OPIATE OF THE MASSES? SOCIAL STATUS, RELIGION, AND POLITICS

Landon Schnabel* Indiana University, Bloomington

Abstract: This study considers the assertion that religion is the opiate of the masses and examines whether this way of understanding religion can be used to account for two key sociological questions: (1) sociodemographic differences in religiosity and (2) apparent inconsistencies in the positionality principle of politics. Using a special module of the General Social Survey, I first demonstrate that groups with less status—women, racial minorities, those with lower incomes, and sexual minorities—receive more compensatory psychological benefits from religion and spirituality and that these compensatory benefits help explain group differences in religiosity. I then demonstrate that religious compensation acts as a suppressor variable between status and politics, suppressing the liberal politics of disadvantaged groups by facilitating and reinforcing religiosity. This study, therefore, provides empirical support for Marx’s general claim that religion is the “sigh of the oppressed creature” and suppressor of political consciousness. It expands and refines the argument, however, by showing that (1) it can apply to disadvantaged social status in addition to disadvantaged economic status; (2) the suppression of progressive politics appears to be due to the conservatizing nature of religiosity as much or more than other-worldly distraction; and (3) this suppression is stronger on issues closely linked to traditional religious values than on economic issues. I conclude that compensatory benefits from religion should be more central to and frequently measured in the study of religion, inequality, and politics.

Key Words: Religion; Politics; Public Opinion; Gender; Race; Ethnicity; Class; Sexuality; Inequality; Well-being; Marx

Last Revised: 6/1/2017 Running Head: Opiate of the Masses? Word Count: 11,996 Tables: 7 Figures: 0

THIS IS A DRAFT. DO NOT CITE OR CIRCULATE WITHOUT AUTHOR PERMISSION.

* The author is grateful to Brian Powell for exceptional feedback. He would also like to thank Art Alderson, Clem Brooks, Youngjoo Cha, Andy Halpern-Manners, Patricia McManus, Chris Munn, Roshan Pandian, Brian Steensland, Evan Stewart, and members of the Indiana University Politics, Economy, and Culture Workshop for helpful comments on this project. Direct correspondence to Landon Schnabel, Department of Sociology, Indiana University, 744 Ballantine Hall, 1020 E. Kirkwood Ave., Bloomington, IN 47405. Email: [email protected].

Opiate of the Masses? 2

Religious suffering is, at one and the same time, the expression of real suffering and a protest against real suffering. Religion is the sigh of the oppressed creature, the heart of a heartless world, and the soul of soulless conditions. It is the opium of the people.

-Karl Marx (1970 [1843])

According to Marx, religion is the opiate of the masses. In other words, religion appeals to the disenfranchised and helps them through suffering, but in so doing distracts from the root causes of their suffering and suppresses the development of political consciousness.1 This study considers the assertion that religion is the opiate of the masses, extending the argument from the materially disadvantaged to the socially disadvantaged to determine whether it can help account for two key sociological questions: (1) sociodemographic differences in religiosity and (2) apparent inconsistencies in the positionality principle of politics.

Although the “opiate” argument is frequently invoked theoretically, the mechanism by which religion is said to operate as an opiate remains underspecified and undertested. I propose and employ compensatory psychological benefits from religion, including comfort received from religion and spirituality, as a way to measure and examine the potential opiate function of religion. The idea that religion and spirituality provide comfort, consolation, and validation is not new to the literature and is central to some theoretical frameworks (such as the deprivation- compensation hypothesis). Despite their theoretical centrality, compensatory benefits from religion and spirituality typically go unmeasured. Previous research on the gender gap in religion has similarly highlighted social psychological benefits to religion to argue that women’s lack of status might contribute to them being more religious, but has not directly measured the proposed compensatory benefits to religion that appeal to women and other disadvantaged groups (Norris and Inglehart 2008; Schnabel 2016c; Walter and Davie 1998).

1 Weber similarly argued that other-worldly religion can alleviate this-worldly suffering and Durkheim argued that religion functions to legitimate social arrangements. Opiate of the Masses? 3

In Marx’s view, religious compensators for material deprivation provide comfort to the disadvantaged and thereby suppress their political consciousness. Several perspectives in the study of public opinion—such as the “underdog” positionality principle and related status and identity theories (Davis and Robinson 1991; Hunt 2007; Robinson and Bell 1978; Schuman and

Harding 1963)—suggest that disadvantaged groups develop a liberal political consciousness because of their marginalized status in society, recognition of structural inequalities, sympathy for and solidarity with other disadvantaged groups, and development of progressive identity and partisanship. Yet, women and racial minorities, to whom the positionality principle has been most frequently applied, are not always more liberal than their counterparts on particular issues, especially those linked in one way or another to religion—for example, abortion, same-sex marriage, and physician-assisted suicide (Schnabel 2016b). A few studies have suggested that religion suppresses the liberalism of women (Barkan 2014) and Blacks (Wilcox 1992) on abortion specifically, but religion may operate as a more general suppressor of progressive politics and thus explain apparent inconsistencies in the positionality principle. Were it not for diversion by that which helps them through hardship, disadvantaged groups may have a stronger political consciousness and hold more consistently progressive social attitudes on issues linked to religious values.

This study first considers whether groups with less status—women, Blacks, Latinos, those with lower incomes, and sexual minorities—receive more compensatory psychological benefits from religion and spirituality. It then examines whether these psychological compensators can help explain one of the most consistently found and frequently debated patterns in the social scientific study of religion: women’s greater religiosity. Finally, this study considers whether and why compensatory psychological benefits from religion suppress the Opiate of the Masses? 4 liberal politics of disadvantaged groups. In the course of testing hypotheses related to the opiate of the masses thesis, I find empirical support for Marx’s general claim that religion is the “sigh of the oppressed creature,” “heart of a heartless world,” and suppressor of political consciousness. I expand and refine the argument, however, by showing that (1) it can apply to disadvantaged social status in addition to disadvantaged economic status; (2) the suppression of progressive politics appears to be due to the conservatizing nature of religiosity as much or more than other-worldly distraction; and (3) this suppression is stronger on issues closely linked to traditional religious values than on economic issues. I conclude that the compensatory benefits provided by religion are a key component in religious processes and, subsequently, the political attitudes, identities, and behaviors of disadvantaged groups.

Deprivation and Religious Compensation

Deprivation-compensation theories of religion have long existed and have frequently been used to make predictions about religiosity. Davis (1948) said that the greater a person’s disappointment and frustration in this life, the more s/he would believe in and focus on a future life. According to Davis, focusing on goals beyond this world allows people to compensate for frustrations trying to reach present goals. In other words, people shift their focus to a better future life as Stark and Bainbridge (1985, 1987) highlighted when they said that religion provides “supernatural general compensators” that replace unavailable material rewards with future spiritual rewards. Therefore, as far as the human experience leads to unmet material needs, people will seek future immaterial compensation for these unmet needs. Even when scholars such as Stark, Iannaccone, and Finke (1996) eventually shifted more explicitly to the rational- choice framework, benefits provided by religion were a key foundation to their economics-of- religion reward-seeking perspective. The importance of considering what people do with and get Opiate of the Masses? 5 from religion in their everyday lives (e.g., receive comfort and make meaning out of hardship) becomes all the more important in light of the more recent cultural turn in the sociology of religion (Edgell 2012).

Class and material deprivation certainly remain relevant, but gender, race, and sexuality can be equally salient to contemporary American religious life (Stewart, Frost, and Edgell 2017;

Wilde and Glassman 2016). Although some applications and tests of the deprivation- compensation principle continue to focus primarily on economic deprivation (Becker and

Woessmann 2013; Wimberley 1984), scholars have highlighted how other social divides and forms of deprivation, such as lack of social status, could also motivate religiosity (Christopher et al. 1971; Ellison and Taylor 1996; Glock 1964; Glock, Ringer, and Babbie 1967; Glock and

Stark 1965). These studies show that factors such as gender and race are consistent with the deprivation hypothesis. In its various forms, the deprivation-compensation hypothesis remains central to social scientific theories about religiousness (Beit-Hallahmi 2014), but is typically used as a theoretical explanation for why disadvantaged groups are more religious rather than explicitly tested using measures of compensatory benefits from religion.

Moving beyond what scholars theorize about the immaterial benefits to religion, we can consider whether the general public thinks religion and spirituality provide psychological benefits. In 2008, the General Social Survey asked Americans whether they think religion provides other people comfort and happiness. Most agreed that religion provides people comfort in hardship (95% agreed, 4% neither agreed nor disagreed, and 1% disagreed) and that religion helps people find peace and happiness (87% agreed, 10% were neutral, and 3% disagreed).

Therefore, both scholars and the public think that religion provides psychological benefits, but these benefits are rarely measured and studied. Opiate of the Masses? 6

Although empirical tests of the “opiate of the masses” concept are much rarer than its theoretical invocation, one study set out to explicitly test one issue related to the argument— higher incomes will increase secularization—using historical data on church attendance in

Prussia from 1886 to 1911 (Becker and Woessmann 2013). The study, which did not measure compensatory benefits from religion, found that regional increases in income (proxied by data on the average annual income of male elementary school teachers) were not causally linked to decreased church attendance. They concluded that religion, therefore, is not the opiate of the masses. Their findings, however, may simply highlight the need to actually measure the opiate function of religion and the importance of considering inequality and social conditions rather than just average improvement of material conditions.

Marx focused on class inequality, but his statement about religion providing compensatory psychological benefits to the disadvantaged could logically extend beyond economic disadvantage to social disadvantage in contemporary American society. In this study, I operationalize and extrapolate on Marx’s argument to consider whether it can apply more broadly to social disadvantage and marginalization.

Hypothesis 1: Disadvantaged groups—including women, blacks, Latinos, those with lower incomes, and sexual minorities—will receive more compensatory psychological benefits from religion and spirituality than will their more advantaged counterparts.

Social Status and Group Differences in Religiosity

A wide range of often incompatible theories exist for why some groups of people are more religious than others. A particularly large and contentious literature has focused on why women seem to be so consistently more religious than men in Christian contexts. Various theories have been used to explain these gender differences in religiosity: for example, biology-based risk preferences (Miller and Stark 2002), socialization-based risk preferences (Collett and Lizardo

2009), social risk (Edgell, Frost, and Stewart 2017), vulnerability (Norris and Inglehart 2011; Opiate of the Masses? 7

Walter and Davie 1998), psychological characteristics (Francis 1997; Freese and Montgomery

2007), and social structure and status (Luckmann 1967; Schnabel 2016c). Schnabel (2015b) argued that gender differences in religiosity should be considered in the context of gender differences in other areas, and here I argue that gender differences in religiosity should be considered in the context of other group differences in religiosity. With the exception of sexual minorities who have been further marginalized by religious institutions (Sherkat 2002, 2017;

Wedow et al. 2017), disadvantaged groups are typically more religious than their more advantaged counterparts in Christian contexts.2 This pattern may be due, at least in part, to religion and spirituality providing compensatory comfort, strength, and validation to people who have less social status and face greater social hardship in their everyday lives (Hastings and

Lindsay 2013; Schnabel 2015b, 2016c).

