Dark side or bright side: The impact of alcohol drinking on the trust of Chinese rural residents

Dong, Jie; Zhao, Qiran; Ren, Yanjun Article — Published Version Dark side or bright side: The impact of alcohol drinking on the trust of Chinese rural residents International Journal of Environmental Research and Public Health Provided in Cooperation with: Leibniz Institute of Agricultural Development in Transition Economies (IAMO), Halle (Saale) Suggested Citation: Dong, Jie; Zhao, Qiran; Ren, Yanjun (2022) : Dark side or bright side: The impact of alcohol drinking on the trust of Chinese rural residents, International Journal of Environmental Research and Public Health, ISSN 1660-4601, MDPI, Basel, Vol. 19, Iss. 10, pp. 1-15, https://doi.org/10.3390/ijerph19105924 , https://www.mdpi.com/1660-4601/19/10/5924 This Version is available at: https://hdl.handle.net/10419/265373 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/   Citation: Dong, J.; Zhao, Q.; Ren, Y. Dark Side or Bright Side: The Impact of Alcohol Drinking on the Trust of Chinese Rural Residents. Int. J. Environ. Res. Public Health 2022,19, 5924. https://doi.org/10.3390/ ijerph19105924 Academic Editors: Charlotte Probst and Shannon Lange Received: 3 March 2022 Accepted: 9 April 2022 Published: 13 May 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). International Journal of Environmental Research and Public Health Article Dark Side or Bright Side: The Impact of Alcohol Drinking on the Trust of Chinese Rural Residents Jie Dong 1, Qiran Zhao 2and Yanjun Ren 3,* 1School of Economics and Business Administration, Chongqing University, Chongqing 400044, China; [email protected] 2College of Economics & Management, China Agricultural University, Beijing 100083, China; [email protected] 3College of Economics and Management, Northwest A&F University, Yangling, Xianyang 712100, China *Correspondence: r[email protected] Abstract: Existing studies have explored the causal effect of social capital on harmful drinking, while the effect of drinking habits on trust is scant. In China, drinking rituals and drinking culture are considered important ways of promoting social interaction and trust, especially in rural areas where traditional culture is stronger. Based on a field survey in rural China in 2019, this paper explores the relationship between drinking habits and trust. First, we found a negative relationship between drinking habits and trust, indicating that those people who drink alcohol are more likely to have a lower trust. Second, we found significant heterogeneity in the effect of alcohol consumption on social trust across various groups. Specifically, the negative effects of alcohol consumption on trust were stronger for the females than for males; drinking alcohol did not reduce the level of trust among the Chinese Communist Party (CCP) in rural China; compared with the Han nationality, we found that the effect of drinking on trust was not significant for the ethnic minority. Third, we observed that the negative effects of alcohol consumption on trust had thresholds across age and income. Among people under 51, the risk of trust from drinking was greater than for those over 51; the negative effect of drinking on residents’ trust was more obvious in low-income families, but not significant in the group with an annual household income of more than CNY 40,000. Our empirical study provides a deeper understanding of drinking culture in rural China from a dialectical perspective. Keywords: dark side; alcohol drinking; trust 1. Introduction In China, the wine culture formed from ancient times has long been an indispensable part of Chinese social interaction [ 1 ]. Like other East Asian countries, drinking rituals and drinking culture in China are considered to be important ways of promoting social interaction and trust [ 2 ]. Chinese people share stories over wine tables, and the relationship gets better when they get tipsy. In modern society, as an important platform for business negotiation, drinking at dinner is often used to maintain good relations between bosses and employees or to promote business cooperation between business partners [ 3 – 5 ]. For example, Huang et al. [ 6 ] found that male CEOs in