POLITICAL ENGAGEMENT IN AND OF CONGREGATIONS

Despite a long history of ambivalence about the separation of church and state, American congregations have long been considered a storehouse of democracy (Tocqueville Reference Tocqueville1994). The definitions of democracy vary, but if we consider the political activity of citizens as the democratic good of interest, then the evidence is nearly unequivocal. Simply put, affiliation with a congregation improves the chances of a member participating in politics (Hougland and Christenson Reference Hougland and Christenson1983; Leege Reference Leege1988; Peterson Reference Peterson1992; Gilbert Reference Gilbert1993; Verba, Schlozman, and Brady Reference Verba, Lehman Schlozman and Brady1995; Calhoun-Brown Reference Calhoun-Brown1996; Harris Reference Harris1999; Wuthnow Reference Wuthnow, Skocpol and Fiorina1999; Jones-Correa and Leal Reference Jones-Correa and Leal2001; Schwadel Reference Schwadel2005; Djupe and Gilbert Reference Djupe and Gilbert2009). And a growing number of studies employing creative designs have demonstrated that this relationship is causal (Bloom and Arikan Reference Bloom and Arikan2012; Gerber, Gruber, and Hungerman Reference Gerber, Gruber and Hungerman2016). This simple relationship, though, obscures a wealth of mechanisms through which congregations enable political activity. Through clergy, small groups, informal networks, and individual beliefs, congregations can help inform, motivate, and recruit people to take part (e.g., McAdam Reference McAdam1982; Rosenstone and Hansen Reference Rosenstone and Hansen1993; Verba, Schlozman, and Brady Reference Verba, Lehman Schlozman and Brady1995; Schwadel Reference Schwadel2005; Beyerlein and Hipp Reference Beyerlein and Hipp2006).

There is considerable nuance in the literature and little agreement on the predominant mechanisms that link religion to political action. The greatest discrepancies are between the approaches that assert connections between religiosity and efficacy (e.g., Harris Reference Harris1999) and those that emphasize organizational experience (e.g., Leege Reference Leege1988; Tate Reference Tate1991; Verba, Schlozman, and Brady Reference Verba, Lehman Schlozman and Brady1995; Jones-Correa and Leal Reference Jones-Correa and Leal2001; Djupe and Gilbert Reference Djupe and Gilbert2009), though see Calhoun-Brown (Reference Calhoun-Brown1996, Reference Calhoun-Brown, Wolfe and Katznelson2010) for a blended perspective. That is, most of this work considers churches to house processes that affect political participation (e.g., Putnam and Campbell Reference Putnam and Campbell2010) rather than to direct those processes. To be clear, the former argues that churches host religious experiences and a variety of small groups in which individuals can exercise civic skills and develop opinions, which are then transportable to the public square. The latter perspective would argue that churches are intentional in their deployment of political groups and information that may help connect congregants with political opportunities outside the church. There need not be an either/or answer, but it is this latter set of activities that is the focus of this paper.

There is existing evidence that some churches direct political action. Most of the research in this area, however, starts from a different vantage point—the clergy and the church organization rather than the individual congregant. A minority of congregations do engage in political activities, with the highest number pursuing officially non-partisan activities like informing congregants about political opportunities and distributing voter guides, with far fewer engaging in active voter registration and protests (Chaves, Giesel, and Tsitsos Reference Chaves, Geisel, Tsitsos, Wuthnow and Evans2002; Chaves Reference Chaves2004). The amount of activity also varies by religious tradition, with congregations specializing in particular types of activities by religious tradition to an extent (Beyerlein and Chaves Reference Beyerlein and Chaves2003; Smith Reference Smith2016). Still, by these measures, it is clear that Catholic and Black Protestant congregations in the United States are the most involved in the political action (see also Morris Reference Morris1984), with evangelicals the least engaged (except with voter guide distribution—Beyerlein and Chaves Reference Beyerlein and Chaves2003, 237). This pattern also obtains at the level of the individual congregant (Campbell Reference Campbell2004). But it is still not clear why some congregations are more political than others within religious traditions. If such activity affects individual political engagement, it is curious why far from all congregations do it.

From the perspective of clergy political engagement, the distributions of activity look quite different and a clearer story emerges. Referencing patterns across religious traditions, old gaps between theologically liberal and conservative clergy appear to be declining (Guth et al. Reference Guth, Green, Smidt, Kellstedt and Poloma1997) and, from the scattered evidence we have, political activity among clergy is up from past levels (Guth Reference Guth, Green, Guth, Smidt and Kellstedt1996; Djupe and Gilbert Reference Djupe and Gilbert2008). Most of the work on clergy has explored their political engagement relative to distributions of opinion in the congregation and community, which together demonstrate a concern for organizational maintenance.

