Who leads the flock? Religion and the radical right among Brazilian migrants

Abstract

Brazilians in the United States voted overwhelmingly for right-wing populist Jair Bolsonaro in 2022. What role did religion play? Based on exit polling, focus groups, and observation of local Brazilian churches, this article explores how Christianity drives support for right-wing populism among Brazilian migrants to the Boston area. Christians, and especially evangelicals, are significantly more likely to vote for Bolsonaro, and the priests and pastors of Brazilian migrant churches are particularly willing to discuss parties and candidates. Yet neither clergy endorsements nor political conversations at church explain this religious effect. I argue that indirect influence within congregations, which reinforces a conservative worldview in non-overtly political ways, helps explain why most observant evangelicals favor Bolsonaro. Migrants potentially influence the voting behavior of friends and family in Brazil, including via transnational religious communities, so their political attitudes can help bolster authoritarian populism in the homeland, as also seen in India and Turkey.

Introduction

In 2022, 65% of Brazilians in the United States who voted from abroad in their country's presidential runoff election supported far-right populist Jair Bolsonaro. Their conservative tendencies were much stronger than those of their compatriots back home, where Bolsonaro narrowly lost with 49% of the runoff vote, or of Brazilian expatriates outside of the United States, where he received 44% (Table 1).¹

Table 1. Brazilian versus expatriate presidential runoff voting results

Source: Tribunal superior electoral. “Registered” and “Valid votes” give raw numbers; other columns give percentages.

In this study, I argue that conservative Christianity, particularly evangelicalism, plays a key role in expatriate Brazilians' support for right-wing populism, due primarily to indirect influence within congregations rather than explicit efforts by clergy or fellow congregants to persuade.² Religious organizations are key institutions in many urban migrant communities, meeting not only spiritual but also material and social needs (Jones-Correa and Leal, 2001; Levitt, 2007, 2008; Manglos-Weber, 2018). They are particularly important for Brazilians, a newer migrant community speaking a different language than others from Latin America. And while Brazil is still a majority-Catholic country, Brazilian migrants have long been disproportionately evangelical (Margolis, 1994; Martes, 2000; Sales, 2003; Marcelli et al., 2009). Back home, Brazilian evangelicals have been growing in numbers and political influence for decades (Smith, 2019; Boas, 2023), and the evangelical vote was crucial for Bolsonaro's 2018 victory (Layton et al., 2021).

Among Brazilian expatriate communities in the United States, I focus on those in Boston, who stand out for their bolsonarista tendencies. Seventy-six percent of Boston-area Brazilians supported Bolsonaro in the 2022 runoff. This figure was surpassed only by Miami (81%), a traditionally conservative community for Latin American expatriates. By contrast, extreme right-wing voting is a novel trend for Brazilians in Boston. In the two elections prior to 2018, when Bolsonaro was not on the ballot, Boston-area Brazilians were the least supportive of the right-wing candidate of any Brazilian expatriate community in the United States (Table 1). While Brazilians in the United States lean conservative in general, those in Boston seem particularly enthused about far-right populism. Thus, their voting behavior in 2022 cannot simply be explained as a continuation of historical patterns, as it might in Miami.

At the same time, election results in Boston are more typical of Brazilian migrant communities in the United States once we take demographic variables into account, as shown in the Appendix. Hence, understanding what drives this community's conservatism ought to offer insights into why U.S.-based Brazilian expatriates have been disproportionately supportive of right-wing populism in general.

This article adopts a multimethod research strategy to address the role of conservative Christianity in Brazilian expatriates' support for Bolsonaro. It draws on an original N = 715 exit poll of Boston-area Brazilians voting in the October 2022 presidential elections; three qualitative focus groups with Christian Bolsonaro supporters; and analysis of the worship services of 10 Brazilian churches in Boston. I show that being Christian, and especially evangelical, is a particularly strong predictor of supporting Bolsonaro. Yet I argue that religion's impact on Boston-area Brazilians' political attitudes is primarily indirect, and that explicit political speech by clergy or fellow congregants does not change many minds. For evangelicals, regular religious worship helps shore up support for Bolsonaro by reinforcing a multifaceted conservative worldview held by many congregants. Meanwhile, for Catholics, religion matters for political attitudes primarily as a group identity, independent of the regularity of practice. The study thus confirms findings from the religion and politics literature about the limits of clergy persuasion and the important role of congregations as a mechanism of political influence, especially indirect (Wald et al., 1988; Gilbert, 1993; Smith, 2008; Djupe and Gilbert, 2009; Bean, 2014). It shows that these conclusions apply even in a context where politicking from the pulpit is much more common than in English-speaking American congregations.

While Brazil's expatriate voters are a small share of the country's massive electorate, their support for authoritarian populism carries import beyond their numbers. Brazilian churches and churchgoers abroad often maintain ties to congregations back home. Such transnational religious communities serve as channels for “social remittances,” whereby political attitudes acquired in the United States are transmitted back to the sending country (Levitt, 1998). And Brazilians abroad, with fewer interests in how their country of origin is governed, tend to form political opinions and vote based on values—including, for many Bolsonaro supporters, a strong skepticism about the value of democracy. Brazilian diaspora voters thus help to shore up authoritarian populism in the homeland, despite holding more progressive political attitudes in their adopted countries—a phenomenon also witnessed in cases such as India and Turkey.

