Abstract
Although the subject has received very little systematic research, media reports of crimes occurring on the property of places of worship appear regularly. While the national media often focus on violent incidents involving religious bias, local media often report on the more common issues that many congregations face, such as vandalism and theft. This article presents analyses of newspaper reports of crimes taking place on religious congregations’ property in 2012. The data indicate that crimes occurring on the property of religious congregations are not an uncommon event. Analyses suggest that congregations of religious minorities do not appear to be more likely to have an incident occur on their property, but the crimes that do occur on their property appear to be more bias-motivated. Future research would benefit from a nationally representative survey of places of worship that included questions measuring experiences with crime.
Places of worship have long been a target of criminal activity. The most well-known cases involve actions that led to significant destruction and/or deaths. For instance, on September 15, 1963 four African-American girls were killed when their Baptist church was bombed with dynamite. On August 5, 2012 a shooting at a Sikh temple in Wisconsin killed six worshippers. While such extreme violence becomes recorded in history, places of worship also face much more frequent and common acts of vandalism, burglary, and other crimes on a daily basis.
Systematic research on crimes occurring on the property of religious congregations, though, is extremely sparse. This is unfortunate, not just because of the prevalence and social importance of this phenomena, but because such research would contribute to rich literatures in social science. In this research we present data collected from newspaper accounts of crimes that occurred on congregational property in 2012 in the United States. While newspaper accounts have potential limitations, which we will acknowledge more below, other potential data sources are even more flawed and problematic. As a result, we look to this newspaper database as a first step in building knowledge in this area of research.
Crime and religion
The literature linking crime and religion is vast in volume but narrow in focus. Most of this research has focused on whether religion affects individuals’ likelihood of committing crime, either as a function of their own individual religiosity or the religiosity of individuals surrounding them (see review by Johnson et al., 2012). Less often, research has examined crime rates in relation to religious adherence or congregation presences in local areas (e.g. Beyerlein and Hipp, 2005; Lee, 2006; Lee and Bartkowski, 2004; Ulmer and Harris, 2013) or criminal punishment and religious contexts (e.g. Fearn, 2005; Ulmer et al., 2008).
Research examining places of worship as the victim of crime is quite sparse and limited. Most of the research that has been conducted has been in the area of what is often called hate or bias crimes. For instance, Soule and Van Dyke (1999) found that higher unemployment rates and more black political officials were positively related to the number of black church arsons. Bartkowski et al. (2002) found that church arsons were more likely in smaller urban areas, contrary to the popular assumption that such crimes are a rural phenomenon. They also found that the density of churches in an area increases arson rates, which they attribute at least partially to the greater opportunities and availability of targets. In a similar analysis, Disha et al. (2011) found that the concentration of Muslims and Arabs in a county increased the raw number of hate crime incidents, which they also explain as a function of target availability. However, they found that the likelihood of a Muslim or Arab being victimized is reduced as the size of the Muslim and Arab population increases. They explain this as the power that comes with larger group size (i.e. offenders may fear retaliation).
Beyond the hate crime literature, research on crime involving places of worship becomes even sparser. Bourns and Wright (2004) surveyed a small number of Southern Baptist and United Methodist churches to evaluate their perceptions of, and experiences with, crime. Interpretations of their results must be made with caution as the sample was highly limited and not representative. Nonetheless, from the responses they received, 49% of the churches had reported an incident of a verbal threat to the church building or clergy, 47% had reported a phone threat, 78% had reported vandalism, and 73% had reported a theft. The Christian Security Network released a report in 2011 stating that between 2009 and 2010 the number of crimes against Christian organizations increased from 1,237 to 1,783 (Griffin, 2011). However, it is unclear exactly how these data were collected. Even if they were systematic and exhaustive of crimes against Christian churches, they still would not tell us anything about crimes against non-Christian places of worship, which could actually be more common relative to their population size.
Existing data
One of the reasons why there is such little research on this topic is the lack of data. The first place one might look is to one of the nationally representative surveys of places of worship, such as the National Congregations Study (Chaves and Anderson, 2008) or the US Congregational Life Survey (Woolever and Bruce, 2009). Unfortunately, these collections have not included items concerning crime experiences.
