Are Shifts in Same-Sex Marriage Attitudes Associated With Declines in Religious Behavior and Affiliation?

Abstract

The anti-LGBTQ politics of the Religious Right has been implicated as one of the critical forces promoting the rise of disaffiliation from religion. The association seems plain given the rapid rise of the nones among younger cohorts of Americans – a group which also holds the most pro-LGBTQ attitudes. However, little work actually tests the link between shifting attitudes on same-sex marriage and declining religious behavior and affiliation. Drawing on the Portraits of American Life Panel study with waves in 2006 and 2012, we use appropriate measures to document the religious effects of changing views on same-sex marriage. We find that while shifting views did have a negative effect on church attendance and affiliation, these effects were not limited to shifts toward support for same-sex marriage and were not limited to liberals.

Keywords

religion and politics public opinion same-sex marriage panel data

Introduction

In this paper we revisit the theorized connection between attitudes toward same-sex marriage and religious behavior. Numerous scholars have placed changes in the American public’s views toward same-sex marriage at the center of explanations for the rise of the religious “nones” (e.g., Hout & Fischer, 2014; Jones, 2016; Putnam & Campbell, 2010). Direct tests of such a linkage are rare, however, and have often come to divergent conclusions regarding the role that same-sex marriage views play in shaping their religious attachments and behavior (Vargas, 2012; Brenner, 2019). Therefore, we turn to panel data from the Portraits of American Life Study (PALS) to explore how shifting attitudes toward same-sex marriage might track with changes in religious behavior – worship attendance (‘attendance’ for short) and leaving congregations. We find some evidence that shifts in support for “traditional marriage” track with changes in attendance, but the effects are not limited to same-sex marriage supporters or to liberals. Moreover, this relationship is most pronounced among those most likely to encounter conflicts over the issue of same-sex marriage: frequent attenders. These are the opposites of most conclusions to this point.

Existing Literature

The early 1990s saw the release of the band R.E.M’s hit single “Losing My Religion” and also, coincidentally, a sharp uptick in the proportion of the American public who, when queried as to their religious affiliation on a survey, answered “none” (Hout & Fischer, 2002, 2014). Most explanations for the surge in religious nones center on the salience of the Religious Right and how its contentious blend of religious dogmatism and conservative politics effectively turned political liberals off of religion (Baker & Smith, 2009; Campbell et al., 2021; Hout & Fischer, 2002).¹ Since the 1990s, the trend toward ever-greater numbers of religious nones has continued in a secular fashion (but see Djupe & Burge, 2020), and at present over a fifth of the American public claims no particular religious identity at all (Brenner, 2019; Newport, 2017), making the nones the largest religious tradition in the U.S. (Burge, 2019).

Organized religion’s reputation for being intolerant toward LGBTQ Americans is often cited as perhaps the chief reason why young Americans are abandoning religion at such a rapid clip (Hout & Fischer, 2014; Jones, 2016; Putnam & Campbell, 2010; Vargas, 2012; cf. Brenner, 2019; Hout & Fischer, 2002, p. 185). Those who came of age at a time when the most visible representation of religion in the public sphere involved the anti-gay agenda put forth by the Religious Right also happen to exhibit the greatest degree of support for gay rights (Putnam & Campbell, 2010). Thus, for many younger Americans, “religion” may function as a shortcut signaling LGBTQ intolerance.

Direct tests of the specific pathways connecting attitudes surrounding gay rights and irreligion have been scant in spite of the attention that the subject has drawn in recent years. Several prominent exceptions to this general observation are worth detailing. Goren and Chapp find, for instance, that “positions on abortion and gay rights lead people to revise their levels of religious commitment” (2017, 123) more generally (see also Putnam & Campbell, 2010). Their study focused broadly on religious commitment – a composite measure of which church attendance is but one part – and did not examine how changes in attitudes toward same-sex marriage influence religious behavior. Still, the evidence that they put forward helps to establish that positions on culture war issues can help to shape future levels of religious commitment. Complimentary findings to this effect come from Vargas (2012). He shows that, while those who expressed negative evaluations of the Religious Right were no more or less likely to leave the faith than anyone else (although they were more likely to consider leaving), those who were more supportive of same-sex marriage were more apt to call it quits on organized religion. Campbell et al. (2021) similarly find that support for same-sex marriage tracks with a rise in secularism.