Theory and hypotheses

Why might members of a migrant community support the political Right in their country of origin? A first common explanation for migrant conservatism concerns social status. Lawson (2003) attributes Mexican migrants' right-wing voting tendencies in expatriate elections to their higher-than-average education levels. Similarly, Bolivian migrants to the United States are whiter and more educated than those who migrate to neighboring South American countries—a likely explanation for their more conservative voting record (Lafleur and Sánchez-Domínguez, 2015). In the case of Brazil, migrants come disproportionately from the wealthier, more developed states in the South and Southeast, and they are whiter, more middle-class, and much more highly educated than the average Brazilian back home (Margolis, 1994; Martes, 2000; Sales, 2003; Rubinstein-Avila, 2005; Levitt, 2007; Lima and Siqueira, 2007). Hence, social status-determining variables such as race, education, and income may influence whether Brazilian migrants support right-wing candidates in Brazil.

Migrants' sources of information about home-country politics also potentially influence their political attitudes. Right-wing populists in Brazil, the United States, and elsewhere have benefited extensively from social media, including its capacity for spreading misinformation (Tucker et al,. 2017). Bolsonaro relied almost exclusively on social media campaigning in his first election in 2018, due to his limited access to television advertising and his hospitalization during much of the campaign (Hunter and Power, 2019). Hence, migrants who favor social media as a source of information about current events might be more likely to support Bolsonaro.

Beyond information sources, religion and religiosity should be particularly important for migrants' support of far-right populists. Christian conservatives, and evangelicals in particular, were disproportionately likely to support Bolsonaro in 2018 (Layton et al., 2021) as well as Trump in 2016 and 2020 (Margolis, 2020; Campbell et al., 2021). More generally, conservative opposition to progressive cultural trends, which is often religiously inspired, has facilitated the rise of right-wing populism in democracies around the world (Norris and Inglehart, 2019). And as “culture war” issues like abortion and LGBTQ rights have become more prominent in national political agendas, religion has become more predictive of left-right voting behavior in Latin America, where materialist conflict use to drive decisions at the polls (Smith and Boas, 2024).

Several distinct aspects of religion potentially matter for Brazilian migrants' political attitudes and voting behavior. First, religion constitutes a social or group identity (Ben-Nun Bloom et al., 2015; McCauley and Posner, 2019) that can potentially influence preferences over parties and candidates. Religious categories define in-groups and out-groups, both of which have implications for voting behavior. All else equal, voters are more likely to favor a candidate with whom they share a politically salient group identity such as religion, thanks to the psychic benefit that it provides and the intrinsic sense of attachment to members of one's “team” (Tajfel and Turner, 1979; McDermott, 2009).

Religious identity was heavily politicized in Brazil's 2022 electoral campaign (Smith, 2022), which ought to increase its salience and weight in voters' decisions. While both major candidates were nominally Catholic, Jair Bolsonaro has long cultivated an ambiguous religious identity that straddles the line between Catholicism and evangelicalism (Oualalou, 2019). The major religious conflict in this election was not between denominations, but rather along Christian versus secular lines. Hence, we would expect those who identify as either Catholic or evangelical to support Bolsonaro in Brazil's 2022 election, while atheists, agnostics, and others without a religious affiliation should favor his opponent, Luiz Inácio Lula da Silva of the Workers' Party.

Religion as a group identity potentially matters even for the nonobservant, but other aspects of religious influence depend upon practice. Regular attendance at worship services brings the faithful into contact with authority figures who can potentially influence their political attitudes. Opinion leaders inserted into local social networks are an important influence on political attitudes and voting behavior in democracies around the world (Baker et al., 2020). Clergy not only have a regular opportunity to communicate their opinions to congregants; they also are endowed with authority by virtue of their religious leadership. Catholic priests combine the authority of the institutional Church with regular contact and an ability to communicate effectively with parishioners (Smith, 2008). Most evangelical churches endow pastors with even greater authority to interpret scripture and make recommendations to the faithful, without the constraints of hierarchy or denominational traditions (Wald et al., 1990; Welch et al., 1993; Djupe and Gilbert, 2009). In surveys, clergy perceive this potential influence over congregants' political opinions, even if they choose not to use it (Djupe and Gilbert, 2009, 31–33).

Despite their potential influence, clergy in the United States rarely try to shape congregants' voting behavior from the pulpit (Guth, 1997; Djupe and Gilbert, 2002, 2003; Beyerlein and Chaves, 2003; Smidt, 2016). In this respect, they have long been constrained by the Johnson Amendment, the provision in the U.S. tax code that prohibits tax-exempt nonprofit organizations from opposing or supporting candidates for office. Overt politicking from the pulpit may also violate important norms within congregations, constituting a line that clergy are reluctant to cross even if they feel they could get away with it (Bean, 2014).

Yet Brazilian clergy in the United States should be more likely than their English-speaking counterparts to endorse candidates, especially those in home-country elections. Some may be unaware of the prohibition on political activity by nonprofit organizations, believe it does not apply to foreign elections, or assume that the chance of legal action against migrant churches is miniscule. Moreover, Brazilians in the United States come from a country where clergy endorsement of political candidates is much more common. In Brazil, legal prohibitions target campaigning in church, but they focus on physical advertisements such as posters rather than speech, and sanctions and enforcement are minimal (Smith, 2019, 21, 86). The greater frequency of discussing partisan politics in Brazilian religious contexts ought to contribute to a distinct norm or “group style” (Bean, 2014) among expatriate congregations, potentially influencing clergy apart from the question of legal constraints.

