Constancy and Changes in the Distribution of Religious Groups in Contemporary China: Centering on Religion as a Whole, Buddhism, Protestantism and Folk Religion

Abstract

Since 1978, mass religious conversion has been a prominent feature of Chinese religious life. In the 1980s and 1990s, the “Five types of believers’ distribution (Wuduo, 五多)” were characterized by the inclusion of “more women”, “more elderly”, “more sick people”, “more rural residents”, and “more people with lower socioeconomic status”, as is the academic consensus. In the 21st century, some scholars have proposed the different view of “Three trends in faith stratification (Sanhua, 三化)”, namely urbanization, rejuvenation, and elitism. However, these claims are either based on theoretical analysis or local cases and data analysis. Based on the Chinese General Social Survey (CGSS) data from 2010 to 2017, this study constructed a comprehensive explanatory framework with macroscopic, mesoscopic, and microcosmic levels to verify the authenticity of the two sets of propositions by taking religion as a whole and also examining some representative religions such as Buddhism, Protestantism and folk religion. The results show that they are partially invalid, and thus, need to be further refined to be explained; the distribution of believer groups varies among religions and between urban/rural areas.

1. Introduction

Since 1978, China’s religions have been expanding rapidly. According to official data, China has nearly 200 million religious believers (The State Council Information Office of the P. R. C. 2008). The number of religious believers can reflect the overall development of China’s religion, but not the contemporary distribution of different religious sub-groups, which may help reveal the changes in China’s religious development, and thus, show the impact of social transformation on religion.

In the 1980s, there was an upsurge of mass religious conversion in China, leading to a “religious fever”. In this period, almost all scholars reached an agreement that Chinese religious believers were characterized by the inclusion of “more women”, “more elderly”, “more sick people”, “more rural residents”, and “more people with lower socioeconomic status” (Zhang and Yin 1988; Liang 1999), which was defined as the “Five types of religious believers distribution”. Entering the 21st century, with the rapid economic development and social differentiation, some scholars have proposed the “Three trends in faith stratification”—urbanization, rejuvenation, and elitism (Duan 2013). However, a unified understanding on the two propositions has not yet been formed (Wu et al. 2020; Liu and Yang 2022).

In general, after more than 40 years of reform and opening up, there is a lack of more rigorous empirical research on the current distribution of religious groups in China, both in domestic and international academic circles. Among the few empirical studies, He and Mencken (2010) examined the group differences in karma belief; Lu and Zhang (2016) present a relatively simple introduction. Therefore, this study aimed to test the existing proposition of the “Five types” and “Three trends” based on the national data of CGSS 2010–2017.

On the basis of the main topics of the academic discussion of “Five types” and “Three trends”, the number of religious believers and the characteristics of CGSS data, we mainly discuss the religious believers’ distribution of religion as a whole and three specific religions, i.e., Buddhism, Protestantism and folk religion. This is because, firstly, these three religions are the most popular in China (The State Council Information Office of the P. R. C. 2008), as reflected in the CGSS data. Buddhism can be regarded as the representative of traditional Chinese institutional religion, while folk religion is the representative of traditional diffusive religion. Secondly, some scholars pointed out that Chinese Catholics have not changed significantly compared to Protestants; for nearly three decades, the gap between Christianity and traditional religion has been largely driven by the rapid growth of Protestantism (Sun 2014; Madsen 2003), but the proportion of Catholics in the CGSS data is lower than 0.5%, which made it difficult to conduct more complex multivariable statistical analysis and valid statistical inference. Therefore, we have to abandon the analysis of the distribution of Catholics. Thirdly, Taoism is also not considered in this paper, as it accounts for a low percentage (about 0.25%) in the CGSS data. Lastly, because of the ethnic and geographical nature of Islam in China, and the CGSS did not specially take this into account in its sampling, its data are not nationally representative.

Accordingly, the paper mainly focuses on religion as a whole. Buddhism, Protestantism and folk religion present the distribution of these religious believers in contemporary China, from which we can also observe the influence of social change on religion and the trends in the religious landscape in the future to a certain extent.

2. Theoretical Background

The current issue of the distribution of religious groups in China includes group differentiation based on the ascriptive role and faith stratification based on socioeconomic status. There are some major differences compared with the West. First, the religious stratification in Western society is mainly based on different denominationalism choices within Christianity, whereas the belief differentiation in Chinese society is affected by competition among different religions (He and Mencken 2010). Second, the differentiation of faith in contemporary China is a result of the massive religious conversion that occurred after 1978 (Yang 2014). Third, the distribution of religious believers is also a manifestation of urbanization and class differentiation at the level of religious conversion during the rapid social change in China.

