Re-examining Restructuring: Racialization, Religious Conservatism, and Political Leanings in Contemporary American Life

Abstract

This paper assesses the continued relevance of Robert Wuthnow’s seminal theory of “religious restructuring” for explaining the relationship between religious conservatism and political allegiances in the contemporary United States. Employing a comparative approach, we evaluate the link between doctrinal conservatism (or liberalism) and political conservatism across the seven largest US religious traditions, including Islam. We find that for most Christians and Jews, doctrinal conservatism continues to be tightly linked with conservative political attitudes, even after adjusting for demographic differences and religiosity. For Muslims, Black Protestants, and Latinx Catholics however, doctrinal conservatism is unlikely to be associated with political conservatism. In short, Wuthnow’s theory still holds, but only for religious traditions that are majority white. We speculate that being “racialized religious traditions” explains the lack of restructuring we observe among Muslims, Black Protestants, and Latinx Catholics. External social and political pressures have kept unifying racialized religious identities salient for each of these traditions, preventing the internal bifurcation still characteristic of other major American religions. Our findings and approaches contribute to the two growing trends within the sociological study of religion—the analytical integration of considerations of race and racial politics into scholarship on religious life (called “complex religion”) and a recognition of the importance of cultural “styles” of religion in shaping political and social behaviors.

Introduction

In an era of political division in American life, as evidenced by the closely contested nature of recent elections, an ideologically fractured media landscape, and the polarizing figure of President Donald Trump, it is important to understand the various coexisting social conditions and cultural forces that contribute to such ideological divergence. The most visible post-2016 election social scientific analyses focused on issues of class grievance or racial identity (e.g., Anderson 2016; Hochschild 2016), but recent research suggests that religion is also bound up with our current political divide. One strain of this scholarship posits an affinity between certain forms of ethno-nationalist religious identity and support for Donald Trump (Edgell 2016; Gorski 2017; Whitehead, Perry, and Baker 2018a). Another suggests that the growing gap in religious participation between the unaffiliated “nones” and the highly religiously committed in America is a strong driver of political differences (Hout and Fischer 2014; Schnabel and Bock 2017). What has been missing is a more comprehensive analysis of the relationship between particular dimensions of religious life and contemporary political attitudes across diverse religious traditions,¹ as well as a consideration of how these religiously informed political attitudes may be simultaneously impacted by issues of racial status and racialized politics (Wilde 2018; but see Edgell and Tranby 2007; Perry and Whitehead 2015).

In two highly influential books, Robert Wuthnow claimed that after the 1960s, there was a dramatic realignment of religious and political affiliations within the United States, such that the most significant ideological and social division became the one between those who take a conservative approach to both religion and politics (“religious conservatives”) and those who take a more liberal approach to both (“religious liberals”) (1988, 1989). He called this process the “religious restructuring” of American life. Using large, nationally representative survey datasets from the Pew Research Center, this paper will test the continued relevance of Wuthnow’s theory and, in particular, assess whether the religious restructuring observed by Wuthnow among majority-white religious traditions is equally evident within racialized religious traditions in the contemporary United States. We use the term “racialized religious traditions” to mean religious traditions whose historical development and social status have been strongly shaped by the minority racialized status of their members and institutions.² By incorporating into our analysis data on three such traditions in the United States—Black Protestants, Latinx Catholics, and (for the first time) American Muslims—we employ the intersectional theoretical lens of “complex religion,” which holds that racial inequalities and religious forces often work together to produce political attitudes, actions, and outcomes (Wilde 2018). While Wuthnow’s theory was historical, cross-denominational, and generally accepted and retested (Hunter 1991; Kellstedt and Smidt 1993; Layman 2001; Ellis and Stimson 2012), it did not take race and racialized religious groups into account in either its original formulation (Wuthnow 1988, 1989; cf. Williams 2018) or later articulations (Hunter 1991; Wuthnow 1996; Ellis and Stimson 2012). In this sense—at least from the perspective of “complex religion”—the theory may be analytically incomplete.

Wuthnow and Religious Restructuring

For over 30 years, the most influential theoretical framework for understanding the relationship between religion and political attitudes in American life—in both sociology and political science—has been Robert Wuthnow’s notion of the “restructuring of American religion” (1988). Wuthnow argued that due to social and political factors including the expansion of higher education, the politically divisive Vietnam War and Civil Rights Movement, the rise of religiously affiliated special interest groups, and the decline of denominationalism, the relationship between religion and political orientations in American life had been fundamentally “restructured.” Whereas, previously, a person’s religious denomination or tradition would be a primary source of identity and would be congruent with his/her political attitudes, after the 1960s, the most fundamental divisions in American life were political rather than religious. And these political divisions now cut through religious traditions. Catholics, for example, who once reliably voted Democratic at rates of 80–90%, became split almost evenly between the two parties after the 1960s (Wald and Calhoun-Brown 2014), and, like the other traditions studied by Wuthnow, have remained internally divided over issues such as abortion, homosexuality, gender rights, and racial inequality ever since (McGreevy 1998; Layman 2001).

The two broader camps that emerged from this transformation—Wuthnow’s “religious conservatives” and “religious liberals”—demonstrated novel alignments of religious life and political orientation. It was not just the reconfigured political split that was significant but the particular relationship between religious style and political attitudes that each side of this division made manifest. On one side, a more traditional approach to religious life—centered mainly on what we call doctrinal conservatism—became linked with right-leaning takes on government spending, defense, and Republican party support. On the other hand, a more pluralistic religious style in which religious interpretation was open and relatively fluid (our “doctrinal liberalism”) was wedded to left-leaning approaches to political issues. This recombination of styles and attitudes was made manifest in increasingly salient social identities based on religious-political ideology rather than denominational affiliation. For Wuthnow, an important part of what made these novel identities possible was the declining significance of denomination as a competing and constraining basis of identity (Wuthnow 1988, 1989).

While Wuthnow’s original work portrayed the ideological division between religious conservatives and liberals as relatively even, clean, and characteristic of US society in general, later scholars established that the qualities of the conservative-liberal divide varied by the specific denominational or racial context being analyzed. Demerath and Yang (1997), for example, found that tradition made a significant difference in precisely how religion was restructured, as evangelical Protestants were almost evenly split between religious conservatives and religious liberals, while members of other traditions clustered more toward the liberal end of the scale. Layman (2001) also finds variation across religious tradition, with evangelical Protestants demonstrating the strongest relationship between doctrinal conservatism and right-leaning politics, Catholics and mainline Protestants showing a weaker version of the same relationship, and Black Protestants instead showing a link between greater religious commitment and Democratic politics. In terms of racial variance, scholars have found that the divide between religious conservatives and liberals is not as straightforward among African Americans as Wuthnow portrays it to be among the general population, with African American religious people more likely than others to simultaneously hold conservative social views and politically liberal attitudes toward economics and civil rights (Davis and Robinson 1996; Edgell and Tranby 2007). This scholarship draws attention to the ways in which Wuthnow’s original analysis did not account for the role of particular traditional and/or racial contexts in shaping religious-political divisions. Our paper follows the leads of these studies by asking the question: how might the racialized status of religious traditions condition the congruence of religious and political conservatisms among modern religious Americans?