Clergy are not the only, or even the most important, source of political influence within churches; social pressure within congregations potentially plays a major role (Wald et al., 1988; Gilbert, 1993; Djupe and Gilbert, 2009; Bean, 2014; Smith, 2019). While clergy's political speech is constrained by tax law, congregants can talk freely during coffee hour, a church picnic, or a Bible study group without fear of sanction. Casual conversation about politics in informal spaces may reach members who would be on guard against a priest or pastor's efforts to persuade. The more a church community is a part of its members' lives, the more powerful congregational influence should be (Wald et al., 1990). In migrant communities, churches tend to play a central role in the lives of their members because they are one of the few institutions capable of providing support and social networks, especially when linguistic barriers exist (Jones-Correa and Leal, 2001; Levitt, 2007, 2008; Manglos-Weber, 2018).

In general, social pressure can be expected to influence an identity group's voting behavior when there is a well-defined group norm, behavior is public or verifiable, social networks are homogeneous, and there is potential for sanctioning deviant behavior or rewarding compliance (White and Laird, 2020, 47). Black Americans' support for Democrats is the quintessential example; White and Laird (2020, 18) argue that their model should also apply to white evangelical Christians in the United States. If so, conservative Christians in the Brazilian migrant community should be even more susceptible to such pressures. A clear group norm exists: in the survey analyzed below, 93% of evangelicals supported Bolsonaro in the runoff, a level comparable to Black Americans' support for Democrats. Linguistic isolation contributes to homogenous social networks, especially compared to Spanish-speaking migrants, who hail from diverse national backgrounds. Voting may be secret, but public displays of political allegiance, such as Bolsonaro campaign paraphernalia, were ubiquitous during the campaign. And sanctioning capacity is inherent in religious organizations that provide spiritual, social, and material support for their members.

Social pressure within congregations may involve direct persuasive efforts by lay opinion leaders, but it can also happen in indirect ways. In the United States, evangelicals participate in a culture in which partisanship has been subtly incorporated into religious identity, contributing to unspoken understandings of which party and candidate “people like them” are expected to support (Bean, 2014, 62). A politically-tinged Manichaean worldview contributes to this process of identity formation, with evangelicals conceiving of themselves in a battle of good versus evil, and blaming “the liberals”—theological, political, or both—for America's decline (Bean, 2014). In polarized Brazil, and among the Brazilian community in the United States, a similar process of evangelical identity formation may be taking place. Negative partisanship—antipathy toward Democrats in the United States or the Workers' Party in Brazil—is a key component of evangelical support for the far right in both countries (Margolis, 2020; Araújo, 2022). If congregations influence political attitudes indirectly, observant Christians in the Brazilian migrant community may incline toward support for Bolsonaro even without explicit attempts to persuade them.

In sum, religion can influence migrants' voting behavior via group identity, clergy persuasion, and two forms of social pressure: direct congregational persuasion and participating in a religious community where a partisan identity is reinforced in more indirect ways. These distinct mechanisms have different observable implications. In all instances, identifying as Christian, especially evangelical, should be associated with support for Bolsonaro. If group identity is the main mechanism, this relationship should not depend on one's level of religious observance. For the other three mechanisms, more frequent church attendance should increase the magnitude of the Catholic or evangelical effect, as voters are potentially exposed to more clergy persuasion and congregational influence. If clergy speech matters, candidate endorsements by one's pastor or priest should be associated with support for Bolsonaro. If direct congregational influence makes a difference, talking about current affairs at church should correlate with support. If indirect congregational influence is the key mechanism, religious group identity should matter for voting behavior, and religiosity should moderate its effects, but neither endorsements nor talking about current affairs at church should be significant predictors.

Methods and data sources

To address these hypotheses, this study draws on three original data sources: an N = 715 exit poll of Brazilians who voted from abroad in the 2022 election at Boston-area polling places; three focus groups with Christian Bolsonaro supporters, recruited from among the survey respondents; and analysis of the livestreamed worship services of 10 Brazilian churches in Boston, both Catholic and evangelical, from August to November 2022. This section describes each data source.

Exit polls

To gather data from a representative sample of Boston-area Brazilians who voted in the October 2022 presidential election, our research team conducted exit polls at both in-person voting locations set up by the Brazilian consulate in Boston. Enumerators waited outside each venue and randomly approached voters who were leaving, asking them to complete a 5-minute survey via a Portuguese-language self-administered paper questionnaire. The enumerators were undergraduates or recent graduates who were fluent in Portuguese; most were Brazilian-Americans who had grown up in the United States. Members of the team were present at each polling place during the majority of voting hours for both the first round election on October 2 and the runoff on October 30.³ Questionnaires were identical across the two rounds, with the exception that the runoff questionnaire asked about vote in both rounds of the election. We surveyed 310 voters in the first round and 405 in the second, a sample size that compares favorably to other exit polls of expatriate voters (Boccagni, 2011; Escobar et al., 2014; Lafleur and Sánchez-Domínguez, 2015) as well as household surveys of immigrant communities in the United States (Guarnizo et al., 2003; Marcelli et al., 2009). We pretested the exit poll questionnaire at the 2022 Brazilian Independence Day Festival in Boston, an annual gathering of the Brazilian community in early September.