2.1. Religious Stratification and Mass Religious Conversion

In the Western literature, the discussion of religious stratification originated with Marx and Weber. Niebuhr (1929) followed and developed this line of thought by exploring social class, racial/ethnic and regional differences between denominationalisms. The topic of religious stratification has been of concern for Western academia (Pope 1953; Demerath 1965; Roof and McKinney 1987; Davidson 2008). Since the 1980s, based on the United States’ experience, many scholars have pointed out that denominational social class lines are becoming increasingly blurred (Wuthnow 1988; Christopher 1989). There are still many scholars who argue that although the gap in socioeconomic status between denominations has tended to narrow, relative positions have remained largely unchanged (Davidson et al. 1995; Pyle 2006). Unlike in the United States, religious stratification has been less pronounced in Europe because of its established church tradition (Berger et al. 2008).

In traditional Chinese society, where Confucianism is linked to the cosmological view, the royal family monopolized belief in the Mandate of Heaven. The upper class focused on the indoctrination of beliefs to maintain power. The intellectual elites were concerned with the philosophical belief, whereas the general public believed in the spirituality of reciprocity between gods and humans (Li 2010, p. 217). After the founding of the People’s Republic of China, religious beliefs were considered to be contrary to the official atheism and were severely suppressed. The “religious fever” that emerged after 1978 manifested in the large number of religious believers. When the “people” in the class struggle discourse are transformed into a different “people” with the division of social interests generated by the market, it is only natural to identify religious diversity and the tendency to form groups.

The explosive growth in beliefs and religious stratification relates to religious conversion. The process model presented by Lofland and Stark (1965) is the first sociological model of religious conversion (Yang and Abel 2014). There are three main perspectives for conversion research in the sociology of religion: (1) Studies grounded on the individual level (Snow and Machalek 1984; Richardson 1985; Kilbourne and Richardson 1989; Bainbridge 1992); (2) Explanations based on religious organizations and institutions (Stark and Finke 2000; Rambo 1993); and (3) Yang’s alternative argument. Yang believed that focusing on the individual level cannot explain phenomena such as the mass conversion in China; instead, one should focus on social and cultural influences (Yang 1998, 2005, 2014; Yang and Tamney 2006).

2.2. Integrated Analytical Framework

The term “conversion”, which emphasizes membership in and the identity of a faith, is primarily based on Western religious realities. However, theoretical appropriateness for different religious traditions must be taken into consideration (Suchman 1992; Jindra 2011). In traditional Chinese religious practice, membership is not a prerequisite for religious activities (Fan and Chen 2014); Adam Yuet (2006) refers to the Chinese way of believing as “doing religion”.

We developed a comprehensive explanatory framework with macroscopic, mesoscopic, and microcosmic levels, and especially focused on the impact of macroscopic factors proposed by Yang and Abel (2014). The individual choice is both synchronically affected by individuality and diachronically subject to social change. The mass conversion that takes place in this period is both a choice of the individual and a product of the times. Naturally, it is also influenced by the operation mode of different religious organizations.

At the macro level, unlike in Western societies, Chinese society has been undergoing rapid social transformation since 1978, with various social structural elements in a state of flux. Differences in the times inevitably led to some changes in the distribution of religious groups while the mass conversion was taking place. Firstly, in the political sphere, with the decline in ideology after 1978, religious affairs were brought under the management of the United Front System, and the citizens’ freedom of religious belief was guaranteed to some extent. These left some room for the development of all religions. Secondly, in the economic sphere, because economic development has replaced the class struggle as the central task, the allocation of resources throughout society has also been changed. Although the state still controls important resources and allocations, society has increasingly become an independent source of opportunities, which constitutes the cornerstone of the reconfiguration of social relations in China after the reform and opening up (Sun et al. 1994). Thirdly, in the social sphere, society has become increasingly polarized, especially in the 1990s. The Gini coefficient exceeded 0.4 for the first time in 1998. This means that social class differentiation is accelerating, with the increasing social wealth spawning a large middle class. Moreover, from the perspective of the spatial distribution of the population, urbanization bears the heavy burden of China’s modernization. China’s urbanization rate exceeded 50% in 2011, an unprecedented change in population distribution in the country’s history.

Choosing different religions should also factor in the meso-level, which affects believers’ conversion, i.e., each religion’s characteristics. Previous studies mostly focused on the rapid growth of Protestantism in China on this view. It has been believed that the doctrines can provide greater stability for believers than traditional Chinese religions, and its ways of believing are more in line with modern society (Fiedler 2010). In addition, the organizational and institutional characteristics of Protestantism provide communal living spaces, interpersonal attachments (Koesel 2013; Sun 2017), and even material and spiritual welfare (Zheng et al. 2010; Ruan et al. 2015). Thus, we can assume that both the middle class and disadvantaged groups would be more inclined to believe in Protestantism than other religions.

