Threat or Godsend? Evangelicals and Democracy in Latin America

Abstract

While Protestant and Evangelical groups started as small minorities in Latin America, they have recently experienced explosive growth and now make up a large proportion of the region’s religious faithful. This unexpected shift led to a spate of scholarship speculating as to how the new Evangelical communities would impact society. Given the perennial concern over the health of democracy in Latin America, much of this work focused on how Evangelicals might differ from Catholics in terms of support for democratic values and civic participation. Some predicted that Evangelicals would be eager supporters of democracy and an active, positive force in the community. Others warned that Evangelicals were too focused on spiritual matters, which would lead to apathy and a passive acceptance of authoritarianism. This paper uses recent survey data from sixteen nations in Latin America to test these theories with the goal of answering the question: are Evangelicals a threat or godsend for democracy in Latin America?

1. Introduction

Over the last fifty years, Latin America has experienced tumultuous religious change. A near-universal Catholic hegemony has been challenged by unprecedented Protestant and Pentecostal growth¹ (De la Torre and Martín 2016; Somma, Bargsted, and Valenzuela 2017). While the growth varied considerably across nations, Evangelicals now make up 19% of the region. Much of this growth is extremely recent—over half of Evangelicals are Catholic converts (Pew Research Center 2014).

These shifts renewed interest in the concept of the “Protestant Ethic.” According to the Weberian tradition, Protestant groups are characterized by the kinds of values and behaviors that are essential for healthy democracies (Weber 1958). As such, scholars predicted that the new Latin American Evangelicals would be more likely than Catholics to hold democratic values and demonstrate behaviors of civic engagement like volunteerism, political campaigning, and voting. This would help stabilize nascent democracies in the region (Willems 1967; Lynch 1998; Smilde 1999; Sherman 1997; Escobar 1994; Stoll 1990). However, scholars were divided in their assessment. Others characterized Latin American Evangelicals as “other-worldly” and detached from their communities, which would naturally lead to political apathy and a tolerance for authoritarianism (d’Epinay 1969; Stoll 1990; Chesnut 1997; Alves 1985; Brouwer 1996; J.-P. Bastian 1993).

Researchers have used survey data to test these competing hypotheses and provide further analysis of the interplay between religion and political behavior in nations across Latin America (Aguilar et al. 1993; Patterson 2004; 2005; Gill 2004; Steigenga 2002; Boas 2021; Parker 2008; Fediakova 2012b). However, relatively little attention has been paid to Central America, even though the region is home to the four most Evangelical countries in all Latin America (Honduras, Guatemala, El Salvador, and Nicaragua). Furthermore, in the past two decades, the religious landscape of the region has continued to experience rapid change, both in religious affiliation and explicit political involvement. Contemporary scholars revisit these questions and call for more research on how Evangelical growth impacts the health of democracy in Latin America. (Boas 2021; Freston 2008; Díaz Domínguez 2020; Cleary 2018; Berger 2010; Parker 2016; Pérez Guadalupe and Grundberger 2019). Using the AmericasBarometer surveys, I analyze recent nationally representation data from sixteen nations, contributing to this body of scholarship with an updated, truly regional analysis of Evangelicals’ democratic attitudes and behaviors to assess the dueling hypotheses in the literature.

2. Evangelical Growth and Democracy: Two Positions Emerge

In the late 1980s and early 1990s, Evangelicalism grew exponentially throughout Latin America (Pérez Guadalupe 2019). Academics interested in religion and Latin America produced a large amount of research analyzing the surprising success of the new Evangelical groups in the region. A flurry of papers and books were published, offering fresh commentary, and analyzing previous research on these “sects” (Stoll 1990; Ireland 1993; Martin 1990; Aguilar et al. 1993; Chesnut 1997; Froehle 1994; Ireland 1992; Garrard-Burnett and Stoll 1993; J. P. Bastian 1990). This conversation was shaped by two landmark works: Is Latin America Turning Protestant? by anthropologist David Stoll (1990), and Tongues of Fire: The Explosion of Protestantism in Latin America by sociologist of religion David Martin (1990). In his forward to Martin’s book, renown sociologist Peter Berger wrote that the work “deals with one of the most extraordinary developments in the world today.” (1990, vii). It was now clear that Evangelicals would play a significant role in the future of Latin America, and researchers wondered how this would impact the region (Steigenga 2002; Parker 2008).

This period of growth coincided with the “expansive context of democratization” in the region (Burity 2020, 6). Given the existing anxiety around authoritarianism and the future of democracy in Latin America at the time, that much of this research attended to questions of political impact and change. How would the growth of Evangelicals affect the trajectory of democracy in Latin America? Would they be active participants in democracy and supporters of human rights, or would they be apolitical and apathetic? Or worse, would they develop anti-democratic characteristics and welcome autocrats? While scholars like Stoll and Martin addressed the nuance of these questions, two competing hypotheses emerged.

