Social Science Research 45 (2014) 1–17

Contents lists available at ScienceDirect

Social Science Research

journal homepage: www.elsevier .com/locate /ssresearch

Structural and cultural sources of community in Americancongregations

0049-089X/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.ssresearch.2013.12.010

⇑ Corresponding author. Address: Department of Sociology, Louisiana State University, 126 Stubbs Hall, Baton Rouge, LA 70803, United States(225) 578 5102.

E-mail address: sstroope@lsu.edu (S. Stroope).

Samuel Stroope a,⇑, Joseph O. Baker b

a Department of Sociology, Louisiana State University, United Statesb Department of Sociology and Anthropology, East Tennessee State University, United States

a r t i c l e i n f o

Article history:Received 28 January 2013Revised 7 December 2013Accepted 22 December 2013Available online 31 December 2013

Keywords:CommunityBelongingSocial networksSocial embeddednessCongregationsReligion

a b s t r a c t

Religious institutions are among the deepest reservoirs of social belonging in America, butwhat determines whether belonging is cultivated in these institutions? Previous researchshows that individuals’ social network composition is a primary predictor of feelings ofbelonging. However, less is known about how group characteristics condition the influenceof social networks on belonging. We use data from the 2001 U.S. Congregational Life Surveyand multilevel modeling to examine how organizational characteristics such as group size,in-group network density, and aggregate ideological uniformity moderate the effects ofindividual social networks on sense of belonging. Results indicate that both structural (net-work density, church size) and cultural (ideology) characteristics of groups significantlycondition the effects of individual social networks on belonging. Smaller group size, net-work density, and ideological unity cultivate contexts that amplify the relationshipbetween personal networks and belonging.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Few needs are as fundamental to social life as feelings of belonging in face-to-face groups. Scholars have long studiedcommunity with an interest in the sources and maintenance of belonging, including the role of religion (e.g. Tönnies[1935] 1957). Durkheim ([1912] 1995) outlined the connections between community belonging and religion in explicitdetail in Les Formes Élémentaires. More recently, Putnam’s (2000) influential analyses of ‘‘the collapse and revival of Americancommunity’’ emphasized the role that religion plays in community efficacy. Recent empirical research also indicates thatfactors such as traditional religious culture and shared moral order promote community feelings (Davis and Robinson,1996; Ryle and Robinson, 2006; Vaisey, 2007).

The kind of community-building cultivated in religious groups is important not just for its emotional qualities; com-munal bonding is associated with a variety of social consequences. At the ecological level, strong belonging in religiouscongregations may be ‘‘too much of a good thing,’’ linked to outcomes within the broader local community such as ele-vated crime rates (Beyerlein and Hipp, 2006), higher mortality rates (Blanchard et al., 2008), reduced economic develop-ment (Mencken et al., 2006; Tolbert et al., 2002, 1998), and social isolation (Emerson and Smith, 2000; Schwadel, 2005).

. Fax: +1

2 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

On the other hand, researchers have found that individuals’ well-being and health are positively influenced by socialintegration in congregations (Ellison and George, 1994; Lim and Putnam, 2010) and feelings of belonging in congregations(Idler et al., 2009; Koenig et al., 2001; Krause and Wulff, 2005). Congregational belonging can potentially have both neg-ative and positive effects. Unraveling the social processes that shape religious participants’ sense of belonging remains animportant puzzle.

Recent scholarship consolidates theoretical explanations of community feeling that are suggestive for the study ofbelonging in religious congregations. In keeping with structuration theory (Giddens, 1984; Sewell, 2005), community-build-ing processes are seen in two primary categories, structure and culture (Smith, 2003a). Prominent structural factors includecharacteristics of social ties, distribution of power, form of hierarchy, and the way objects and spaces are organized (seeMartin, 2009). A general statement of the structural sources of belonging is the idea that community feelings flow from socialencounters and networked connections among people. Empirical research on congregational belonging is consistent withthis perspective, showing that indicators of structural integration such as congregation-based social relationships andservice attendance are associated with stronger individual sense of belonging (Krause and Cairney, 2009). As confirmationsof the effects of network embeddedness accumulate, new questions about the structural sources of belonging arise. Of par-ticular interest is understanding under what conditions network embeddedness facilitates feelings of belonging more or lesseffectively.

Structural explanations of community-feeling, however, may not fully account for culture (Mische, 2010), anotherproposed determinant of belonging feelings. Cultural explanations of belonging are consistent with Etzioni’s (2001) argu-ment that the experience of community does not simply arise out of social structural relations, because there are ‘‘moralunderpinnings of trust and social capital’’ embedded in culture (Vaisey, 2007, p. 865). Cultural approaches argue that sharedbeliefs are necessary to undergird belonging because it is collective cultural schema that help actors interpret social andstructural relations as meaningful (Smith et al., 1998).

Scholars of religion argue that a primary function of religious congregations is cultivating ‘‘a place of community andbelonging’’ (Ammerman, 2009, p. 572). Although extant theory and research outlines the connections between structuraland cultural processes of community-feeling, as well as individual-level sources of congregational belonging, much remainsunknown about links between group and individual levels of analysis and feelings of belonging. Using data on 78,895 wor-shippers in 399 congregations, we assess how organizational structure and culture interactively influence individuals’ senseof belonging. We draw on and contribute to sociological literatures on community, religion, organizations, and social capitalby explicitly testing whether and how group characteristics affect belonging and condition the influence of social networkson feelings of belonging.

2. Background

2.1. Structure and belonging

Clearly, ‘‘strong social ties, relatedness, and a sense of belonging’’ are often strongly connected (Kanter, 1972, p. 70).1

However, it is important not to equate a sense of belonging with the kinds of social ties from which belonging arises (Paxtonand Moody, 2003; Moody and White, 2003). These elements are distinct; relational ties and community feelings are notsynonymous. One reason why the presence of social relations should be differentiated from feelings of belonging is that it ispossible to have relationships with people without also feeling belonging among the relations. These constructs are distinctand partitioning the two is necessary for examining the relationship between them, and presently for examining the relation-ship between congregation-based social networks and sense of belonging.

2.1.1. Social network embeddednessThe literature on structural relations and ‘‘social capital’’ features a prominent split between studies focusing on the

effects of ‘‘network closure’’ and ‘‘structural holes’’ (see Burt, 2005). Closure refers to the degree to which an individual’ssocial network is ‘‘dense,’’ that is, that each node in a network is connected to the others. Conversely, structural holesexamine brokerage across otherwise unconnected social networks. Structural holes are vital for business productivity be-cause they facilitate innovation, change, and flexibility (Burt and Ronchi, 2007), while network closure facilitates trust andstability (Burt, 2001a). As is often the case, much of the debate over the relative influence of these dimensions of socialnetworks hinges on two interrelated issues: (1) what researchers mean when employing the concept of social capital; and(2) what outcome is being examined. Indeed, extensive reviews of the literature on social capital highlight a multiplicity ofuses for the concept and a corresponding divergence of findings depending on the topic of examination (see Portes, 1998,2000).

1 Although advances in communication technology have challenged the need for physical co-presence, teleconferencing in for holiday dinners has yet tobecome common. Similarly, while beaming a preacher’s sermon into a satellite worship is on the rise in some churches, this form of remote pastoral contact isunlikely to benefit participants’ sense of belonging as much as a sermon delivered in person. This is one of the reasons that multi-site megachurches attempt toaddress this issue by hiring ‘‘site pastors’’ and aggressively integrating members into smaller cell groups, which enhance belonging, participation, and financialgiving (Dougherty and Whitehead, 2010).

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 3

Because our outcome of interest is feelings of belonging, or what has been termed ‘‘collective’’ social capital (Brunie,2009), we focus on network closure, particularly James S. Coleman’s neo-Simmelian (esp. 1902, 1905) theorizing aboutthe influence of network closure and norms on individuals and social groups.2 In the twilight of Coleman’s career(1988b, 1990a,b,c), he became a leading advocate for the concept of social capital, and early returns from the poll of scholarlycitations indicate that social capital has become his most influential intellectual contribution (Marsden, 2005). Notably, Cole-man does not give social capital the generally positive connotation it would attain in later applications (Field, 2008). Rather,Coleman defined social capital as the action potentiated by relational arrangements, absent a normative endorsement: ‘‘A gi-ven form of social capital that is valuable in facilitating certain actions may be useless or even harmful for others’’ (1988b, p.S98). With regard to network arrangement, he paid particular attention to the role of network closure for facilitating socialtrust and the emergence and enforcement of social norms (Coleman, 1988a,b, pp. S102–S108, 1990a, esp. chaps. 10–12,1990b,c). We use this work to generate hypotheses about the interactive relations between individual network embedded-ness and group-level attributes.

In empirical work in the sociology of religion, network embeddedness predicts a variety of outcomes. Envelopmentin congregation-based social ties bolsters religiosity (Stroope, 2012), recruits and retains converts (Stark and Bain-bridge, 1980), and provides social control (Smith, 2003b,c). Scholars have also shown that congregation-based socialrelations are important for shoring up emotional support (Ellison and George, 1994), giving life a sense of religiousmeaning (Krause, 2008), and influencing personal well-being and health (Krause, 2006; Lim and Putnam, 2010). It re-mains unclear, however, the extent to which embeddedness in congregation-based friendship networks influences con-gregants toward heightened belonging feelings, although such a social involvement-belonging connection is oftenimplicitly made by tying religious participation to emotional rewards of solidarity fostered by network ties (Collins,1997; Stark and Bainbridge, 1980). Although this specific relationship has not been empirically studied, extant findingsare suggestive. Recent work shows a positive link between individual-level emotionally supportive social ties and senseof belonging at church (Krause and Cairney, 2009; Krause and Wulff, 2005). Theory and empirical evidence lead to ourfirst hypothesis:

Hypothesis 1. Individual social embeddedness will be positively related to feelings of group belonging.

2.1.2. Congregational network densityIn addition to the influence of individuals’ social networks, structural characteristics at the organizational level may di-

rectly influence belonging feelings and condition the degree to which person-level social embeddedness impacts feelings ofbelonging. We consider two such factors; the first is the density of social networks within the congregation. Just as individ-uals can be more or less embedded in a network, so too can an individual’s network be more or less integrated into a larger,densely-connected group of networks (Paxton and Moody, 2003, p. 36).

When a person participates in a densely-networked congregation, s/he should be more likely to have a sense ofbelonging. A primary mechanism explaining this pattern is the increased monitoring and ‘‘policing’’ of norms thataccompanies densely connected groups, which forces weak, ‘‘free-riding’’ members out of the group (Iannaccone,1992, 1994). Dense network structures correlate strongly with strictness levels in religious groups (Iannaccone,1988). Further, empirical tests demonstrate that strictness also correlates strongly with positive collective emotionalityexpressed through intense religious experiences during worship (Baker, 2010), a factor that should be closely related toa sense of group belonging.

The idea that densely networked groups are better able to maintain social control also underlies Coleman’s (1988a,b)theorizing about the positive benefits of social capital for schoolchildren and young adults. Concerning religious groups,the extent to which people’s social networks are embedded in a congregation, and networks within the group are inter-linked, people may come to see their own networks as extensions of the group, and vice versa. Hence, the degree towhich one’s personal network is connected to an organization with high network density may influence one’s view ofthe larger organization as a place of strong community and belonging. The above considerations lead to the followinghypotheses:

Hypothesis 2. Greater organizational network density will be positively related to feelings of group belonging.

Hypothesis 3. The positive relationship between individual social embeddedness and feelings of group belonging will beamplified by greater organizational network density.

2 Simmel’s work on group structure and norms mirrors Coleman’s emphasis on the structural properties of groups and central preoccupation with normcreation and enforcement as the meso-level link between individuals and groups. Stinchcombe’s (1992) critical review of the Simmelian themes in Coleman’scentral theoretical treatise is insightful because he effectively makes connections in the text that have been obscured by others (cf. Coleman, 1992), while alsosimultaneously highlighting both the useful and myopic dimensions of formal rational choice theory. Ultimately, we share Stinchcombe’s (befitting) view ofrational choice theory, which is pragmatic and utilitarian—use it where it works, omit it where it does not.

4 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

2.1.3. Congregational sizeOrganizational size is another important aspect of group-level structure, such that individuals’ sense of belonging may

decline as the scale of operations in a congregation increases. Sociologists have shown that in larger organizations the man-ner in which business is conducted and how members perceive and interact with the organization are altered substantially.With increased organizational size, prescribed ways of interacting displace informal, ad-hoc, and more relational ways ofaccomplishing tasks. In larger structures, specific tasks are divided and assigned to units, each contributing a part to theoverall goal (Blau, 1970; Knoke, 1981). Sociologists of religion suggest that the scale of a religious congregation impactsthe life of a group and the way congregants experience belonging (Finke, 1994, p. 8; Stark and Finke, 2000), with empiricalevidence documenting these effects (Dougherty and Whitehead, 2010). Even the anticipation of supportive interaction be-comes less likely in congregations surpassing 2500 members (Ellison et al., 2009). Rather than gather with a small number ofattendees who are more easily known, members of big congregations are more likely to become anonymous in a sea ofnameless faces. With so many unknown others in the group, members can begin to feel distanced from the congregation(Schaller, 1982). Personal interaction with leaders, often a key point of congregational identification for attendees, also be-comes less feasible in larger churches.

Returning to Coleman’s theory of network closure and norm enforcement, the ability of individuals to interact frequentlywith other members of the group, and therefore to police and enforce group norms, declines precipitously as group size in-creases (Coleman, 1986). Consequently, feelings of solidarity may become diluted by the presence of spotty attendees orother free riders who become more prevalent in larger congregations (Iannaccone, 1992, 1994). Not only free riders, but peo-ple with physical and emotional needs are less likely to be detected, supported, and reminded that they belong in large con-gregations (Finke, 1994, pp. 7–9). While personal friendship network embeddedness and overall congregational networkdensity help lessen the negative effects of large church size (Dougherty and Whitehead, 2010; Stark and Finke, 2000), theemotional liabilities of large congregations go beyond the presence or absence of close congregant network ties.

Just as the context of density may shape how relational ties are experienced, so too the context of size may moderatethe effects of personal networks (Paxton and Moody, 2003). Specifically, increased congregation size can directly dimin-ish individuals’ sense of belonging, but may also reduce the positive effects of personal embeddedness on feelings ofbelonging. This is because congregation size taps other aspects of social disconnection such as weakness of ties or infre-quency of extended interaction. Even if most of a person’s friends go to his or her congregation, s/he may encounter thesefriends less frequently in a larger organization both because of the overall number of congregants and because thesefriends’ attendance may be more irregular where social sanctions are fewer (Moody and White, 2003). Additionally,the relationship between friends and sense of belonging may be stronger in smaller congregations because the definitionof friendship network and congregation could have considerable overlap in the mind of a member of a smallcongregation.3

Hypothesis 4. The positive relationship between individual social embeddedness and feelings of group belonging will beattenuated in larger organizations.

2.2. Culture and belonging

Community sociology often privileges social structural explanations (e.g. Field, 2008; Delanty, 2009). Community issaid to arise principally out of social connections and the organization of places and spaces. This perspective views culturalschemas, beliefs, and feelings as determined by material sources such as physical environments and social networks(Emirbayer and Goodwin, 1994). In addition to structure, culture too plays an important role, and recent research in cul-tural sociology highlights the influence of salient cultural schemas on social structural outcomes (Lizardo, 2006; Vaiseyand Lizardo, 2010). Rather than being viewed as competing perspectives, structure and culture are better understoodas interactive factors.

In keeping with structuration theory’s vision of the duality of social processes (Giddens, 1984; Sewell, 2005), the inter-relationship between structure and culture is central in the study of religion (Smith, 2003a, p. 21).4 Scholars have used thestructure-culture conceptualization to explain a variety of religious outcomes including conversion (Smilde, 2005), denom-inational switching (Sherkat, 1998), biblical literalism (Hoffmann and Bartkowski, 2008; Stroope, 2011b), and aggregatecommunity-feeling in religious and secular communes (Vaisey, 2007). Stated directly, cultural schemas condition howstructures are enacted (Martin, 2002), while the ‘‘shape’’ of structures also influences the content of cultural schemas(Martin, 2011).

Consider a congregation with divided views on salient religious issues, such as the interpretation of sacred texts. Thoughmembers may feel belonging through their ties to close friends in the group (structure), the ideological fractures of the con-gregation would not escape them (culture), potentially diminishing the degree to which they identify the closeness of theirinterpersonal relations with the larger, ideologically disjointed congregation. In some cases, ideological disunity leads to

3 We are grateful to an anonymous reviewer for this insight.4 Ironically, but perhaps not surprisingly, the power of beliefs to influence is largely assumed rather than explicitly or critically examined in the sociology of

religion (Chaves, 2010).

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 5

schisms (see Ammerman, 1995; Becker, 1999). These processes feature prominently in church-sect theory, both classical(Niebuhr, 1929; Weber [1930] 2002, pp. 53–101) and contemporary (Finke and Stark, 2005; also Johnson, 1963).

Conversely, shared ideology can shape communities above and beyond structural features. Durkheim ([1897] 1951) the-orized that religious contexts—moral communities—determine individual attitudes and actions. Empirical research in recentdecades has found support for the importance of ‘‘moral communities’’ for outcomes as varied as delinquency (Regnerus,2003; Stark, 1996), coital onset (Adamczyk and Felson, 2006), religious participation (Hill, 2009), suicidality (Maimon andKuhl, 2008), volunteering (Lim and MacGregor, 2012), and gender attitudes (Moore and Vanneman, 2003). Even formal orga-nizations and economic markets are thought to depend on embeddedness in a ‘‘moral system’’ of ‘‘values, norms, and mean-ings’’ (Baker and Forbes, 2006, p. 23; Baker and Faulkner, 2009, p. 1532). Shared cultural beliefs of groups provide anadditional layer of motivation for participants to nurture within-group friendship ties and view such friendships as takingon deep and far-reaching importance (Berger, 1967, p. 38; Smith et al., 1998). This is consistent with the idea that, in general,people have greater difficulty enacting social structures without shared cultural schemas (Martin, 2009, pp. 16–20; Ridge-way, 2006, p. 7).

In a similar vein, Scheitle and Adamczyk (2009) find that shared theology at the congregation level has a direct effect oncongregational friendship embeddedness, with more particularistic theology at the group level associated with greater con-centration of friendship ties within the congregation. Similarly, Vaisey (2007) shows that ideological uniformity is the closestpredictor of greatest magnitude when assessing community feelings. Drawing on emotional theories of community, Vaiseyproposes a cultural approach centered on shared beliefs as a source of community feeling. This proposition follows Etzioni’scontention that social network infrastructure alone is not an adequate condition for the production of community. Socialnetworks’ ability to cultivate belonging is subject to the level of cultural uniformity (Etzioni, 2001, p. 223; Vaisey, 2007).Dovetailing these arguments and findings, shared ideology in congregations—measured as heuristic views of scripture—islinked directly to individuals’ sense of belonging (Stroope, 2011b). This line of reasoning and research have particular appli-cability to religious groups, given their use of sermon, song, recitation, and sacrament explicitly linked to scriptural and tes-timonial narratives.

Consequently, one way of identifying shared cultural belief in religious organizations is by looking to the group’s generalapproach to sacred texts (Stroope, 2011b).5 Variation in scriptural views exists across a spectrum of religious groups(Woodberry and Smith, 1998). Researchers have repeatedly found that scriptural views shape a variety of political, social,and cultural attitudes (Froese and Bader, 2010; Sherkat and Ellison, 1997; Wolkomir et al., 1997), lending support to the ideathat a shared view of scriptures—belief unity (Vaisey, 2007, p. 858) or consensus (Martin, 2002)—shapes other aspects of sociallife, such as how a person views or experiences his or her personal friendships and a related experiences of belonging. Culturalbelief unity may amplify the positive effects of personal relationships for sense of belonging because cultural orientations canshape the composition of social networks (Lizardo, 2006).6 For example, in recent longitudinal research, Vaisey and Lizardo(2010) find that respondents’ moral and theistic worldviews predict friendship network composition three years later. Inaddition to shaping the composition of networks, culture may also influence how networks are experienced.

Even Coleman, whose work is often derided for being a-cultural (e.g. Alexander, 1992; Rawls, 1992; cf. Adams, 2010),explicitly recognized the potentially critical role played by cultural consensus, linking it to the enforcement of norms: ‘‘ifthe consensus leads to a collective decision to expel the offender. . .then the second-order public good problem [free-riding]is overcome,’’ and that ‘‘consensus also increases the effectiveness of incremental sanctions. . .for these sanctions taken to-gether may be very powerful’’ (Coleman, 1990a, p. 284). He also added a brief section on ideology to explain how social cap-ital is generated before his heavily cited article (1988b) became a central chapter in Foundations of Social Theory (1990a, p.320). Indeed, ideological uniformity is a central component of what makes a group strict or permissive (Kelley, 1986).

Finally, alluded to above, ideological unity at the group level may have a stronger link to an individual’s sense of belong-ing when he or she has many friends in the congregation. Individuals that are less socially connected to the congregationmay have limited awareness of the cultural unity of the congregation, and therefore their sense of belonging would be lessinfluenced by the existence of unity. Taken together, the foregoing considerations lead to:

Hypothesis 5. The positive relationship between individual social embeddedness and feelings of group belonging will beamplified in more ideologically uniform groups.

To be clear, network emdeddness is our primary predictor of interest as we assess the degree to which group-level attri-butes influence the relationship between individuals’ networks and sense of belonging. We focus on social networks as theprimary independent variable because of their role as a proximate cause of belonging feelings. That is, the theoretical per-spective we outline above places individual social networks as a factor more immediately related to belonging than organi-zational characteristics, based on the idea that social ties a person values as ‘‘friendships’’ are a more proximate source ofbelonging feelings, while group attributes provide the (relatively) more distal context in which such relations occur.Fig. 1 provides a graphic depiction of the hypotheses we test concerning the structural and cultural sources of belongingin religious congregations. Of course, it would also be feasible to hypothesize and examine the degree to which individual

5 Kanter (1972, pp. 3–4) pointed out that the ‘‘earliest American community-builders,’’ many of whom were pietist immigrants intent on founding religiousutopian communities, were shaped by a vision to ‘‘return, for example, to a literal interpretation of the Bible.’’

6 At the contextual level, Baker and Faulkner (2009, p. 1532) note that ‘‘both networks and culture are bases of action, sometimes working together, impellingpeople in the same direction.’’

Fig. 1. Hypothesized relationship between network embeddedness and group structure and culture for feelings of group belonging.

6 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

networks moderate (or mediate) the influence of group level attributes on sense of belonging (see Berry et al. (2012) on thesymmetry of interaction effects). There are likely interesting and relevant effects to be mapped from such analyses; however,the theory we have outlined would still view the moderation (or mediation) of group attributes by individual social networksfrom the same perspective—as moving from more distal (group attributes) to more proximate (network) causes.

3. Data and methods

The data for this study come from the 2001 administration of the U.S. Congregational Life Survey (USCLS). The USCLS iscomprised of surveys of 424 congregations and their members (member N = 122,404). In drawing the USCLS’ sample, hyper-network sampling was utilized (McPherson, 1983). This national random sample (N = 1214) was taken by the National Opin-ion Research Center (NORC), located at the University of Chicago. Thirty-four percent of potential respondents completedcongregation questionnaires and member questionnaires. Although a response rate above 34% is desirable, research indi-cates that response rates in such samples do not meaningfully impact regression coefficients (AAPOR, 2008; Singer,2006). Additionally, the USCLS compares favorably to data from the 1998 National Congregations Study (Hill and Olson,2009). Individuals under the age of eighteen and those identifying as visitors are excluded from analyses. The analytic sam-ple below consists of 78,895 individuals nested in 399 congregations.7 Additional details on the USCLS data can be found indocumentation by Woolever and Bruce (2002).

3.1. Response measure

Sense of belonging is measured as a respondent’s subjective evaluation of feeling at home in her or his religious con-gregation. This item asked, ‘‘Do you have a strong sense of belonging to this congregation?’’ Respondents chose amongseven answers: 1 = Yes, a strong sense of belonging that is growing (48%), 2 = Yes, a strong sense—about the same as lastyear (25%), 3 = Yes, but perhaps not as strong as in the past (8%), 4 = No, but I am new here (5%), 5 = No, and I wish I did bynow (4%), 6 = No, but I am happy as I am (5%), and 7 = Don’t know or not applicable (4%). Given our focused interest inwhether or not an individual possesses a sense of belonging, we utilize a binary response measure (1, 2, and 3 = 1, else0). Follow prior work (e.g., Dougherty and Whitehead, 2010; Stroope, 2011b; Martinez and Dougherty, 2013), this ap-proach separates those who report having a sense of belonging (81%) from all other respondents. A limitation of the USCLSdata and extant work is that participants’ sense of belonging is measured using only one item. Given that single-item mea-sures generally have lower reliability, future research can build on the present study by using a multi-item measure ofbelonging.8

3.2. Key independent variables

Key structural independent variables are individual network embeddedness, organizational social network density, andcongregation size. Network embeddedness/density uses an item that asked individuals about friendships in their congrega-tion. Possible answers were: 1 = little contact with people, 2 = some friends, 3 = some close friends, and 4 = most of closest

7 This complete case approach to missing values in the USCLS data allows us to compare results to most USCLS studies which use a complete case approach(e.g., Scheitle and Finke, 2008; Dougherty and Whitehead, 2010; Scheitle and Dougherty, 2010). See Appendix A for replication of this analysis using multiplyimputed data.

8 Alternate coding strategies do not substantively change results. Relationships between variables were not significantly changed by alternate codingstrategies including coding the dependent variable as one if categories one and two are true, or if just category one is true. Additionally, in ancillary models weexcluded response category six, ‘‘No, but happy as I am.’’ Results were not significantly impacted.

Table 1Descriptive statistics.

Mean SD Min Max

Dependent variableSense of belonging 0.81 0.39 0 1Level 1 (individual level)Age 50.69 16.25 18 100Female 0.60 0.49 0 1Race

(White) 0.79 0.41 0 1Black 0.04 0.21 0 1Hispanic 0.11 0.31 0 1Asian 0.04 0.19 0 1Other 0.02 0.15 0 1

Income 3.65 1.47 1 6Married 0.58 0.49 0 1Child 0.58 0.49 0 1Education 5.65 1.71 1 8Church attendance 5.77 1.10 1 7Longevity of attendance 4.91 1.83 1 7Bible view 4.21 0.59 1 5Network embeddedness 2.65 0.90 1 4

Level 2 (congregation level)Religious tradition

(Catholic) 0.23 0.42 0 1Black Protestant 0.04 0.19 0 1Mainline 0.38 0.49 0 1Jewish 0.01 0.11 0 1Evangelical Protestant 0.31 0.46 0 1Other 0.04 0.18 0 1

Size (LN) 5.89 1.26 2.30 8.88Social network density 2.78 0.25 1.79 3.50Bible belief unity 0.49 0.12 �0.14 1.00

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 7

friends. The individual-level network embeddedness variable is coded as above. This single-item measure of networkembeddedness is a limitation of the USCLS. However, in analyses using subjects’ single-item religious network metricsand objective ego network metrics in regressions predicting religious identity and participation, both kinds of measuresbehave ‘‘remarkably similarly’’ (Kadushin et al., 2012, p. 172). At the organization level, social network density is a measureof the congregational average of the individual network embeddedness measure. Congregation size, another measure ofstructure, is the number of regular congregation participants (logged).9

Ideological unity is measured using a measure of scriptural views. Following prior research (Stroope, 2011b), we usedresponses to a question that asked study participants about their perception of the Bible. Although a limitation of this mea-sure is that it is a single-item, view of the Bible is a key measure of religious culture widely used in the literature (for a reviewsee Stroope, 2011a). We constructed a scale ranging from low to high levels of belief regarding the divine inspiration or con-temporary relevance of the Bible (range: 1–5). To measure the degree of congregation-level ideological uniformity aroundbiblical views, we use the measure’s standard deviation inversed. Greater values for this variable indicate increased levelsof ideological uniformity. This strategy of measuring a key ideology indicator (biblical views) is suitable because: (1) formeasuring variation in the case of a nominal indicator, Herfindahl or entropy indices are preferable measurement strategies,but not in the case of an ordinal variable (biblical view is measured as an ordinal indicator) where the standard deviation is asuitable indicator of variability among individuals nested within an organization; (2) this measurement strategy is in linewith its repeated use in organization research (e.g. McPherson, 1983); and (3), this measurement approach is in keeping withprior congregation research (Smith, 2010; Stroope, 2011b).

3.3. Control variables

Denominational affiliation of congregations is measured as a system of dummy indicators for: black Protestant, evangel-ical Protestant, mainline Protestant, Catholic, Jewish, and other tradition (Steensland et al., 2000). Correlation matrices ofcontextual and individual-level independent variables and multiple regression diagnostics did not indicate that models suf-fer from multicollinearity-related problems (all values <.60; VIFs <2.5). Control variables include gender (1 = female), age (inyears), race (a series of binary variables: white, Hispanic, black, Asian, and other race), parent (1 = child at home), married(1 = married), educational attainment (range: 1 = none to 8 = Master’s degree or above), household income (range:

9 In ancillary analyses (not shown), coding size as a binary variable (>2500) or leaving size as a continuous measure did not produce substantively differenceresults.

8 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

1 = <$10,000 to 6 = >$100,000), worship service attendance (1 = first time to 7 = more than weekly), and attendance longevity(the length of time the respondent has attended the congregation: 1 = less than a year to 6 = 20+ years). Descriptive statisticsfor study variables are shown in Table 1.10

3.4. Analytic strategy

Because individuals are nested within congregations, these data do not match the assumptions of basic modeling ap-proaches. In contrast to basic approaches, hierarchical modeling techniques recognize the within-group respondent covari-ance when producing group-level parameter estimates. We use hierarchical modeling in order to simultaneously estimategroup-level and individual-level effects and more accurately assess how group and individual processes interact. With theexception of dichotomous indicators, this study’s explanatory measures are grand-mean centered. The dependent measureis dichotomous, therefore we use hierarchical logistic regression (Raudenbush and Bryk, 2002).

We retained a dichotomous dependent variable in primary analyses because it allowed us to make a clear distinctionbetween those with and without a strong sense of belonging, and also because separate models using ordered logistic regres-sion analyzing a quasi-continuous version of the dependent variable violated the proportional-odds assumption, producingpotentially biased estimates (Agresti, 2002). We also decided against using hierarchical multinomial logistic regression,which estimates a different set of logits for each value of the dependent variable in contrast to a suppressed value.11 Thismethod results in a large number of contrasts and produces results in which general patterns are difficult to identify. Multino-mial logistic regression also obscures the basic theoretical arguments in this study, which focus on presence or absence ofbelonging rather than the growth, decline, or stability of belonging over time. Overall, binary logit models are easier to interpret,more broadly accessible, and more consistent with the proposed hypotheses.

Model 1 considers person-level effects. The coefficients for the following variables are allowed to vary across congrega-tions: social embeddedness, income, service attendance, and attendance longevity. The second model considers the role ofcongregational characteristics, while simultaneously taking into account individual-level effects on sense of belonging.Cross-level interaction effects are added in Model 3. This slopes-as-outcomes model examines the degree to which congre-gational characteristics condition the effect of individual embeddedness on sense of belonging.

4. Results

Prior to testing focal hypotheses, we estimated a null model (not shown), which tests for the existence of between-groupvariation in the response variable (v2 = 6341.179, p < .001). The intra-class correlation (ICC = .30) indicates that 30% of thetotal variance in belonging feelings is found at the group level. Thirty percent is substantial given that five to twenty percentis common in multilevel modeling (see Snijders and Bosker, 1999, p. 46).

The first model of Table 2 shows that the individual-level measure of structure, social network embeddedness, is stronglyand positively related to feeling belonging, net of individual-level controls. These results support Hypothesis 1. The magni-tude of the effect is also considerable. In terms of standardized effect, a one unit (standard deviation) increase in networkembeddedness increases the odds of belonging by 83%. Individual network embeddedness is the third largest effect in themodel (c = .671, p < .001) after longevity and frequency of attendance. Model 1 allows the embeddedness variable to varyacross congregations. This specification of a random effect to the embeddedness slope tests whether the effect of embedd-edness varies over congregations. The significant random effect indicates that there is variation in worshippers’ belongingfeelings across congregations (variance component = .042, p < .001). This random effect persists across all models.

Briefly, significant person-level effects in Model 1 also included the following. As age increases, the odds of feeling a senseof belonging increases (standardized effect = 4%). Being female diminishes the odds of belonging feelings by 9%. Blacks,Hispanics, and Asians have higher odds of feeling belonging than do whites (standardized effects = 24%, 18%, and 27% respec-tively). ‘‘Other’’ race individuals show no difference to whites. Educational attainment has a negative effect on belongingfeelings (standardized effect = 8%). Income does not significantly influence sense of belonging. Individuals who have at leastone child have 12% greater odds of feeling belonging. Frequency of service attendance has a positive and significant effect:with one standardized unit increase in service attendance, the odds of belonging increase by 85%. Finally, longevity ofattendance exhibits the largest effect in the model, more than doubling of the odds of belonging with a one standardizedunit increase in longevity.

The second model in Table 2 introduces contextual effects. A diminished AIC value in this model indicates enhancedmodel fit compared to the individual-level model. The contextual-level cultural variable, ideological uniformity, is linkedto greater odds of member belonging. Here, members of congregations with the greatest ideological uniformity have roughlytwice (e1.00�.605/e�.136�.605) the odds of reporting belonging compared to members of congregations with minimum ideolog-ical uniformity (standardized effect = 7%). Increased congregation size significantly reduces odds of belonging. Illustratively,

10 The USCLS also includes a measure of religious exclusivity belief. This study’s findings are not meaningfully impacted by controlling for this variable at theindividual level (p < .0001) and as exclusivity belief uniformity at the congregation level (p = .457). See Appendix A for separate analyses with additional controlvariables included at the congregation level.

11 Results from multinomial models showed that contrasts between the higher values on the sense of belonging variable were not as distinct as contrastsbetween having no belonging, declining belonging, and stable or growing belonging.

Table 2Sense of belonging on network embeddedness, congregational size, and belief unity: hierarchical logistic regression models (coefficients and standard errors reported).

Model 1 Model 2 Model 3

b SE b SE b SE

Intercept 1.930 0.04*** 1.902 0.06*** 1.927 0.06***

Level 1 (individual)Agea �0.002 0.00** �0.002 0.00* �0.002 0.00*

Female �0.092 0.02*** �0.093 0.02*** �0.094 0.02***

Race(White)Black 0.216 0.07** 0.161 0.08* 0.163 0.08*

Hispanic 0.167 0.04*** 0.213 0.04*** 0.214 0.04***

Asian 0.236 0.06*** 0.275 0.06*** 0.278 0.06Other �0.097 0.06 �0.091 0.07 �0.087 0.07

Incomea �0.021 0.01 �0.017 0.01 �0.017 0.01Married 0.044 0.02* 0.051 0.02* 0.051 0.02*

Child 0.111 0.02*** 0.108 0.02*** 0.109 0.02***

Educationa �0.050 0.01*** �0.053 0.01*** �0.054 0.01***

Church attendancea 0.558 0.01*** 0.585 0.01*** 0.586 0.02***

Longevity of attendancea 0.388 0.01*** 0.416 0.01*** 0.420 0.01***

Bible viewa 0.240 0.02*** 0.270 0.02*** 0.272 0.02***

Network embeddednessa 0.671 0.02*** 0.745 0.02*** 0.839 0.02***

Level 2 (congregational)Religious tradition

(Catholic)Black Protestant 0.594 0.16*** 0.599 0.16***

Mainline Protestant 0.461 0.07*** 0.464 0.07***

Jewish 1.065 0.30** 1.088 0.31**

Evangelical Protestant 0.072 0.08 0.078 0.08Other 1.106 0.16*** 1.108 0.16***

Size (LN)a �0.114 0.02*** �0.122 0.03***

Social network densitya 0.013 0.12 0.070 0.12Bible belief unitya 0.605 0.29* 0.721 0.30*

Cross-level interactions (level 2 � level 1)Size (LN) � individual network embeddedness �0.045 0.01**

Social network density � individual network embeddedness 0.300 0.07***

Bible belief unity � individual network embeddedness 0.475 0.16**

Deviance 197983.9 197851.8 197808.7AIC 198045.9 197929.8 197892.7BIC 198135.8 198042.8 198014.4

Congregation N = 399; Individual N = 78,895.a Variable centered around its grand mean.

* p < .05.** p < .01.*** p < .001.

S.Stroope,J.O.Baker/Social

ScienceR

esearch45

(2014)1–

179

10 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

individuals in the smallest congregations have 2.1 times (e2.303��0.114/e8.882��0.114) greater odds of belonging than membersof the largest congregations. With a one standard deviation increase in congregation size, the odds of belonging decline by13%. Turning to the other contextual variables, the positive effect of network density is in the hypothesized direction, but isnot statistically significant, failing to support Hypothesis 2.

Among contextual-level controls, all religious traditions are significantly and positively associated with greater odds ofbelonging compared to Catholics, with the exception of evangelical Protestants. Evangelical affiliation is positively but notsignificantly associated with the dependent variable in comparison to Catholic affiliation.

Model 3 examines the degree to which the effect of individual-level structure (i.e., social network embeddedness) is mod-erated by congregation-level characteristics. The AIC value of this model shows that the inclusion of the interaction termsimproves model fit. In terms of the specific interaction effects, the model shows that, net of individual and group attributes,the main effect of individuals’ personal network embeddedness is conditioned by congregation size, such that the influenceof embeddedness on the probability of feeling belonging diminishes as congregation size increases (c = �0.045, p < .01).Conversely, this interaction also implies that individual embeddedness is more strongly related to belonging in smaller con-gregations. Fig. 2 illustrates the predicted probabilities of sense of belonging in the interplay between social embeddednessand congregation size. This interaction between the congregation size and the individual-level network aspect of structuresupports Hypothesis 4.

The next interaction effect in Model 3 supports Hypothesis 3. The interaction between congregation-level network den-sity and individual network embeddedness is positive and significant (c = 0.300, p < .001). Fig. 3 displays differences in theslopes of individual embeddedness effects and shows that in low network density congregations, individuals with low socialembeddedness actually have greater odds of belonging than more socially embedded individuals. However, the likelihood ofbelonging is increased in high density congregations for individuals with higher levels of individual embeddedness, such thattheir likelihood of belonging surpasses that of less embedded individuals. The shape of these slopes show that with anincrease in individuals’ embeddedness, the gap in belonging feelings between members of high and low density congrega-tions converge.

Finally, the effect of individual network embeddedness is significantly conditioned by congregation-level ideologicalunity (c = 0.475, p < .01). Specifically, the positive effect of network embeddedness on sense of belonging is amplified forpeople in congregations with greater ideological uniformity. Put differently, there is a smaller effect of personal networkembeddedness in less ideologically unified congregations and a greater effect in more unified congregations. This interactionimplies that the differences between less embedded and more embedded people are smaller in ideologically mixed congre-gations compared to ideologically unified congregations. Fig. 4 shows that network embeddedness positively affects sense ofbelonging in the least ideologically unified context, but that the effect is more strongly positive in ideologically unifiedcongregations. The interaction supports Hypothesis 5, that highly embedded members of ideologically coherent churcheshave greater odds of belonging.

Overall, network embeddedness has a substantial effect on sense of belonging in religious groups, and its effects are con-ditioned by organization-level structural and cultural characteristics of congregations.

4.1. Supplementary analyses and limitations

Missing values were present on some study variables. The percentages missing were: belonging (3%), education (1.5%),gender (4.5%), married (1.2%), income (10.9%), longevity of attendance (0.9%), attendance (2.7%), biblical view (5.6%), and

0.53

0.65

0.77

0.88

1.00

Sens

e of

Bel

ongi

ng

-1.64 -0.90 -0.16 0.58 1.33

Network Embeddedness

Large CongregationMean Congregation SizeSmall Congregation

Fig. 2. Interactive effects of individual network embeddedness and congregation size on the probability of feeling a sense of belonging.

0.24

0.43

0.62

0.81

1.00

Sens

e of

Bel

ongi

ng

-1.64 -0.90 -0.16 0.58 1.32

Network Embeddedness

High Network DensityMean Network DensityLow Network Density

Fig. 3. Interactive effects of individual network embeddedness and congregation network density on the probability of feeling a sense of belonging.

0.61

0.71

0.80

0.90

1.00

Sens

e of

Bel

ongi

ng

-1.64 -0.90 -0.16 0.58 1.32

Network Embeddedness

High Belief UnityMean Belief UnityLow Belief Unity

Fig. 4. Interactive effects of individual network embeddedness and congregation belief unity on the probability of feeling a sense of belonging.

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 11

network embeddedness (3.4%). We applied multiple imputation using PROC MI in SAS 9.2 and conducted the following anal-yses on five imputed datasets. Continuous independent variables were grand-mean centered.

Table A1 in Appendix A repeats the analysis seen in Table 2, while also adding controls for region of the country (North-east, West, Midwest, and South (reference)). Results for focal variables (i.e., individual network embeddedness and the inter-action terms and significant lower-order effects) in Models 1 through 3 of Table A1 are the same as the models in Table 2 interms significance levels and the direction of effects for significant variables. Figs. A1–A3 graph the focal interaction effects,which show no meaningful differences compared to those of the complete case models (Figs. 2–4). Model 4 in Table A1 addsadditional potentially confounding variables. Added at the individual level is a measure of religious exclusivity and at thecongregation level the following aggregate variables: average religious exclusivity, average levels of service attendance,average education, average income, and average attendance longevity. The focal effects in Model 4 are similar to those ofModel 3.12

One of the limitations of our study is that we are constrained by the single-item measurement for sense of belonging inthe USCLS. Although we implemented a variety of coding strategies in analyzing sense of belonging, we are unable to movebeyond this fundamental limitation of the data. To assess the predictive validity of using the single-item metric of belongingin the USCLS, we examined the binary belonging measure with other, related measures of group involvement in the USCLS.The belonging measure co-varied with other metrics in expected ways. People who reported a sense of belonging were much

12 Key effects were not significantly changed in binary logistic regression models using an alternate coding of the dependent variable (1 = growing sense ofbelonging, else 0) and the same sets of controls in Models 3 and 4 of Table A1. One exception is the interaction term between network embeddedness andcongregation size, which becomes nonsignificant.

12 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

more likely to attend frequently, be formal members, have a long tenure in the group, give more money, report optimismabout the future of the group, experience spiritual growth and joy from participation, and consider themselves decision-makers within the organization.

We also compared the results of our study to results in similar extant literature. At present, we are only aware of ahandful of other studies examining sense of belonging in congregations as an outcome. Similar to Krause and Cairney’s(2009) three-item index of belonging, we find that non-whites, high attenders, and people with embedded social networksare more likely to report congregational belonging. The closest study to our own is that of Ryle and Robinson (2006),whose findings for education and marital status in relation to congregational belonging match those presented here. Ina different setting, but consistent with our organization-level structural findings, Vaisey (2007) found that increasedgroup size in communes decreased levels of group belonging. These comparisons along with the expected relationshipswithin the USCLS indicate that the single-item measure is picking up meaningful aspects of belonging. Still, while sin-gle-item metrics have utility (see Dollinger and Malmquist, 2009), they are clearly less than ideal. Future research shouldexpand both the scope and measurement of belonging in congregations, as it is a central concept in the study of individ-ual-group relations.

Future large-N, multilevel surveys that sample congregations and members can build on the present understanding ofbelonging by developing and fielding a more comprehensive battery of questions on the concept. Such multi-item measure-ment can help reduce the possibility of low reliability and potentially identify different dimensions of communal integration.Relatedly, although we have focused specifically on the importance of biblical views as an aspect of congregational culture,subsequent studies should ask questions about a broader set of beliefs that might more fully tap the ideological culture ofreligious congregations. Research on communal groups utilizing multi-item scales of both cultural antecedents of commu-nity feeling and community feeling itself (Zablocki, 1980) provide a fruitful model for advancing measurement goingforward.

Additionally, the single-item measure of social network embeddedness in the USCLS is a limitation.13 The relationship be-tween individual network embeddedness and feelings of belonging may be complicated by the degree to which members of anindividual’s network are attached to the congregation themselves. For example, members of some friendship networks in agroup may feel disenfranchised or on the periphery of the social life of a congregation because core members have few rela-tional nodes available for reaching out to less integrated individuals. Embeddedness would then only promote belonging tothe degree that an individual’s personal network is in the structural core of the larger group. In other words, having friendsin the center of organizational life would increase sense of belonging in the organization more than having friends on theperiphery of organizational life. A more thorough explication of the effects of network embeddedness on feelings of belongingcould be attained by assessing the degree to which members of an individual’s network are integrated into the power structureof a congregation. Criminologists have examined the religious attributes reported by friends of respondents (Adamczyk and Fel-son, 2006), and were data available, a similar method could help clarify and extend the present analysis. Future research shouldutilize more granular, multi-item measurement strategies.14

A final limitation is that available multilevel data do not allow us to definitively show that individual or congregational-level structural and cultural characteristics cause feelings of belonging. One could imagine that first possessing feelings ofbelonging would allow individuals the motivation to forge better social relationships, which then provides the relationalinfrastructure for the development of unified beliefs. It is also possible that people who value belonging seek out smaller,more relationally connected congregations where they can feel at home with others who have similar goals. Clearly, theseinfluences may work in both directions reciprocally in aggregate data. Still, the most plausible understanding of these pro-cesses locates the bulk of causation with the social ties of individuals and contextual features of groups, rather than emo-tional states of belonging. Group interaction creates emotional states in participants (Collins, 2005). This causal directionis also the most consistent with the theoretical background presented at the outset. Methodologically, hierarchically nesteddatabases of individuals and congregations tracked and measured over time could shed more light on the direction of cau-sation; but even with longitudinal data, many of the processes of interest are coterminous, precluding a definitive empiricaladjudication of causal direction.

5. Discussion

Community and social capital literatures suggest that contextual factors (both structural and cultural) interact with indi-vidual structural position, but the empirical patterns and specifics of these dynamics are only beginning to be understood. Inaddition, the outcome of interest dictates how aspects of structure, culture, and group dynamics matter. The findings of thepresent study add to this effort. For assessing the structural dimensions of success in business environments, where theoutcomes of interest are competitiveness, innovation, and breadth of networking, ‘‘structural holes’’ are central because theyfacilitate brokerage and flexibility, while network closure plays a supplemental role by making organizations more stable

13 The sense of belonging item in the USCLS came from an ongoing survey of Australian Protestant and Catholic congregations, the National Church Life Survey(NCLS). The NCLS has surveyed Protestants since 1991.

14 One approach that would generate richer data would entail asking participants to list their closest friends. Information could then be gathered about thesefriends’ congregation memberships and levels of belonging.

Fig. A1. Interactive effect of individual network embeddedness and congregation size on the probability of feeling a sense of belonging.

Fig. A2. Interactive effect of individual network embeddedness and congregation network density on the probability of feeling a sense of belonging.

Fig. A3. Interactive effect of individual network embeddedness and congregation belief unity on the probability of feeling a sense of belonging.

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 13

(see Burt, 2001a). For assessing feelings of belonging in religious groups, theorizing about network closure and the creationand enforcement of group norms play a more prominent role.15

At the individual level, we find that congregation-based social network embeddedness is among the strongest predictorsof belonging feelings; however, these network effects operate differently across congregations due to characteristics at thegroup level. The evidence presented shows that whether congregations foster belonging depends not only on members’ net-work embeddedness in a congregation, but also the presence of a unified ideology and structural dimensions of the group,especially size and network density. This clarifies some of the specific mechanisms in the interplay between group context,

15 There is also a downside to tight-knit groups, such as more intense expressions of exclusion and control (see Burt, 2001b). This aspect of network closurealso explains much of the ‘‘dark side’’ of social capital noted in previous research, as it usually pertains to ‘‘closed’’ communities. Tight in-group cohesionnecessarily means strong out-group exclusion.

Table A1Multilevel binary logistic regression models of sense of belonging (coefficients and standard errors reported).

Model 1 Model 2 Model 3 Model 4

b SE b SE b SE b SE

Intercept 1.169 0.030*** 1.002 0.056*** 0.986 0.056*** 1.140 0.074***

Level 1 (individual)Age 0.051 0.001 0.038 0.001 0.037 0.001 0.019 0.001Female �0.053 0.017** �0.057 0.017** �0.057 0.018** �0.065 0.018***

Black (Ref. = White) 0.245 0.059*** 0.186 0.067** 0.187 0.065** 0.164 0.065*

Hispanic (Ref. = White) 0.230 0.030*** 0.256 0.031*** 0.254 0.031*** 0.245 0.031***

Asian (Ref. = White) 0.276 0.047*** 0.260 0.049*** 0.259 0.049*** 0.232 0.049***

Other race (Ref. = White) �0.078 0.048 �0.093 0.049 �0.094 0.049 �0.095 0.049Income �0.003 0.007 �0.001 0.007 �0.001 0.007 �0.001 0.007Married 0.031 0.018 0.022 0.018 0.022 0.018 0.027 0.018Child 0.065 0.017*** 0.067 0.018*** 0.069 0.018*** 0.069 0.018***

Education �0.045 0.006*** �0.047 0.006*** �0.047 0.006*** �0.043 0.006***

Longevity of attendance 0.188 0.005*** 0.186 0.005*** 0.186 0.005*** 0.187 0.005***

Church attendance 0.588 0.008*** 0.592 0.009*** 0.591 0.009*** 0.596 0.009***

Bible view 0.216 0.015*** 0.220 0.016*** 0.218 0.016*** 0.228 0.016***

Network embeddedness 0.599 0.016*** 0.587 0.016*** 0.641 0.018*** 0.642 0.019***

Religious exclusivity �0.063 0.008***

Level 2 (congregational)Northeast (Ref. = South) �0.032 0.065 �0.019 0.066 0.010 0.062Midwest (Ref. = South) �0.076 0.055 �0.060 0.054 �0.008 0.052West (Ref. = South) �0.060 0.061 �0.049 0.060 0.012 0.058Mainline Protestant (Ref. = Evangelical Protestant) 0.389 0.059*** 0.399 0.060*** 0.172 0.090Black Protestant (Ref. = Evangelical Protestant) 0.457 0.137*** 0.479 0.138*** 0.312 0.137*

Catholic (Ref. = Evangelical Protestant) 0.109 0.073 0.108 0.074 �0.179 0.117Jewish (Ref. = Evangelical Protestant) 1.363 0.252*** 1.394 0.252*** 1.216 0.287***

Other religion (Ref. = Evangelical Protestant) 0.886 0.144*** 0.884 0.144*** 0.614 0.159***

Size (logged) �0.083 0.024*** �0.091 0.024*** �0.079 0.026**

Social network density �0.078 0.106 �0.032 0.109 0.250 0.131Bible belief unity 0.734 0.259** 0.865 0.258*** 1.125 0.281***

Religious exclusivity �0.295 0.066***

Church attendance 0.062 0.095Education �0.006 0.050Income �0.039 0.062Longevity of attendance �0.220 0.041***

Cross-level interactions (level 2 � level 1)Size (logged) � individual network embeddedness �0.040 0.013** �0.038 0.014**

Social network density � individual network embeddedness 0.152 0.063* 0.174 0.065**

Bible belief unity � individual network embeddedness 0.515 0.147*** 0.464 0.157**

Note: Multiply imputed data are used. Age betas are multiplied by 100 for ease of presentation.* p < .05.** p < .01.*** p < .001.

14S.Stroope,J.O

.Baker/SocialScience

Research

45(2014)

1–17

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 15

individuals, and belonging feelings by outlining how social networks interact with organizational characteristics. We antic-ipate that these patterns also operate in other types of groups with regard to sense of belonging.

We find that the strong positive influence of individual congregation-based social relationships is significantly degradedin larger congregations. Organizational researchers have offered reasons why larger groups tend to be less effective in pro-moting feelings of community, and our findings are consistent with those arguments. Larger congregations face challenges inintegrating the average attendee, not only because leaders must become more task-specialized and less holistically focusedon individuals, but also because the general scale of the physical environment feels more cavernous and impersonal. As Cole-man noted, it is much more difficult to establish and police group norms effectively as the size of the organization increases.The implications of the effects of congregation size on sense of belonging are particularly noteworthy because the share ofworshippers attending large congregations is growing in the United States (Chaves, 2006, 2011). The ability of congregationsto offset the negative impact of size on belonging may also have consequences for the degree to which church attendancemaintains its association with well-being outcomes. Voluntary organizations must compete for individuals’ scarce resources,so the erosion of belonging feelings by increased size could also diminish congregations’ fiscal health by dampening moti-vation for giving to faith communities.

Congregational network density amplifies individual network effects on sense of belonging, we find, but this relationshipis not as anticipated. Only at extreme levels of individual network embeddedness do members of more densely networkedcongregations experience stronger belonging than members of less densely networked congregations. In fact, at low levels ofindividual embeddedness, worshippers in less densely networked congregations are more likely to feel belonging. These re-sults are somewhat surprising, but interpretable. Less socially integrated worshippers at densely networked congregationsmay ‘‘find such churches ‘cliquish’ because high-tenure members have as many friends as they want or can manage’’ (Olson,1989, p. 1). In contrast, a worshipper will likely sense openness and welcome in a congregation with a preponderance of lesssocially integrated individuals, a group of people looking for new, weak ties themselves.

Contrary to Coleman’s generalized theory that higher group density leads to belonging, it is alignment between individualand organizational network characteristics that produces a greater sense of belonging in these data, while a mismatch ofthese dimensions produces less positive emotion derived from group participation, at least at low levels of network integra-tion. This applies especially to groups operating with what Edgell termed a ‘‘house of worship’’ organizational model. Suchgroups focus on creating a connection between worshippers and the divine, but focus less on connecting individuals in thegroup to one another. Congregations operating with this type of model do not ‘‘see religion as something that overlaps andreinforces ties of friendship, affection and community, nor as a publically oriented ‘player’ in local affairs’’ (Becker, 1999, p.75). In these contexts, a social network that is not integrated with the group actually aligns with the organizational model,and thereby produces comparatively more sense of belonging. These differential effects equalize near the mean of socialnetwork embeddedness, after which there are generally equal, positive returns on belonging for high network integration.

Sense of belonging is a central component of integration into a group. The factors that lead to this collective and individ-ual feeling should be of particular interest to researchers devoting attention to how organizations foster or inhibit groupbuy-in and individual sacrifice. In this regard, the degree to which members’ social networks are integrated into an organi-zation must be considered focal; however, the structural and cultural dimensions of groups condition how well networkintegration translates into feelings of belonging, demonstrating that social connections interact with organizationalcharacteristics to foster feelings of belonging.

Acknowledgments

We thank Kevin Dougherty, Paul Froese, Andrew Whitehead, anonymous reviewers, and the editor for comments on aprevious version. Thanks to Cynthia Woolever, Ruth Powell, and Bob Dixon for helpful correspondence regarding surveyitems in the USCLS and NCLS.