Some research on gender and other sociodemographic differences in religion emphasizes the greater material deprivation and physical insecurities women face around the world (Norris and Inglehart 2008, 2011), but other research (Schnabel 2016c) argues that lack of social status may be more important than lack of material resources for understanding why women and other disadvantaged groups are often more religious than men. This same literature also suggests that the American religious context, where Christianity predominates, promotes gender and other social status differences in religiosity because Christianity is perceived as a sympathetic underdog religion that, as noted by Marx (1970), appeals to the disadvantaged (Schnabel 2015a,

2015b). Therefore, gender and other sociodemographic gaps in religiosity may be the result of

(Christian) religion providing comfort to the disenfranchised. Although past research has argued

2 Relevant to Marx’s argument about disadvantage and otherworldly comfort, sexual minorities tend to be more likely to believe in an afterlife even though they are no more (and often less) religious than heterosexuals (Sherkat 2017). Opiate of the Masses? 8 that Christian religion provides comfort, strength, and support to the disadvantaged, these potential compensatory benefits to everyday lived religion typically go unmeasured. In this study, I will contribute to these debates by explicitly measuring the potential “opiate” benefits provided by religious beliefs, examining social status gaps in these benefits, and testing the extent to which group differences in religiosity are a function of these psychological compensators.

Hypothesis 2: Sociodemographic differences in religiousness are in part a function of psychological compensation.

Religion and Apparent Inconsistences in the Positionality Principle

In popular usage, the phrase “opiate of the masses” is frequently employed to refer just to consolation and happiness premiums provided by religion. But in context Marx was not simply saying that people seek comfort and gain happiness in religion because of hardship, but also that such psychological compensators suppress progressive politics and prevent proletarian revolution against the bourgeoisie. According to Marx (1970), religion is an “inverted consciousness” that responds to disadvantageous material conditions with nonmaterial compensators. This compensatory spiritual focus of Christian religion then limits the development of consciousness and action to address the disadvantageous conditions that made religion and spirituality more appealing.3 Complementing Marx’s argument, Simmel (1971) said that subordinated groups are more likely to accept social domination—and less likely to develop political consciousness and status-based solidarity—if those above them in the social hierarchy are perceived as spiritually subordinate to a caring, personal, and fair higher authority in which the subordinated find support. Similarly, more recent cultural and social psychological

3 Marx (1970) went so far as to argue that to struggle against religion is also to struggle against the unequal material and social conditions that contribute to the appeal of religion as a coping mechanism. Opiate of the Masses? 9 approaches to the study of religion highlight not just how people use religion to make meaning out of their everyday lives, but also how religion as a legitimating force can promote exclusionary symbolic boundaries (Edgell 2012) and motivate or constrain social action (Vaisey

2009) in ways that justify and exacerbate inequalities (Jost et al. 2014). Marx’s general argument about religion suppressing political progressive politics may therefore explain apparent inconsistencies in group position theories of politics.

According to positionality and status theories that build on the foundational concept of the “other” (Box-Steffensmeier, Boef, and Lin 2004; Davis and Robinson 1991; Hoffman 2000;

Hunt 1996, 2007; Kinder and Winter 2001; Manza and Brooks 1998; Robinson and Bell 1978;

Schuman and Harding 1963), group position and a greater awareness of structural reasons for inequality make disadvantaged groups less likely to believe in legitimating ideologies of racism, sexism, nationalism, etc. that justify social hierarchies and inequalities (Frost and Edgell 2017;

Sidanius, Pratto, and Bobo 1994), more likely to recognize and distrust oppressive norms, policies, and institutions (Collins 2000), and, therefore, more likely to develop progressive politics across a broad range of issues (Davenport 2016). This “underdog” positionality principle is generally compatible with Marx and his idea of group-based consciousness, but differs in its focus on social status instead of economic status and is typically applied to gender, race and, more recently, sexuality.

The status, positionality, and politics literature has consistently shown that women and racial minorities are more liberal on issues related not just to their own wellbeing, but also to the wellbeing of other marginalized groups (e.g., women tend to hold less racist attitudes than men).

A recent study applied the positionality principle to sexual minorities—who tend to be less religious and/or practice more inclusive forms of religion due to their marginalization by many Opiate of the Masses? 10 religious groups—and found that they tend to be consistently more liberal than their counterparts across hundreds of attitude items in the General Social Survey (Schnabel 2016b). In the same study, women and Blacks also tended to be more liberal than their more advantaged counterparts across most topics, but on some specific issues—frequently those tied to traditional religious values such as abortion, sexuality, and suicide—women and Blacks were just as if not more conservative than their counterparts.

Scholars working from status-based frameworks for public opinion and politics—for example, the underdog positionality principle, conflict theory, other identity and group interest theories, and even just utilitarian self-interest approaches—have frequently neglected the potential symbolic power of religion as a cultural force to shape the values of disadvantaged groups. Although religion as a suppressor of liberal politics has not been broadly examined, cultural approaches to religion have highlighted the identity and outlook-shaping power of religion (Edgell 2012; Perry and Schnabel 2017; Tranby and Zulkowski 2012) and a few studies have suggested religion suppresses the liberalism of women (Barkan 2014) and Blacks (Wilcox

1992) on abortion specifically. Although not framed as a test of Marx’s concept of religion as the opiate of the masses, these studies provide some indication that religion may account for why some disadvantaged groups are not more consistently liberal.

In terms of Marx’s view that religious compensators suppress the development of political consciousness, we can think of the opiate function of religion as a passive force that yields political quietism by distracting disadvantaged groups from the root causes of their suffering. Moving from the theoretical to the empirical and directly testable, a suppressor variable is a measure that conceals the strength of a relationship, such as the relationship between

“underdog” status and liberal politics. This occurs when a variable, such as religious Opiate of the Masses? 11 compensation, is associated with an independent variable and a dependent variable in opposite directions. If disadvantaged groups receive more comfort from religion and comfort from religion is associated with conservative politics, then this comfort, as Marx proposed, suppresses what would otherwise be a stronger link between positionality and politics.

This study will test Marx’s argument and determine whether psychological compensation from religion is keeping disadvantaged groups from developing an even more liberal political consciousness, which might otherwise lead them to challenge the material and social root causes of the marginalization that makes religion and spirituality more appealing.

Hypothesis 3: Religious compensation will suppress liberal politics among disadvantaged groups.

According to Marx (1970), religion suppresses political consciousness by means of distraction. More specifically, abstract other-worldly distraction diverts attention from material this-worldly concerns. But recent cultural approaches to the study of religion suggest that it is not just a passive distraction. Instead, religion is a meaningful construct and resource that when used by people acts back upon them with substantial symbolic power (Edgell 2012). Religion and conservative politics are closely linked (Norris and Inglehart 2011; Putnam and Campbell 2010), especially in the contemporary United States with its politicized religiosity (Bean 2014; Goren and Chapp 2017; Hout and Fischer 2002, 2014; Steensland and Wright 2014; Wuthnow 1988), and research in the social psychological study of religion suggests that religion promotes system- justifying beliefs that enforce exclusionary moral boundaries and highlight individual responsibility over systemic causes of inequality (Jost et al. 2014; Rankin, Jost, and Wakslak

2009).

Therefore, when religion and religious beliefs are used as comfort-providing and meaning-making cultural tools by the socially disenfranchised, the symbolic power of religion Opiate of the Masses? 12 could legitimate power structures and promote inequalities (Edgell 2012; Lizardo 2009). Yes, some small countercultural religious groups promote progressive values (e.g., Unitarian

Universalism) and larger religious groups run by marginalized groups have at times mobilized for certain progressive issues, such as racial equality in the Civil Rights Movement. But, as a whole, religion is linked to conservative identity and conservative values on gender, sexuality, and other issues closely related to religious teachings. Supporting the possibility that American religion is an actively conservatizing rather than just passively distracting force, it is primarily issues linked to religion and religiosity on which disadvantaged groups are no more liberal and sometimes more conservative than their counterparts (Adamczyk, Boyd, and Hayes 2016;

Barkan 2014; Schnabel 2016b; Wilcox 1992). Consequently, I expect religion to suppress progressive politics not simply through passive other-worldly focus but also by promoting traditional morality and generally conservative politics through active religiosity.

Hypothesis 4: The suppression of progressive politics will be as much or more a function of the conservatizing nature of religiosity than of other-worldly focus.

Thus far, I have focused on the suppression of liberalism on general politics and specific issues on which some disadvantaged groups are not more liberal. Marx, however, emphasized economic concerns, such as redistribution and state-provided social welfare. Unlike issues such as abortion, same-sex relationships, and physician-assisted suicide, disadvantaged groups tend to be consistently more liberal on economic issues (Kinder and Winter 2001; Robinson and Bell

1978; Schnabel 2016b). Therefore, in addition to issues that demonstrate apparent inconsistencies in the positionality principle, I will also consider whether and how religious compensators relate to economic attitudes and whether and how this alters the relationship between social status and economic attitudes. Opiate of the Masses? 13

If we were to develop predictions based on Marx alone we might say that religious compensators would have an especially strong negative impact on economic politics. However, religion and religiosity are not monolithic political forces (Greeley and Hout 2006), are not as closely linked to economic conservatism as social conservatism in the majority-Christian context of the United States (Felson and Kindell 2007; Schnabel 2016a; Steensland and Wright 2014;

Tranby and Zulkowski 2012), and in fact can even promote compassion and care for the poor in some circumstances (Blouin, Robinson, and Starks 2013; Davis and Robinson 1999, 2012). Due to my expectation that it is not just distraction but the link between (Christian) religiosity and politics in the United States that suppresses progressive politics, I predict that religious compensation will not suppress liberal economic attitudes as clearly as attitudes more closely related to religious teachings and values (on which disadvantaged groups are not already more consistently liberal). If so, religion is more than just a passive opiate that promotes political quietude through distraction, it is also an active conservatizing force that diverts and then redirects the focus of disadvantaged groups toward traditional social values.

Hypothesis 5: The impact of religious compensation will vary across attitudinal domains and have less of an influence on economic attitudes than social attitudes.