Conversely, experimental work by Brenner shows that liberals who answered a series of survey questions about “the civil rights of and discrimination against LGBQ individuals” (2019, 157) before being asked about their religious identity actually claimed to be religious nones at a lower rate than those not primed. Previous work, however, has stressed the importance of salient controversy – policy debates over moral issues that serve as a necessary catalyst for religious change stemming from political forces (Djupe et al., 2018; see also Margolis, 2018, 3). That is, action by the Religious Right backing anti-gay ballot measures drove up the rate of religious nones in the states in the critical 2004–2010 period. With the Supreme Court effectively bringing the fight for marriage equality to a close in 2015 and the Religious Right’s subsequent shift toward religious liberty as its primary mass appeal (e.g., Lewis, 2017), it is plausible that liberal persons of faith would be less likely to associate organized religion in general with opposition to same-sex marriage after Obergefell.

What is more, the available evidence suggests that changes in religious identity are more likely only after shifts in religious behavior, such as drawing down one’s participation in religious services or leaving a house of worship altogether (Djupe, Neiheisel, & Sokhey, 2018). This ordering provides a possible explanation as to why experimental work on religious identity finds no (negative) effect of a prime reminding study participants about their attitudes towards gay rights on religious de-identification post-Obergefell (Brenner, 2019), while support for same-sex marriage predicts falling from the faith in earlier periods (Vargas, 2012).

This conjecture drives us to re-examine the influence of shifts in same-sex marriage attitudes on key aspects of religious behavior: worship attendance and leaving a congregation. If Djupe and colleagues (2018) are correct that changes in religious behavior precede shifts in religious identity (e.g., from some broad category of religious identification to “none”), then it seems reasonable to expect that changes in salient attitudes would be linked to declines in attendance, which would serve as a necessary precursor to leaving a congregation and perhaps leaving religion altogether.

The other critical disagreement in this literature centers on how to conceptualize groups. Many pieces focus attention on the social groupings “liberals and moderates” since they are likely to disagree most with the politics of the religious right (Hout & Fischer, 2002; 2014; Margolis, 2018; Putnam & Campbell, 2010). Hence, “The increase in ‘no religion’ responses was confined to political moderates and liberals...This political part of the increase in ‘nones’ can be viewed as a symbolic statement against the Religious Right” (Hout & Fischer, 2002, 165).

An alternate view is that the role of politics in driving down attendance is premised on actual exposure to political difference in congregations (Djupe et al., 2018). In this read, sorting is organized by dynamics within face-to-face communities and therefore should be occurring to some extent across the political spectrum, which is what they find. The former approach expects most movement to be on the ideological left. The latter approach suggests we should find the greatest decline in attendance among high attenders and those who changed their mind on this issue since they have the most exposure to potential disagreement. Fortunately, our data provide evidence to test both expectations.

Theory

In our examination of the linkages between changes in attitudes toward same-sex marriage and shifts in religious behavior, we draw inspiration from a theoretical framework set forth by Margolis (2018; see also Campbell et al., 2021). In her book, Margolis argues for what she terms a “life-cycle theory of religion and politics” in which the influence of political considerations on religious behavior can be explained with reference to the fact that partisan identities often solidify at precisely the same time when many (younger) Americans are beginning to distance themselves from organized religion. As they age, these individuals are then confronted with a decision as to whether to re-engage with a faith tradition having already internalized a particular political worldview in which, depending on the context in which their partisan attachments were formed, organized religion might be seen as anathema to their political commitments.

The life-cycle theory of religion and politics therefore explains who is most likely to fall from the faith for political reasons (Democrats or liberals who are confronted with the decision to re-enter a faith at a time when religion is seen as something that is more for conservatives or Republicans). At its core, however, we would argue that the life-cycle theory also offers an explanation as to why such shifts occur that is far less mechanistic and less limited in time than it might at first seem.

Sociologists have pointed to a number of explanations for why younger Americans are disproportionately likely to disaffiliate from organized religion. As Margolis notes, for instance, leaving home and developing new peer groups can “often lead people to cut ties with their childhood religious institutions” (2018, 44). Leaving home and making new friends, whether at college or in the context of a place of work, frequently throws individuals into contact with others who are different from themselves. One of the subjects who Campbell, Layman, and Green interviewed for their recent study of secularism was such an individual. He said that it was “in college [where] social views turned me away from religion, pushed me away from religion on things like marriage equality and abortion” (Campbell et al., 2021, 85).