At the micro level, on the one hand, while public life is deteriorating and economism is gradually dominating, a triple atomization tendency of politics, economy, and value has emerged in contemporary China. Although there is room for the revival of traditional elements, such as clan relations and the principle of reciprocity after the Cultural Revolution, it is the instrumentalization and commodification of interpersonal relationships that are the most important features (Gold 1985). The utilitarianism of interpersonal relationships and the anomie of values create great uncertainties and risks to individuals’ lives, and “survival anxiety in social transition” (Guangjin 2013) has become a common phenomenon. On the other hand, rapid social transformation has brought different changes in life opportunities for different people, and the change in the stratification mechanism has caused some class interests to be damaged or subjectively perceived as not gaining enough; therefore, there is a general sense of relative deprivation in society, especially for the lower-class and marginalized groups. The proposition of the “Five types” is based on this logic. Scholars have tended to employ the “deprivation theory” or “marginalization theory” to explain conversion, arguing that poor or marginalized people are more likely to seek spiritual support and comfort in religion (Dai and Peng 2000, pp. 77–87).

According to the above discussion, compared with the previous “Five types” proposition, we consider that the distribution of China’s religious believers is both stable based on individual ascriptive factors and variable due to social changes. This variability is ascribed by the “Three trends” proposed by academics.

As an ascriptive role, gender and age are less affected by social changes. There are two main paradigms to explain why women are more inclined to convert to religion. One is the deprivation theory, which emphasizes that women experience greater social inequalities, leading to a higher sense of deprivation. They are, therefore, more likely to seek religious solace. The second is the risk preference theory. It argues that unbelief is a typical risk-taking behavior, and that men take more risks than women; thus, females are more likely to be religious than males (Miller and Hoffmann 1995; Freese 2004). Miller and Stark (2002) also found that gender differences in belief exist in different types of religions. The highly institutionalized religions of Christianity, Islam, and Orthodox Judaism have greater differences than Buddhism, Shintoism, and other Jewish denominations with more diffused characteristics. Based on these, we hypothesize that more women are also religious than men in China today.

The risk analysis paradigm is also used to explain why the elderly and the sick are more likely to be religious (Zheng et al. 2010). The theory of individual life cycle proposes that as one enters old age, there are more mental health demands to deal with, presenting greater risks, especially death. Religious teachings and social support can be important resources to cope on the crises (Bosworth et al. 2003). Moreover, given that Protestantism is so organized that it can provide stronger social support than traditional Chinese religions, we hypothesize that the elderly are more likely to be religious and more inclined to believe in Protestantism.

Religion can become an important source of spiritual solace and social support in response to the physical suffering caused by disease (Idler 1987). This is further reinforced by the utilitarian nature of Chinese faith. Relevant empirical studies have shown that one of the most important reasons why Chinese people believe in religion is because either they or their family members are sick (Zhou and Sun 2017).

The urban and rural distribution of religious believers is also quite common. As the two main forms of economic and social organization, urban and rural areas, determine the way of life, thus influencing the identities and choices of different individuals. Classical secularization theories concern the “less urban religious believer”. Cox (2013) argued that the increase in urban civilization is a direct cause of religious decline, and the roots of secularization should, therefore, be ascribed to urbanization. The rural represents a “communal” way of life based on traditional spiritual order, low mobility, high integration, and emotion, which is more conducive to the survival of religion.

Urbanization has led to a change in the geographic distribution of the population in China. Some scholars affirm that urban residents dominate the group of religious people (Wu et al. 2020; Duan 2013), but others still argue otherwise (Liu and Yang 2022). We hold that as the urban population increases, more and more rural people, especially younger people, flow into the cities, and the transition anxiety is more pronounced among the urban population. Therefore, the number of urban and younger believers will be higher.

In addition, Dong and Yang (2014) emphasized the difference between northern and southern China, arguing that Protestantism is prevalent in the rural north; however, because they did not use a rigorous large-scale sample, these findings need to be further tested.

In terms of socioeconomic status distribution, firstly, Chin’s economic development has generally raised the income levels of the Chinese people, with the middle class and the new affluent class growing. Official estimates suggest that the current middle-income group in China has reached 400 million. Secondly, the sense of relative deprivation and existential anxiety is prevalent among all classes, especially among the middle class (Zhou 2005), and it is common among the affluent class to pursue the meaning of life, producing “boss Christian” and the opinion that “Buddha only helps the rich”. Therefore, we hold that the number of religious believers will increase, especially in cities, the youth, and those with a higher socioeconomic status.

Taken together, we deduce that “more women”, “more elderly”, and “sicker and more people with lower socioeconomic status” are still constituent aspects of religious believers; “more rural residents” is no longer a factor. The trends of urbanization and rejuvenation are taking place. These are distinct profiles in different religions. However, these hypotheses must be tested based on Chinese religious practice.