Doctrinal Conservatism (and Liberalism) as Religious Styles

Like other social scientists, we disentangle three primary dimensions of religion within our analysis: the tradition to which one belongs, one’s level of commitment to religious life,³ and one’s approach to religious doctrine or theology (conservative or liberal) (Davis and Robinson 1996; Layman 2001; Ellis and Stimson 2012; Schnabel and Bock 2017). While our analysis includes all three dimensions of religion in order to assess their relative importance, our primary relationship of interest is the one between doctrinal conservatism (or liberalism) and political attitudes. Although the dimensions of religious commitment and religious affiliation have garnered greater attention in recent scholarly and journalistic attempts to understand American political division, we consider doctrinal conservatism (or liberalism) more directly relevant to questions of political attitudes based on its theoretical and empirical import in previous scholarship. It is doctrinal approach, rather than religious affiliation or religious commitment, that Wuthnow pinpoints as the most central religious factor distinguishing religious conservatives from religious liberals (1988, 133, 202, 213; 1989, 178) and contributing to the intra-tradition divisions that characterize the religiously restructured landscape. Indeed, one’s approach to religious doctrine has been clearly and consistently linked to political attitudes in empirical tests, even when dimensions of religious affiliation (tradition) and commitment (religiosity) are also considered (Kellstedt and Smidt 1993; Layman 2001; Ellis and Stimson 2012; Perry and Whitehead 2015; Stewart, Edgell, and Delehanty 2018).

Following sociologists of religion, we define doctrinal conservatism as the tendency to treat the texts, teachings, and traditions of one’s religion as self-evidently true, strongly authoritative, and lacking the need for extensive human interpretation (Wuthnow 1988; Hunter 1991; Ellis and Stimson 2012). Doctrinal liberalism, in contrast, involves treating the texts, teachings, and traditions of one’s religion as historically situated, in dialogue with other bases of cultural authority in modern life (i.e., science, rationality), and open to multiple credible interpretations (Wuthnow 1988; Hunter 1991; Aslan 2006). In our usage, doctrinal conservatism and liberalism are “styles” of religious life (Wuthnow 1989; Brophy 2016)—cultural modes of religion involving particular approaches to practice and/or belief that are neither identical to nor definitively measurable by denominational affiliation, level of religiosity, or religious subgroup (Stewart, Edgell, and Delehanty 2018, 18).⁴ As transposable forms of culture (Edgell and Tranby 2007), religious styles can spread and be found across different traditions and religious and demographic subgroups (i.e., Stewart, Edgell, and Delehanty 2018, 25; Whitehead, Schnabel, and Perry 2018b, 6). Although their proportions vary, sociologists have identified doctrinal conservatives and doctrinal liberals among American Jews (Cadge and Davidman 2006), Muslims (Read 2008), Evangelicals (Bielo 2009), and Catholics (Dolan 2002).

Implicit in Wuthnow’s original project is the assertion that certain styles of religion can become aligned with certain political attitudes. Sociologists of religion and culture have recently shown that three cultural styles of religion in contemporary American life—Christian nationalism, public religious expression, and secularized Evangelical discourse—are all strongly predictive of conservative political attitudes and voting behaviors, especially among white Americans (Frost and Edgell 2017; Delehanty, Edgell, and Stewart 2018, 17; Stewart, Edgell, and Delehanty 2018; Whitehead, Perry, and Baker 2018a).⁵ This work establishes the analytical importance of religious styles, demonstrates the congruence of certain contemporary styles with conservative politics, and suggests that the effect of these styles on politics is often conditioned by race (Edgell and Tranby 2007; Frost and Edgell 2017). Our findings contribute to the growing evidence that religious styles in American life are effective at predicting political views, attitudes, and behaviors, in ways that are empirically independent from religious affiliation, level of religious practice, and even religious subgroup (Frost and Edgell 2017; Delehanty, Edgell, and Stewart 2018; Stewart, Edgell, and Delehanty 2018; Whitehead, Perry, and Baker 2018a).

Race, Complex Religion, and the Restructuring Thesis

Our paper addresses a question hinted at but not answered in the previous scholarship: does the relationship between doctrinal conservatism and political attitudes differ by whether a religious tradition is racialized? Do the members of minority racialized religious traditions demonstrate the same congruence between religious style and political attitudes that Wuthnow treated as a generally American social phenomenon? While no one has answered this question with a comprehensive comparison, previous studies focusing on particular communities or comparing a few groups suggest mixed results. Research assessing the relationship between doctrinal conservatism and political attitudes among Black Protestants has found no straightforward relationship between the two (Layman 2001; Hirschl, Booth, and Glenna 2009), but instead a complex combination of social conservatism (on issues such as abortion, sexuality, gender, and religion in public life) and political liberalism (including support for social welfare policies and the Democratic party) (Davis and Robinson 1996; Wald and Calhoun-Brown 2014). Research on Latinx Catholics reveals combinations of conservatism on moral issues and liberalism on economic policy, minority rights, and voting (Stevens-Arroyo 1998; Kelly and Kelly 2005; Wald and Calhoun-Brown 2014). Recent studies comparing black and white conservative Protestants (as well as white and Latinx Catholics) show racial minority membership to be a powerful predictor of people’s views of racial inequality, despite sharing the same “religious cultural tool kit” (Edgell and Tranby 2007; Frost and Edgell 2017). These studies suggest that racialized religious traditions in the United States may have a different relationship with the restructuring forces theorized by Wuthnow than do majority-white traditions. We aim to settle this question through a systematic comparative analysis across seven American religious traditions—four majority-white and three racialized traditions, including American Muslims.

Questioning whether religious restructuring may vary by the racialization of the religious tradition, we employ a theoretical orientation toward the intersectional study of religion and race (and other inequalities) called “complex religion.” Conceptualized by Melissa Wilde, complex religion urges sociologists to attend to the ways that religious identities and institutions are frequently “deeply intertwined” with powerful structures of inequality within US society—race, class, and gender (Wilde 2018). This empirical complexity means that social phenomena that initially appear to be either about religion or race, class, or gender may in fact involve religion and one or more of these structures of inequality working in tandem. For Wilde, this situation calls for an intersectional approach attuned to the potential impact of multiple social forces at the “starting point” of analysis (Wilde 2018, 296). In our case, recent surveys of large probability samples of US Muslims (Pew Research Center 2007a, 2011) allow us to interrogate the intersection of race and religion with politics, by including a total of three racialized religious traditions into our analysis.

While American Muslims, Black Protestants, and Latinx Catholics have different histories, social trajectories, and internal diversities, these traditions share similar experiences of, racialization being marked by race or ethnicity as a salient and socially consequential master status; social marginalization, being seen as inherently different by much of mainstream, white American society; and institutional segregation, practicing religion in spaces and places separate from white Americans (Stevens-Arroyo 1998; Emerson and Smith 2000; Pew Research Center 2007b; Hirschl, Booth, and Glenna 2009; Selod and Embrick 2013; Wald and Calhoun-Brown 2014; Lipka 2017; Selod 2018). Each of these traditions and/or their members has been the target of active symbolic boundary work and discriminatory political discourse by white conservative politicians and associated organizations, some with an explicitly Christian right-wing agenda (Edgell and Tranby 2007, 266–7; Crespino 2008; Bail 2015; Cobb, Perry, and Dougherty 2015, 192; Balmer 2017; Braunstein 2017; Whitehead, Perry, and Baker 2018a). Given these similar experiences of racialization and symbolic marginalization, it seems plausible that Black Protestants, Latinx Catholics, and Muslims will show little of the congruence between religious styles and political attitudes observed by Wuthnow. We hypothesize instead that heightened and unifying racialized religious identities will prove more powerful than internal political differences, leaving these traditions lacking clear divisions between religious-political conservatives and liberals. If we do find such variation by group racialization, it will suggest that the emphasis placed by Wuthnow on religious restructuring as central to contemporary American social life may need to be reconsidered, taking the effects of race—and the diversity of the modern American religious landscape—more fully into account.