Social psychologists have long argued that contact with members of different social groups can, under the right circumstances, lead to greater levels of tolerance (e.g., Allport, 1954). Indeed, research has found that contact with gays and lesbians can increase support for same-sex marriage, even among those who report being affiliated with a faith tradition (Bramlett, 2012; see also Becker, 2012; Lee & Mutz, 2019). And while contact effects may be blunted by ties to conservative religious communities (Merino, 2013), others have similarly enlisted social norms as a reason why opinion toward same-sex marriage has shifted at a rate that precludes pure generational change as an explanation (Kreitzer et al., 2014).

Taken together, these findings suggest that encounters with those who are different in any number of salient ways – or, by extension, with new arguments or information – can force a re-evaluation of one’s opinion on culture wars issues such as same-sex marriage. In turn, shifts in opinion of this nature may also precipitate a reconsideration of one’s “fit” with cultural institutions (like churches) conditional on the salience of the issue and relevant messaging.

It is possible that this process unfolded largely among younger Americans who came of age during a time when religion was perceived as being something to which conservatives and Republicans disproportionately hew (as the life-cycle theory of religion and politics predicts). But, at root, it is contact with difference that may drive attitudinal change across a host of issues, including same-sex marriage, that may occur at any age. And Americans of all ages were shifting their opinions on same-sex marriage in this period. We expect these changes to track with shifts in religious behavior, particularly when differences in opinion on gay rights and same-sex marriage are encountered within the context of organized religion. It is in this way that we build upon the life-cycle theory of religion and politics to stipulate why an individual – young or old – may shift his or her religious behavior in the context of changing personal attitudes and confrontations with differences in opinion.

Data and Methods

To test the linkage between same-sex marriage attitudes and religious behavior we enlist the Portraits of American Life Study (PALS)² – a two-wave panel that initially interviewed a random sample of 2610 American adults in 2006 and re-interviewed 1314 of them in 2012.³ Although there are a number of panel datasets that might be up to the task, the PALS is explicitly focused on documenting religious trends in the U.S., and therefore contains excellent measures of a host of different variables that are important to our inquiry. It also has the great value of taking place at the center of the cultural battle over same-sex marriage.

The key independent variable asks about same-sex marriage attitudes consistently across the waves in the following fashion: “The only legal marriage should be a marriage between one man and one woman.” For ease of discussion, we will refer to the likert response items as ranging from strongly supporting same-sex-marriage (strongly disagreeing with the statement – low value) to strongly opposing same-sex marriage (strongly agreeing with the statement – high value). Please see the appendix for all other variable coding.

We begin descriptively, documenting that some people did, in fact, change their opinion toward same-sex marriage across the two waves of the panel. We then detail results from a series of models (OLS or logit, where appropriate given the dependent variable) that seek to explain changes in attendance and religious exit (leaving a particular congregation) as a function of shifts in support for same-sex marriage (difference from wave 1 to wave 2 – negative values signal more support for same-sex marriage).

As such, we employ a series of lagged dependent variable models in our analysis – what Finkel (1995) refers to as “conditional change” models. We opted to model wave 2 attendance as a function of the change in attitudes toward same-sex marriage between wave 1 and wave 2 while controlling for wave 1 attendance and a series of controls.⁴ After controlling for lagged values of the dependent variable, our key independent variables can then be thought of as predicting changes in the outcome of interest (church attendance, in our application) over and above what would be expected based solely upon the starting value of the dependent variable (see Eveland & Thomson, 2006). It is worth noting that this is the algebraic equivalent of modeling a change dependent variable (w2-w1). One distinct advantage of modeling the wave 2 variable (rather than a change score) is that we retain natural units rather than changes. The interpretation, however, remains the same. Almost all other variables are measured at time 1, with the exception of partisanship for which we use a change measure (w2-w1). All models include a host of relevant controls, including age, gender, change in party identification (w2-w1), and a series of religious tradition dummies (e.g., Catholic, Jewish, mainline Protestant, etc.), as well as attendance at wave 1. Full model results are available in Supplementary Appendix Tables A2–A6. All variable coding is in the appendix.

Given the discussion of debates in the literature above, we specify the following tests:

• Does worship attendance weaken at wave 2, controlling for wave 1 attendance, for those who became more supportive of same-sex marriage?