After Campbell and Pettigrew's (Reference Campbell and Pettigrew1959) pathbreaking study that described clergy deeply constrained by disagreement over civil rights, researchers have continued to focus on the importance of congregational disagreement. “Storms in the churches” during the civil rights movement and beyond suggested clergy engaged in politics at their peril and pain of losing members (Hadden Reference Hadden1969; Quinley Reference Quinley1974; Guth et al. Reference Guth, Green, Smidt, Kellstedt and Poloma1997; Friedland Reference Friedland1998; Djupe, Neiheisel, and Sokhey Reference Djupe, Neiheisel and Sokhey2018; see also McDaniel Reference McDaniel2008). Indeed, churches that were cohesive (and “strong”—Kelley Reference Kelley1972) combined traits of social ties, theological conservatism, and tension with the world—the seeming opposite of those engaged with current politics (Wald, Owen, and Hill Reference Wald, Owen and Hill1990). However, more recent work carved out room for a prophetic role for clergy in their congregations, showing that greater disagreement with the congregation promoted more political speech (Djupe and Gilbert Reference Djupe and Gilbert2003).

Some of the differences between these studies were resolved by turning to richer measures of clergy engagement with politics. That is, once researchers captured the content of their public argumentation, they found that clergy speech does not shrink in the face of disagreement, in part because clergy can rely on other reference groups (Calfano Reference Calfano2009), but especially because their speech tends to be more ideologically balanced (Djupe and Calfano Reference Djupe and Calfano2012). From this perspective, the balanced presentation of issues is more common among mainline Protestant clergy than other clergy because they face more diverse congregations (Olson, Djupe, and Cadge Reference Olson, Djupe, Cadge, Rayside and Wilcox2011). This behavior can be tied to congregational maintenance strategies since pursuing balanced policy speech is simultaneously linked to political division in the congregation and congregational outreach toward new members (Djupe and Calfano Reference Djupe and Calfano2012). Presenting balanced accounts of a policy issue is a way to build more informed, thoughtful citizens (e.g., Price, Capella and Nir Reference Price, Cappella and Nir2002), but is also a way to maintain engagement by people who disagree with each other (Guttmann and Thompson Reference Guttmann and Thompson1996). That is, clergy facing diversity in the pews shift their political presentations in ways to serve member retention and the demands of outreach to grow the membership.

Disagreement in the congregation can be managed from another perspective. Political disagreement is seldom confronted directly in organizational societies. In Olson's (Reference Olson1965) view, members belong to groups for the selective benefits, which are typically tangential to the public purpose of the organization. Friendship and cheap insurance attract and maintain members and not the pursuit of clean air (but see Rothenberg Reference Rothenberg1988). Congregations may work similarly, though the benefit packages obviously vary. If clergy politicking costs members because of a perceived inattention to the needs of the congregation (e.g., Hadden Reference Hadden1969), then it is no surprise that political action is enabled as a byproduct of attention to congregational needs (Djupe and Gilbert Reference Djupe and Gilbert2008). That is, clergy will be more politically engaged when their members are satisfied with the religious and social life of the congregation.Footnote ²

This is not to say that researchers have been inattentive to the role of the community in shaping clergy and congregational political presence. The earliest work was not particularly sanguine about the role of churches in reshaping their communities, finding them (too) well integrated into the economic life of mill towns (Pope Reference Pope1942), a finding echoed later in work on churches in civil rights struggles (Campbell and Pettigrew Reference Campbell and Pettigrew1959; see Friedland Reference Friedland1998 for an overview). While a concern for the relationship of the church to the community lay dormant for decades, more recent work has placed it front and center. Especially in areas left organizationally barren, clergy stepped in to play representative roles, serving as liaisons to local government and even enlisting their congregations as service providers (McAdam Reference McAdam1982; Olson Reference Olson2000; Owens Reference Owens2007; but see McRoberts Reference McRoberts2003). More generally, Djupe and Gilbert (Reference Djupe and Gilbert2003) find that clergy engage in more political activity when their congregations hold minority views in the community, reinforcing their roles as representatives when needed (see also Djupe, Burge, and Calfano Reference Djupe, Burge and Calfano2016).