Data, Measures, and Methods

Data

The General Social Survey is a repeated nationally-representative survey of the non- institutionalized American public with consistently fielded questions and special modules fielded in one or a few years. The 2004 GSS—which included a special module on religion with measures of compensatory benefits from religion and spirituality—provides the data for this study. Opiate of the Masses? 14

Measures

Compensatory Psychological Benefits from Religion and Spirituality

Compensatory psychological benefits from religion and spirituality (shortened as “psychological compensators” throughout) are measured with seven questions about the everyday psychological benefits respondents receive from religion. The items asked how frequently (never or almost never=1, once in a while=2, some days=3, most days=4, every day=5, and many times a day=6) people have the following experiences:

1. I find comfort in my religion or spirituality. 2. I find strength in my religion or spirituality. 3. During worship, or at other times when connecting with God,4 I feel joy which lifts me out of my daily concerns. 4. I ask for God's help in the midst of daily activities. 5. I feel guided by God in the midst of daily activities. 6. I feel God's love for me, directly. 7. I feel God's love for me, through others.

I initially conducted a factor analysis, anticipating that different measures would fit better with one another and produce a multi-factor measure. Instead, fit statistics provided strong support for a one-factor solution (illustratively, in a two-factor solution the first factor’s eigenvalue was 5.3 and the second 0.3). I conducted all analyses first with a one-factor solution and then a summative scale (a = .96)5 and found the same patterns (the summative scale and the one-factor solution are correlated at .998). I present the summative scale because the results are the same and this method is simpler and more familiar to some scholars.

4 The GSS included the following text before these questions: “A number of items use the word 'God.' If this word is not a comfortable one, please substitute another idea that calls to mind the divine or holy for you.” 5 I also checked fit statistics for summative scales of various combinations of items and found that combining all items together was supported. Additional analyses using individual items and then using subsets of the items, including a set that could be labeled a narrower “comfort” scale instead of the more general “psychological compensators” scale, yielded the same patterns. Models that included multiple measures or multiple scales for different combinations of these items at the same time suffered from multicollinearity. Opiate of the Masses? 15

I use “psychological compensators” instead of the previous terminology of “supernatural general compensators” for this construct because I diverge from the deprivation-compensation literature by focusing on—and actually measuring—everyday compensatory psychological benefits from religion and spirituality, including perceived comfort, strength, connection, support, guidance, love, and validation. Descriptive statistics for this scale are presented in Table

1.

[Table 1]

Religiosity

Religiosity is measured with three key religious salience and practice measures standard to past research: (1) strength of religious affiliation (no affiliation=1 to strong affiliation=4), (2) attendance frequency (never=1 to more than once a week=9), and (3) prayer frequency (never=1 to several times a day=6). I first considered each measure individually and found similar patterns across measures. I then conducted factor and scaling analyses and determined these measures fit together well and load on one factor. The one-factor solution and a standardized summative scale

(a =.80) provide the same patterns and are correlated with one another at .987. I use the summative scale for the main analyses and include results for the one-factor solution and individual religiosity measures in ancillary analyses. This standardized summative scale of salience, attendance, and prayer follows previous research using the GSS, including Barkan’s

(2014) study on religiosity as a suppressor of women’s support for abortion (also see Barkan

2006; Eggebeen and Dew 2009; Mockabee, Monson, and Grant 2001).

Politics

This study considers several measures of political identities, attitudes, and behaviors, including general measures of political ideology (self-placement on a scale from extremely conservative=1 Opiate of the Masses? 16 to extremely liberal=7), political affiliation (not Republican=1), and voting behavior (voted for

Gore in the 2000 presidential election). I also include attitudes about specific topics on which some marginalized groups are not consistently more liberal selected based on previous research, including a recent study that compared the impacts of sexuality, gender, race, and class across

199 attitude items in the General Social Survey (Schnabel 2016b): abortion attitudes (a seven- item scale), the morality of same-sex relationships (always wrong=1 to not wrong at all=4), other sex-related attitudes (a six-item scale), gender attitudes (a four-item scale), free speech (a 15- item scale), prayer in public schools (approve Supreme Court ruling against required prayer in public schools=1), physician-assisted suicide (support physician-assisted suicide for person with incurable disease=1), and corporal punishment (strongly agree with “a good, hard spanking”=1 to strongly disagree=4). Further analyses presented in the appendix consider additional attitude measures.

Although disadvantaged groups tend to be consistently more liberal on economic issues, which are thus not a site for apparent inconsistency in the positionality principle, these issues are important to Marx’s argument about the suppression of political consciousness. Therefore, I also consider whether psychological compensators suppress attitudes toward redistribution and government spending on societal welfare (a 12-item scale). The scale includes the following items selected based on theoretical relevance of available items and factor analysis6: government should (1) reduce income inequality, and government should spend more on (2) welfare, (3) social security, (4) the environment, (5) the nation’s health, (6) solving problems of big cities, (7)

6 The GSS government spending series includes additional items excluded in factor analysis because they are negatively associated with the others spending measures, were not supported by fit statistics, and/or were not relevant to Marx’s argument. These include spending on military, foreign aid, space exploration, science, roads, and transit. Including these items does not alter the patterns. Excluding theoretically distinct items that were included—

Opiate of the Masses? 17 parks and recreation, (8) assistance for childcare, (9) addressing crime, (10) addressing drug addiction, (11) improving the conditions of Blacks, and (12) improving education. These economic attitude items—which were asked of the same people asked the psychological benefits items—measure rather moderate forms of redistribution and government spending, but some of

Marx’s tenets were more radical. Therefore, further analyses presented in the appendix consider related processes on additional economic attitude measures.

Gender, Race, Class, and Sexuality

Gender is measured as a binary (female=1) and race/ethnicity is measured in four categories

(non-Latino white, non-Latino Black, Latino, and non-Latino other race) with Black vs. non-

Latino white and Latino vs. non-Latino white as the focus comparisons. Class is measured with earnings in tens-of-thousands of dollars, but some analyses also consider class identification (low class=1 to upper class=4) and subjective social status (perception of family income as far below average=1 to far above average=5). Sexuality is measured in three categories (heterosexual

[opposite-sex partners only in the last five years], LGB [had a same-sex partner in the last five years], and no sexual partner information) with LGB vs. heterosexual as the focus comparison.

From 2008 forward the GSS has included a measure of self-reported sexual orientation, but this behavior measure is the available and standardly used measure of sexuality prior to 2008

(Sherkat 2002, 2017; Smith 2006).

such as aid everyone gets rather than just those in need (i.e., social security) and spending on crime and drugs—also does not alter the pattern. Opiate of the Masses? 18

Additional Covariates

The models also include the following standard sociodemographic covariates: education

(binaries for highest degree attained), age (in years), marital status (married=1), parental status

(parent=1), region (South=1), and rurality (rural=1).

Analytic Strategy

I will first examine whether disadvantaged groups score higher on religious compensators. I will then explore the extent to which group differences in religiosity are a function of these compensators. Finally, I will examine the argument that psychological compensators suppress progressive politics, considering (1) general political identities, attitudes, and behaviors, (2) social attitudes linked to religious values, and (3) economic attitudes. In addition to determining whether religious compensation suppresses progressive politics, I will also consider possible explanations for why it might do so.

This study is limited to observations with data on compensatory benefits from religion and on religiosity, each of which are treated as dependent variables in different models. Sample sizes for other analyses vary by the availability of the given outcome measure. Missing data on covariates are imputed using chained equations.7 OLS regression is used for the continuous outcomes and Sobel-Goodman mediation tests are used to determine the percentage and significance of the total effect of an independent variable explained by mediator variables.8

7 Race, gender, marital status, region, and rurality had no missing data, and those with missing data on sexuality were included in the group with no partnering information. Data were imputed on income for 159 observations with missing data (12% of 1323 cases) and for less than 1% of the sample on parental status (missing on three observations), age (two observations), and education (one observation). The main regression analyses utilized multiply-imputed data where the number of imputations equaled 20. Analyses computing percent-of-effects mediated by compensatory benefits from religion utilized single imputation because the sgmediation Stata user- written program is not compatible with multiple imputation. 8 These were conducted using the sgmediation Stata user-written program and cross-checked with the khb user- written program. This method first regresses the dependent variable on the independent variable, then the mediator/suppressor variable on the independent variable, and then the dependent variable on both the independent

Opiate of the Masses? 19

Results

Do the Disadvantaged Receive More Psychological Compensation from Religion?

I first examine the relationship between social status characteristics and compensatory benefits received from religion or spirituality (i.e., “psychological compensators”). Model 1 of Table 2 presents compensatory benefits received from religion and spirituality by gender, race, class, and sexuality. Women, Blacks, Latinos, those with lower income, and sexual minorities each receive more compensatory benefits from religion than their counterparts. The psychological compensators scale is standardized, so women receive .44 of a standard deviation more compensatory benefits from religion and spirituality than men, and Blacks .59 and Latinos .40 of a standard deviation more compensatory benefits than non-Latino whites. Those with less income receive .02 of a standard deviation more compensatory benefits for every $10,000 less they earn, and sexual minorities receive .41 of a standard deviation more compensatory benefits than heterosexuals.

[Table 2]

For each sociodemographic status, the more disadvantaged group receives more compensatory benefits, and receiving these greater benefits are not simply due to disadvantaged groups being more religious. Additional analyses presented in Table A1 demonstrate that each disadvantaged group receives significantly more compensatory benefits among just those with a religious affiliation from which to get benefits and also when controlling for religiosity.9 Further additional analyses presented in Table A2 demonstrate that the “opiate” function of religion that

variable and the mediator/suppressor variable. These models also account for the covariates. Additional analyses with bootstrapped standard errors for the Sobel-Goodman tests did not alter the results. 9 Further supplemental analyses including a categorical measure of religious affiliations demonstrated that these patterns and others presented in the paper are not due to type of affiliation differences across advantage/disadvantage, but disadvantaged groups can find certain strains of religion more appealing than others

Opiate of the Masses? 20 helps compensate for inequalities translate into general happiness premiums. Among people disadvantaged on at least one factor, compensatory benefits from religion are closely linked to greater happiness.10 But, consistent with arguments about religious compensation being more important for the disadvantaged, these benefits from religion are more weakly and not significantly linked to happiness among those advantaged on every factor (i.e., heterosexual white men with higher earnings). Overall, Marx’s claim that religion provides compensatory psychological benefits to the downtrodden is supported, though this extends beyond Marx’s focus on those with less wealth (i.e., material deprivation) to those with less social status (i.e., social deprivation).