• Does the negative effect of growing support for same-sex marriage on wave 2 attendance vary depending on wave 1 attendance? Is it greater for more frequent attenders?

• Is the negative effect on wave 2 attendance from shifts in same-sex marriage attitudes limited to the left side of the ideological spectrum?

• Does the probability of leaving the congregation increase for those who became more supportive of same-sex marriage (and is that conditional on ideology)?

Volatility on Same-Sex Marriage Attitudes

Recent years have seen tremendous shifts in the public’s attitudes toward same-sex marriage. According to Pew Research Center, Americans went from 60 percent opposed to same-sex marriage in 2004 to over 60 percent supportive of same-sex marriage in 2019.⁵ Aggregate changes might still mask a great deal of individual-level opinion stability, however, and it is worth documenting that at least some individual survey respondents updated their views on the issue over time (see also Kreitzer et al., 2014).

As the visual crosstab showing the interaction between attitudes toward same-sex marriage at waves 1 and 2 indicates (see Figure 1), there was an enormous amount of change in the population across this time period and in both directions. The majority of respondents in 2006 were against same-sex marriage (57 percent, which dropped to 54 percent by 2012), which helps to explain two things: why people might shift to less support for same-sex marriage (majorities are coercive), and why any change in attitudes from the right (to more support) was more consequential for the population distribution than among the minority on the left (to less support). Even sizable portions of the extremes showed movement – a third of those strongly against same-sex marriage at wave 1 shifted to a more supportive stance, while 44 percent of those strongly in favor of same-sex marriage in wave 1 shifted to less supportive positions. Those starting wave 1 in the middle three positions demonstrated even more fluidity – roughly two-thirds of each moved in either direction. Notably, just a bit more of this change was toward greater acceptance of same-sex marriage (or, at least, less resistance to it). For instance, a third of those who somewhat disagreed with same-sex marriage in wave 1 agreed with it by wave 2.

Figure 1.

The Considerable Shifts in Same-Sex Marriage Attitudes from Waves 1 to 2. Source: 2006–2012 PALS Panel.

Although such opinion change always incorporates error due to nonattitudes, instrumentation, and ambivalence (Zaller, 1992), these patterns nevertheless provide prima facie evidence that same-sex marriage attitudes were in flux over our period of observation (and for all age groups – see Supplementary Appendix Figure A1). What is more, recent research has called into question the durability of not just attitudes, but also identities that have long been believed to exhibit a great degree of resistance to change (e.g., Clarke & McCutcheon, 2009; Gerber et al., 2010). The movement that is seen in these data is therefore plausible, and even likely, given what is now known about the instability of attitudes and identities.

Multivariate Results

As expected, increases in support for same-sex marriage correspond with lower levels of wave 2 attendance, controlling for wave 1 attendance and the raft of controls discussed above (see Figure 2 below). Those who effected a complete shift to support from their opposition of same-sex marriage showed the lowest levels of predicted wave 2 attendance. Among those who decreased or did not change their support for same-sex marriage, predicted levels of attendance climbed. Though the effect is modest – suggesting that respondents began skipping multiple months by 2012 – the relationship on display in Figure 2 is precisely what we might expect if increasing acceptance of same-sex marriage served to encourage a drawdown in attendance at religious services.⁶

Figure 2.

The Estimated Effect of Shifting Same-Sex Marriage Attitudes on Wave 2 Worship Attendance (Model-based estimates from Supplementary Appendix Table A2). Source: 2006–2012 PALS Panel.

One important implication of these results is where the shifts in attendance are concentrated. Given the variable construction, the greatest changes in attendance are concentrated among those who held the most anti same-sex marriage attitudes at wave 1 and shifted considerably by wave 2. This is not what we would expect given the common narratives that liberals were most likely to leave organized religion, but of course we will test the conditional effect of ideology explicitly (below).

Our second test examines the relationship between shifts in attitudes toward same-sex marriage and religious behavior conditional on wave 1 attendance. We expect that declining attendance will be greater among those attending at high levels in wave 1 because they are more likely to be exposed to messages they might disagree with. Figure 3 plots predicted wave 2 attendance for those who decreased, increased, and did not change their support for same-sex marriage.⁷

Figure 3.

Wave 2 Attendance Lags Among Those Who Increased Their Support for Same Sex Marriage (Model-based estimates from Supplementary Appendix Table A3). Source: 2006–2012 PALS Panel.

