Family Structure, Market Competition and Local Culture: The Mechanisms of Religious mobility beyond Christianity

Abstract

Based on pooled data drawn from a representative Taiwan Social Change Survey dataset, this article analyzes the pattern of transmission of religious belief from parents to adult children in Taiwan from 1999 to 2009. As a society imbued with diverse religious traditions, Taiwan as a region of China serves as an ideal point of departure to examine religious mobility beyond the Western and monotheistic world. The article has three main findings. First, there is a gendered pattern in exchange mobility. Mothers play a stronger role than fathers overall, but Christian families have an inverse situation. Sons are more likely to adopt their parents’ religious belief than daughters. Second, a parent’s capacity to transmit his/her religious belief greatly depends on the religious group to which he/she belongs. Daoism shows more robustness than other religions in transgenerational transmission, which contradicts the prediction drawn from religious economy theory. Third, religious communities significantly differ from each other in structural mobility. Buddhism has the highest retentive capacity for reducing religious switching, and folk religions have the highest attractive capacity to win over switchers from other religious communities.

Introduction

Passing religiosity from parents to offspring serves as a key mechanism to retaining the vitality of religions. A religious group that makes the new generation inherit their parents’ religious belief will gain advantage in the competition for members and consequently change the landscape of religious groups. Scholars have long been interested in the way that religiosity is transmitted across generations, and many empirical studies have been conducted (Hunsberger and Brown 1984; Erickson 1992; Bader and Desmond 2006).

Despite its richness, the literature still has noteworthy weaknesses. First, most results stem from the homogeneity of samples—white American Protestant or Catholic adolescents (Keeley 1976; D’Antonio 1985; Ellison and Sherkat 1993; Bao et al. 1999; Hoge et al. 1982; Gunnoe and Moore 2002). Even outside the USA, the populations under investigation are still located in the Western world, which is characterized by the monotheistic tradition of Christianity, such as Australia (Hayes and Pittelkow 1993), Canada (Hunsberger and Brown 1984), and Britain (Taris and Semin 1997). It is unknown whether the structural factors that shape the characteristics of transmission work in Asian societies, which are distinguished by religious diversity and deep-rooted polytheism. This cultural and geographical limitation mirrors today’s lack of productivity in the sociology of religion, which has almost been transformed into a “sociology of Christianity,” in Turner’s words (Turner 1983, 5). Although a few scholars have extended scholarship to Hong Kong (Lang and Ragvald 1993), Taiwan (Hu and Leamaster 2013; Liu 2008), and Mainland China (Hu and Leamaster 2015; Yang 2006), the research community still marginalizes the religious phenomena of Chinese societies. The scarcity of survey data and quantitative works contrasts sharply with the diversity of religious traditions and with the complexity of polytheistic religions in China, issues that would enrich the sociology of religion.

Second, most works focus on adolescent religious retention but very few pay attention to adults (Smith and Sikkink 2003; Hoge et al. 1982; Bao et al. 1999; Flor and Knapp 2001). The neglect of the adult group rests on an implicit assumption about the parent-child relationship, that is, children enjoy high-level autonomy and self-sufficiency but low-level moral compliance to their parents. However, the power structure and gender roles in Chinese families that interweave with Confucian codes may not conform to that assumption drawn from the social life of the occidental world. The divergent cultural prescriptions on family and parenthood require more attention on adult children (Bellah et al. 1985; Tobin et al. 1991).

Third, the religious economy theory predicts that deregulation of the religious market creates a favorable advantage for institutional religions whose memberships are clear and exclusive, while bringing about unfavorable conditions for noninstitutional ones that separate purchase from supply and take a noncollective way of producing religious goods (Stark 2006; Stark et al. 2005). Using statistical data on Taiwan, Lu et al. (2008) support that prediction by showing that the abolition of regulations forced folk religions to develop into more structural organizations. However, Hu’s work challenges that prediction by pointing out the prominent rise rather than decrease of folk religions in Taiwan. These findings provide insight into understanding the complexity of religious life in Chinese society. It is still unclear whether nonexclusive, noninstitutional religions will retain new-generation believers after the religious market becomes free and highly competitive.