We proceed, first, by assessing the prevalence of doctrinal conservatism (and liberalism) within the seven largest religious traditions in the United States: mainline Protestantism, non-Latinx Catholicism, Latinx Catholicism, evangelical Protestantism, Black Protestantism, Judaism, and Islam. We show the degree to which demographic inter-religious differences explain variation in doctrinal conservatism, thereby updating current knowledge on these relationships and extending it to include American Muslims. Next, we look at the relationship between doctrinal conservatism and political conservatism within each tradition while still accounting for religiosity and demographic differences. This allows us to test whether the restructuring relationships hypothesized by Wuthnow are evident, such that doctrinal and political conservatisms are aligned.

Data

The Pew Research Center conducted the first ever surveys of a probability sample of Muslims in the United States in 2007 (N = 1,050) and 2011 (N = 1,033). We combine these data on American Muslims with another Pew Research Center survey that samples the general population—the US Religious Landscapes Survey (N = 35,556).⁶ This gives us the unique advantage of comparing across religious traditions while including Muslims, since several questions were identical across the general and Muslim questionnaires. After listwise deletion of respondents missing on either dimension of religious conservatism or any of our demographic controls, our sample size was 25,286, including 6,463 mainline Protestants, 8,038 evangelical Protestants, 5,536 non-Hispanic⁷ Catholics, 1,370 Hispanic Catholics, 588 Jews, 1,673 Black Protestants, and 1,618 Muslims.

Variables

Doctrinal conservatism⁸

We sought a measure of doctrinal conservatism that has been used in other literature and that would hold comparable meaning across the religious traditions in our analysis. The most common proxy for doctrinal conservatism is Biblical literalism. While measures of Biblical literalism and authority are suitable for capturing religious conservatism in a Christian context, this has been critiqued as inappropriate for other contexts, even when the question wording is broad (e.g., referencing “Word of God” or literalism in “Holy Books”) (Casanova 1995; Mohamed 2011). Equating Quranic literalism with Biblical literalism is particularly problematic, because the Quran and the Bible have such radically different histories as texts as well as play different roles within their respective faiths. For example, reading and reciting the Quran is a central part of Muslim worship across all sects, whereas scholars have argued that reading the Bible is a more recent phenomenon in Christian traditions.⁹

Instead, in keeping with the definitions of doctrinal conservatism and liberalism specified above and formed in consultation with the sociology of religion, we chose a proxy measure that placed emphasis on the respondent’s general orientation toward the central sources of meaning within one’s religion, whether they be practices, texts, or traditions (glossed in the survey question as “teachings”).

Therefore, we use responses to the following question: Now, as I read a pair of statements, tell me whether the FIRST statement or the SECOND statement comes closer to your own views even if neither is exactly right. There is only ONE true way to interpret the teachings of [my religion/Islam],¹⁰  OR there is MORE than one true way to interpret the teachings of [my religion/Islam]. Those that answered one true way were coded as 1 on the conservatism variable, whereas those who answered more than one true way were coded as 0.¹¹ This measure captures the difference between those who tend to view their religion’s teachings as stable in nature, self-evidently true, and lacking the need for extensive human interpretation and those who allow for multiple, changing, and historically situated understandings of their religion’s teachings, but without hinging this distinction on the nature of the holy book in one’s faith.

Political conservatism

We use two measures of political conservatism—self-identified political ideology and party identification.¹²

Survey takers asked: In general, would you describe your political views as very conservative, conservative, moderate, liberal, or very liberal? Responses were reverse coded such that the resultant measure ranges from 1 to 5 where 5 is very conservative and 1 is very liberal. Respondents were also asked: In politics TODAY, do you consider yourself a Republican, Democrat, or Independent? If respondents answered independent, no preference, other, or do not know, they were asked: As of today, do you lean more to the Republican party or more to the Democratic party? Our measure combined responses to these two questions, such that those who said they were Republican or leaned Republican were given a 3, those who answered independent or neither in both the initial question and the follow-up were given a 2, and those who said they were Democrat or leaned Democrat were given a 1. Those who refused to answer the follow-up question were coded as missing.

Religious tradition and religiosity

Our primary independent variable is a modified version of religious tradition, for which we include a set of indicator variables with mainline Protestant as the reference category. Other groups in the analysis are evangelical Protestant, Catholics (divided into Hispanic and non-Hispanic),¹³ Black Protestant,¹⁴ Muslim, and Jewish.¹⁵ We also include a variable for religiosity, the construction of which is detailed in Supplementary material, Appendix A.

Demographic variables

In both parts of our analysis, we control for demographic variables that may differ within traditions and may be associated with the conservatisms—both doctrinal and political—in which we are interested. We control for age, race, gender, level of education, whether the individual is foreign-born, marital status, the presence of children in the household, and region of residence. Both political attitudes and engagement with and approach to religious life may vary with each of these variables (Layman 2001). Table 1 displays the means and standard errors on all variables for the analytic sample as a whole and divided by religious tradition. Though the compositional differences between groups are marked, we found that demographic controls held little explanatory power when it came to doctrinal or political conservatism for each group.

Table 1.

Means and Standard Errors on All Variables

	Mainline protestants	Evangelical protestants	Non-hispanic catholics	Hispanic catholics	Jews	Black protestants	Muslims	All
Doctrinal conservatism	0.13	0.45	0.18	0.23	0.04	0.41	0.34	0.28
	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
Political ideology (continuous)	3.22	3.60	3.27	3.09	2.74	3.17	2.88	3.31
	(0.01)	(0.01)	(0.01)	(0.03)	(0.04)	(0.02)	(0.02)	(0.01)
Political ideology (categorical)
Very liberal	0.03	0.02	0.03	0.06	0.09	0.07	0.07	0.04
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
Liberal	0.15	0.08	0.13	0.21	0.30	0.16	0.22	0.13
	(0.00)	(0.00)	(0.00)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Moderate	0.43	0.31	0.43	0.39	0.40	0.40	0.49	0.39
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Conservative	0.34	0.45	0.35	0.29	0.17	0.28	0.18	0.36
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Very conservative	0.05	0.14	0.06	0.06	0.03	0.09	0.04	0.08
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.00)	(0.00)
Party identification (continuous)	2.01	2.22	1.95	1.63	1.54	1.28	1.37	1.94
	(0.01)	(0.01)	(0.01)	(0.02)	(0.03)	(0.02)	(0.02)	(0.01)
Party identification (categorical)
Democrat	0.44	0.34	0.47	0.60	0.69	0.83	0.75	0.48
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Independent or no preference	0.10	0.10	0.11	0.16	0.08	0.07	0.13	0.10
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
Republican	0.45	0.56	0.42	0.23	0.23	0.11	0.12	0.42
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Religiosity	−0.04	0.76	0.13	0.13	−1.04	0.88	0.14	0.31
	(0.02)	(0.01)	(0.02)	(0.03)	(0.05)	(0.02)	(0.03)	(0.01)
Female	0.55	0.54	0.55	0.56	0.49	0.61	0.44	0.55
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Immigrant	0.04	0.04	0.05	0.52	0.09	0.03	0.70	0.11
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.00)	(0.01)	(0.00)
Age
18–29	0.07	0.10	0.09	0.25	0.10	0.19	0.24	0.11
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
30–39	0.12	0.14	0.13	0.26	0.10	0.15	0.24	0.15
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
40–49	0.19	0.20	0.21	0.20	0.17	0.20	0.26	0.20
	(0.00)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)

(Continued)

Table 1.