At the extreme low end of wave 1 attendance, predicted levels of wave 2 attendance were indistinguishable among these groupings. As wave 1 attendance rose, those who shifted their positions on same-sex marriage began to differentiate their wave 2 attendance. Large shifts in favor of same-sex marriage (solid line) entailed a modest lag in wave 2 attendance as wave 1 attendance grew larger. One implication of these results is that they lend support for the notion that exposure to disagreement is important in driving religious change. The gap in attendance is not consistent across the range, but is somewhat greater at the top, among the most frequent wave 1 attenders.

Although floor effects are another possible explanation, it seems plausible that infrequent attenders who came to support same-sex marriage did not further draw down their level of attendance because they simply did not have many opportunities to notice how their views differed from others in a congregational setting. Greater initial levels of attendance at religious services would have offered more chances for conflict to occur surrounding the issue of same-sex marriage and LGBTQ rights (see Djupe et al., 2018). If it were the case, however, that a general sense that organized religion is hostile to gay rights was driving down religious engagement, attendance levels should matter little and attitudes would be the sole determinant. Instead, we observe the most pronounced gaps in wave 2 attendance at high wave 1 attendance.

This pattern raises the question of just what messages such individuals were hearing in the congregation. It seems likely that the modal attending individual likely encountered messages expressing opposition to same-sex marriage. In fact, wave 2 of the Baylor Religion Study conducted in 2007⁸ asked this very question: “By your best guess, how would your current place of worship feel about each of the following behaviors? Homosexual behavior.” The overwhelming majority (75 percent) sensed that their place of worship would discourage it in some way – 42.4 sense that their house of worship would “forbid” it, 24 percent “strongly discourages” it, and 9.8 percent “somewhat discourages” it. Only 11 percent report that their house of worship would encourage homosexual behavior (0.4 percent) or “isn’t concerned” (10.5 percent). The balance indicated that they “don’t know” (13 percent). This is clearly consistent with the relationships we see in the data from the PALS across this time period.⁹ Moreover, another study of key informants in a nationally representative sample of congregations surveyed for the National Congregations Study conducted in 2006-07 found very similar results: “In the full NCS sample, 6 percent of congregations have a formal statement of welcoming. Three percent of U.S. congregations allow for gay couples to become members and leaders and also have a formal statement of welcoming” (Whitehead, 2017). “Welcoming” is a well-known term for being accepting of LGBTQ members. Though more congregations might have informally allowed openly gay people to become members, the fact that so few congregations had formal policies is highly supportive of the distribution of popular perceptions gathered in the Baylor survey.¹⁰

We next investigate the common notion that this change was concentrated on the political left – that the religious right drove liberals and moderates out of houses of worship with their prejudicial politics. There are multiple ways to look at the evidence,¹¹ however, we believe the most appropriate test is whether changes in same-sex marriage attitudes resulted in the same religious attendance effects across ideological identity categories. We should not forget that sizable numbers of individuals actually changed their position and behavior during this time span. So, group-level summaries obscure a great degree of churn among individuals.

The results of the test shown in Figure 4 reveal that shifts in same-sex marriage attitudes had no effect on wave 2 attendance for those at the ideological poles. Both strong liberals and strong conservatives continued to be heterogeneous in their religious behavior as their same-sex marriage attitudes changed – some increased, some decreased. However, the “somewhat” ideological categories display strong changes – each suggests that change in the direction of an anti-same-sex marriage position is linked to an increase in their religious attendance by 0.1–0.2 points. And the change was greater for those identifying as “somewhat conservative,” though not distinguishably so. The change for those in the “middle of the road” was similarly positive (more anti-same-sex marriage entails more attendance), though just narrowly significant at the 90% level. This pattern may result from several mechanisms, but perhaps the most plausible is that strong ideologues are less likely to attend houses of worship with which they profoundly disagree. Moderates would not place such a premium on political agreement, meaning they are more likely to find diversity of opinion where they worship.

Figure 4.

The Marginal Effects on Wave 2 Attendance of a Unit Change in Same-Sex Marriage Attitudes Conditional on Wave 1 Ideology (Model-based estimates from Supplementary Appendix Table A4). Source: 2006–2012 PALS Panel.