The distinction between objective material conditions, group identification, and subjective status cannot be teased out for each of the key status characteristics, but the GSS includes alternative measures of class that make it possible to compare objective material conditions (i.e., family income), identity (i.e., low, working, middle, or upper class identification), and subjective social status (i.e., subjective interpretation of one’s economic status in comparison to others ranging from far below average to far above average). Although interrelated (correlations among the three measures range from .45 to .56), each of these items provides a distinct way to measure class and what it is about being lower in the economic hierarchy that links most directly to receiving more compensatory benefits from religion. When including each of these measures in Model 2 of Table 2, only lower subjective status is significantly linked to receiving more compensatory benefits from religion. Therefore, status as

(Hoffmann and Bartkowski 2008). Moreover, propensity score matching analyses further demonstrate significant status differences in compensatory benefits received from religion. 10 Additional analyses demonstrated that the relationship between psychological compensators and happiness operates net of typical religiosity measures known to be linked to happiness (including strength of affiliation and frequency of attendance and prayer). Opiate of the Masses? 21 perceived in comparison to others appears to be just as if not more important than material deprivation and group identity for seeking psychological compensation from religion.11 The consistent and parallel patterns across gender, race, ethnicity, class, and sexuality, as well as the importance of social status when considering class, provide clear confirmation for the hypothesis that groups with less social status will receive more compensatory benefits from religion and spirituality (Hypothesis 1).

Do Psychological Compensators Account for Group Differences in Religiosity?

Religion provides psychological benefits to the disadvantaged, but do these benefits account for sociodemographic differences in religiosity? Table 3 presents the relationship of gender, race, class, and sexuality with a standardized measure of religiosity in two models. Model 1 presents the relationships without psychological compensators and Model 2 includes psychological compensators.

[Table 3]

Model 1 of Table 3 shows that women are, on average, more religious than men and that

Blacks and Latinos are more religious than non-Latino whites. These relationships are both significant and substantial: women are .36 of a standard deviation more religious than men and

Blacks are .65 and Latinos .27 of a standard deviation more religious than non-Latino whites.

Family income and sexuality are not significantly associated with religiosity in this model.12

When accounting for psychological compensators in Model 2 of Table 3, women are no longer significantly more religious than men and the previously large effect for Black vs. white is substantially smaller.

11 Additional analyses using subjective class status instead of objective income demonstrate that this measure operates similarly to the more commonly used income measure in the results presented throughout the paper. Opiate of the Masses? 22

Table 3 also includes results from a mediation analysis, presenting the percentage of each sociodemographic gap that is a function of psychological compensators. Each of these results were calculated in a separate model with religiosity as the outcome, the given predictor as the effect to be explained, psychological compensators as the mediator, and all other measures included as covariates. First, we see that 86% of the gender gap in religiosity is a function of psychological compensators. To put this percentage in context, Miller and Hoffmann’s (1995) risk preferences explanation, currently the dominant explanation in the literature, accounted for

40% of the gender gap in religiosity and women continued to appear significantly more religious than men when including risk preferences in the model.13

In addition to explaining 86% of the gender gap in religiosity, compensatory benefits from religion and spirituality also account for a significant and substantial percentage of race and ethnicity gaps in religiosity. Although psychological compensators account for 65% of the large race gap, a significant difference persists as demonstrated in Model 2 of Table 3. Psychological compensators mediate 105% of the more moderate ethnicity gap, which means that the coefficient for Latino vs. non-Latino white has reversed. Although there were not significant differences on sexuality or class, psychological compensation is still a significant mediator that reverses the coefficients for both measures.

Table A3 in the online appendix presents the mediation patterns for a religiosity measure created using factor analysis and for each of the religiosity measures individually instead of the summative scale presented in the main analyses. The patterns for the religion factor are

12 The bivariate coefficient for sexuality on the religiosity scale is negative, but is positive in the full model. 13 The large proportion of the gender and others gaps explained by psychological compensators could be due, in part, to compensation-seeking resulting from broad social processes related to status and power, and does not preclude the explanatory power of likely mutually-reinforcing factors such as the performance of gender and sanctions women face for being nonreligious (Edgell et al. 2017; Schnabel 2015b). Opiate of the Masses? 23 equivalent to those for the summative scale, but there are some differences of note on the individual measures despite similar overall patterns. For example, psychological compensators account for more of the Black vs. white gap in prayer than of strength of affiliation or attendance frequency, with the opposite true for gender where more of a prayer gap persists. These patterns suggest that gendering processes over and above compensatory benefits from religion contribute to women enacting an internal and affective form of religious practice more frequently (Schnabel

2015b; Sullins 2006), whereas racial processes over and above compensatory benefits from religion contribute to Blacks being more closely tied to and more likely to be frequently involved with their religious congregations (Ellison and Sherkat 1995). Each of the coefficients for the disadvantaged groups are typically toward more religiosity, but the coefficient for sexual minorities is toward less attendance and accounting for compensatory benefits from religion exaggerates this trend toward secularism.

Do Psychological Compensators Suppress the Liberal Politics of Disadvantaged Groups?

Thus far, I have demonstrated that religion plays the consolation and happiness-providing opiate function Marx proposed. But just because psychological compensators make people happier does not mean that an abstract solution (i.e., religion) will keep people from being concerned about material problems (i.e., inequality). Yet, disadvantaged groups are not always more liberal than their counterparts as would be predicted by status-based theories of politics, and the issues on which some disadvantaged groups are no more liberal (and sometimes more conservative) are frequently linked to religion and religious values. For example, women and Blacks are just as if not more conservative than their counterparts on certain issues of gender/sexuality, free speech, science and the environment, physician-assisted suicide, and corporal punishment (Schnabel

2016b). Opiate of the Masses? 24

To examine whether psychological compensators suppress progressive politics as Marx suggested, I will first present more detailed results for general political ideology and then provide summary results for more measures. The results presented in Table 4 examine whether psychological compensators suppress liberal political ideology on a measure that ranges from extremely conservative=1 to extremely liberal=7. Model 1, which shows the relationships before psychological compensators are added to the model, demonstrates that disadvantaged groups tend toward more liberal political views but that this trend is not significant for Latinos or those with less income. Model 2 includes psychological compensators and each of the patterns are amplified. In other words, religious psychological compensation appears to be suppressing liberal political views. This table also includes results from Sobel-Goodman tests that demonstrate this suppression of liberal political views by psychological compensation is substantial and significant.14 For example, the already large Black-white gap in liberal political views would be 60% larger if it were not for the comfort Blacks find in religion.

[Table 4]

Table 5 presents summary results for additional outcome measures. The table focuses on whether liberal politics for each group are suppressed by psychological compensators (or, alternatively, whether conservative politics are mediated). This suppression is calculated by whether Sobel-Goodman mediation analyses provide evidence for a significant amplification of liberal politics when accounting for compensatory benefits from religion. I first present measures of general politics in the form of self-placement on a conservative-to-liberal identity scale (which

I presented in more detail above), party identity, and voting behavior. I then present patterns for some key sociopolitical attitudes on which previous work has shown one or more disadvantaged

14 Mediation can be significant without the original effect being significant when the relationships between the

Opiate of the Masses? 25 group to be no more liberal or even possibly more conservative than their counterparts. Sample sizes, and thus statistical power, vary by the availability of each outcome measure.

[Table 5]

As shown above and reiterated in Table 5, religious psychological compensation suppresses liberal political ideology for women, Blacks, Latinos, those with lower incomes, and sexual minorities. The same pattern holds for political affiliation, with psychological compensation suppressing disadvantaged people’s likelihood of identifying as something besides

Republican (Democrat, independent, or other party). On the third general politics measure, psychological compensators significantly suppress the likelihood that women, Blacks, Latinos, and those with less income voted for Gore in the 2000 presidential election. Similar patterns appear on attitudes toward specific issues that can appear to contradict the positionality principle

(i.e., issues on which some disadvantaged groups are not always more liberal). On each of these issues psychological compensators suppress liberal attitudes for women and Blacks, and on some issues psychological compensation also suppresses liberal attitudes for Latinos, those with less income, and sexual minorities. Therefore, psychological compensators suppress liberal attitudes and can be used to understand why disadvantaged groups are not always more liberal than their counterparts. Moreover, these compensatory benefits from religion frequently suppress what would otherwise be even more liberal attitudes when disadvantaged groups are already more liberal.

Because compensatory benefits from religion were fielded as part of a GSS special module that did not overlap with some key attitudinal items, I conducted additional analyses with religiosity as a potential suppressor of progressive politics on same-sex marriage, science and the

predictor and mediator and between the mediator and outcome are strong. Opiate of the Masses? 26 environment, the legislation of morality, and beliefs about the right to civil disobedience15 using

GSS data from 2000-2014. As presented in Table A4, religiosity suppresses progressive politics

(and/or mediates conservative attitudes) among women and Blacks on each additional outcome measure, and among Latinos and those with less income on most outcomes.16

Why Does Religious Psychological Compensation Suppress Progressive Politics?

Does religion suppress progressive politics only because of the other-worldly distraction highlighted by Marx, or also because religiosity is a generally conservatizing force in the contemporary United States? Table 6 explores whether other-worldly distraction (proxied by belief in an afterlife) and/or general religiosity help account for the strong link between religious psychological compensators and conservative political ideology. Model 1 highlights the relationship between psychological compensation and conservative ideology before including afterlife belief or religiosity. Models 2 and 3 add just afterlife belief and then just religiosity, and

Model 4 includes both.17

[Table 6]

Based on the patterns presented in Table 6, it appears to be the interrelationship between psychological compensators and active religiosity that is the key reason psychological

15 In addition to beliefs about the right to civil disobedience, the GSS has fielded two older measures on confidence to protest against government action the respondents strongly oppose. These questions, likelihood to attend a public protest meeting and likelihood of attending a protest march, were fielded in 1996 (ethnicity was not fielded until 2000). Similar to the right to civil disobedience item presented in Table A4, religiosity suppressed women and Blacks’ confidence in protesting against government action they strongly opposed, but did not suppress liberal responses among those with lower incomes or sexual minorities. 16 Although sexual minorities receive substantially more compensatory benefits from religion and spirituality than do heterosexuals, they do not tend to be more religious and are frequently less religious on conventional measures of religiosity. Therefore, religiosity does not suppress their liberal consciousness, which appears to be part of why sexual minorities are so consistently more liberal than heterosexuals (Schnabel 2016b). 17 Collinearity diagnostics indicate acceptable VIF levels even with compensatory benefits from religion (VIF=2.54), afterlife belief (VIF=1.21), and religiosity (VIF=2.45) all in the same model (Model 4). Opiate of the Masses? 27 compensation suppresses progressive politics.18 In other words, psychological compensators suppress progressive politics primarily because they facilitate and reinforce religiosity among disadvantaged groups. This pattern, which is not particularly surprising in a political context that links religiousness and conservative politics, expands on Marx’s notion of religion as a passive distraction to suggest that religion can also be an active conservatizing force.