areas with a strong drinking culture had more social connections, both inside and outside the company. Hao et al. [ 7 ] also believe that social drinking is an essential skill for managers, which can relieve tension and embarrassment and promote social interaction. However, there is not much commercial activity in rural areas, and social drinking is different from that in urban areas. On the one hand, the social capital (social trust and social network) of Chinese rural residents is relatively simple. Drinking is not as utilitarian as commercial drinking tables; on the other hand, the traditional culture in rural areas is more preserved, and the culture of drinking tables during festivals is one of the important ways of maintaining rural social networks. Int. J. Environ. Res. Public Health 2022,19, 5924. https://doi.org/10.3390/ijerph19105924 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2022,19, 5924 2 of 15 Therefore, the main motivation of this study was to explore the impact of drinking on the social capital or social network of rural residents. Existing studies have explored the link between social capital and harmful drinking (HD) or alcohol consumption [ 8 – 11 ]. Moreover, most of these studies focus on areas with drinking habits and cultures. For example, researchers have looked at the impact of alcohol consumption on social relations in northern Europe, where drinking is well known within Europe [ 12 ]. Ahnquist et al. [ 13 ] found that lack of trust in institutions increased the likelihood of harmful alcohol consumption in Sweden. However, another study from Sweden demonstrated that social capital at the contextual level showed very weak effects on alcohol consumption for teenagers [ 14 ]. In Denmark, friends often get drunk as a sign of mutual respect [ 15 ]. Similarly, China also has a longstanding drinking culture; for example, there are many ancient Chinese poems related to drinking. In the context of wine culture, drinking is also very popular and common in China. Several studies have examined the impact of social capital on HD in China [ 16 – 18 ]. For instance, a study found that a high level of social capital may promote HD among the residents of Chinese neighborhoods [ 16 ]. However, taking Chinese migrant workers as the research object, Gao et al. [ 17 ] concluded that higher social capital reduces the possibility of problematic drinking among migrant workers. Additionally, in Taiwan, China, Chuang et al. [ 19 ] found that social engagement promoted drinking in both men and women. Others focused on adolescents and found that trust is significantly associated with drinking [ 20 – 24 ]. In addition, a large number of studies have found the influence of peer effects of adolescent social interaction on drinking habits [25–28]. Previous studies analyzed the effect of social capital on alcohol intake—social capital or social trust is the cause. For instance, Gao et al. [ 17 ] studied the influence of social capital on problematic drinking among migrant workers in China and found that higher individual-level social capital may protect against HD. However, studies on the effect of drinking habits on social trust are scant. In particular, we do not know how and to what extent alcohol consumption could affect trust among Chinese. Unlike the extensive literature focusing on the health effects of alcohol consumption [ 29 – 31 ], the impact of alcohol consumption on social capital has not received much attention. Meanwhile, in recent years, the Chinese government has implemented the “rural revitalization” strategy, which aims to improve the sense of contentment and happiness of rural Chinese. Good social capital and trust relationships are the premises behind enhancing people’s wellbeing [ 32 ]. In rural China, the social relationship among residents is not complicated, and information transmission is not as fast as that between cities (e.g., the Internet and smartphone usage rates are relatively low). As a result, their level of trust also differs from that of urban residents. Moreover, trust can be divided into vertical trust and horizontal trust. Vertical trust—namely, institutional trust—refers to residents’ trust in the institutional environment, involving