Taking account of these three flaws reviewed above, Taiwan was chosen as the case under investigation to expand the scope of scholarship to the non-European/American world. This study has several advantages over previous investigations on religious mobility. First, the data analyzed comes from the 1999–2009 Taiwan Social Change Surveys (TSCS), which provide both more and recent cases to compare religious mobility over three periods. Second, the work analyzes the pattern of religious retention of adult children (age > 18) rather than teenagers, a group neglected in most research on religious transmission. Third, I compare fathers’ with mothers’ influence on the religious retention of children and present a gender pattern of continuity of religious affiliation. Fourth, the retentive capacity of each main religious community is depicted according to data on switching on the collective level.

This article compares the theoretical expectations of social learning theory, religious capital theory, and religion market theory. Each of these views generates unique expectations about (1) levels of and trends in retention across denominations and between generation, (2) patterns of father’s and mother’s hold on offspring believers, and (3) the relative gains and losses from switching experienced by different religious groups across periods and cohorts.

The remainder of the article briefly reviews three theoretical lenses into parent-child religious transmission. The following part introduces the background of the case of Taiwan. Data and variables drawn from the Taiwan Social Change Survey are presented in the third part, which also includes descriptive statistics, period-cohort-age distribution of religious identity, the religious distance matrix, and the switching pattern of religious communities. Analytical models address the causal relationship between parental religious affiliation and adult child religious retention.

Literature review

In principle, the similarities and differences between children’s and parents’ religious identities can be attributed to two distinct social forces: exchange mobility and structural mobility (Hu and Leamaster 2015). The former is meant to discover transmission mechanisms at the micro level, such as parenthood, gender, and ethnicity, that result in systematical religious inheritance or deviation across generations due to people’s choices. The latter focuses on macrolevel structural change, such as transition of the religious market or fashion, which regulates the extent of opportunities for religious choice for the two generations. Social learning theory and religious capital theory pay close attention to the former, and religious economy theory provides analytical framework for the latter.

Social learning theory and religious retention

Social learning theory argues that the similarity between parents and children depends on the quantity and quality of their interaction. By observing two of the most important role models in the family, children imitate and internalize parents’ habits, attitudes, behavior modes, and even personality (Bandura 1977; Bandura and McDonald 1963). Religious learning is no exception. Parental religious participation influences the child’s preferences through two mechanisms: frequency and motivation (Cornwall 1988). The more parents participate in worship, praying, and scripture study, the higher the likelihood that their child will have religious involvement at the same level. Parker and Gaier (1980) find that the frequency of parental participation in religious activities can explain more than 60% of variance among high school students. Besides the frequency of learning, how the child interprets parental practices also conditions the succession of religiosity. Tight affective bonding with the parent gives the child stronger incentives to learn and internalize their practices (Roest et al. 2009). In contrast, conflict and tension in the family are likely to retard this process (Smetana 1997; Kochanska and Thompson 1997).

The mechanisms of both frequency and motivation magnify the mother’s influence on religious upbringing since she is more likely to be the caregiver, spend more time with the child, and establish a close affective relationship with them (Coltrane 2000; Goldscheider 2000). All these can become pathways for transmitting religiosity. Hunsberger and Brown (1984) find that students who enjoyed a good relationship with the mother during childhood were less likely to leave the religious community in which they were raised. However, the quality of the child’s relationship with the father did not have a significant effect. Bao et al.’s work supports this finding by pointing out that if the child felt accepted by the mother rather than the father, this significantly mediated mother-child religious similarity (Bao et al. 1999).

Empirical examinations of religious retention have generated mixed findings regarding the father’s role in religious upbringing. For example, the Intergenerational Panel Study of Parents and Children, a classic survey, did not cover any information on the father’s religiosity (Pearce and Thornton 2007). Likewise, Myers (1996) averages fathers’ and mothers’ religious participation instead of measuring them separately. Recent findings problematize this conventional practice. Hu and his colleagues find that the father and mother develop divergent paths of transmitting religiosity to offspring. The paternal effect is more powerful in Christian and Islamic groups, while the maternal effect is more powerful on passing on no belief (Hu and Leamaster 2015). Some scholars even argue that overall fathers play a more important role than mothers in the religious retention of children (Baker-Sperry 2001). Taking account of this inconsistent evidence, it is necessary to weight the maternal and paternal effects on religious upbringing separately.

Some scholars propose that there is a gendered pattern of religious socialization made up of parental labor division of religious socialization and daughter-son differences in the continuity of religious affiliation. By investigating parents and their firstborn sons in Protestant families in Virginia, Clark et al. (1988) argue that fathers shape a son’s religious behavior while the mother shapes their belief. Bra and his colleagues find that a mother and father only shape the religiosity of a child of the same gender. No significant mother-son or father-daughter effects were found, but both parents exerted a stronger influence on daughters than on sons. Social learning theories, specifically the literature on attachment and internalization, imply that (1) the mother has a greater hold on adult children than her spouse and (2) the mother and father only shape the religious affiliation of a child of the same gender.