Continued

	Mainline protestants	Evangelical protestants	Non-hispanic catholics	Hispanic catholics	Jews	Black protestants	Muslims	All
50–59	0.23	0.22	0.23	0.15	0.25	0.21	0.16	0.22
	(0.01)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
60–69	0.18	0.17	0.18	0.08	0.19	0.15	0.08	0.16
	(0.00)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Over 70	0.22	0.17	0.17	0.06	0.20	0.11	0.03	0.16
	(0.01)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.00)	(0.00)
Race
White	0.95	0.89	0.94	0.69	0.97	0.03	0.30	0.81
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.00)	(0.01)	(0.00)
Black	0.02	0.05	0.02	0.05	0.01	0.96	0.22	0.11
	(0.00)	(0.00)	(0.00)	(0.01)	(0.00)	(0.00)	(0.01)	(0.00)
Asian	0.01	0.01	0.02	0.02	0.00	0.00	0.29	0.03
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(.)	(0.01)	(0.00)
Other	0.02	0.04	0.01	0.19	0.01	0.01	0.18	0.04
	(0.00)	(0.00)	(0.00)	(0.01)	(0.00)	(0.00)	(0.01)	(0.00)
Do not know/refused	0.01	0.01	0.01	0.05	0.02	0.00	0.01	0.01
	(0.00)	(0.00)	(0.00)	(0.01)	(0.01)	(0.00)	(0.00)	(0.00)
Education
Less than high school	0.04	0.10	0.04	0.27	0.01	0.12	0.05	0.08
	(0.00)	(0.00)	(0.00)	(0.01)	(0.00)	(0.01)	(0.01)	(0.00)
High school diploma	0.24	0.32	0.28	0.30	0.11	0.35	0.19	0.28
	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
Some college or vocational training	0.27	0.29	0.27	0.21	0.17	0.29	0.23	0.27
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
College degree	0.25	0.19	0.25	0.15	0.29	0.15	0.28	0.22
	(0.01)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Graduate degree	0.19	0.10	0.16	0.07	0.42	0.08	0.26	0.15
	(0.00)	(0.00)	(0.00)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Marital status
Married	0.62	0.63	0.63	0.55	0.63	0.36	0.65	0.61
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Previously married	0.24	0.24	0.21	0.16	0.19	0.29	0.10	0.22
	(0.01)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Never married	0.14	0.13	0.17	0.29	0.18	0.35	0.25	0.17
	(0.00)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)

(Continued)

Table 1

Continued

	Mainline protestants	Evangelical protestants	Non-hispanic catholics	Hispanic catholics	Jews	Black protestants	Muslims	All
Income
Under $30K	0.18	0.26	0.17	0.39	0.08	0.36	0.26	0.23
	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
$30K–50K	0.18	0.21	0.17	0.19	0.09	0.23	0.18	0.19
	(0.00)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
$50K–100K	0.30	0.28	0.30	0.18	0.26	0.19	0.25	0.27
	(0.01)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Over $100K	0.20	0.13	0.21	0.10	0.37	0.09	0.22	0.17
	(0.00)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Do not know/refused	0.14	0.13	0.15	0.14	0.20	0.13	0.09	0.13
	(0.00)	(0.00)	(0.00)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Children in household	0.26	0.31	0.30	0.49	0.26	0.35	0.62	0.33
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Region
Northeast	0.18	0.09	0.33	0.16	0.40	0.11	0.30	0.19
	(0.00)	(0.00)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Midwest	0.31	0.25	0.32	0.10	0.14	0.19	0.21	0.27
	(0.01)	(0.00)	(0.01)	(0.01)	(0.01)	(0.01)	(0.01)	(0.00)
South	0.33	0.50	0.21	0.32	0.26	0.62	0.31	0.37
	(0.01)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
West	0.17	0.16	0.14	0.43	0.20	0.08	0.19	0.17
	(0.00)	(0.00)	(0.00)	(0.01)	(0.02)	(0.01)	(0.01)	(0.00)
Observations	6463	8038	5536	1370	588	1673	1618	25286

Standard errors are in parentheses.

Empirical Approach

Our analytical approach is two-fold. First, we examine the relationship between religious tradition and doctrinal conservatism using logistic regression models where religious tradition is the primary independent variable and doctrinal conservatism is the outcome. We nest these models to see the degree to which the differences in doctrinal conservatism across traditions are explained by either religiosity or the demographic composition of the religious tradition. This portion provides an update on the prevalence of doctrinal conservatism within American religious traditions, demonstrating where Muslims fit within this framework.

Next, we examine the relationship between doctrinal conservatism and right-leaning political attitudes for each religious tradition. This allows a reassessment of Wuthnow’s theory of religious restructuring given today’s religious landscape. Hypothesis 1 (H1), emerging from Wuthnow, is that we would see a strong relationship between doctrinal conservatism and conservative political attitudes for all groups. Alternatively, adopting an intersectional approach, we hypothesize that mainline Protestant, evangelical Protestant, non-Hispanic Catholic, and Jewish groups (those overwhelmingly racialized as white) will have substantively and statistically significant positive relationships between doctrinal conservatism and right-leaning political attitudes (H2A). For Hispanic Catholics, Muslims, and Black Protestants, we hypothesize the relationship between doctrinal conservatism and political conservatism will be insignificant or low in magnitude (H2B). To test this, we use multivariate ordinary least squares regression models where the outcomes are conservative political ideology and party identification and the primary independent variable is doctrinal conservatism, which is interacted with religious tradition to allow for differing relationships across groups. Again, we nest the regression models to see whether these relationships (between doctrinal conservatism and political conservatism, for each religious tradition) can be explained by demographic composition and/or religiosity.

Results

Predicting doctrinal conservatism

In the first portion of our analysis, we use logistic regressions to predict doctrinal conservatism, primarily from religious tradition. The first model does not include any controls, but the second model includes demographic controls: age, gender, race, immigrant status, education, income, marital status, whether there are children in the household, and region. Some may wonder whether the relationship between tradition and religious conservatism varies by these controls. For example, could the association between religious tradition and doctrinal conservatism differ by class? In a separate analysis, we tested for interaction effects between religious tradition and each of the controls listed here. We find none. Interactions were insignificant, and we confirmed this by calculating and comparing predicted probabilities. Similarly, in the second part of the analysis, where religious tradition and religious conservatism are used to predict political attitudes, we test for interactions and, again, find none. Hence, we treat all the demographic controls as additive. This effectively means that, while several variables (such as class as measured by education and income) had their own relationships with doctrinal conservatism (see Supplementary material, Appendix B, table B) and political conservatism (see Supplementary material, tables C-1 and C-2 in Appendix B), those variables were not a central part of our results because either (1) they did not strongly mediate the relationships of interest (between religious tradition and doctrinal conservatism and between doctrinal and political conservatism, respectively) or (2) the relationships of interest did not differ across levels of those variables.

Comparing results from Model 1 and Model 2 allows us to see whether and to what extent the between-tradition differences in doctrinal conservatism, observed in Model 1, can be explained by the composition of each group. Finally, the third model adds religiosity. Since coefficients in a logistic regression are difficult to interpret (representing changes in the log likelihood of a positive outcome on the dependent variable) and it is not appropriate to compare them across models (Mood 2010), we discuss our results in the form of predicted probabilities,¹⁶ which are displayed in table 2.