government credibility in administration, judicature, taxation, and so on; horizontal trust is the general value of non-institutional trust—that is, the trust between friends, relatives, and neighbors in the general sense [ 33 – 36 ]. In order to make the research more comprehensive, we considered both vertical trust and horizontal trust in the construction of trust indicators. Based on the above background, our research questions also came out: does drinking affect the trust of rural residents? If so, is it negative (dark side) or positive (bright side)? In addition, is there heterogeneity in the effect among different subgroups? Based on a field survey in rural China, this research explored the relationship between drinking habits and social trust. The rest of the article is organized as follows: Section 2 deals with theories and hypotheses; Section 3presents data, variables, and the model; Section 4presents the regression results; Section 5discusses the results; and Section 6closes with our conclusion. Int. J. Environ. Res. Public Health 2022,19, 5924 3 of 15 2. Literature Review A large body of literature has also demonstrated that drinking alcohol can enhance the social capital of residents, including social trust and social networks [ 37 – 42 ]. Drinking is more of a social culture than a personal habit or preference, and it is deeply ingrained around the world [ 6 , 43 ]. For the most part, drinking is seen as a social lubricant that can be used as a medium to enrich people’s social networks. Some economists argue that drinking can induce people to reveal (unwillingly) information about themselves, thus pulling people into social distancing [ 44 – 46 ]. Some studies find that drinking alcohol can promote trust. For example, in studies set in Denmark, increased drinking among adults has been accompanied by an increase in trust [ 37 , 38 ]. Sayette et al. [ 39 ] claim that alcohol consumption promotes emotion-related behaviors at the individual and group levels in two ways: it enhances positive behaviors and decreases negative behaviors. Frank et al. [ 46 ] found that drinking alcohol does not necessarily mean increased trust, but moderate drinking does. Other studies have found that drinking strengthens social networks. For example, Bray [ 47 ] found that moderate drinking increases wage returns and accumulation of social experience and social capital. Similarly, another study, from Germany, showed that alcohol consumption increased wage returns and strengthened social networks [ 48 ]. Moreover, Groh et al. [ 49 ] from the opposite perspective, found that abstinence harms social networks among friends. In addition, studies have shown that drinking alcohol increases friendships and produces a sense of connection with others [ 50 – 52 ]. For example, MacLean [ 50 ] suggested that drinking alcohol enhances intimacy and demonstrates trust, especially at higher levels of intoxication, based on interviews of those aged 18–24 years in Australia. A second branch of the literature shows that alcohol consumption harms social capital. Some studies have shown that alcohol-dependent people have significant emotional empathy deficits that make it difficult for them to trust others [ 53 , 54 ]. Moreover, drinkers tend to fall into a vicious cycle of self-centeredness: drinking leads to self-centeredness, which in turn leads to alcoholism [ 55 , 56 ]. Furthermore, drinking can weaken rationality, and when people are not completely rational in social interactions, lying and cheating can occur [ 57 , 58 ]. Ahnquist et al. [ 13 ] found that low levels of institutional trust were associated with an increased likelihood of dangerous alcohol use among adults in Sweden. Similarly, Lindstrom [ 59 ] found that alcoholics in Sweden were generally less trusting. Other studies have shown that drinking tends to have negative social effects. For example, Fielding et al. [ 60 ] proved that drinking alcohol makes people less generous. Schweitzer et al. [61] conducted a scenario simulation to explore the impact of drinking on personal decision-making and found that drinkers were more likely to make radical choices and make mistakes. The third branch of studies has found that drinking does not affect people’s social capital. Bregu et al. [ 62 ] suggested that alcohol consumption has little systemic effect on economic behavior. From a business perspective, Brañas-Garza et al. [ 63 ] found that drinking alcohol does not affect the outcome of negotiations. Another interesting study reports that drinking increased males’ promises to others, but had no effect on their fulfillment— suggesting that drinking does not, at the least, increase people’s trust levels [64]. 