Introduction

Extensive literature on disasters and mental health has shown that disasters can cause serious psychological health problems in survivors, such as anxiety, depression, and post-traumatic stress disorder (PTSD), and that the effects may persist years after the disaster (Kokai et al., 2004; Newnham et al., 2022; North et al., 2021; Seto et al., 2019). The May 12, 2008 Wenchuan earthquake in China caused 7,789,100 houses to collapse and killed 69,227 people (National Bureau of Statistics, 2009). This disaster left survivors not only grappling with property loss as well as the injury and death of relatives and friends, but also caused them great psychological trauma and varying degrees of depression (Chen et al., 2020; Fu et al., 2021; Tanaka et al., 2016; Wang et al., 2009; Xu & Song, 2011; Zhang et al., 2012). Survivors in rural areas are especially susceptible to more serious psychological trauma due to the lack of resources for psychological recovery. However, the long-term mechanism underlying recovery from depression among rural disaster survivors remains unclear. Therefore, this study used longitudinal survey data collected after the Wenchuan earthquake to elucidate the long-term changes in post-disaster depression recovery among rural survivors and its influencing mechanisms.

Disasters and post-disaster depression

Studies on the consequences of disasters have shown that many victims may experience clinically significant symptoms due to exposure to stressful events, loss, social disintegration, disruption of social networks, and destruction of community structures or traditional support mechanisms (Kokai et al., 2004; Seto et al., 2019). The most commonly found specific psychological problems include anxiety, distress, grief, depression, and PTSD (Hackbarth et al., 2012; McFarlane et al., 2012; Norris & Murrell, 1988; North, 2016; Wang et al., 2011), followed by non-specific psychological distress, as well as various health problems and worries (Madakasira & O’Brien, 1987; Norris et al., 2002; Young et al., 1999). In addition, scholars have studied the longitudinal changes in the mental health of survivors after a disaster and found that mental health tends to deteriorate sharply after a disaster but gradually recovers over time (Calvo et al., 2015; Ni et al., 2015; Priebe et al., 2011; Sugano, 2016; Wang et al., 2000).

Social-systemic mechanism of post-disaster depression

The mechanism of post-disaster recovery is discussed in the framework of the ecosystem model. Proposed by Bronfenbrenner (1979), the ecosystem model argues that individuals are significantly affected by interactions among many nested systems. These systems constitute the social context that determines the effect of life events on individuals and the individuals’ response to these events (Hoffman & Kruczek, 2011). As a life event, ecological models can facilitate understanding of the impact mechanism of disaster on psychological trauma. For example, some scholars believe that ecosystem models more accurately provide a conceptual framework that can explain the complexity of trauma and facilitate the development of a comprehensive and systematic approach to trauma coping at the individual, family, community, and societal levels (Hoffman & Kruczek, 2011). Therefore, this study adopted the ecosystem model as the framework to analyze the impact mechanism of post-disaster depression recovery at the individual, community, and social levels.

At the individual level, studies on victims’ depression trajectories after natural disasters usually explain the impact mechanism in the context of social capital theory. Social capital is usually defined on the basis of networks. Scholars believe that social capital is “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationship[s] of mutual acquaintance or recognition” (Bourdieu, 1986: 249). In short, social capital is defined as social networks and the resources they contain (Lin, 2001). It is precisely because the resources embedded in social networks bring specific opportunities and advantages to individuals, that individuals can recover from the state of vulnerability more quickly.

Thus far, the empirical evidence is inconclusive on the associations between individual social networks and depression. Research has shown that a social network can have both positive and negative effects on the depression of people receiving psychological services and other interventions (Almedom, 2005; Noel et al., 2018). One view is that there is a close correlation between an individual’s mental health and changes in social network conditions. For example, studies have found that the loss of social capital and distortion of relationship networks negatively affect psychological states (Kaniasty & Norris, 1995; Litwin & Levinsky, 2022; Zhao, 2011). A second view is that social networks and the social resources within them can reduce depression. Social capital can prevent symptoms such as anxiety and depression after disasters by reducing stress (Kawachi & Berkman, 2000; Sasaki et al., 2020; Wind et al., 2011). In other words, social networks can have a mitigating effect in that they help survivors recover from depression by providing them with psychological support. A third view is that social network capital may be harmful to individuals’ recovery from depression. For example, a study examining the link between social capital and mental illness revealed a potential negative correlation between individual cognitive social capital (ICSC) levels and mental disorders (De Silva et al., 2005). Wong et al. (2019) further found that structural social capital negatively affected the depression of Ya’an earthquake survivors and failed to relieve their psychological trauma. In their study based on data from four countries (Peru, Ethiopia, Vietnam, and India), De Silva et al. (2007) found that although ICSC was associated with the reduction of common mental disorders in pregnant women, the impact of structural social capital was mixed and context-specific.

