People of Color, People of Faith: The Effect of Social Capital and Religion on the Political Participation of Marginalized Communities

U.S. immigration policy over the last 100 years has changed the onus of political acculturation from public programs to private groups like churches. After this significant policy change, how do religion, social capital, and nativity intersect in the political mobilization of racial minorities? Furthermore, after the 1965 Hart-Celler Act, the country of origin of immigrants shifted from European countries to Latin America and Asia. Scholars have theorized that churches play a pivotal role in the socialization of immigrants by providing a place of belonging and a community willing to teach newcomers about the goings-on of American political society. How have these acculturation policies worked under new immigration populations? Previous scholarly work has connected social capital with churches, though their relationship to political participation has been minimal. We hypothesize that social capital and religious tradition have a multiplicative effect on the participation rates of believers, but that race mitigates that effect. The positioning of racial groups in broader society impacts the significance and role of churches within these communities. We use Collaborative Multiracial Post-Election Survey (CMPS) 2016 data to examine the connection between social capital, religion, and political behavior in a novel attempt to systematically identify the unique role of churches in the mobilization of racial minority communities. We use these results to suggest that the current policies of privatizing political acculturation have had less success with more recent waves of immigrants.

the relationship between religion and politics still hold true. Specifically, we consider whether places of worship augment community members' social capital, leading to increased political participation of congregants.
Churches have long been noted for being the primary source of political socialization for immigrants. The process of socialization in the 19th century was much clearer than it is today-immigrants were approached the moment they stepped off of the boat from Europe by church groups holding bibles and offering a place to live, jobs to earn money, and-importantly-encouragement to vote and engage with party machines (McAvoy 1969;Cogley 1973;Dolan 1977Dolan , 1992. In this historical environment, the process of political socialization for immigrants-and many subsequent generations-was easily identifiable, though not quantified by modern standards. Today the intermingling of churches and political parties is less blatant, but churches are still assumed to be sources of political socialization and engagement, particularly for immigrant and ethno-racial minority groups. Scholars have yet to conclude if churches are having the same, if any, effect on these new groups (Jones-Correa and Leal 2001;Wald and Wilcox 2006;Levitt 2008). Furthermore, the mechanism of that socialization is less established. Most established theories emphasize the importance of denomination, devotion, or civic skill development, although few (if any) have considered all simultaneously and few clear patterns have been identified.
It is possible that religion works to promote political learning and engagement instead because it is a source of both identity and significant political cues (Wilson 2008). This understanding, then, begets a consideration of race as well, given that race and religion are co-constituted, meaning they are "defined, in part, by each other, both in their meanings and in their political effects" (Jacobson and Wadsworth 2012, p. 4). If religious identity is a potential explanation for political socialization and mobilization, then so too must we consider the impact of social capital, which is a measure of one's dedication to and relationship with their community-and those who help provide their identity (Alba et al. 2009;Gordon 1964;Haddad et al. 2003;Lin 2001). Furthermore, places of worship have been argued to be the largest repository of social capital in the U.S. (Gordon 1964;Brady et al. 1995;Putnam 2000), and social capital has consistently been found to promote civic engagement (Wong et al. 2011;Djupe and Calfano 2012;Ocampo et al. 2018).
Here, we investigate whether a more complete rendering of the relationship between religion (that includes denomination and devotion as well as race, nativity, and social capital) and political engagement will provide clearer cues about the process of socialization through churches. The literature offers few clear clues as to expectations save one: that increased social capital correlates with higher levels of political participation. Given the foundation of social capital in churches, we theorize that social capital is the mechanism that "turns on" the religious variables. We also find that social capital is a positive indicator of political participation for all racial groups, including immigrants. But when we moderate this affect with religious affiliation, immigrants and congregants of color are less likely to participate politically. Finally, we find that respondents who expressed high levels of devotion to their faith and faith communities-even if they were infrequent attendees-demonstrated positive correlations with increased political participation if they perceived religious discrimination. These results question the extent to which traditional religious indicators, such as church attendance, are effective measures of religious devotion and predictors of political participation, particularly among non-white believers. We first test the correlation between frequent attendance and political participation in order to examine the effectiveness of traditional theories to predict religion's impact on political participation of non-white communities. Our second set of models test our theory that religion may "turn-on" social capital for already religious respondents by examining the effect of religious tradition and social capital on political participation. Our third set of models controls for both frequent attendance and religious tradition, mediated by social capital in order to test both traditional religion and politics models against our theory of social capital's relationship with religion and political participation.
We take advantage of the large over-sampling of Black, Asian, and Latino respondents in the 2016 Collaborative Multiracial Post-Election Survey (CMPS) dataset to conduct a nationally representative examination of these three traditionally neglected groups. Such robust samples allow us to examine each group separately to identify any variations in their relationships between religion, social capital, and political participation. What we find is that social capital is the only reliable predictor of political participation across racial groups, while denomination, devotion, and nativity are insignificant or negatively correlated with participation. This indicates one of two things: that religion does not relate to political outcomes for non-white populations, a dubious conclusion given the wealth of data and history demonstrating the opposite. The second possibility is that our current framework for assessing the role of religion is inadequate for this new wave of immigrants and needs to be reconsidered to more directly address the function of churches and faith in these communities.