3. Data and Method 3.1. Source of Data The data for this study came from a primary survey in 2019 by the Center for Human Capital and Geoeconomics (CHCG) at China Agricultural University. The survey was approved by the Ethics Committee of China Agricultural University, and a study on the link between alcohol and depression also use these data [ 65 ]. This survey covered 50 villages from 7 provinces of mainland China (Heilongjiang, Henan, Zhejiang, Yunnan, Xinjiang, Shandong, Anhui). Ten households were randomly selected from each village. Before data collection, all respondents voluntarily signed an informed consent form after receiving the questionnaire for scientific research. Moreover, they were also told that the Int. J. Environ. Res. Public Health 2022,19, 5924 4 of 15 feedback was confidential. After matching all the variables and dropping observations with missing covariates, the final sample consisted of 5207 rural adults. 3.2. Variables 3.2.1. Dependent Variable The explained variable of this research was the trust level of rural residents. Trust can be divided into vertical trust and horizontal trust [ 33 ]. Vertical trust represents institutional trust, which refers to residents’ trust in the institutional environment, involving government credibility; horizontal trust is the general value of non-institutional trust—that is, the trust between friends, relatives, and neighbors in the general sense [ 34 ]. Therefore, we also considered both vertical trust and horizontal trust in the design of the questionnaire. In China’s rural areas, village cadres are the policy transmitters and managers that villagers contact directly, so the trust of village cadres to a large extent represents the vertical trust of residents. Horizontal trust includes the trust of neighbors, kin, and friends, which are the most common social objects in rural China. In the questionnaire, we designed four indexes about the trust degree both from vertical trust and horizontal trust: the trust degree of village cadres (vertical trust), neighbors, kin, and friends (horizontal trust). As shown in Table 1, each trust indicator for trust levels from lowest to highest is presented on a scale of 1–10. Table 1. Definition and statistics of variables. Variable Variable Definitions Mean S.D. Min Max Dependent variable Trust in cadres The degree of trust to village cadres: 0–10; low–high 8.712 1.973 0 10 Trust in neighbors The degree of trust to neighbors: 0–10; low–high 8.246 1.805 0 10 Trust in kin The degree of trust to kin: 0–10; low–high 9.079 1.351 0 10 Trust in friends The degree of trust to friends: 0–10; low–high 8.643 1.832 0 10 Trust Comprehensive indicators of trust obtained by PCA 17.286 2.640 1.934 19.950 Independent variables B_drink Do you currently drink alcohol? 1 = yes; 0 = no 0.694 0.461 0 1 Control variables Individual factors Gender 1 = male; 0 = female 0.845 0.362 0 1 Ethnic 1 = Han; 0 = minorities 0.822 0.382 0 1 Age Years 50.618 10.822 19 80 Edc Education years 7.932 3.340 0 15 Health Excellent–poor: 1–5 1.956 1.015 1 5 Interpersonal factors Friends Number of friends 18.182 10.736 0 100 WeChat Do you use the social networking APP WeChat? 1 = yes; 0 = no 0.893 0.309 0 1 Organizational factors Agriculture Is your family engaged in agriculture? 1 = yes; 0 = no 0.899 0.302 0 1 F_mem Number of family members 3.934 1.668 1 15 Community factors Distance_V Distance from household to village committee (km) 1.590 11.449 0 100 Distance_T Distance from household to county center (km) 22.710 19.203 0 200 Public policy factors CCP Are you a member of the Chinese Communist Party (CCP)? 