I conducted further analyses on the other political outcomes and found similar patterns.

Therefore, religious compensation appears to be a general suppressor of progressive politics not simply because of an other-worldly focus that operates as a passive distraction to promote political quietude,19 but also because of the actively conservatizing force that is American religiosity.

Does Psychological Compensation Suppress the Progressive Politics of Disadvantaged Groups on Economic Issues?

Thus far the study has focused on general politics and specific issues that appear to be inconsistent with the underdog positionality principle. As a further test of Marx’s idea of the opiate of the masses, I consider a specific topic on which disadvantaged groups already tend to be consistently more liberal: economic redistribution and government spending on societal welfare. When Marx spoke about the opiate of the masses, he used “the masses” to refer primarily to those who were not economically elite, not those with less social status. And when

18 Additional mediation analyses demonstrate that when religiosity is included as a covariate, compensatory benefits from religion are no longer a significant suppressor of progressive politics on any sociodemographic divide. When afterlife belief is included, compensatory benefits from religion are still a significant suppressor of liberal ideology for gender, race, and ethnicity and a marginally significant suppressor for earnings and sexuality. 19 Rather than simply promoting political quietude, American religion is highly politicized, and politicized in such a way that it can be an actively conservatizing force. Additional analyses on measures that do not overlap with the psychological compensators scale show that religiosity is associated with discussing politics more frequently and trying to persuade others more frequently. This pattern is likely influenced by religion providing a community of people with whom to discuss politics. In addition to conservatizing mechanisms highlighted in past research, political discussion among congregants provides a further mechanism by which religiosity can facilitate conservative politics. Opiate of the Masses? 28 he explained how religion would suppress the political consciousness of the masses due to abstract spiritual distraction from their concrete material conditions, he spoke primarily about their economic political consciousness. Table 7, therefore, examines whether psychological compensation suppresses progressive politics on economic issues.

[Table 7]

Model 1 of Table 7 shows that apart from sexual minorities, disadvantaged groups are significantly more liberal on economic issues. Although the 30 sexual minorities in this analysis—a small proportion of the sample that yields limited statistical power20—are not significantly more liberal like the other disadvantaged groups, the coefficient is in the predicted direction. Model 2, which adds psychological compensators to the model, shows that psychological compensation is negatively associated with liberal economic attitudes, but that this association is not as strong as that for the other political outcomes. Mediation analysis demonstrates that psychological compensation, while a consistent suppressor of liberal politics in other areas, does not significantly suppress liberal economic attitudes for any of the five disadvantaged groups. Additional analyses on individual items from the scale, including the

“government should reduce inequality” item which was fielded separately from the other measures, yielded the same patterns.

As noted earlier, because compensatory benefits from religion were fielded as part of a

GSS special module they does not overlap with some key attitude items. In fact, the measures about income redistribution and government spending considered above could be considered neoliberal economics rather than the more radical and emancipative economics espoused by

20 Previous research with larger samples from the GSS has shown sexual minorities are significantly more liberal on welfare issues. Of 26 economy and role of government issues considered by Schnabel (2016b), the coefficient was in the direction of sexual minorities being more liberal on all 26 items and these effects were significant on 18 items. Opiate of the Masses? 29

Marx. Therefore, I considered whether religiosity, which is more consistently available and thus overlaps with more attitude outcomes, suppresses liberal views of economic inequality and class conflict on additional items. As presented in Table A5, religiosity can suppress liberal views on economic inequality, but only on certain measures, each of which are associated with religiosity and provide relatively large samples. For example, religiosity is positively and significantly associated with the belief that income inequality generates prosperity (which is consistent with

Weber’s argument about religion and capitalism), and religiosity subsequently suppresses liberal attitudes on this issue among women, Blacks, Latinos, and those with lower incomes. However, on other measures less closely linked to religiosity—including cognizance of class conflict and the importance of an adequate standard of living for all—religiosity is not a significant suppressor of liberal economic attitudes among disadvantaged groups. In short, religion suppresses the progressive politics of disadvantaged groups on issues with notable religious divides, but not on issues unrelated to religion.

Although this pattern of less consistent suppression on economic issues appears to be inconsistent with Marx’s argument, it is consistent with this study’s finding that religion suppresses liberal politics not because of distraction but because of the conservatizing power of religion on certain issues. Whereas religiosity is a particularly strong conservatizing force on issues of traditional morality, there is a weaker link between religion and certain economic attitudes. Subsequently, religious compensation does suppress progressive politics, but less consistently on the issues Marx emphasized than on other issues more strongly linked to religiosity. Opiate of the Masses? 30

Discussion

This study considered Marx’s claim that religion is the opiate of the masses and whether this argument can be applied to the contemporary United States. I applied Marx’s general idea beyond “the masses” (i.e., the poor) to the socially disenfranchised and to issues beyond economics. I used this extrapolation on the “opiate of the masses” concept to help explain two key social scientific puzzles: the gender gap in religiosity and why some disadvantaged groups are not consistently liberal across sociopolitical issues. Disadvantaged groups receive more compensatory psychological benefits from religion, and religion as an “opiate” of the disenfranchised has four key functions: (1) provides comfort and happiness to the disadvantaged,

(2) motivates and/or reinforces religiosity, (3) promotes sociodemographic religiosity gaps, and

(4) suppresses progressive politics.

American religion appeals to the disadvantaged and suppresses their progressive politics, but it does so in ways beyond Marx’s focus. Religious coping mechanisms are important to not just the poor, but also the socially disenfranchised. Religious compensation is not just a distraction, it facilitates and reinforces religiosity and its conservatizing influences. And religion is an important suppressor of progressive politics on issues tied to traditional religious morality and general political polarization, but is not an important suppressor of progressive economic attitudes. Insofar as religion appeals to the poor, provides comfort, and suppresses progressive politics, it is the opiate of the masses. It is more than that, however, because it also appeals to the socially disadvantaged and actively redirects their focus toward traditional values. Therefore, it may be more accurate to call contemporary American religion the diversion of the disenfranchised rather than the opiate of the masses. This extrapolation of Marx’s proposition can be used to rethink religion in the social sciences, explain previously unexplained Opiate of the Masses? 31 sociodemographic differences in religiosity, and even account for apparent inconsistencies in the positionality principle.

Sociodemographic differences in religiosity, and especially gender differences in religiosity, have been contentiously debated for decades. The research has typically focused on one sociodemographic gap at a time and thus not situated these individual gaps within a broader and simultaneous consideration of gender, race, ethnicity, class, and sexuality. I found that disadvantaged groups in the realms of gender, race, class, and sexuality all use religion and spirituality as coping mechanisms and this psychological compensation accounts for group religiousness gaps. Of the group differences in religiosity considered in this study, only Black vs. white differences persist after accounting for psychological compensation. Although Blacks remain more religious than whites after accounting for compensatory benefits from religion, this racial gap is a third of what it was before accounting for this compensation. These findings place the gender gap in religiosity within a larger framework of social status differences and may finally lay to rest essentialist arguments about gender differences in religiosity.

The positionality principle of politics and similar status-based theories of social attitudes argue that disenfranchised groups will be consistently more liberal. Therefore, scholars have long wondered why disadvantaged groups, and especially women, do not have more consistently liberal attitudes on issues like abortion (Barkan 2014; Wilcox 1992). This study provides support for the positionality principle of social attitudes, explains what at first appear to be inconsistencies, and highlights the previously underemphasized importance of how religion complicates the relationship between status and progressive politics. Marginalized groups are both more liberal and more likely to receive comfort from religion, and this comfort from religion suppresses what would otherwise be even more liberal politics. According to Marx Opiate of the Masses? 32

(1970), this suppression is due to religion being an “inverted consciousness,” or focus on other- worldly compensation to address this-worldly disadvantage. But at least in the United States, religion and politics are linked in such a way that religion does not make people less political,21 it changes their politics (Edgell 2012; Perry and Schnabel 2017; Tranby and Zulkowski 2012).

Subsequently, I showed that rather than just other-worldly distraction from material inequality, psychological compensators appear to suppress progressive politics by facilitating and reinforcing religiosity, which subsequently suppresses progressive politics.

Psychological benefits from religion are a central factor in why some people are more religious than others. I do not mean to suggest that psychological benefits are the only reason people are religious, but it is clear that religion provides important compensatory benefits that appeal to the disenfranchised in society. To call American religion an opiate is not to deem it primitive or otherwise denigrate it, nor does it mean that all religion is an opiate (the opiate function Marx referred to was specific to Christianity and its appeal to the oppressed). In fact, one could argue that thinking about religion as the opiate of the masses highlights its value to society. For example, religion helps people, especially the downtrodden and disadvantaged, gain well-being and make meaning out of hardship and thus provides, as Marx said, a “heart for a heartless world.” If we were to think about religion as a social and psychological good and meaning-making cultural tool, which has already been hinted at in the religion-as-rational-choice and cultural lived-religion literatures, it should not be surprising that people who get more benefit from it would be more religious. Religion is not an impotent cultural resource, however, and will justify and reinforce social boundaries, thereby conservatizing the identity and outlook

21 There are some specific cases, such as the Jehovah’s Witnesses and Amish, where small religious groups reject earthly politics (and thus do not vote, run for office, etc.) for a reason that matches Marx’s opiate argument well: a belief, based on a biblical statement, about God’s kingdom being not of the world. Opiate of the Masses? 33 of those who rely on it in ways that exacerbate inequalities (Edgell 2012; Pachucki, Pendergrass, and Lamont 2007).

Although religion often promotes traditional values because religious teachings tend to get fixed in a point in time and the symbolic boundaries promoted by religion tend to harken back to the past (Edgell 2012; Tranby and Zulkowski 2012), religion is dynamic and the social message it carries with it is at once both traditional and contemporary. Rather than fixed in perpetuity, the particular “traditional” issues by which religion (or, more accurately, religions) defines itself shift over time and are both time and place specific (Schnabel 2016a). Therefore, the landscape of Christian religion that Marx referred to was distinct from the landscape of

Christian religion of contemporary American society, which is closely linked to socially conservative politics (Bean 2014; Goren and Chapp 2017; Hout and Fischer 2002, 2014; Putnam and Campbell 2010; Steensland and Wright 2014; Wuthnow 1988). This study indicates that

Marx’s statement about religion in his particular historical context can still be applied in principle, but that the interrelationship between disadvantaged status, religious compensation, and political outlook is fluid and specific to a particular status-religion-politics nexus. If religion were more closely linked to economically conservative politics and less closely linked to socially conservative politics in the contemporary United States, I suspect that the patterns demonstrated in this study would have fallen more in line with what one might have predicted based upon

Marx’s argument.