3. Methods

3.1. Data

We used data from the Chinese General Social Survey (CGSS) for 2010, 2012, 2013, 2015, and 2017; the CGSS is conducted by the China Survey and Data Center of Renmin University of China. The sampling of these five waves was based on the same sampling frame, and the questions and the answers were all coded consistently. Considering the heterogeneity of Chinese urban and rural societies, the CGSS sampling frame settings and the topics discussed in this paper, our statistical analyses incorporated the official urban–rural weights provided by the CGSS in order to obtain more accurate conclusions.

3.2. Measures

Dependent variables. As mentioned above, the aim of this study was to test the proposition of the “Five types” and “Three trends” by taking religion as a whole and also examining each religion such as Buddhism, Protestantism and folk religion. Thus, there were two sets of dependent variables: (1) Religious believers as a whole. The CGSS asked respondents “Which religion do you belong to”. We created a binary indicator for those who held religious beliefs (coded 1) and those who did not (coded 0); (2) Specific religious believers. We only examined differences in the distribution of the three chosen types of religious groups: Buddhists, Protestants and folk religion followers. Respondents were asked to choose between “yes” (coded 1) and “no” (coded 0).

Key independent variables. Based on the consideration of “Five types” and “Three trends”, there were five key independent variables: (1) Gender: we coded “male” as 0 and “female” as 1; (2) Age: we constructed a continuous variable by subtracting the year of survey from the year of birth of the interviewees; (3) Health: we used the subjective health self-assessment in the questionnaire to measure the respondents’ physical condition. The options of this question ranged from “very unhealthy” to “very healthy”, to which we assigned values of 1 to 5; (4) Socioeconomic status: we measured this from three aspects—subjective class, education level and family income. First, many Chinese sociologists believe that due to rapid social changes and cultural values, the correlations between the main indicators of objective stratification (e.g., income, education, occupation, and consumption) and class status identity are not very strong (Li 2005). Therefore, compared with the objective dimension of stratification, the subjective dimension is a better way to reveal social changes and social differentiation in people’s subjective perceptions and values (Qin 2016). Based on this, we measured the subjective self-assessment of identity through the question “On which social class level do you think you are currently?” This question was answered on a scale of 1–10, with 1 meaning the bottom and 10 meaning the top. In addition, we measured the level of education by querying participants on the time spent in formal education. Finally, household income was taken as the total annual income of the whole family; we used the natural logarithm; (5) Area: values of 0 and 1 were attributed to urban and rural residents, respectively.

Control variables. We controlled for a variety of factors, including ethnicity (Han Chinese = 0; other = 1), marriage (married = 1; unmarried = 0) and the survey waves. The study involves multiple periods of data collected in different years; therefore, we created a continuous variable to show the effects of time.

3.3. Method

To test the authenticity of the “Five types” and the “Three trends”, firstly, we presented the frequency distribution of the above-mentioned groups, tested whether there were significant differences by bivariate analysis and showed the tendency by urban/rural area, age, and socioeconomic status. Secondly, this study used binary logistic regression to explore different groups’ likelihoods of believing in religion with a full sample of data. Thirdly, considering the socioeconomic and cultural differences between urban and rural areas in China, we also used the urban–rural data as a sub-sample to further examine whether there were urban/rural differences in the belief mechanisms.

4. Results

4.1. Descriptive Statistics and Bivariate Analyses

Table 1. Descriptive statistics of the proportion of each religious group.

	Buddhist	Taoist	Catholic	Protestant	Folk Religion Believer	Religious Believer
CGSS2010	5.48%	0.25%	0.23%	1.93%	2.89%	13.74%
CGSS2012	6.02%	0.26%	0.20%	2.31%	3.43%	14.44%
CGSS2013	5.66%	0.32%	0.32%	1.88%	2.05%	11.44%
CGSS2015	4.74%	0.27%	0.18%	2.11%	1.68%	10.63%
CGSS2017	4.66%	0.23%	0.20%	1.41%	2.11%	10.61%
Total	5.31%	0.26%	0.23%	1.92%	2.44%	12.18%

shows that those who declared holding religious beliefs accounted for more than 10% of all respondents in each period of data collection. In terms of the specific religions, self-proclaimed Buddhists had the highest percentage, accounting for approximately 5%. Folk religion believers and Protestants came second. Taoists and Catholics comprised approximately 0.2% of the whole sample.

Following the existing assessment with the propositions of the “Five types” and “Three trends”, we combined the five waves into one value and presented the distribution of religious believers. At the same time, to test the correlation between different subgroups and tendencies, we built bivariate lists and use chi-squared tests for categorical variables and t-tests for continuous variables.