The first hypothesis was rooted in Christian Lalive d’Epinay’s criticism of the Pentecostal movement in Chile (1969). He claimed that in the face of urbanization and social change, Pentecostal groups sought to recreate the paternal hacienda model of the past. The pastors assumed the role of the strongman patron, demanding total obedience and extinguishing any lay ambition or movement toward reform. This “ethic of passivity” had real-world consequences for the public square as it taught “its initiates withdrawal and passivity in political matters.” (d’Epinay 1969, 145; Fediakova 2012a). As this movement was inherently anti-progress, D’Epinay predicted that it would play an “extremely conservative” role in legitimizing the status quo of inequality and autocratic leadership in the region (Fediakova 2012a, 127).

Later proponents adopted and expanded this frame, asserting that Evangelical groups across Latin America were inwardly focused and politically apathetic (Brouwer 1996; Alves 1985; J.-P. Bastian 1993). Their dualistic theology that shunned “worldly matters” meant that Evangelicals wouldn’t concern themselves with democratic participation or societal improvement (Parker 2012, 33). This retreat from civic engagement could result in a passive acceptance of creeping authoritarian rule, as demonstrated by the high-profile examples of Evangelicals support for autocrats like Rios Montt in Guatemala, Alberto Fujimori in Peru, and Augusto Pinochet in Chile (Deiros 1991; Gill 2004, 45; Ireland 1992; 1993; Parker 2008).

Across academic circles, Latin American Evangelicals were soon routinely characterized as a negative force for democracy. While the details varied between contexts, the message was clear: “Whether as obtuse, or as opportunists, or as a threat, they were seen as indicators of a democratic sickness” (Burity 2020, 6). Decades later, this remains a common belief about Latin American Evangelicals that scholars continue to test and challenge (Burity 2020; Parker 2012, 41; Mansilla 2014).

Sociologist Emilio Willems, a contemporary of D’Epinay, came to a different conclusion in his research on Protestant groups in Brazil and Chile (1967). He argued that Protestantism represented a rejection of the traditional social order, and that the inherent egalitarianism in the movement would encourage civic and political participation. As these minority groups grew, they would naturally begin to demand rights and recognition—this would be the “first moment of Evangelical politization,” a process that implied democratic participation (Burity 2020, 19; Fediakova 2012b, 26).

Scholars built on Willems’ argument, rejecting the conceptualization of Evangelicals as apolitical or authoritarian by pointing to case studies across the region where they had been an active and positive force for democracy (Froehle 1994; Smilde 1999; Steigenga 2002; Wilson 1994).

With a Weberian framework they proposed an alternate hypothesis: Positive elements of the new Evangelical groups naturally encourage support for democracy and an active citizenry Specifically, the participatory church model encourages leadership development, associational ties, and civic and political participation (Lynch 1998; Smith 1994; Sherman 1997; Escobar 1994; Gómez 1999).² Sociologist Christian Smith summarized this view in his prediction that these groups may “emerge as a significant positive force to helping to foster genuine democracy” in Latin America (1994, 119).

2.1. Testing the Theories

Most of the evidence presented in support of these hypotheses came in the form of theoretical analysis or qualitative research. However, some researchers have tested these ideas with quantitative data. In 1993, Aguilar et. al evaluated “conventional wisdom” about Evangelical groups in El Salvador against survey data, including the postulate that they “might become politically emboldened and more capable of fulfilling the role of the proactive democratic citizen” (Aguilar et al. 1993, 130). They found that Evangelicals were less likely to vote in national elections than Catholics but more likely to vote than those of similar socioeconomic status who claimed no religion, leading the authors to conclude: “Protestantism is not necessarily an electorally demobilizing experience” (Aguilar et al. 1993, 136).

Steigenga found that in Guatemala and Costa Rica, religious affiliation was not a reliable predictor of political attitudes and activities” (2002, 122). Gill (2004) considered potential differences between Catholics and Protestants on a “civil participation index.” With null findings, he concluded that “Weber is not at work in Latin America” and that both traditional hypotheses about Latin American Protestants are “according to the evidence seen here, overdrawn” (Gill 2004, 60).

Patterson analyzed political participation among Protestant and Catholics in Brazil and Chile (2005) and political attitudes in both religious groups in Argentina and Chile (2004). He found that Protestants were not more likely to be disengaged, otherworldly, or prone to authoritarianism. Furthermore, Protestants were more likely to be devout than Catholics and more likely to belong to religious organizations. Devotion positively correlated with support of democracy and organization membership positively correlated with political participation (Patterson 2004; 2005). Therefore, Patterson predicted that “as this segment of the population grows it should have positive consequences” (2005, 357).

Chilean sociologist Cristián Parker determined that the tendency for Chilean Evangelicals to maintain dualistic doctrines and avoid political participation was only happening at the lowest education levels (2008). As education levels rose, Evangelicals were more likely to engage civically and politically. Similar to Patterson, Parker predicted that this was good news for “social society and democracy” if Evangelicals remained a “growing force that constructs social capital and networks that knit together democratic citizenship” (2008, 320).

In 2012, Fediakova found that Chilean Evangelicals still maintained some of the apolitical attitudes of their past. However, they were not “totally depoliticized,” instead demonstrating local behaviors of civic engagement—a “democracy from below” (Fediakova 2012b, 45) She concluded that “Evangelicals have great potential for the formation of civil society and the development of democracy in Chile” (Fediakova 2012b, 44). Four years later, Parker concluded that in contemporary Chile, religious choices are not predictive of attitudes on democracy and authoritarianism, nor of overall political engagement (Parker 2016).