The data available for this study were cross-sectional, and the link between religion and socially conservative politics is so strong in the contemporary United States that some liberals Opiate of the Masses? 34 select out of religion (Hout and Fischer 2002, 2014).22 While politics-to-religion selection is important and likely explains some of the relationship between religiosity and conservative politics for the population as a whole, there is not a clear reason to expect disadvantaged groups are simply selecting into more religiosity because of their politics (which tend to be more liberal). It would appear, then, that religion likely has a conservatizing influence on the politics of disadvantaged groups. The data focused on the United States and did not have a sufficient sample size to separately consider the patterns for non-Christian religions.23 Christianity has been framed by many, including Marx, as a sympathetic underdog religion that appeals to the disadvantaged. Other religions may or may not provide the same compensatory psychological benefits and may or may not suppress progressive politics. It is also possible that some specific forms of Christianity—such as Marxist liberation theology that emphasizes this-worldly salvation in the form of social justice instead of hope for the hereafter among Catholics in Latin

America (Smith 1991)—would not suppress progressive politics. In addition to possible heterogeneity by religious tradition, the patterns could also vary by other societal factors, such as economic development. For example, Marx’s argument about religion suppressing economic class consciousness may be more clearly borne out where people are more economically destitute in absolute terms. The data were also limited in that the GSS, like other major surveys, does not ask about some of the more radical economic tenets set forth by Marx, and therefore

22 Although the compensatory benefits from religion items are only available cross-sectionally, in recent years the GSS included a panel design. Analyses presented in Table A6 of the online supplement show that religiosity similarly suppresses liberal political ideology in panel models. 23 Additional analyses by religious affiliation suggest that the patterns may operate differently for non-Christians. Although there were only 50 available in the sample, the patterns for these 50 non-Christians were notably different than for the Christians and the unaffiliated (many of whom hold Christian beliefs and/or are former Christians). Among people in non-Christian religious traditions, disadvantaged groups do not receive more psychological compensation. In fact, the coefficient for Black and sexual minority are the reverse of what they were among Christians and among the unaffiliated (i.e., toward Blacks and sexual minorities receiving less psychological

Opiate of the Masses? 35 some aspects of class consciousness and support for class revolution were not measured. I hope that future research will examine further measures, groups, and contexts.

I drew on Marx to argue that religion can be understood, at least in part, as a social and psychological good, and that religiousness can be predicted by people getting more from it. We could likewise predict that people will be less religious where they have less need for existential comfort. Marx was an early proponent of the as yet unsubstantiated secularization hypothesis, assuming that religion would decline into obscurity as society advanced. Whereas others tended to argue that the advancement of science would lead to the decline of religion, Marx argued that it was not just science but material security and equality that would lead to religion’s demise.

Although we have seen religion decline most dramatically in particularly egalitarian Democratic

Socialist contexts with strong social safety nets,24 neither the egalitarian revolution nor the global demise of religion have occurred. Therefore, we might expect that as long as hardships and inequalities persist—and secular people have low fertility (Hackett et al. 2015)—the dual-edged sword that is religion will also persist, providing comfort and strength to the downtrodden at the same time that it diverts them from the real cause of their suffering.

compensation than their counterparts). Therefore, the patterns demonstrated in this study may be specific to Christians and/or Christian contexts. 24 It is possible that American religion’s unique, yet still declining (Voas and Chaves 2016), vitality in the context of comparable countries is in part a function of similarly unique inequalities, policies (e.g., lack of national healthcare that disproportionately affects women, Blacks, etc.), and subsequent vulnerability of disadvantaged groups. Opiate of the Masses? 36

REFERENCES

Adamczyk, Amy, Katharine A. Boyd, and Brittany E. Hayes. 2016. “Place Matters:

Contextualizing the Roles of Religion and Race for Understanding Americans’ Attitudes

about Homosexuality.” Social Science Research 57(1):1–16.

Barkan, Steven E. 2006. “Religiosity and Premarital Sex in Adulthood.” Journal for the

Scientific Study of Religion 45(3):407–17.

Barkan, Steven E. 2014. “Gender and Abortion Attitudes.” Public Opinion Quarterly 78(4):940–

50.

Bean, Lydia. 2014. The Politics of Evangelical Identity: Local Churches and Partisan Divides in

the United States and Canada. Princeton, NJ: Princeton University Press.

Becker, Sascha O. and Ludger Woessmann. 2013. “Not the Opium of the People: Income and

Secularization in a Panel of Prussian Counties.” American Economic Review 103(3):539–

44.

Beit-Hallahmi, Benjamin. 2014. Psychological Perspectives on Religion and Religiosity. New

York: Routledge.

Blouin, David, Robert V. Robinson, and Brian Starks. 2013. “Are Religious People More

Compassionate and Does This Matter Politically?” Politics and Religion 6(3):618–45.

Box-Steffensmeier, Janet M., Suzanna De Boef, and Tse-min Lin. 2004. “The Dynamics of the

Partisan Gender Gap.” American Political Science Review 98(3):515–28.

Christopher, Stefan, John Fearon, John Mccoy, and Charles Nobbe. 1971. “Social Deprivation

and Religiosity.” Journal for the Scientific Study of Religion 10(4):385–92.

Collett, Jessica L. and Omar Lizardo. 2009. “A Power-Control Theory of Gender and

Religiosity.” Journal for the Scientific Study of Religion 48(2):213–31.

Opiate of the Masses? 37

Collins, Patricia Hill. 2000. Black Feminist Thought: Knowledge, Consciousness and the Politics

of Empowerment. New York: Routledge.

Davenport, Lauren D. 2016. “Beyond Black and White: Biracial Attitudes in Contemporary U.S.

Politics.” American Political Science Review 110(1):52–67.

Davis, Kingsley. 1948. Human Society. New York: MacMillan.

Davis, Nancy J. and Robert V. Robinson. 1991. “Men’s and Women’s Consciousness of Gender

Inequality: Austria, West Germany, Great Britain, and the United States.” American

Sociological Review 56(1):72.

Davis, Nancy J. and Robert V. Robinson. 1999. “Their Brothers’ Keepers? Orthodox

Religionists, Modernists, and Economic Justice in Europe.” American Journal of Sociology

104(6):1631–65.

Davis, Nancy J. and Robert V. Robinson. 2012. Claiming Society for God: Religious Movements

and Social Welfare in Egypt, Israel, Italy, and the United States. Bloomington, IN: Indiana

University Press.

Edgell, Penny. 2012. “A Cultural Sociology of Religion: New Directions.” Annual Review of

Sociology 38(1):247–65.

Edgell, Penny, Jacqui Frost, and Evan Stewart. 2017. “From Existential to Social Understandings

of Risk: Examining Gender Differences in Non-Religion.” Social Currents Online 1st.

Eggebeen, David and Jeffrey Dew. 2009. “The Role of Religion in Adolescence for Family

Formation in Young Adulthood.” Journal of Marriage and Family 71(1):108–21.

Ellison, Christopher G. and Darren E. Sherkat. 1995. “The ‘Semi-Involuntary Institution’

Revisited: Regional Variations in Church Participation among Black Americans.” Social

Forces 73(4):1415–37. Opiate of the Masses? 38

Ellison, Christopher G. and Robert Joseph Taylor. 1996. “Turning to Prayer: Social and

Situatioal Antecedents of Religious Coping Among African Americans.” Review of

Religious Research 38(2):111–31.

Felson, Jacob and Heather Kindell. 2007. “The Elusive Link between Conservative Protestantism

and Conservative Economics.” Social Science Research 36(2):673–87.

Francis, Leslie J. 1997. “The Psychology of Gender Differences in Religion: A Review of

Empirical Research.” Religion 27(1):81–96.

Freese, Jeremy and James D. Montgomery. 2007. “The Devil Made Her Do It? Evaluating Risk

Preference as an Explanation of Sex Differences in Religiousness.” Pp. 187–230 in

Advances in Group Processes: The Social Psychology of Gender, edited by Shelley J.

Correll. Oxford: Elsevier.

Frost, Jacqui and Penny Edgell. 2017. “Distinctiveness Reconsidered: Religiosity, Structural

Location, and Understandings of Racial Inequality.” Journal for the Scientific Study of

Religion In Press.

Glock, Charles. 1964. “The Role of Deprivation in the Origin and Evolution of Religious

Groups.” in Religion and Social Conflict, edited by Robert Lee and Martin E. Marty. New

York: Oxford University Press.

Glock, Charles, Benjamin Ringer, and Earl Babbie. 1967. To Comfort and to Challenge.

Berkeley: University of California Press.

Glock, Charles and Rodney Stark. 1965. Religion and Society in Tension. Chicago: Rand

McNally and Company.

Goren, Paul and Christopher Chapp. 2017. “Moral Power: How Public Opinion on Culture War

Issues Shapes Partisan Predispositions and Religious Orientations.” American Political Opiate of the Masses? 39

Science Review. In press.

Greeley, Andrew and Michael Hout. 2006. The Truth about Conservative Christians: What They

Think and What They Believe. Chicago: Chicago University Press.

Hackett, Conrad, Marcin Stonawski, Michaela Potanč oková, Brian J. Grim, and Vegard

Skirbekk. 2015. “The Future Size of Religiously Affiliated and Unaffiliated Populations.”

Demographic Research 32(April):829–42.

Hastings, Orestes P. and D.Michael Lindsay. 2013. “Rethinking Religious Gender Differences:

The Case of Elite Women.” Sociology of Religion 74(4):471–95.

Hoffman, Martin L. 2000. Empathy and Moral Development: Implications for Caring and

Justice. Cambridge: Cambridge University Press.

Hoffmann, John P. and John P. Bartkowski. 2008. “Gender, Religious Tradition, and Biblical

Literalism.” Social Forces 86(3):1245–72.

Hout, Michael and Claude S. Fischer. 2002. “Why More Americans Have No Religious

Preference: Politics and Generations.” American Sociological Review 67(2):165–90.

Hout, Michael and Claude S. Fischer. 2014. “Explaining Why More Americans Have No

Religious Preference: Political Backlash and Generational Succession, 1987-2012.”

Sociological Science 1(October):423–47.

Hunt, Matthew O. 1996. “The Individual, Society, or Both? A Comparison of Black, Latino, and

White Beliefs about the Causes of Poverty.” Social Forces 75(1):293–322.

Hunt, Matthew O. 2007. “African American, Hispanic, and White Beliefs about Black/White

Inequality, 1977–2004.” American Sociological Review 72(3):390–415.