According to the frequency analysis and chi-squared test results presented in

Table 2. Bivariate analysis of the “Five types” (I).

		Religious Believer	Protestant	Buddhist	Folk Religion Believer
Gender	Man	39.32%	28.92%	37.86%	48.98%
	Woman	60.68%	71.08%	62.14%	51.02%
	χ2	339.94 ***	216.83 ***	215.679 ***	1.688
Urban/rural	Urban	61.76%	50.64%	67.31%	59.42%
	Rural	38.24%	49.36%	32.69%	40.58%
	χ2	0.414	44.435 ***	50.425 ***	2.078

Note: *** p < 0.001 (2-tailed).

, women were the majority of religious believers of Protestantism and Buddhism. However, there was no gender difference among the followers of folk religion. This showed that “more women” are common in institutional religions such as Protestantism and Buddhism.

Regarding urban and rural geographical distribution, the results showed that there are far more Buddhists in cities than in rural areas, whereas Protestants numbered slightly more in cities. There was no urban–rural difference in the number of religious believers or followers of folk religion. These disprove the previous proposition that there are “more rural” religious believers.

Table 3. Bivariate analysis of the “Five types” (II).

		Age	Health	Education	Income	Subjective Class
Religion as a whole	Believers	50.608	3.420	7.667	10.218	4.210
	None	47.293	3.622	9.185	10.278	4.171
	T-test	−14.175 ***	13.388 ***	22.546 ***	0.751	−1.613
Protestantism	Believers	52.656	3.320	7.396	9.828	4.076
	None	47.482	3.607	9.030	10.277	4.185
	T-test	−10.128 ***	8.662 ***	11.084 ***	8.667 ***	1.963 **
Buddhism	Believers	50.880	3.451	8.095	10.365	4.297
	None	47.397	3.610	9.049	10.263	4.177
	T-test	−11.155 ***	7.826 ***	10.130 ***	−3.151 **	−3.852 ***
Folk religion	Believers	48.611	3.420	6.931	10.217	4.169
	None	47.556	3.606	9.050	10.269	4.183
	T-test	−2.322 *	6.327 ***	16.195 ***	1.292	0.320

Note: *** p < 0.001, ** p < 0.01,* p < 0.05 (2-tailed).

shows that the average age of religious believers, Buddhists, Protestants, and folk religion believers, was higher than that of non-believers; therefore, the claim that they are “more elderly” still holds true.

In terms of physical health, firstly, based on the mean health self-assessment values of the respondents, we found that believers’ health assessment, in general or in all three specific religious groups, was significantly lower than that of non-believers. Secondly, the mean value of those with faith was around 3.4, which was lower than those without faith (3.6), but still exceeded the theoretical mean of 2.5. Although it was difficult to directly conclude whether there were more sick people, we maintain that there are more believers with poorer health than in the general population.

As far as socioeconomic status is concerned, in the field of education, the religious believers had spent significantly fewer years in education than non-believers, which manifested in general, as well in the three faith groups. Buddhists had average years of schooling, followed by Protestants and folk religion believers. In the aspect of income and subjective class, both Protestant household income and subjective class identity were significantly lower than non-believers’, whereas Buddhists were the opposite. There were no substantial differences between religious believers and non-believers, or between folk religious followers and non-believers.

It can be concluded that there are four commonalities among all kinds of believers: more women, more elderly, more people with poorer than average health, and more people with a low level of education. However, whether people were living in rural areas was no longer an influencing characteristic of the distribution of religious believers in China. Of course, there were also some differences among the various religions.

The above discussion is only a result of the statistical analysis; “urbanization”, “rejuvenation”, and “elitism” form a dynamic process. The “three trends” do not necessarily mean “more urban residents”, “more young people”, and “more elite people”; therefore, this should be examined from a comparative perspective.

The proposition of “urbanization” is strongly supported by “more urban residents”; this paper also analyzed the dynamic changes in the proportion of urban residents among the followers of each religion in the five periods of data. In order to present changing trends more concisely, we focus on the data of 2010 and 2017.

As seen in Figure 1, the proportion of urban Buddhists among all Buddhists remained above 60% from 2010 to 2017. Although Protestantism has had some peaks and troughs, its “urbanization” trend has increased from 44.51% in 2010 to 51.18% in 2017. If the ambiguity of respondents’ subjective understanding of folk religion is disregarded, the trend shows that the proportion of urban folk religion believers was 40.80% in 2010, although it has been above 50% since 2012, and 70.53% in 2017. Accordingly, the “urbanization” of believers in recent years is indeed supported by national statistics.

Figure 2 presents the changes in the age composition of the four categories of religious adherents in the 2010–2017 data. In the first and last waves, although the proportions of religious believers over 60 years old were similar, the proportion of believers under 30 years old increased by approximately 10%. This means that the tendency of “rejuvenation” has taken place.