The next option could be official crime data. While there are no centralized statistics for crimes involving places of worship in general, the Federal Bureau of Investigation (FBI) does compile reports of hate crimes with an anti-religious motivation. These hate crime statistics are part of the larger Uniform Crime Report (UCR), through which law enforcement agencies from around the United States submit their crime data to the FBI. Not all anti-religious hate crimes are necessarily against a congregation. Of the 6,222 hate crime incidents reported in 2011, 1,233 incidents (20%) were motivated by some religious bias and a ‘religious organization’ was the victim in 204 of those incidents (3.2%).
There are several major weaknesses with these hate crime data. First, the categorization of a particular crime as a hate crime is subject to the interpretation of local law enforcement agencies and officers. As Cronin et al. (2007) observe in their study of hate crime determinations in police departments, making distinctions between regular crime and hate crime is challenging because it depends on making a judgment about motivation. Such judgments are fraught with ambiguity and uncertainty (see also Martin, 1995). Such judgments are also influenced by organizational and contextual characteristics of the law enforcement agency and community (Jenness and Grattet, 2005; McVeigh et al., 2003).
Many advocacy groups, such as the Southern Poverty Law Center, have criticized the hate crime statistics as severely underestimating the amount of hate crimes occurring (Potok, 2012). Indeed, a study by the Bureau of Justice Statistics found that, based on the National Crime Victimization Survey (NCVS), the actual number of hate crime incidents was almost 30 times higher than the official FBI numbers (Harlow, 2005). Some of this difference is due to the methodological differences in the data from which these numbers are derived, as victimization surveys often find more crime than official police reports since some crime goes unreported. However, the size of the UCR–NCVS gap is much larger for hate crimes than other crimes.
Regardless of these issues, even if the FBI’s hate crime data were flawless, it does not solve the problem that it only represents one type of crime (and probably a relatively rare type) that a place of worship may encounter. It does not represent the everyday vandalism, theft, or other type of crime that might take place on a congregation’s property without any type of bias motivation. Given this, we turn to examining newspaper records as a potential data source that might represent a larger range of crime. Before explaining our data collection method, though, it is worth acknowledging the potential issues with this data source.
Potential biases of newspaper data
There are two potential issues presented by the use of newspaper records. First, for a newspaper to report on a crime it must first know or hear about it. The probability of reporting is likely to increase if the crime is reported to the police. The probability of reporting is also likely to increase if the crime is unusual in whom or what it targets (e.g. the elderly, children) or particularly violent (Sorenson et al., 1998). When looking at crimes involving places of worship, though, these two forces could pull in opposite directions. There has not been much research on variations in crime reporting and relationships with police by religion. However, based on research examining such issues in racial and other minority populations (Cao, 2011; Tyler, 2005) there is reason to hypothesize that religious minorities might be less likely to report crimes. On the other hand, crimes involving religious minorities might appear more newsworthy and more likely to generate media attention.
One factor that might mitigate some of the concern about the biasing effect of ‘newsworthiness’ is that many newspapers include a police log or blotter that lists all crimes reported in the community. This does not, however, mitigate the concern about crimes being disproportionately unreported across types of places of worship. Unfortunately, without any clear external benchmark to assess actual crimes involving congregational property relative to crimes reported in newspapers, we have no way to assess these issues. All we can do here is keep in mind that we are looking at the frequency and pattern of crimes that are reported to the police and/or receive media attention.
Identifying articles
We began by having software randomly select a date from the 2012 calendar. We then sampled every tenth day from that date to generate a sample of 36 dates. We chose this systematic method versus a random selection of dates so that we could ensure an even distribution of dates across months. We then tested three newspaper databases (ProQuest, Lexis Nexis, and NewsBank) to see which produced the most coverage of news for our purposes. These test searches found that the NewsBank database was the most comprehensive. For example, a search for ‘vandalism’ and “church” for one date produced one article from Lexis Nexis and ProQuest. NewsBank produced 13 articles, including the two that the other databases had found.
After choosing a database we then conducted searches on our selected dates combining the keywords shown in Table 1. Each search combined a term from the first and second columns.