At the community level, communities are an important source of support during disasters, providing a collective environment for common recognition and solidarity (Bowe et al., 2022). The beneficial association between community social capital and community belonging, community trust, and depression recovery was supported by some empirical studies (Almedom, 2005; Berry & Welsh, 2010). For example, some scholars found that trust and interaction in collective social capital had a significant effect on disaster victims’ post-traumatic stress (Xu & Hong, 2015), and those in communities with higher social capital suffered less post-traumatic stress (Wind & Komproe, 2012). A broad sense of community belonging is also crucial in post disaster recovery, helping to provide effective and resonant disaster response measures to people of different cultures and languages (Marlowe, 2015). Other studies found that coordinated communities can reduce depression and anxiety by building a sense of community identity and solidarity during the COVID-19 pandemic (Bowe et al., 2022).

At the social level, this study focuses on the allocation of social support resources in disaster relief. After a disaster, many relief resources such as materials and money from all sectors of society are poured into the disaster area, which helps alleviate the survivors’ material losses and supports their psychological reconstruction and recovery. The social support from disaster relief resources is important in post-disaster reconstruction, as it can alleviate the psychological pressure caused by the disaster. The impact of the allocation of these social relief resources on the survivors’ recovery from depression has also received special attention. For example, scholars have found a significant relationship between social policies related to income distribution and social security, and subjective well-being (Sun & Xiao, 2012). After the Wenchuan earthquake, the government was responsible for allocating disaster relief resources, and the earthquake victims’ perceived fairness of and satisfaction with the government’s allocation were found to be significantly associated with their health and welfare (Bai et al., 2009; Hong & Zhao, 2019), which may have also been related to their mental health. Therefore, this study assumes that the social support resources in post-earthquake disaster relief will have an important effect on the depression of survivors.

Research hypothesis

In sum, extensive literature has examined changes in survivors’ mental states after disasters from psychological and sociological perspectives. However, studies on the depression of survivors are mainly based on snapshot data, and scant attention has been paid to the changing trajectory of disaster victims’ depression over time. Therefore, this study aims to use three-phase longitudinal survey data collected after the Wenchuan earthquake to analyze the changes in the trajectory of survivors’ depression as well as the effect of social capital and social support for disaster relief. Specifically, this study answers the following questions. First, at the individual level, how do individual community networks affect the trajectory of depression recovery of rural survivors of the Wenchuan earthquake? Second, at the community level, how does community social capital affect the change in the post-disaster depression of rural survivors? Finally, at the social level, how does resource relief and allocation from all aspects of society affect the change trajectory of the depression of rural survivors? Accordingly, we established the following research hypothesis.

  • Hypothesis 1: At the personal level, individual social networks are negatively related to the depression of rural survivors of the Wenchuan earthquake. That is, the depression trajectory of rural survivors after the earthquake will change with their social network levels over time.

  • Hypothesis 2: At the community level, community social capital (community trust and community belonging) is negatively associated with the depression of rural survivors of the Wenchuan earthquake. The depression trajectory of rural survivors will vary with their community social capital.

  • Hypothesis 3: At the social level, the disaster relief resources and allocation provided by society are negatively related to the depression of rural survivors. The depression trajectory of rural survivors will vary depending on the amount of social relief resources they receive.

Methods

Data

The data employed in our research are derived from the “Reconstruction Research Project after Wenchuan Earthquake” survey conducted by the “Xinyi Community Construction Research Center” of Tsinghua University. The survey lasted for four years, with data intervals of about a year and a half to allow the different stages of reconstruction to be investigated and tracked. This survey has passed the ethical guidelines review (IRB no. 041913016), and all participants provided informed consent. Employing a structured questionnaire designed by the research team, the study mainly collected information on respondents’ demographic characteristics, disaster damage, availability of relief resources, social network, and health. The interviewees comprised doctoral and master’s students majoring in sociology from Tsinghua University, who had received professional training before the interviews. Most participants completed the questionnaires in the presence of researchers or interviewees.