Figure 3 shows the changes in the subjective class identity of the four categories of believers. Comparing the data from 2010 to 2017, we found that the subjective class identities of all types of believers increased: religious believers increased from 4.16 to 4.26; Buddhists increased from 4.24 to 4.34; Protestants increased from 3.95 to 4.12; the numbers of folk religion believers fluctuated more considerably, with only a slight increase. The questionnaire used a subjective identity scale between 1 and 10; these figures mean that believers all show a tendency of moving up from the lower middle class to the middle class. Nevertheless, we cannot conclude that there is an elitist trend, only a middle-class trend.

4.2. Binary Logistic Regression of the Full Sample

The above analyses did not control for other variables. The dependent variable is the presence or absence of faith; therefore, according to the method described above, we constructed two sets of binary logistic regression models based on the full sample and urban/rural sub-samples to test the “Five types” and “Three trends” hypotheses.

From the results presented in

Table 4. Binary logistic regression model for the likelihood of believing in religion.

	Religious Believer	Buddhist	Protestant	Folk Religion Believer
Gender	0.408 *** (0.039)	0.477 *** (0.052)	1.004 *** (0.089)	−0.297 *** (0.079)
Age	0.003 + (0.001)	0.007 *** (0.002)	0.013 *** (0.003)	−0.0147 *** (0.003)
Health	−0.096 *** (0.019)	−0.079 ** (0.024)	−0.074 + (0.039)	−0.128 *** (0.036)
Education	−0.058 *** (0.005)	−0.038 *** (0.007)	−0.009 (0.010)	−0.125 *** (0.010)
Income	0.068 *** (0.017)	0.108 *** (0.026)	−0.040 (0.023)	0.114 ** (0.035)
Subjective class	0.035 ** (0.011)	0.036 * (0.015)	−0.002 (0.025)	0.0407 + (0.022)
Area	−0.203 *** (0.045)	−0.378 *** (0.062)	0.378 *** (0.091)	−0.187 * (0.088)
Ethnicity	0.410 *** (0.066)	−0.069 (0.093)	−1.137 *** (0.205)	0.662 *** (0.101)
Marriage	−0.103 * (0.047)	−0.191 ** (0.061)	−0.094 (0.099)	0.113 (0.103)
Wave	−0.047 *** (0.008)	−0.029 ** (0.011)	−0.027 + (0.015)	−0.090 *** (0.017)
Constants	0.003 + (0.001)	0.007 *** (0.002)	0.013 *** (0.003)	−0.015 *** (0.003)
Observations	50,322	51,660	51,660	51,660
Pseudo R2	0.026	0.02	0.046	0.039

Notes: The standard errors are shown in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05, + p < 0.1 (2-tailed).

, after controlling for other variables, it can be concluded that women are more likely to be religious, Buddhist and Protestant, than men. However, they are less likely to choose a folk religion. Age also has a positive effect on all kinds of religious groups. The difference is because age is positively associated with the probability of being religious, Buddhist and Protestant; nevertheless, there is a negative correlation with folk religion. Those with lower self-health assessment were more likely to be religious in all models. Education is negatively related to religious beliefs; Buddhism and folk religion, family income, and subjective class identity were positively related to holding religious beliefs, although they all had no significant relationship in the likelihood of believing in Protestantism. Rural residents were less likely than urban residents to believe in religion, Buddhism and folk religion, but they were more likely to believe in Protestantism.

As shown in Table 4, compared with Han Chinese, ethnic minorities are more likely to believe in religion or folk religion, but less likely to believe in Protestantism. The probability of believing in Buddhism shows no clear distinction between them. Marriage does not have a significant effect on Protestant and folk religion believers, but it is negatively related to Buddhist. In the wave variables, a slowly decreasing trend in the likelihood of people believing in religion was observed as time changed.

4.3. Binary Logistic Regression of the Sub-Sample of the Urban/Rural Data

From the results presented in

Table 5. Binary logistic regression model of urban and rural areas.