Altogether, these concerns about how clergy react to congregation and community can stand integration. A growing chorus finds congregations pursuing politics in ways that demonstrate an abiding concern for membership maintenance. Clergy are also mindful about their place in the community in a manner that appears to balance the needs of the congregation with their reputation in the community. One way forward is to draw on a religious economies approach, which is explicitly concerned with how congregations and clergy entrepreneurs develop strategies to attract and retain members in a religious marketplace.

THE RELIGIOUS ECONOMY MODEL

Described by Warner (Reference Warner1993) as one aspect of an emerging “new paradigm” in the sociology of religion (see also Sherkat and Wilson Reference Sherkat and Wilson1995; Jelen and Wilcox Reference Jelen and Wilcox1998), the religious economies model draws upon economic concepts to argue that competition between houses of worship makes religious organizations stronger and increases participation in religious practices (Finke and Stark Reference Finke and Stark1988, Reference Finke and Stark1989, Reference Finke and Stark2005). The core assumption of the religious economies model is that individuals assign a value to religious belief and will maximize the benefits they receive when choosing a particular religious home (Gill Reference Gill2001). The incentives, then, are for religious entrepreneurs (typically clergy) to offer an array of religious models in order to meet a diversity of needs through congregations. A denser array of options in the religious economy should drive up the intensity of services offered to recruit and maintain members (Jelen and Wilcox Reference Jelen and Wilcox1998). Hence, free markets marked by dense pluralism should and do have greater public participation in them (Finke and Stark Reference Finke and Stark2005; McCleary and Barro Reference McCleary and Barro2006).

While certainly not without its critics (e.g., Breault Reference Breault1989; Olson Reference Olson1998; Voas, Crockett, and Olson Reference Voas, Crockett and Olson2002; Norris and Inglehart Reference Norris and Inglehart2004; Hill and Olson Reference Hill and Olson2009), the religious economies perspective has opened up fruitful new avenues of research and bears the promise of connecting institutional and contextual factors with decision-making among individuals. It has been employed, with some success, in explaining why strict churches are often so successful (Kelley Reference Kelley1972; see also Wald, Owen, and Hill Reference Wald, Owen and Hill1990)—a puzzle with which other models struggled to contend (see Iannaccone Reference Iannaconne1988, Reference Iannaconne1992, Reference Iannaconne1994).

Even though the economic approach is appealing in many respects, the evidence in support of canonical statements of the religious economies model is spotty (Chaves and Gorski Reference Chaves and Gorski2001). Early research (Finke and Stark Reference Finke and Stark1989) has been roundly criticized for a series of mistakes that affected the conclusions about the relationship between competition in the religious marketplace and religious practice (Breault Reference Breault1989; Olson Reference Olson1998; Norris and Inglehart Reference Norris and Inglehart2004), while more recent work has gone after the measurement strategies used to gauge whether the model can explain religious adherence across countries and over time (Voas, Crockett, and Olson Reference Voas, Crockett and Olson2002). However, there are stable findings—a bevy of analyses have replicated the relationship that small marketshare congregations enjoy greater religious commitment (Stark and McCann Reference Stark and McCann1993; Jelen and Wilcox Reference Jelen and Wilcox1998; Stark Reference Stark1998; Perl and Olson Reference Perl and Daniel2000; Brewer, Jozefowicz, and Stonebraker Reference Brewer, Jozefowicz and Stonebraker2006; Hill and Olson Reference Hill and Olson2009). Relatedly, religious participation is linked to state deregulation of religion, which is necessary for the free market to function (Fox and Tabory Reference Fox and Tabory2008). On the other hand, analyses pitched at the congregational level have found little to suggest that low market-share congregational clergy “try harder” to attract new adherents by engaging in outreach efforts (Hill and Olson Reference Hill and Olson2009).

If competition drives congregations to alter their behavior, there is no reason those behaviors are limited to religious activities. Indeed, a religious free market marked by competition is argued to be a primary driver of innovative practices in the search for adherents (Finke and Stark Reference Finke and Stark2005). Politics can be one such tool that congregations may use to reach out and distinguish themselves in the marketplace. This is just the argument that a select few political scientists have made with respect to the political orientation of the Catholic Church under authoritarian regimes (see Gill Reference Gill2001). The argument is straightforward—the Catholic Church will support the regime (or remain quiescent) until religious competition from Protestants compels the Church to take more progressive and populist stands against the regime (Gill Reference Gill1998). For instance, the Catholic Church in Mexico was pressured into supporting indigenous causes by the expansion of Protestantism to the poor by U.S. missionaries (Trejo Reference Trejo2009). Catholic priests in Brazil similarly turned to deliver social justice messages from the pulpit, as opposed to lines of argumentation focused on personal morality, when primed to think about competition from other religious “firms” (Smith Reference Smith2016).