Jost, John T. et al. 2014. “Belief in a Just God (and a Just Society): A System Justification

Perspective on Religious Ideology.” Journal of Theoretical and Philosophical Psychology Opiate of the Masses? 40

34(1):56–81.

Kinder, Donald R. and Nicholas Winter. 2001. “Exploring the Racial Divide: Blacks, Whites,

and Opinion on National Policy.” American Journal of Political Science 45(2):439–56.

Lizardo, Omar. 2009. “The Devil as Cognitive Mapping.” Rethinking Marxism 21(4):605–18.

Luckmann, Thomas. 1967. The Invisible Religion: The Problem of Religion in Modern Society.

New York: Macmillan.

Manza, Jeff and Clem Brooks. 1998. “The Gender Gap in U.S. Presidential Elections: When?

Why? Implications?” American Journal of Sociology 103(5):1235–1266.

Marx, Karl. 1970. Marx’s Critique of Hegel’s Philosophy of Right. Cambridge: Cambridge

University Press.

Miller, Alan S. and Rodney Stark. 2002. “Gender and Religiousness: Can Socialization

Explanations Be Saved?” American Journal of Sociology 107(6):1399–1423.

Mockabee, Stephen T., Joseph Quin Monson, and J.Tobin Grant. 2001. “Measuring Religious

Commitment among Catholics and Protestants: A New Approach.” Journal for the

Scientific Study of Religion 40(4):675–90.

Norris, Pippa and Ronald Inglehart. 2008. “Existential Security and the Gender Gap in Religious

Values (Unpublished Working Paper).”

Norris, Pippa and Ronald Inglehart. 2011. Sacred and Secular: Religion and Politics Worldwide.

2nd ed. New York: Cambridge University Press.

Pachucki, Mark A., Sabrina Pendergrass, and Michele Lamont. 2007. “Boundary Processes:

Recent Theoretical Developments and New Contributions.” Poetics 35(6):331–51.

Perry, Samuel and Landon Schnabel. 2017. “Seeing Is Believing: Religious Media Consumption

and Public Opinion toward Same-Sex Relationships.” Social Currents Online 1st. Opiate of the Masses? 41

Putnam, Robert D. and David E. Campbell. 2010. American Grace: How Religion Divides and

Unites Us. New York: Simon & Schuster.

Rankin, Lindsay E., John T. Jost, and Cheryl J. Wakslak. 2009. “System Justification and the

Meaning of Life: Are the Existential Benefits of Ideology Distributed Unequally across

Racial Groups?” Social Justice Research 22(2–3):312–33.

Robinson, Robert and Wendell Bell. 1978. “Equality, Success, and Social Justice in England.”

American Sociological Review 43(2):125–43.

Schnabel, Landon. 2015a. “How Religious Are American Women and Men? Gender Differences

and Similarities.” Journal for the Scientific Study of Religion 54(3):616–22.

Schnabel, Landon. 2015b. More Religious, Less Dogmatic: Reexamining Gender Differences in

Religion. Paper presented at the Annual Meetings of the American Sociological

Association.

Schnabel, Landon. 2016a. “Gender and Homosexuality Attitudes across Religious Groups from

the 1970s to 2014: Similarity, Distinction, and Adaptation.” Social Science Research

55(1):31–47.

Schnabel, Landon. 2016b. Sexuality and Social Attitudes. Paper presented at the Annual

Meetings of the American Association for Public Opinion Research.

Schnabel, Landon. 2016c. “The Gender Pray Gap: Wage Labor and the Religiosity of High-

Earning Women and Men.” Gender & Society 30(4):643–69.

Schuman, Howard and John Harding. 1963. “Sympathetic Identification with the Underdog.”

Public Opinion Quarterly 27(2):230–41.

Sherkat, Darren E. 2002. “Sexuality and Religious Commitment in the United States: An

Empirical Examination.” Journal for the Scientific Study of Religion 41(2):313–23. Opiate of the Masses? 42

Sherkat, Darren E. 2017. “Sexuality and Religious Commitment Revisited: Exploring the

Religious Commitments of Sexual Minorities from 1991-2014.” Journal for the Scientific

Study of Religion. Online 1st.

Sidanius, Jim, Felicia Pratto, and Lawrence Bobo. 1994. “Social Dominance Orientation and the

Political Psychology of Gender: A Case of Invariance?” Journal of Personality and Social

Psychology 67(6):998–1011.

Simmel, Georg. 1971. On Individuality and Social Forms. Chicago, IL: University of Chicago

Press.

Smith, Christian. 1991. The Emergence of Liberation Theology: Radical Religion and Social

Movement Theory. Chicago: Chicago University Press.

Smith, Tom W. 2006. GSS Report: American Sexual Behavior: Trends, Socio-Demographic

Differences, and Risk Behavior. General Social Survey Topical Report No. 25. National

Opinion Research Center, Chicago.

Stark, Rodney and William Sims Bainbridge. 1985. The Future of Religion: Secularization,

Revival, and Cult Formation. Berkeley: University of California Press.

Stark, Rodney and William Sims Bainbridge. 1987. A Theory of Religion. New York: Peter

Lang.

Stark, Rodney, Laurence R. Iannaccone, and Roger Finke. 1996. “Linkages between Economics

and Religion: Religion, Science, and Rationality.” The American Economic Review

86(2):433–37.

Steensland, Brian and Eric L. Wright. 2014. “American Evangelicals and Conservative Politics:

Past, Present, and Future.” Sociology Compass 8(5):705–17.

Stewart, Evan, Jacqui Frost, and Penny Edgell. 2017. “Intersectionality and Power.” Secularism Opiate of the Masses? 43

& Nonreligion 6(6):1–3.

Sullins, D.Paul. 2006. “Gender and Religion: Deconstructing Universality, Constructing

Complexity.” American Journal of Sociology 112(3):838–80.

Tranby, Eric and Samantha E. Zulkowski. 2012. “Religion as Cultural Power: The Role of

Religion in Influencing Americans’ Symbolic Boundaries around Gender and Sexuality.”

Sociology Compass 6(11):870–82.

Vaisey, Stephen. 2009. “Motivation and Justification: A Dual-Process Model of Culture in

Action.” American Journal of Sociology 114(6):1675–1715.

Voas, David and Mark Chaves. 2016. “Is the United States a Counterexample to the

Secularization Thesis?” American Journal of Sociology 121(5):1–40.

Walter, Tony and Grace Davie. 1998. “The Religiosity of Women in the Modern West.” The

British Journal of Sociology 49(4):640–60.

Wedow, Robbee, Landon Schnabel, Lindsey Wedow, and Mary Ellen Konieczny. 2017. “‘I’m

Gay and I’m Catholic’: Negotiating Two Complex Identities at a Catholic University.”

Sociology of Religion (Forthcoming).

Wilcox, Clyde. 1992. “Race, Religion, Region and Abortion Attitudes.” Sociology of Religion

53(1):97.

Wilde, Melissa and Lindsay Glassman. 2016. “How Complex Religion Can Improve Our

Understanding of American Politics.” Annual Review of Sociology 42(1):407–25.

Wimberley, Dale W. 1984. “Socioeconomic Deprivation and Religious Salience: A Cognitive

Behavioral Approach.” The Sociological Quarterly 25(2):223–38.

Wuthnow, Robert. 1988. The Restructering of American Religion: Society and Faith Since World

War II. Princeton, NJ: Princeton University Press. Opiate of the Masses? 44

TABLES

Table 1: Descriptive Statistics for Key Measures Mean/ Measures Metric N Proportion SD Range Primary Measures Psychological Compensators Standardized Seven-Item Scale (a = .96) 1,323 0.000 1.000 -1.969–1.457 Religiosity Scale Standardized Three-Item Scale (a = .80) 1,323 0.000 1.000 -1.916–1.499 Female Female=1 1,323 0.535 Race 1,323 White, Non-Latino White=1 0.750 Black, Non-Latino Black=1 0.116 Latino Latino=1 0.091 Other Race, Non-Latino Other Race=1 0.043 Family Income In Tens-of-Thousands of U.S. Dollars 1,323 5.182 4.300 0.047-15.105 Sexuality 1,323 Heterosexual Opposite-Sex Partners Only in Last Five Years 0.661 LGB Had a Same-Sex Partner in Last Five Years 0.023 No Partnering Information No Partnering Information (No Partners or Missing) 0.317 Additional Class Measures Class Identification Lower Class=1 to Upper Class=4 1,323 2.428 0.670 1-4 Subjective Class Status Far Below Average=1 to Far Above Average=5 1,323 2.914 0.877 1-5 Additional Key Outcome Measures Political Ideology Extremely Conservative=1 to Extremely Liberal=7 1,294 3.771 1.406 1–7 Economic Attitudes Standardized 12-item Redistribution and Government 1,322 0.000 1.000 -4.136–2.382 Spending Scale (a = .72) Source: 2004 General Social Survey Note: Family income was missing on 12% of cases, which were imputed using chained equations where the number of imputations equals 20. Descriptive statistics for family income are equivalent before and after imputation. Opiate of the Masses? 45

Table 2: Disadvantaged Social Status Predicting Religious Psychological Compensators Model 1 Model 2 Key Status Measures Female 0.436*** 0.434*** (0.052) (0.052) Black 0.590*** 0.587*** (0.085) (0.085) Latino 0.400*** 0.395*** (0.093) (0.093) Family Income (in tens-of-thousands of dollars) -0.019** -0.013 (0.007) (0.008) LGB 0.422* 0.385* (0.175) (0.176) Alternative Measures of Economic Standing Class Identification 0.021 (0.046) Subjective Class Status -0.072* (0.036)

Covariates Yes Yes Constant -0.811 -0.690 N 1,323 1,323 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Opiate of the Masses? 46

Table 3: Disadvantaged Social Status and Psychological Compensators Predicting Religiosity % of Total Difference Model 1: Model 2: Mediated by Without With Psychological Compensators Compensators Compensators Female 0.362*** 0.050 86%*** (0.053) (0.038) Black 0.647*** 0.224*** 65%*** (0.085) (0.061) Latino 0.271** -0.015 105%*** (0.094) (0.067) Family Income (in tens-of-thousands of dollars) -0.010 0.004 145%* (0.007) (0.005) LGB 0.172 -0.130 176%* (0.177) (0.125) Psychological Compensators 0.717*** (0.020) Covariates Yes Yes Constant -1.069 -0.487 N 1,323 1,323 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. Sobel-Goodman tests used to determine percentage and significance of religiosity mediated by psychological compensators. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Opiate of the Masses? 47