Religious capital theory and religious choice

Scholars employing the rational actor framework explain religious retention and patterns of exchange by appealing to religious capital theory. Religious capital theory is driven by the assumption that the family works as a quasi-factory organization that produces household commodities for family members (Becker 1965). Household commodities include both tangible things like meals and intangible ones like health, love, and religious experience. Similar to commercial goods, household commodities are the outcome of essential productive factors and limited resources, including physical capital and human capital like time, labor, and education. That is exactly the case with religious production, which is in nature a process of investing time, voluntary work, mediation, money, skill, and experience in cultivating religious satisfaction and identity through participating in religious activities. Religious knowledge, friends, network, and understanding of a religious culture can be regarded as a collection, conceptualized as “religious capital” (Iannaccone 1990; Stark and Finke 2000). Religious capital seems to be both the precondition and the outcome of religious participation.

Parents and religious institutions are cardinal sources for gaining religious capital, since parents teach children on religious attendance and daily practices, and institutions provide religious members and peers to structure children’s social network. All of them promote children’s capacity of increasing religious capital and of consuming certain religious goods. The process of capital accumulation usually starts at the very beginning of childhood and reaches its threshold in adulthood. One strategic move to maximize the amount of religious capital is to keep the original membership instead of switching to another religious group; the latter requires the process of accumulation to start from scratch and the capital has to reset. Passing religiosity to the next generation is essentially intergenerational transmission of religious capital. The maximization of capital influences children’s choices about retention and switching. Hence, religious capital theory implies that (1) religious continuity should remain more prevalent than religious switching and (2) individuals who switch will tend to choose denominations similar to those from which they originated.

Religious economy theory and religious mobility

Religious economy theory applies rational choice theory to the field of religion and recodes religious phenomena within an economic framework. The fundamental rule of market economy—the supply-demand relationship—is used to model and explain religious behavior and the development or decline of religious groups. This theory is driven by the assumption that a religious individual who practices or identifies with any particular religion is naturally risk averse, just like a rational consumer who weighs cost versus benefits to make a consumption choice in a free market. A religious group provides particular religious services to maximize the number of current and potential adherents, just like a firm that intends to expand its market share in a competitive business world (Sherkat 1997, 1998; Sherkat and Ellison 1999; Sherkat and Wilson 1995). The basic social process at the heart of religious life is exchanges between buyers and sellers.

Religious economy theory consists of two parts: the supply-side theories focusing on religious regulation and competition as well as the productive capacity of congregations (Finke and Stark 1992), and the demand-side theories focusing on the preference dynamics and individual-level choices (Sherkat and Wilson 1995). The former indicates that religious groups or denominations develop two types of strategies to decrease the risk perceived by consumers. The first type is to endow formal membership and mobilize adherents as a community in order to produce religious service collectively. It requires adherents to input personal resources, such as time and money, formally and continuously. Religious attendance, doctrine studies, and personal connections are all built on and conditioned by salient denominational boundaries and formal memberships. The second type is the opposite. Religious groups separate production from consumption in order to make religious consumption informal and private. Neither the production of religious service nor the consumption of religious goods is constrained by membership (Iannaccone 1990, 1992, 1994). According to the typology given by Iannaccone (1990), institutional and congregational religions such as Christianity characterize the former type, while noninstitutional and nonexclusive religions like Daoism, Buddhism, and folk religions characterize the latter type.

Scholars following the line of religious economy theory find that restrictive religious groups are more successful at gaining and retaining members, which has been proposed as the “strict church” thesis (Finke and Stark 1992; Iannaccone 1992, 1994) The strict church thesis demonstrates that sectarian boundaries prevent free riding and promote collective production efforts, implying more commitment, collectivity, and religious benefits. This makes exclusive and strict denominations better at retaining current members and attracting new ones. Some empirical works support that thesis. For instance, Sherkat (2001) finds that from 1973 to 1982 there was an overall trend of religious mobility from liberal denominations to conservative denominations in the USA. Given this, my hypothesis is that exclusive religions will (1) be more retentive of their members and (2) gain members at the expense of their nonexclusive colleagues through religious switching.

Integrating the framework of religious economy with organizational ecology, researchers ascribe the expansion and contraction of a religious group to its structural location among its peers who make up an organizational population and ecology. Niche overlap impedes the group’s development because it has to compete with its peers for the same target consumer group and similar resources. Individuals find alternatives easily and will circulate among these groups rather than stay loyal. Hence, I expect that (1) religious groups with high distinctiveness will have higher rates of retention, (2) religious groups with little distance from secular society will have lower rates of retention and will lose members through switching, and (3) religious groups similar to a high number of other religious organizations will have lower rates of retention.

Data and methods

The data comes from the 1999, 2005, and 2009 waves of the Taiwan Social Survey (TSCS). Since 1985, the TSCS has tracked social changes in Taiwan. Every 5 years, the TSCS rotates selective modules for religion, family, politics, stratification, lifestyle, culture, and various other topics. By 2014, six waves had been conducted, among which only 1999, 2005, and 2009 provide information about the religious affiliation of respondents and their parents. The sampling procedures first involved deciding on the number of target respondents for each of ten strata of cities and townships, with the number selected proportionate to the population size in those strata. Precincts (the smallest administrative unit in cities) or villages (the smallest administrative unit in rural areas) were then randomly selected for Taipei and Kaoshiung and for other townships within the strata. Finally, 20 registered adults, aged over 18, were randomly selected for a face-to-face interview in each of the precincts. The response rate of the 1999 TSCS is 45% (N = 1925), the 2005 TSCS is 48% (N = 1881), the 2009 TSCS is 43% (N = 1927), and the final data set was weighted to reflect population parameters in Taiwan’s census in the year before the wave was conducted. After deleting missing data, our sample size was 5730.

Dependent variables

Religious affiliations were separated into five categories: (1) No Religion (no religious belief), (2) Buddhism (Buddha worship, Jingtu Zong, Chan Zong, Mi Zong, dual practicing of Chan, dual practicing of exoteric and esoteric teachings, other Buddhism), (3) Daoism, (4) Christianity (Catholicism, Protestant Christianity, other denominations), and (5) Others (self-identified, worships the gods, not clearly specified as such, other religions). All categories were identified by the respondents themselves and clearly indicated by the questionnaire.

Retention and mobility patterns were examined across the five affiliation categories for three periods (1999–2003, 2004–2008, 2009–2013) and four cohort groupings (pre-1944, 1945–1970, 1971–1985, post-1986). The periods were evenly spaced over the decade of the TSCS. Cohorts were defined largely by historical breakpoint. The eldest cohort went through the retreat of Kuomintang and large immigration from Mainland China to Taiwan. The 1945–1970 cohort experienced the urbanization and industrialization of Taiwan at unprecedented speed. The 1971–1985 cohort entered an era of free regulations over the religious market, and the youngest cohort embraced the highly competitive market as well as the rise of new religions.

Independent variables

The two main independent variables were “father’s religious affiliation” and “mother’s religious affiliation,” which had the same categories as the dependent variables. Among parents, folk religion had the largest number of adherents, and Buddhism was the second-most prevalent religion, 21.59% and 24.05% respectively. For the elder generation, Christianity is the least prevalent religion. Only 4.33% of fathers and 4.71% of mothers identified as Christian.

Control variables

Respondents’ education, age, income, and marriage status were controlled in all models because they can confound the relationship between parental religious identity and child’s religious preferences. For example, research found that education improved our comprehensive ability of “complex system of symbols,” changed our understanding of religiosity, and then affected choices about conversion or retention (Pollner 1989, 94). Some ethnic groups assign a religious role to the child and constrain the opportunities for religious choice (Sandomirsky and Wilson 1990). Also, scholars have found that religious affiliation is associated with social status since religious membership signals one’s cultural capital and socioeconomic resources (Alston 1971; Lauer 1975). In addition, marriage increases the likelihood of accepting parental religious identity (Smith and Sikkink 2003).

Since macrolevel factors such as local politics and public religious settings may influence an individual’s behavior at the micro level, the ethnicity of respondents was controlled (Fukienese of Taiwan, Hakka of Taiwan, Mainlander, Aborigine), and the annual statistics data on religious settings was analyzed to ensure there was only a moderate change (results not presented here). To control other factors like institutional and cultural change, the fixed effect of 24 counties of Taiwan was also controlled, which caught all characteristics of these areas invariant with time. The fixed effect of time (1999, 2004, 2009) was also controlled to exclude the potential effect of periodic policies on religious life. Table 1 provides descriptive statistics of the variables.

Table 1 Descriptive statistics

Model

Binary logistic regression was used to examine the father’s and mother’s effect on the child’s religious choice regarding affiliation. As the formula shows, P_ik is the likelihood that child i believes in religion k. The dummy variables father_ij and mother_ij measure if child i’s father and mother believe in religion j; j could be apostasy, Buddhism, Daoism, or Christianity. β_ij and ρ_ij respectively represent the regression coefficient of the father and mother believing in religion j on child believing in k. X_i are control variables, including gender, age, household income, and marriage status. φ_t and μ_c respectively measure the fixed effect of time and county. ε_i is the error term.

$$ \log \left(\frac{P_{ik}}{1-{P}_{ik}}\right)=\sum \limits_{j=1}^4{\beta}_{ij}{\mathrm{father}}_{ij}+\sum \limits_{j=1}^4{\rho}_{ij}{\mathrm{mother}}_{ij}+{\alpha}_i{X}_i+{\varphi}_t+{\mu}_c+{\varepsilon}_i $$

Religious membership by period and cohort

Table 2 presents the religious membership of respondents by period and cohort. Period analysis shows that the percentage of adult child No Religion and Taoists did not change significantly, but that of Buddhists and Christians fluctuated and declined continuously. Compared to 1999, followers of Buddhism and Christianity in 2009 decreased by 24.9 and 24.5% respectively. Rather than stagnating or declining, the numbers of those believing in other religions continued to increase at a high speed. The period distribution of the father and mother shares a similar pattern with the child, that is, a small fluctuation in percentage of apostates and Taoists, an obvious decline in Buddhists and Christians, and a strong increase in believers in other religions. Table 2 also points out that the mother is more likely to be involved in religious groups since the percentage of mothers identifying as No Religion is lower than that of fathers identifying as No Religion over all three periods, which supports the popular statement that females are more likely to be religious than males (Miller and Stark 2002).

Table 2 Membership by denomination, period, and cohort

The different cohorts show divergent religious preferences. First, the cohorts of the 1970s and 1980s had a higher percentage of apostates than their predecessors because they grew up when industrialization, urbanization, and secularization in Taiwan accelerated and the collective level of religiosity declined as a result (Qu 2006). Meanwhile, the vigorous expansion of higher education enlarged the group of No Religion by activating the negative correlation between school education and religious devotion (Guo 2009).

The prevalence of Buddhism shrank significantly among younger cohorts. The percentage of Buddhists significantly declined since the mid-1970s. For those born after 1985, Buddhism continuously lost its attraction. Some scholars have pointed out geographic distribution mattered in this process of decline. Qu (2006) indicates that traditionally Buddhist temples were located in areas whose residents had a higher level of education because local gentry were their main donors. The rites of Buddhism were not as common as the folk religions that existed throughout the rural areas or as pervasive as Christianity, which was widespread throughout the mountainous regions. From 1950 to 1980, the slow growth of total numbers compounded the institutional disadvantage of Buddhism. As a result, the absorptive capacity of Buddhism was gravely weakened.

The cohort change in Christianity was related to the dramatic social transition of Taiwan since the 1950s. From the late 1940s, clergy and believers migrated from mainland China to Taiwan after the civil war. Many aborigines living in the mountainous regions made the next two decades a golden age for Christianity. In the 1950 and 1960 cohorts, the number of followers increased vigorously. However, this golden age ended around the 1970s, characterized by an annually declining number of church members per 10,000 people until 1980 (Qu 2006). On the one hand, the social networks of the followers originally living in mountainous regions lost its influence as urbanization increasingly accelerated domestic migration; on the other hand, the rise of the middle class and the educated, and the reform of evangelical society partially attracted some young people and thus offset Christianity’s loss of the older generation.

The percentage of Taoists stayed at a stable level across cohorts older than 1944. This stability related to the openness of Taoist doctrines, the diversity of Taoist teachings, and the extensive distribution of Daoism rites. Daoism kept its vigor due to its focus on the changing daily life of the ordinary and coexistence with secularity resulting from modernization. The continuity of ancestor worship and related rituals in rural areas also fueled Daoism’s consistent strength.