The Ongoing Political Relevance of Churches
Despite a noted decline of religious affiliation and attachment, a full 75% of the U.S. population continues to identify with established religions (Pew Research Center 2014), and most attend religious services anywhere from a few times a year to once a week, signifying that religion continues to be significant to a clear majority of the U.S. population.
Moreover, the importance of religious signaling in political preferences also has not abated, and nor has religious signaling by political parties and candidates decreased. Religion in politics is still salient, specifically to the level of political and policy support from religious groups (e.g., white evangelicals). The media interviewed White Evangelical leaders throughout the 2016 election about their presidential vote choice and offered their perspective of how their faith and religion influenced the balancing out of options (Whitehead et al. 2018;Margolis 2020;Martí 2019). Furthermore, religious institutions are still making public political statements and mobilizing their followers. The Catholic Church has a long history of such mobilization, most notably in its support of immigrant rights (Barreto et al. 2009;Ramirez 2013), and the annual March for Life protests in Washington, D.C. Lastly and most notably, Black churches have a long history of educating their congregations and facilitating explicitly political engagement (such as organizing drivers to the polls for early voting) that continues to this day (Barnes 2006;Calhoun-Brown 1996;McKenzie 2004;Warren 2001). All told, religion remains a central component of American politics, warranting continued study to understand how it functions within our increasingly diverse communities.

The Theories of Religion and Politics
Religion is a central component to understanding political outcomes because "Americans do not hermetically seal off their religious identities from their political attitudes and behavior" (Wald and Calhoun-Brown 2007, p. 37). As such, scholars have historically focused on the relationship between common religious indicators (religious tradition and devotion) and political outcomes with mixed results (Carter 1994;Conkle 1993;Kaufmann 2004). We argue this is because such analyses may oversimplify the significance of religion. A more thorough investigation will account for the varying significance of one's church and church community to one's daily life, variation that can be accounted for through one's race, nativity, and levels of social capital. This theoretical approach logically follows from the development of religion and politics literature to date.
Researchers initially focused on the significance of religious tradition (Herberg 1955), but soon grew to include what they termed "the Religious Factor", of a religious tradition, involvement, and orthodoxy that impacts not just faith, but all parts of life (Lenski 1961). Vernon (1962) argued that, while religiosity was significant, it could not be reliably measured by attendance or other behaviors because such activities do not connote the same strength of devotion amongst various believers. Others noted that it has been difficult to accurately assess the impact of religion because most surveys measure beliefs poorly, which then precludes precise analysis. Green (2007) theorized an improved measure of religion using the three B's (Belief, Behavior, and Belonging) in order to address the complexity of religious devotion and the various ways it can be expressed. Verba et al. (1995) expanded the discussion to include a consideration of the mechanism for religion's positive influence on politics: churches offer an opportunity for congregants to learn civic skills that can facilitate and increase the rate of successful political engagement. In this way, more frequent church attendance became a proxy for exposure to and opportunities for civic skill development. Catholics, coming from a perceived hierarchical religious tradition, were thus deemed disadvantaged in comparison to Protestants who were thought to have more opportunities for leadership and participation . However, other studies found that Catholicism had a more positive correlation with political engagement precisely because of their frequent offering of civic engagement opportunities and organizational support to their communities (Barreto et al. 2009;Beyerlein and Chaves 2003;Kurien 2007;Menjivar 2003;Mora 2013). Peggy Levitt's investigation of immigrant communities in Boston advanced the understanding of churches as skill developers. She found that "even when religious institutions did not have explicit political agendas, people learned about fundraising, organizing, and leadership by participation, which they applied to other settings" (Levitt 2008, p. 778). In spite of these conclusions, there has been little consensus on how to systematically measure such exposure to political opportunities or the learning of civic skills through churches beyond church attendance (perhaps in part due to the infrequency of such pointed questions on large, national surveys). We agree that skill development is an important component to religion's influence on congregants' political engagement, but that not all believers will experience these benefits equally. We argue that race and social capital moderate religion's influence on political participation.

Religion and Immigrants
The significance of the Black church to the political participation of Black communities is mirrored among immigrant groups, who rely on churches as their central socializing institution (Gordon 1964;Haddad et al. 2003;Alba et al. 2009). As with communities of color, immigrants suffer a dearth of attention from traditional, formal institutions such as political parties (Wong 2006;García-Castañon et al. 2019), thereby increasing the ability of churches to influence immigrants' policy preferences, civic engagement, and willingness to participate politically (Gordon 1964;Jones-Correa and Leal 2001;Putnam and Campbell 2010;Djupe and Neiheisel 2012). This is because churches are trustworthy sources of information, which can be important when they do not know where else to turn for help with finding a school for their children, looking for work, or asking questions about immigration documentation (Cadge and Ecklund 2006;Cadge and Ecklund 2007;Putnam and Campbell 2010;Martinez et al. 2012). This trust facilitates mobilization, as we saw during the 2006 Immigrant Rights Marches (Barreto et al. 2009;Ramirez 2013;Wong 2006). Furthermore, the relationships among congregants is also a source of information and mobilization for immigrants, who learn a great deal about contemporary political issues and how the government functions through the casual conversations after religious services, group meetings, or other gathering times (Djupe 2009;Neiheisel et al. 2009;Djupe and Neiheisel 2012). Many scholars have noted the number and level of civic skills that immigrants learn by participating in and organizing church activities Jones-Correa and Leal 2001). Such political learning takes place "even when religious institutions [do] not have explicit political agendas," allowing immigrant congregants to learn "about fundraising, organizing, and leadership by participation, which they applied to other settings" (Levitt 2008, p. 778).
As of yet, there has been little research to investigate if and how this outcome varies by levels of social capital even though its presence is implicit through the creation of, reliance on, and contribution to immigrants' church communities. This investigation seeks to fill this gap, testing whether the theories about the relationship between places of worship and political participation are applicable to the newest wave of immigrants from outside of Europe and their descendants who make up communities of color in America.

Race, Ethnicity, and Churches
It is not possible to fully understand churches' role in promoting political engagement without acknowledging how race, a "central axis" of American society (Omi and Winant 1994), fundamentally changes the interaction of marginalized groups with religious and political institutions. Furthermore, many scholars have shown how race mediates the likelihood and quality of interacting with religious institutions (Maiden 2019; St. Jean 2016; Bracey and Moore 2017). Rev. Martin Luther King Jr. once said, "it is appalling that the most segregated hour of Christian America is eleven o'clock on Sunday morning", a truism that still holds. Racial and ethnic groups have distinct theologies, styles of worship and church government, and beliefs about the role of faith in a society that preclude widespread cohesion, even within religious traditions. In other words, religion is the "major way that a group communicates its culture" (Wald and Calhoun-Brown 2007, p. 283). Furthermore, "the meaning and the political effects of racial categories have often been channeled through religion" (Jacobson and Wadsworth 2012, p. 3) such that race and religion are co-constituted (Jacobson and Wadsworth 2012, p. 4; see also Oskooii 2018). This means that religious traditions may have distinct patterns of political socialization conditional on race, which would then also impact the role of devotion and church attendance on the political outcomes of that particular religious community.
Indeed, research has come up with varying conclusions about the relationship between religion and politics for different racial groups, though often this is due to differing emphases. For example, it is still unclear whether Catholicism can in part explain Latinos' lower rates of political participation, as theorized by Verba et al. (1995). They argued that church attendance only increased Protestant political participation, and so Latinos, being mostly Catholic, experienced little political benefit from their devotion. Jones-Correa and Leal (2001) also found that church attendance had a positive impact on political engagement, but that it impacted Latino Catholics as much as Latino Protestants. This investigation was hardly the last word, as DeSipio (2007) found that Latino Catholics vote at higher rates than non-Catholics, making it impossible to conclude that religious tradition has a clear directional impact on Latino political engagement. Researchers have not found a difference in political participation by religious affiliation for whites (Jones-Correa and Leal 2001).
Studies of Asian communities and religions can be difficult given the establishment of religious communities based on national origin, rather than a pan-ethnic identity. However, some sectors of conservative evangelical Protestantism have succeeded in creating "Asian American niche markets," particularly among college students (Busto 1999;Kim 2004). The growth of these organizations also led to the emergence of pan-Asian Protestant congregations whose ranks are filled mostly by second-generation Asian Americans of various nationalities (Jeung 2005). There is no other religious group that employs the tactic of attaching a pan-ethnic label to its religious clientele, potentially creating a new Asian American evangelicalism (Park 2008;Lien 2010;Wong et al. 2011). Such conclusions are bolstered by the fact that being Born Again does not appear to result in a conservative political agenda (Republican affiliation, belief in and prioritization of anti-abortion policies) for Asians in the same way it does for whites (Wong 2015), demonstrating the unique relationship between race, religion, and American politics.
In contrast, studies of Black religions tend not to emphasize religious tradition and devotion, but have clearly established the Black church's historical role as a vital political resource to the African American community (McKenzie 2004;Tate 1991). The Black church itself is thought to be explicitly political, given its foundation in response to slavery and segregation. Furthermore, Black churchgoers have been found to be more likely to attend church meetings on political issues than those of other racial groups, perhaps due to the church culture that teaches "about their obligation to the larger community and to their own people" (Wald and Calhoun-Brown 2007, p. 288; see also Harris 1999;Brown and Brown 2003). Research has also emphasized the significance of political discussion among congregants and messaging from the pulpit to increases in Black political participation (Wald and Calhoun-Brown 2007;McKenzie 2004;Brown 2011).
Previous research has explained the relationship between religion and political engagement for each ethno-racial group, though each body of work has emphasized a different aspect of religion or politics. Here, we contribute by completing the same analysis for each racial or ethnic group-examining not only the role of denomination and devotion, but also social capital.