While compelling, this research has not yet gelled into a comprehensive theoretical framework applicable outside autocratic regimes (or countries with an autocratic past). First, this work has focused on responses by the Catholic Church to the exclusion of other religious groups (for a new, notable exception see Smith Reference Smith2016). This is, of course, a sensible analytic choice given the dominance of the Catholic Church in many countries with autocratic regimes that might then open up to religious competition. But if the religious economies model has empirical legs, it must apply outside of autocratic regimes to groups at varying levels of dominance. And we know little about how Protestants respond to competitive pressures (e.g., Bellin Reference Bellin2008; though see Smith Reference Smith2016).

Second, to be applicable to developed democracies and to prove the mechanism, the unit of analysis must shift. While previous work in Latin America has considered the Catholic Church as a unitary religious body (e.g., Gill Reference Gill1998, 9), that simplification does not hold water in the United States where there is a strong decentralizing tendency, even among the most hierarchical groups (Hatch Reference Hatch1989; see also Philpott Reference Philpott2007). There are also good theoretical reasons why the unit of analysis should shift—the congregation is the primary unit where competitive pressures are active (Finke and Scheitle Reference Finke, Scheitle, Stolz and Usunier2013), though broader religious groups are not unaware of them. Moreover, the democratic contributions of religion are rarely posited to emanate from aggregated religious bodies (though see Wood and Bloch Reference Wood and Bloch1995), but instead from congregations as discussed above. Therefore, a more efficacious religious economies claim explaining political engagement should be pitched at the congregational level. Lastly, to be broadly applicable, the political ends pursued by congregations should be diverse and not limited to progressive causes. The claim that competition results in exclusively progressive political movement by religious groups is clearly a non-starter (e.g., Gerring, Hoffman, and Zarecki Reference Gerring, Hoffman and Zarecki2018; Toft, Philpott, and Shah Reference Toft, Philpott and Shah2011).

Conceptualized more broadly, we lay hold of the primary mechanism of the religious economies approach—under pressure, churches work to provide services to their members to attract new adherents and stave off defections to other religious firms (see also Jelen and Wilcox Reference Jelen and Wilcox1998). In an effort to maximize benefits and minimize costs, religious consumers may seek out churches that offer the most in the way of services (support groups, daycare, etc.). When options are numerous and religious pluralism high, the religious market is for the buyer and services will be plentiful. There is no reason to limit the nature of these services (within reason), which modern megachurches highlight—there are activities for an eye-widening array of interests, including groups for mechanics, sports of all kinds, and every form of grief and relationship type. Politics has never been out of bounds among most religious groups in America and may be considered an additional good that churches and other houses of worship might provide. Much as is the case with individual clergy, who are permitted to engage in political activities in return for satisfying the religious needs of their congregants (Djupe and Gilbert Reference Djupe and Gilbert2008), opportunities for political activism in the church are likely driven by a desire to provide a wide range of services to current and prospective members. Therefore, politics is likely to be pursued by congregations under competitive pressure as just one among many interests they may cater to.

HYPOTHESES

Our main hypothesis is that congregations under pressure of competition will offer political activities at greater rates. We test this notion through four variables that capture evidence of competition and one (theology) that conditions its effects. First, we expect that greater religious pluralism in an area (in this case, the county) will drive up political activity. It is worth acknowledging that an alternate hypothesis suggests that religious pluralism cuts down on congregational politicking owing to a dearth of like-minded confederates (e.g., other churches that share a denominational affiliation) in the region. Second, greater religious marketshare could either undercut political activity by encouraging lazy monopolistic behavior or could increase political activity as a measure of intra-religious group competition, particularly given that most religious switching occurs within denominations (Sherkat Reference Sherkat2014). Third, religious bodies with a monopoly over the production of spiritual goods might take this power as a license to make political demands of their members (see Stedman Reference Stedman1964).

However, we also believe that competition will drive congregations to offer programs conditional on what they believe is an appropriate response (what might be termed their “role orientation”). In particular, religious conservatives have tended to adhere to the notion that they should render unto Caesar that which is Caesar's; religious liberals have tended to embrace politics as an appropriate mission (e.g., Guth et al. Reference Guth, Green, Smidt, Kellstedt and Poloma1997). Put somewhat differently, theological conservatives often focus on “‘otherworldly rewards’ over an emphasis on the everyday demands of the social gospel” (McAdam Reference McAdam1982, 91). Thus, under competitive pressure, we expect religious liberals and conservatives will diverge in their offerings in a brand-consistent way—low marketshare liberals will be more heavily political while low marketshare religious conservatives will be less political.