2.2. Toward an Updated, Regional Analysis

Previous studies, however rich, are limited in their comparative scope. Latin America is large and diverse. It is difficult to extrapolate findings from a handful of countries, especially when little analysis has been conducted of the large Evangelical populations in Central America. Moreover, thirty years have passed since Stoll and Martin heralded the coming changes to the Latin American religious landscape. Evangelical growth has outpaced even the more ambitious predictions and these communities are not the small, novel, and marginalized ones they once were. This topic remains relevant—scholars of Latin America continue to produce empirical work (Boas 2021; Freston 2008; Parker 2008; Parker and Freston 2012; Parker 2016; Bermúdez 2019; Nevache 2019) on Evangelicalism and democracy and call for more research, specifically quantitative, on the subject (Cleary 2018; Berger 2010; Pérez Guadalupe and Grundberger 2019).

In the edited volume, Evangélicos y Poder en América Latina (Evangelicals and Power in Latin America) featuring authors from across the region, Peruvian sociologist José Luis Pérez Guadalupe claims that 2018 “signified the consolidation of Evangelical churches as the new political actors in Latin America” (2019, 19). Sociological, political, and theological changes in the past decades have enabled Evangelicals to break of their apolitical past accurately described by D’Epinay (Pérez Guadalupe 2019, 49–57). Is this indeed the case? And is this new political action a threat or an advantage for electoral democracy in the region? It is reasonable to assume that a generation later, the early predictions and about the trajectory of Evangelicals and democracy in the region will be adequately testable. A wide-scoping, regional, updated analysis is warranted to test these competing hypotheses once again.

3. Contemporary Concerns about Democracy in Latin America

The designers of the AmericasBarometer surveys utilize the concept of “electoral democracy” in their measures and analysis, as it is the “predominant framework for politics in Latin America” (Zechmesiter and Lupu 2019, 9). Like many social scientists, they draw from Joseph Schumpeter’s classic definition: “The democratic method is that institutional arrangement for arriving at political decisions which realizes the common good by making the people itself decide issues through the election of individuals who are to assemble in order to carry out its will” (1942, 269).³

We find ourselves in a similar time as the 1980s and 1990s in that many are once again sounding the alarm about the future of democracy in Latin America. While these fears are expressed in mainstream new sources and political magazines (Almagro 2019; Schenoni and Mainwaring 2018), scholars also decry the “democratic backsliding” of the region (Bermeo 2016; Lührmann and Lindberg 2019; Rice, Plattner, and Diamond 2015). In a report accompanying the dataset I use for my analyses, the authors begin: “The pulse of democracy in the region remains weak” (Zechmesiter and Lupu 2019, 2). They note that in their most recent wave of data collection, support for democracy declined, satisfaction with the performance of democracy declined, and hypothetical support for coups increased.

The Varieties of Democracy (V-Dem) Research Project echo these concerns. As one of the world’s largest social science data collection projects on democracy, the international, interdisciplinary research team uses over four hundred and fifty indicators to measure democracy around the world. In the most recent report, “Autocratization Turns Viral,” (Alizada et al. 2021), the authors claim that liberal democracy has been in steep decline for the last decade. This decline is “especially prominent” in Latin America (Alizada et al. 2021, 13). These worrisome indicators led one political scientist to conclude: “At present, prospects for democracy in Latin America, whether participatory or representative, are dim” (Goldfrank 2017, 157). It is imperative to understand how the changing religious makeup might impact the anti-democratic trajectory of Latin America. In this study, I consider survey data from sixteen nations to evaluate the following hypotheses:

• H1: Latin American Evangelicals are more likely than Catholics to hold democratic values and demonstrate democratic behaviors

• H2: Latin American Evangelicals are less likely than Catholics to hold democratic values and demonstrate democratic behaviors

4. Methods

4.1. Data

In this analysis, I use data collected from the AmericasBarometer, a survey conducted by the Latin American Public Opinion Project (LAPOP), housed at Vanderbilt University. This regular survey is conducted across thirty-four countries in the Americas and is intended to gauge the public’s experiences with democratic governance. The surveys are national probability samples of voting-age adults involving face-to-face interviews. Participants provide their opinions on public issues related to governance, democracy, and social problems. They also answer questions about personal religious affiliation and practices.

For the purposes of this project, I analyze Spanish-speaking nations in the Americas and Brazil. However, LAPOP does not collect data from Cuba or Puerto Rico, and I decided to exclude Venezuela and Nicaragua as responses to political questions will undoubtedly be shaped by the ongoing political and human-rights crises in those nations. This leaves the sixteen nations of Mexico, Guatemala, El Salvador, Honduras, Costa Rica, Panama, Dominican Republic, Colombia, Ecuador, Peru, Bolivia, Chile, Argentina, Brazil, Paraguay, and Uruguay. I pooled the two most recent sets of data for each nation, collected in 2016/7 and 2018/9. As recommended for analysis of multiple AmericasBarometer surveys, I weighted each individual wave and organized the merged set according to the stratification and clustering included in the survey design (Castorena 2021).

