Abstract
Research has demonstrated that concentrated disadvantage and other measures are strongly associated with aggregate-level rates of violence, including across racial and ethnic groups. Less studied is the impact of cultural factors, including religious contextual measures. The current study addresses several key gaps in prior literature by utilizing race/ethnic-specific arrest data from California, New York, and Texas paired with religious contextual data from the Religious Congregations and Memberships Survey (RCMS). Results suggest that, net of important controls, (1) religious contextual measures have significant crime-reducing associations with violence, (2) these associations are race/ethnic-specific, and (3) religious contextual measures moderate the criminogenic association between disadvantage and violence for Blacks. Implications for future research are discussed.
Keywords: lence, religion, race, ethnicity
Extant research exploring aggregate-level patterns of crime and violence consistently demonstrates the robustness of structural predictors – especially concentrated disadvantage – as key factors explaining differences across communities in exposure to crime (Peterson and Krivo 2005; Pratt and Cullen 2005). It is now well known that communities (as well as neighborhoods, counties, and other larger units of analysis) plagued by high levels of disadvantage are typified by higher rates of crime and violence. Yet, this now substantial body of literature also suggests that differences in structural disadvantage and other key measures across macro-level units do not completely account for differences in overall rates of crime (Pratt and Cullen 2005). Relatedly, while theoretical efforts to explain aggregate-level violent crime have emphasized concentrated disadvantage and similar macro-structural characteristics, cultural mechanisms have gone somewhat unexplored even though “cultural explanations of crime do have a long tradition in sociology” (Kirk and Papachristos 2011:1194; for exceptions see Chamlin and Cochran 1997; 2001). Taken together, there now exists the need to expand macro-level research to include cultural factors in addition to traditional structural covariates in order to more fully explicate the relationship between crime and aggregate-level context.
Notably, religion at the macro-level (or the religious context) – including the rate of religious adherence and the relative civic engagement of religious adherents – is one factor that has received some attention but which remains relatively underdeveloped within criminology (DiIulio 2009; Jang and Johnson 2010; Rosenfeld 2011). This is an unfortunate oversight because, as sociologists following Emile Durkheim have long recognized, religion is a unique institution in its emphasis on normative prescriptions for personal behavior and its capacity for informal social control through normative socialization and other mechanisms (Jang and Johnson 2010; Smith 2003; Stark and Bainbridge 1996). Moreover, that characteristics of the macro-level religious context might affect crime is consistent with a number of prominent sociological frameworks, including civic community, social disorganization, and institutional anomie perspectives of community social control. Indeed, the focus on local religious contexts coincides with the civic community perspective's view that links to strong local institutions (including religious ones and adherence to them) facilitate community social capital and social control (Tolbert et al. 1998; 2002; Lee and Thomas 2010), as well as social disorganization theory's emphasis on the necessity of both economic and non-economic local conventional institutions – like families, schools, and churches – to produce normative coherence, cohesion, and social control (see especially Stark 1987). Notably, the institutional anomie perspective in particular draws attention to the strength of non-economic institutions, including local religious organizations, in balancing economic institutions through normative control (Chamlin and Cochran 1995; Messner and Rosenfeld 1994; 2001). That is, institutional anomie identifies kinds of institutional configurations that are conducive to anomie, where economic norms and roles dominate the cultural landscape and are not counterbalanced by non-economic institutions (Messner 2012).
Compounding the relative paucity of empirical research exploring the relationship between religion and crime at the macro-level, extant research suffers from a number of prominent shortcomings, as well. First, while there has been limited empirical research examining macro-level religious contextual measures as they impact crime, much of this research has centered on differences in religious civic-engagement across geographic units (Lee 2006; Lee and Bartkowski 2004). As such, little is known about whether and how other religious contextual measures (e.g., overall religious adherence or the diversity of adherents) impact crime and violence at the aggregate-level, suggesting the need to expand research to include other dimensions of religion at the macro-level. Second, there is an absence of race/ethnic-specific analyses. Despite a significant body of literature detailing variation in religious experience and participation across racial and ethnic groups in their respective communities (Chatters, Taylor, and Lincoln 1999; Hunt and Hunt 2001; Taylor et al. 1996), little is known about how religious contextual measures are associated with crime uniquely for particular racial and ethnic groups (net of other key macro-structural characteristics). Indeed, to our knowledge, no study has yet to simultaneously incorporate religious contextual measures into race/ethnic-specific analyses of crime. This is vital because, while there are theoretical and empirical reasons to expect that aggregate-level religious prevalence may discourage violence for all groups, religious contexts may be protective to a greater or lesser extent for some race/ethnic groups than others. Third, there has been virtually no research that explores how religious contextual measures interact with other key macro-structural predictors of crime (e.g., concentrated disadvantage). For example, religious contextual measures might moderate the relationship between disadvantage and violence and do so in ways unique to particular racial and ethnic groups as suggested by a long line of research on minority communities and religious participation.
The goal of the current study is to address these shortcomings in several important ways. Using county-level racially/ethnically-disaggregated arrest data from California, New York, and Texas as well as data from the U.S. Census and the Religious Congregations and Memberships Survey (RCMS), we, first, build off of and extend prior research by exploring how religion at the macro-level, an important cultural mechanism, impacts violent crime. Specifically, we assess the extent to which three dimensions of macro-level religion – total religious adherence, civically-engaged religious adherence, and religious homogeneity – impact aggregate violence rates and ask the following: are characteristics of the religious context associated with violent crime at the macro-level? Second, whereas prior research has focused on overall (total) crime, our analysis specifically seeks to disentangle whether religion at the macro-level is associated with violent crime in unique ways for particular racial and ethnic groups. Specifically, we ask: are characteristics of the religious context associated with violence uniquely for Whites, Blacks, and Latinos? Third, we move beyond prior research by investigating the extent to which religious contextual measures differentially moderate the criminogenic effect of structural disadvantage on violence (and whether it does so uniquely for racial/ethnic groups) by asking: to what extent do religious contextual measures moderate the impact of concentrated disadvantage on violence at the macro-level and do so in race/ethnic-specific ways? That is, we examine a theme noted by Lee (2006) and others – that religious institutions convey mainstream norms and values that may be of particular importance to disadvantaged communities to the extent that the criminogenic effect of disadvantage is attenuated in religious contexts.
In the following sections we, first, review the literature on religion and crime with a focus on recent empirical examinations that have moved beyond the individual-level to examine how religion permeates the broader social landscape and overlaps with prominent macro-oriented sociological theories. Second, we examine the (still rather limited) empirical research assessing the impact of religion on crime at the macro-level. Third, we note several prominent gaps in knowledge and, subsequently, build upon our review of prior empirical research and theorizing to outline the current study's key research questions and contributions. Fourth, we present details of the current study, including the sources of data, measurement of our three key religious contextual measures, analytic strategy, and our results and robustness checks. Finally, we conclude with a discussion of our findings and their importance for theory and future research.
Religion and Crime at the Macro-Level
Sociologists since Durkheim have long recognized that religion is a core element of normative culture and a powerful potential motivator of and control on behavior at the individual, community, and societal levels (Smith 2003; see also Vaisey 2009). Religion is relatively distinct from other cultural forces because it focuses so strongly on purveying moral norms of behavior backed by sacred meanings and occasional sanctions (Smith 2003; Pargament et al. 2005). Indeed, as Christian Smith (2003) argues, religion provides individuals, groups, communities, and societies with distinctive moral narratives and cultural meanings connected to behavior. Given the uniqueness of religion and religious institutions, the study of religion seems to be enjoying a renewed vitality in sociology, psychology, and the broader social sciences (Pargament 2008; Geyer and Baumeister 2005).
Regarding the social scientific study of religion and crime, extant research has focused predominately on the individual-level. A substantial body of research suggests that religiosity inhibits youth delinquency and adult crime, though the latter is less researched than the former (see Johnson and Jang 2010). For example, church attendance among youth is inversely related to delinquent behaviors including drug use, fighting, and both violent and non-violent crime (Johnson and Jang 2010; Chitwood et al. 2008; Johnson et al. 2001; Regnerus 2006), which some have argued is achieved through the promotion of conventional social capital (Glanville et al. 2008). Religiosity also seems to insulate youth from the criminogenic influences of social disorganization (Jang and Johnson 2001).
Increasingly however, scholars have begun to extend the study of the religion-crime nexus beyond the individual level, as well. As Stark (1996: 164) urges, religion should not simply be viewed as an individual-level trait but as a “group property” that includes “persons in a given ecological setting who are actively religious.” That is, Stark and others (e.g., Smith 2003) encourage a view of religion that extends to the broader macro-level context, in which religion is an institution “comprised of regulative, normative, and cultural-cognitive elements that, together with associated activities and resources, provide stability and meaning to social life” (Scott 2008: 48).1 Indeed, across aggregate units (including larger units such as counties or states), religion is seen as a “repository of cultural resources that communities may use to their collective benefit” to the extent that places in which religion is pervasive might experience less crime because religious institutions and their adherents augment social control (Lee and Bartkowski 2004: 1007). According to Lee (2006: 310), “the central thesis is that crime rates will be lower where larger proportions of the population are actively religious...because widespread adherence to religious moral values should deter potential offenders from engaging in crime and deviance.”
Religious Contexts and Prominent Theoretical Frameworks
That religion at the macro-level should impact crime dovetails with at least three complementary sociological perspectives of social control that posit different sets of “micro instantiations of [macro-level] theorized processes” or mechanisms (Messner 2012: 15). While it is beyond the scope of the current study to fully test these specific theories, the notion that the religious context is an important cultural correlate of crime is broadly consistent with key themes from each of the following perspectives.
First, the civic community perspective depicts religion as an important local institution that shapes the social capital and social control capacity of a community. In particular, the civic community framework emphasizes religion as a source of “bridging” social capital. According to Putnam's (2000) discussion of the two types of social capital, bridging social capital is “defined by engagement with the community outside the circle of membership...aimed at improving the community as a whole and may cross ideological lines” (Ovadia and Moore 2010: 324). Adherence to religious traditions that typically exhibit greater civic engagement fosters bridging social capital in a locale, which produces and strengthens between-group ties, resulting in augmented social control within locations where such civically engaged religious traditions are more prevalent (Beyerlein and Hipp 2005; Lee 2008; Lee and Thomas 2010). As Desmond et al. (2010: 39) explain:
Communities where citizens frequently interact with each other, both formally and informally, show lower levels of crime...Frequent interaction increases interconnectedness, and in turn fosters an interest in the health of the community. Communities with stronger social networks also show a greater propensity for community action, such as neighborhood watch, and they tend to take a more active role in resolving conflict within the community.
In contrast, bonding social capital entails strong ties among in-group members that “reflects strong in-group solidarity, but does little to build community, since bonding only occurs between members of the group” (Desmond et al. 2010: 39). Religious traditions that emphasize bonding capital tend to do less outreach, which can decrease ties at the community level (Desmond et al. 2010; Ovadia and Moore 2010; Beyerlein and Hipp 2005). Bonding social capital might hinder the formation of ties across groups that foster collective efficacy, or the sense that other community members will act in the face of problems such as norm violations (Sampson, Morenoff, and Earls 1999). Bridging social capital fosters more ties and social interaction between community members, and thus may increase the social control capacity and collective efficacy of a locale (see Beyerlein and Hipp 2005). Importantly for the current study, adherence to mainline Protestant traditions and the Catholic Church is said to be a strong proxy for the prevalence of bridging social capital, and its resulting social control potential (see Beyerlein and Hipp 2005).