Appendix A

See Figs. A1–A3 and Table A1.

References

Adamczyk, A., Felson, J., 2006. Friends’ religiosity and first sex. Soc. Sci. Res. 35 (4), 924–947.Adams, J., 2010. The unknown James Coleman: culture and history in foundations of social theory. Contemp. Sociol. 39 (3), 253–258.Agresti, A., 2002. Categorical Data Analysis, second ed. John Wiley & Sons, Hoboken, New Jersey.Alexander, J.C., 1992. Shaky foundations: the presuppositions and internal contradictions of James Coleman’s Foundations of Social Theory. Theory Soc. 21

(2), 203–217.American Association for Public Opinion Research, 2008. Do Response Rates Matter? <http://www.aapor.org> (accessed 23.11.10).Ammerman, N.T., 1995. Baptist Battles: Social Change and Religious Conflict in the Southern Baptist Convention. Rutgers University Press, New Brunswick.Ammerman, N.T., 2009. Congregations: local, social, and religious. In: Clark, P.B. (Ed.), The Oxford Handbook of the Sociology of Religion. Oxford University

Press, New York, pp. 562–581.Baker, J.O., 2010. Social sources of the spirit: connecting rational choice and interactive ritual theories in the study of religion. Sociol. Religion 71 (4),

432–456.Baker, W., Faulkner, R.R., 2009. Social capital, double embeddedness, and mechanisms of stability and change. Am. Behav. Sci. 52 (11), 1531–1555.

16 S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17

Baker, W., Forbes, M., 2006. Moral values and market attitudes. Society 43, 23–26.Becker, P.E., 1999. Congregations in Conflict: Cultural Models of Local Religious Life. University of Chicago Press, Chicago.Berger, P.L., 1967. The Sacred Canopy: Elements of a Sociological Theory of Religion. Doubleday, New York.Berry, W.D., Golder, M., Milton, D., 2012. Improving test of theories positing interaction. J. Politics 74 (3), 653–671.Beyerlein, K., Hipp, J.R., 2006. Social capital, too much of a good thing? American religious traditions and community crime. Soc. Forces 84 (2), 995–1013.Blanchard, T.C., Bartkowski, J.P., Matthews, T.L., Kerley, K.R., 2008. Faith, morality and mortality: the ecological impact of religion on population health. Soc.

Forces 86 (4), 1591–1620.Blau, P.M., 1970. A formal theory of differentiation in organizations. Am. Sociol. Rev. 35 (2), 201–218.Brunie, A., 2009. Meaningful distinctions within a concept: relational, collective, and generalized social capital. Soc. Sci. Res. 38 (2), 251–265.Burt, R.S., 2001a. Structural holes versus network closure as social capital. In: Lin, N., Cook, K., Burt, R.S. (Eds.), Social Capital: Theory and Research. Aldine De

Gruyter, New York, pp. 31–56.Burt, R.S., 2001b. Bandwidth and echo: trust, information, and gossip in social networks. In: Casella, A., Rauch, J.E. (Eds.), Networks and Markets:

Contributions from Economics and Sociology. Russell Sage Foundation, New York, pp. 30–74.Burt, R.S., 2005. Brokerage and Closure: An Introduction to Social Capital. Oxford University Press, New York.Burt, R.S., Ronchi, D., 2007. Teaching executives to see social capital: results from a field experiment. Soc. Sci. Res. 36 (3), 1156–1183.Chaves, M., 2006. All creatures great and small: megachurches in context. Rev. Religious Res. 47 (4), 329–346.Chaves, M., 2010. Rain dances in the dry season: overcoming the religious congruence fallacy. J. Sci. Study Religion 49 (1), 1–14.Chaves, M., 2011. American Religion: Contemporary Trends. Princeton University Press, Princeton.Coleman, J.S., 1986. Social structure and the emergence of norms among rational actors. In: Diekman, A., Mitter, P. (Eds.), Paradoxical Effects of Social

Behavior. Physica-Verlag, Heidelberg, pp. 55–83.Coleman, J.S., 1988a. Free riders and zealots: the role of social networks. Sociol. Theory 6 (1), 52–57.Coleman, J.S., 1988b. Social capital in the creation of human capital. Am. J. Sociol. 94 (S), 95–120.Coleman, J.S., 1990a. Foundations of Social Theory. Belknap/Harvard University Press, Cambridge.Coleman, J.S., 1990b. The emergence of norms. In: Hechter, M., Opp, K.-D., Wippler, R. (Eds.), Social Institutions: Their Emergence, Maintenance, and Effects.

Aldine De Gruyter, New York, pp. 35–60.Coleman, J.S., 1990c. Norm-generating structures. In: Cook, K.S., Levi, M. (Eds.), The Limits of Rationality. University of Chicago Press, Chicago, pp. 250–272.Coleman, J.S., 1992. The problematics of social theory: four reviews of Foundations of Social Theory. Theory Soc. 21 (2), 263–283.Collins, R., 1997. Stark and Bainbridge, Durkheim and Weber: theoretical comparisons. In: Young, L.A. (Ed.), Rational Choice Theory and Religion: Summary

and Assessment. Routledge, New York, pp. 161–180.Collins, R., 2005. Interaction Ritual Chains. Princeton University Press, Princeton.Davis, N.J., Robinson, R.V., 1996. Are the rumors of war exaggerated? Religious orthodoxy and moral progressivism in America. Am. J. Sociol. 102 (3),

756–787.Delanty, G., 2009. Community, second ed. Routledge, London.Dollinger, S.J., Malmquist, D., 2009. Reliability and validity of single-item self-reports: with special relevance to college students’ alcohol use, religiosity,

study, and social life. J. Gen. Psychol. 136 (3), 231–241.Dougherty, K.D., Whitehead, A.L., 2010. A place to belong: small group involvement in religious congregations. Sociol. Religion 72 (1), 91–111.Durkheim, E., [1897] 1951. In: Simpson, G. (Ed.), Suicide (J.A. Spaulding, Trans). Free Press, New York.Durkheim, E., [1912] 1995. The Elementary Forms of the Religious Life (K.E. Fields, Trans). Free Press, New York.Ellison, C.G., George, L.K., 1994. Religious involvement, social ties, and social support in a southeastern community. J. Sci. Study Religion 33 (1), 46–61.Ellison, C.G., Krause, N., Shepherd, B.C., Chaves, M., 2009. Size, conflict, and opportunities for interaction: congregational effects on members’ anticipated

support and negative interaction. J. Sci. Study Religion 48 (1), 1–15.Emerson, M.O., Smith, C., 2000. Divided by Faith: Evangelical Religion and the Problem of Race in America. Oxford University Press, New York.Emirbayer, M., Goodwin, J., 1994. Network analysis, culture, and the problem of agency. Am. J. Sociol. 99 (6), 1411–1454.Etzioni, A., 2001. Is bowling together sociologically lite? Contemp. Sociol. 30 (3), 223–224.Field, J., 2008. Social Capital, second ed. Routledge, London.Finke, R., 1994. The quiet transformation: changes in size and leadership of Southern Baptist churches. Rev. Religious Res. 36 (1), 3–22.Finke, R., Stark, R., 2005. The Churching of America, 1776–2005: Winners and Losers in Our Religious Economy, revised ed. Rutgers University Press, New

Brunswick.Froese, P., Bader, C., 2010. America’s Four Gods: What We Say About God and What That Says About Us. Oxford University Press, New York.Giddens, A., 1984. The Constitution of Society: Outline of the Theory of Structuration. California University Press, Berkeley.Hill, J.P., 2009. Higher education as moral community: institutional influences on religious participation during college. J. Sci. Study Religion 48 (3), 515–

534.Hill, J.P., Olson, D.V.A., 2009. Market share and religious competition: do small market share congregations and their leaders try harder? J. Sci. Study Religion

48 (4), 629–649.Hoffmann, J.E., Bartkowski, J.E., 2008. Gender, religious tradition and biblical literalism. Soc. Forces 86 (3), 1245–1272.Iannaccone, L.R., 1988. A formal model of church and sect. Am. J. Sociol. 94 (S), 241–268.Iannaccone, L.R., 1992. Sacrifice and stigma: reducing free-riding in cults, communes, and other collectives. J. Political Econ. 100 (2), 271–291.Iannaccone, L.R., 1994. Why strict churches are strong. Am. J. Sociol. 99 (5), 1180–1211.Idler, E.L., Boulifard, D.A., Labouvie, E., Chen, Y.Y., Krause, T.J., Contrada, R.J., 2009. Looking inside the black box of attendance at services: new measures for

exploring an old dimension in religion and health research. Int. J. Psychol. Religion 19 (1), 1–20.Johnson, B., 1963. On church and sect. Am. Sociol. Rev. 28 (4), 539–549.Kadushin, C., Wright, G., Shain, M., Saxe, L., 2012. How socially integrated into mainstream America are Young American Jews? Contemp. Jewry 32 (2), 167–

187.Kanter, R.M., 1972. Commitment and Community: Communes and Utopias in Sociological Perspective, first ed. Harvard University Press, Cambridge.Kelley, D.M., 1986. Why Conservative Churches are Growing. Mercer University Press, Macon.Knoke, D., 1981. Commitment and detachment in voluntary associations. Am. Sociol. Rev. 46 (2), 141–158.Koenig, H.G., McCullough, M.E., Larson, D.B., 2001. Handbook of Religion and Health, first ed. Oxford University Press, New York.Krause, N., 2006. Exploring the stress-buffering effects of church-based and secular social support on self-rated health in late life. J. Gerontol. Ser. B: Psychol.