Search terms used to identify articles.
An undergraduate research assistant examined each article returned by each search to see if it fulfilled the following criteria: Did the article mention a crime that occurred in 2012? Did the article mention a crime that occurred in the United States? Did the article mention a crime that occurred on the property (inside or out) of a place of worship?
Crimes did not have to be clearly ‘against’ the place of worship to be included. The criteria simply asked whether the crime took place on the property of the congregation or faith group. In some cases a crime took place in the grounds or in the parking lot of the place of worship, but the crime did not seem to involve that building directly. For instance, two people might have met in a church parking lot simply because it was vacant and that meeting ended in a robbery. We included these cases, though, because our overall goal is to understand how places of worship are affected or touched by crime, and if a crime occurs on the grounds of such a property it could often affect the congregation even if it was not the direct victim in the crime (e.g., by increasing fear of crime in the neighborhood). However, in most the cases the place of worship and/or its members were the clear victim of the crime.
Identified articles came in two forms. The first were short, police ‘blotter’ entries that listed all the crimes in that community for a specific time period. Often one or two sentences, these articles usually did not provide much detail other than the type of crime and the place of worship’s name, although sometimes the latter was just ‘church.’ As an example of such a blotter entry: Larceny of a $5,000 three-ton Trane air conditioning unit was reported Friday at Flatrock Methodist Church.
Other articles were full newspaper articles, usually consisting of a few hundred words.
In most cases the decision to include an article was straightforward. However, as is always the case with data collection such as this, there were some cases that presented challenges. In these cases the undergraduate coders and the principal investigator discussed these cases to make a determination. For instance, a couple of articles mentioned a crime that involved places of worship but not in the context of any specific property. One article quoted a city councilperson as stating: ‘Churches have been vandalized, cars broken into and home burglaries have occurred. We need to have infrastructure and neighborhoods safe for residents in the city.’ Because this did not mention a specific place of worship, we did not include it in our data. We also excluded a couple of articles that took place on the property of congregations that were no longer active (i.e. the article stated that the church has been closed for months). Another special case concerned articles about clergy abuse and sexual assault. We did not include such articles unless it was clear that the assault occurred in the grounds of the place of worship. Most of these articles, though, were excluded because they were discussing crimes that occurred before 2012.
There were also a handful of articles that presented questions of whether it should be counted as a crime or not. As an example, one article discussed the death of a woman in the parking lot of a church in Texas after a van ‘went out of control.’ However, there was no mention of foul play or other sign that the authorities or anyone else viewed this as a crime, so this case was excluded. On the other hand, a different article from North Carolina discussed a man who was killed by a vehicle while he was cutting the lawn of a congregation’s property. In this case, though, the driver was charged with reckless driving and causing a death by motor vehicle. Because of this information it was determined that this was a crime, which we classified as a homicide/manslaughter/wrongful death. In cases of arson, we included the article if it stated that arson was ‘suspected’ or if there were ‘signs’ of arson.
Coding articles
A team of three undergraduate coders then began examining each eligible article to collect relevant variables from the text. The variables included the type(s) of crime the place of worship experienced, the congregational building’s name, its broad religious tradition (i.e., Christian, Jewish, Muslim, etc.), the location of the property, what was stolen (if a theft/burglary), and whether the article mentioned that the police were considering it a bias or hate crime.
In coding crime types we began by using the categories used in the violent and property crime indexes used by the FBI’s UCR. However, during the training and test coding phase of the project it was determined that most articles did not provide enough information to distinguish between burglary and larceny/theft. 1 As a result, we combined these into a general theft/burglary code. We also often did not have enough information to make more nuanced classifications, such as whether an assault would be considered aggravated or simple in nature. As a result, we simply included a general assault category.
There were 114 articles that did not provide the specific name of the place of worship. In many of these articles the congregation was simply referred to as a ‘church.’ We assumed in these cases that the congregations were Christian. However, there were eight cases among the 114 without a name that were specified as non-Christian (specifically, five Muslim, one Jewish, one Hindu, and one Buddhist congregation). We did not try to classify the religious tradition of the places of worship beyond the level of major world religions. While we thought there might be variations in crime experiences between a Christian church and a Muslim mosque, we did not have any theoretical reason to think that a Lutheran church would be different from a Presbyterian church. Furthermore, given that 18% of the entries did not have a specific place of worship’s name attached to them, we would have been faced with a substantial amount of missing data in creating a more specific denominational tradition measure.