The survey was divided into three waves and conducted in Deyang City and Mianyang City of Sichuan Province. The baseline survey data were collected in 2009. First, 12 villages in Deyang and Mianyang were randomly selected. Second, taking the list of households in the village as a sampling frame, 33 households were randomly selected from each village. Then, one adult per household was randomly selected as the interviewee, using the Kish grid. In total, 558 questionnaires were distributed in 12 villages, and 466 valid questionnaires were collected, which amounted to an effective response rate of 83.5%. The follow-up survey was conducted in November 2010, with 313 respondents tracked across all 12 villages. In April 2012, the 12 villages were revisited and 225 questionnaires were collected. The final sample obtained from the longitudinal survey comprised 221 respondents.Footnote 1 The sample characteristics of the baseline survey are shown in Table 1.

Table 1 Sample statistics of the baseline survey (n = 221)
Full size table

Variables

Dependent variable

The dependent variable of this study was the depression of survivors, and it was measured using the Chinese Health Questionnaire-12 scale. This scale is a Chinese version of the health questionnaire developed by Cheng et al. (1990) based on the General Health Scale. It consists of 12 items that ask how often the respondent experiences a certain symptom, specifically, “Have you recently been suffering from headache or pressure in your head?”; “Have you recently had palpitations and been worried about heart trouble?”; “Have you recently felt discomfort or pressure in your chest?”; “Have you recently suffered from shaking or numbness of your limbs?”; “Have you recently lost much sleep over worry?”; “Have you recently been taking things hard?”; “Have you recently been getting along well with your family and close relatives?”; “Have you recently been losing confidence in yourself?”; “Have you recently been feeling nervous and high-strung?”; “Have you recently been feeling hopeful about your future?”; “Have you recently been worrying about your family or close relatives?”; and “Have you recently been feeling that life is entirely hopeless?” For each item, respondents were to answer on a four-point scale, wherein 1 = never, 2 = occasionally, 3 = sometimes, 4 = often. The Cronbach α coefficient of this scale was 0.681. After reversing the responses to the positive questions, the sum of the 12 answers was considered the respondent’s depression score. The minimum value of this variable was 11, and the maximum value was 48. The higher the score, the higher the level of depression.

Independent variables

Individual social network

This study adopted the network-based definition of individual social capital and measured it from the perspective of social networks. Bian (2004) believed that in the Chinese context, “Chinese New-Year Greeters’ Networks” could reflect the capacity and extent of the personal relationship network and serve as a valid method for measuring individual social networks. This measurement method has been adopted by many scholars, such as Wei and Han (2018), Zhao (2011), and Hong and Zhao (2019). Therefore, this study followed Bian’s “Chinese New-Year Greeters’ Networks” method to measure individual social capital. Specifically, individual social network was operationalized as the total number of relatives, good friends, and other people who visited the respondents’ homes in various ways during the Chinese New Year. This study measured the social network using the following three items: “the number of relatives who greeted your family during the Chinese New Year,” “the number of good friends who visited your family during the Chinese New Year,” and “the number of other guests who visited your family during the Chinese New Year.” This study obtained the size of the social network by aggregating the responses to all three questions. The Cronbach α coefficient of the scale was 0.698.

Community social trust

Community social capital includes community trust and community belonging. Community trust is measured via the following three items: “the degree of trust of respondents in their families,” “the degree of trust that respondents have in the people who live around them,” and “the respondent’s trust in village cadres/community cadres.” The answers to these three questions were one of the following: no trust at all (= 1), less trust (= 2), more trust (= 3), and complete trust (= 4). The sense of community belonging was measured via the following items: “whether the respondent plans to continue living in the current village in the next few years,” “whether the respondent feels that he/she will live in the current village for a long time,” “whether the respondent feels that the current village is suitable for him/her to live in,” “whether the respondent feels that living in the current village is important for him/her,” “whether the respondent feels that living in the current village is very comfortable and that it evokes a sense of home,” “whether the respondent often realizes that he/she is a resident of the village,” “whether the respondent often feels that the identity of being one of the villagers is important for him/her,” and “whether the respondent feels that he/she is a member of the village.” The answers to these questions took the form almost never (= 1), sometimes (= 2), often (= 3), and almost always (= 4). The Cronbach α coefficient of this variable was 0.919.

Post-disaster relief resource allocation

This study reflected the relief resource allocation after the earthquake through three variables. The first was the amount of relief money received by survivors, which refers to the sum of consolation funds, temporary living assistance funds, transitional housing subsidies, housing construction subsidies and other subsidies/compensations/donations received by respondents from governments, work units, NGOs, and other social sectors after the earthquake. This variable is a continuous variable with a mean value of 55,177.32 yuan. This study used the log form of this variable, after which the average value derived was 10.659 yuan. The second variable was the survivors’ perception of fairness in disaster relief after the earthquake, which refers to the survivors’ perception of how fair the allocation of relief materials in their areas was. This variable had four options: very fair (= 1), mostly fair (= 2), not very fair (= 3), and very unfair (= 4). The third variable was the satisfaction of survivors with post-disaster relief. This variable comprised the following items: the survivors’ satisfaction with the performances of the central government, provincial government, and city/county government in disaster relief, separately, as well as their satisfaction with the performances of the township/town/subdistrict government and village/neighborhood committee and other organizations in disaster relief and reconstruction, separately. Each question includes four options: very satisfied (= 4), relatively satisfied (= 3), not very satisfied (= 2), very dissatisfied (= 1). The Cronbach α coefficient of this variable was 0.761.

Covariates

The Covariates in our study were gender and education level. Gender was a dummy variable (0 = male, 1 = female). Educational level included two categories: elementary school and below (= 1) and middle school or above (= 2). In addition, because of their biased distribution, marital status, hukou type, and political status variables were not included in the HLM model.

Statistical analysis

The Hierarchical Linear Model (HLM) was used in our study. HLM can simultaneously test the effect of individual-level and community/social-level explanatory variables on the outcome variable and can effectively eliminate the deficiencies of variance homogeneity while enhancing the accuracy of model parameter estimation (Potthoff & Roy, 1964). In this study, the intra-individual differences between different survey time points were regarded as the lower level of the data structure, while the inter-individual differences were regarded as the upper level of the data structure. We established the following four models.

Model 1 is an unconditional mean model, including only the repeated measurement of \({\mathrm{Y}}_{\mathrm{ij}}\). The model equations are:

$$\mathrm{Level}-1\mathrm{ \;model}: {\mathrm{Y}}_{\mathrm{ij}}={\upbeta }_{0\mathrm{i}}+{\upvarepsilon }_{\mathrm{ij}}, {\varepsilon }_{ij}\sim N(0,{\sigma }_{\varepsilon }^{2})$$
$$\mathrm{Level}-2\mathrm{ \;model}: {\upbeta }_{0\mathrm{i}}={\upgamma }_{00}+{\upzeta }_{0\mathrm{i}}, {\zeta }_{0i}\sim N(0,{\sigma }_{0}^{2})$$

Model 2 is an unconditional growth model, only incorporating the repeated measurement of \({Y}_{ij}\) with the time-related parameter. Model 3 further includes the covariates of gender and education level. Model 4 is the full model to examine the effect of individual social capital, community social capital and social support resources on the changing trajectory of survivors’ depression after the earthquake.

The equations for the full models are:

$$\mathrm{Level}-1\mathrm{ \;model}:{\mathrm{Y}}_{\mathrm{ij}}={\upbeta }_{0\mathrm{i}}+{\upbeta }_{1\mathrm{i}}\left({\mathrm{Survey\;Time}}_{\mathrm{ij}}\right)+{\upvarepsilon }_{\mathrm{ij}}$$
$$\mathrm{Level}-2\mathrm{ \;model}: {\upbeta }_{0\mathrm{i}}={\upgamma }_{00}+{\upgamma }_{01}\left({\mathrm{Gender}}_{\mathrm{i}}\right)+{\upgamma }_{02}\left({\mathrm{Education}}_{\mathrm{i}}\right)+{\upgamma }_{03}\left({\mathrm{Social\;Network}}_{\mathrm{i}}\right)+{\upgamma }_{04}\left({\mathrm{Community\;Trust}}_{\mathrm{i}}\right)+{\upgamma }_{05}\left({\mathrm{Community\;Belonging}}_{\mathrm{i}}\right)+{\upgamma }_{06}\left(\mathrm{Resource}{\mathrm{ \;Relief}}_{\mathrm{i}}\right)+{\upzeta }_{0\mathrm{i}}$$
$${\upbeta }_{1\mathrm{i}}={\upgamma }_{10}+{\upgamma }_{11}\left({\mathrm{Gender}}_{\mathrm{i}}\right)+{\upgamma }_{12}\left({\mathrm{Education}}_{\mathrm{i}}\right)+{\upgamma }_{13}\left({\mathrm{Social\;Network}}_{\mathrm{i}}\right)+{\upgamma }_{14}\left({\mathrm{Community\;Trust}}_{\mathrm{i}}\right)+{\upgamma }_{15}\left({\mathrm{Community\;Belonging}}_{\mathrm{i}}\right)+{\upgamma }_{16}\left({\mathrm{Resource\;Relief }}_{\mathrm{i}}\right)+{\upzeta }_{1\mathrm{i}}$$