Cohort analysis shows that the growth of the categories No Religion and Christianity occurred in the 1940s. This coincidence resulted from the first-wave inflow of mainlanders into Taiwan during this period. Although the total number of mainlanders was only 12.1 million (13% of Taiwan’s population) (Lin 2009), they greatly changed the landscape of religion in Taiwan because most of these mainlanders were in either the No Religion or Christian categories. This clearly distinguished them from the natives. Data drawn from TSCS 1999 to 2009 shows that among fathers of the 1940 and older cohorts, the percentage of No Religion reached 37.14% and Christians 20%, which far surpassed the prevalence of these two religions among the whole population. Chen explains why they tended to have no religion: “Previous involvement of mainlanders into folk religions was severed after they moved to Taiwan and left the original geographical and kindred relationship. The cruelty of war also pushed them toward Religious None and atheism” (2002, 130). Qu (2006) and Madison (2012) explain why the percentage of mainlander believers joining Taiwan churches was much higher than the percentage of all mainlanders on the island. They regard immigrants’ psychological need for religious service, the influx into Taiwan of religions professionals speaking Mandarin Chinese, and the preaching strategies taken by churches as the three main factors driving the conversion to Christianity among mainlanders. In addition, the influence of these factors was magnified by the socioeconomic characteristics of the group of mainlanders since their three main occupations were soldier, civil servant, and technician.

Membership retention by denomination

Table 3 depicts the overall trend of religious mobility. First, among all five categories of fathers’ identity, No Religion has the lowest transmission rate (42.62%) and Daoism the highest (92.55%). Although Christianity has a higher level of institutionalization and collectivity than the others, its retention rate is only 58.73%. This does not support the expectation regarding its competitive advantage in retention. Transmission of the mother’s religious identity presents a similar pattern as the father’s, that is, 93.06% of Daoist mothers have transmitted Daoism to their child, but that rate is much lower for Christians. As an exclusive religion that endows clear membership, Christianity transmitted from mother to child is not as high as theorists predicted. Only 61.9% of mothers successfully retained their child within their Christian group, which appears to be lower than those belonging to folk religions.

Table 3 Cross-tabulation of parent and child religious membership

In addition, Table 3 depicts the direction of mobility. Scholars have found that such direction was conditioned by differences between religions along multiple dimensions, such as socioeconomic status, theological doctrines, worship, and rites (Stark and Glock 1968). These differences construct a religious distance matrix whose basic law is that people are most likely to switch to a religion with the smallest distance from their previous one. For example, members are more likely to flow between the Catholic, Episcopalian, and Lutheran denominations than toward Presbyterian, Methodist, or Baptist denominations (Coombs 1964; Doughty and Rodgers 1998).

Table 4 calculates the distance of religion based on data about intergenerational mobility with the method proposed by Parkman and Sawyer (1967). The matrix shows that even if the mother and father share the same religious affiliation, religious distance is highly persistent. For No Religion, the ascending order is Buddhism, Daoism, and Christianity; for Buddhism, the order changes to No Religion, Daoism, and Christianity; for Daoism, it is No Religion, Buddhism, and Christianity; for Christianity, it is No Religion, Buddhism, and Daoism. On average, No Religion is closest to the others and Christianity has the greatest distance. Buddhism and Daoism fall between the closest and the farthest. Based on the matrix, a religious continuity was constructed, embedded within the circumstances of Taiwan.

Table 4 Religious distance matrix

The religious distance matrix shows the general relationship between religious groups at the collective level, and Table 5 further canvasses the matrix by presenting the preference dynamics at the individual level, that is, the religious-belief switch pattern. It portrays the outcome of the structural as well as the exchange mobility of religious affiliations. The diagonal of Table 5¹ indicates that Buddhism has the largest retentive capacity. Only 20% of believers leave this community and switch to other religious groups. The most popular choice for switchers is the category of “Others,” mainly composed of folk religions. This finding partially challenges the hypothesis based on religious market theory that predicts noninstitutional religions have less hold on members and thus lose vitality in the competition with institutional religions. Buddhism not only surpasses Christianity in the respect of retentive capacity but also shows a much greater hold than Daoism and folk religions. Although Buddhism, Daoism, and folk religions are well known for their affinity, Table 5 points out that this affinity decreased after the deregulation of Taiwan’s religious market.

Table 5 Religious belief switch pattern

The martial law system was lifted in 1987. Buddhism in Taiwan gradually broke away from the political framework and the level of organization increased significantly (Jiang 2010). Buddhist organizations emerged quickly, while regional small groups also gradually transformed into large organizations. For example, Fagu Mountain and Zhongtai Temple were both started after martial law was abolished. Without the restrictions and interference of external political control, Buddhist organizations began to flourish and the process of institutionalization sped up. Many organizations used the model of nonprofit organizations to promote their operations, which enabled Buddhist groups to become a legal entity that could exercise its rights and thus expand the legal space for its development. As nonprofit organizations, they enjoyed both reduced public taxes and the greater flow of private funds (Jones 2009). The formalization trend increasingly distinguished Buddhism from other noninstitutional religions. This trend lowered membership turnover and strengthened the capacity for retention. Over 79.4% of Buddhists stay within the religious community and only one fifth become switchers, most of which turned to folk religions, whose pace of institutionalization was the slowest among all religious affiliations.