4.2. Dependent Variables

In this paper, the task is to operationalize the terms in H1 and H2: “support for democracy” and “democratic behaviors.” There are several AmericasBarometer questions that measure democratic attitudes and behaviors consistent with an electoral democracy framework and previous literature. I include three variables that measure attitudes: explicit support for democracy (Patterson 2004; Dion and Díez 2017) and commitment to democratic norms (Booth and Richard 1998b).

I also include three variables that measure democratic behaviors: voting, protesting (Patterson 2005) and participation in community meetings (Carreras and Bowler 2019; Seligson 1999; Klesner 2007). When necessary, I recoded the variables so that the positive measure or dummy variable indicates the “more democratic” answer.

For the first, most explicit attitudinal variable, participants were asked: “Democracy may have problems, but it is better than any other form of government. To what extent do you agree or disagree with this statement?” (On a Likert scale of seven). To measure acceptance of the “basic rules of electoral democracy”, the designers of AmericasBarometer ask about the justification of a military coup. A randomly drawn half of respondents in each sample were asked the first follow-up question, and the other half was randomly assigned to receive the second question. (1) “In your opinion would a military coup be justified when there is a lot of corruption?” (2) “In your opinion would a military coup be justified when there is a lot of crime?” For both questions, the answer was: “Yes, it is justified” or “Not, it is not justified.” I generated a new dichotomous variable where respondents who answered “Yes, it is justified” to either question were coded as 1, indicating support for a hypothetical military coup.

The third attitudinal variable measures support for political free speech and the right to dissent. I operationalize this concept by constructing an additive scale derived from four questions where participants were prompted to respond from one (strongly disapprove) to ten (strongly approve). The four questions are as follows: “There are people who only say bad things about the (country) form of government, not just the current (incumbent) government but the system of government. How strongly do you approve or disapprove of such people’s right to vote? (1). How strongly do you approve or disapprove that such people be allowed to conduct peaceful demonstrations in order to express their views? (2). “Still thinking of those who only say bad things about the (country) form of government, how strongly do you approve or disapprove of such people being permitted to run for public office?” (3) “How strongly do you approve or disapprove of such people appearing on television to make speeches?” (4). I added them together to create a composite measure ranging from 4 to 40, with a mean of 22.7. The variable was internally consistent, with a mean Cronbach’s alpha coefficient of 0.76.⁴

The first variable for democratic behaviors measures voting.⁵ Respondents were asked “Did you vote in the last presidential elections of (year of last presidential elections)?” (1) “Yes” (2) “No.” I generated a dichotomous variable where 1 = voted. I generated a similar dichotomous variable indicating “Protested,” drawn from the responses to the question: “In the last 12 months, have you participated in a demonstration or protest march?”

Key to any discussion of political participation is the concept of social capital, the “features of social organization such as networks, norms, and social trust that facilitate coordination and cooperation for mutual benefit” (Putnam 1995, 66). Capital formed in networks of social interaction is “spent” participating in and improving society through civic engagement. In Central America specifically, Booth and Richard found that formal group membership associated positively with every measure of political engagement they tested (1998). As a proxy for social capital, I draw the last behavioral variable from a question measuring frequency of participation in “meetings of a community improvement committee or association.” Respondents answered that they attended them (1) once a week (2) once or twice a month (3) once or twice a year, or (4) never. Most respondents answered “never,” so I recoded this as a dichotomous variable indicating any participation in community meetings.

4.3. Independent Variable

The independent variable in this study is religious affiliation, specifically Evangelical affiliation. Survey participants were asked, “What is your religion, if any?” The interviewer then categorized their response as Catholic, Protestant/Mainline Protestant/Non-Evangelical Protestant, Evangelical and Pentecostal,⁶ Non-Christian Eastern Religions, Traditional or Native Religions, Jehovah’s Witness, Mormon, Jewish, or Other. The interviewers were instructed that if a respondent said that they did not have a religion, they were to probe to see if they should be categorized as Agnostic/Atheist, or None (Believes in a Supreme Entity but does not belong to any religion). In my analyses, I dropped all categories except Catholic, Protestant, and Evangelical. I combined the Protestant and Evangelical categories and generated a dichotomous variable indicating “Evangelical,” with Catholics as the reference group.

4.4. Control Variables

I use a set of demographic characteristics that are standard as control variables for assessing political behavior: sex (male = 1, female = 0), marital status (married = 1, all others = 0), ethnicity, age (continuous, in years), and education (continuous, in years). I also add an indicator for population density (national capital (metropolitan area) = 1, large city = 2, medium city = 3, rural area = 4), as studies have shown that rural communities in Latin America have disparate access to political resources and opportunities (Carreras and Bowler 2019; Carreras and Castañeda-Angarita 2014). As income can be an unreliable proxy for socioeconomic status in low and middle income countries, surveys usually include a series of questions with which to create an asset-based index of household wealth (Poirier, Grépin, and Grignon 2020). Following the recommendation of Córdova (2009) for analysis of the AmericasBarometer, I add quintiles of wealth to my full regression models. Finally, in Latin America, political interest has a significant impact on explicit political behaviors like voting (Carreras and Castañeda-Angarita 2014; Patterson 2005). Therefore, I add a political interest variable to the models for voting and protesting (“How much interest do you have in politics: a lot, some, little, or none?”)