Second, social disorganization and systemic control models emphasize the importance of religion as an institution that augments informal social control, as well (Lee and Thomas 2010). If social disorganization erodes a community's ability to exert social control and enforce its own norms, then a strong local prevalence of religious adherence might mitigate such disorganization and both motivate and augment community informal capacity to communicate disapproval of normative violations and/or encourage conventional normative behavior. Aside from the potential for certain kinds of religious adherence to foster bridging social capital, the permeation of overall religious adherence into the macro-social environment may help to attenuate the “moral cynicism” (Stark 1987) and “legal cynicism” (Kirk and Papachristos 2011) fostered by disadvantage and social disorganization, and which is said to erode social control and foster crime and violence. That is, greater religious prevalence may help to attenuate the cynicism-producing effect of disadvantage and disorganization, which can undermine social control within geographic areas. It is important to note that, in regards to social disorganization, religious contexts might augment social control within larger aggregate units (such as cities or counties), too – that institutions can foster social control through normative cultural influence and social capital ties beyond the immediate areas, such as neighborhoods, that are often characterized as communities and extend to larger surrounding areas Indeed, as originally formulated, the social disorganization perspective was not restricted to the neighborhood level – Thomas and Znaniecki (1918) spoke of social disorganization at the societal level just as Park et al. (1925) explicitly spoke of disorganization processes at levels beyond neighborhoods, especially for cities and regions. In fact, recent empirical research utilizing disorganization frameworks has used data from larger units such as counties and census places (Light and Harris 2012; also see Steffensmeier et al. 2010 for further discussion of units of analysis and testing macro-level theories of crime).
Third, institutional anomie theory emphasizes the role of ties to non-economic institutions, including religious ones, in reducing anomic pressures by providing moral, normative balance to the acquisitive prescriptions of economic institutions (Merton 1964; Messner and Rosenfeld 2001). According to Rosenfeld (2011: 10), “a latent tendency for anomie is contained within the ethos of modern capitalism, which may be become criminogenic if not curbed by robust, counterbalancing social institutions.” In other words, given that a strong capitalistic ethos is pervasive throughout U.S. society, local variation in the strength of counterbalancing non-economic institutions, including religious ones, would be important in explaining variation in local anomie and crime rates. Chamlin and Cochran (1995: 418) note the potential importance of religious contexts in connection with institutional anomie by suggesting, “like families, religious institutions are important transmitters of values and norms that can counteract the anomic pressures produced by the economy.” That is, religion at the macro-level may work to offset the “utilitarian logic of the marketplace” and provide a moral compass that buffers residents in various geographic units from anomie and resultant crime (Messner and Rosenfeld 1997: 1396). Indeed, religion might be uniquely well suited to do so, given the moral narratives and emphasis on normative prescriptions for behavior motivated by sacred meanings discussed above. According to Messner (2012), non-economic institutions such as religion likely shape a locale's normative cultural milieu, or the context in which normative socialization occurs (which is synonymous with what Stark and Bainbridge [1996] term a “moral community” – see below), affecting not just acquisitive crimes but violent crimes as well (Messner and Rosenfeld 1997; Messner, Thome, and Rosenfeld 2008; Messner 2012; see also Currie 1997).
Importantly, according to Messner and Rosenfeld (1997; see also Rosenfeld 2011), the effects of anomie and the importance of counterbalancing non-economic institutions are not just – or even primarily – at the neighborhood level, but can extend to larger levels of scale. Drawing on Durkheim, Messner and colleagues note that institutional anomie theory posits that “egoistic or disintegrative individualism provides the cultural foundation for economic dominance in the institutional balance of power and widespread anomie” at a variety of levels of scale (Messner, Thome, and Rosenfeld 2008: 172). For our purposes, at a variety of levels of aggregation, a cultural milieu enhanced by pervasive religious adherence may help to reduce the egoistic individualism, which can lead to anomie and violence.
It is important to emphasize that, although institutions (or congregations) are important in that they are frequently the sites of organized outreach activities (Beyerlein and Hipp 2005), our focus, as with most studies (e.g., Lee and Bartkowski 2004; Beyerlein and Hipp 2005 and others), is on the population of religious adherents, and this is consistent with the theoretical frameworks described above. Institutions are not synonymous with organizations – organizations are arenas that “house” institutions, and in which institutional action is often (but not always) carried out (Rosenfeld 2011; on institutions and organizations generally, see Scott 2008). We focus on religious adherence here because it is collectivities of adherents that more frequently “carry” and enact institutions. That is, it is the level of adherence by which normative culture permeates and which affects the level of social capital (see Beyerlein and Hipp 2005). In sum, key themes from the theoretical perspectives described above are consistent with the expectation that religious contexts might mitigate crime at the macro-level as a result of the cultural milieu produced by a prominent population of religious adherents.
Similarly, we stress that the theoretical arguments above do not commit the ecological fallacy. We are not analyzing whether religious people commit less violence, or positing that would-be violent offenders must be irreligious. Rather, our argument is that the aggregate prevalence of certain kinds of religious adherents shapes the levels of social capital and social control (according to the civic community perspective), and therefore decreased the risk of crime, in a locale. In addition, collective religious adherence should shape the socialization context and the normative cultural milieu and counteract collective anomie (according to institutional anomie theory) or augment social control and decrease collective moral cynicism (as in the social disorganization vein).
Prior Research Exploring Religious Contexts and Crime
Building off of these theoretical frameworks and responding to the call of Stark and Bainbridge (1996) and others, empirical research exploring how religion at the macro-level impacts rates of crime has begun to emerge. Centered on the “moral community hypothesis” – which argues that the effect of an individual's religiosity on delinquent and criminal behavior is dependent upon the degree to which religion permeates the surrounding culture (Stark and Bainbridge 1996) – some of this research utilizes multi-level modeling procedures and observes that the preventative-effect of religion on juvenile marijuana use and binge drinking is enhanced in school environments where students are more religious (Wallace et al. 2007) and, similarly, that conservative Protestant homogeneity in the local community intensifies the degree to which individual religiosity prevents delinquency (Regenerus 2003).
Working more exclusively at the macro-level, Lee and Bartkowski (2004) draw upon the moral communities framework and the civic community perspective in their study of county-level juvenile homicide rates, emphasizing the presence of civically-engaged denominations as a particular feature of the religious context that should impact violence (see also Desmond et al. 2010). As envisioned by Stark and his associates, the concept of moral communities more broadly refers to the degree to which religion permeates the culture of a locale, and even large-scale religious contexts have been shown to shape collective and individual morality (Finke and Adamczyk 2008). Indeed, the prevalence of civically-engaged denominations, which Tolbert et al. (1998) distinguish as religious groups that are actively involved in community outreach and service programs beyond proselytizing, are found to buffer against juvenile homicide in rural counties because “horizontal social networks may be strengthened, normative consensus on acceptable and unacceptable behaviors may be elevated...and the community's ability to express and pursue collective goals may be bolstered.” (Lee and Bartkowski 2004: 1003). Other research confirms this relationship for homicide and the broader index of violent crime (Lee 2006; 2008). Thus, consistent with the frameworks outlined above, evidence now exists that macro-level religious contextual measures (specifically the presence of civically-engaged denominations) appear to solidify conventional normative order within aggregate units, enhance “bridging” social capital, and produce greater social control (Lee 2006; Lee and Thomas 2010; Beyerlein and Hipp 2005).2
Prominent Gaps in Knowledge
While empirical research on the impact of religion on crime at the macro-level has made significant headway and constitutes an important advance in bringing cultural dimensions into the macro-level study of crime, a number of prominent gaps in knowledge remain. First, extant research has largely centered on how differences in religious civic-engagement at the macro-level impact crime rates across geographic units (Lee and Bartkowski 2004; Lee 2006; 2008). While this is important, little is known about how other religious contextual measures are associated with crime, including (1) overall religious adherence and (2) the diversity of the religious context (i.e., the dominance of particular denominational groups among all religious adherents). That is, we know of no research that compares the impact of overall religious adherence, civically-engaged adherence, and the diversity of the religious population on violence rates at the aggregate-level. For example, religious homogeneity – or the relative dominance of a particular religious denomination – might be important for reducing anomie (i.e., uncertainty, normlessness) by promoting value consensus or cohesion to a greater extent than religious adherence or the civic engagement of the religious adherents in a given aggregate unit.3 Indeed, this is a key factor emphasized by both civic engagement and institutional anomie theories, and some evidence of such an effect is provided by Ulmer, Bader, and Gault (2008) who observe that county religious homogeneity (but not religious adherence or the percent of conservative Christians) is associated with more severe punishment of criminal defendants. Likewise, Ellison, Burr, and McCall (1997) find that religious homogeneity appears to reduce suicide rates to a greater extent than other characteristics of the religious context.
Second, prior research has focused on the impact of religious contextual measures on overall rates of crime with little recognition of how this relationship might be conditioned by race and ethnicity. It is well known that the nature of religious participation varies across racial and ethnic groups (Taylor et al. 1996; Chatters, Taylor, and Lincoln 1999), as does the extent to which religion permeates racial and ethnic communities more broadly (Bartkowski 2003; Hunt and Hunt 2001). Despite a significant body of literature detailing such differences, little is known about how religion at the macro-level might affect crime uniquely for particular racial and ethnic groups, especially net of other key macro-structural characteristics such as concentrated disadvantage. To our knowledge, no study has yet simultaneously incorporated characteristics of the religious context into race/ethnic-specific analyses of crime at the macro-level.
Building off of prior research and theorizing, however, there are a number of reasons to suspect that religious contexts might have a unique impact on crime for specific racial and ethnic groups. Black Protestant and Catholic churches are known to be central institutions in Black and Latino communities, respectively (Johnson et al. 2000a; Shihadeh and Winters 2010; Desmond et al. 2010). Indeed, the Moynihan Report (1965) suggests the importance of normative institutions such as African American churches as a bulwark against poverty, anomie, and social disorganization (and, by extension, the crime they often produce). Likewise, W.E.B. DuBois (1978) eloquently described African American Protestant churches as crucial normative and social organizational institutions for Black communities. As DuBois (1978: 216-217) explained, African American churches historically have provided social organization, social capital, and normative socialization that Black communities often lacked due to racism and segregation:
Entertainments, suppers, and lectures are held beside the five or six regular weekly religious services. Considerable sums of money are collected and expended here, employment is found for the idle, strangers are introduced, news is disseminated and charity distributed...Back of [the] more formal religion, the Church stands as a real conserver of morals, a strengthener of family life, and the final authority of what is Good and Right. Thus one can see in the Negro church of to-day, reproduced in microcosm, all that great world from which the Negro is cut off by color prejudice and social conditions.
In a similar vein, religious contexts might particularly affect crime among Latinos at the macro-level, as well. As Shihadeh and Winters (2010) note, because of Latinos’ deeply rooted Catholic tradition, the important role of the church in Latino community social organization and daily life, and also the recent rapid growth of evangelical Protestant adherence among Latinos, religion as a macro-level phenomenon might be particularly protective against violence in Latino communities. Their own county-level analysis of Latino homicide victimization confirms that Catholic adherence is associated with reduced homicides against Latinos, particularly in “new destination” counties. Unfortunately, we could find only two studies that assess the effects of religious contexts on race/ethnic-specific violence (see Lee et al. 2010 and Shihadeh and Winters 2010; but see Harris and Feldmeyer 2013 for more discussion of traditional and “new” destinations), and neither of these studies undertakes a comparison across Whites, Blacks, and Latinos. As such, it remains an open question as to whether and how various dimensions of religion at the macro-level impact crime uniquely for specific racial and ethnic groups (e.g., Whites, Blacks, and Latinos).
A third persistent gap in knowledge is the lack of research exploring how religion at the macro-level interacts with other robust predictors of violent crime, especially concentrated disadvantage and disorganization. For instance, it is now well established that concentrated disadvantage is strongly and positively associated with violence rates (Pratt and Cullen 2005; Peterson and Krivo 2005), but there is almost no research on the question of whether religious contextual measures might alter the effect of disadvantage. According to Lee (2006), religious institutions convey mainstream norms and values (as well as a context within which to develop social and cultural capital) that may be of particular importance to disadvantaged communities. As Lee and Ousey (2005: 35) state, “the main means by which economically disadvantaged residents can participate in, and learn the norms and values of, mainstream society is via these non-economic institutions.”
Other scholars from the institutional anomie perspective have argued that the influence of economic conditions on crime depends on the strength of non-economic institutions (Chamlin and Cochran 1995) and the pervasiveness of moral and legal cynicism (Stark 1987; Kirk and Papachristos 2011). This suggests that the local prevalence of religion might moderate the effect of disadvantage on violence. In fact, this dovetails with a key proposition of institutional anomie theory: ties to normative institutions (such as religion) are important as a balance against the competitive and acquisitive logic of economic institutions, thus dampening their anomic effects. Supporting this institutional anomie proposition, Chamlin and Cochran's (1995) state-level analysis finds that, among other indicators of the strength of non-economic institutions, state religious participation dampened the criminogenic effect of poverty.4 Similarly, Jang and Johnson (2001) find that individual religiosity moderates the effect of neighborhood disorder on drug use by augmenting the social control of youth living in disadvantaged and disorganized communities, just as Pearce et al. (2003) observe that personal religiosity reduces the criminogenic effect of exposure to neighborhood violence. Taken together, these latter studies provide suggestive evidence that individual religiosity might buffer against the effects of neighborhood disorder and disadvantage on criminal behavior, but the extent to which religious contexts at the macro-level might moderate the effects of disadvantage remains an open question.
Research Questions and Key Contributions
Our goal is to address these persistent gaps in knowledge by answering three related questions. First, we ask: are characteristics of the religious context associated with violent crime at the macro-level? We move beyond prior literature, which has largely focused on civically-engaged adherence, by assessing the potentially unique relationships between violence rates and three key dimensions of the religious landscape: total religious adherence, civically-engaged religious adherence, and homogeneity of religious adherence.
Second, we ask: are characteristics of the religious context associated with violent crime at the macro-level in different ways for Whites, Blacks, and Latinos? This is an advance beyond extant knowledge because we focus on the ways in which macro-level religious measures are associated with violence in unique ways for Whites, Blacks, and Latinos. Whereas prior research has focused on overall crime, our analysis specifically seeks to disentangle whether religion at the macro-level is associated with violent crime in unique ways for specific racial and ethnic groups. We return to this issue as it relates to the inclusion of Latinos (an often overlooked demographic group within criminology) when we discuss our sources of data.
Third, we ask, to what extent do religious contextual measures moderate the impact of concentrated disadvantage on violence at the macro-level, and do so in race/ethnic-specific ways? Specifically, we extend prior research by investigating the extent to which religion moderates the effect of structural disadvantage on violence at the macro-level and whether it does so uniquely for each racial and ethnic group.
The Current Study
Data are taken from three sources. First, information on race-disaggregated index-violence arrests is drawn from the crime reporting programs of California, New York, and Texas for the years 1999 through 2001. A key advantage of using arrest data from California, New York, and Texas is that they classify arrestees into White, Black, and Latino groups (as well as American Indian and Asian), thus allowing us to examine non-Latino White, non-Latino Black, and Latino differences in the effects of religious contexts on violence rather than focusing only on one racial/ethnic group (e.g., Blacks in southern counties) or on lethal offending (which has been the major focus of prior macro-level religion-crime scholarship). Additionally, their populations are large – together they are home to more than 38 million Whites, over 7 million Blacks, and 20 million Latinos (U.S. Census Bureau 2008) – and diverse.5 Also, arrests for violent index crimes in these states make up a sizable share (about 41%) of all arrests for violent crime in the U.S. Finally, California, New York, and Texas contain a significant number of counties in which all three race/ethnic groups (Whites, Blacks, and Latinos) reside together, making them ideal for comparing differences across groups while retaining a sufficiently large sample size (but note that we review some disadvantages of arrest data below).6 Second, 2000 US Census Summary Files 1 and 4 are used to generate information about social and economic characteristics of the White, Black, and Latino populations in California, New York, and Texas.
Third, religious contextual measures are derived from the Religious Congregations and Memberships Survey (RCMS) for the year 2000. The RCMS provides a county-by-county enumeration of religious bodies in the U.S. as tallied by the National Council of Churches and the Glenmary Research Center and sponsored by the Association of Statisticians of American Religious Bodies (Jones et al. 2002). Each participating religious body supplies the number of churches, full members, adherents (affiliated members of a congregation or those who regularly attend services), and other church population estimates, which we utilize to construct our key religious contextual measures. Relevant for our discussion of key religious contextual measures, we rely on estimates of adherents because (a) including only full members misses a large proportion of the religiously-affiliated population who might help to exert social control in the community, (b) prior research on religious contexts has almost exclusively used counts of adherents (e.g., Lee and Bartkowski 2004; Lee et al. 2010) and one of our goals is to build off of and extend this line of research, and (c) membership is determined by the by-laws of each participating congregation, which makes estimating the member population in a given unit sensitive to the particularities of congregations present to a greater degree than the measuring the adherent population (Jones et al. 2002).
Though we are unaware of any viable alternative data sources for assessing the link between the religious context and race/ethnic-specific crime across a large number of units at the macro-level, we recognize that the RCMS data are far from perfect. One limitation of the RCMS is that it likely undercounts religious adherents, particularly African Americans and adherents to traditional African American Protestant denominations (Finke and Scheitle 2005). To address this issue we utilize Finke and Scheitle's (2005) recommended revised estimates in order to account for religious adherents undercounted in the RCMS data and to limit the undercount bias among minorities in particular.7 Additionally, while the RCMS data is meant to be the most complete count of people affiliated with congregations and the most comparable count of people across all participating groups, we are unable to capture the level of participation or more nuanced meanings of adherence (a point to which we return in our discussion of directions for future research). Nevertheless, we rely on the population of religious adherents as a measure of the institutional permeation of religion in a locale because it is such collective adherence that is said to shape normative cultural milieu and foster social capital and social control (see Beyerlein and Hipp 2005).
Units of Analysis
Our units of analysis are counties, which we use for three reasons. First, they are the smallest aggregate units for which religious contextual data are available. The RCMS provides tallies of church membership for counties and larger units, thus restricting analysis to larger macro-level units. While some may see this as a disadvantage, we note that the theoretical frameworks reviewed above (civic community, social disorganization, institutional anomie, Stark's moral communities) are not restricted to neighborhoods or smaller units, but have in many instances been applied to larger units of varying sizes (e.g., counties, states, and even nations) (see for example Light and Harris 2012). Second, related to data availability, counties provide a large enough sample size to include an adequate number of covariates in our models and still retain statistical power to detect effects, particularly when making comparisons across racial and ethnic groups for rare events (like violent crime). Third, we utilize counties because prior research examining religious contexts and crime has relied on these units of analysis (e.g., Lee and Bartkowski 2004; Lee 2006; 2008; Lee et al. 2010; Lee and Thomas 2010; Shihadeh and Winters 2010). For the current study, we restrict our sample of counties to those with 1000 of each race/ethnic group resulting in a sample size of 182.8
Dependent Variables
The dependent variables in this study are county-level White, Black, and Latino violent index arrest rates (sum of arrests for homicide, aggravated assault, forcible rape, robbery) averaged across the 1999-2001 period. We average the three-year arrest figures to add stability to the rates and to ensure adequate arrest counts for statistically rare offenses (e.g., homicide). Because they have skewed distributions with some counties having particularly high rates, the rates are log transformed to normalize their distributions and account for any non-linearity in the relationships between our predictors and crime outcomes.
Arrest data are subject to many well-known criticisms, but violent index crimes are viewed as more serious than most other crimes and generally have a higher likelihood of being reported to the police and resulting in arrest than property crimes (Mosher, Miethe, and Phillips 2002). However, as we discuss in more detail below, we also ran supplemental analyses utilizing homicide arrest counts (arguably the most reliably reported offense and most likely to result in arrest) as a test of the robustness of our violent index models, which produced substantively similar results. Other scholars suggest utilizing race/ethnic-specific homicide victimization data. However, doing so would (a) limit our analysis to homicide and preclude examining violent crime more broadly and (b) necessitate assuming that homicide offenders were the same race or ethnicity as victims (on the logic that violent crime is typically intra-racial). In light of these limitations of victimization data, we rely on race/ethnic specific arrest rates.
Key Independent Variables and Controls
We focus on three unique dimensions of the religious contexts in our sample of counties. First, total religious adherence is measured as the proportion of the county's population that adheres to a religious institution recorded by the RCMS, as indicated by affiliation or regular attendance in a congregation. Second, civically-engaged religious adherence is measured as the proportion of the county's population that adheres to a religious institution recognized as being more civically-engaged than the national average according to the General Social Survey (see Tolbert et al. 1998; Lee and Bartkowski 2004).9 Third, religious homogeneity is measured as the relative diversity of the religious adherents within a county. The measure is calculated using a multi-group entropy score ranging from 0 to 1 (see Reardon and Firebaugh 2002), which captures the extent to which religious adherents are from the same religious denominational grouping (i.e., larger values suggest that religious adherents are more homogenous).10 In calculating our measure of religious homogeneity, we adapted the RELTRAD classification to define five religious groups (Catholic, Jewish, evangelical Protestant, mainline Protestant, and other) (Steensland et al. 2000; Tolbert et al. 1998). Given the multi-group nature of our religious homogeneity measure, we performed supplemental analyses below to further disentangle this religious contextual measure. With regard to our homogeneity measure, we recognize that the RCMS does not separate out Black denominations, but instead counts them among other categories (Catholic, Jewish, evangelical Protestant, etc.). As such, we are unable to explore the full range of racial diversity in religion under the traditional RELTRAD scheme. We return to this point in our conclusion.
Additionally, we include four race/ethnic specific disadvantage indicators – poverty, unemployment, education, and female headship – that have emerged as important macro-structural characteristics in criminological theory and prior empirical research. Poverty is measured as the percentage of county residents below the poverty line. Unemployment is measured as the percentage of the civilian labor force between the ages of 16 and 59 that is unemployed. Female headship is measured as the percentage of families with children under 18 years old that are headed by a female. Low education is measured as percentage of residents over 25 years old with less than a high school education (or equivalent). Because poverty, unemployment, low education, and female-headed families tend to be highly correlated in aggregate data, estimating their unique effects, and race/ethnic differences in those effects, can be problematic due to multicollinearity (correlation matrices available upon request). We address this by using principal components methods (see Land, McCall, and Cohen 1990) to extract race/ethnic-specific disadvantage index factors based on the four race/ethnic-specific concentrated disadvantage measures (i.e., one disadvantage factor represents the combined influence of poverty, unemployment, low education, and female headship for each group).11 As Jolliffe (2002: 1) summarizes, “the central idea of principal component analysis (PCA) is to reduce the dimensionality of a data set consisting of a large number of interrelated variables, while retaining as much as possible of the variation present in the data set.” In our case, we were able to combine the inter-related measures of disadvantage into separate White, Black, and Latino disadvantage factors, with race/ethnic-specific sample means of zero and standard deviations of one. This practice is now standard in the macro-structural literature (see also Land, McCall, and Cohen 1990; Steffensmeier et al. 2010).
In addition, because of their demonstrated relevance in previous studies, we include as controls: population density (residents per square mile); race/ethnic-specific residential instability (percentage of Black, White, or Latino households that experience housing turnover during the 1995-2000 period); and racial/ethnic heterogeneity as measure of population diversity.12 We do not include the proportion of the population that resides in an urban area as a result of our sample selection criteria (see endnote 9). Last, we include police per capita as a control for variations across counties in law enforcement activity.
Alternative models including (a) state dummy variables, (b) the relative size of the male crime-prone population (ages 15-25), (c) offense clearance rates as a substitute for police per capita, and (d) population structure (percent urban, population density, etc.) were estimated. The substantive results from these models were identical to what we present below (available from authors). The similarity of results when controlling for clearance rates in particular give us more confidence in the use of arrest data than we would have otherwise.
Analytic Techniques
After providing descriptive statistics, we first estimate seemingly-unrelated regression models regressing (logged) violent index offending rates disaggregated by race/ethnicity on each of the religious contextual measures and other key macro-structural characteristics. We use seemingly unrelated regression (SUR) because White, Black, and Latino measures are derived from the same set of counties, violating the ordinary least squares (OLS) assumption of independence. As a result, SUR is more appropriate for our analysis because it takes into account the correlated errors associated with shared, unmeasured predictors across groups and provides more robust standard errors for testing the equality of coefficients across groups drawn from the same sample of counties (for more detailed discussions and similar applications of SUR see Feldmeyer 2010; Ousey 1999, Steffensmeier and Haynie 2000; Schwartz 2006). SUR models provide coefficient-specific F-tests during estimation, satisfying a key goal of the study in addressing whether (and how) key religious contextual measures impact crime uniquely for specific racial and ethnic groups.
Second, we construct SUR interaction models regressing (logged) violent index offending rates disaggregated by race/ethnicity on each religious contextual measure (as a main effect), concentrated disadvantage (as a main effect), the interaction of disadvantage and each religious contextual measure (as an interaction term), and other key macro-structural characteristics. Our goal here is to assess the extent to which macro-level religion impacts the relationship between disadvantage and violent crime in race/ethnic-specific ways.
Third, we construct supplemental models to further parse out the relationships established in our primary SUR models and to assess the robustness of our key findings. In particular, we utilize an alternative outcome (homicide arrest counts) paired with negative binomial regression models, as well as disaggregated denominational homogeneity measures (Evangelical Protestant and Catholic) to disentangle how religious homogeneity impacts race/ethnic-specific violence. More details are presented in the supplemental results section.
Findings
We begin by describing the distribution of key measures in our analysis as displayed in Table 1. First, there are notable race/ethnic differences in rates of violence across race/ethnic groups with Black and Latino violence rates both substantially greater than White rates. Second, mean county-level exposure to disadvantage (and other control variables) varies significantly by race/ethnicity. With the exception of low education, Black exposure to disadvantage outpaces exposure by Latinos and particularly Whites (all differences are significant at p<.05).
Table 1.
Descriptive Statistics for Sample of Counties by Race/Ethnic Group (N=182)
| Means |
Standard Deviations |
|||||
|---|---|---|---|---|---|---|
| White | Black | Latino | White | Black | Latino | |
| Dependent Variables (Race/Ethnic-Specific): | ||||||
| Homicide Count | ||||||
| Violent Index Rate | 254.30B | 2184.72W | 1940.68 | 213.67 | 11654.83 | 17435.09 |
| Violent Index Rate (ln) | 5.28 B | 6.68W L | 5.52 W B | 0.71 | 1.14 | 1.23 |
| Key Independent Variables (Not Race/Ethnic-Specific): | ||||||
| Total Religious Adherence | 0.64 | 0.64 | 0.64 | 0.17 | 0.17 | 0.17 |
| CE Religious Adherence | 0.13 | 0.13 | 0.13 | 0.05 | 0.05 | 0.05 |
| Total Religious Homogeneity | 0.33 | 0.33 | 0.33 | 0.01 | 0.01 | 0.01 |
| Evangelical Homogeneity | 0.28 | 0.28 | 0.28 | 0.25 | 0.25 | 0.25 |
| Catholic Homogeneity | 0.21 | 0.21 | 0.21 | 0.24 | 0.24 | 0.24 |
| Race/Ethnic-Specific Control Variables: | ||||||
| Concentrated Disadv. (discrete components below):a | 0.00 | 0.00 | 0.00 | 1.00 | 1.00 | 1.00 |
| Poverty | 9.42 B L | 28.09 W L | 24.80 W B | 3.17 | 10.47 | 7.83 |
| Unemployment | 2.80 B L | 6.04 W L | 5.10 W B | 0.93 | 4.36 | 2.55 |
| Female Headship | 7.91 B L | 30.13 W L | 15.25 W B | 1.69 | 9.63 | 7.71 |
| Low Education | 15.99 B L | 31.30 W L | 51.80 W B | 5.55 | 14.06 | 12.87 |
| Residential Instability | 43.40 B L | 54.20 W | 55.57 W | 7.09 | 13.58 | 9.58 |
| Global (Not Race/Ethnic-Specific) Control Variables: | ||||||
| Race/Ethnic Heterogeneity | 0.65 | 0.65 | 0.65 | 0.21 | 0.21 | 0.21 |
| Population Density (ln) | 4.91 | 4.91 | 4.91 | 1.55 | 1.55 | 1.55 |
| Police Per Capita (ln) | 2.60 | 2.60 | 2.60 | 0.63 | 0.63 | 0.63 |
The concentrated disadvantage index combines poverty, unemployment, female headship, and low education using principal component analysis (PCA) where a single factor was extracted on which all four variables loaded. The PCA method produces race/ethnic-specific standardized scores for each unit of analysis with the sample having a mean of zero and standard deviation of one which were saved and used in our subsequent regression analyses (Jolliffee 2002; Land, McCall, and Cohen 1990; see
significantly different from whites at p<.05,
significantly different from blacks at p<.05,
significantly different from Latinos at p<.05 using Welch's t-tests for differences
Third, Table 1 illustrates significant variation across counties in religious contextual measures. On average, roughly 64% of the population of our counties identifies as a member of a religious institution, though this varies considerably (note standard deviations). For civically-engaged denominations, the average proportion is considerably smaller (13%), and varies less across counties. In contrast, we observe fairly narrow religious homogeneity (mean of .33, standard deviation of .01), indicating greater religious diversity than homogeneity in the average county and few counties are far from this mean level.
Religious Contextual Measures and Violence
Next, to explore whether religious contexts are related to violent arrests and race/ethnic-specific in these relationships, Table 2 presents SUR models regressing (logged) violent crime on each of our religious contextual variables. Model 1 presents the results for total religious adherence, Model 2 displays findings for civically-engaged religious adherence, and Model 3 presents results for religious homogeneity (all models include controls).
Table 2.
Unstandardized Coefficients from Seemingly Unrelated Regression of Logged Violent Crime Rates on (1) Total Adherence, (2) Civically Engaged Adherence, and (3) Religious Homogeneity and Other Key Measures by Race/Ethnic
|
Model 1
|
Model 2
|
Model 3
|
|||||||
|---|---|---|---|---|---|---|---|---|---|
| White | Black | Latino | White | Black | Latino | White | Black | Latino | |
| b | b | b | b | b | b | b | b | b | |
| Total Adherence | −11 191*** | −1.364** | −1.784 | - | - | - | - | - | - |
| (.336) | (.506) | (.549) | - | - | - | - | - | - | |
| Civically Engaged Adherence | - | - | - | −4.519*** B L | −1.453W | −1.260W | - | - | - |
| - | - | - | (.942) | (1.527) | (1.655) | - | - | - | |
| Religious Homogeneity | - | - | - | - | - | - | −1.424 B L | −1.762** W | −1.624* W |
| - | - | - | - | - | - | (.440) | (.651) | (.747) | |
| Concentrated Disadv. | .151** | .342*** | .211** | .122** | .278*** | .208** | .138** | .315*** | .204** |
| (.045) | (.067) | (.072) | (.043) | (.066) | (.072) | (.047) | (.063) | (.072) | |
| Race/ethnic hetero. | .859** | −.943* | .920† | .672** | −1.119** | .826† | .514* | −1.399*** | .581 |
| (.262) | (.389) | (.473) | (.244) | (.387) | (.466) | (.261) | (.392) | (.474) | |
| Residential Instab. | −.006 | −.014** | .006 | −.007 | −.009† | .008 | .004 | −.013** | .007 |
| (.007) | (.005) | (.008) | (.006) | (.005) | (.008) | (.006) | (.005) | (.008) | |
| Pop. Density (ln) | −.003 | .252*** | −.009 | −.010 | .240*** | −.009 | −.031 | .171** | −.078 |
| (.036) | (.054) | (.060) | (.035) | (.055) | (.061) | (.041) | (.060) | (.068) | |
| Police per capita (ln) | −.072 | −.244† | −.214 | −.091 | −.319* | −.247 | −.169* | −.291* | −.230 |
| (.094) | (.135) | (.158) | (.088) | (.134) | (.157) | (.093) | (.131) | (.153) | |
| Constant | 5.937*** | 8.315*** | 5.679*** | 5.979*** | 7.626*** | 5.382*** | 5.490*** | 8.782*** | 6.278*** |
| (.431) | (.688) | (.837) | (.392) | (.676) | (.820) | (.477) | (.765) | (.926) | |
| R2 | .157 | .223 | .099 | .194 | .201 | .091 | .103 | .227 | .112 |
| Breusch-Pagan | X2 = 194.136, p<.001 | X2 = 209.027, p<.001 | X2 = 194.852, p<.001 | ||||||
Note: Standard errors in parentheses
significantly different from Whites,
significantly different from Blacks,
significantly different from Latinos at p<.05 using coefficient-specific F-test:
p<.10,
p<.05,
p<.01,
p<.001
First, total religious adherence is negatively related to violent crime for Whites and Blacks, net of other key measures. That is, the greater the proportion of a county's population that is religious, the lower the violent crime rate for Whites and Blacks (we note also that the effect for Latinos is in the expected direction, though not significant at p<.05). However, there are no significant differences across Whites, Blacks, and Latinos in the relationship between total religious adherence and violence. F-tests for differences were all non-significant (p>.10, two tailed), suggesting that religious adherence has roughly equivalent associations with violence across race/ethnic groups.13
Second, civically-engaged religious adherence has a statistically significant, negative association with White violent crime (p<.001), but not Black or Latino violent crime. As we noted, prior research has focused in particular on the impact of civically-engaged religious adherence on violent crime (Lee and Bartkowski 2004). Consistent with the general pattern in extant literature, the greater the proportion of a county's population belonging to a civically-engaged denomination, the lower the White (but not Black or Latino) violent arrest rate. Moreover, the relationship between civically-engaged religious adherence and violence is significantly stronger for Whites than for Blacks or Latinos (p<.05). That is, civically-engaged religious adherence differs across race/ethnic groups in its association with violent arrests.
Third, religious homogeneity is associated with statistically significant reductions in violence for Blacks (p<.05) and Latinos (p<.01), but not Whites. Put another way, Black and Latino violence is lower in counties where adherents belong to similar types of religious institutions. Importantly, the impact of religious homogeneity on violence is significantly stronger for Blacks and Latinos than for Whites (p<.05); however, the difference between Blacks and Latinos is non-significant, suggesting that religious homogeneity has roughly equivalent relationships with violence for these two groups.
Fourth, other key macro-structural characteristics, particularly concentrated disadvantage, have the expected criminogenic (positive) association with violence for all three racial and ethnic groups. That is, a greater confluence of poverty, unemployment, female headship, and low education in a county is associated with increased violent crime rates for Whites, Blacks, and Latinos. Likewise, racial/ethnic heterogeneity (or the diversity of a county's population) also has important criminogenic effects for Whites in all three models, but appears to reduce Black violence net of religious contextual characteristics and other key controls.
Overall, Table 2 suggests that (1) religious contexts are associated with lower rates of violence for Whites and Blacks, and (to a lesser extent) Latinos, net of other key macro-structural predictors of violence, but (2) there are important differences across racial and ethnic groups in the impact of religious contextual measures on violent crime. More simply, religion as a macro-level phenomenon relates to violence in important ways, though these relationships are contingent upon both specific dimensions of the religious context and the race/ethnic group in question. Even where we fail to observe significant associations between our religious contextual measures and violence, these relationships are all in the expected direction (i.e., violence reducing) for each racial and ethnic groups.
Interaction of Religious Contextual Measures and Concentrated Disadvantage
We turn now to our third key question: whether the relationship between concentrated disadvantage and violence is conditioned by macro-level religious contextual measures and in ways that are race/ethnic-specific. More simply, we ask whether the criminogenic effect of concentrated disadvantage on violence is attenuated in communities where religion is more prominent or homogenous and whether it does so in ways that are unique to Whites, Blacks, and Latinos. Table 3 presents the coefficients for the interaction terms for models interacting concentrated disadvantage and total religious adherence (Model 1), civically-engaged religious adherence (Model 2), and religious homogeneity (Model 3). For parsimony, we present only the coefficients for the interaction terms though a full set of controls and mainline effects are also included (see footnote in Table 3; available from authors upon request).
Table 3.
Unstandardized Coefficients from Interactions Between Concentrated Disadvantage and Key Religious Contextual Variables on Logged Violent Crime (N=182)
| Model 1 |
Model 2 |
Model 3 |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| White | Black | Latino | White | Black | Latino | White | Black | Latino | |
| b | b | b | b | b | b | b | b | b | |
| Disadvantage * Total Adherence | .356† | −.715* | .841* | ||||||
| (.196) | (.338) | (.396) | - | - | - | - | - | - | |
| Disadvantage * Civically Engaged Adherence | - | - | - | .328 | −.801 | 1.975 | - | - | - |
| - | - | - | (.638) | (1.203) | (1.317) | - | - | - | |
| Disadvantage * Total Religious Homogeneity | - | - | - | - | - | - | −.373 | −1.273** | .504 |
| - | - | - | - | - | - | (.271) | (.375) | (.470) | |
| R2 | .156 | .238 | .091 | .190 | .197 | .065 | .110 | .250 | .133 |
| Breusch-Pagan | X2 = 199.670, p<.001 | X2 = 192.750, p<.001 | X2 = 188.355, p<.001 | ||||||
Note: Each model includes mainline effects for both interacted variables and a full set of controls; standard errors in parentheses
p<.10,
p<.05,
p<.01,
***p<.001
Our key findings are, first, that at least two religious contextual measures – total religious adherence and religious homogeneity – appear to moderate the relationship between concentrated disadvantage and violence for Blacks. The impact of disadvantage on Black violence is significantly reduced in counties with greater levels of religious adherence (Model 1), suggesting that disadvantage is less strongly related to Black violence when a greater proportion of a county's population belongs to a religious institution. Similarly, in counties with greater religious homogeneity, the relationship between Black disadvantage and violence is reduced significantly. In contrast, the interaction between disadvantage and religious contexts is non-significant (or marginally significant in the case of total adherence and disadvantage on White violence), suggesting that the moderation by religious contextual measures of the disadvantage-violence relationship is primarily a Black phenomenon.14
To further illustrate the conditioning effect of religious contexts on the disadvantage-violence relationship for Blacks, we plot predicted (logged) Black violent crime rates using the interaction models presented in Table 3 (see endnotes in Figure 1 for details). We predict values for low (two standard deviations below the mean) and high (two standard deviations above the mean) levels of disadvantage and our key religious contextual variables, while holding all other control measures constant at their means.
Figure 1.
Predicted Logged Black Violent Index Rates for Models of Interaction Between Disadvantage and (A) Total Adherence and (B) Religious Homogeneity
Figure 1 displays the statistically significant interactions between total adherence and disadvantage (panel A) and religious homogeneity and disadvantage (panel B) in predicting Black violence. We note that the slope for the predicted violence rates by level of disadvantage at low levels of religious adherence is steeper than the slope for predicted violence rates by level of disadvantage at high levels of religious adherence. In sum, disadvantage has a weaker relationship with Black violence in counties with greater overall adherence rates and in counties with greater religious homogeneity.
Assessing the Robustness of Our Key Findings
To reiterate, the results from our primary violent index models displayed in Tables 1-3 and Figure 1 indicate that (1) religion at the macro-level (the religious context) has important relationships with violence, (2) these relationships are often conditioned by race/ethnicity, and (3) elements of the religious context moderate the criminogenic association between disadvantage and violence for Blacks (but not Whites or Latinos).
However, several issues warrant consideration. First, we recognize that our measure of religious homogeneity may mask important denominational differences. That is, our homogeneity measure captures the overall breadth of religious representation (or lack of religious diversity) but does not reveal which religious denominational groups (e.g., Evangelical Protestants, Catholics) dominate the religious affiliations of county adherents. While it is beyond the scope of this study to fully resolve this issue and is an important direction for future research, our goal here is to further parse out the relative dominance of specific denominations amongst adherents in order to clarify our key finding from Table 2 that overall religious homogeneity is associated with reduced Black and Latino violence.
Second, arrest data are subject to well-known criticisms, including potential bias from under-reporting and differences across jurisdictions in the use of police discretion in charging (Mosher et al. 2010). Given the breadth of the violent index and the inclusion of offenses such as aggravated assault – which are more likely to be influenced by under-reporting and enforcement practices – it is possible that our results may be unduly influenced by our primary outcome variable. As such, we assess whether religious contexts yield similar violence-reducing associations with the more reliably reported violent crime of homicide, which is arguably the most reliably reported offense. Due to the rare nature of homicide (particularly in smaller counties), our supplemental analysis of homicide necessitates the use of negative binomial (count-based) models that account for the over-dispersion of the data rather than the SUR models used in our main analysis (Osgood 2000).
Table 4 presents the results of our supplemental analyses. Model 1 displays results utilizing SUR procedures to regress violent index arrests (logged) on Evangelical Protestant homogeneity and controls, while Model 2 displays results for similar models with Catholic homogeneity plus controls. Models 3-5 display supplemental results using negative binomial modeling procedures regressing homicide arrest counts on our three primary religious contextual measures plus controls.
Table 4.
Unstandardized Coefficients from Robustness Checks Examining Denominational Effects and Homicide as an Alternative Dependent Variable (N=182)
| White | Black | Latino | |
|---|---|---|---|
| b | b | b | |
| Parsing out religious homogeneity (denominational effects): | |||
| Model 1: Logged Violent Index Rates on Evangelical Protestant Homogeneity | −.796* | −.446 | −.606 |
| (.320) | (.428) | (.484) | |
| Model 2: Logged Violent Index Rates on Catholic Homogeneity | −.385† | −.665† | −1.056** |
| (.233) | (.356) | (.404) | |
| Alternative dependent variable (homicide arrest counts): | |||
| Model 3: Homicide on Total Adherence | −1.162† | −.431 | −2.275* |
| (.645) | (.908) | (1.103) | |
| Model 4: Homicide on Civically Engaged Adherence | −5.580** | −2.628 | −7.199** |
| (1.781) | (2.368) | (2.455) | |
| Model 5: Homicide on Religious Homogeneity | −.170 | .366 | .665 |
| (.757) | (1.270) | (1.677) |
Note: Each model includes a full set of controls (and mainline effects for both interacted variable in models 6-8); standard errors in parentheses
Wsignificantly different from whites at p<.05,
Bsignificantly different from blacks at p<.05,
Lsignificantly different from Latinos at p<.05 using coefficient-specific F-tests
p<.10,
p<.05,
p<.01,
***p<.001
We note two key findings. First, greater Evangelical Protestant religious homogeneity (i.e., when Evangelical Protestants comprise a greater proportion of the adherent population relative to other religious groups) is associated with reduced White violence, but not Black or Latino violence. Catholic religious homogeneity is associated with reduced Latino violence, but not Black or White violence (though the association with Black violence is marginally significant). In sum, these results help to clarify our findings from Table 2 and suggest that there are indeed denomination-specific relationships for particular racial and ethnic groups. Specifically, when religious uniformity favors Evangelical Protestants or Catholics, there are associated reductions in White and Latino violence, respectively. While religious homogeneity has important crime-reducing associations that are race/ethnic-specific on the whole (as displayed in Table 2), these relationships appear to be driven by the predominance of specific religious groups within the adherent population.
Second, findings for our homicide models generally parallel our violent index findings. Specifically, total religious adherence is negatively associated with homicide arrests for Latinos and (to a lesser extent) Whites, net of other controls. Likewise, civically-engaged religious adherence appears to reduce both White and Latino homicide, net of controls, while religious homogeneity has no statistically significant effects on homicide for any racial/ethnic group. The latter is unsurprising given the rare nature of homicide and the small sample size of the current study (N=182) in that the data offer such limited variation in the dependent variable and highly-skewed homicide distributions that the estimation of models may be problematic even when using negative binomial procedures. In light of these concerns, we acknowledge the null findings from the homicide models of homogeneity and view them as an important supplement to our analysis, but we place greater emphasis on the violent index findings. In either case, religious contextual measures appear to have important, race/ethnic-specific relationships with homicide as in our violent index models presented above.
CONCLUSIONS
We investigated whether religious contextual variables predicted county-level violent crime and whether they did so in race/ethnic-specific ways. Moreover, we examined whether and how local religious contexts moderated the criminogenic effects of race- and ethnic-specific measures of concentrated disadvantage. We found, first, that county-level religious adherence was associated with reduced White and Black violence (in fact, the size of the association rivaled the relationships for concentrated disadvantage in some cases), while civically-engaged religious adherence was associated with reduced White violence (but not Black or Latino) violence. Religious homogeneity, a relatively untapped dimension of the religious context, was associated with reduced Black and Latino (but not White) violent crime. Clearly, religious contexts have important relationships with violence in important ways, net of structural disadvantage and other important predictors of violence, and in turn point to the role of normative-cultural factors in buffering communities from violence. Moreover, the fact that macro-level religious characteristics differentially impacted Black, White, and Latino violence suggested that relationships between dimensions of the local religious context and violence at the macro-level are conditioned by race and ethnicity. This is unsurprising in light of the well-known tendency for U.S. churches to be racially and ethnically homogeneous (Lundskow 2008; Scheitle and Dougherty 2010).
That civically-engaged religious adherence was negatively associated with White (but not Black or Latino) violence would seem to coincide with Lee's (2008) finding that the presence of civically-engaged denominations reduced rural (and likely predominately White) but not urban (and more racially/ethnically mixed) violence. Since race- and ethnic-specific religious adherence data are not available, we cannot definitively address this issue. However, the current study makes a significant contribution to the extant literature by exploring the broader impact of religious contexts on race/ethnic-specific violent crime. Indeed, an important first step is to identify the larger pattern of effects upon which future research may build. As such, we note that the adherents of civically-engaged denominations in our data are predominantly White. Of the twelve denominations deemed “civically-engaged,” only one (African Methodist Episcopal) is known to have a predominantly Black membership (but see endnotes 8 and 10). If these civically-engaged denominations have predominantly White membership, our findings would indicate that such groups primarily enhance bridging social capital and cohesion intra-racially, but do not provide notable violence reduction benefits among blacks and Latinos. Alternatively, civically-engaged denominations may foster more social capital and cohesion among Whites to a greater extent than other predominantly White denominations that are not civically engaged. In contrast, as suggested by Ellison and Sherkat (1995) and Sherkat (2002), choices of religious affiliation may be “semi-involuntary” for Blacks (and Latinos). That is, minorities are more likely to be constrained by racial and ethnic social pressure in their choice of religious affiliation, resulting in under-representation in civically-engaged denominations as defined in our data. Rather Black and Latino religious adherents may belong to predominantly Black or Latino churches that might foster as much social capital as other denominations but which are not themselves defined as civically engaged. In this case, our civically engaged denomination measure would not differentiate Black (and perhaps Latino) denominational adherence that was as rich in social capital as those primarily White denominations identified in our data. This reasoning is consistent with our finding that overall local religious adherence reduced Black violence, but civically-engaged adherence does not.
Likewise, the pattern for religious homogeneity implies that religious normative consistency and consensus might also be an important dimension of communities (Ulmer, Bader, and Gault 2008) associated with reduced crime, and that it may specifically impact some groups (e.g., minorities) more than others. Why might this be the case? Again, we lack the data to definitively respond to this question, but the answer might lie with the possible race/ethnic-specific effects of religious pluralism, along with the tradition-specific nature of homogeneity (as suggested by our supplemental analyses). Local religious heterogeneity may increase religious participation for Whites (Finke and Stark 2005), producing beneficial social control effects in terms of White violence. For minorities, local religious homogeneity may provide a more unified community with greater informal social control and enhanced cultural cohesion and value consensus, particularly for African Americans and Latinos.