However, the negative correlation between institutionalization and turnover rate is not tenable for Christianity. Among all five categories, Christianity is the only exclusive and monotheistic religious group in terms of doctrines as well as organizational form. However, its hold on members seems weaker than Buddhism and Daoism. If Buddhism wins believers through expansion of its organizational mode, what is Daoism’s tactic for preventing switchers? The answer may be its advantage over the exchange mobility between generations, as presented in the next part of the article.

Folk religion has the lowest level of retentive capacity. Only 4.05% of previous believers stay within the group, and nearly 96% have switched to other affiliations, among which Buddhism and No Religion become their preferred choices. Over half of the switchers became Buddhists and one tenth joined Daoism, which demonstrates that the boundary between folk religions, Daoism, and Buddhism is much more complicated than the conventional works have proposed. Although Buddhism, Daoism, and folk religions tangled with each other in the Chinese culture, the divergence of these three is manifested in the modern religious market.

Almost half of those in the No Religion category stay in this community, while half flow into other communities as switchers. Surprisingly, with no collective activities or hierarchical system of organizations, the retentive capacity of the No Religion category is almost as strong as that of Christianity. Madison (2012) indicates that urbanization and secularization resulted in the decline of Catholic churches in Taiwan since the 1960s. However, little attention has been given to Christianity in Taiwan after the 1990s. It is possible that the small portion of Christians in the total population and the rapid mobility of the population have loosened the close network of members and the treelike system of churches.

Analytical statistics

Table 6 presents the logistical models of adult children’s religious preferences regressed by paternal and maternal affiliations and controlled by the child’s marriage, gender, ethnicity, household income, and other socioeconomic variables. The reference group is the category of Other Religion. The results show that both paternal and maternal effects are statistically significant (p < 0.01). Children are more likely to inherit their parent’s religious identity rather than selecting a new one. This finding is consistent with the pattern of intergenerational transmission presented in the section on descriptive statistics.

Table 6 Logistic coefficients regressing adult child’s religious affiliation on father’s and mother’s religious preference

Table 6 highlights the difference between the paternal and maternal effects. Overall, the maternal effect is larger, which supports the proposition derived from social learning theory about the positive correlation between daily interaction and religious influence. However, the paternal effect varies with religious affiliation. Compared to their spouse, the father enjoys a greater capacity for passing on Christianity but shows vulnerability in passing on Daoism or Buddhism. How can this difference be explained?

Religiosity has been found to be an important factor in shaping the understanding of gender roles. The literature shows that religious involvement and customs can strengthen male and female stereotypes. Christian tradition emphasizes the father’s priority and responsibility in religious upbringing. Consistent with this tradition, churches in Taiwan also justify the father’s leadership in the religious socialization of the child (Chen 2016). Compared to Christianity, Daoism represents greater gender equality. The relationship between men and women is not hierarchical but dialectical, as yin and yang. Femininity is also manifested in the importance of goddesses in the Taoist lineage of gods. In application, “the spirit of Kun” (kundao jingshen) pays attention to female practice and cultivation, which may strengthen the maternal influence on religious socialization (Liu 2008). In Taiwan, the femininity associated with Daoism is enhanced by bonding with the goddess Matsu, whose relationship with followers was believed to be a reflection of the relationship of mother and son (Su 2011).

In addition, Table 6 reveals that the probability of an adult child believing in Christianity decreases by 88.8% if he or she has a Buddhist mother. The negative relationship of religious transmission between Buddhism and Christianity across generations may demonstrate the overlap of the maternal effect and the religious effect. On one hand, as caregivers, mothers are more likely to make the child take Buddhism and private worship, such as wearing Buddha beads and enshrining Bodhisattva, for granted as part of daily life, and thus unconsciously facilitate the effectiveness of religious socialization. On the other hand, despite its compatibility with other Oriental religions, Buddhism in Taiwan has pursued institutionalization and formalization over the past few decades by increasing the purity of belief and explicit membership, which may explain its greater power of transmission (Hu and Leamaster 2013).