This conditional hypothesis explicitly acknowledges the moderating role that theology plays in shaping the responses of congregations (and congregational leaders) to market pressures in the local religious economy. Political theology, as Amy Erica Smith notes (2016), may help determine clergy's stance toward mixing political activity with other facets of the church experience (see also Philpott Reference Philpott2007; Toft, Philpott, and Shah Reference Toft, Philpott and Shah2011). That is, we would expect churches’ organizational reactions to competition to depend, at least in part, on the theological tradition with which they are oriented. Protestant clergy (which are almost universally theologically conservative in the Latin American case), for instance, have been known to respond to market forces not by changing the messages that they relay from the pulpit, but rather by encouraging the sorts of social ties that limit members from seeking out other houses of worship (see Smith Reference Smith2016). Theologically conservative churches, on balance, simply demand more of their members and may, therefore, be able to discourage membership losses by fostering strong social ties among the laity (Wald, Owen, and Hill Reference Wald, Owen and Hill1990). By contrast, theologically liberal churches may have to compensate for the relative lack of close-knit connections among their members by offering a wide array of services.

Lastly, we include two measures of felt competition—congregational behaviors that should respond to competitive pressures (which we provide evidence for later in the paper). Congregations under competition should have to work harder to attract and retain members. Therefore, we expect that congregations offering a (1) greater diversity of non-political activities and (2) more outreach activities should be more likely to engage in political activities. We discuss these measures in more detail below.

An alternative hypothesis suggests that religious organizations are active in the political realm only in situations where their involvement might prove determinative. Much as is the case with individuals, groups (including religious groups) are perhaps most likely to participate in politics when they are explicitly asked to do so (Rosenstone and Hansen Reference Rosenstone and Hansen1993). Organizations that are ostensibly non-political in nature, such as churches, are often asked to venture into the business of politics by campaigns and political parties that are strategic about the business of mobilizing voters. While the Republican Party receives a great deal of the credit for bringing the faithful into the political fold to turn so-called “values voters” out to the polls, most political campaigns make overtures aimed at mobilizing churchgoers, including Barack Obama's campaign in 2012 (Dallas Reference Dallas2014). These efforts by campaigns are likely focused in areas in which the returns are expected to be the greatest—geographic regions in which the candidates are in tight contests. In testing this conjecture we include measures of party competition and the Democratic share of the two-party vote for president at the county level in all of the models. Again, the response to the political environment is likely to be brand consistent—theological conservatives are more likely to engage in a majority Republican area, with theological liberals engaging more in Democratic areas.

Although there has been a great deal of debate regarding the appropriate level of aggregation at which to examine the nexus of religious and political competition (see Jelen and Wilcox Reference Jelen and Wilcox1998), we focus on counties owing to their importance as political entities. As Masket notes, “many local political structures are established at the county level, and it is there that many electoral contests are waged” (Reference Masket2009: 1026). Moreover, if the campaigns are successful at mobilizing the faithful, then the effects of such efforts should be localized to the county level—the unit at which many campaign field operations are organized (see Masket Reference Masket2009).

Another alternative explanation for why only some congregations offer political goods is that clergy are often limited in their ability to delve into the political arena by the people in the pews. We term this the “gathering storm” thesis.Footnote ³ Specifically, this perspective holds that mismatches between the policy preferences of clergy and those to whom they minister are likely to present barriers not just to clergy, but congregational politicking. When congregants are opposed to mixing politics with religion, or are opposed to the political messages coming from the pulpit, political goods are unlikely to be provided by the congregation.

A final alternative hypothesis views congregations as part of organizational fields filled with other service providers and community-engaged organizations. In this social movement perspective, church service provision is not the result of filling representational gaps, but instead results from connections to available community initiatives and mobilization from community partner organizations (e.g., Fernandez and McAdam Reference Fernandez and McAdam1988; Kitts Reference Kitts1999; Baldassarri and Diani Reference Baldassarri and Diani2007).