In level 1 of the full model, the survey time points (Survey Time) were employed to explain the change in individuals’ depression at different time points after the earthquake. In level 2, this study included independent variables and covariates to explain the initial state (at the time the earthquake occurred) and the change trajectory of the survivors’ depression. The parameter estimation uses the Restricted Maximum Likelihood Estimation. All data processing and analysis was performed using STATA 15.0 software.

Results

Results of descriptive statistical analysis

Table 2 shows the change trajectory of survivors’ depression scores after the earthquake. The survivors’ average depression score decreased from 25.502 in the first wave to 24.606 in the third wave (also see Fig. 1). This shows that from the time of the earthquake to the four years after the earthquake, the survivors’ depression decreased, and their mental health continued to recover. Based on the results of the t-tests in Table 2, this study found differences in depression scores among survivors in each wave by education level.

Table 2 Depression scores of earthquake survivors with different characteristics among different waves
Full size table
Fig. 1
figure 1

Changes of survivors’ depression scores after the earthquake

Full size image

Results of HLM analysis

Table 3 shows the HLM results. We interpreted the results based on Model 4. Concerning the fixed effect, survivors’ depression scores varied depending on their education level. The difference in depression scores between survivors in elementary school or below and those in middle high school or above at the time of the earthquake (initial state) was 2.242. However, gender had no significant effect on the initial state of the survivors’ depression score.

Table 3 Results of hierarchical linear modeling with depression scores as the dependent variable
Full size table

Regarding individual social capital, individual social networks had no significant effect on the survivors’ initial depression scores but had a significant effect on the depression trajectory. The survivors’ depression scores decreased by 0.015 for each unit of increase in individual social network, which indicates that individual social networks can help survivors recover from psychological trauma over time.

From the perspective of community social capital, community trust had a significant effect on the initial state and trajectory of survivors’ depression scores. Community trust had an effect of -1.099 per unit on these scores, and survivors with high community trust had lower depression scores than those with low community trust had. For each additional unit of survivors’ community trust, the survivors’ depression scores increased by 0.302. However, the impact of community belonging on the initial state and growth rate was not significant.

Regarding the allocation of relief resources, the amount of relief funds received by survivors, fairness perception in resource relief, and satisfaction with resource relief had no significant effect on the survivors’ depression scores when the earthquake occurred. However, perception of fairness in resource relief had a significant effect on the change in survivors’ depression scores after the earthquake. That is, the difference in the effect on depression scores between survivors who considered the post-disaster resource allocation very unfair versus those who considered it very fair was 2.207; the difference between those who considered it not very fair and those who considered it very fair was 2.194; and the difference between those who considered it mostly fair and those who considered it very fair was 1.592. In addition, the amount of relief funds received by survivors and satisfaction with resource relief had no significant impact on the change in survivors’ depression scores over time.

Regarding random effects, the intra-individual variance of Model 4 decreased compared with that of Models 1–3, indicating that Model 4 explains more intra-individual variation when incorporating the variables of individual social network and resource relief.

Marginal effect analysis

This study also conducted marginal effect analysis to obtain a typical fitting trajectory. Specifically, we defined large individual social networks (average plus one standard deviation), small individual social networks (average minus one standard deviation), high community trust (average plus one standard deviation), and low community trust (average minus one standard deviation), and substituted these values into Model 4 to plot the marginal effects.