On metrics that allow for a comparison, including gender and age, the sample is fairly representative of Boston-area Brazilians who voted in the 2022 election, as highlighted in the Appendix. In terms of religious variables, 32% of respondents reported that they were Catholic, while 44% self-identified as Protestant, evangelical, Pentecostal, or other non-Catholic Christians. By comparison, in Marcelli et al.'s (2009) household survey of Boston-area Brazilians, fielded in 2007, 48% of respondents were Catholic and 37% were Protestant. The difference could reflect continued growth of the Protestant/evangelical population—also a phenomenon in Brazil—as well as the fact that evangelicals may have been particularly motivated to vote in this election. Nearly half of all respondents reported that they attend church once a week or more.

Focus groups

To gain qualitative insight into the political opinions of Bolsonaro supporters and the role that religion plays in their attitude formation, I conducted focus groups, a common approach in other studies of religion and politics in Brazil (Smith, 2019) and among immigrants to the United States (Wong, 2018). When completing the questionnaire, all respondents had the option to leave their contact information to potentially receive an invitation to participate in a focus group in exchange for a $50 gift card; 45% did so. After each round of the survey, including the pretest at the September festival, a Brazilian-American research assistant invited nearly all respondents who were religiously observant Bolsonaro-supporting Christians (either Catholic or evangelical) to participate in a focus group. Groups were held in meeting rooms in local public libraries on a Saturday morning or Sunday afternoon and ran for 1.5–2.5 hours; six participants attended each one. While those who show up to focus groups are an inherently self-selected sample (Boas, 2024), the focus group participants were not markedly different from those who were invited but did not attend, as highlighted in the Appendix.

As recommended by Cyr (2019), focus groups were moderated by a member of the research team who was demographically similar to participants—a first-generation immigrant to the greater Boston area who had grown up in an evangelical household. I attended all focus groups, introduced myself at the start, and asked a few follow-up questions during the conversations, but the Brazilian-American moderator was running them. Focus group topics included people's sources of information about Brazil and the election, their opinions about Brazilian and American politics, and their religious lives, including how often politics is discussed in church; specific questions are reproduced in the Appendix. All quotes taken from the focus groups use pseudonyms.

Brazilian church services

In order to obtain a direct measure of how much political information is conveyed during Brazilian church services in Boston, I analyze the livestreamed weekly worship services of six evangelical and four Catholic churches in the Boston area between August 13 and November 6. In selecting churches, I sought to identify the largest and most popular Brazilian churches in Boston, as described in the Appendix. Many churches began livestreaming their services during the COVID pandemic and continued this practice even as in-person worship resumed. For each church, I downloaded videos of weekly services from their Facebook or YouTube pages, ran them through Trint's automated video transcription service, and read through the transcripts, looking for any mentions of politics and extracting relevant quotes. I also conduct text-as-data analysis of the transcribed worship services.

Compared to visiting churches in person, analyzing videos and transcripts of worship services has both advantages and disadvantages. The present approach allows me to follow more churches, on a more regular basis, than would be possible through in-person visits, even if every member of the research team were visiting a Brazilian church every Sunday morning. There are no concerns about Hawthorne effects (research subjects modifying their behavior in response to being observed); a scholar analyzing a public video is an electronic “fly on the wall.” One avoids the awkwardness and potential inaccuracies of scribbling notes during a worship service, as well as the ethical concerns that might arise from recording or publicizing a message that may only have been intended for a limited, in-person audience. On the other hand, analyzing livestreamed services is clearly not participant observation; one sees and hears what is taking place on stage or behind the altar but not how participants are reacting in the pews or what conversations might be taking place before and after worship services.

Analysis

Predicting support for Bolsonaro

Consistent with the actual electoral results, exit poll respondents were strongly supportive of Jair Bolsonaro, with 63 and 70% reporting a vote for him in the first round and the runoff, respectively. What explains Boston-area Brazilians' support for right-wing populism?

To test the hypotheses outlined in section “Theory and hypotheses,” I estimate a probit regression in which the dependent variable is the respondent saying that they voted for Bolsonaro in the first round of the election. To operationalize social status, I use respondents' self-declared race (indicator variables for Black, brown, and other race, with white as the reference category), education, and income. Religious group membership is measured in terms of identifying as Catholic or as evangelical, Protestant, Pentecostal, or non-Catholic Christian, with no or another religion as the reference category. Both religious indicator variables are interacted with frequency of church attendance. To test the hypothesis about clergy influence, I use an indicator for reporting that the respondent's pastor or priest spoke in favor of or against a candidate in Brazil's 2022 presidential election. To measure the potential for direct congregational influence, I include an indicator for mentioning church (among several nonexclusive options) as a place where the respondent talks about “what's going on in the world.” To test the hypothesis about political information sources, I include an indicator for relying on social media as a major source of information about Brazilian current events. Finally, I include controls for age (in years) and an indicator for male respondents (versus female/other gender). Age, education, income, and church attendance are standardized, so estimates represent the effect of a one standard deviation change, and the estimates for the evangelical and Catholic indicators represent the effect for those with average levels of church attendance.