1 = yes; 0 = no 0.281 0.450 0 1 P_news Do you follow political news? 1 = yes; 0 = no 7.826 2.811 0 10 To better quantify the trust degree of villagers, we used principal component analysis (PCA) to reduce the dimensionality of each trust variable, while minimizing the loss of information. First, we examined whether the data support PCA. As shown in Table 2, the KMO value was 0.738 (higher than the threshold of 0.7), indicating that the data support Int. J. Environ. Res. Public Health 2022,19, 5924 5 of 15 the PCA method. Moreover, we see that only the eigenvalue of Comp1 is 2.330, greater than 1, indicating that the Comp1 can be used as a linear combination of variables. Table 2. Statistics of principal components. Component Eigenvalue Difference Proportion Cumulative Comp1 2.330 1.573 0.583 0.583 Comp2 0.758 0.295 0.189 0.772 Comp3 0.463 0.013 0.116 0.888 Comp4 0.450 —— 0.112 1.000 KMO 0.738 To further prove that Comp1 is the only principal component, we also drew a scree plot. As shown in Figure 1, the abscissa represents the number of principal components and the ordinate represents the eigenvalues. When the x-coordinate exceeds 2, the eigenvalues begin to flatten out, so it is appropriate to choose an eigenvalue. In conclusion, it is convincing to choose Comp1 as the principal component of the variable. Int. J. Environ. Res. Public Health 2022, 19, x 5 of 16 To better quantify the trust degree of villagers, we used principal component analysis (PCA) to reduce the dimensionality of each trust variable, while minimizing the loss of information. First, we examined whether the data support PCA. As shown in Table 2, the KMO value was 0.738 (higher than the threshold of 0.7), indicating that the data support the PCA method. Moreover, we see that only the eigenvalue of Comp1 is 2.330, greater than 1, indicating that the Comp1 can be used as a linear combination of variables. Table 2. Statistics of principal components. Component Eigenvalue Difference Proportion Cumulative Comp1 2.330 1.573 0.583 0.583 Comp2 0.758 0.295 0.189 0.772 Comp3 0.463 0.013 0.116 0.888 Comp4 0.450 —— 0.112 1.000 KMO 0.738 To further prove that Comp1 is the only principal component, we also drew a scree plot. As shown in Figure 1, the abscissa represents the number of principal components and the ordinate represents the eigenvalues. When the x-coordinate exceeds 2, the eigenvalues begin to flatten out, so it is appropriate to choose an eigenvalue. In conclusion, it is convincing to choose Comp1 as the principal component of the variable. Figure 1. Scree plot of PCA. Finally, we have the loading value of Comp1, as shown in the abscissa of Figure 2, which can intuitively present the impact of each variable on the principal component, which is, according to the loading value in ascending order: village cadres (0.443), friends (0.504), kin (0.513), and neighbors (0.535). Thus, we achieved the purpose of dimensionality reduction for the four trust variables and obtained a composite indicator, which is the variable Trust. The statistic of the Trust variable is shown in Table 1. 1.5 2.50.5 1.0 2.0 Eigenvalues 1 2 3 4 Number 95% CI Eigenvalues Scree plot of eigenvalues after pca Figure 1. Scree plot of PCA. Finally, we have the loading value of Comp1, as shown in the abscissa of Figure 2, which can intuitively present the impact of each variable on the principal component, which is, according to the loading value in ascending order: village cadres (0.443), friends (0.504), kin (0.513), and neighbors (0.535). Thus, we achieved the purpose of dimensionality reduction for the four trust variables and obtained a composite indicator, which is the variable Trust. The statistic of the Trust variable is shown in Table 1. 3.2.2. Independent Variable The core independent variable of this study was whether residents drink alcohol and was defined as “B_drink”. In the field study, the question was “Do you currently drink alcohol?” Participants responded with “Yes = 1; No = 0”. The core independent variable of this study was whether residents drink alcohol. As shown in Table 1, the proportion of rural residents who drink alcohol was close to 70%, indicating that the proportion of rural residents who drink alcohol was relatively high in rural China. 