DATA AND METHODS

To examine the relationship between religious competition and the provision of political goods in churches we principally turn to two datasets that are particularly well-suited to the task: the 1998 wave of the National Congregations Study (NCS) and the 2001 wave of the U.S. Congregational Life Survey. The 1998 NCS collected responses from a key informant from 1,236 congregations snowballed from 1998 General Social Survey (GSS) respondents (see Chaves et al. Reference Chaves, Ellen Konieczny, Beyerlein and Barman1999; Chaves Reference Chaves2004)—a procedure known as hyper-network sampling (McPherson Reference McPherson1982). Although not without its difficulties owing to its reliance on key informants (see Schwadel and Dougherty Reference Schwadel and Dougherty2010), the NCS has an ideal set of questions for this inquiry, gathering a nearly comprehensive set of information about the congregation, its composition, and its community, though it is (unfortunately) light on the information content flowing through churches. Critical to this effort, the NCS collects data on the diversity of organizationally-based services offered to members as well as a few political activities the congregation might engage—chief among them being the sponsorship of a voter registration drive.Footnote ⁴ Though the activity list is long, it is missing clergy initiatives (especially their speech) as well as any informal politicking that takes place. The U.S. Congregational Life Study (USCLS) covers essentially the same ground as the NCS, generated by the same snowball sampling method (but using the 2000 GSS).Footnote ⁵ To both datasets, we attach county-level data on religious membership from the 2000 Religious Congregations and Membership Study (RCMS) and county-level data from the National Center for Charitable Statistics to test the social movement hypothesis.

We create two measures of market pressure from these religious census data. For each denominationally affiliated group in each sample, we attached their county-level marketshare (total group adherents divided by total adherents in the county).Footnote ⁶ We also constructed a Herfindahl-style measure of market concentration which sums the squared proportion of each religious group in the county—a higher value (toward 1) suggests that adherence is concentrated in one or a few groups, and values toward 0 suggest equally-sized group pluralism.Footnote ⁷

Two other measures of “felt competition” are of special note. The first indexes several activities that congregations may use to reach out to potential members. The items are similar across the two studies, asking about explicit new member recruitment, placing paid ads, encouraging members to invite others, conducting a survey in the community, mailing a flyer, follow-ups with visitors, and forming a special membership committee. In the NCS, the mean outreach is 4.4 (out of 7 with a S.D. of 2). In the USCLS there are 14 items with a mean of 7 (S.D. of 3).

The second measure is a count of the number of different group services the congregation sponsors. The USCLS asked whether congregations engaged in 23 different forms of service, including such things as senior citizens assistance, support groups, preschool, emergency relief, sporting activities, music programs, and craft groups. Sample congregations engaged a mean of 7 (S.D. of 4, range from 0 to 18). Two of these, not included in the services variable, were explicitly political activities, including “voter registration or voter education” and “political or social justice activities.” This constitutes one of our dependent variables.

The NCS asked about more group types at a much finer grain. Of the 43 types asked about in 1998 (six of which were political), the mean number ascribed was 5.5 (S.D. of 3.6). It is important to note that this measure is not the same as the raw number of groups meeting in a church (i.e., there could be many sections of a group type). In another part of the survey, informants in the NCS were asked to record the total number of groups meeting in the congregation. The correlation between that figure and our service diversity figure is 0.26 (p < 0.01). Our theory is not that the raw number of groups matter, but that some congregations are consciously generating different group types to attract and maintain members with a wider selection of interests—they are seeking a broader share of the market. We will test both measures below, expecting only service diversity to be significantly linked to political activity, and thus the raw number of groups serves as a placebo test. The six political services asked about in the NCS, which constitutes our other dependent variable, were: groups or activities to discuss politics, to register voters, to lobby elected officials, to participate in a march or demonstration, to distribute voter guides, and to tell members about political opportunities.

To operationalize the social movement alternative hypothesis, we draw on data available from the National Center for Charitable Statistics. The 1998 core data capture public charity filings for the United States, which we aggregate by the county to attach to the NCS and USCLS. Our alternative hypothesis holds that more community initiatives should boost the opportunity for congregations to engage and represent more partners that can help mobilize congregations to act.

RESULTS—DESCRIPTIVE EVIDENCE

First, we examine the distribution of congregations engaging in up to six political activities among the four major religious traditions represented in the 1998 NCS (see Figure A1 in the Appendix). Overall, 56% engaged in at least one activity, though the figures are quite a bit different depending on the religious tradition. Less than majorities of evangelical and mainline Protestant congregations engaged in at least one activity and their distributions look almost identical. More Catholic and Black Protestant congregations were active and their overall activity levels were higher. Just under three-quarters of Catholic parishes reported at least one political activity, with Black Protestant congregations trailing closely with 67%. That pattern is reflected in the average amount as well—among those congregations that engaged in at least one political activity, Catholics (2.47) and Black Protestants (2.36) engaged in at least 0.5 more activities than evangelical (1.84) and mainline Protestants (1.86). While these religious tradition differences are significant, they do not reveal why they are different and clearly leave considerable variance to be explained.