The results of this analysis, summarized in Figure 1 and in an Appendix table, confirm that religion is a major factor in explaining Brazilian migrants' support for right-wing populism. At an average level of church attendance, evangelicals and Catholics are significantly more likely than those with no religion, or a non-Christian religion, to vote for Bolsonaro. The estimated effect of these religious indicators dwarfs that of any other variable in the model.

Figure 1. Predicting Brazilian migrants' support for Bolsonaro.

Note: Icons show point estimates and lines give 95% confidence intervals from a probit regression of vote for Bolsonaro in the first round in 2022 (versus other/none/blank/null). Age, education, income, and church attendance are standardized, so estimates represent the effect of a 1 standard deviation change. White is the reference category for race and none/other is the reference category for religion. Church attendance is interacted with religion; in place of the main effect and interaction terms, I report the marginal effects for evangelicals and for Catholics. N = 479.

For evangelicals, the results suggest that indirect persuasion within congregations matters most for support of Bolsonaro. Leaving other covariates at their observed values and varying church attendance across its interquartile range, an evangelical who attends church once or twice a month has a 77% chance of supporting Bolsonaro, whereas one who attends more than once a week has a 90% chance. However, neither clergy endorsements nor talking about current affairs at church are significant predictors of supporting Bolsonaro. This suggests that church attendance matters for evangelicals because of indirect influence rather than explicit efforts, by either clergy or fellow congregants, to persuade.

For Catholics, church attendance does not significantly boost support for Bolsonaro. Rather, simply identifying as Catholic, even nonpracticing, is what makes a difference. Hence, it seems like group identity matters most, rather than political influence within places of worship. This finding contrasts with Brazilians' voting behavior in the 2018 election (Layton et al., 2021), where being Catholic was not a significant predictor of support for Bolsonaro. In contrast to Brazil, Catholicism is a minority religious tendency in the United States, as it is among the Brazilian community in Boston. Those with no religion are also a much larger share of the population—around 30% in the United States, versus 12% in Brazil (Balloussier, 2022; Pew Research Center, 2022)—and large urban areas like Boston are a particularly secular environment. While identifying as Catholic is simply the default in Brazil, retaining one's Catholic identity after immigrating appears to be a more politically meaningful choice.

In contrast to religious variables, information sources and social status-related variables have smaller and inconsistent associations with support for Bolsonaro. Black respondents are less likely to vote for him, though these effects are imprecisely estimated due to their small share of the sample. Higher incomes have no significant effect. Education matters somewhat, but in the opposite direction as predicted by social status explanations for migrant conservatism; those with less education are more likely to favor Bolsonaro, as has also been shown for Brazilians in Brazil (Layton et al., 2021). Finally, relying on social media as an information source has no significant association with support for Bolsonaro, perhaps reflecting its ubiquity and users' tendency to consume information that reinforces existing views. This finding echoes research on Brazil's 2018 election showing that social media use did not consistently benefit Bolsonaro (Rennó, 2020).

The negative relationship between education and support for Bolsonaro helps explain why, among major Brazilian expatriate communities in the United States, those in Boston went from being the least supportive of the Right in 2010 and 2014 to the second most supportive in 2018 and 2022 (Table 1). As shown in the Appendix, Brazilians in Boston are a more working-class migrant community, with substantially lower levels of education, than those in other major U.S. cities (Martes, 2000; Lima and Siqueira, 2007). In the 2010 and 2014 elections, with conventional right-wing candidates, lower levels of education predicted voting for the Left rather than the Right (Smith, 2011; Baker et al., 2020). This pattern flipped once Bolsonaro was on the ballot in 2018 and 2022. Given their demographic profile, Boston-area Brazilians were ripe for being wooed by a right-wing populist, much as Donald Trump did with many blue-collar, formerly Democratic voters in the United States. Of course, favorable demographic profiles merely set the stage for other sources of influence, such as churches, to potentially affect vote choice.

Further evidence on politics in church

Analysis of the exit poll survey data suggests that religious variables matter a great deal for Brazilian migrants' support for Bolsonaro. However, direct political persuasion, in the form of clergy speech about parties and candidates or congregational discussion of current events, has no effect. Indirect influence within congregations seems to be the major reason why evangelicals favor Bolsonaro, and group identity, rather than any persuasive aspects of the worship experience or congregational life, accounts for why Catholics support him. In this section, I draw on evidence from the focus groups and church service analysis, combined with descriptive statistics from the survey, to delve deeper into how religion and religiosity may influence migrants' political attitudes.

Clergy political speech

Based on responses to the survey, Brazilians in Boston are much more frequently exposed to explicitly political messages in church than their American counterparts in the United States more broadly (Table 2). I asked if the respondent's priest or pastor had spoken in favor of or against a candidate in Brazil's 2022 election. Among respondents who attend church at least one to two times a month, 27% of evangelicals and 15% of Catholics answered yes. These figures are not as high as in Brazil's highly polarized 2018 election, when Bolsonaro first ran for president, but they are higher than in the more ordinary Brazilian elections of 2014 and 2010. They are also far above figures from the United States in the polarized elections of 2000 and 2016, where no more than about a tenth of Protestants report clergy endorsements. Some of the Brazilian respondents attend English-language American congregations, so the rate of explicitly political speech in Boston-area Brazilian churches is almost certainly higher than reported here.