Table 4: Disadvantaged Social Status and Psychological Compensators Predicting Liberal Political Ideology % of Total Difference Model 1: Model 2: Mediated by Without With Psychological Compensators Compensators Compensatorsa Female 0.235** 0.361*** -55%*** (0.078) (0.079) Black 0.294* 0.465*** -60%*** (0.128) (0.128) Latino 0.165 0.269 -68%*** (0.140) (0.139) Family Income (in tens-of-thousands of dollars) -0.010 -0.015 -44%* (0.011) (0.011) LGB 0.661* 0.787** -19%* (0.260) (0.256) Psychological Compensators -0.285*** (0.041) Covariates Yes Yes Constant 4.430 4.196 N 1,294 1,294 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. Sobel-Goodman tests used to determine percentage and significance of political ideology mediated by psychological compensators. a Negative mediation is suppression (i.e., psychological compensators are suppressing liberal ideology). A mediation can be significant without the original effect being significant when the relationships between the predictor and mediator and between the mediator and outcome are strong. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Opiate of the Masses? 48

Table 5: Psychological Compensators as a Suppressor of Liberal Politics among Disadvantaged Groups across Measures Liberal Politics Suppressed by Psychological Compensators Lower Sexual Political Measures N Female Black Latino Income Minority General Politics Liberal Political Ideology 1,294 * * * * * Not Republican 1,319 * * * * * Voted for Gore in 2000 794 * * * * Attitudes That Sometimes Appear to Contradict the Positionality Principle Abortion Attitudes Scale (7 items; a = .90) 884 * * * Same-Sex Relationship Morality 862 * * * Other Sex-Related Attitudes Scale (6 items; a = .66) 1,319 * * * * * Gender Role Attitudes Scale (4 items; a = .66) 892 * * * * * Free Speech Scale (15 items; a = .90) 890 * * Required Prayer in Public Schools 863 * * * * * Physician Assisted Suicide 860 * * * * * Corporal Punishment 883 * * * * Source: 2004 General Social Survey Models include gender, race, class (family earnings), sexuality, education, age, marital status, parental status, region, and rurality. * indicates suppression of liberal response or mediation of conservative response by psychological compensators (p < .05 in Sobel-Goodman tests using sgmediation in Stata 14)

Opiate of the Masses? 49

Table 6: Psychological Compensators, Afterlife Belief, and Religiosity Predicting Liberal Political Ideology Model 1 Model 2 Model 3 Model 4 Psychological Compensators -0.285*** -0.256*** -0.072 -0.057 (0.041) (0.043) (0.057) (0.058) Belief in an Afterlife -0.239* -0.189 (0.107) (0.107) Religiosity -0.297*** -0.286*** (0.057) (0.057) Female 0.361*** 0.371*** 0.376*** 0.384*** (0.079) (0.079) (0.078) (0.078) Black 0.465*** 0.440*** 0.531*** 0.509*** (0.128) (0.128) (0.127) (0.128) Latino 0.269 0.249 0.269 0.254 (0.139) (0.139) (0.137) (0.137) Family Income (in tens-of-thousands of dollars) -0.015 -0.014 -0.013 -0.013 (0.011) (0.011) (0.011) (0.011) LGB 0.787** 0.796** 0.749** 0.757** (0.256) (0.256) (0.253) (0.253) Other Covariates Yes Yes Yes Yes Constant 4.196 4.375 4.056 4.201 N 1,294 1,294 1,294 1,294 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Opiate of the Masses? 50

Table 7: Disadvantaged Social Status and Psychological Compensators Predicting Economic Attitudes Scale % of Total Difference Model 1: Model 2: Mediated by Without With Psychological Compensators Compensators Compensators Female 0.175*** 0.200*** -14% (0.055) (0.056) Black 0.544*** 0.578*** -6% (0.089) (0.090) Latino 0.157 0.180 -15% (0.097) (0.098) Family Income (in tens-of-thousands of dollars) -0.026*** -0.027*** -4% (0.008) (0.008) LGB 0.085 0.109 -29% (0.183) (0.184) Psychological Compensators -0.057* (0.029) Covariates Yes Yes Constant 0.246 0.199 N 1,322 1,322 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. Sobel-Goodman tests used to determine percentage and significance of economic attitudes mediated by psychological compensators. a Negative mediation is suppression (i.e., psychological compensators are suppressing liberal attitudes). * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Online Appendix for “Opiate of the Masses?” 1

ONLINE APPENDIX

Table A1: Disadvantaged Social Status Predicting Psychological Compensators Accounting for Religious Affiliation and Religiosity Model 1: Model 2: Among Among Full Religiously Sample, Affiliated Controlling for Only Religiosity Key Status Measures Female 0.384*** 0.181*** (0.052) (0.037) Black 0.519*** 0.135* (0.083) (0.061) Latino 0.338*** 0.209** (0.092) (0.066) Family Income (in tens-of-thousands of dollars) -0.024*** -0.012* (0.007) (0.005) LGB 0.424* 0.301* (0.187) (0.123)

Religiosity 0.703*** (0.019)

Covariates Yes Yes Constant -0.402 -0.060 N 1,125 1,323 Standard errors in parentheses Source: 2004 General Social Survey Note: Models also include an “other race” category and a sexuality category for no partnering information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Online Appendix for “Opiate of the Masses?” 2

Table A2: Correlation between Psychological Compensators and General Happiness by Social Status Correlation between Psychological Compensators and Happiness N Among Full Sample 0.153*** 1,320 Among Heterosexual White Men with Top 50% Earnings 0.119 215 Among People Marginalized on at Least One Factor 0.187*** 1,105 Source: 2004 General Social Survey Note: Additional analyses considered straight white men in the top earning category instead of the more broadly encompassing top 50% of earnings: the correlation between compensatory benefits from religion and happiness among this more elite group of 59 men was only .10. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Online Appendix for “Opiate of the Masses?” 3

Table A3: Percentage of Gender, Race, Class, and Sexuality Differences in Religiosity Mediated by Psychological Compensators % of Total Difference % of Strength % of % of in Religiosity Factor of Religious Attendance Prayer Mediated by Affiliation Frequency Frequency Compensators Difference Mediated Difference Mediated Difference Mediated Gender (Female vs. Male) 97%*** 155%*** 93%*** 63%*** Race (Black vs. Non-Latino White) 61%*** 65%*** 52%*** 82%*** Ethnicity (Latino vs. Non-Latino White) 106%*** 1378%*** 76%*** 74%*** Class (Family Earnings) 138%** 179%** 121%** 170%** Sexuality (LGB vs. Heterosexual) 294%* 260%* -566%*a 79%* Source: 2004 General Social Survey. Religion Factor N=1,302; Strength of Affiliation N=1,306; Attendance N=1,320; Prayer N=1,316. a The coefficient for sexuality indicates that sexual minorities attend slightly, though not significantly, less frequently before accounting for psychological compensation, which is a suppressor of sexual minorities’ tendency to attend services less frequently. In each case, including psychological compensators in the model moves sexual minorities, and each other group presented here, toward appearing less religious. Note: Sobel-Goodman tests used to determine percentage and significance of religiosity mediated by psychological compensators. Models include gender, race, class (family earnings), sexuality, education, age, marital status, parental status, region, and rurality. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)

Online Appendix for “Opiate of the Masses?” 4

Table A4: Religiosity as a Suppressor of Liberal Politics among Disadvantaged Groups, 2000-2014 GSS Liberal Politics Suppressed by Religiosity Lower Sexual N Female Black Latino Income Minority Same-Sex Marriage 7,763a * * * * Science and Environment Scale (15 items; a = .78) 3,528 * * * * Legislation of Morality Scale (4 items; a = .40b) 6,013 * * * * Right to Civil Disobedience Important 2,309 * * Source: 2000-2014 General Social Survey a Sample size varies by availability of the outcome measure, each of which was only fielded in certain years and questionnaire forms. Missing data were handled with listwise deletion for the additional analyses presented in this table. b The alpha for this scale is low, but I present it instead because this was a GSS-formulated scale and the patterns are the same for the individual measures. Note: Models include gender, race, class (family earnings), sexuality, education, age, marital status, parental status, region, rurality, and a series of binary measures for survey year. * indicates suppression of liberal response or mediation of conservative response by psychological compensators (p < .05 in Sobel-Goodman tests using sgmediation in Stata 14)

Online Appendix for “Opiate of the Masses?” 5

Table A5: Religiosity as a Suppressor of Liberal Economics among Disadvantaged Groups, 2000-2014 GSS Liberal Economics Suppressed by Religiosity Lower Sexual N Female Black Latino Income Minority Consciousness of Class Conflict Scale (5 items; a =.74) 1,006a All Citizens Need an Adequate Standard of Living 2,417 Union Importance Scale (2 items; a = .79) b 1,265 Workers Need Strong Unions 5,144 * * * Confidence in Organized Labor 9,556 Large Income Inequality Not Necessary for Prosperityc 4,489 * * * * Inequality Persists Because It Benefits the Rich 4,413 * * * * Inequality Persists Due to Lack of Solidarity 2,370 Unacceptable for Few to Accumulate Wealth While 1,178

Many Live in Poverty Source: 2000-2014 General Social Survey a Sample size varies by availability of the outcome measure, each of which was only fielded in certain years and questionnaire forms. Missing data were handled with listwise deletion for the additional analyses presented in this table. Note: Models include gender, race, class (family earnings), sexuality, education, age, marital status, parental status, region, rurality, and a series of binary measures for survey year. b The sexuality measure did not overlap with this measure, so sexual minority was not included in these models. c This item and the two following were fielded together in the GSS, but they do not scale together well and are thus presented separately (the alpha for a summative scale of the three items is .22). * indicates suppression of liberal response or mediation of conservative response by psychological compensators (p < .05 in Sobel-Goodman tests using sgmediation in Stata 14)

Online Appendix for “Opiate of the Masses?” 6

Table A6: Disadvantaged Social Status and Religiosity Predicting Liberal Political Ideology in Random-Effects GLS Regression, 2010-2014 GSS Panel Model 1: Model 2: Without With Religiosity Religiosity Female 0.233*** 0.353*** (0.059) (0.057) Black 0.234** 0.413*** (0.089) (0.085) Latino 0.059 0.174 (0.101) (0.096) Family Income (in tens-of-thousands of dollars) -0.004 -0.005 (0.006) (0.006) LGB 0.408** 0.390** (0.135) (0.131) Religiosity -0.410*** (0.030) Covariates Yes Yes Constant 4.368 3.952 N 4,344 4,344 Standard errors in parentheses Source: 2010-2014 General Social Survey Note: Sexuality is measured as self-reported sexual orientation. Models also include an “other race” category and a sexuality category for no orientation information, as well as the following covariates: education, age, marital status, parental status, region, and rurality. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed)