including the following: the highest scores belong to Mingchenshijia and Kangsheng Garden, and the middle scores belong to the Xishan apartments, whereas the lowest scores belong to Blue Valley. The lowest scores belong to reform housing communities, including Banqiao Xincun and Shuiwen Garden, whereas the highest score of social capital belongs to the affordable housing community of Golden Leaf Garden. Despite the limited number of communities surveyed in this study, the trend of social capital level is obvious (Figure 3).

3.2. Different Dimensions of Social Capital

Figure 3 illustrates the social capital across different dimensions within the relevant communities. In terms of cohesion, the reform housing communities exhibit varying levels of cohesion, with Banqiao Xincun scoring the highest and Shuiwen Garden scoring the lowest, indicating significant internal differences. The middle scores belong to affordable housing, including Golden Leaf Garden. The lowest scores belong to commercial housing, with great differences, including the highest for the Xishan apartments and the lowest for Kangsheng Garden and Blue Valley. Comparing reform housing to commercial housing, the relationships of neighborhoods in reform housing are relatively closer than those in commercial housing; therefore, reform housing has a high level of cohesion.

In the dimension of collective efficacy, the highest score of collective efficacy belongs to commercial housing communities, with great internal differences, including the highest for Mingchengshijia and the Xishan apartments, and the lowest for Blue Valley. According to the survey, Mingchengshijia belongs to a commercial housing community, and the property corporation does a good job in flood emergencies after disasters, which is reflected in the timely high cooperation between residents and the property corporation to alleviate flood situations. Additionally, the Xishan apartment community is located in a flood-prone area with low terrain and close proximity to the river. Therefore, the government pays much attention to this community and prepares sufficient flood control materials to alleviate flood situations. However, Blue Valley is blacklisted in Yuhuatai District due to refusal to pay property fees and is in severe conflict with the property corporation. The property corporation failed to implement their responsibilities, resulting in a low score. The lowest scores belong to reform housing communities and affordable housing communities.

In the dimension of sense of belonging, the highest score of sense of belonging is for commercial housing, with great internal differences, including the highest for Mingchengshijia and the lowest for the Xishan apartments. The community with a mid-level score is Kangsheng Garden. The sense-of-belonging scores are lower for reform housing and affordable housing, with significant internal differences in reform housing. This includes the highest score for Banqiao Xincun and the lowest score for Shuiwen Garden.

According to the survey, Mingchengshijia is a newly built commercial housing community which has a good environment, a high green rate and sufficient support facilities. However, the Xishan apartment community is an old community with a long history of construction, a low green rate and insufficient community support facilities. Residents of the Xishan apartment community have a low education level but high rental rate. Most of them completed primary school and junior high school. Therefore, the sense of belonging is relatively weak.

In the dimension of trust and reciprocity, the highest score of trust and reciprocity is for commercial housing, with great internal differences, including the highest for Blue Valley and the lowest for Mingchengshijia. The community with a mid-level score is Kangshen Garden. Despite some internal distinctions, reform housing has the same level of reciprocity and community trust as commercial housing. The highest (Shuiwenyuan) and the lowest (Banqiao Xincun) scores belong to reform housing communities. Golden Leaf Garden belongs to the affordable housing community and has a low level of trust and reciprocity. According to the survey, flooding is a constant problem in Blue Valley communities, with the highest scores in the dimension of trust and reciprocity. Residents in this community report floods to the government, which increases the trust among residents in the process of jointly solving flood problems. This scenario is true despite the fact that cooperation between residents and property in the community is low, and solving the community flood problem is difficult. However, the fact that the residents are not intimately related is the key reason explaining the low comprehensive score of trust and reciprocity in Mingchengshijia.

In the dimension of informal social control, the score of informal social control in commercial housing is higher than in affordable housing and reform housing. The highest belongs to Mingchengshijia, and the lowest scores belong to the Xishan apartments, Blue Valley and Kangsheng Garden. The informal social control of reform housing shows polarization, including the highest for Shuiwenyuan and the lowest for Banqiao Xincun. The informal social control of affordable housing of Golden Leaf Garden is relatively low. According to the survey, Mingchengshijia has the highest score of informal social control because the residents in this community primarily have bachelor’s degrees or above, and their income levels are relatively high. Both the Xishan apartments and Banqiao Xincun have the lowest scores for informal social control because the residents in this community have lower education and a higher proportion of tenants (Figure 4).

3.3. Community Resilience Measurement

To assess the comprehensive score of community resilience, the factor analysis method was employed to identify eight indicators of community resilience. These indicators were primarily drawn from existing literature on community resilience and floods. The test results revealed a KMO statistic of 0.642. Additionally, a Bartlett sphericity test (p = 0.000 < 0.05) was conducted, indicating that the correlation coefficient matrix significantly differs from an identity matrix and that a factor analysis is necessary and appropriate for this analysis. The result of the factor analysis identified three principal components, which accounted for a cumulative variance contribution rate of 55.887%, encompassing most of the original variables. The first principal component (F1), termed organization resilience, includes neighborhood help, committee help and government help. The second principal component (F2), termed social resilience, comprises resident participation, shelter availability, insurance purchase behavior and rainstorm information access. Lastly, the third principal component (F3), named economic resilience, encompasses expense reduction strategies (

Table 5. Factor analysis of community resilience.

Factor	Evaluation Index System	Load Factor	Eigenvalue	Variance Contribution Rate (%)	Cumulative Variance Contribution Rate (%)
F1 Organization resilience	Neighborhood help	0.353	1.770	2.122	22.122
	Committee help	0.871
	Government help	0.836
F2 Social resilience	Resident participation	0.655	1.566	19.576	41.698
	Shelter	0.611
	Purchase insurance	0.390
	Rainstorm information	0.738
F3 Economic resilience	Reduced expenses	0.899	1.135	14.189	55.887

).

The comprehensive factor score of each community can be calculated using the weighted sum method. These scores reflect the level of community resilience across various indicators, and a score below zero indicates that the community is performing below average.

Z = 0.02122F₁ + 0.19576F₂ + 0.14189F₃

(5)

The highest scores belong to Mingchenshijia, Kangshen Garden and the Xishan apartments, whereas the lowest scores belong to Golden Leaf Garden, Banqiao Xincun, Shuiwenyuan and Blue Valley.

3.4. Impact of Social Capital on Community Resilience

A stepwise regression model was utilized for calculation using SPSS 24.0. The individual attributes, risk perception and social capital were each examined in relation to community resilience, resulting in the derivation of multiple linear regression equations.

Y₁ = −0.002 X₁ − 0.150 X₄ − 0.165 X₆ + 0.449

(6)

Y₂ = 0.051 X₇ + 0.147 X₈ + 0.033 X₉ + 0.054 X₁₁ + 0.000

(7)

After incorporating all three sets of variables into the model, the results indicate that in the regression equation, R² = 0.371 and F = 12.47 (Sig = 0.000). This suggests that the regression equation demonstrates a high degree of fitting and effectiveness (

Table 6. Regression analysis of the impact of social capital on community resilience.

		ModelI	Model II	Model III
Individual	Age X1	−0.002 * (−1.69)		−0.003 ** (−2.44)
Attributes	Gender X2	0.039 (1.00)		−0.006 (−0.17)
	Education X3	0.013 (0.896)		0.009 (0.74)
Risk Perception	Increased savings X4	−0.150 *** (−2.78)		−0.111 ** (−2.35)
	Property transfer X5	0.027 (0.50)		0.019 (0.41)
	Community preparation X6	−0.165 *** (−2.57)		−0.01 (−0.19)
Social Capital	Cohesion X7		0.051 *** (3.35)	0.070 *** (4.08)
	Collective efficacy X8		0.147 *** (9.65)	0.124 *** (6.93)
	Sense of belonging X9		0.033 ** (2.17)	0.039 ** (2.21)
	Trust and reciprocity X10		0.020(1.33)	0.023 (1.27)
	Informal social control X11		0.054 *** (3.56)	0.046 *** (2.72)
	Constant	0.449 *** (3.34)	0.000 (0.00)	0.271 ** (2.24)
	Adjusted R2	0.152	0.306	0.371
	F value	7.10 ***	24.70 ***	12.47 ***

Note: t values are in brackets; *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

). Based on the results of the regression analysis, we can derive the following multiple regression equation.

Y = −0.003 X₁ − 0.111 X₄ + 0.070 X₇ + 0.124 X₈ + 0.039 X₉ + 0.046 X₁₁ + 0.271

(8)

In the context of community resilience, the regression model comprises a total of six factors. The regression coefficients for these influencing factors do not exceed 0.5, indicating relatively low significance levels and highlighting the complexity of their impact on community resilience. The explanatory powers of individual attributes, risk perception and social capital for community resilience are 0.152, 0.306 and 0.371, respectively. Therefore, social capital has the greatest impact on community resilience, followed by individual attributes and risk perception.

The explanatory powers of individual attributes, risk perception and social capital on community resilience (R²) are 0.152, 0.306 and 0.371, respectively. Although the R² value is not high, it is evident that the dependent variable Y has been influenced by a lot of complicated factors. Furthermore, the core variable remains significant in the context. As can be seen, social capital has a higher impact on community resilience than individual attributes and risk perception. Considering the combined impact of both groups of variables, the explanatory power of Model 3 reaches 0.371. The impact of individual attributes and risk perception on community resilience primarily focuses on age, savings transfer and community preparation, which are closely related to community resilience. The existing literature indicates a connection between age and the capacity to cope with risks. As individuals age, their ability to manage risks tends to decrease, potentially impacting community resilience. A correlation exists between risk awareness and community resilience. Risk awareness includes increasing savings and community preparation. Increased savings indicate that respondents possess a strong awareness of disaster risk prevention. Community preparation involves facility construction and lecture propaganda, among others. By enhancing residents’ risk perception, overall community resilience is strengthened.