3.2.3. Control Variables This research added control variables according to the ecological model [ 66 ]. In our case, factors affecting residents’ trust come from five dimensions: individual factors, interpersonal factors, organizational factors, community factors, and public policy factors. This paper first controlled individual factors (gender, ethnicity, age, education, health) [20,67,68]. Int. J. Environ. Res. Public Health 2022,19, 5924 6 of 15 Second, we used the number of friends and whether respondents used the social APP WeChat as interpersonal variables. Third, we controlled for organizational factor variables: whether the family was engaged in agriculture and the number of family members. Fourth, we added the two distance variables as community factors: distance from household to village committee and distance from household to county center. The closer they lived to the village committee, the wider their social network within the community. In addition, the closer the villager was to the county center, the more extensive and the faster the information they received. Last, we used whether residents were Chinese Communist Party (CCP) members and whether they cared about political news as public policy factors. Because in rural China members of the CCP often serve as village officials, they tend to have a broader social network [ 69 , 70 ]. The definition and statistics of control variables are shown in Table 1. Int. J. Environ. Res. Public Health 2022, 19, x 6 of 16 Figure 2. Loading diagram for PCA. 3.2.2. Independent Variable The core independent variable of this study was whether residents drink alcohol and was defined as “B_drink”. In the field study, the question was “Do you currently drink alcohol?” Participants responded with “Yes = 1; No = 0”. The core independent variable of this study was whether residents drink alcohol. As shown in Table 1, the proportion of rural residents who drink alcohol was close to 70%, indicating that the proportion of rural residents who drink alcohol was relatively high in rural China. 3.2.3. Control Variables This research added control variables according to the ecological model [66]. In our case, factors affecting residents’ trust come from five dimensions: individual factors, interpersonal factors, organizational factors, community factors, and public policy factors. This paper first controlled individual factors (gender, ethnicity, age, education, health) [20,67,68]. Second, we used the number of friends and whether respondents used the social APP WeChat as interpersonal variables. Third, we controlled for organizational factor variables: whether the family was engaged in agriculture and the number of family members. Fourth, we added the two distance variables as community factors: distance from household to village committee and distance from household to county center. The closer they lived to the village committee, the wider their social network within the community. In addition, the closer the villager was to the county center, the more extensive and the faster the information they received. Last, we used whether residents were Chinese Communist Party (CCP) members and whether they cared about political news as public policy factors. Because in rural China members of the CCP often serve as village officials, they tend to have a broader social network [69,70]. The definition and statistics of control variables are shown in Table 1. 3.3. Model and Preliminary Statistical Analysis 3.3.1. Model The following econometric model was estimated: 𝑇𝑟𝑢𝑠𝑡=𝛼+𝛽𝐵_𝑑𝑟𝑖𝑛𝑘 +𝛾𝑋 +𝜀  (1) Cadres Neighbours Kins Friends 0-0.5 0.5 1.0 Component 2 0.44 0.46 0.48 0.50 0.52 0.54 Component 1 Component loadings Figure 2. Loading diagram for PCA. 3.3. Model and Preliminary Statistical Analysis 3.3.1. Model The following econometric model was estimated: Trusti=α+βB_drinki+γXi+εi(1) where the subscript i indicates the individuals; Trust is the comprehensive trust index of residents by PCA; B_drink is the core explanatory variable based on the question “Do you currently drink alcohol?” and replying “yes = 1, no = 0”. Xis a vector of variables that controls the five dimensions of the ecological model mentioned above; ξ is a random disturbance term. 3.3.2. Preliminary Statistical Analysis As shown in Table 3, we divided the residents into