Indeed, recall that greater Evangelical Protestant religious homogeneity was associated with reduced White violence, but not Black or Latino violence, while Catholic religious homogeneity was associated with reduced Latino (but not Black or White) violence – though the association with Black violence was marginally significant. Specifically, when religious uniformity favors Evangelical Protestants or Catholics, there are associated reductions in White and Latino violence, respectively. This findings may appear at odds with some studies that show a positive association with Evangelical adherence and crime in some circumstances (Ellison et al. 2003; Lee et al. 2010; Shihadeh and Winters 2010), though these studies were focused on homicide (and sometimes homicide victimization), overall crime rates (Beyerlein and Hipp 2005), and the link to adherence rather than religious diversity. Beyerlein and Hipp (2005) and others argued Evangelical Protestant denominations mostly foster bonding social capital (see also Stroope 2011) that leads to insularity and comparatively less social control for the larger community. However, Evangelical denominational dominance may offset some of this insularity and help to counteract the development of anomie, may decrease moral and legal cynicism, and may increase social control because group insularity means less in highly homogenous religious contexts (that is, in places where there are fewer non-Evangelical Protestants to civically engage). Our findings suggest that this may be especially true for White and Latino violence. The finding of a negative effect of Evangelical adherence on Latino violence is particularly intriguing and worthy of future research, and more generally points to the need for additional scholarship that parses out the Evangelical Protestant category that prior research has established as being quite diverse (e.g., Blanchard et al. 2008). In addition, with better data on traditional Black denominations, one might find a stronger effect of traditional Black denomination adherence on Black violence (see below). In sum, future research should examine how religious homogeneity is associated with reduced violence in race/ethnic-specific ways, and how these relationships might be affected by the predominance of specific religious groups within the adherent population (see our discussion below regarding related directions for future research).
Another key finding is that religious contexts appeared to moderate the disadvantage-violence relationship for Blacks but not Whites or Latinos. That is, the disadvantage-violence relationship was weaker for Blacks when total religious adherence was greater and when counties were more religiously homogenous. This is an important finding and supports Johnson et al.'s (2000a; 2000b) contention that religiosity plays a protective role against crime in disadvantaged African Americans communities especially. The findings also suggested that institutional anomie theory's argument that non-economic institutions are important in buffering the effects of economic factors, particularly socioeconomic disadvantage, may be particularly germane for Black violence. Dimensions of the local religious context appeared to provide normative guidance and social control as a counterbalance to the anomie-fostering effects of economic deprivation consistent with the Moynihan Report's (1965) emphasis on cultural institutions, such as family and church, as buffers against poverty and social disorganization in Black communities.
Beyond this point, our findings have broader implications for institutional anomie theory. Most generally, our findings are consistent with the notion that ties to religious institutions might shape local normative culture, counteracting anomie, and perhaps also the kind of moral cynicism Stark (1987) described. Messner (2012) speculates that such a collective normative context, or in Stark's terms a moral community, might even shape individual decision making in a manner similar to that described in situational action theory (Wikstrom 2010), in that the collective moral socialization context shapes individual moral options and choices, including violent ones (see also Messner, Thome, and Rosenfeld 2008 and endnote 2). According to Messner (2012:12, emphasis added): “[Situational action theory] recognizes the critical role of social institutions such as the family, the schools, and religion in delivering the kinds of socialization that promote prosocial behavior and inhibit crime.” That is, there appear to be micro-macro linkages by which non-economic normative institutions, such as religion, foster individual-level pro-social socialization and behavior, perhaps by rendering violence morally “unthinkable” in the situational choices of individuals, in the manner described by situational action theory (Wikstrom 2010; for fuller discussions of such micro-macro linkages, see Messner 2012; Rosenfeld 2011).
Directions for Future Research
Building on the logic of examining micro-macro linkages, we note several important limitations of the current study that suggest related directions for future research. First, in addition to exploring the specific micro-level (e.g., social psychological) processes by which religious institutions and their adherents institute social control, greater concern for specific denominational adherence combined with religious homogeneity may be a particularly fruitful avenue for future scholarship. Our findings suggested that religious homogeneity was an important dimension of the religious landscape at the macro-level (particularly for minorities), but it was beyond the scope of the current study to examine whether belonging to specific denominations impacts violence for individuals living within locales of varying religious homogeneity. That is, research is needed which asks which denominations impact crime for individual adherents more or less strongly in homogeneous areas (and for particular race/ethnic groups). Answering this question would perhaps require a multi-level modeling framework that can incorporate individual and contextual level interactions. This would help bring together the growing macro-level literature on religion and crime with extant research focusing on the religiosity of individuals.
As a second concern, we recognize that arrest data are not perfect, and that there are well-known criticisms of them as a source of crime information (see Mosher, Miethe, and Hart 2010), a point which future research can rectify by replicating our analysis using alternative data sources (e.g., victimization or self-report surveys). Likewise, our analysis is limited to the predictors available from U.S. Census data and to non-race/ethnic-specific religious contextual measures. Unfortunately, regarding the latter, we know of no such data in existence, as least for a relatively large number of aggregate units.
Third, and related to the point above, we recognize that it would be beneficial to have religious adherence measures, including more nuanced measures that reflect levels of participation and personal religiosity, at other units of analysis, such as cities, census places, or even neighborhoods. This is one (of several) notable limitation of the RCMS data, but at present such data are not available across a large number of units. For example, while Desmond et al. (2010) looked at congregational data at the neighborhood-level, their analysis was limited to a single city. Some critics may suggest that counties are too large to accurately capture the permeation of religion into communities, though we note that Land et al. (1990) argue that macro-level effects on crime should be apparent at varying levels of analysis (whether it be neighborhoods, cities, counties, or states). Likewise, it is also worth noting that religious participation is far from a purely neighborhood-level phenomenon. Religious individuals do not necessarily (in fact, increasingly do not) attend church and interact with their fellow congregants in their own neighborhoods, but are often “commuter congregants” who travel outside their neighborhoods or perhaps even municipalities for religious participation (Desmond et al 2010).
Recall also that the RCMS data tends to undercount traditional African American denominational adherence in particular. In spite of our use of revised estimates for addressing such minority undercount, there still may be some minority undercount bias for our civically-engaged adherence measure (see also endnote 8). For example, traditionally African American denomination adherents might be undercounted, and these denominations might exhibit comparable civic engagement to other groups in our civically engaged adherence measure. This potential for undercount bias likely means our results are a conservative estimate of civically engaged adherence on black or Latino violence. That is, if our civically engaged adherence measure leaves out adherents to traditionally African American Protestant churches who might be strongly civically engaged (or leaves out Latino adherents to any predominantly Latino Protestant churches that might exhibit strong civic engagement) this would mean our results for the effect of civically engaged denominations on Black and Latino violence are understated. Thus, it may be that more complete, inclusive measures of adherence would show a stronger preventative association with Black or Latino violence.
Future research should continue to examine the effects of religion on overall and racially-and ethnically-specific violence and other types of crime. In particular, given the importance of Black churches noted in the sociology of religion literature, research should explore how traditionally African American denominational adherence affects Black violence, especially in predominantly Black locales. Such traditionally Black denominations (e.g., African Methodist Episcopal, African Methodist Episcopal Zion, Church of God in Christ, National Baptist Convention USA, and others), tend to be strict in their prescriptions regarding social behavior, much like White evangelicals, and they tend to engage is considerable outreach and service in Black communities. As Johnson et al. (2000a: 480) note, “numerous scholars agree that the African American church has been an important agency of social control and organization among Black Americans.” Black churches may be especially important in preventing violence and moderating disadvantage, “given the higher levels of religious involvement and the symbolic centrality that religious institutions, especially Black churches, occupy within African American communities” (Jang and Johnson 2005: 326-327).
Few would dispute the important role of concentrated disadvantage in producing heightened violence rates. However, we hope this study will stimulate further research into the role of local cultural/normative-contextual factors, particularly dimensions of the local religious context, in explaining race-ethnic specific violent crime and in conditioning the effects of structural factors on violence.
Acknowledgments
This research made is possible by National Science Foundation Grant SES-0719648
Footnotes
In our view, the religious context includes religious adherents, their beliefs and prescriptions for behavior, as well as the institutions to which they belong. Regarding the latter, Scott (2008: 48) defines institutions as “multifaceted, durable social structures made up of symbolic elements, social activities, and material resources” that include norms and behavioral regulation (social control) as central ingredients. As we elaborate throughout, our empirical examination focuses on the population of adherents and the relative diversity of the adherent population that is consistent with much prior empirical research (e.g., Lee and Bartkowski 2004; Lee et al. 2010) and has been shown to impact crime net of institutional/denominational structural measures (e.g., Beyerlein and Hipp 2005). Moreover, as we also detail, the theoretical frameworks utilized throughout the current study are not inconsistent with adherent-centered measures of the religious context in that adherents are those who act out the prescriptions of religious institutions on a day-to-day basis (or rather, put the words of institutions into action) and are, therefore, central to the link between religious contexts and crime. Indeed, even Stark's (1996: 164) statement encourages social scientists to focus on “the proportion of persons in a given ecological setting who are actively religious” (see also Lee and Bartkowski 2004).
While much prior research and theorizing suggests that more religious contexts may be associated with reduced rates of crime and violence, some scholars suggest that this may vary by denomination (e.g., Evangelical Protestant) and context (Ellison et al. 2003; Shihadeh and Winters 2010; Desmond et al. 2010; Lee, Thomas, and Ousey 2010). For instance, Ellison et al. (2003) find that evangelical Protestant adherence is associated with greater violence in Southern (but not Northern) cities, while Beyerlein and Hipp (2005) find that evangelical adherence was positively associated with higher overall crime rates. These authors argue that while mainline and civically-engaged adherent populations, along with perhaps Catholic adherents, foster “bridging” social capital, evangelical Protestant adherents mostly foster “bonding” social capital that leads to insularity and comparatively less social control for the larger community. Similarly, Lee et al. (2010) find that evangelical Protestant adherence was associated with more argument homicides for Whites and Blacks in both the North and South, though the effects were smaller for Blacks. While it is beyond the scope of the current study to do so, such scholarship points to the need for more empirical research examining specific types of denominations in relation to violent crime, since there are good reasons to suspect that not all have the same effects on crime (as well as differing effects across racial and ethnic groups). We return to this point in our supplemental analyses and in discussing directions for future research that might include disentangling how various denominational groups foster violence in specific contexts.
Research on religious pluralism, particularly its impact on religious participation, is a growing body of literature in the sociology of religion (see Finke and Stark 2005). Yet, our review reveals virtually no research on the link between religious pluralism/homogeneity and social problems, such as crime or violence (for one exception see Ellison, Burr, and McCall 1997 who examined religious homogeneity as it impacted suicide at the macro-level).
On the other hand, some scholars have argued that religious contexts should partially mediate the relationship between disadvantage and crime (Maume and Lee 2003). A full treatment of whether religious contexts mediate the effect of disadvantage on violence is beyond the scope of this paper, but supplemental analyses (available from authors) find no significant mediation of the disadvantage-violence relationship through religious contexts.
For the year 2000, the overall racial/ethnic composition for the United States was roughly 69% white, 12% black, and 13% Latino. The racial/ethnic composition of the combination of California, New York, and Texas is comparable at roughly 52% white, 10% black, and 28% Latino.
One limitation of our data is that we are unable to capture communities with the largest concentration of Black residents (e.g., the southern “Black Belt”) and, thus, may be unable to fully examine the link between the religious context and violence for Blacks. However, the central goal of the current study is to compare the impact of religious contextual measures on broader violence (versus lethal crime only) and across three racial/ethnic groups (Whites, Blacks, and Latinos). Restricting our sample to counties in the South with larger Black populations would not provide enough Latinos for comparison and, moreover, there are significant problems with the race/ethnic coding of the data used in much prior research (e.g., UCR) that provide information on criminal offending in the South (as well as many other areas of the United States) (for a review, see Steffensmeier et al. 2011). As such, our California, New York, and Texas data is instrumental in (a) broadening the criminal landscape beyond homicide, while (b) allowing for a comparison across Whites, Blacks, and Latinos without confounding these racial/ethnic groups and (c) still capturing sizable populations of all three race/ethnic groups (which is necessary to compare across each). In short, our data may not be ideal for the analysis of any one racial/ethnic group (a point which we note is a direction for future research), but are perhaps the most valuable source of crime data for addressing the shortcomings of prior research we have identified and for answering our specific research questions.
Even utilizing the revised estimates, these data are known to undercount Black Baptists and Latino and nondenominational congregations in the Latino Pentecostal realm that are not part of the Glenmary process (see for example Finke and Scheitle 2005). Some attempts have been made to correct these problems, but minorities in marginal denominations or those which are non-denominational often get overlooked. Again, we know of no comparable alternative sources of data, and while we are not focusing specifically on denominational effects, it is important to keep in mind that many religious adherents (particularly minorities) may still be missing despite efforts to capture them. We thank a reviewer for drawing our attention to this cautionary issue and note that the undercount of certain minority and small denominational members may actually work to conservatively bias our results – by missing certain minority and smaller denominational adherents, the relationship between total adherence, civically-engaged adherence, and religious homogeneity may more strongly reduce violence for Blacks and Latinos than we observe here.
An analysis of influential cases suggests that our results are not sensitive to the sample selection criteria. Cook's D values are all well below 1.0 (highest is .421) indicating no consistently influential cases (Agresti and Finlay1997). We also note here that, as a result of our selection criteria, our sample of counties is overwhelmingly urban. As such, we do not control for percent urban (or other similarly invariant macro-structural traits associated with an almost exclusively urban sample) as there is so little variation in this measure. Because our sample selection criteria eliminates many smaller, rural counties from the analysis, we utilized 3-year averaged violence rates rather than 5-year (or longer) averages because (1) diagnostic analyses revealed no problematic outliers or influential cases in our sample (and, hence, no counties whose size appeared to skew the crime data) and (2) the authors’ examination of violence rates using additional years of data (5 years) revealed virtually identical rates to those employed in the current study. Moreover, much prior macro-structural criminological research has relied upon 3-year averaged rates to the point that this has now become somewhat standard as long as reasonable population criteria are imposed (see for example Steffensmeier et al. 2010). Regarding the temporal period covered (1999-2001), we recognize that more recent data would be ideal; however, crime data are often delayed for several years before release to academics and the general public and, as such, we utilized data from the most recent time point at this writing for which religion and crime data were available for all three states.
We follow prior research in measuring the civically engaged religious population. However, we also recognize that civic engagement measured nationally (via the General Social Survey) may miss important regional heterogeneity within a denomination. That is, denominations deemed to be civically engaged may not be so in all places, just as non-civically engaged denominations in some communities may indeed exhibit significant community outreach. Unfortunately, capturing such heterogeneity (to our knowledge) is impossible with the current religious data and the extent of such patterns (and subsequently, their implications for the findings presented here and in prior research) remains unsettled. Nevertheless, we utilize this measure as (1) it has been argued to be the central way that religious adherents and their institutions leverage social capital and impact crime in communities (as argued by the civil society literature – see Lee and Bartkowski 2004) and (2) it has been the primary focal variable in prior research against which we are able to compare findings. We also emphasize here that we cannot control for denominations and adherent populations simultaneously because of collinearity. Diagnostics reveal that because our models are race/ethnic specific (with small crime counts in some instances) we cannot control for strongly correlated measures like congregational presence (i.e., per capita prevalence of specific congregations) and rates of adherence in the same models. Moreover, we reiterate that it is not our goal here to disentangle the impact of congregations relative to adherents (see for example Beyerlein and Hipp 2005), but rather demonstrate the salience of religious contextual measures (here, at the adherent-level) across racial and ethnic groups on rates of violence.
More specifically, for each race/ethnic group we ran principal component analyses for poverty, unemployment, female headship, and low education with factors retained at a cutoff Eigenvalue of 1. In all cases, poverty, unemployment, and female headship loaded on a single factor. This process, in sum, yielded three new factor score variables (one for each race/ethnic group) where each set of scores had a standardized sample mean of zero and standard deviation of one, which we then saved and used in our subsequent regression models as our indexes of concentrated disadvantage (see for example Land, McCall, and Cohen 1990). After creating our three disadvantage indexes, all variables in our models are correlated at .41 or below and all variance inflation factors (VIFs) are below 2.0 for all variables in each model (well below the threshold of 10 that is frequently used – see Kutner, Nachtsheim, and Neter 2004), indicating that there is little evidence of problematic multicollinearity in our final models.
We also ran supplemental models examining the effects of the separate RELTRAD groupings (Steensland et al. 2000) on race/ethnic-specific violence. For White violence, the results for Mainline Protestant and Jewish adherence paralleled the strong, negative effects of civically engaged denomination adherence in Table 3. For Black violence, evangelical Protestant adherence and Catholic adherence both showed marginally significant negative effects. For Latino violence, mainline and evangelical Protestant adherence exhibited negative but marginally significant effects.
We note that the R-squared values observed in Table 3 exhibit small changes from those observed in Table 2. Though it is beyond the scope of the current study to fully expound on this issue, we emphasize, first, that in most cases the changes in the R-squared values are trivial. Second, we are hesitant to place too much emphasis on the R-squared values and are cautious in drawing conclusions from these statistics given that some scholars (e.g., McElroy 1977) question their validity in SUR models and treat them primarily as goodness-of-fit measures that depend on the correlation of errors across equations. Third, and related to our second point, adding additional variables (like interaction terms) in SUR models will necessarily yield a new error variance-covariance matrix with correspondingly unique explained variance values that are not comparable to other other SUR models with their own unique matrix.
Contributor Information
Jeffery T. Ulmer, The Pennsylvania State University Department of Sociology and Crime, Law, and Justice 211 Oswald Tower University Park, PA 16802.
Casey T. Harris, University of Arkansas Department of Sociology 211 Old Main Fayetteville, AR 72701 caseyh@uark.edu
REFERENCES
- Agresti Alan, Finlay Barbara. Statistical Methods for the Social Sciences. 3rd Edition. Prentice Hall; Upper Saddle River: 1997. [Google Scholar]
- Bartkowski John P. Charitable Choices: Religion, Race, and Poverty in the Post-Welfare Era. New York University Press; New York: 2003. [Google Scholar]
- Beyerlein Kraig, Hipp John. Social Capital, Too Much of a Good Thing? American Religious Traditions and Community Crime. Social Forces. 2005;84:995–1006. [Google Scholar]
- Blanchard Troy C., Bartkowski John P., Matthews Todd L., Kerley Kent R. Faith, Morality, and Mortality: The Ecological Impact of Religion Reconsidered. Social Forces. 2008;86:1591–1620. [Google Scholar]
- Chamlin Mitchell B., Cochran John K. Assessing Messner and Rosenfeld's Institutional Anomie Theory: A Partial Test. Criminology. 1995;33:411–430. [Google Scholar]
- Chamlin Mitchell B., Cochran John K. Social Altruism and Crime. Criminology. 1997;35:203–226. [Google Scholar]
- Chamlin Mitchell B., Cochran John K. Social Altruism and Crime Revisited: A Research Note on Measurement. Journal of Crime and Justice. 2001;24:59–72. [Google Scholar]
- Chatters Linda M., Joseph Taylor Robert, Lincoln Karen D. African American Religious Participation: A Multi-Sample Comparison. Journal for the Scientific Study of Religion. 1999;38:132–145. [Google Scholar]
- Chitwood Dale D., Weiss Michael L., Leukefeld Carl G. A Systematic Review of Recent Literature on Religiosity and Substance Use. Journal of Drug Issues. 2008;38:653–688. [Google Scholar]
- Currie Elliott. Market, Crime, and Community: Toward a Mid-Range Theory of Post-Industrial Violence. Theoretical Criminology. 1997;1:147–172. [Google Scholar]
- Desmond Scott, Morgan Kristopher, Kikuchi George. Congregations and Crime: Is the Spatial Distribution of Congregations Associated with Neighborhood Crime Rates? Journal for the Scientific Study of Religion. 2010;49:37–55. [Google Scholar]
- DiIulio John D. More Religion, Less Crime? Science, Felonies, and the Three Faith Factors. Annual Review of Law and Social Science. 2009;5:115–133. [Google Scholar]
- DuBois WEB. On Sociology and the Black Community. University of Chicago Press; Chicago: 1978. [Google Scholar]
- Ellison Christopher G., Burr Jeffrey, McCall Patricia. Religious Homogeneity and Metropolitan Suicide Rates. Social Forces. 1997;76:273–299. [Google Scholar]
- Ellison Christopher G., Burr Jeffrey, McCall Patricia. The Enduring Puzzle of Southern Homicide: Is Regional Religious Culture the Missing Piece? Homicide Studies. 2003;7:326–352. [Google Scholar]
- Ellison Christopher G., Sherkat Darren. The Semi-Involuntary Institution Revisited: Regional Variation in Church Participation Among Black Americans. Social Forces. 1995;73:1415–1437. [Google Scholar]
- Finke Roger, Adamczyk Amy. Cross-National Moral Beliefs: The Influence of National Religious Context. The Sociological Quarterly. 2008;49:617–652. doi: 10.1111/j.1533-8525.2008.00130.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Finke Roger, Guest Avery, Stark Rodney. Mobilizing Religious Markets: Religious Pluralism in the Empire State, 1865. American Sociological Review. 1996;61:203–218. [Google Scholar]
- Finke Roger, Scheitle Christopher. Accounting for the Uncounted: Computing Correctives for the 2000 RCMS Data. Review of Religious Research. 2005;47:5–22. [Google Scholar]
- Finke Roger, Stark Rodney. Religious economies and sacred canopies: Religious mobilization in American cities, 1906. American Sociological Review. 1988;53:41–49. [Google Scholar]
- Finke Roger, Stark Rodney. The Churching of America, 1776-1990: Winners and Losers in our Religious Economy. Rutgers University Press; New Brunswick: 2005. [Google Scholar]
- Harris Casey T., Feldmeyer Ben. Latino Immigration and White, Black, and Latino Violent Crime: A Comparison of Traditional and Non-Traditional Immigrant Destinations. Social Science Research. 2013;42:202–216. doi: 10.1016/j.ssresearch.2012.08.014. [DOI] [PubMed] [Google Scholar]
- Hunt Larry L., Hunt Mathew O. Race, Region, and Religious Involvement: A Comparative Study of Whites and African Americans. Social Forces. 2001;50:605–631. [Google Scholar]
- Glanville Jennifer L., Sikkink David, Hernandez Edwin I. Religious Involvement, and Educational Outcomes: The Role of Social Capital and Extracurricular Participation. The Sociological Quarterly. 2008;49:105–138. [Google Scholar]
- Jang Sung Joon, Johnson Byron R. Neighborhood Disorder, Individual Religiosity and Adolescent Use of Illicit Drugs: A Test of Multilevel Hypotheses. Criminology. 2001;39:109–144. [Google Scholar]