The impact of social capital on community resilience primarily focuses on cohesion, collective efficacy, sense of belonging and informal social control. Cohesion includes the closeness of community residents’ communication and cooperation, significantly influencing community resilience. Collective efficacy involves the government, neighborhood committees and property management guiding the community to fight against floods. Sense of belonging includes the residents’ sense of identity to the community and livability. Informal social control includes community collective quality, public health and acceptance of immoral behavior.

When both sets of variables are simultaneously incorporated into the model, an enhancement in model correlation indicates that individual characteristics, risk perception and social capital all have an impact on community resilience. Age, increased savings, cohesion, collective efficacy, sense of belonging and informal social control significantly influence community resilience.

4. Conclusions and Policy Implications

4.1. Social Capital at the Community Level

These findings reveal that social capital plays a crucial role in enhancing community resilience during flood disasters. Social capital varies among reformed housing, commercial housing and affordable housing within different communities. Our findings indicate that cohesion and collective efficacy are the most significant factors contributing to social capital in Nanjing. There is evidence suggesting that communities with strong collective efficacy are better equipped to learn from past flood experiences and effectively utilize the social capital present within the community, including individuals, groups and organizations [33,34,35]. Furthermore, the most crucial criteria for enhancing community resilience in developing countries include effective neighborhood communication, resident cooperation, mutual assistance and trust in local committees.

4.2. Influencing Factors of Community Resilience

Communities with moderate to high levels of social capital are more likely to exhibit resilience in the face of flooding events. This suggests that social capital plays a crucial role in enhancing a community’s ability to recover from flood-related challenges. For instance, residents living in reformed housing generally exhibit higher cohesion scores compared to those residing in commercial or affordable housing; however, internal variations persist within each type of community. The majority of reformed housing residents consist of employees from the same workplace, resulting in close-knit neighborly relationships. In times of flood disasters, mutual assistance among neighbors living in reformed housing is more prevalent than within other types of communities. Previous studies on flood risk assessment have traditionally focused primarily on the physical factors of flooding while neglecting the socio-economic factors, such as income, age and race, as important aspects affecting the impact of urban flood disasters. Therefore, it is essential to integrate the social impact of flood disasters into the risk assessment process [36,37,38,39]. The relationships among community stakeholders and community participation are crucial social assets for coping with community disasters. Specifically, the transformation of social capital is evident in the response, recovery and reconstruction processes which directly influence the development and consolidation of capacities for community adaptation [40,41,42].

4.3. Policy Implications

Our findings have several policy implications for enhancing community resilience. Firstly, it is crucial to emphasize the importance of community committees and property services in building resilience against flood disasters. Both stakeholders require a strong ability for informal social control and collective efficacy [27]. Our interviews with local residents revealed that property services play a vital role in cleaning garbage from drains and preparing emergency flood control resources. They also disseminate warning information to ensure that all community residents are aware of potential waterlogging. A well-functioning property service has the potential to enhance trust and social cohesion among residents through activities such as focus groups and social events. Additionally, some community committees have implemented various volunteering projects, including activities for children and the establishment of community gardens, with the aim of fostering trust among residents.

Secondly, different types of communities demonstrate varying levels of social capital. Reformed housing communities tend to exhibit high levels of cohesion, especially those that are densely populated with predominantly elderly residents in urban and peri-urban areas [43,44]. In order to improve disaster preparedness, the community committee should gather basic information about residents and establish a database to identify vulnerable groups. Regular meetings for disaster prevention should be held in public squares and playgrounds. The community committee or property service should also provide sandbags and shovels to residents, especially those living on the ground floor who are more vulnerable to floods [45,46]. By contrast, commercial housing communities generally demonstrate a high level of collective efficacy, and a sense of belonging, trust and reciprocity. Strengthening these aspects of social capital in commercial housing communities can help reduce social differentiation whilst enhancing cooperation between residents and property management. This finding contributes to increased collective efficacy through improved community governance. The level of social capital in affordable housing communities is lower than in the other two types of communities. Additional efforts are required to boost collective efficacy and mutual trust in these communities.

Thirdly, the Chinese government should improve the dissemination of hazard information to the public and implement measures to mitigate the impact of flooding. Utilizing flood risk maps can play a crucial role in increasing public awareness and promoting safety education. The existing government-led disaster management model in China necessitates structural reforms to transition from a top-down approach to a bottom-up model that encourages social connections and community involvement. Establishing emergency response practices at the community level is imperative, ensuring that local areas have sufficient rescuers, resources and technical support for effective risk reduction.