Table 2

Predicted Probabilities from Logistic Regression Models Predicting Doctrinal Conservatism from Religious Tradition, Religiosity, and Controls

	Model 1a	Model 2b	Model 3c
Mainline protestant	0.13	0.14	0.17
	[0.12, 0.14]	[0.13, 0.15]	[0.16, 0.18]
Evangelical protestant	0.45	0.44	0.38
	[0.44, 0.47]	[0.43, 0.45]	[0.37, 0.39]
Non-hispanic catholic	0.18	0.19	0.20
	[0.17, 0.19]	[0.18, 0.20]	[0.19, 0.22]
Hispanic catholic	0.23	0.22	0.23
	[0.21,0.25]	[0.20, 0.25]	[0.21, 0.26]
Jewish	0.04	0.05	0.12
	[0.03, 0.06]	[0.03, 0.08]	[0.08, 0.16]
Black protestant	0.41	0.33	0.31
	[0.39, 0.43]	[0.30, 0.37]	[0.28, 0.34]
Muslim	0.34	0.36	0.38
	[0.32, 0.36]	[0.33,0.39]	[0.35, 0.41]
Observations	25286	25286	25286

95% confidence intervals in brackets; coefficients from full models are in Supplementary material, Appendix A, table B.

^aModel 1 is a bivariate model wherein doctrinal conservatism is predicted only from religious tradition.

^bModel 2 adds demographic controls: age, gender, race, immigrant status, education, household income, and marital status.

^cModel 3 adds the religiosity index, which combines religious service attendance, prayer, and importance of religion.

Looking at Model 3 with all controls, the religious traditions are split into two groups. On the liberal end are Jews (12% predicted probability of doctrinal conservatism), mainline Protestants (17%), and Catholics (20% for non-Hispanic and 23% for Hispanics), while on the conservative end are Muslims (38%), evangelical Protestants (38%), and Black Protestants (31%). These predictions suggest that doctrinal conservatism is a minority position, even among the most doctrinally conservative traditions. Evangelicals, for example, are at a predicted probability of 45% in Model 1, and after controls the predicted probability is little more than a third (38%).

To understand how the predicted probabilities of doctrinal conservatism shift across columns in table 2, let us consider how the distance between each religious tradition and the reference group (mainline Protestants) changes from model to model. Evangelical Protestants are a distance of 32 percentage points from mainline Protestants in Model 1 without accounting for religiosity or controlling for demographic differences. This gap is about the same in Model 2 (30 percentage points), meaning that demographic differences between the two groups do not explain the gap in doctrinal conservatism. That is, we cannot say that Evangelicals are more conservative than mainline Protestants simply because they are younger, less educated, or more concentrated in the South than mainliners. However, once we control for religiosity, the difference between Evangelicals and mainline Protestants drops to 21 percentage points. This means that about a third of the original difference between the two traditions could be attributed to differences in average religiosity. As table 1 shows, the average religiosity of mainline Protestants was −0.04 on a scale from −3.5 to 1.8, whereas the Evangelical average was 0.76, and religiosity is correlated with higher religious conservatism.

Similarly, the difference between Black Protestants and mainline Protestants decreases from 28 percentage points in Model 1 to 19 percentage points in Model 2, a 32% drop. In other words, about a third of the difference in doctrinal conservatism between Black Protestants and mainline Protestants is attributable to differences in demographic composition between the two groups. Adding religiosity further reduces the difference (to 14 percentage points), meaning that about 18% of the difference in Model 1 is explained by variation in religiosity. Muslims remain about 20 percentage points higher than mainline Protestants across the three models. Non-Hispanic Catholics are consistently close to mainline Protestants in estimated doctrinal conservatism. Hispanic Catholics are slightly more doctrinally conservative than their non-Hispanic counterparts (5 percentage points in Model 1). About half of this small difference is mediated by demographic variables in Model 2, and then it loses significance once religiosity is included in Model 3. On the other hand, the small difference between Jews and mainline Protestants (8 percentage points) decreases and disappears in significance in Model 3.¹⁷

Predicting political conservatism

We turn next to the relationship between doctrinal conservatism and political conservatism. Our two political measures are ideology and party identification. We predict each outcome from doctrinal conservatism, religious tradition, and the interaction between the two (which is significant for several groups) using ordinary least squares regression.¹⁸ Again, we use models with and without controls to show the degree to which demographic variables and religiosity mediate our relationships of interest.¹⁹ First, in figure 1A, we show the overall distribution of political ideology as predicted from Model 1, without religiosity or demographic controls. As mentioned above, ideology ranges from 1 to 5 with 5 being very conservative and 1 being very liberal.

Figure 1

(A) Predicted conservative political ideology by religious tradition from OLS regression. ⁺Computed from Model 2 in Supplementary material, Appendix B, table C-1, which includes an interaction between doctrinal conservatism and religious tradition as well as the following controls: religiosity, age, gender, race, immigrant status, education, income, marital status, presence of children in the household, and region. Conservative political ideology outcome measure ranges from 1 to 5, with 5 being very conservative and 1 being very liberal. (B) Predicted Republican party identification by religious tradition from OLS regression. ⁺Computed from Model 2 in Supplementary material, Appendix B, table C-2, which includes an interaction between doctrinal conservatism and religious tradition as well as the following controls: religiosity, age, gender, race, immigrant status, education, income, marital status, presence of children in the household, and region

Comparing figure 1A to table 2, we can see that a religious tradition’s level of doctrinal conservatism is not always consistent with its level of political conservatism. The consistent groups are Evangelicals and Black Protestants. Evangelical Protestants come first in both political conservatism (3.45 on a scale of 5) and doctrinal conservatism (0.38 on a 0–1 variable). Jews are also aligned, but on the opposite end of the scale—they are both doctrinally and politically more liberal than other traditions. Though tied with Evangelicals for most doctrinally conservative, Muslims have the lowest estimated political conservatism (not significantly different from Jews). Mainline Protestants are the second most politically conservative group, falling slightly above the mean, despite being on the liberal end of religious conservatism. Non-Hispanic Catholics are more doctrinally liberal and more politically conservative than Hispanic Catholics.

Figure 1B shows the overall distribution of party affiliation as predicted from Model 1, without religiosity or demographic controls. This measure ranges from 1 to 3 with 3 being Republican, 1 being Democrat, and 2 being independent. The overall sample mean on this measure, as shown in table 1, is 1.94.

Evangelical Protestants lean most heavily toward being Republican, followed by mainline Protestants then non-Hispanic Catholics and Black Protestants, who do not differ significantly from one another. Jews, Hispanic Catholics, and Muslims are the least likely to identify as Republican. Overall, the two outcome measures (ideology and party identification) place traditions into a similar pattern relative to each other. Already these results drive a wedge into the restructuring thesis.

Figure 2A and B shows the degree to which doctrinal conservatism is associated with political conservatism for each group, directly adjudicating between Hypothesis 1 and Hypothesis 2. In other words, each bar represents the change in political conservatism associated with moving from doctrinally liberal to doctrinally conservative for each group. There are two bars for each religious tradition; on the left is the effect from Model 1 (with only religious tradition, doctrinal conservatism, and their interaction), and on the right is the effect from Model 2 (adding religiosity and controls).

Figure 2

(A) Average marginal effects of doctrinal conservatism on conservative political ideology by religious tradition. ⁺Computed from Model 1 and Model 2 in Supplementary material, Appendix B, table C-1. Model 1 includes doctrinal conservatism, religious tradition, and the interaction between the two. Model 2 includes the same and adds the following controls: religiosity, age, gender, race, immigrant status, education, income, marital status, presence of children in the household, and region. (B) Average marginal effects of doctrinal conservatism on Republican party identification by religious tradition. Computed from Model 1 and Model 2 in Supplementary material, Appendix B, table C-2. Model 1 includes doctrinal conservatism, religious tradition, and the interaction between the two. Model 2 includes the same and adds the following controls: religiosity, age, gender, race, immigrant status, education, income, marital status, presence of children in the household, and region.