	Religious Believer		Buddhist		Protestant		Folk Religion Believer
	City	Rural	City	Rural	City	Rural	City	Rural
Gender	0.375 *** (0.050)	0.522 *** (0.065)	0.466 *** (0.063)	0.556 *** (0.095)	0.863 *** (0.123)	1.178 *** (0.131)	−0.289 ** (0.104)	−0.176 (0.121)
Age	−0.001 (0.002)	0.010 *** (0.003)	0.005 * (0.002)	0.014 *** (0.004)	0.010 ** (0.004)	0.017 *** (0.005)	−0.027 *** (0.004)	0.002 (0.005)
Health	−0.113 *** (0.025)	−0.087 ** (0.0275)	−0.105 *** (0.031)	−0.053 (0.040)	−0.123 * (0.058)	−0.046 (0.053)	−0.135 ** (0.051)	−0.136 ** (0.052)
Education	−0.084 *** (0.006)	−0.002 (0.009)	−0.055 *** (0.008)	0.012 (0.013)	−0.037 ** (0.012)	0.034 * (0.016)	−0.182 *** (0.012)	−0.036 * (0.018)
Income	0.073 ** (0.026)	0.074 ** (0.023)	0.096 ** (0.037)	0.146 *** (0.039)	−0.041 (0.035)	−0.036 (0.031)	0.168 ** (0.062)	0.084 * (0.041)
Subjective class	0.061 *** (0.015)	−0.003 (0.0174)	0.059 ** (0.019)	−0.008 (0.025)	0.012 (0.036)	−0.016 (0.033)	0.080 ** (0.029)	−0.006 (0.033)
Ethnicity	0.205 + (0.113)	0.539 *** (0.083)	−0.028 (0.130)	−0.114 (0.131)	−0.855 ** (0.281)	−1.308 *** (0.291)	−1.053 ** (0.321)	1.327 *** (0.123)
Marriage	−0.177 ** (0.058)	0.028 (0.082)	−0.241 *** (0.072)	−0.066 (0.120)	−0.298 * (0.130)	0.172 (0.155)	0.159 (0.136)	0.054 (0.157)
Wave	−0.048 *** (0.011)	−0.049 *** (0.013)	−0.045 ** (0.014)	0.001 (0.018)	−0.025 (0.0210)	−0.030 (0.022)	−0.068 ** (0.022)	−0.142 *** (0.028)
Constants	95.045 *** (21.102)	95.106 *** (25.261)	86.590 ** (27.462)	−6.971 * (35.249)	47.347 * (42.303)	55.937 * (43.907)	134.711 ** (44.60)	280.597 *** (56.30)
Observations	30,448	19,874	31,227	20,433	31,227	20,433	31,227	20,433
Pseudo R2	0.035	0.021	0.026	0.015	0.043	0.046	0.060	0.050

Notes: The standard errors are shown in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05, + p < 0.1 (2-tailed).

, after controlling for the other variables, it can be concluded that women are more likely to be religious, Buddhist, and Protestant than men, regardless of urban and rural area, but less likely to be folk religion believers if living in cities. Gender has no clear effect on belief in folk religion in rural areas. Age has a positive effect on being Buddhist and Protestant in all areas, a positive effect of people being religious only in rural areas, a negative effect on belief in folk religion in cities, and no effect on following any religion in cities or folk religion in rural areas. There is no significant correlation between belief in Buddhism or Protestantism and health status in rural areas; thus, health is negatively correlated with belief. The level of education is negatively related to all religious beliefs in cities and folk religion in rural areas, whereas it has a positive effect of being a Protestant in rural areas. Income is positively correlated with the likelihood of belief in religion, Buddhism, and folk religion in urban and rural areas; subjective class identity only shows the same profile in urban areas. Meanwhile, these two variables were not significantly related to the likelihood of being a Protestant regardless of urban and rural areas.

5. Conclusions and Discussion

The analyses present the distribution of religious groups and the trends in faith stratification. Compared with the proposition of the “Five types”, some changes occurred. First, “more rural residents” was no longer a characteristic of the geographical distribution of religions in China, and each religion showed either “more urban believers” or “no significant difference between urban and rural areas”. Second, although the number of religious believers who assessed themselves as less healthy was large, their self-assessed health was still higher than the average; therefore, the proposition that “more sick people” would be religious believers may not be accurate. We propose the term “more people with poorer health condition” instead. Third, except for folk religion, the characteristics of “more women”, “more elderly”, “more less healthy people” and “more less educated people” remained the same among religion as a whole, Buddhism and Protestantism. Fourth, there was another prominent change. There are “more people with higher socioeconomic status” (“more rich—man” and “more people with higher class identity”). Lastly, the composition of Buddhists differed the most from the “Five types”.

Based on the conditions of the different religions and areas, we may question the theoretical and logical support of the existing reasons of religious conversion in China. Concerning gender, “more women” were present in Buddhists and Protestants, regardless of the full sample or urban and rural samples, respectively. Women were less likely than men to believe in folk religion, both nationwide and in cities, which is consistent with Miller and Stark’s (2002) findings of gender differences between institutional and diffused religions, but there is no gender difference in rural areas. In terms of age, we find that “more elderly” fits mainly into institutional religious groups. However, age is negatively correlated with folk religion in both national and urban contexts, but not so in rural areas. In terms of health status, although this study found that “more people in poorer health” was a more precise expression than “more patients”, which could be interpreted by the “risk analysis paradigm”, there was no significant correlation between health self-assessment and the likelihood of practicing Buddhism or Protestantism in rural areas.