We turn finally to the effect of changes in attitudes toward same-sex marriage on leaving the congregation.¹² A fifth of wave 1 attenders left their congregation across the six-year span, indicating that leaving is common in American religion. Figure 5 shows that for the total sample (upper left panel) – those who shifted their views on same-sex marriage were no more or less likely to effect a change in place of worship between the two waves. Note, however, that those who came to oppose same-sex marriage between waves 1 and 2 exhibited an ever-so-slightly higher propensity to leave their congregation (by a not significant 4 percent).

Figure 5.

Estimated Effects of Shifting Same-Sex Marriage Attitudes on Leaving a Wave 1 Congregation for the Total Sample and Conditional on Ideology (Model-based estimates from Supplementary Appendix Tables A5 and A6). Source: 2006–2012 PALS Panel.

The remainder of Figure 5 reports the results of an interaction between same-sex marriage attitude shifts and wave 1 ideology (collapsed to three categories).¹³ Those results show that shifts in same-sex marriage attitudes worked in ways counter to previous reports. Wave 1 liberals and moderates who shifted toward more anti-same sex marriage positions were more likely to leave their congregations (Supplementary Appendix Figure 3 shows the distributions of attitude shifts by ideology to indicate that these people actually exist). Conservatives who shifted toward more supportive same-sex marriage attitudes were more likely to leave their congregations as well. These results are consistent with our argument that difference experienced in the congregation drives decisions about exit. Conservatives, for instance, were likely in more conservative congregations in wave 1, so when some shifted to take more liberal positions on same-sex marriage, they found disagreement around them, which increased the probability of congregational exit.

Overall, liberals show a higher rate of leaving congregations than conservatives, though given the inclusion of the attitude in the model, this intercept shift is due to factors outside of the same-sex marriage context. There are many explanations; one of which is perhaps that liberals and moderates are considerably more likely to have moved in this time period (see Supplementary Appendix Figure A4 in the Appendix).¹⁴

Discussion

The estimates from panel data are not causal, but we can gain confidence in the relationship between religious behavior and same-sex marriage attitudes by eliminating some other possible explanations. One possibility is that other attitudes were changing at the same time, so our findings are actually the result of some other attitude shift or a general ideological drift. There were several other attitudinal measures in the PALS data and we examined two in the same manner as the model that produced Figure 2 (see Supplemenatry Appendix Table A2). One presented the statement, “Immigrants coming into the US are taking too many jobs away from other American citizens.” The other read, “It’s OK to have a country where the races are basically separate from one another, as long as they have equal opportunity.” Over 60 percent of the sample shifted their opinion on immigration, while just under 60 percent shifted their opinion on racial separation. In both cases, the distributions were symmetrical as about 30 percent shifted in more liberal and more conservative directions. Shifts in these two variables across time are only weakly correlated with changes in same-sex marriage attitudes and each other (all r < .21), so there is a good evidence that most of the movement in these variables is independent.

The results shown in Supplementary Appendix Figure A5 (in the Appendix) show no systematic movement in worship attendance at wave 2 given movement on these attitudes. In both cases, the estimate is effectively a flat line, showing no difference with the sample mean attendance at any level of attitude change. Again, this evidence is not definitive, but does add confidence to the argument that there is something special about changes in same-sex marriage attitudes across this time period.

Conclusion

A number of researchers have speculated that organized religion’s association with opposition to same-sex marriage has turned away many from the faith who disagree with such a stance. Surprisingly, then, only a couple of studies have emerged in recent years that have sought to test the nexus between support for gay rights and declining rates of attendance and religious adherence. They have arrived at divergent conclusions (Brenner, 2019; Djupe et al., 2018; Goren & Chapp, 2017; Vargas, 2012). Previous studies, however, did not make claims about how changes in support for same-sex marriage correlate with changes in religious behavior. Using the Portraits of American Life Study – a two-wave panel survey, both waves of which were administered prior to the Court’s decision in Obergefell – we conducted what we believe is the most faithful test to date of the thesis that changing same-sex marriage attitudes have been shaping Americans’ commitment to religious institutions.

We find that those who came to support same-sex marriage between the two waves of the panel study exhibited lower levels of attendance at wave 2, controlling for attendance at wave 1. As such, our findings lend some support for the popular view that changing attitudes toward same-sex marriage and LGBTQ rights more generally are encouraging some Americans to re-evaluate their fit with organized religion.