Churches and Social Capital
Social capital is an understudied factor in the race-religion-politics relationship. It is defined as the "social networks and [their] associated norms of reciprocity" necessary for the functioning of government (Putnam 2000, p. 21). Alternate definitions recognize the meaning provided by such relationships that then motivates individuals to act for their benefit (Granovetter 1995;Nakhaie 2008). Congregational Social capital includes accessing the social networks of places of worship for potential goal-oriented action (Schneider 1999). Social capital is an established component to political participation; it has consistently been found that those who engage with community or volunteer organizations (including churches) have higher rates of civic and political participation (Wolfinger and Rosenstone 1980;Verba et al. 1995;Tossutti 2003;García-Castañon et al. 2019). It then makes sense that Putnam argues churches are the "single greatest repository of social capital" in the U.S. (2001), as places of worship often organize volunteer opportunities and offer prime (and ongoing) sources of both material and emotional support to members, allowing for the development of relationships and a community, which form the backbone of social capital (Manzano 2007;Brown and Brown 2003). The higher levels of trust created from these beneficial activities direct worshippers to act within the norms of their religious tradition even outside of that context, like in the context of politics (Cnaan et al. 1999).
Connecting churches with social capital for distinct racial groups is not new. Mosqueda (1986), found Chicanos valued the compadrazo often found in churches, even though their culture and relationships were not dependent on the church or formal Catholicism. This connection suggests that churches may facilitate the creation of community and social capital that can positively contribute to political participation but is not necessarily the result of devotion or any specific religious affiliation. Leal et al. (2016) ventured a related hypothesis to explain their uneven findings regarding the relationship between church attendance and political participation: such focus masks whether "some immigrants join churches that promote a bridging social capital that promotes greater civic and political engagement, while others become members of isolated congregations and consequently experience a form of segmented religious assimilation" (Leal et al. 2016, p. 136). Examinations of the Black church have found that churches "empower their members politically by increasing social capital" (Wald and Calhoun-Brown 2007, p. 288; see also Brown and Brown 2003;Harris 1999).
Although rarely explicitly stated by scholars, immigrants rely significantly on churches as a source of social capital, particularly in highly concentrated immigrant communities. Immigrants come to rely on each other (often through churches) for the myriad of informational, social, emotional, and material needs they have. This process of resource exchange facilitates the creation and maintenance of relationships that result in social capital formation (Ebaugh and Chafetz 2000;Richman 2005;Stepick et al. 2009). Other scholars have noted the importance of bridging social capital-connecting immigrants with the native-born congregants in their church-as the key to promoting immigrant political participation (Levitt 2008;Djupe and Neiheisel 2012).
Churches also contribute to social capital creation through their participation in broader, formal organizational efforts, founding, or joining groups like the Industrial Areas Foundation (IAF) and Communities Organized for Public Service (COPS) (Warren 2001;Hinojosa 2014). These are community-based grassroots organizations that not only encourage voting, but also broader engagement to solve community problems, run for office, and build cross-racial cross-denominational alliances, allowing for the creation and development of civic skills that can have a direct and positive impact on members' ability to participate politically (Warner 2001;Putnam and Campbell 2010;Wilson 2008). Importantly, these groups operate within or among churches, and, in many cases, are exclusively in the service of communities of color as populations marginalized outside of the political mainstream. By creating a broader sense of kinship, increasing both positive political outcomes as well as feelings of political efficacy among members, such organizations are able to facilitate the creation of social capital specifically due to the foundational role churches play in those communities.