4.5. Plan of Analysis

To evaluate H1 and H2, I analyzed six related dependent variables with one independent variable indicating religious affiliation for each nation. I use binary logistic models for the dichotomous variables, ordinary least squares regression for the continuous free speech and dissent scale, and ordinal regression for the variable evaluating democracy on a Likert scale. For each variable, I estimate two models. The first model only includes religious affiliation as a predictor variable to determine the gross differences between Catholics and Evangelicals for each dependent variable. In the second model, I include appropriate socio-demographic control variables to assess the net group differences. As there are minimal missing data in my predictor variables, I exclude missing cases in all models.⁷

5. Findings

I begin with a basic description of the religious makeup of the region, and socio-demographic characteristics of Catholics and Evangelicals in each nation. Then I present the findings from the simple and multiple regression models for each dependent variable in each nation.

5.1. Religious Makeup

In this region, 82.6% of participants claimed some form of mainstream Christian religion. However, there are significant differences in the religious makeup of the sixteen nations (See Table 2⁸). The wide range in the Evangelical population is demonstrated by the neighboring countries of Guatemala (42%) and Mexico (9%). Most (thirteen) countries are still majority Catholic, but in Brazil and the Central American nations of Guatemala, Honduras, El Salvador and Panama, Evangelicals make up a significant portion of the population. These data from 2016–2019 demonstrate further shifts since 2014, when the scoping and influential Pew Research Center study “Religion in Latin America: Widespread Change in a Historically Catholic Region” published data on religious makeup in the region. The study categorized nations as “Predominantly Catholic (> 70%)”, “Majority Catholic (> 50%)”, “Half Catholic,” and “Less than Half Catholic.” In my sample, five nations change categories, reflecting a shrinking population of Catholics in each case. Bolivia, Argentina, and Panama shifted from “Predominantly Catholic” in 2014 to “Majority Catholic.” Brazil moved from “Majority Catholic” to “Half Catholic,” and El Salvador moved from “Half Catholic” to “Less than Half Catholic.”

Table 2.

Religious makeup of nations in study

Nation	Catholic	Protestant	None	Other
Paraguay	82.22	10.99	3.06	3.73
Mexico	73.44	8.91	8.62	9.03
Peru	71.47	15.16	3.93	9.44
Ecuador	71.31	15.85	7.98	4.86
Colombia	70.63	14.18	8.02	7.17
Bolivia	66.39	21.68	7.19	4.74
Argentina	62.08	16.34	11.42	10.17
Panama	59.99	30.32	4.85	4.85
Costa Rica	59.94	27.05	7.85	5.16
DR	51.86	28.39	14.75	4.99
Chile	51.79	20.5	20.85	6.86
Guatemala	50.58	42.44	3.85	3.13
Brazil	50.39	32.27	8.99	8.35
El Salvador	47.03	36.61	10.24	6.12
Honduras	44.61	38.52	7.65	9.22
Uruguay	36.98	11.36	28.28	23.39

As in other studies, Uruguay is quite an outlier. Uruguay is very secular, with a long history of church and state division. It consistently demonstrates the lowest levels of religious commitment or participation in Latin America and the highest support for liberal policies such as legalized same-sex marriage and abortion (Pew Research Center 2014, 18). In this sample, the “Other” group is so large due to the category of “Agnostic or Atheist” (17.85% of population). Taken together, agnostics, atheists, and “Nones” are the largest category in Uruguay, representing almost half of the population (46.13%).

Table 2 displays sociodemographic characteristics of Catholics and Evangelicals in each nation, including age, education, income, and household wealth quintile.⁹ While there are some exceptions, broad patterns emerge: Evangelicals are younger and less educated than Catholics. They are also less wealthy and have lower incomes, a finding consistent with other research in the region (Boas 2021). For each variable, group differences were statistically significant in most nations.

5.2. Regression Results

To begin the analysis, I ran simple regression models estimating the relationship between each dependent variable and the independent variable of religious affiliation, for all sixteen nations. The results are displayed in Table 3. The first columns display raw coefficients from logistic regression models, and the last display coefficients from ordinal and OLS regression models. The independent variable in all cases indicates “Evangelical,” with Catholics as the reference group.

Table 3.