As is often the case when dealing with complex social processes, the overall lesson that our analysis imparts is more nuanced. Attitudes were changing rapidly across this time period in all directions, introducing conflict where it had likely not been before. From this perspective, it is not surprising that leaving the congregation is due to mismatches – leaving is more common among liberals who shifted to take less supportive positions on same-sex marriage and among conservatives who shifted to a more supportive stance. But this is also a very different story than liberals and moderates taking symbolic stands against the Religious Right.

Generally, individuals either gravitate toward religious communities that reflect their own social values or, more likely, are so completely shaped by the church context that they come to adopt the attitudes shared by those around them via a form of social contagion (e.g., Wald et al., 1988). But these selection and socialization processes are not nearly perfect. There are times when political differences are imposed from the outside and are forced to the fore in a religious community. Gay rights is one excellent, but not isolated example, which featured outside pressure from politics as well as from religion – denominations addressing LGBTQ policies had the same effect. To counter these pressures, houses of worship often provide a mix of benefits to adherents that can help parishioners to brush away disagreement or otherwise minimize the importance of such differences (see Djupe & Neiheisel, 2019; Djupe, Neiheisel, & Sokhey, 2018). However, frequent contact with divergent viewpoints may force a reconsideration of one’s continued involvement with a particular congregation. Importantly, these processes do not vary much by age, so this religious reconsideration does not appear to be generational (see Supplementary Appendix Figure A6).

Paradoxically, then, high levels of attendance may very well be a necessary precondition for changing attitudes about same-sex marriage to influence religious behavior. Frequent attenders simply have more opportunities to encounter views on same-sex marriage, or anything else, that depart from their own. And while a more complete test would include direct measures of the kinds of messages that individual churchgoers in the dataset received from the pulpit and from other parishioners concerning issues of gay rights, we argue that the pattern of results presented here is strongly suggestive of a more localized process than simple versions of the thesis that “conservatives drove liberals out of religion” otherwise posit. That is to say that the available evidence is by and large inconsistent with a general process dividing liberals and conservatives. Instead, local exposure to conflict over the issue of same-sex marriage drove down attendance among those who came to support and oppose same-sex marriage. Those people were located on the left and the right.

Supplemental Material

Supplemental Material - Are Shifts in Same-Sex Marriage Attitudes Associated With Declines in Religious Behavior and Affiliation?

Supplemetary Material for Are Shifts in Same-Sex Marriage Attitudes Associated With Declines in Religious Behavior and Affiliation? by Paul A. Djupe and Jacob Neiheisel in American Politics Research

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Paul A. Djupe

Jacob Neiheisel

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Allport

G. W.

(1954). The nature of prejudice. Addison-Wesley

Baker

J. O.

Smith

(2009). None too simple: Examining issues of religious nonbelief and nonbelonging in the United States. Journal for the Scientific Study of Religion, 48(4), 719–733. https://doi.org/10.1111/j.1468-5906.2009.01475.x

Becker

A. B.

(2012). Determinants of public support for same-sex marriage: Generational cohorts, social contact, and shifting attitudes. International Journal of Public Opinion Research, 24(4), 524–533. https://doi.org/10.1093/ijpor/eds002

Bramlett

B. H.

(2012). The cross-pressure of religion and contact with gays and lesbians, and their impact on same-sex marriage opinion. Politics & Policy, 40(1), 13–42. https://doi.org/10.1111/j.1747-1346.2011.00337.x

Brenner

P. S.

(2019). Sexuality, political polarization, and survey reports of religious nonaffiliation. Politics and Religion, 12(1), 153–170. https://doi.org/10.1017/s1755048318000433

Burge

R. P.

(2019). Growth and decline in American religion over the last decade. Religion in Public. https://religioninpublic.blog/2019/07/09/growth-and-decline-in-american-religion-over-the-last-decade/

Campbell

D. E.

Layman

G. C.

Green

J. C.

(2021). Secular surge: A new fault line in American politics. Cambridge University Press.

Clarke

H. D.

McCutcheon

A. L.

(2009). The dynamics of party identification reconsidered. British Journal of Political Science, 73(4), 704–728. https://doi.org/10.1093/poq/nfp051

Djupe

P. A.

Burge

R. P.

(2020). The decline of religion may be slowing. Religion in Public. https://religioninpublic.blog/2020/02/10/the-decline-of-religion-may-be-slowing/

10.