Perhaps the most appropriate place to start the analysis is with a gut check of the religious economies theory. In Figure 1, we present the raw distributions of the diversity of services offered by congregations in the two samples (gray area) along with the amount of congregational political activity across the service-level distribution (black line). Panel A shows the proportion of congregations engaging in voter registration given the diversity of other services they offer in the USCLS data, which is overlaid on the distribution of congregation service-provision. The proportion engaging in voter registration climbs steadily across the range of services offered. The correlation between congregational service diversity and engaging in voter registration is 0.27 (p < 0.01) in the USCLS data (the correlation between outreach activities and voter registration is insignificant (r  =  0.08, p  =  0.12)). Panel B shows the NCS variant, which uses a political activity index rather than simply voter registration (though the voter registration variable itself produces a similar pattern). While the NCS has a higher number of congregations offering none of these services, it otherwise shows the same association between service diversity offered and the average number of political activities pursued. The correlation in these data is a bit smaller at r  =  0.16 (p < 0.01; the correlation between outreach activities and political activities is r  =  0.11, p < 0.01). Thus, there is consistent evidence from these two datasets that the provision of political activities is linked to the diversity of services offered.Footnote ⁸

Figure 1.

The Distribution of Congregational Services and the Relationship of Congregational Political Activity to Service Provision. (a) U.S. Congregational Life Study. (b) National Congregations Study

RESULTS—MODELS

With a plausible connection between the two established, we turn to estimates in a model predicting the provision of political activities by churches. We begin with the National Congregations Study data in Table 1, which shows the results of two models—one of engaging in a voter registration effort (logit—20% pursued voter registration) and another of a 6-item political participation index (negative binomial).Footnote ⁹ The results are largely in accord, but most importantly religious competition variables are implicated in both models. At the top of the table, both the diversity of services offered and the number of outreach activities engaged have positive and significant effects.Footnote ¹⁰ In the voter registration model, a shift from the typical lower end of service diversity (mean minus 1 S.D.) to the typical high end (mean plus 1 S.D.) boosts the probability of voter registration by 0.08; the same shift in outreach activities has the same effect on voter registration (0.08). In the index model, this shift from low to high outreach boosts the number of political activities by 0.65 and by 0.39 from that same shift in service diversity. These measures of felt competition are important determinants of the political activism of congregations.

Table 1.

Logistic Regression Estimates of Voter Registration and Negative Binomial Regression Estimates of a Congregational Political Activity Count Measure

***p < 0.01, **p < 0.05, *p < 0.10.

Their effects are comparable with the size of the other significant effects in the model. In fact, the outreach effect (of 0.65 on average) is the strongest in the model with the activity index as the dependent variable. The effect of being denominationally affiliated boosts activity by 0.48. Being affiliated with a mainline Protestant denomination drops activity by the same margin. A higher than average proportion of whites in the congregation similarly depresses the rate of political activity. Seeking a government permit increases activity by 0.41, while the other significant effects are operating at about half of this rate. A congregation with a higher than average poverty level has 0.26 more political activities, a higher than average congregational income boosts activity by 0.2, a higher than average number of members with a long drive to reach the congregation tends to have more political activity, as does having a lower than average proportion of seniors (by 0.27).

It is essential to note that the effects of the diversity of group services provided are not achieved by a measure of the number of groups active in a congregation. Rather than counting the types of groups, this measure counts the raw number of groups operating in the congregation (we also tested a per capita variant). This measure is not close to significant in these same models (nor is the per capita variant—result not shown). Moreover, the percent of the congregation that is exercising leadership is a significant determinant only in the index model but points consistently toward fewer political activities, not more as the civic voluntarism model might predict. As an aside, including the number of groups meeting in the congregation does not affect the estimate of service diversity either. Thus, congregational engagement in politics is not a function of civic skill exercise or simple associationalism in these data. Instead, it is the orientation of the church toward providing a diversity of services to attract and retain members that is important.

Other contextual measures of competition are significantly related to congregational political activity as well and are relatively consonant in their effects across the two models. Greater religious concentration (the opposite of pluralism) undermines the probability of engaging in voter registration and is just outside statistical significance for the general political participation index (p  =  0.13). This is what a religious economies approach would expect as the trend toward monopolism undermines competition.Footnote ¹¹ Notably, this particular result runs counter to the expectation that religious organizations nested within an environment studded with other churches of the same denomination provide the kinds of resources necessary to engage in political action.