Table 2. Clergy speech on candidates/parties: comparative statistics

Note: Figures are percent of churchgoing (at least once a month) respondents of each religion reporting that their clergy spoke about presidential candidates or their parties during the campaign. Data sources: Boston Brazilians Survey 2022; Democracy on the Ballot: Brazil 2018 Survey (using sampling weights due to the online sample); Brazilian Electoral Panel Study 2010 and 2014; Pew American Trends Panel Wave 18 (2016); American National Election Studies 2000 Time Series.

On the other hand, even in those churches where pastors speak about politics, the vast majority of worship time is devoted to strictly religious topics. Figure 2 summarizes results from a word count analysis (discussed further in the Appendix) of transcribed livestreamed worship services from 10 Brazilian churches. Common worship terms, such as “blessing” and “hallelujah,” appear much more frequently than political terms such as “vote,” “election,” and “candidate.” References to political terms peaked on the two election days, October 2 and October 30, but many of these mentions were entirely nonpartisan, such as clergy mentioning at the start of the service that people would be trickling in late because of long lines at the polling place. A pastor or priest who speaks about parties or candidates is likely to do so via a prayer, announcement, or isolated comment during a sermon on election day, rather than a constant drumbeat of political content throughout the campaign season.

Figure 2. Religious and political terms in Boston-area Brazilian church services.

Note: Based on automated text analysis of church service transcripts; see text for details.

Qualitative examples from the livestreamed Brazilian church services give a sense of what form clergy speech about candidates might take. In four of the six evangelical churches (and none of the Catholic churches), clergy offered pro-Bolsonaro or anti-Lula comments on election day or the Sunday before. At one church on election day, the pastor closed the service with “God Bless Bolsonaro, brothers, and yes, we're going to pray for him…It's crazy for anyone, any believer, to vote for someone else.” At another church, the pastor criticized Bolsonaro's major opponent: “I'm not for Lula, no way. Lula has to get out of there by any means, I wouldn't vote for him even if I were dead.” Others endorsed Bolsonaro in more of a tongue-in-cheek, wink-wink, nod-nod fashion. At one church on election day, before offering a closing prayer for the nation, the pastor asked for a Brazilian flag to be displayed on the video screen behind him. A Bolsonaro 2022 campaign poster promptly appeared, and the pastor remarked, to applause and laughter, “There we go! Oh, not that one, sorry!” The Brazilian flag replaced it, and the prayer began.

These examples of pro-Bolsonaro or anti-Lula comments offer insight into why clergy endorsements, though more common than in English-speaking American churches, may not actually affect Brazilian migrants' voting behavior. Even with the more serious, less tongue-in-cheek comments, one gets the sense that clergy knew they were preaching to the converted and did not need to expend much effort to win votes for Bolsonaro. Comments about candidates and parties were always brief, even on election day. The only extended discussion of the election, running about three and a half minutes, came from a pastor who did not endorse any of the candidates, but instead lamented the divisiveness of the campaign and called on Christians to move beyond their political differences.

Similarly, nearly all of the Bolsonaro supporters who participated in the focus groups reported that explicit discussion of politics in church was rare. According to José, a Catholic, the topic only came up indirectly: “in the church that I attend, they ask God to enlighten politicians…I've never heard [my priest] talk about parties, never.” Maria, an evangelical, said “I think it is really great that my pastors, they don't show, there's no way to know who they voted for.” Some participants had encountered political discussion before; Adriana, an evangelical, said she used to attend a Brazilian church in Boston where talking about politics was more common than in her current English-language church. But most participants reported that clergy never talk explicitly about politics during church, whether because there are too many different political opinions within a congregation, too little time during the service, or simply that the purpose of worship is not politics.

Congregational influence

While analysis of livestreamed worship services can capture clergy's formal messages about politics, a lot of potential persuasion happens in more informal spaces of socialization, such as a church picnic or the ubiquitous coffee hour after worship. Several focus group participants mentioned that politics is more commonly discussed in these informal spaces, especially at election time. According to Adriana, an evangelical: “I think that it depends a lot on the moment…now, in electoral season, it's going to be a common subject. Outside of election season, maybe soccer is more the topic.” Paulo, an evangelical, mentioned that during campaign season, a common discussion topic was how people who purport to believe in biblical principles can support leftist candidates who go against those principles. Different congregations have different norms, however. Ana, a Catholic, said that in her parish, politics “is not a common topic of discussion, because it is very polemical and everyone has an opinion.”

Yet it was also apparent from the focus groups that there is ample potential for indirect influence within congregations, as people's interactions with one another reinforce multiple elements of a distinct conservative worldview. As Brazilian migrants and churchgoing Christians, the six participants in each of the three focus groups were people who, despite not knowing each other (with one exception) before setting foot in the discussion room, could plausibly have been members of one or a couple congregations.⁴ As such, their interactions during the focus groups constitute something of a microcosm of Brazilian migrant churches and the conversations that might emerge therein. While the focus group moderator asked specific questions, there was plenty of freeform discussion that followed, allowing for particular topics, and points of consensus, to emerge spontaneously.

Below, I list elements of this conservative worldview that emerged spontaneously in one or more focus groups, along with representative quotations. Comments on these topics often engendered nodding, murmurs of approval, and supportive interjections from other group members.