Socio-demographic characteristics by religion

Nation	Group	Age	Education	Quintile
Mexico	Catholic	41.5	9.58	2.96
	Protestant	39.7	9.53	2.87
Guatemala	Catholic	37.7	8.32	2.98
	Protestant	37.5	8.10	3.04
El Salvador	Catholic	40.6*	9.28*	3.01*
	Protestant	39.2	8.70	2.9
Honduras	Catholic	39.4*	7.90	3.00
	Protestant	36.6	7.77	2.96
Dominican Republic	Catholic	41.9*	9.65	2.98
	Protestant	37.7	9.98	2.98
Costa Rica	Catholic	43.1*	9.58*	3.13*
	Protestant	35.7	8.70	2.63
Panama	Catholic	41.6*	11.05*	3.02
	Protestant	36.2	10.63	2.96
Bolivia	Catholic	39.9*	11.45*	3.04*
	Protestant	37.4	11.02	2.85
Brazil	Catholic	41.3*	8.68	2.93
	Protestant	37	8.67	2.85
Chile	Catholic	43.9*	11.23*	2.99*
	Protestant	42.4	10.21	2.72
Colombia	Catholic	40.7	9.83*	2.96*
	Protestant	39.5	9.31	2.69
Argentina	Catholic	44*	10.90*	3.01*
	Protestant	39.8	10.00	2.43
Ecuador	Catholic	39.2*	11.59*	3.00*
	Protestant	37.3	11.03	2.86
Peru	Catholic	39	11.65*	3.03*
	Protestant	39	10.23	2.67
Paraguay	Catholic	40.2*	9.64	3.00
	Protestant	37.2	9.28	3.03
Uruguay	Catholic	50.2*	10.00*	3.04*
	Protestant	45.6	8.56	2.50

^*Indicates statistical significan

With sixteen nations and six dependent variables, there are a total of ninety-six individual regressions to assess the opposing hypotheses about Evangelicals and democracy in Latin America. Overall, the direction of the coefficients is negative—about two-thirds of coefficients (66) are negative. Twenty-four of these are statistically significant, representing 25% of all regressions. There are thirty-one positive coefficients and only three of those are statistically significant, representing 3.1% of all regressions. These results offer no support for H1 and some support for H2. Evangelicals clearly do not enjoy a democratic advantage in the region—in fact, they are significantly less democratic in some contexts, considering certain variables.

Statistically significant differences were most evident in voting behavior. Fifteen out of sixteen nations have negative coefficients, and they are statistically significant in ten. Beyond this, however, there are few patterns of significance by variable or nation. Support for democracy and support for dissent were significantly negative in four nations each, support for a coup was significant in three, and community meetings and protesting were split with equal positive and negative significance. At the national level, Mexico, Guatemala, Colombia, Peru, and Ecuador had no negative statistically significant differences for any variable. Of the remaining eleven nations, four displayed statistically significant differences in three of the six variables. In only one nation, Costa Rica, were Evangelicals significantly negatively correlated with a majority of the democratic variables.

In the second model, I introduce the full set of control variables including sex, marital status, ethnicity, age, education, population density, and wealth quintile. The results are displayed in Table 4.¹⁰ For voting and protesting, I also include political interest. While the direction of the coefficients remains largely unchanged, sixteen of the twenty-four negative coefficients lost their significance.

Table 4.

Simple regression of democratic behaviors and values on Evangelical affiliation

Nation	Voting (logistic)	Protesting (logistic)	Community meetings (logistic)	Support for coup (logistic)	Support for democracy (ordinal)	Support for dissent (ols)
MEX	−0.157	−0.029	−0.274	−0.123	−0.014	0.009
GUA	0.014	−0.258	−0.070	0.113	0.039	−0.004
SAL	−0.265**	−0.143	−0.161	−0.086	−0.065	0.000
HON	−0.208*	−0.268	−0.228*	−0.026	−0.203**	−0.004
DR	−0.525***	−0.407**	0.039	0.141	0.105	−0.019*
CR	−0.608***	−0.116	−0.407***	−0.354***	−0.351***	0.002
PAN	−0.381***	0.350*	−0.009	−0.050	0.007	0.011
BOL	−0.233***	0.139	0.102	−0.035	0.079	−0.015*
BRA	−0.294**	0.147	−0.165	−0.189	0.136	0.014
CHI	−0.229*	−0.205	0.258*	−0.197	−0.121	−0.023*
COL	−0.122	−0.054	0.039	−0.108	0.115	0.015
ARG	−0.290*	0.361	−0.043	−0.307*	−0.322**	−0.001
ECU	−0.207	−0.296	−0.213	0.069	0.031	−0.006
PER	−0.195	0.259	0.279*	−0.063	−0.089	0.011
PAR	−0.553***	−0.042	−0.216	−0.041	0.137	0.024
URU	−0.391	−0.278	−0.440	−0.549**	−0.487***	−0.032*

^***p<0.01,

^**p<0.05,

^*p<0.1

This suggests that most of these differences are due to the distinct demographic profiles of each group, not due to their religious affiliation.

Two negative coefficients also gained statistical significance with the addition of the control variables, leaving ten statistically significant negative coefficients, or 10.4% of all regressions. While two positive coefficients lost their significance, two more gained statistical significance in model 2, leaving three statistically significant positive coefficients.

These are largely null findings. There is insufficient evidence to support either hypothesis that Latin American Evangelicals are more or less likely than Catholics to hold democratic values and demonstrate democratic behaviors This leads me to reject both H1 and H2. Additionally, when democracy was evaluated directly, there were no significant differences between Evangelicals and Catholics in any nation. In model 1, Evangelicals were significantly less approving of democracy in four nations, but all four of these coefficients lost their significance with the addition of control variables. Therefore, while Evangelicals are less approving of democracy in those nations, this is again due to their distinct demographic characteristics, not their religious affiliation.