Figure 2 shows the relationship between individual social networks and the change in survivors’ depression scores after the earthquake. It appears evident that from the time of the earthquake to four years after, the depression scores of survivors with large social networks showed a downward trend, while those with small social networks showed an upward trend. This shows that survivors with large social networks can better recover from psychological trauma, and thereby supports Hypothesis 1 of this study. However, the effect of community trust on changes in depression over time was different. The depression scores of survivors with high community trust were significantly lower than those of survivors with low community trust at the time of the earthquake, however, the depression scores of survivors with high community trust increased over the four years following the earthquake and approached the average level of depression scores of survivors (see Fig. 3). This shows that, at the beginning of the earthquake, high community trust protected the survivors’ mental health, however, this protective effect began to decrease with the passage of time.

Fig. 2
figure 2

Individual social network and survivors’ depression scores after the earthquake

Full size image
Fig. 3
figure 3

Community trust and survivors’ depression scores after the earthquake

Full size image

Figure 4 depicts the relationship between resource allocation and changes in survivors’ depression scores. Survivors who considered the allocation of disaster relief resources to be very fair had a decreasing depression score trajectory. That is, these survivors had decreasing depression scores in the four years following the earthquake. In addition, the depression scores of survivors who believed that the resources allocation was not very fair or very unfair rose slightly in the four years following the earthquake. This finding partially supports Hypothesis 3 of this study.

Fig. 4
figure 4

Graded allocation of resource and survivors’ depression scores after the earthquake

Full size image

Discussion

Many studies have shown that social capital can enhance the connection and trust between disaster survivors and inspire their mutual help and support, thus promoting the survivors’ recovery from depression (Noel et al., 2018; Wind et al., 2011, 2021). The results of this study further support these conclusions, having found that individual social capital also has a long-term effect on the depression of survivors and can promote their recovery from depression symptoms. However, this study does not support the impact of community social capital on the survivors’ recovery from depression; thus, this effect requires further study.

In the process of disaster relief and post-disaster reconstruction, survivors’ perception of fairness regarding the allocation of disaster relief resources directly affects their depression, and thus affects their psychological rehabilitation. In the Chinese context, the government holds a large number of disaster relief resources and is responsible for the overall allocation of these resources (Li et al., 2015). In other words, the amount of disaster relief resources that survivors receive and the timeliness and fairness of relief resource allocation have a significant impact on their depression recovery trajectory (Cunningham et al., 2021; Hong & Zhao, 2019). Therefore, strengthening the government’s resource allocation capacity can increase vulnerable survivors’ (including rural residents) access to psychotherapy resources, which could accelerate the survivors’ recovery from depression. In addition, we should look for alternative sources of resources, such as non-governmental organizations and rural autonomous groups. As important disaster relief forces, these organizations have closer contact with survivors and can play a useful complementary role in the allocation of disaster relief resources and in the survivors’ psychological recovery.

This study has several limitations. First, the survey was mainly conducted in rural disaster-stricken areas; hence, the findings may only be applicable to the psychological recovery of survivors in similar areas. Second, the measurement of individual social capital adopted the method of “Chinese New-Year Greeters Network,” which was mainly developed for measuring social networks among urban residents. The validity of applying this measurement tool to disaster victims in rural areas needs to be further verified. Furthermore, the Cronbach’s alpha value of the depression scale in this study was slightly lower than 0.7, which may affect the reliability of the conclusions to a certain extent.

Conclusions

Using the longitudinal data collected after the Wenchuan earthquake, this study found that survivors’ perceptions of fairness regarding the distribution of disaster relief resources as well as individual and community social capital, had significant effects on the depression of survivors and could play positive roles in their recovery from depression after the disaster. Therefore, in the post-disaster relief process, we should not only provide money and material resources for the survivors, but also pay attention to the construction of their post-disaster social networks. First, survivors’ psychological recovery after a disaster is a long process. Policymakers should consider this in the development of post-disaster psychological support policies and formulate long-term policies to provide continuous psychological support services for survivors. Second, the government should provide more relief resources to survivors as much as possible, and ensure the timely, fair, and reasonable distribution of relief resources. Third, in the process of post-disaster psychological construction, we should give full play to the important role of social capital. We should strengthen the emotional connection among the survivors by reconstructing their social networks and cultivating community emotion, as this could support their recovery from depression.