Using either measure,²⁰ mainline Protestants, evangelical Protestants, non-Hispanic Catholics, and Jews have clear, significant, and positive relationships between doctrinal conservatism and political conservatism. This lends support to Hypothesis 2A that majority-white traditions would be split politically along the lines of doctrinal conservatism. The association between each type of conservatism is highest for Jewish Americans. Doctrinally conservative Jews are about a full point higher on the conservative ideology scale (range 1–5) and a half point higher in Republican party identification (range 1–3) than their doctrinally liberal counterparts. That Jewish communities exhibit the strongest bifurcation makes intuitive sense given the vast lifestyle differences and even physical segregation between Orthodox and Reform Jews.

The associations between doctrinal and political conservatism are also positive and significant for the other majority-white religious traditions: non-Hispanic Catholics, mainline Protestants, and evangelical Protestants. The magnitude of the associations is similar across these three traditions, though we note that the addition of demographic variables seems to lower effects for ideology more so than for Republican identification. In other words, the average marginal effect of doctrinal conservatism on political conservatism is lower after the addition of controls in a model with political ideology as outcome. When the outcome is Republican party identification, adding controls makes little difference.

For Hispanic Catholics, Black Protestants, and Muslims, the relationships between doctrinal conservatism and political conservatism are either weaker, as is the case with ideology, or nonexistent, as is the case with Republican party identification. After the addition of controls, being doctrinally conservative raises estimated political conservatism by 0.13 for Muslims and by 0.11 for Black Protestants. Though these are statistically significant and positive associations, they are about half as large as the effect for mainline Protestants and about a third as large as the effect for evangelical Protestants. For Hispanic Catholics, there is no significant relationship between doctrinal conservatism and ideologically leaning conservative.

When the outcome is Republican party identification, the racial split becomes even more apparent. There are strong, positive associations between being doctrinally conservative and leaning Republican for every majority-white (non-Hispanic) religious tradition, and there are no significant associations between being doctrinally conservative and leaning Republican for any of the racialized religious groups. These results lend support for hypothesis 2B, which predicts that racialized traditions would not be restructured in the manner theorized by Wuthnow.²¹

Discussion

In reassessing Wuthnow’s theory, we find that four of the seven religious traditions analyzed—mainline Protestants, evangelical Protestants, non-Latinx Catholics, and Jews—demonstrate continued congruence between doctrinal style and political attitudes. This results in an intra-religious distinction between “religious conservatives”—who wed conservative religious styles with conservative political attitudes—and “religious liberals,” who demonstrate liberal religious styles and liberal political attitudes. For these four, majority-white faith traditions, the theory of restructuring holds, though we note that the split between “religious liberals” and “religious conservatives” within most traditions is far from even. Our results highlight the significance of racialization to the religious restructuring process. The three racialized religious traditions in our analysis—Black Protestants, Latinx Catholics, and Muslims—do not demonstrate the internally realigned relationship between religious styles and political attitudes that Wuthnow theorized as a generally American phenomenon and that we find still present in the four majority-white traditions. While each of the racialized traditions possesses intragroup fractures and tensions (Ellison 1991; Jackson 2005; Pew Research Center 2007a, 2007b; Chen and Jueng 2012; Shelton and Cobb 2017; Guhin 2018), their racialized status prevents the formation of an internal division between religious conservatives and religious liberals of the kind conceptualized by Wuthnow. Our discussion here will reconsider the particulars of Wuthnow’s theory while taking group racialization more explicitly into account, in an attempt to speculate about the specific ways that racialized religious traditions have been differently effected by the restructuring forces outlined by Wuthnow.

Wuthnow identified four central factors that led to religious restructuring—the decline of denominational identities, an increasingly salient intra-denomination divide between a college-educated elite and a non-college-educated lay membership, politically divisive events (i.e., Vietnam War and Civil Rights Movement) that pitted Americans against one another along ideological rather than religious lines, and the rise of small “special-purpose groups” through which Americans organized around political issues at an intra-religious level (Wuthnow 1988, 1989). In terms of Wuthnow’s first factor, our findings suggest that while declining denominationalism did have a lasting effect on the internal political dynamics of majority-white religious traditions, for racialized traditions, religious identities continued to hold strongly because in these contexts they were undergirded by and intertwined with socially reinforced, politicized racial identities.²² While white ethnic and denominational identities came to matter less in comparison to ideology- and issue-based political ones (Herberg 1955; Roediger 2003; Calvillo and Bailey 2015; Williams 2018), such a realignment was simply not possible and/or desirable for members of racialized religious traditions. For these congregants, racial-religious collective identities were consistently reinforced by and served as a strategic bulwark against a conservative politics promoting an anti-minority rights and pro-white agenda (Emerson and Smith 2000; Crespino 2008; Wald and Calhoun-Brown 2014; Balmer 2017; Braunstein 2017; Whitehead, Perry, and Baker 2018a).

The continued salience of racialized religious identities for Black Protestants, Latinx Catholics, and Muslims also helps us understand these traditions’ resistance to Wuthnow’s second factor promoting religious restructuring—differences in education level within traditions. Again, while this may have been consequential for majority-white traditions, our analysis suggests that the unifying power of racial politics and racialized identities serves as a countervailing force against potentially divisive demographic differences such as educational background. While Black Protestants, Muslims, and Latinx Catholics do demonstrate internal variation by educational background—as well as class and, for Muslims and Latinx Catholics, race, ethnicity, and national background (Jackson 2005; Pew Research Center 2007a, 2007b; Chen and Jueng 2012; Shelton and Cobb 2017, Guhin 2018)—our analysis, as well as supporting evidence from qualitative studies, suggests that a shared history of racialized identity and politics overpowers such differences, particularly when it comes to political attitudes (see Wald and Calhoun-Brown 2014 for a review; also Lincoln and Mamiya 1990; Ellison 1991; Stevens-Arroyo 1998; Love 2017; O’Brien 2017; Selod 2018).

Racialized religious traditions’ shared history of being on the receiving end of and working to resist anti-minority politics also helps to explain why Wuthnow’s remaining two factors—internally divisive national-level politics and ideologically divergent intra-tradition special-purpose groups—have sharply different effects on these traditions than the majority-white ones. Regarding divisive politics, the very same issues that often divide majority-white traditions along lines of political ideology—school desegregation, welfare benefits for racial minorities, and stances on Latinx and Muslim immigrants—often politically unify members of racialized religious traditions in a progressive, Democratic, pro-minority rights direction (Kelly and Kelly 2005; Crespino 2008; Chen and Jueng 2012; Wald and Calhoun-Brown 2014; Balmer 2017) When they do not, they at least complicate the clean restructuring theorized by Wuthnow (Davis and Robinson 1996; Hinojosa and Park 2004; Edgell and Tranby 2007; Cobb, Perry, and Dougherty 2015; Shelton and Cobb 2017). Regarding Wuthnow’s final factor, while white traditions saw an emergence of politically divergent special-purpose groups within the same denomination (Wuthnow 1988), special-purpose groups within racialized religious traditions served to consolidate and amplify the relatively unified political stances of these communities around race-specific issues. Such issues have included racial and economic equality for Black Protestants (Wald and Calhoun-Brown 2014, 280–290); resisting racism, Islamophobia, and American foreign policy in the Middle East for Muslims (Wald and Calhoun-Brown 2014, 297–304; Love 2017); and economic equality, civil rights, and immigration issues for Latinx Catholics (Stevens-Arroyo 1998; Pew Research Center 2007b).²³