The measure of group marketshare works in conjunction with the theological orientation of the congregation as expected by role orientation theory. The results are displayed in Figure 2 with confidence intervals; the black line represents predictions for theological conservatives, the x-axis shows changes in marketshare (logged), and the y-axis shows the predicted number of political activities. Theological conservatives respond to their marketshare differently than others. While others (liberals and moderates) engage in less political activity as their marketshare increases (from over 2 down to 1), conservatives increase theirs modestly.

Figure 2.

Conditional Probability of Congregational Political Activity as a Function of Theological Conservatism and Marketshare (National Congregations Study—see Table 1). Note: Comparing two sets of confidence intervals constitutes a 90% test following Payton, Greenstone, and Schenker (Reference Payton, Greenstone and Schenker2003)

The pattern among liberals and moderates corresponds to Djupe and Gilbert's (Reference Djupe and Gilbert2003; see also Olson Reference Olson2000) formulation that (mainline) clergy take on a representative (or prophetic) role when their congregations are not well represented locally. But this misses the larger story—the difference is more important. The result displayed in Figure 2 suggests that under market pressure (low marketshare) liberals and conservatives differentiate their offerings in expected ways—conservatives maintain separation from politics while liberals engage the world. When in average (around −2 on the logged scale) to more concentrated situations, their political presences converge to become indistinguishable.

It is important to note what is not significant as well. Congregations, on average, appear not to engage politics strategically. If it were strategic, then the deployment of resources would take place when they are likely to make a difference—when there is actual competition between the two parties in the United States. While party competition has no bearing on the extent to which congregations get involved, the partisan cast of the county does. In Figure A3, we show that theologically conservative congregations offer more political activities in Republican majority areas, while theological liberals have more politically active congregations in majority Democratic areas. Moreover, the social movement hypothesis that an organizational field saturated with public charities finds some supportive evidence here, but only that they are linked to voter registration drives.Footnote ¹² However, that result flips to negative in the USCLS (voter registration) model described next.

We also attempt to replicate, as near as is practicable, the NCS models using the 2001 USCLS. While the model specifications differ since the two surveys did not ask the exact same complement of questions, many key variables are present in both studies. With a pair of logit models, we used these data to examine the determinants of whether a congregation hosted a voter registration drive in the year prior to the survey, and whether the church reported sponsoring other political or social justice-oriented activities (see the models displayed in Table A1 in the Appendix). Both models are rather sparse in terms of significant determinants, which may not be surprising given that the sample is half the size of the NCS, and yet they agree in one key way. Importantly, both models show that the diversity of group services offered is positively and significantly related to hosting a voter registration drive and sponsoring a political activity. A higher than average diversity of service provision augments the likelihood of registering voters by 0.15 and of a political/social justice activity by an even greater 0.20.

While there is little evidence from any model to suggest that the provision of political activities by houses of worship is directly related to competition from other denominations in the surrounding religious marketplace (religious pluralism), it is possible that the impetus to offer more diverse services (political and non-political alike) to their members flows from market forces. A series of simple Pearson correlations (see Table A3 in the Appendix) speaks to this conjecture, showing the relationships between service diversity and outreach activities, and between each of those to marketshare and religious concentration. Collectively, these results provide the proof of concept that service diversity and outreach (what we call “felt competition”) are related to religious competition. Service diversity is significantly and positively correlated with outreach activities in both datasets, consistent with the idea that programmatic churches do more and that congregations under pressure engage in more activities to keep and attract members. Service diversity is inconsistently related to objective measures of religious competition. Higher religious concentration in a county is negatively related to service diversity, though only significantly so in the NCS data, and negatively and significantly to the number of outreach activities pursued. These patterns could reasonably be produced by monopolistic environments discouraging activity. On the other hand, service diversity is positively linked to marketshare, though only significantly so in the USCLS data, while marketshare is negatively related to the number of outreach activities (significantly in both datasets). That is, it is reasonable to expect that churches might expend less energy trying to poach other congregants and more energy to retention in high marketshare environments.

In fact, we would expect that denominations would encourage a policy of not coveting thy neighbor's congregants. This is what we find in a simple model predicting outreach activities using an interaction of being denominationally affiliated and marketshare. Those results show that the outreach activities of denominationally-affiliated congregations decline as their marketshare grows, whereas the outreach of non-denominational congregations rapidly increases (see Figure A4 in the online Appendix). Whether by policy, but more likely by tacit agreement, denominations discourage competitive practices when they are numerically dominant. This would make member retention all the more important, explaining the positive relationship between marketshare and service diversity.