Theoretical Framework
This project works to identify whether religious denomination and devotion, when moderated by social capital, affects levels of political participation among communities of color. We expect these results to vary by race given the co-constitution of religion and culture. Furthermore, we account for the real-life variation in the role places of worship play their attendants (Brown and Brown 2003). This concept is already broadly accepted when considering immigrant populations, who are theorized to rely more heavily on their church communities than native-born congregants (McLellan and White 2005;Mitra 2012;Ley 2008;McAndrew and Voas 2014). We test whether this is the case, and whether race and ethnicity further vary the relationship between religion and political outcomes. Moreover, significant to understanding religion's influence is a consideration of social capital, which impacts attendees' willingness and ability to participate politically. Churches, as one of this nation's "largest repositor[ies] of social capital" (Putnam 2000), can then also be considered one of the most significant contributors to American political socialization and engagement. Indeed, scholars have long lauded this role, particularly in the socialization of immigrant groups (Cao 2005;Hirschman 2004). We argue that race changes the socializing influence of churches and religious belief. Communities otherwise neglected by dominant political and social institutions may rely on churches more for their socialization and political development, particularly for the devout who are already more trusting and willing to invest in their church community. This creates an interactive effect that can easily be missed when considering issues of devotion, affiliation, or levels of social capital individually.
Prior research shows that churches can successfully motivate political activity through teaching civic skills, encouraging political action, and increasing knowledge of the issues through discourse, although it appears that attendance alone is an insufficient condition for reaping those benefits. Instead, we argue that congregants with higher levels of social capital are more likely to be the beneficiaries of such positive political externalities through church attendance. This is because social capital is a measure of one's investment in and dedication to one's community. Greater investment indicates stronger relationships; this would increase the influence of both clergy and congregant discourse (McKenzie 2004;Brown 2011;Neiheisel et al. 2009), receptivity to mobilization efforts, and the likelihood of volunteering, which would result in the learning of civic skills (McClerking and McDaniel 2005;Brown and Brown 2003;Verba et al. 1995;Levitt 2008). We theorize that this relationship is not based in frequent attendance, alone, but is conditional on the presence of social capital. 1 Hypothesis (H1). Frequent Attendance 2 to faith services, conditional on social capital, correlates with increased levels of political participation. 1 We assume a moderating effect because our data-generating process assumes a moderating effect, which we found limiting and have attempted to account for with multiple models. First, we would argue that a mediating effect assumes a temporal order (a→b→c) that, while we agree is important, our data limit us from disentangling. Because respondents are asked about their level of social capital and religious preferences at the same time of the survey, we can neither assume nor model whether their social capital level caused or was caused by their adherence to a religious tradition. Furthermore, after a lengthy search, we found no questions in the survey that might help us control for this temporal order. Therefore, we must assume these variables are concomitantly present and model our analyses as such: that respondents who currently adhere to a religious tradition also have a current level of social capital (a ∪ ∆b→c). We attempt to address this moderating versus mediating affect by using three different models to test the relationship between social capital and religious tradition on political participation. We also include a model in the appendix of social capital on political participation with standard errors clustered around religious tradition.
2 A basic correlation test shows an insignificant negative correlation between the social capital variable and the frequent attendance of −0.0218.
Past research presents contrary conclusions about the relationship between denomination and political outcomes. We consider whether clearer patterns may emerge when taking into account variation in both racial and religious diversity in the U.S. The disagreement between Verba et al. (1995) and Jones-Correa and Leal (2001) about the role of Catholicism on political participation, and Wong's (2015) study of born-again Evangelicals highlight the importance of race in determining the relationship between religion and political participation. The lack of consensus on the relationship between religion and political participation may signal that previous models, though providing important findings, may have been underspecified. Social capital as a moderating influence may explain which religious traditions correlate to higher rates of political participation for non-white and immigrant groups.
Hypothesis (H2). Religious affiliation, conditional on social capital, correlates with increased levels of political participation.
We also address the significance of nativity to political outcomes. While the importance of churches to immigrant political socialization has been well-established historically, we consider whether this remains the case in the present day. Immigrant groups tend to rely on churches for a myriad of services and support; those who come to churches more frequently for assistance may also participate in other groups at a higher rate, which would work to develop social capital. As such, we theorize that religiously affiliated immigrants with higher levels of social capital will be more likely to engage politically than their unaffiliated and low social capital counterparts. Further, this relationship will be stronger than it is for native born congregants who have access to a greater number of services and support groups than do immigrants. Here we include both religious tradition and attendance of religious services because social capital may require identification with, participation in, and dedication to a particular community in order to "turn on" the positive effects of religious affiliation.
Hypothesis (H3). Religiously affiliated immigrants with higher levels of social capital will participate politically at higher rates than both their native-born counter parts, and immigrants who are religiously unaffiliated.

Data and Methods
We use the Collaborative Multiracial Post-Election Survey (CMPS), a national survey conducted from 3 December 2016 to 15 February 2017. A total of 10,145 completed interviews were collected online via a self-administered format. The adult sample included non-registered voters and non-citizens of white, Black, Latino, and Asian race and ethnicities. 3,4 This survey contains an oversample of non-white respondents as well as foreign-born respondents, increasing the precision of our findings. While religion is not the central focus of this survey, we use it because it oversamples racial and ethnic groups as well as noncitizens. It allows us to make inquiries into racial, immigrant, and religious groups usually overlooked due to the small sample size of non-white respondents in other studies focused on religion.