Table 7 shows that the likelihood of religious retention significantly increases when father and mother share the same affiliation. Religious capital can be used to explain this phenomenon. With the same religious preference, the husband and wife can maximize the utility of limited resources in religious upbringing by eliminating mutual competition and making common efforts to cultivate the child’s religiosity. This mode accelerates the accumulation of religious capital for the next generation and thus promotes their religious devotion. In addition, parents sharing the same religious preference may affect the child’s religious identity through the mechanism of religious distance. For example, a child who grew up in a Buddhist or Christian family is more likely to resist other religions because these two religions have the largest distance from the others on average. In a similar way, Daoism does not reject a child switching from other groups, including Christianity.

Table 7 Logistic coefficients regressing adult child’s religious affiliation on parent’s common religious preference

Table 8 reveals the difference in religious retention between daughter and son. Generally, either the father or mother has a stronger influence on a son than on a daughter. Based on empirical evidence in the Western world, many scholars find that daughters are more likely to be influenced since they are socialized to be more compliant and submissive, unlike their rebellious and self-centered brothers (Feather 1980). However, when we bring the context back, we see an inverted pattern. In Taiwan, the traditional culture expects sons to participate in ancestor worship and continue the lineage, and thus, their religiosity gains more attention from parents. In other words, social expectations magnify the parental effect on sons. Hence, the hypothesis that the mother or father only shapes the religious affiliation of children of the same gender is not supported.

Table 8 Logistic coefficients regressing daughter’s and son’s religious affiliation on father’s and mother’s religious preference

Conclusions

Using the survey data of TSCS 1999, 2004, and 2009, this article provides a general view of the landscape of religion in Taiwan and analyzes transmission and switch patterns over the past two decades. The category of No Religion seemed more attractive to the younger generation; Buddhism continuously lost prevalence among cohorts younger than 1970; Christianity had its golden age in the 1950s and 1960s and has started to revive since the 1980s; Daoism developed in a stable way and was distributed quite evenly among different cohorts.

There are both paternal and maternal effects in each religious group, but gender and religious differences matter in the transmission process. Overall, compared with fathers, mothers are better at passing on their religious preference to their child, especially to sons. However, fathers affiliated with Christianity are the exception. They have a stronger influence on both sons and daughters. This complexity of religious transmission between father (mother) and son (daughter) deserves more attention in the future. One feasible explanation is that in Chinese society religious belief and family identity mutually constitute each other. Males and extended families along with the paternal line are the most important unit in collective religious activities (Wolf 1974; Jordan 1972), which means religious life is partially coded by patriarchy and cultural scripts about male successors. It is therefore highly likely that religious retention of sons and daughters differs from what research based on Western experience suggests.

The switch pattern at the macro level points out the divergence in retentive capacity between religious affiliations. Among the five categories, Buddhism had the strongest hold on members and lowest membership turnover. Less than 21% of Buddhist believers switched to other religious groups; most switchers turned to folk religions. Daoism retained its believers quite strongly, though not as strongly as Buddhism. Less than 39% of Daoist believers replaced their previous religious choices with new ones, the most prevalent of which was Buddhism. Folk religions had the lowest level of retentive capacity and the highest turnover rate. Almost 95% of believers turned to other groups. No Religion, Buddhism, and Daoism respectively absorbed 32.09, 51.64, and 10.44% of switchers from folk religions. Considering the switch pattern, the boundaries between Buddhism, Daoism, and folk religions are far more complicated than the conventional literature states. Although all of these religions share common elements, this commonality may ground on textual doctrines rather than rituals or organizational forms. Statistical findings here call more attention to the relationship between these noninstitutional religions in action.

One potential contribution of this article may be the analysis of the switching and transmission patterns of Christianity in Taiwan. Contrary to predictions derived from religious market theory, the exclusive and highly institutional religion does not show an advantage over either structural or exchange mobility since only 56% of believers keep their original beliefs. Buddhism is a more attractive destination than folk religions, No Religion, or Daoism. How can this disadvantage of Christianity be explained? One reasonable explanation is that the continuity with the uniform dimension ranging from conservative to liberal is incompatible with China’s religious landscape. Religious mobility measured or explained by this continuity thus provides only limited insights into understanding the mobility pattern of diversified religions. For Chinese people, religious membership is cumulative and alternative rather than exclusive and restrictive, resulting in a common ambiguity of religious identity at the aggregate level (Yang 1961; Jordan 1993). This accumulation may result in a disadvantage for Christianity.