Ferraro and LaGrange defined fear of crime as “an emotional response of dread or anxiety to crime or symbols that a person associates with crime” [3]. More specifically, individuals fear the harm or loss resulting from criminal acts, and not simply the existence of actions labeled as “crimes” [4]. This fear is produced by one's perceived risk of crime, which is “a recognition of a situation as possessing at least potential danger, real or imagined” [3]. It is important to distinguish between these concepts: fear of crime refers to an emotive state of being afraid of being victimized by crime while perceived risk of crime refers to a cognitive assessment of one's chances of being victimized. As measured variables, perceived high risk of crime should increase levels of fear of crime, although the two are not necessarily highly correlated. Some people, for example, may perceive high risk of crime but not be highly afraid of it [5]. Further, these concepts should not be confused with general concern over crime [5], which has more to do with recognition of crime as a social problem. General concern does not necessarily equate with fear and perceived risk. For example, a person can believe that a community or society has a significant crime problem without thinking that s/he has a significant chance of being personally victimized or being very afraid of crime.

Past research suggests that fear of both violent crime and property crime is widespread across the public and is a social problem separate from crime itself, although most people are not likely highly fearful of crime on a daily basis [6]. Zhao, Lawton, and Longmire [7] found that living in close proximity to crime events increased residents' fear of crime. Other research shows a lack of correspondence between fear of crime and official crime rates, with fear appearing to over- or under-estimate actual crime threats [8]. Regardless, it is apparent from research on fear of crime that such fear is not simply a reaction to area crime rates and personal experiences with victimization.

A number of studies suggest that fear of crime is also the result of several personal as well as ecological and community level factors [5,9]. While personal and demographic variables such as age, gender, income, education, race, health, and victimization are frequently studied, religious orientation or behavior variables are rarely included in models examining fear of crime. Previous research showed that religiosity is an important factor of public perceptions of criminal behavior in that the strength of religious belief is positively related to punitive and morally indignant responses to crime [10]. It appears that more religious persons may have more “tough on crime” attitudes, but how fearful do they tend to be of crime? This oversight in the literature is curious, given that religious affiliation ties persons to social networks offering resources that can be used to respond to threats to well-being such as crime. Additionally, and perhaps most crucially, religious affiliation has been found to build generalized trust (trust in others unknown to us) in participants [11-18]. This form of trust has been characterized as “the foundation of a civil society… (that) eases the way toward getting people to work together” [19]. Religious participation or affiliation is seen as a central pathway for the generation of such trust, a contention extending back in American social thought to Alexis de Tocqueville's Democracy in America almost two centuries ago. Contemporary scholars maintain that religious institutions play a vital role in the development of generalized trust today [16-18].

This trust in others beyond our immediate social milieu is critical in our complex, specialized world, reducing social risk and fears and enabling day-to-day social action to continue relatively unimpeded [20,21]. Excessive fear of external actors and social conditions, both known and unknown, would paralyze the modern social world, as individuals would be too fearful or mistrusting to engage in any sort of “risky” action. Thus, elevated levels of generalized trust should engender lower levels of distrust or misanthropy, including that which is directed towards a general fear of crime. Excessive fear of crime can cause such social paralysis, as it leads to significant changes in individual and collective behavior and can truly alter social relations. Despite the apparent link between religion, trust, and perceptions of crime, only a few studies examining the relationship between religion and fear of crime could be found.

Lotz tested for the relative impacts of experienced victimization versus indicators of resistance to social change that included religion and political conservatism [22]. In a sample of residents in Washington state in the US, Lotz found that religion and political conservatism were slightly associated with increased concern over but not fear of crime while victimization was slightly associated with increased fear of but not concern over crime. The results suggest that fear results from practical crime threats but concern results from socio-political ideology [22].

O'Mahony and Quinn tested if community-related factors explain fear of crime better than individual factors, including one's religious participation [23]. Prior research using national samples suggest that Protestants are more worried about terrorist attacks than Catholics in Northern Ireland. Using community-based survey data, however, O'Mahony and Quinn found in a multivariate model that community type and respondent satisfaction with the community significantly determined fear of crime while individual characteristics, including one's religion, did not [23]. It is important to recognize here that the community type measure they used is partly defined by religious composition (groupings included “Catholic Lower Working Class Urban,” “Protestant Small Towns,” etc.) [23]. Thus, while their study suggests that individuals' religious affiliation alone does not determine their fear of crime, it does not suggest that religion plays no part in determining fear of crime.

Sacco and Nakhaie included religiosity among several demographic variables in their analysis of age and crime-coping adaptations in a Canadian national sample [24]. The researchers argue for and used behavioral, rather than perceptual, outcome measures of fear of crime. These behaviors included “carrying something for self-defense, locking car doors when alone, checking the back seat of the car for intruders, planning their route with safety in mind, and staying at home at night because they are afraid” [24]. Results showed that among nonelderly respondents only, the religious (compared to the non-religious) were more likely to engage in some of the behaviors—lock car doors, plan a safe route, and stay home for fear. However, Sacco and Nakhaie's behavioral measures of fear of crime are controversial as they may actually indicate a separate subjective reaction to a crime construct, behavioral reactions to crime, and not fear of crime that is widely understood as an emotional state [24].

Mohammed, Saridakis, and Sookram included a dichotomous measure of religion (religious affiliation/no religious affiliation) as a control variable in their analysis of the impact of victimization upon fear of crime—also a dichotomous measure (fearful/not fearful of crime) in a national sample from Trinidad and Tobago [25]. The researchers report that persons with a religious affiliation were more likely to be fearful of crime than those without a religious affiliation, a result they found to be counter-intuitive.

Clearly, more studies are needed if the relationship between religion and fear of crime is to be understood. Aside from being too few, religion was not the focus of prior studies and they used dated or questionable measures of fear of crime, such as those noted in the review above.

The current study tests for a possible association between religious involvement and fear of crime. Frequency of attendance at religious services and religious orientation will serve as the primary independent variables of interest in models examining fear of crime as an outcome. Because of the paucity and inconsistency of past research on religion and fear of crime, a positive or negative relationship between the two will not be hypothesized. Generalized trust is included to see if it has effects independent of religious attendance and religious orientation. Also, an index of civic engagement is included here, as those who participate more broadly in social life should also generally be more trusting and less fearful. Further, political orientation, community tenure, and civic participation are added as potentially competing independent variables. If religious involvement is associated with fear of crime, it will be important to explore the possibility that liberal vs. conservative identity or community ties in general, not religious activity specifically, are actually affecting people's fear of crime. As O'Mahony and Quinn point out, it may be the lived experiences within the community that determines fear of crime more than religion and other personal factors [23]. Finally, prior victimization is included as a potential independent variable, even though past research does not indicate that objective experience with crime is a strong predictor of fear of crime [5].

The models also contain a set of demographic control variables: age, gender, race, and education. Past research shows that fear of crime varies according to personal factors [5]. Two key factors here are age and gender. Earlier studies suggested that older persons have higher fear but more rigorous recent research has revealed complex relationships. For example, Ferraro found a curvilinear relationship with age: the young, not elders, had the highest levels of fear while middle-aged adults had the lowest [5]. Concerning gender, studies consistently show that women have higher levels of fear of crime, including crimes in which women are not more likely to be victimized [5]. It is possible that fear of sexual assault elevates women's fear of other crimes. Because women are at higher risk of sexual assault, it may operate as a master offense that increases their fear of other crimes, especially those with the potential to lead to sexual victimization [26]. Research by Ferraro [26] and Fisher and Sloan [27] supported hypotheses that women's fear of general crime shadows their fear of sexual assault. However, this research focused little on how risk shapes men's fear of crime. Reid and Conrad [28] found that women reported higher levels of fear of sexual assault but also that perceived risk had a greater cumulative impact on men's fear of robbery, a crime in which men are more likely to be victimized. The shadow effect in women's fear of crime appears to be part of a larger gender differences picture. As Schafer, Huebner, and Bynum's [29] research shows, men and women's experiences in fearing victimization are likely determined by different factors. Also, gender differences may lie in the reporting itself. To protect masculine self-identity, men may under-report their levels of fear. Sutton and Farrall [30] found evidence that men's, but not women's, reports of fear of crime is influenced by social desirability. At any rate, in this analysis of the impact of religious participation on fear of crime, it is important to consider the impact of gender.

Finally, fear of crime could be an outcome of the extent to which one finds the neighborhood to be a safe place. Assessments of the presence of danger in the community, where people spend a great deal of time, are likely to influence fear of crime. Several studies find that neighborhood characteristics and perceptions of one's neighborhood affect fear of crime [ see for example 7,9,31-33]. Further, like community tenure and civic participation, feeling safe in the neighborhood could be an indicator of community attachment—a “safe place” could be part of a favorable definition of a community. Thus, a measure of how safe one feels in the neighborhood will also be included. A total of four models will be tested: two examining fear of violent crime and two examining fear of property crime.

In Fall 2010, the Survey Research Center at the University of West Georgia launched the West Georgia Area Survey (WGAS). The 2010 WGAS was a first of its kind social scientific phone survey of the seven county core of the West Georgia region, an area of roughly 650,000 people on the western fringe of the larger Atlanta metropolitan area. These counties were selected due to their geographical proximity and general demographic similarities with Carroll County, GA, which is the home county of the University of West Georgia. The WGAS was designed to follow from decades-long studies like the General Social Survey (National Opinion Research Center at the University of Chicago), the Houston Area Survey (Rice University), the Baton Rouge Area Survey (Louisiana State University), the Mississippi Poll (Mississippi State University) and other similar social science attempts to empirically assess attitudes and behaviors regarding topical issues of the day. Another study of particular influence was political scientist Robert Putnam's Social Capital Benchmark Survey. This study was conducted with a nationally representative sample and with dozens of communities across the United States. Topics and questions were drawn from these previous studies as appropriate to fit the purposes of the WGAS principal investigators.

A landline random digit dialing sample was generated for the Survey Research Center by Survey Sampling Incorporated, and the survey was administered using Computer Assisted Telephone Interview software over a six week period from late October to early December. The sample was first drawn proportionally from each of the seven counties, based on the most recent Census population estimates. Households were then randomly drawn from within each county. The most recent birthday was used as the final step in sampling individuals within the households contacted.

The survey contained questions on a wide range of topics, including: demographics and contextual variables; community life and social trust; civic involvement; environmental concern and water issues; and fear of crime and victimization. The average completion time for respondents was approximately 20 minutes. The final number of completed surveys was 380, with a response rate of 22%. While the response rate was lower than ideal (though not excessively so today), non-response bias was counteracted by generating a true random digit dialing sample, utilizing refusal conversion, and matching our demographics to local Census estimates, weighting those that were significantly different (such as age). The final dataset was roughly representative of major demographic characteristics in the study region except for the age of respondents. This is a fairly consistent reality today for those who engage in landline-based telephone surveys, as older respondents tend to be more likely to both have a landline and to use it as compared to younger respondents. Thus the decision was made to weight the data based on the age of the survey respondent. It should be noted that the models were run by the researchers both with and without weighting, with no major differences emerging.

The dependent variables utilized in the regression models include an index of fear of property crime and an index of fear of violent crime. Ferraro points out that sound measures of fear of crime should (1) tap into the emotion of fear or worry, not judgments or concerns, (2) refer to the type of victimization, not “crime” generically, (3) take aim at the subject's everyday life, not hypothetical situations or ones subjects may intentionally avoid, and (4) avoid double-barreled items, asking what subjects hypothetically or actually do for example [5]. Also, to best measure fear of crime in survey research, multi-item measures are essential [5]. The current analysis uses the fear of crime measure developed by Ferraro and Lagrange to respond to these criteria [5].

Two indices were constructed from a series of ten statements where respondents were asked to rate their fear of each on a scale of 1–10, where 1 indicated no fear at all, and 10 meant that the respondent was very afraid. The indicators of fear of property crime included:

being approached on the street by a beggar or panhandler

The indicators of fear of violent crime included:

having someone break into your home while you are there

The possible values for each index range from 5–50. The mean score for the property crime index was 22.02, and the mean score for the violent crime index was 23.19. Both indices exhibit a high degree of reliability: for the property crime index, the Cronbach's alpha was 0.831, while the Cronbach's alpha for the violent crime index was 0.935.

The demographic variables in the regression models include race (dichotomized to white and black/other); education, a six-point ordinal variable ranging from less than high school to graduate or professional degree; gender; political ideology, which includes conservative, moderate, liberal and those who “haven't thought about it” years lived in community, a six-point ordinal variable ranging from “less than one year” to “all your life”; a civic participation index, measuring participation in a wide range of community life, ranging from 0–15 types of activities; and generalized trust, recoded into two dummy variables (“people can be trusted” and “depends,” with “you cannot be too careful” as the reference group). Descriptive statistics for all the variables used in the regression models, are found in Table 1.