drinking samples and non-drinking samples to observe the statistical differences between the two groups. It can be seen that the explained variable Trust in this paper had a difference of 0.203 between the two groups and was significant at the 5% level, which means that residents who do not drink have higher levels of trust. Table 3. Mean differences in Trust between drinking and non-drinking. Mean of Non-Drinking (ND) Mean of Drinking (D) Mean of D-ND Trust 17.427 17.224 −0.203 ** (2.667) (2.625) [2.565] Note: Standard deviation in parentheses; T values in square brackets; *** p< 0.01, ** p< 0.05, * p< 0.1. Int. J. Environ. Res. Public Health 2022,19, 5924 7 of 15 4. Results 4.1. Baseline Results The effect of drinking on trust is shown in Table 4. In Column 1, the coefficient of drinking for individuals is − 0.203 at a significant 1% level. Adjusting for individual factors (Column 2), the association between drinking and trust is strengthened, with a coefficient value of − 0.409, which is significant at the 1% level. The effect of drinking on trust is also significantly decreased with additional Friends and WeChat variables (Column 3). It shows that the negative effect of drinking on trust still exists after controlling for the confounding interpersonal factors. After controlling organizational factors, the coefficient of the core explanatory variable is − 0.350 and is significant at the 1% level (Column 4). After the community factors are added, the core explanatory variable decreases to − 0.320, but it is still significant at the 1% level (Column 5). Finally, after controlling all variables, the coefficient of the core explanatory variable is − 0.354, which is significant at the level of 1%, indicating that drinking has a negative relationship with residents’ trust with full control. Table 4. The results of baseline regression. (1) (2) (3) (4) (5) (6) Variables Trust Trust Trust Trust Trust Trust B_drink −0.203 * −0.409 *** −0.337 ** −0.350 ** −0.320 ** −0.354 *** (0.104) (0.121) (0.138) (0.137) (0.129) (0.125) Gender 0.104 0.170 0.181 0.103 −0.081 (0.188) (0.205) (0.204) (0.192) (0.194) Ethnic 0.332 * 0.418 ** 0.383 * 0.348 * 0.446 ** (0.168) (0.200) (0.202) (0.197) (0.202) Age 0.007 0.012 ** 0.010 * 0.010 * −0.002 (0.005) (0.005) (0.006) (0.005) (0.005) Edc 0.077 *** 0.073 *** 0.071 *** 0.063 ** 0.006 (0.020) (0.026) (0.026) (0.026) (0.029) Health −0.310 *** −0.336 *** −0.349 *** −0.331 *** −0.293 *** (0.052) (0.062) (0.063) (0.060) (0.061) Friends 0.002 *** 0.002 *** 0.002 *** 0.001 *** (0.001) (0.000) (0.000) (0.000) WeChat 0.037 0.031 0.031 −0.007 (0.217) (0.218) (0.207) (0.206) Agriculture −0.107 −0.119 −0.127 (0.192) (0.194) (0.191) F_mem −0.076 −0.071 −0.077 (0.049) (0.049) (0.050) Distance_V 0.003 * 0.003 * (0.002) (0.002) Distance_T −0.002 −0.001 (0.004) (0.004) CCP 0.411 *** (0.121) P_news 0.142 *** (0.025) Constant 17.427 *** 16.840 *** 16.466 *** 17.037 *** 17.205 *** 17.084 *** (0.103) (0.319) (0.483) (0.601) (0.592) (0.616) Observations 5207 5207 5207 5207 5207 5207 R-squared 0.001 0.030 0.034 0.036 0.032 0.058 Note: Standard errors in parentheses (clustered at the county level). *** p< 0.01, ** p< 0.05, * p< 0.1. 4.2. Identifying Causal Effects In the baseline regression model, the main source of endogeneity problems is reverse causality. To deal with this endogeneity, we employed the technique by Lewbel [ 71 ] to identify causality. The products of exogenous covariance and heteroscedastic error can be Int. J. Environ. Res. Public Health 2022,19, 5924 8 of 15 used as effective instrumental variables to identify endogenous parameters when effective instrumental variables cannot be found. In our case, we used the following equations: Trust =α1X+β1B_drink +ε1(2) B_drink =α2X+ε2(3) where β1 is the parameter of the core independent variable, B_drink is the endogenous variable, and X is a vector of control variables. We assumed there was not a valid IV for B_drink and that the error ε2 was heteroscedastic—that is, CovX, ε2 26= 0. According to the deduction, X−Xε2would be the valid IV for B_drink. Table 5reports the results of the regression using Lewbel’s method. Column 1 is the first-stage regression, and column 2 is the second-stage regression. First, we