• • Patriotism

  • ◦ João (evangelical): “As long as the schools don't go back to teaching patriotism, no one is getting ahead.”

  • ◦ Antonia (evangelical): “Patriotism. We have to first create it at home, then in the schools.”

• • Anticommunism and the military regime

  • ◦ Juliana (evangelical): “I am against communism. Why?…We can't say we're going to equalize the world. Equality doesn't exist. We are different.”

  • ◦ José (Catholic): People criticize the military regime “because they had prisons, and all that stuff about killing 100,000 people. But it wasn't like that at all…they were against communism.”

  • ◦ João (evangelical): “Who saved Brazil? It was the military.”

• • Leftist indoctrination in schools

  • ◦ Maria (evangelical): “Here, in elementary school, my son, when Trump was in office, they were teaching partisanship,⁵ speaking disrespectfully.”

  • ◦ Pedro (evangelical), on his education in Brazil: “It was sixteen years of brainwashing, from kindergarten through the university.”

  • ◦ Carlos (Catholic): “The universities in Brazil, mainly the federal ones, they are created within an ideological system.”

Participants did not agree on every aspect of a conservative worldview. Some were much more favorable toward the welfare state, while others espoused free market attitudes. Most, understandably, supported more liberal immigration laws. But enough elements of this worldview emerged as seemingly consensus positions in the focus groups that, if conversations at church picnics and coffee hours tend to flow along similar lines, there is definite potential for influence and persuasion. Without ever mentioning candidates, elections, and parties, discussion of moral and political issues within congregations can encourage ideological conformity by clearly indicating the positions that “people like them” are expected to hold. Right-wing voting behavior has the potential to emerge naturally within such a community simply because the majority of members cannot imagine doing anything else. Indirect social pressure thus plays a similar role among churchgoing Brazilian migrants as it does among Black Americans, who “have come to see black support for the Democratic Party as just something that ‘black people do’” (White and Laird, 2020, 45).

My interpretation of the survey and focus group evidence is that indirect persuasion by fellow congregants serves to reinforce a conservative worldview among churchgoing Brazilian migrants, but it is worth considering the possibility of reverse causality—that those with a conservative worldview choose churches that align with their preexisting attitudes. While politics frequently drives religious practice in the United States (Margolis, 2018), where houses of worship span the ideological spectrum and many opt out of religion entirely, there is less opportunity for choice among the Brazilian migrant community. Few Brazilian migrant churches lean left in home-country politics; only two survey respondents reported that their pastor or priest supported Lula or criticized Bolsonaro. Switching to a more liberal English-speaking congregation requires language skills, and even migrants who face no language barrier often prefer the cultural familiarity of an ethnic church (Manglos-Weber, 2018). And while staying home is always an option, the nonpracticing miss out on the crucial community-building and social support functions that churches fulfill, especially for the newly arrived. These incentives underlie the high rates of religious attendance among Brazilian migrants: 48% of exit poll respondents reported attending church once a week or more, versus 23% of Americans in 2022 (Public Religion Research Institute, 2023). While there is undoubtedly some politically-driven self-selection in terms of who attends Brazilian churches, there is also ample opportunity for social pressure to reinforce conservative attitudes among those who might otherwise be inclined to waver.

Discussion

Brazilian migrants' support for Bolsonaro is theoretically significant for what it reveals about the mechanisms of political persuasion within congregations, but it is also substantively important for Brazilian democracy. In this section, I offer some tentative thoughts about how migrants' values-driven attitudes about Brazilian politics may serve to shore up authoritarian populism, beyond the direct impact of their votes.

If voting from abroad is one way that migrants maintain ties to their country of origin, participating in transnational religious communities is another. Religious Brazilians who migrate to the United States often remain members of and continue to donate to churches in Brazil, and diaspora churches are often linked to home-country churches of the same denomination (Levitt, 2007). In the livestreamed church services analyzed for this study, visiting pastors from Brazil often made guest appearances at congregations in the Boston area, such as Brazilian pastor and federal deputy Marco Feliciano, who served as guest preacher at Revival Church in Everett in mid-November 2022. Membership in transnational religious communities of this sort facilitates participation in home-country politics—not only voting from abroad, but also “social remittances,” in which migrants transmit newly acquired attitudes and opinions back to their country of origin (Levitt, 1998, 2008).

While Brazilian migrants are unlikely to change an election outcome with their votes alone, their influence in the form of social remittances is potentially greater. A case in point concerns Governador Valadares, Brazil's major migrant-sending city and one with deep historical ties to Boston (Rubinstein-Avila, 2005). Like Brazilians in Boston, Governador Valadares voted disproportionately for Bolsonaro in 2022 and 2018 but was more typical in 2014, which featured a conventional rather than populist right-wing candidate (see the Appendix). While a conclusive analysis is beyond the scope of this article, the parallel trends in these deeply interconnected cities could certainly be a product of reciprocal influence via migrants' ties to family and friends back home.