Djupe

P. A.

Neiheisel

J. R.

(2019). Political mobilization in American congregations: A religious economies perspective. Politics and Religion, 12(1), 123–152. https://doi.org/10.1017/s1755048318000512

11.

Djupe

P. A.

Neiheisel

J. R.

Conger

K. H.

(2018). Are the politics of the christian right linked to state rates of the nonreligious? The importance of salient controversy. Political Research Quarterly, 71(4), 910–922. https://doi.org/10.1177/1065912918771526

12.

Djupe

P. A.

Neiheisel

J. R.

Sokhey

A. E.

(2018). Reconsidering the role of politics in leaving congregations: The importance of affiliation. American Journal of Political Science, 62(1), 161–175. https://doi.org/10.1111/ajps.12308

13.

Eveland

W. P.

Thomson

(2006). Is it talking, thinking, or both? A lagged dependent variable model of discussion effects on political knowledge. Journal of Communication, 56(3), 523–542. https://doi.org/10.1111/j.1460-2466.2006.00299.x

14.

Finkel

S. E.

(1995). Causal analysis with panel data (sage university paper series on quantitative applications in the social sciences. Sage

15.

Gerber

A. S.

Huber

G. A.

Washington

(2010). Party affiliation, partisanship, and political beliefs: A field experiment. American Political Science Review, 104(4), 720–744. https://doi.org/10.1017/s0003055410000407

16.

Goren

Chapp

(2017). Moral power: How public opinion on culture war issues shapes partisan predispositions and religious orientations. American Political Science Review, 111(1), 110–128. https://doi.org/10.1017/s0003055416000435

17.

Hout

Fischer

C. S.

(2002). Why more Americans have no religious preference: Politics and generations. American Sociological Review, 67(2), 165–190. https://doi.org/10.2307/3088891

18.

Hout

Fischer

C. S.

(2014). Explaining why more Americans have no religious preference: Political backlash and generational succession, 1987-2012. Sociological Science, 1(24), 423–447. https://doi.org/10.15195/v1.a24

19.

Jones

R. P.

(2016). The end of white Christian America. Simon and Schuster

20.

Kreitzer

R. J.

Hamilton

A. J.

Tolbert

C. J.

(2014). Does policy adoption change opinions on minority rights? The effects of legalizing same-sex marriage. Political Research Quarterly, 67(4), 795–808. https://doi.org/10.1177/1065912914540483

21.

Lee

Mutz

D. C.

(2019). Changing attitudes toward same-sex marriage: A three-wave panel study. Political Behavior, 41(3), 701–722. https://doi.org/10.1007/s11109-018-9463-7

22.

Lewis

A. R.

(2017). The rights turn in conservative Christian politics: How abortion transformed the culture wars. Cambridge University Press. https://doi.org/10.1017/9781108278171

23.

Margolis

M. F.

(2018). From politics to the pews: How partisanship and the political environment shape religious identity. University of Chicago Press. https://doi.org/10.7208/chicago/9780226555812.001.0001

24.

Merino

S. M.

(2013). Contact with gays and lesbians and same-sex marriage support: The moderating role of social context. Social Science Research, 42(4), 1156–1166. https://doi.org/10.1016/j.ssresearch.2013.02.004

25.

Newport

(2017). 2017 Update on Americans and religion. Gallup. https://news.gallup.com/poll/224642/2017-update-americans-religion.aspx

26.

Putnam

R. D.

Campbell

D. E.

(2010). American grace: How religion divides and unites us. Simon and Schuster

27.

Vargas

(2012). Retrospective accounts of religious disaffiliation in the United States: Stressors, skepticism, and political factors. Sociology of Religion, 73(2), 200–223. https://doi.org/10.1093/socrel/srr044

28.

Wald

K. D.

Owen

D. E.

Hill

S. S.

(1988). Churches as political communities. American Political Science Review, 82(2), 531–548. https://doi.org/10.2307/1957399

29.

Whitehead

A. L.

(2017). Institutionalized norms, practical organizational activity, and loose coupling: Inclusive congregations’ responses to homosexuality. Journal for the Scientific Study of Religion, 56(4), 820–835. https://doi.org/10.1111/jssr.12484

30.

Zaller

J. R.

(1992). The nature and origins of mass opinion. Cambridge University Press. https://doi.org/10.1017/cbo9780511818691

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.35 MB