Conclusion

This paper has examined seven religious traditions within the contemporary United States to assess the extent to which the relationship between doctrinal approach and political attitudes shapes intra-tradition political divisions across the diverse landscape of American religious life. We find that the answer varies by the racialized status of the religious tradition. While the expectation of Wuthnow’s restructuring theory—that doctrinal approach and political attitudes will be tightly linked and result in a configuration of “religious conservatives” and “religious liberals”—still holds true for the four mostly white traditions, the three traditions for whom religious identity remains highly salient and racialized—Muslims, Black Protestants, and Latinx Catholics—show little or no relationship between doctrinal approach and political attitudes. We speculate that the social forces hypothesized by Wuthnow to contribute to restructuring—changes in education, denominationalism, and political division—are buffered by the continued salience of unified racialized religious identities and support for associated pro-minority rights politics within the un-restructured traditions. Relatedly, we speculate that white Evangelicals demonstrate such high levels of doctrinal and political conservatism because their ranks include substantial unified support for a conservative, white, pro-Christian nationalist, anti-minority rights agenda (Whitehead, Schnabel, and Perry 2018b), a politics in direct opposition to that of the minority racialized religious traditions in our analysis.

It is important to note that the process of resistance to restructuring that we observe by members of racialized religious traditions is distinct from what is experienced by racial minority members of non-racialized traditions, congregations, or denominations. For example, black members of multiracial congregations are less likely to hold liberal political views on racial inequality than African Americans who attend racially homogenous congregations (Cobb, Perry, and Dougherty 2015). Similarly, the growing numbers of religiously and politically conservative Latinx Protestants in the United States (Kelly and Kelly 2005; Wald and Calhoun-Brown 2014) demonstrate that it is not simply being Latinx that predicts a lack of congruence between religion and politics. Rather, it is the particular interaction between one’s own racial status and the racialized religious tradition one belongs to—in this case, Latinx Catholicism, with its very unique history of ethnic identity, social justice work, and conservative morality (Stevens-Arroyo 1998; Chen and Jueng 2012; Calvillo and Bailey 2015)—that predicts a resistance to the congruence of religious and political conservatisms.²⁴

With this paper, we join a growing chorus of sociologists who conceptualize doctrinal conservatism and liberalism as “styles” of religion—cultural forms of religion analytically distinct from religious group membership and affiliation that can impact belief and behavior in “non-religious” areas of social life (Whitehead, Schnabel, and Perry 2018b, 9). Our conceptualization of doctrinal approach as a religious style that predicts politics—and this style’s variation by racial identity—reinforces two recent findings about the cultural content of American religious styles. First, our paper adds to the evidence that American religious styles often rely upon a particular approach to textual interpretation as a core component of their cultural meaning (Perry and Whitehead 2015, 1673; Stewart, Edgell, and Delehanty 2018, 19; see also Crapanzo 2001; Beilo 2009; Tripodi 2018). Second, our findings align with other studies asserting that contemporary religious styles are often bound up with racial-religious identities and racial politics (Perry and Whitehead 2015; Delehanty, Edgell, and Stewart 2018, 17; Stewart, Edgell, and Delehanty 2018, 33; Whitehead, Perry, and Baker 2018a).²⁵ In short, this paper joins recent scholars of religion and culture in finding that religious styles matter for politics, that they often centralize particular doctrinal approaches, and that they frequently involve racial boundary work and racialized identities.

This study also strengthens the gathering consensus that racialized “structural location” is a powerful factor in mediating the relationship between religious styles and politics (Edgell and Tranby 2007; Frost and Edgell 2017; Perry and Whitehead 2015; Whitehead, Schnabel, and Perry 2018b). Yes, religious styles matter, but the particular ways in which they matter are shaped by the specific racial and ethnic contexts in which they are transposed, articulated, and situated. These findings reflect our use of the emerging theoretical orientation of “complex religion,” in which racial and other structural inequalities are integrated into analyses of nominally religious social and political phenomena. Only by considering the potential importance of race “at the starting point” of an analysis—as we have done by including Muslims, Black Protestants, and Latinx Catholics here—can researchers capture important racial dynamics that might otherwise be overlooked (Wilde 2018, 296).

This acknowledgment of racial politics as a line of fracture through the American religious and political landscape returns us to contemporary politics and the election of Donald Trump. While some were surprised by the steadfast support of Evangelicals for Trump’s candidacy despite his personal moral failings and lack of visible religious commitment (Green 2016), others noted one area in which the politics of conservative white Evangelicals and Trump did overlap (other than a concern for securing conservative Supreme Court nominees)—an emphasis on prioritizing the well-being of Christian white Americans and an active marginalization of racialized “others” (Edgell 2016; Braunstein 2017; Whitehead, Perry, and Baker 2018a). Reconsidering Wuthnow’s restructuring thesis from the perspective of complex religion helps us make better sense of this contemporary political reality. Rather than a restructuring in which particular social forces raised the salience of political identities over denominational ones and led to the internal fracture of all US religious traditions, the power of racial identities, racial politics, and racism have complicated this process more than Wuthnow and most following him have noticed or predicted. This has left particular religious traditions—those prioritizing racialized identities—internally united within a nationally divided, racially infused politics.

About the authors

John O’Brien is an Assistant Professor of Sociology in the Social Sciences Division of NYU Abu Dhabi. His first book, Keeping It Halal: The Everyday Lives of Muslim American Teenage Boys (2017), won the Distinguished Book Award from the Religion Section of the American Sociological Association. His articles have been published in Sociological Theory, Poetics, and Social Psychology Quarterly. His current book project explores the cultural dispositions of Western professional-class migrants to the Arabian Gulf.

Eman Abdelhadi is a Provost’s Postdoctoral Fellow at the University of Chicago in the Department of Comparative Human Development. Her research is at the intersection of religion, gender, race, and inequality, and her recent work has been published in Socius, Research in Social Stratification and Mobility and the British Journal of Sociology. She is writing a book on how gender shapes American Muslims’ constructions of community and identity.

Endnotes

1

Following standard practice within sociology, we use “traditions” to mean major religious groups within the society under study and “denominations” to mean smaller divisions within these traditions.

2

Our use of “racialized” when referring to minority and socially marginalized religious traditions is not intended to connote that “whites” are not also members of a racial group. Following Bonilla-Silva (1997), we treat racialization as a process in which “racial groups” are both socially constructed and placed in hierarchies. In racialized societies, the racialized status of subordinate groups is more visible and used as a basis for social marginalization, while the racialized status of the (usually white) socially dominant group is generally unmarked yet still used as a basis of privilege.

3

Also referred to as religiosity or, more recently, “intense religion” (Schnabel and Bock 2017).

4

Doctrinal conservatism, while often presenting itself as eternal and unchanging (Aslan 2006; Avishai 2008), is actually able to be adopted or cast off (e.g., Brophy 2016). While it is true that certain traditions, denominations, congregations, and communities demonstrate an affinity for a doctrinally conservative approach to religious life with some consistency over a given time period (Ammerman 1987; Schielke 2009), scholars agree that even long-term adoptions of doctrinal conservatism are historically contingent and the result of social and historical processes, not the result of something inherent to the particular religion itself (e.g., Aslan 2006; Asad 2009).