Measuring Participation, Religion, and Social Capital
Our primary dependent variable measures respondent political participation. We created this variable from a battery of twelve questions regarding respondent engagement in both formal and informal political activities in the past year, such as discussing politics with their family or contributing money to a candidate (Cronbach's α = 0.8) (Porter 2016;Siegfried et al. 2018;Ciftci 2018;Shamaileh 2019) 5 . We collapsed this variable into three levels of participation: low participation (zero to three activities), medium participation (four to eight activities), and high participation (nine or more). Collapsing these variables allows us a more concise measure of the overall level of engagement of each respondent without artificially valuing the significance of one activity over another and still measuring some level of political participation for comparison. Including informal modes of participation also allows us to more fully account for non-traditional modes of engagement that are easier for non-citizens to access.
Our primary predictor variables concern religious tradition and social capital. Our measure for social capital is a dichotomous variable from the question, "Do you participate in one or more than one social, cultural, civic, political group or union or do you not participate in the activities of any such groups?" 6 Potential responses were "none", "one", or "more than one". We coded "none" responses as 0, and "one", or "more than one" responses as 1. No social capital is the reference category in all regression analyses. Because our measure of social capital considers membership, we consider this a conservative proxy for the skill development and access to resources that are offered through community-based organizations (La Due Lake and Huckfeldt 1998; Todd and Houston 2013; Wong 2006). 7 We include measures for the six most common religions both in the U.S.: Catholic, Protestant/ Christian, Muslim, Hindu, and Buddhist. 8 Born-Again Christians include those who identify themselves as both Protestant/Christian and Born-Again; the two categories are mutually exclusive. We interact each of these religious traditions with our social capital variable in order to measure the combined effect of religious affiliation and social capital.
We measure devotion by frequency of attendance to religious services. This is coded to be a dichotomous variable, where "one" represents weekly attendance of faith services or more. Discussed politics with family and friends? 2.
Worked for a candidate, political party, or some other campaign organization? 3.
Contributed money to a candidate, political party, ballot issue, or some other campaign organization? 4.
Wore a campaign button, or posted a campaign sign or sticker? 5.
Contacted an elected representative or a government official in the U.S. in any way-such as through writing a letter, emailing, calling, or in person-about a policy or issue you care about? 6.
Contacted in any way, such as by letter, telephone, internet, or in person, a government office about a problem you have or to get help or information? 7.
Worked or cooperated with others to try to solve a problem affecting your city or neighborhood? 8.
Attended a meeting to discuss issues facing the community? 9.
Discussed a candidate or political issue on social media like Facebook, Twitter, WeChat, Kakao Talk, LINE, Zalo, or Babble? 10.
Attended a protest march, demonstration, or rally? 11.
Signed a petition regarding an issue or problem that concerns you? 12.
Boycotted a company or product for political reasons? 6 We recognize this measure is limited concerning the type of social capital at work here, specifically bonding or bridging. We hope this preliminary investigation will lead future scholars to overcome this initial data limitation by including interactions to measure the different features of social capital (see Manzano 2007). 7 While it is possible that our variable also captures the trust or friendship that may result from such organizational membership, we do not make that assertion. 8 Additionally, in observance of the Central Limit Theorem, we dropped religious traditions that had less than 50 observations in each racial category to reduce the possibility of a type I error (false positive). We dropped Latino respondents who identified as Hindu, Muslim, or Buddhist. Further, we dropped 43 Latino traditional Protestants in the dataset, as minimal variation in the social capital of led to inflated and unreliable coefficients. We also dropped Black respondents who identified as Hindu or Buddhist as these grouped observations totaled less than the required minimum number of observations for normally distributed data to have reasonable confidence bounds on the variance estimate. 9 The CMPS does not ask respondents about their belief in Jesus Christ or their interpretation of the Bible, and so such measures or their proxies are not included in this analysis. Given our emphasis on religious tradition, inclusive of non-Christian religions, we are unconcerned by the absence of such Christian-specific measures in this project.
This operationalization allows us to make a distinction between respondents that are devoted to their religious communities, versus those who attend only on holidays or from other types of familial obligations that would result in less frequent attendance 10 . Our study assumes racial groups have had distinct experiences with the American political system resulting from America's embedded racial hierarchy and that this relationship can impact political engagement (Bowler and Segura 2012;Masuoka and Junn 2013). We want to ensure that this mechanism is not being masked nor misattributed to another variable in our analysis. As such, we include a dummy variable measuring if respondents ever perceived racial or ethnic discrimination. Perceived discrimination is related to but distinct from linked fate, which measures the extent to which a respondent feels personally impacted by the well-being of her racial group (Dawson 1994). While linked fate is traditionally associated with Black American political participation, we include it to compare full models across all racial groups, and verify its relationship (or lack thereof) for other communities 11 .
In order to more fully account for the possible ways that religion may motivate political participation, we include two other less traditional measures. The first is perceived religious discrimination, which we measure with a dichotomous variable. Religious discrimination could potentially motivate political participation in the same way as racial discrimination and is something that may be felt by those who are more devout. We also control for whether the respondent's political views are similar to their church's political views. Mobilizing congregants may be easier if they already agree with their church's political rhetoric. Lastly, we include a dummy variable for a respondent's nativity, along with other standard controls: ideology (using the variable "liberal"), respondent interest in politics, gender, wealth (via homeownership), education, and age.
Tables 1-3: Summary Statistics for Asian, Black, and Latino Respondents. 10 Running these models with full ordinal frequent attendance variables resulted in similar results. We include a dichotomous variable here for ease of interpretation and explanation. 11 While over two-thirds of respondents identified themselves with a religious tradition, less than half of respondents answered questions about the extent of religious discrimination they have experienced and so were dropped from our analysis.