found that the null hypothesis of homoscedasticity was strongly rejected (the chi-squared statistic was 113.10, at a 1% significance level), conforming to the prerequisites of the method. Second, the coefficient was − 0.701 at a 10% level of significance. This indicates that drinking alcohol does have a decreasing effect on trust. Table 5. Identifying causal effects based on Lewbel’s method. (1) (2) Variables B_drink Trust B_drink −0.701 * (0.398) Gender −0.084 (0.194) Ethnic 0.441 ** (0.201) Age −0.001 (0.006) Edc 0.003 (0.031) Health −0.291 *** (0.061) Friends 0.001 *** (0.000) WeChat 0.014 (0.208) Agriculture −0.125 (0.190) F_mem −0.078 (0.050) Distance_V 0.003 * (0.002) Distance_T −0.002 (0.004) CCP 0.414 *** (0.121) P_news 0.141 *** (0.024) Error*c_Gender −0.471 (0.314) Error*c_Ethnic 0.470 (0.723) Error*c_Age −0.051 *** (0.016) Error*c_Edc 0.126 *** (0.047) Error*c_Health −0.018 (0.113) Int. J. Environ. Res. Public Health 2022,19, 5924 15 of 15 69. Meng, T.G.; Chen, H. A multilevel analysis of social capital and self-rated health: Evidence from China. Health Place 2014 , 27, 38–44. [CrossRef] [PubMed] 70. Jiang, J.; Wang, P. Is Linking Social Capital More Beneficial to the Health Promotion of the Poor? Evidence from China. Soc. Indic. Res. 2019,147, 45–71. [CrossRef] 71. Lewbel, A. Using Heteroscedasticity to Identify and Estimate Mismeasured and Endogenous Regressor Models. J. Bus. Econ. Stat. 2012,30, 67–80. [CrossRef] 72. Cochrane, J.; Chen, H.H.; Conigrave, K.M.; Hao, W. Alcohol use in China. Alcohol Alcohol. 2003,38, 537–542. [CrossRef] 73. Wang, J.X.; Rao, Y.L.; Houser, D.E. An experimental analysis of acquired impulse control among adult humans intolerant to alcohol. Proc. Natl. Acad. Sci. USA 2017,114, 1299–1304. [CrossRef] 74. Hansen, B.E. Sample splitting and threshold estimation. Econometrica 2000,68, 575–603. [CrossRef] 75. Cooper, M.L. Motivations for alcohol use among adolescents: Development and validation of a four-factor model. Psychol. Assess. 1994,6, 117. [CrossRef] 76. Gonzalez, V.M.; Collins, R.L.; Bradizza, C.M. Solitary and social heavy drinking, suicidal ideation, and drinking motives in underage college drinkers. Addict. Behav. 2009,34, 993–999. [CrossRef] 77. Christiansen, M.; Vik, P.W.; Jarchow, A. College student heavy drinking in social contexts versus alone. Addict. Behav. 2002 , 27, 393–404. [CrossRef] 78. Creswell, K.G. Drinking together and drinking alone: A social-contextual framework for examining risk for alcohol use disorder. Curr. Dir. Psychol. Sci. 2021,30, 19–25. [CrossRef] 79. Yu, M.M.; Gu, L.B.; Jiao, W.J.; Xia, H.Z.; Wang, W.R. Predictors of self-neglect among community-dwelling older adults living alone in China. Geriatr. Nurs. 2019,40, 457–462. [CrossRef] 80. Steele, C.M.; Josephs, R.A. Alcohol Myopia-Its Prized and Dangerous Effects. Am. Psychol. 1990,45, 921–933. [CrossRef] 81. Macdonald, T.K.; Zanna, M.P.; Fong, G.T. Decision-Making in Altered States-Effects of Alcohol on Attitudes toward Drinking and Driving. J. Personal. Soc. Psychol. 1995,68, 973–985. [CrossRef] 82. Denton, K.; Krebs, D. From the Scene to the Crime-the Effect of Alcohol and Social-Context on Moral Judgment. J. Personal. Soc. Psychol. 1990,59, 242–248. [CrossRef] 83. Wynne, L.A. ‘Just Right’: Balance, tranquility, and drunkenness in rural Yucatan. Glob. Public Health 2020 ,15, 704–714. [CrossRef] 84. Blume, S.B. Women and Alcohol—A Review. JAMA 1986,256, 1467–1470. [CrossRef] 85. Ma, L.; Christensen, T. Government Trust, Social Trust, and Citizens’ Risk Concerns: Evidence from Crisis Management in China. Public Perform. Manag. Rev. 2018,42, 383–404. [CrossRef] 86. Christensen, T.; Fimreite, A.L.; Laegreid, P. Crisis Management: The Perceptions of Citizens and Civil Servants in Norway. Adm. Soc. 2011,43, 561–594. [CrossRef] 87. Zhang, C. Family support or social support? The role of clan culture. J. Popul. Econ. 2018,32, 529–549. 88. Brandt, M.J.; Wetherell, G.; Henry, P.J. Changes in Income Predict Change in Social Trust: A Longitudinal Analysis. Political Psychol. 2015,36, 761–768. [CrossRef] 89. Ananyev, M.; Guriev, S. Effect of Income on Trust: Evidence from the 2009 Economic Crisis in Russia. Econ. J. 2019 ,129, 1082–1118. [CrossRef]