5

Christian nationalism has proven highly predictive of conservative political views, attitudes, and behaviors including voting for Donald Trump (Whitehead, Perry, and Baker 2018a); opposing gun control (Whitehead, Schnabel, and Perry 2018b); opposing racial intermarriage (Perry and Whitehead 2015); espousing anti-Muslim attitudes (Braunstein 2017); opposing economic regulations, welfare, and affirmative action (Froese and Mencken 2009); and supporting African American-targeted harsh penalties for crime and deviance (Davis 2018). Public religion, as conceptualized by Stewart, Edgell, and Delehanty (2018), is highly correlated with conservative political attitudes such as intolerance of various social out-groups (Stewart, Edgell, and Delehanty 2018), opposition to “race-focused movements and policies” (Delehanty, Edgell, and Stewart 2018, 17), and reduced support for structural explanations for racial inequality among white Americans (Frost and Edgell 2017).

6

The data we use are publicly available on the Pew Research Center’s website: www.pewresearch.org. Analysis was conducted using Stata 16, and all do files and tables associated with the project are available upon request. More details about the sampling techniques used for each survey and the process of winnowing down our sample are available in Supplementary material, Appendix A to this article.

7

As the variable section below will detail, the survey used the word Hispanic. When directly describing our data, we replicate this choice in order to maintain accuracy. However, we recognize that this term has a problematic and controversial history (Alcoff 2005). Out of respect for Latino/a communities’ recent embrace of the term Latinx, we use it in our broader discussions throughout the paper (Salinas and Lozano 2017, Morales 2018).

8

The surveys we use here do not contain self-identification as a religious or theological conservative, which would have been ideal. However, other surveys that do contain these measures do not include enough Muslim respondents to allow for the unique cross-religion comparison we provide here.

9

In Supplementary material, Appendix C, we describe results from supplemental models using literalism as the measure of religious conservatism. We also present results from other potential measures of religious conservatism in the surveys. We find consistent patterns across these sensitivity checks, which gives us confidence that doctrinal conservatism is serving as an adequate proxy for religious conservatism. Ultimately the paucity of measures for doctrinal conservatism is a limitation of these data, but we believe it is offset by the unique comparative advantage they offer.

10

My religion in the Religious Landscapes Survey and Islam in the Muslim American Survey.

11

A limitation of this measure is that it treats doctrinal conservatism as binary, much as Wuthnow does. However, as Supplementary material, Appendix C shows, the overall story in our results remains the same if we replace this measure with a composite, continuous scale of other potential proxies for doctrinal conservatism.

12

An earlier iteration of this article combined these two variables with belief in hard work as a guarantor of success and desire for a small government into a political index using principle component analysis. The patterns and results we describe here were nearly identical using this index. Results are available upon request.

13

Separate from the race question, respondents were asked if they were Hispanic. We used their answer to separate Hispanic and non-Hispanic Catholics, because research in the United States suggests Latinx communities build separate congregations (Stevens-Arroyo 1998; Pew Research Center 2007b).

14

Meaning the respondent belongs to a church categorized as part of the Black Protestant tradition, not that the respondent is simply black and Protestant.

15

Churches included under each Christian category are listed in Supplementary material, table A of Appendix B.

16

The full logistic regression models are included in Supplementary material, Appendix B, table B.

17

Though demographic controls did little to explain the differences in doctrinal conservatism between groups, many were significantly associated with doctrinal conservatism. For example, young people (between ages 18 and 29) were the least doctrinally conservative. Immigrants were more doctrinally conservative, on average, than those born in the United States. Women were more doctrinally liberal than men, and both education and income had negative, monotonic effects on doctrinal conservatism. Marital status, region, and race were mostly not significant. See Supplementary material, table B in Appendix B for the full regression results.

18

We treat these categorical variables as continuous, because this provides the easiest results to interpret. However, as a sensitivity test, we used multinomial logistic regression to ensure that our story holds across different types of models. We find the same patterns we describe here—namely, strong relationships between doctrinal and political conservatism for mainline Protestants, evangelical Protestants, non-Hispanic Catholics, and Jews and no or very weak associations between doctrinal and political conservatism for Muslims, Hispanic Catholics, and Black Protestants.

19

Supplementary material, tables C-1 and C-2 in Appendix B show coefficients from these regressions. The first model in each table predicts political conservatism from only doctrinal conservatism, religious tradition, and the interaction between the two. The second model in each table adds religiosity and demographic controls. We add religiosity to each of the second models at the same time as demographic controls, because when added on its own, religiosity did not mediate the effects found in Model 1 of either table. Furthermore, we found no interaction between religiosity and religious tradition or between religiosity and doctrinal conservatism in predicting either of the political conservatism measures.

20

We focus here on political measures of conservatism, but the readers may wonder whether social conservatism such as views on abortion or LGBTQ rights would exhibit the same patterns. Indeed, Davis and Robinson argue sexual and gender politics are especially impacted by religious conservatism (1996). We elected to focus on political conservatism, because there were not enough measures of social views across all three surveys. However, in Supplementary material, Section III of Appendix C, we share results from the one social question shared across all samples—how respondents felt about homosexuality. Based on this question, there is mixed evidence for whether our patterns also hold for social conservatism, and given that it is only one measure, we cannot make any claims in either direction.

21

The readers may wonder whether doctrinal conservatism predicts political conservatism among racial minorities within Mainline, Evangelical, and non-Hispanic Catholic traditions. We conduct an analysis where we test for a three-way interaction between denomination, doctrinal conservatism, and race to see whether this may be the case. We do not find evidence of a racial difference within these traditions, which suggests the unifying effect of racialization happens at the institutional rather than individual level.

22

Another group in which a racialized identity has continued to matter is among a subset of white Evangelicals who have actively embraced Evangelicalism as a subcultural identity (Smith 1998), as well as sometimes wedded this identity to a pro-Christian nationalism and anti-minority rights conservative agenda (Crespino 2008; Whitehead, Perry, and Baker 2018a).

23

Just as racialized religious traditions unified around political issues and expressed this support through special-purpose groups on the liberal side, an active subset of white Evangelicals used special-purpose groups to advocate for a conservative and often anti-minority rights political agenda (Balmer 2017; Whitehead, Perry, and Baker 2018a).

24

Some scholars even suggest that religious congregations comprised completely of minority members may not be “ethnicized” at all depending on the extent to which they undergo a process of actively “shedding ethnic languages, theologies, and worship cultures, to adopt the theology, music, and worship practices” (2012, 465) of white traditions (Kurien, cited in Calvillo and Bailey 2015, 60).

25

For Christian nationalists, racial boundaries and exclusion have been linked with the religious style from the beginning, as it “draws its roots from ‘Old Testament’ parallels between America and Israel, who was commanded to maintain cultural and blood purity” and its appeals often “imply the exclusion of other religious faiths of cultures” (Whitehead, Perry, and Baker 2018a, 150). Unlike Christian nationalism, public religious expression (PRE) is not conceptualized as having race at the root, although it does often correlate with race-based boundary and exclusion work in practice (Delehanty, Edgell, and Stewart 2018; Stewart, Edgell, and Delehanty 2018). When this style is carried into the public sphere—including a concern with the application of “God’s laws”—it is highly correlated with prejudice and intolerance (Stewart, Edgell, and Delehanty 2018, 33), the “belief that leaders of racial minority groups have too much power” (Delehanty, Edgell, and Stewart 2018, 17), and a “distinct and relatively narrow vision of belonging in American society” (Stewart, Edgell, and Delehanty 2018, 32).