**1. Introduction**

As proposed by Sartorius [1] (p. 101), mental health is a ''the state of balance that individuals establish within themselves and between themselves and their social and physical environment". As per the report by Global Forum for Health Research and World Health Organization [2], mental and neurological disorders account for 13% of the global burden of disease. According to Zhang [3] and National Health and Family Planning Commission [4], 2.7% of the population in China was afflicted with severe mental illness in 1950s, a figure which rose to 5.4% in the 1970s, 11.1% in the 1980s, 13.47% in the 1990s, and further to 15% in the 2010s. Currently, over 30 million people have a diagnosis of severe mental illness (especially depression) in China [4]. Mental illness has been found to be the prime

cause of the high suicide rate in China, which stands at approximately 10 per 100,000 in 2016 [5]. These statistics call for an urgen<sup>t</sup> need to deepen our understanding on the determinants of Chinese mental health [6]. Research indicates that mental health is determined by socioeconomic, environmental, intersectoral, and civic security factors [7]. Mental health policy and practice remain pivotal to the discussion on social capital [8], because China's surge in mental illness problems is probably due to its rapid urbanization, which is usually associated with a range of social health hazards and risks that can lead to the development of neuropsychiatric illness [4]. In fact, prior studies all over the world report that social capital is a crucial determinant of both physical and mental health issues at the individual and the aggregate level and their interactions [9–19]. Social capital can influence an individual's health by opening information channels, promoting collective action, addressing detrimental cultural norms, and fostering the development of support systems serving as a source of self-esteem and mutual respect [14,20]. These mechanisms of social capital also work in the context of mental health [7,8].

However, most prior studies were focused on western developed economies and presented mixed findings [21]. Research in the transition economies alluded to a positive relationship between social capital and improved health [22]. As argued by Ichida et al. [23], the relationship between high social capital and good health indicators can vary owing to the cultural differences among countries, justifying the need for a systematic study in the different cultural contexts. Recently, relevant research has been conducted in East Asia (for example by Fujisawa et al. [24]; Suzuki et al. [25]; Miller et al. [26], Yip et al. [21], Sun et al. [27]; and Yamaoka [28]). Although these works have advanced our understanding of the social capital-general health nexus, little attention has been particularly devoted to compare how the influence of social capital on mental health behaviors. To our best knowledge, there is little research addressing how social capital influences mental health differently in Chinese urban and rural settings. Such study is important, as prior works indicate that there is a huge urban-rural health disparity in China and call for further studies [29–32].

The major objective of this study is to fill the above-mentioned gap. Specifically, we examined empirical relationships of four dimensions of social capital, i.e., civic participation, civic trust, political participation, political trust, on self-rated mental health. We followed the composite hypothesis, which argued that the mental health of an individual was influenced not only by his or her human capital (i.e., education, income and social status) but also by his or her social capital, as people are usually involved in a variety of analytically distinct but nonexclusive social networks [33–35]. We also followed the multi-level framework proposed in [36]. By community, we referred to counties (*xian* in Chinese) or districts (*qu* in Chinese), the third level of China's administrative hierarchy. Our sample individuals were nested within communities and an individual's mental health status was a function of a set of individual covariates and human capital (i.e., gender, age, education, marriage, unemployment, social-economic status, household annual income, household wealth and household size), individual-level social capital (i.e., civic trust, civic participation, political trust, political participation), and community-level social capital. The multilevel linear regression technique was chosen as the most appropriate design tool to explain the variation in mental health status of the population by both within-community and between-community differences. To our best knowledge, the sample size of our study is the largest of its kind at the individual and community levels.

### **2. Materials and Methods**