The focal independent variables of interest include two measures of religious life often used by researchers, frequency of religious attendance and religious orientation. The frequency of religious attendance variable asked “(n)ot including weddings and funerals, how often do you attend religious services?” Options ranged from every week to less often than once a year. Just fewer than forty percent of respondents reported attending religious services every week, while each of the other categories fell in the ten to twenty percent range. The variable was reverse coded for the regression analyses. The orientation question captured broad religious preferences, including Protestantism, Catholicism, another type of Christianity, Judaism, some other religion, or no religion. Protestantism was the most frequently cited preference, with 64.1% choosing this option. Other Christians were 15.5% of the respondents, while those with no religion were 9.3%, Catholics made up 7.5%, and those representing other religions were 3.6% of total respondents in the survey.

The other focal variables of interest for the researchers were previous victimization (either of a property crime or a violent crime respectively), and the respondent's feeling of safety at night in one's neighborhood. Over 42% of respondents reported having been a victim of property crime at some point in their lives, while 18.9% reported having been a violent crime victim. Feeling of safety at night in one's neighborhood is a measure that was used in earlier research to measure fear of crime, though Ferraro and Lagrange contend that it is likely a better indicator of perceived risk than fear [5].

While feeling of safety at night in one's neighborhood is moderately correlated with both the fear of property crime index and the fear of violent crime index, VIF diagnostics run for each regression model indicate no problems with multicollinearity. Slightly less than half of the respondents (45.2%) reported that they felt very safe in their neighborhood at night, while 41.2% felt somewhat safe, and only 13.6% reported feeling somewhat or very unsafe.

Ordinary Least Squares (OLS) regression analysis is used to examine whether the frequency of religious attendance and/or the respondent's religious orientation are predictors of fear of property crime and fear of violent crime, net of the effects of the respondent's general feeling of nighttime safety in their neighborhood and of the demographic variables. As noted above, Table 1 presents the descriptive statistics for all variables included in the analysis. Table 2 includes two models displaying the results of the OLS regression of the predictor variables on fear of property crime. In Table 3, there are two models presented which show the results of OLS regression of the predictor variables on fear of violent crime. In both Table 2 and Table 3, the first model includes the demographic variables, the previous victimization measure and the respondent's feeling of safety. The second model in both tables incorporates the measures from the first model, while adding in religious attendance and religious denomination.

Table 2 includes the results of the OLS regression models of the predictors of fear of property crime. Model 1 focuses on the demographic variables, previous property crime victimization, and the respondent's feeling of safety. Several demographic variables emerge as significant predictors of fear of property crime in this model. Gender (β = −0.122, p < 0.05) displays a significant and negative relationship with fear of property crime, indicating that male respondents are much less fearful of property crime than female respondents, controlling for all other variables in the model. Also significant and negative is one of the comparison groups for political ideology, those who “ haven't thought about” where they stand ideologically (β = −0.130, p < 0.05). Respondents in this category are much less afraid of property crime than the reference group of ideological conservatives. Years lived in community is also significant and negative (β = −0.123, p < 0.05), thus those respondents who lived in their community longer showed less fear of property crime than those who had been there a shorter time. Also, those who exhibit higher levels of trust were significantly less fearful of property crime than those who believe you cannot be too careful with others (β = −0.212, p < 0.001). Being the victim of a property crime was not statistically significant. However, the respondent's feeling of safety during the night in the neighborhood was highly significant and negative for both groups versus the reference group. Those who felt very safe (β = −0.457, p < 0.001) and those who felt somewhat safe (β = −0.225, p < 0.01) expressed lower levels of fear of property crime compared to those who felt somewhat unsafe or unsafe in their neighborhood at night.