- Jang Sung Joon, Johnson Byron R. Gender, Religiosity, and Reactions to Strain among African Americans. The Sociological Quarterly. 2005;46:323–357. [Google Scholar]
- Johnson Byron, Joon Jang Sung. Crime and Religion: Assessing the Role of the Faith Factor. In: Rosenfeld R, Quinet K, Garcia C, editors. Contemporary Issues in Criminological Theory and Research: The Role of Social Institutions. Wadsworth; Belmont, CA: 2010. pp. 117–150. [Google Scholar]
- Johnson Byron, Sung Joon Jang Spencer De Li, Larson David. The ‘Invisible Institution’ and Black Youth Crime: The Church as an Agency of Local Social Control. Journal of Youth and Adolescence. 2000a;29:479–498. [Google Scholar]
- Johnson Byron R., David Larson Spencer D. Li, Joon Jang Sung. Escaping from the Crime of Inner Cities: Church Attendance and Religious Salience among Disadvantaged Youth. Justice Quarterly. 2000b;17:377–391. [Google Scholar]
- Johnson Byron R., Sung Joon Jang David B. Larson, De Li Spencer. Does Adolescent Religious Commitment Matter?: A Reexamination of the Effects of Religiosity on Delinquency. Journal of Research in Crime and Delinquency. 2001;38:22–44. [Google Scholar]
- Jolliffe IT. Principal Component Analysis. 2nd Edition. Springer; New York: 2002. [Google Scholar]
- Jones Dale E., Doty Sherri, Horsch James E., Houseal Richard, Lynn Mac, Marcum John P., Sanchagrin Kenneth M., Taylor Richard H. Religious Congregations and Membership in the United States 2000: An Enumeration by Region, State and County Based on Data Reported by 149 Religious Bodies. Glenmary Research Center; Nashville, TN: 2002. [Google Scholar]
- Kirk David, Papachristos Andrew. Cultural Mechanisms and the Persistence of Neighborhood Violence. American Journal of Sociology. 2011;116:1190–1233. doi: 10.1086/655754. [DOI] [PubMed] [Google Scholar]
- Kubrin Charis E., Stucky Thomas D., Krohn Marvin D. Researching Theories of Crime and Deviance. Oxford University Press; New York: 2009. [Google Scholar]
- Kutner Michael, Nachtsheim Christopher, Neter John. Applied Linear Regression Models. 4th edition McGraw-Hill Irwin; 2004. [Google Scholar]
- Land Kenneth, McCall Patricia, Cohen Lawrence. Structural Covariates of Homicide Rates: Are There Any Invariances Across Time and Social Space? American Journal of Sociology. 1990;95:922–963. [Google Scholar]
- Lee Matthew R. The Religious Institutional Base and Violent Crime In Rural Areas. Journal for the Scientific Study of Religion. 2006;45:309–24. [Google Scholar]
- Lee Matthew R. Civic Community In the Hinterland: Toward a Theory of Rural Social Structure and Violence. Criminology. 2008;46:447–78. [Google Scholar]
- Lee Matthew R., Bartkowski John P. Love Thy Neighbor? Moral Communities, Civic Engagement, and Juvenile Homicide in Rural Areas. Social Forces. 2004;82:1001–1035. [Google Scholar]
- Lee Matthew R., . Ousey Graham C. Institutional Access, Residential Segregation, and Urban Black Homicide. Sociological Inquiry. 2005;75:31–54. [Google Scholar]
- Lee Matthew R., Thomas Shaun A. Civic Community, Population Change, and Violent Crime in Rural Communities. Journal of Research in Crime and Delinquency. 2010;47:118–47. [Google Scholar]
- Lee Matthew R., Bankston William B., Hayes Timothy C., Thomas Shaun A. Revisiting the Southern Culture of Violence. The Sociological Quarterly. 2007;48:253–276. [Google Scholar]
- Lee Matthew R., Thomas Shaun A., Ousey Graham. Southern Culture and Homicide: Examining the Cracker Culture/Black Rednecks Thesis. Deviant Behavior. 2010;31:60–96. [Google Scholar]
- Light Michael T., Harris Casey T. Race, Space, and Violence: Exploring Spatial Dependence in Structural Covariates of White and Black Violent Crime in US Counties. Journal of Quantitative Criminology. 2012;28:559–586. [Google Scholar]
- Lundskow George. The Sociology of Religion: A Substantive and Transdisciplinary Approach. Pine Forge Press; Los Angeles: 2008. [Google Scholar]
- Maume Michael O., R Matthew. Social Institutions and Violence: A Sub-National Test of Institutional Anomie Theory. Criminology. 2003;41:1137–1172. [Google Scholar]
- McElroy Marjorie B. Goodness of Fit for Seemingly Unrelated Regression: Glahn's R2 and Hooper's R2. Journal of Econometrics. 1977;6:381–387. [Google Scholar]
- Merton Robert K. Anomie, Anomia, and Social Interaction. In: Clinard MB, editor. Anomie and Deviant Behavior: A Discussion and Critique. The Free Press; New York: 1964. pp. 213–242. [Google Scholar]
- Messner Steven. Morality, Markets, and the ASC. Criminology. 2012;50:5–25. [Google Scholar]
- Messner Steven, Thome Helmut, Rosenfeld Richard. Institutions, Anomie, and Violent Crime: Clarifying and Elaborating Institutional-Anomie Theory. International Journal of Conflict and Violence. 2008;2:163–181. [Google Scholar]
- Messner Steven, Rosenfeld Richard. Crime and the American Dream. Wadsworth; Belmont, CA: 1994. [Google Scholar]
- Messner Steven, Rosenfeld Richard. Political Restraint of the Market and Levels of Criminal Homicide: A Cross-National Application of Institutional-Anomie Theory. Social Forces. 1997;75:1393–1416. [Google Scholar]
- Messner Steven, Rosenfeld Richard. An Institutional Anomie Theory of Crime. In: Paternoster R, Bachman R, editors. Explaining Crime and Criminals. Roxbury; Los Angeles: 2001. pp. 151–160. [Google Scholar]
- Moody Carlisle, Marvell Thomas. The Debate on Shall Issue Laws, Continued. Econ Journal Watch. 2009;6:203–217. [Google Scholar]
- Mosher Clayton, Miethe Terrance, Phillips Dretha. The Mismeasure of Crime. Sage Press; Thousand Oaks: 2002. [Google Scholar]
- Office of Planning and Research (Moynihan, Daniel Patrick) The Negro Family – The Case for National Action. U.S. Department of Labor; Washington, D.C.: Mar, 1965. 1965. [Google Scholar]
- Ousey Graham. Homicide, Structural Factors, and the Racial Invariance Hypothesis. Criminology. 1999;37:405–426. [Google Scholar]
- Park Robert E., Burgess Earnest W., MacKenzie Roderick. The City. University of Chicago Press; Chicago: 1925. [Google Scholar]
- Pearce Michelle J., Jones Stephanie M., Schwab-Stone Mary E., Ruchkin Vladislav. The Protective Effects of Religiousness and Parent Involvement on the Development of Conduct Problems Among Youth Exposed to Violence. Child Development. 2003;74:1682–1696. doi: 10.1046/j.1467-8624.2003.00631.x. [DOI] [PubMed] [Google Scholar]
- Peterson Ruth, Krivo Lauren. Macrostructural Analyses of Race, Ethnicity, and Violent Crime: Recent Lessons and New Directions for Research. Annual Review of Sociology. 2005;31:331–56. [Google Scholar]
- Pratt Travis, Cullen Frances. Assessing Macro-Level Predictors and Theories of Crime: A Meta-Analysis. In: Tonry M, editor. Crime and Justice: A Review of Research. University of Chicago Press; Chicago: 2005. pp. 373–450. [Google Scholar]
- Putnam Robert D. Bowling Alone: The Collapse and Revival of American Community. Simon and Schuster; New York, NY: 2000. [Google Scholar]
- Reardon Sean F., Firebaugh Glenn. Measures of Multigroup Segregation. Sociological Methodology. 2002;32:33–67. [Google Scholar]
- Regnerus Mark. Moral Communities and Adolescent Delinquency: Religious Contexts and Community Social Control. The Sociological Quarterly. 2003;44:523–554. [Google Scholar]
- Regnerus Mark. Adolescent Delinquency. In: Ebaugh HR, editor. Handbook of Religion and Social Institutions. Springer; New York, NY: 2006. pp. 265–282. [Google Scholar]
- Rosenfeld Richard. The Big Picture: 2010 Presidential Address to the American Society of Criminology. Criminology. 2011;49:1–26. [Google Scholar]
- Scheitle Christopher, Dougherty Kevin. Race, Diversity, and Membership Duration in Religious Congregations. Sociological Inquiry. 2010;80:405–423. [Google Scholar]
- Schwartz Jennifer. Effects of Diverse Forms of Family Structure on Female and Male Homicide. Journal of Marriage and Family. 2006;68:1291–1312. [Google Scholar]
- Sherkat Darren. African American Religious Affiliation and in the Late 20th Century: Cohort Variations and Patterns of Switching 1973-1998. Journal for the Scientific Study of Religion. 2002;41:485–493. [Google Scholar]
- Shihadeh Edward S., Winters Lisa. Church, Place, and Crime: Latinos and Homicide in New Destinations. Sociological Inquiry. 2010;80:628–649. doi: 10.1111/j.1475-682x.2010.00355.x. [DOI] [PubMed] [Google Scholar]
- Smith Christian. Moral, Believing Animals: Human Personhood and Culture. Oxford University Press; New York: 2003. [Google Scholar]
- Stark Rodney. Deviant Places. Criminology. 1987;25:893–909. [Google Scholar]
- Stark Rodney. Religion as Context: Hellfire and Delinquency One More Time. Sociology of Religion. 1996;57:163–173. [Google Scholar]
- Stark Rodney, Sims Bainbridge William. Religion, Deviance and Social Control. Routledge; New York: 1996. [Google Scholar]
- Stark Rodney, Doyle Daniel P., Kent Lori. Rediscovering Moral Communities: Church Membership and Crime. In: Hirschi Travis, Gottfredson Michael., editors. Understanding Crime. Sage; Beverly Hills, CA: 1980. [Google Scholar]
- Steffensmeier Darrell, Ulmer Jeffery T., Feldmeyer Ben, Harris Casey T. Scope and conceptual issues in testing the race-crime invariance thesis: Black, white, and Hispanic comparisons. Criminology. 2010;48:1133–70. doi: 10.1111/j.1745-9125.2010.00214.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Steffensmeier Darrell, Feldmeyer Ben, Harris Casey T., Ulmer Jeffrey. Reassessing Trends in Black Violent Crime, 1980-2008: Sorting Out the ‘Hispanic Effect’ in Uniform Crime Reports Arrests, National Crime Victimization Survey Offender Estimates, and U.S. Prisoner Counts. Criminology. 2011;49:197–251. [Google Scholar]
- Steensland Brian, Park Jerry Z., Regnerus Mark D., Robinson Lynn D., Bradford Wilcox W, Woodberry Robert D. The Measure of American Religion: Toward Improving the State of the Art. Social Forces. 2000;79:291–318. [Google Scholar]
- Stroope Samuel. How Culture Shapes Community: Bible Belief, Theological Unity, and a Sense of Belonging in Religious Congregations. The Sociological Quarterly. 2011;52:568–593. doi: 10.1111/j.1533-8525.2011.01220.x. [DOI] [PubMed] [Google Scholar]
- Taylor Robert Joseph, Chatters Linda M., Jayakody Rukmalie, Levin Jeffrey S. Black and White Differences in Religious Participation: A Multi-sample Comparison. Journal for the Scientific Study of Religion. 1997;35:403–410. [Google Scholar]
- Thomas WI, Znaniecki Florian. The Polish Peasant in Europe and America. University of Chicago Press; Chicago: 1918. [Google Scholar]
- Tolbert Charles M., Lyson Thomas A., Irwin Michael D. Local Capitalism, Civic Engagement, and Socioeconomic Well-Being. Social Forces. 1998;77:401–427. [Google Scholar]
- Tolbert Charles M., Irwin Michael D., Lyson Thomas A., Nucci Alfred R. Civic Community in Small-Town America: How Civic Welfare is Influenced by Local Capitalism and Civic Engagement. Rural Sociology. 2002;57:90–113. [Google Scholar]
- Ulmer Jeffery T., Bader Christopher, Gault Martha. Do Moral Communities Play a Role in Criminal Sentencing? Evidence From Pennsylvania. The Sociological Quarterly. 2008;49:737–768. doi: 10.1111/j.1533-8525.2008.00134.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ulmer Jeffery T., Steffensmeier Darrell. Reply to Lee: Is there No Place for Culture in a Sociology of Illegal Enterprise? American Sociological Review. 71:162–166. [Google Scholar]
- United States Census Bureau . Annual Estimates of the Population by Race, Sex, and Hispanic Origin for the United States, 2000–2007. Washington, DC.: 2008. [Google Scholar]
- Wallace John M., Ryoko Yamaguchi Jerald Bachman, Patrick O'Malley John Schulenberg, Johnston Lloyd. Religiosity and Adolescent Substance Use: The Role of Individual and Contextual Influences. Social Problems. 2007;54:308–327. [Google Scholar]
- Wikstrom Per-Olof H. Explaining Crime as Moral Action. In: Hitlin S, Vaisey S, editors. Handbook of the Sociology of Morality. Springer; New York: 2010. pp. 211–239. [Google Scholar]