Sci. Soc. Sci. 61 (1), S35–S43.Krause, N., 2008. The social foundation of religious meaning in life. Res. Aging 30 (4), 395–427.Krause, N., Cairney, J., 2009. Close companion friends in church and health in late life. Rev. Religious Res. 51 (2), 181–200.Krause, N., Wulff, K.M., 2005. Church-based social ties, a sense of belonging in a congregation, and physical health status. Int. J. Psychol. Religion 15 (1), 73–

93.Lim, C., MacGregor, C.A., 2012. Religion and volunteering in context disentangling the contextual effects of religion on voluntary behavior. Am. Sociol. Rev.

77 (5), 747–779.Lim, C., Putnam, R.D., 2010. Religion, social networks, and life satisfaction. Am. Sociol. Rev. 75 (6), 914–933.Lizardo, O., 2006. How cultural tastes shape personal networks. Am. Sociol. Rev. 71 (5), 778–807.Maimon, D., Kuhl, D.C., 2008. Social control and youth suicidality: situating Durkheim’s ideas in a multilevel framework. Am. Sociol. Rev. 73 (6), 921–943.Marsden, P.V., 2005. The sociology of James S. Coleman. Annu. Rev. Sociol. 31, 1–24.

S. Stroope, J.O. Baker / Social Science Research 45 (2014) 1–17 17

Martin, J.L., 2002. Power, authority, and the constraint in belief systems. Am. J. Sociol. 107 (4), 861–904.Martin, J.L., 2009. Social Structures. Princeton University Press, Princeton.Martin, J.L., 2011. The Explanation of Social Action. Oxford University Press, New York.Martinez, Brandon C., Dougherty, Kevin D, 2013. Race, belonging, and participation in religious congregations. J. Sci. Study Religion 52 (4), 713–732.McPherson, M., 1983. An ecology of affiliation. Am. Sociol. Rev. 48 (4), 519–532.Mencken, F.C., Bader, C.D., Polson, E.P., 2006. Integrating civil society and economic growth in Appalachia. Growth Change 37 (1), 107–127.Mische, A., 2010. Relational sociology, culture, and agency. In: Scott, J.P., Carrington, P. (Eds.), Social Network Analysis: A Handbook. Sage Publications,

Thousand Oaks.Moody, J., White, D.R., 2003. Structural cohesion and embeddedness: a hierarchical concept of social groups. Am. Sociol. Rev. 68 (1), 103–127.Moore, L.M., Vanneman, R., 2003. Context matters: effects of the proportion of fundamentalists on gender attitudes. Soc. Forces 82 (1), 115–139.Niebuhr, H.R., [1929] 1984. The Social Sources of Denominationalism. Peter Smith, Gloucester.Olson, D.V.A., 1989. Church friendships: boon or barrier to church growth? J. Sci. Study Religion 28 (4), 432–447.Paxton, P., Moody, J., 2003. Structure and sentiment: explaining emotional attachment to group. Soc. Psychol. Q. 66 (1), 34–47.Portes, A., 1998. Social capital: its origins and applications in modern sociology. Annu. Rev. Sociol. 24, 1–24.Portes, A., 2000. The two meanings of social capital. Sociol. Forum 15 (1), 1–12.Putnam, R.D., 2000. Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, New York.Raudenbush, S.W., Bryk, A.S., 2002. Hierarchical Linear Models: Applications and Data Analysis Methods. Sage Publications, Thousand Oaks.Rawls, A.W., 1992. Can rational choice be a foundation for social theory? Theory Society 21 (2), 219–241.Regnerus, M.D., 2003. Moral communities and adolescent delinquency. Sociol. Q. 44 (4), 523–554.Ridgeway, C.L., 2006. Linking social structure and interpersonal behavior: a theoretical perspective on cultural schemas and social relations. Soc. Psychol. Q.

69 (1), 5–16.Ryle, R.R., Robinson, R.V., 2006. Ideology, moral cosmology, and community in the United States. City Community 5 (1), 53–69.Schaller, L.E., 1982. Growing Plans: Strategies to Increase Your Church’s Membership. Abingdon Press, Nashville.Scheitle, C.P., Adamczyk, A., 2009. It takes two: the interplay of individual and group theology on social embeddedness. Journal for the Scientific Study of

Religion 48 (1), 16–29.Scheitle, C.P., Dougherty, Kevin, 2010. Race, Diversity, and Membership Duration in Religious Congregations. Sociol. Inquiry 80 (3), 405–423.Scheitle, Christopher P., Finke, Roger, 2008. Maximizing Congregational Resources: Selection Versus Production. Soc. Sci. Res. 37 (3), 815–827.Schwadel, P., 2005. Individual, congregational, and denominational effects on church members’ civic participation. J. Sci. Study Religion 44 (2), 159–171.Sewell Jr., W.H., 2005. Logics of History: Social Theory and Social Transformation. University of Chicago Press, Chicago.Sherkat, D.E., 1998. Counterculture or continuity? Competing influences on baby boomers’ religious orientations and participation. Soc. Forces 76 (3),

1087–1114.Sherkat, D.E., Ellison, C.G., 1997. The cognitive structure of a moral crusade: conservative Protestantism and opposition to pornography. Soc. Forces 75 (3),

957–980.Singer, E., 2006. Special issue on nonresponse bias in household surveys. Public Opin. Q. 70 (5), 637–645.Smilde, D., 2005. A qualitative comparative analysis of conversion to Venezuelan Evangelicalism: how networks matter. Am. J. Sociol. 111 (3), 757–796.Smith, C., 2003a. Moral, Believing Animals: Human Personhood and Culture. Oxford University Press, New York.Smith, C., 2003b. Theorizing religious effects among American adolescents. J. Sci. Study Religion 42 (1), 17–30.Smith, C., 2003c. Religious participation and network closure among American adolescents. J. Sci. Study Religion 42 (2), 259–267.Smith, B.G., 2010. Acceptance of other religions in the United States: an HLM analysis of variability across congregations. Soc. Compass 57 (1), 127–142.Smith, C., Emerson, M., Gallagher, S., Kennedy, P., Sikkink, D., 1998. American Evangelicalism: Embattled and Thriving. University of Chicago Press, Chicago.Snijders, T.A.B., Bosker, R.J., 1999. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. Sage Publications, Thousand Oaks.Stark, R., 1996. Religion as context: hellfire and delinquency one more time. Sociol. Religion 57 (2), 163–173.Stark, R., Bainbridge, W.S., 1980. Networks of faith: interpersonal bonds and recruitment to cults and sects. Am. J. Sociol. 85 (6), 1376–1395.Stark, R., Finke, R., 2000. Acts of Faith: Explaining the Human Side of Religion, first ed. University of California Press, Berkeley.Steensland, B., Park, J.Z., Robinson, L.D., Wilcox, W.B., Woodberry, R.D., 2000. The measure of American religion: toward improving the state of the art. Soc.

Forces 79 (1), 291–318.Stinchcombe, A.L., 1992. Simmel systematized: James S. Coleman and the social forms of purposive action in his Foundations of Social Theory. Theory Soc.

21 (2), 183–202.Stroope, S., 2011a. Education and religion: individual, congregational, and cross-level interaction effects on biblical literalism. Soc. Sci. Res. 40 (6), 1478–

1493.Stroope, S., 2011b. How culture shapes community: Bible belief, theological unity, and a sense of belonging in religious congregations. Sociol. Q. 52 (4),

568–592.Stroope, S., 2012. Social networks and religion: the role of congregational social embeddedness in religious belief and practice. Sociol. Religion 73 (3),

273–298.Tolbert, C.M., Lyson, T.A., Irwin, M.D., 1998. Local capitalism, civic engagement, and socioeconomic well-being. Soc. Forces 77 (2), 401–427.Tolbert, C.M., Irwin, M.D., Lyson, T.A., Nucci, A.R., 2002. Civic community in small-town America: how civic welfare is influenced by local capitalism and

civic engagement. Rural Sociol. 67 (1), 90–113.Tönnies, F., [1935] 1957. Community and Society (C.P. Loomis, Trans. and Ed.). Michigan State University Press, East Lansing.Vaisey, S., 2007. Structure, culture, and community: the search for belonging in 50 urban communes. Am. Sociol. Rev. 72 (6), 851–873.Vaisey, S., Lizardo, O., 2010. Can cultural worldviews influence network composition? Soc. Forces 88 (4), 1595–1618.Weber, M., [1930] 2002. The Protestant Ethic and the Spirit of Capitalism (T. Parsons, Trans.). Routledge, New York.Wolkomir, M., Futreal, M., Woodrum, E., Hoban, T., 1997. Substantive religious belief and environmentalism. Soc. Sci. Q. 78 (1), 96–108.Woodberry, R.D., Smith, C., 1998. Fundamentalism et al: conservative Protestants in America. Annu. Rev. Sociol. 24, 25–56.Woolever, C., Bruce, D., 2002. A Field Guide to US Congregations: Who’s Going Where and Why. John Knox Press, Louisville.Zablocki, B., 1980. Alienation and Charisma: A Study of Contemporary American Communes. The Free Press, New York.