In addition to the overall evaluation of the hypotheses, there are several patterns of interest in these data. First, Evangelicals are significantly less likely to vote in five nations: El Salvador, Honduras, Costa Rica, Dominican Republic, and Paraguay. Holding all else constant, Evangelicals in these countries range from 13% to 32% less likely¹¹ to have voted in the last election, in the unit of odds. Voting is the only variable with a negative pattern across nations and it describes the majority of nations in Central America and the Caribbean, but only one nation in South America.

At the national level, there are few patterns. In fourteen of the nations, there is no pattern of significance (more than variable in either direction). However, in Costa Rica, Evangelicals are significantly less likely to vote and attend community improvement meetings. In the Dominican Republic, Evangelicals are less likely to vote, protest, and support the right to dissent. However, they are also less likely to justify a potential military coup.

6. Discussion

As Evangelical growth exploded throughout Latin America, scholars predicted how this growth would impact the social and political climate of the region. While some anticipated that the movement would produce active citizens and supporters of democracy, others feared that it would lead to large communities of passive, disinterested citizens tolerant of authoritarianism. This study tested these theories with recent AmericasBarometer survey data from sixteen countries across the region.

I found insufficient evidence that Evangelicals are less likely than Catholics to hold democratic values and demonstrate democratic behaviors. If Latin American Evangelicals began as small, inwardly focused minoritized communities uninterested in the public square (Burity 2020), it is clear that is no longer the case. As Pérez Guadalupe concludes, “Evangelicals are now present in political parties in every country on the continent and they are political actors each of them” (2019, 164). However, there is also insufficient evidence to conclude that Evangelicals in Latin America are more likely than Catholics to hold democratic values and demonstrate democratic. The hopes and fears that the Evangelicals movement would have a significant effect (negative or positive) on democratic participation in the region were, it seems, overstated.

Given the disruptive nature of Evangelical growth and the massive influence of this movement in family life, society, and even politics (Boas 2020), these results are perhaps surprising. In the earliest periods of Evangelical growth, scholars made bold predictions about the exponential growth and political ramifications of the movement. Surprisingly, even the most outrageous predictions of growth fell short of the reality—Evangelicals grew faster and larger than almost anyone expected. However, the similarly confident claims about their impact on democracy would prove misguided. This could be a lesson for scholars studying “sects” or other new religious and social movements. Despite their novelty or fervor, growing groups are still made up of individuals immersed in the culture and norms of their context. The passage of time, the transmission of values, and the rule of the regression to the mean will temper the impact of even the most divergent groups. Perhaps this realization might even temper the sweeping pronouncements of the professional prognosticators.

This is not to say that religion is inconsequential to democracy and civic engagement in the region. Several patterns in the data indicate that there are indeed more “stories to be heard” about the interplay of Evangelical religion and democracy in Latin America. First, there is strong evidence that Evangelicals in Central America and the Caribbean are less likely than Catholics to vote. This is true for both Costa Rica and Panama, and Honduras and El Salvador, which represent two poles of the religious landscape in Central America. El Salvador is a “new center of global Evangelicalism” (Offutt 2015). As in Honduras, Evangelicals have sky-rocketed to almost 40% of the population and hold considerable political and cultural influence. On the other hand, Costa Rica and Panama are the least Evangelical nations in this sub-region and Catholicism is the official state religion in both nations. The consistency of this finding across different religious landscapes is particularly compelling.

Pérez Guadalupe offers three “regional models” to explain Evangelical political participation in Latin America: Central American, South American, and Brazilian (2019). He considers the differences between how Evangelicals across Latin America decide how to organize politically, what to prioritize at the bal-lot box, and who to vote for. In the smaller Evangelical communities of South America, religious discourse is not central to electoral decision-making. In Brazil, Evangelical churches or organizations sponsor official parties and candidates that vie for the votes of the faithful. In Central America, however, larger, and more religious groups of Evangelicals are motivated by a “moral agenda,” forming a large bloc of value voters, together with traditional and charismatic Catholics (Pérez Guadalupe 2019, 172). These prolife, pro-family movements are usually led by Evangelicals, who channel these values into electoral politics with more skill than Catholics. Central America is the region where, in the future, Evangelicals will most likely form part of the electoral majority and themselves be elected to national office (Pérez Guadalupe 2019).

While Pérez Guadalupe does not consider the decision to vote (or abstain), these issues are closely related. It is telling that my findings also suggest that Central American Evangelicals are distinct from their regional counterparts in their electoral decision-making. One of the best-known theories on the decision to vote is the pivotal voter model, the “probability that a citizen votes increases with the probability that her vote is pivotal” (Downs and Downs 1957; Blais and Hortala-Vallve 2016, 117). This rational-choice framework also explains the “underdog effect,” where voters supporting the less-popular option participate at higher rates (Levine and Palfrey 2007). Although they are still a minority in most of the sub-region, Evangelicals are more likely to lead and represent the large Catholic-Evangelical value-voter blocs. A possible explanation for this paper’s finding on voting is that Evangelicals in Central America feel well-represented by their movement’s leadership and therefore reason that their individual vote is less pivotal. Central American Catholics, on the other hand, would have a rational motivation to continue to participate, to ensure that their distinct interests were represented in a political and national context where Evangelicals pose a perennial threat to Catholic monopoly (Pérez Guadalupe 2019, 165).