Model 2 of Table 2 adds the respondent's religious participation and religious attendance to the predictors utilized in Model 1. Regarding the religious participation variables, religious attendance is significant and negative (β = −0.137, p < 0.05). Those who attend religious services more frequently exhibit less fear of property crime than those who attend less frequently, net of the other variables in the model. Further, Protestants (β = 0.239, p < 0.05) and Catholics (β = 0.162, p < 0.05) each reported significantly higher levels of fear of property crime when compared to the reference group of the non-religious. However, the inclusion of this variable did not reduce to insignificance the variables that were significant in the previous model. Gender (β = −0.138, p < 0.01) was again significant and negative, with male respondents displaying less fear of property crime than female respondents1. Those who “haven't thought about” where they stand ideologically were again less fearful of property crime than conservatives (β = −0.147, p < 0.05). Years lived in community was again significant and negative as well (β = −0.126, p < 0.05). Those who exhibit higher levels of trust were again significantly less fearful of property crime than those who believe you cannot be too careful with others (β = −0.210, p < 0.001). Finally, the respondent's feeling of safety during the night in the neighborhood was highly significant and negative for both groups versus the reference group. Those who felt very safe (β = −0.472, p < 0.001) and those who felt somewhat safe (β = −0.243, p < 0.01) again expressed lower levels of fear of property crime compared to those who felt somewhat unsafe or unsafe in their neighborhood at night.

The results of the OLS regression models of the predictors of fear of violent crime are found in Table 3. Again, Model 1 focuses on the demographic variables, previous property crime victimization, and the respondent's feeling of safety. Only gender, generalized trust, and feeling of safety at night in the neighborhood emerged as significant predictors of fear of violent crime. Gender (β = −0.278, p < 0.001) has a highly significant and negative relationship with fear of violent crime, displaying an even more powerful effect than it did on fear of property crime. Male respondents are much less fearful of violent crime than female respondents, controlling for all other variables in the model. Those who exhibit higher levels of trust were significantly less fearful of violent crime than those who believe you cannot be too careful with others (β = −0.184, p < 0.001), while those who said that it depends were also less fearful of violent crime than the reference group (β = −0.100, p < 0.05). Finally, he respondent's feeling of safety during the night in the neighborhood was highly significant and negative for those who felt very safe (β = −0.369, p < 0.001) versus the reference group.

Model 2 of Table 3 incorporates the religious participation variables along with the predictors of fear of violent crime utilized in Model 1. Neither religious attendance nor any of the denomination variables are statistically significant. Gender remained highly significant and negative (β = −0.272, p < 0.001), with male respondents once again displaying less fear of violent crime than female respondents. Those who reported higher levels of trust were again significantly less fearful of violent crime (β = −0.176, p < 0.01), and the respondent's feeling of safety during the night in the neighborhood was highly significant and negative for those who felt very safe (β = −0.378, p < 0.001) versus the reference group.

As noted in previous sections of the manuscript, this study suffers from certain limitations. The sample size of 380 is acceptably large for many types of analysis, but it can create issues when attempting to examine variations across many categories, such as religious denominations. This limited the analyses of religious differences to those of major affiliations, which obscures important denominational differences, as noted in the previous section. For example, those Protestants that dhere to a more conservative theology may have quite different belief systems regarding fear of crime than rather liberal Protestants [34]. Relatedly, more in depth measures of religious beliefs could help more effectively elucidate differences in fear of crime, in a similar manner to recent research which has demonstrated differences in gender beliefs rooted in religious ideology [35].

However, the results reported here indicate that more frequent religious involvement, as measured by worship attendance, is inversely related to fear of property crime. This lends credence to the idea that religious activity provides individuals with access to social networks that provide fear-reducing resources and even facilitate higher levels of generalized trust. If this link is genuine, it represents another positive outcome of religious activity. As discussed earlier, such generalized trust is a crucial element of life in a complex, modern world [20,21]. Even though fear of crime may be beneficial to the extent that it motivates a person to take precautions against victimization, thereby increasing feelings of safety, fear is an undesirable emotion that potentially reduces one's well-being and quality of life [31].

In addition to further expanding our understanding of correlates with fear of crime, future research should examine other important victimization variables as well. It remains to be seen how religious involvement impacts other subjective appraisals of crime threats (e.g., perceived risk and general concern) as well as actual victimization. Also, more research is needed to specify what it is about religion that reduces fear of crime. The trust generated by participation in religious activities is one possibility, but other dimensions of religious life could also be involved in determining levels of fear. More personal religious activities such as prayer, meditation, and reading sacred texts, for example, may serve as routine activities that alleviate one's fears. Individualized religious coping activities such as prayer have also been found to reduce distress [36,37]. However, the current study does suggest that religious involvement is not spuriously related to fear of crime via community attachment or conservative identity. Future researchers should seek to explore the nuances of the general relationship between religion, trust, and fear of crime.