Disentangling Immigration, Race, Religion, and Social Capital: Our Methods
We measure political participation with a three-category ordinal variable. As such, we used an ordered logit model (Borooah 2002) but also ran an ordered probit as a robustness check, which yielded similar results 12 . We present the odds ratios of our models to facilitate a discussion of the regression coefficients and include the z statistics (the number of standard deviations the coefficient lies from the mean) to simplify comparison across models. In order to identify the unique relationship between religion, social capital, and political participation, we ran separate regression models for each racial 12 We also undertook a negative binomial regression model to test an additive index of all 12 indicators. We use a negative binomial regression because the dependent variables in this model are counts (total number of political participation activities) and the model is over-dispersed as the variance is greater than the mean. We did not use a Poisson regression because Poisson models assume that the variance is equal to the mean. The negative binomial model offers similar results in this to our ordered logit model in Table 6 with one addition. In the Negative binomial model using the indexed dependent variable, we were able to detect a small but statistically significant relationship between Black Born-Again Christians with no social capital and an increase in political participation. Black Born-Again Christians with no social capital were 26% more likely to have a higher level of political participation relative to Black respondents with no social capital who did not identify as Born-Again Christians (p < 0.05).
group (Asian, Black, and Latino). Dividing the models by race also allows us to accurately account for the racial variations across the values of our control variables (Masuoka and Junn 2013). We present three sets of models from our most parsimonious model to the most complex model to show the results as we add control variables. The first set of logit regressions model the effect of frequent attendance and social capital on political participation with robust standard errors clustered around religious tradition. We interact our social capital and attendance variables because we expect that social capital initiates the positive influence of religion. However, it could be that individuals attend religious services, but do not engage with other congregants, thereby eschewing the opportunity to create social capital.
Our second set of models address the impact of religious tradition, specifically on political participation with robust standard errors. While religious tradition alone has been shown to be an unreliable predictor of political participation, we argue that religious tradition positively relate to increased political participation when partnered with social capital-a hypothesis that we test directly using an interaction term (Duriez et al. 2002;Todd and Houston 2013;Lewis and De Bernardo 2010).
The third set of models accounts for both behavior and affiliation in order to more completely model the ways that religion may help to create social capital, and thereby positively impact rates of political participation. Much in line with past research, we find that there are few consistent trends by levels of devotion, religious tradition, ethno-racial group, or even nativity. While there is a strong tendency to omit non-findings from publications, we wish to be a part of a growing body of researchers who acknowledge the importance of non-findings to the scientific process (Franco et al. 2014;Leal et al. 2016;Mervis 2014). Instead of negating the importance of religion and religious communities, we believe that our results further extol the need for a re-evaluation by social scientists of the relationship between places of worship and political participation.

Testing the Effect of Behavior on Political Participation
Our first hypothesis concerns religious behavior-specifically the relationship between attendance of religious services, social capital, and political participation. We find that frequent attendance, in and of itself, does not affect participation, in line with past theories. Social capital, alone, is a positive indicator of political participation for all racial groups. However, our hypothesis posits that the positive relationship of frequent attendance to political participation is conditional on social capital. We did not find this to be the case, as frequent attenders with social capital are not statistically significantly associate with higher levels of political participation low-attending, low social capital counterparts (Table 4).
We pushed the theoretical bounds of our models to account for the various ways religious devotion could be related to political participation. To this end, we included a measure of perceived religious discrimination, a variable that would only achieve significance is one's religion were important to them. Here we find that perceptions of religious discrimination are positively correlated to increased engagement, though only for Latinos. While far from conclusive, this offers preliminary support for the argument that religious attendance may be an ineffective or incomplete measure of religious devotion, particularly when examining non-white or post-1965 immigrant believers (Vernon 1962).
Further variations by racial group emerge. Linked fate, traditionally found to positively correlate with Black political participation, only, here also demonstrates a positive relationship for Asian respondents, as well. Furthermore, only Asians who report co-ethnic discrimination are significantly less likely to participate. Neither variable is statistically significantly related to Latino levels of political engagement. Robust se Eform in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.

Testing Religious Tradition on Political Participation
Our second hypothesis considers the relationship between religious tradition, social capital, and political participation. In these models shown in Table 5, we find that, similar to our analysis of attendance, the significance of tradition alone varies greatly by race. While Black Catholics are more than twice as likely to participate politically compared to their non-Catholic peers (p < 0.05), Latinos who are Catholic or Born-Again are 42% and 50% less likely to politically engage (p < 0.01). Similarly, Asian respondents who are followers of Hinduism are almost eighty percent less likely to participate in politics relative to their non-Hindu peers (p < 0.01). Interestingly, in we find no statistically significant relationship between traditional Protestantism and political participation for any racial or ethnic group. This result contradicts the literature that argues the Protestant tradition has a stronger relationship with political participation than other religious traditions ; but see footnote 9). Robust se Eform in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.
It is important to note that these prior works did not control for social capital, which we find to be consistently positive and significant across all categories of respondents. The presence of social capital increased the likelihood of political participation seven-fold for both Black and Asian respondents as compared to their no social capital peers (p < 0.001), and three-fold for Latinos (p < 0.001). However, we find no significant relationship between our interactive terms and political participation, disproving our second hypothesis.