There is also a “national story” to be told about the Dominican Republic, which alone represented more than one-third of the statistically significant multiple regression coefficients. Here, Evangelical are less likely to vote, protest, and support the right to dissent than Catholics. However, they are also less likely to justify a potential military coup. While it is out of the scope of the present study, there are undoubtedly rich details about how the religious and political complexities of this context could lead to such stark, and divided, findings. While further quantitative work is warranted in this case, rich qualitative analysis is essential to elucidate these findings. This example of the Dominican Republic is a strong reminder that there are rich data in individual Latin American contexts through which to explore the complexities around religion, democracy, and civic engagement—including and beyond the research questions addressed in this paper. Scholars should continue to prioritize the kind of local, careful analysis that has been essential for understanding these topics (Brusco 1993; Alves 1985; Sherman 1997; Garrard-Burnett and Stoll 1993; Freston 2008; Bermúdez 2019; 2019; Zúñiga Ramírez 2019).

Lastly, this study has several important limitations. In many cases, there are significant differences between how different Evangelical communities approach political engagement. Survey data that combines all Evangelical and Pentecostal groups together obscures important distinctions between denom-inational groups. Second, this study excludes the “Nones,” or those who report no religious affiliation. Nones are also experiencing unprecedented growth in Latin America, albeit slower than Evangelicals (Somma, Bargsted, and Valenzuela 2017; Johnson and Crossing 2020). The group has doubled (from 4 to 8%) in the last generation (Pew Research Center 2014) and is projected to grow 44% between 2010 and 2050, adding twenty million more Nones and challenging both Evangelical and Catholic influence (Hackett et al. 2015). Religious affiliation and behaviors are linked to higher voting rates and higher participation in civic activities (Driskell, Embry, and Lyon 2008; Jones-Correa and Leal 2001; Verba, Schlozman, and Brady 1995) and in Latin America religious affiliation is also a predictor of political party identification (Díaz Domínguez 2020). To understand how religious change will impact the health of democracy in Latin America, it is imperative to include an empirical focus on the Nones.

Table 1.

Descriptive statistics

Variable	Description	Mean or %	SD
Democracy is the best form of Government	1 = completely disagree, 7 = completely agree	4.78	1.71
Justification of hypothetical coup	1 = justify coup	59.11	—
Support for free speech and dissent	Index; min = 4 to max = 40	22.66	8.40
Voting	1 = voted in last election	75.01	—
Protesting	1 = protested in the last 12 months	10.53	—
Community Meetings	1 = attends community improvement meetings	32.15	—
Protestant	1 = Protestant	27.84	—
Male	1 = male	50.15	—
Married	1 = married	31.02	—
Mestizo	1 = mestizo	46.92
White	1 = white	27.56	—
Indigenous	1 = indigenous	6.90	—
Black	1 = black	5.58	—
Mulato	1 = mulato	6.55	—
Other	1 = other	6.51	—
Age	In years; min = 16 to max = 112	39.87	16.23
Education	In years; min =	10.05	4.25
Capital city metropolitan area	1 = capital city	22.24	—
Large city	1 = large city	19.91	—
Medium city	1 = medium city	17.57	—
Small city	1 = small city	14.12	—
Rural area	1 = rural area	26.17	—
Income	1 = lowest, 4 = highest	2.31	1.16
Quintiles of wealth	1 = lowest, 5 = highest	2.97	1.40
Political interest: none	1 = none	35.60	—
Political interest: small amount	1 = small amount	30.72	—
Political interest: some	1 = some	21.10	—
Political interest: a lot	1 = a lot	12.58	—

Table 5.

Multiple regression of democratic behaviors and values on Evangelical affiliation

Nation	Voting (logistic)	Protesting (logistic)	Community meetings (logistic)	Support for coup (logistic)	Support for democracy (ordinal)	Support for dissent (ols)
MEX	−0.081	−0.010	−0.338*	−0.057	0.067	0.235
GUA	0.078	−0.283	−0.034	0.116	0.053	−0.160
SAL	−0.252*	−0.267	−0.144	−0.035	0.016	−0.047
HON	−0.133*	−0.323	−0.146	0.007	−0.116	−0.093
DR	−0.380**	−0.462**	0.090	0.261*	0.132	−1.164**
CR	−0.224*	−0.042	−0.249*	−0.011	−0.139	0.300
PAN	0.012	0.259	0.171	−0.050	0.120	0.717
BOL	−0.031	0.051	0.038	0.093	0.111	−0.456
BRA	−0.183	0.295	−0.068	−0.259*	0.175	0.156
CHI	−0.120	−0.143	0.294*	−0.069	−0.015	−0.428
COL	−0.017	−0.193	0.078	−0.066	0.164	0.972
ARG	−0.088	0.428*	0.018	−0.057	−0.046	0.415
ECU	−0.056	−0.196	−0.123	0.120	0.058	−0.326
PER	−0.039	−0.053	0.194	0.040	−0.048	0.410
PAR	−0.460*	−0.053	−0.150	−0.022	0.189	0.910
URU	−0.026	−0.043	−0.289	−0.213	−0.089	−0.656

^***p<0.01,

^**p<0.05,

^*p<0.1