If migrants are sending home attitudes about Brazilian politics, these social remittances derive largely from values rather than interests, including a strong skepticism of the value of democracy. In the focus groups, participants explained their support for Bolsonaro by reference to his persona—that he was honest, spoke the truth, and delivered on his promises—rather than his policy positions (with the exception of abortion). This contrasted with their stance on U.S. politics, where many participants favored Democrats—especially Massachusetts gubernatorial candidate Maura Healey, an open lesbian—and justified their support by reference to issues such as health care and immigrants' rights that affected their day-to-day lives in the United States. A strong contrast also emerged in the exit poll, where respondents were asked whether a military coup could be justified in each country in the face of major corruption. As shown in the Appendix, Brazilians in Boston were significantly more supportive of a Brazilian coup than their counterparts back home but significantly less supportive of a coup in the United States than the American public.

In sum, the case of Brazilians in Boston shows how a migrant community can help shore up authoritarian populism in their country of origin despite adopting more progressive and pro-democratic political attitudes in their new home. One observes a similar pattern in other cases (Mishra, 2021; Böcü and Panwar, 2022; Prasad et al., 2023; Saglam and Paarlberg, 2023). Indian-Americans overwhelmingly vote for Democrats in the United States, yet they have also been a key source of financial support for Prime Minister Narendra Modi. Turkish migrants to Europe tend to vote for the Left in their adopted countries but have strongly favored President Recep Tayyip Erdoğan when voting from abroad in Turkish elections. In each case, conservative religious values influence attitudes toward country-of-origin politics, whereas interests may be driving a more progressive stance abroad. This finding suggests important limits to theories and policy goals of immigrant integration (Saglam and Paarlberg, 2023).

Conclusion

In recent years, the United States and Brazil have both experienced extreme political polarization and the coming to power of right-wing populists who posed serious threats to democratic stability both in and out of office. In both countries, an emerging cleavage between Christian conservatives and secular progressives is contributing to this polarization, and conflict around “culture war” issues and identities has helped secure a loyal base of support for the radical right.

Regardless of whether they arrived in the United States three decades or 3 months ago, Brazilians in major immigrant communities like Boston have hardly escaped the climate of political polarization in their home country. Bolsonaro 2022 bumper stickers were visible on cars around Boston during the campaign, and campaign posters were on display in Brazilian restaurants. On election day, many voters arrived at the polling place dressed in the colors of their candidate—green and yellow for Bolsonaro or red for Lula. And these physical manifestations of polarization surely pale in comparison to social media, where partisan and fake news content shared by family and friends flows readily across borders via WhatsApp groups.

Brazil's 2022 campaign was particularly polarized along religious lines, with the leading candidates visiting churches, seeking endorsements from televangelists, and trading accusations of making pacts with or being possessed by the devil. Local priests and pastors often waded into the fray, supporting a candidate or criticizing his opponent from the pulpit. In this respect, Brazilian communities in the United States were more similar to their homeland than their adopted country. While most clergy in the United States rarely make political endorsements that could jeopardize their churches' tax-exempt status, Brazilian clergy in Boston do so much more freely, as shown by the survey analyzed in this article.

Religion mattered in particular for Brazilian migrants' voting behavior in the 2022 election. In this study, I show that being a Christian, whether evangelical or Catholic, is a strong predictor of supporting Bolsonaro over his opponent Lula. For evangelicals, these results echo prior findings about home-country voting behavior in the 2018 election (Layton et al., 2021). For Catholics, this finding is new; it suggests that Catholic identity abroad may be more politically meaningful than in Brazil, where it is still the default.

Yet I argue that direct persuasion within churches—in the form of either clergy endorsement from the pulpit or explicit conversation about the election within congregations—does not explain evangelical and Catholic support for Bolsonaro among Boston-area Brazilians. Neither clergy endorsements nor mentioning church as a place where one talks about current affairs is predictive of the vote. Analysis of livestreamed services from 10 Brazilian churches in Boston suggest that endorsements, when they happen, are short, fleeting, and sometimes tongue-in-cheek, with the distinct feel that clergy are preaching to the converted. Participants in focus groups report that, in some churches, both clerical and lay discussion of politics violates important norms.

Rather, what seems to matter most for Brazilian migrants' right-wing voting behavior is indirect influence within congregations, where day-to-day interactions serve to reinforce a conservative worldview in ways that are not explicitly political or even religious per se, but that nonetheless contribute to a shared understanding about the types of candidates that Christians should support. In focus groups, several areas of consensus emerged naturally and independently—the importance of teaching patriotism to children, the role of Brazil's military regime in opposing communism, and charges of leftist indoctrination within schools. These issue stances are all closely associated with Bolsonaro, but participants did not link their positions to his presidency or candidacy for reelection. To the extent that participating in evangelical worship, and evangelical culture more broadly, reinforces this conservative worldview, it helps explain how church attendance can increase the magnitude of evangelical support for Bolsonaro, even if clergy endorsements and explicit discussion of current events in church have no direct effects.

This study thus extends Bean's (2014) argument about evangelical political identity in the United States, showing that congregational subcultures can be a key mechanism of indirect political influence even in churches where overt discussion of candidates and elections is not necessarily taboo. It also underscores the limits of clergy political influence in a context where clergy are not shy about speaking up. In English-speaking congregations in the United States, direct clergy effects on voting behavior are the dog that didn't bark: there is little evidence that priests and pastors influence how their congregants vote (Smith, 2008), but also little evidence that they try to do so (Guth, 1997; Djupe and Gilbert, 2002, 2003; Beyerlein and Chaves, 2003; Smidt, 2016). Here, I show that even among a community where direct clergy endorsements are much more common, indirect influence within congregations matters more.