Testing, Belief, Behavior, and Nativity
Our third hypothesis is that religion mediated by social capital be significantly correlated with political behavior among foreign-born respondents. We theorize that given immigrants' increased reliance on churches for support and information, in combination with the neglect shown them by more formal institutions would correlate significantly with political participation. We test this hypothesis by presenting a set of models that control for attendance, religious tradition, and nativity and present the results below in Table 6. Being foreign born remains significantly negatively correlated to political participation for Black respondents in all models. Black foreign-born respondents are 62% less likely to politically participate compared to their native-born, non-Black peers (p < 0.001). The nativity variable has been significant and negative for Black foreign-born respondents in the all three sets of models, suggesting that first-generation Black respondents may have different patterns of religious affiliation and political participation than native-born of Black churchgoers.  Robust se Eform in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1.
The lack of a significant coefficient for Asian and Latino respondents who are foreign born demonstrates the importance of considering race and ethnicity in these analyses. It is possible that there is a longer history of Asian and Latino immigration that negates the negative impact of foreign-born status on participation. Having mixed-status communities facilitates the political socialization of immigrants, particularly when given opportunities to participate through religious organizations. Examining the relationship between places of worship and immigrant versus mixed-status communities will require more targeted research of these traditionally marginalized and overlooked populations.
The significance of religious identification increases in this full model. Asian respondents who experienced religious discrimination are almost twice as likely to participate than Asian respondents who claim no religious discrimination (p < 0.05). Latinos who claimed religious discrimination were almost sixty percent more likely to participate more in politics than Latinos who claimed no religious discrimination (p < 0.05). This again highlights the importance of exploring more nuanced ways of measuring religious identification and devotion.

Discussion and Conclusions
We theorized that accounting for social capital would facilitate a more complete understanding of religion's relationship with political participation for communities of color and immigrant groups. We first hypothesized that frequent attendance, thought to be a proxy for civic skill development, would more reliably predict political engagement given higher levels of social capital. The presence of social capital would theoretically indicate a stronger involvement with one's church community, and more opportunities to develop said civic skills. Instead we found no evidence for this hypothesis. While social capital, in and of itself, was consistently and positively significant for all racial groups, our attendance-social capital interaction term was not significant in any model.
Next we tested whether religious tradition, conditional on social capital, would be positively correlated to higher rates of political participation. For respondents of all racial groups who affiliated with a religion, social capital does predict higher rates of political participation. However, the same relationship was not significant for interactions of social capital and specific religious traditions. While Black Catholics were more likely to participate, we found negative relationships for Latinos who identify as Catholic or Born Again, and for Asians who are Hindu.
Lastly, we explicitly examined the importance of religion to immigrants' level of political engagement, theorizing that a more complete model that accounted for tradition, attendance, and social capital would predict increased participation levels. We found support for this hypothesis among Black foreign-born respondents only and not for Latino or Asian foreign-born respondents. This could signify that the connection between places of worship and social capital has changed, and that churches are no longer the primary source of social capital for immigrant and immigrant-derived communities. Another alternative is that communities of color, including those who are primarily foreign born, express their devotion and dedication to their communities in ways that we are not yet capable of measuring.
To that end, we recognize that this investigation suffers from data limitations that could be rectified by future research. First, the survey we used was not explicitly constructed for a discussion of religion and politics, but was chosen to take advantage of the large oversample sample of non-white respondents and the wide range of political outcomes. Because of these limits, we attempted to be conservative in our coding and analyses. We look forward to examining future surveys that can more completely address questions of religious habits and levels of engagement, ask more detailed questions regarding social capital, and still include a sufficient sample of communities of color.
All told, these findings have important implications for the study of religion in traditionally marginalized groups in the United States (Li and Jones 2020). While we know much about the political socialization work faith communities do in general, we show that there is variation conditional on a respondents' social capital. The political participation of congregants can be augmented or decreased by affiliation alone, and not only through repetition of exposure (as measured by church attendance). Anecdotally, this makes sense if we think of respondents who may identify as culturally affiliated with a religion, what we call the Archie Bunker paradox. The television character Archie Bunker rarely attended religious services but regularly professed strong Christian values. Another example is the increasing percentage of Americans who identify as culturally Catholic or culturally Jewish. These folks occasionally attend worship, do not consider their "practice sufficient" to claim this religion, but were raised in that faith and allow it to influence their political participation (Mullen 2015).
We also add more nuanced information to our understanding of religion in the lives of immigrant communities. Although nativity was significant for Latinos in our initial model, this correlation became insignificant when controlling for the religious tradition. Overall, nativity was not a significant indicator of political participation among people of color. This result might be a signal that places of worship are fulfilling a different role for people of color and immigrants in the 21st century than it did in times past. In the 19th century, places of worship were a place for the political socialization of poor immigrants; perhaps now they are places of political refuge for better educated, wealthier and more politically socialized native-born groups (Masuoka et al. 2019). While these questions cannot be answered with the information presently at hand, these analyses provide a stepping-stone for future scholars to investigate these relationships more deeply and broadly, and with a full appreciation of the potential difference among racial and ethnic groups.