As far as the three variables of socioeconomic status are concerned, in terms of the level of education, the traditional explanation is that people with a lower level of education are more likely to be religious, either because they may lack scientific knowledge or because they feel more relative deprivation due to their lower social status. This explanation only applies in cities, not rural areas. The level of education is positively correlated with the likelihood of believing in Protestantism. On the income side, the distribution of Protestant groups also shows different characteristics from theoretical derivations. Income has a positive effect on the likelihood of belief in Protestantism. There is a negative relationship between income and the likelihood of being a Protestant, but no significant correlation in either urban or rural areas. In terms of subjective class identity, it is positively correlated with the likelihood of believing in religion, Buddhism, and folk religion, but not Protestantism.

This shows the importance of the comprehensive explanatory framework with macroscopic, mesoscopic, and microcosmic levels in explaining the current mass religious conversion in China. The specific explanations of these issues are beyond the scope of this paper, and need to be discussed one by one.

Regarding the proposition of the “Three trends”, “urbanization” can be directly confirmed, which is a natural consequence and an epiphenomenon of China’s accelerated urbanization process, population migration and the growth of the middle class. However, we also find that the “urbanization” level varies among different religions. Buddhism is the most urbanized, followed by Protestantism and folk religion. In addition, “rejuvenation” is tentatively supported, which is based on the fact that the average age of believers is lower than that of non-believers, with an increasing proportion of believers under the age of 30. At the same time, the trend toward the “rejuvenation” of folk religion believers is obvious. Although the statement of the tendency toward “elitism” is not accurate, we argue that it can be described as the tendency toward the “middle class”. For religion in general, Buddhism and folk religion, it is evidently clear that not only has the gap between believers and non-believers narrowed or even disappeared in terms of household income class identification, but it also shows a positive correlation between income and class and the likelihood of believing in them. The lower income and lower class categories are dominated by Protestants.

Through these comparisons, we can see that these religious landscape and development trends of different religions have both similarities and differences; the latter reveals an urban/rural distinction, as well as religious differences among institutionalized religion, folk religion, Protestantism and traditional Chinese religions after decades of social change.

From the above findings, we can put forward the following statements predicting the trends of their future developments.

First, with respect to urban–rural geography, the growth rate of urban believers of Buddhism and folk religion will be faster than that of rural believers, and the growth rate of rural believers in Protestantism will be faster than that of urban believers (Wu et al. 2020); however, with the expanding urbanization process, “more urban believers” will become a stable pattern in the future.

Second, institutional religions, such as Buddhism and Protestantism, have similar profiles in terms of their believers’ health and education. The tendency for those who are less healthy and educated to believe in these two religions will be strengthened nationwide, as well as in urban areas. Folk religion shows the same feature on all three levels, namely, national, urban, and rural areas.

Third, for Buddhism and folk religion, the proportion of the believers with more wealth and higher social status will increase. However, Protestantism shows a diffusive trend at the national level, i.e., there are no clear differences among Protestants. This seems to be consistent with the phenomenon of Christian pluralism in recent years, such as “cultural Christians”, “migrant workers’ Christians”, and “boss Christians”.

Beyond these contributions, the paper also brings to attention some areas for further research. The first is more general and relates to the measure of Chinese religious belonging. On the one hand, some Chinese believers may know which deity they are worshipping, but they rarely care which religion that deity belongs to due to a lack of religious awareness towards religious affiliation (Zhang and Lu 2020). On the other hand, exclusivity is the core character of Western religions, while traditional Chinese religion is characterized as pantheism. Correspondingly, additive (Jordan 1993) or multiple religious belonging (Oostveen 2019; Yang and McPhail 2023) are some of the most salient characteristics of some Chinese religious practices. However, since the respondents of CGSS are required to only choose a single option of their religious affiliation, this study only aimed to reveal one aspect of the changing distribution of religious groups in contemporary China, given the scarcity of national religious data based on scientific sampling. Thus, we propose to conduct nationwide research by exploring more precise measurement methods, as well as conducting qualitative case studies in the future.

The second is about the heterogeneity of the urban religious adherents. We acknowledge that more and more rural people migrate to cities as a result of urbanization and social mobility. The religious believers in urban areas include both urban and rural hukou-registered residences during the current period of institutional transformation in China. Some believers may still be rural in all other characteristics but are counted in the urban population as a result of their place of residence. Though the religious believers’ distribution in rural and urban areas is a reflection of the official data, we still hold that it can reveal the characteristics of the urban and rural distribution of religious believers in current China, as well as predict the future to a certain extent. A further study could aim at the comparative analysis of the distribution of rural/urban religious groups to show each group’s distinct nature.