Although we did collect information on the date of the crime, these dates often had to be estimated. Articles would often provide no specific date of crime or a vague estimate (e.g. ‘over the weekend’). Because of this imprecision, when examining temporal patterns below, we limit our analyses to the month the crime occurred as this broader time period is subject to less error.
After collecting all of our data we then systematically examined the data for duplicate entries, as sometimes an article would be a follow-up to a case that was coded in an earlier article. To do this cleaning we sorted the file by the name of the place of worship, the city, and other variables to identify duplicate entries. However, not all multiple entries were necessarily duplicate entries, as some places of worship were involved in different crimes over the course of the year. To make these determinations we would examine the original articles to see whether the articles were discussing the same crime multiple times or different crimes.
Results
In the end, we identified 624 unique incidents occurring on the property of places of worship. Based on our 10% sample of days, this would project to over 6,000 incidents for the year. Of course, this is undoubtedly an underestimate given that we are looking at newspaper reports and some crimes, especially minor ones, might not be reported in any way. Given this, we must be careful not to extrapolate some total number or rate of crimes occurring on the property of places of worship.
Table 2 shows the religious tradition of places of worship in our data. Overwhelmingly, the newspaper reports were for Christian churches. Just over 93% of our incidents involved Christian properties, and given that Christian churches represent just under 93% of US places of worship, this is almost perfectly proportional to the population of buildings. Muslim mosques represented the second most common religious tradition among our incidents (2.7%), followed by Jewish synagogues (1.8%). However, as we will discuss below, these numbers might hide some important differences in the types of crime taking place at these places of worship.
Distribution of religious traditions among place of worship crime incidents in 2012.
a2006–2007 National Congregations Study, w2 weight applied so percentages represent the congregational population and not the attendee population.
Table 3 shows the distribution of crimes that took place in these incidents. Note that each incident could involve multiple crimes (e.g. a simultaneous vandalism and burglary). In total, our 624 incidents involved 684 crimes. About a quarter of the incidents in 2012 involved vandalism, while 65% involved a theft or burglary. The most common target of theft or burglary was wallets, cash, jewelry, purses, and other smaller items (110 or 27% of the theft/burglary incidents). However, we were surprised to find that a common target of thefts and burglaries was copper, which was often stripped from congregations’ air-conditioning units, or in some cases the air-conditioning unit was stolen entirely. Of the theft or burglary incidents, 91 (22%) involved copper. After initially seeing this pattern in the articles we quickly discovered that this was a popular target because the price of copper had increased, therefore many criminals were stripping wiring out of buildings and air-conditioning units to sell as scrap metal. Clearly, this market-driven criminal activity affected places of worship as well. Many of these articles reported that the damage done by this particular crime often stretched into the thousands of dollars. Given that the median place of worship has about $1,000 in savings, such an expense could be devastating (Chaves, 2004: 36).
Distribution of crime types across place of worship crime incidents.
aIncludes incidents like trespassing and threats.
While most of the crime that took place on congregational property in 2012 consisted of relatively minor property crimes, places of worship were not immune from being touched by more serious crimes. Incidents involving arson amounted to 6%, while 2% involved a wrongful death. For the latter, the most well-publicized incident during 2012 was the shooting at the Wisconsin Sikh temple that was mentioned at the beginning of this article. There is evidence in our data that more serious crimes disproportionately occurred on the properties of religious minorities. For instance, as shown in Table 4, while only 6% of all incidents involved arson, 24% of incidents at Muslim mosques and 36% of incidents at Jewish synagogues involved arson. Similarly, Jewish, Mormon, and Muslim places of worship were more likely to be vandalized. On the other hand, these religious minority buildings were less likely to be victims of theft or burglary while the majority of crimes on the property of Christian churches were theft or burglary. All of this suggests that the crimes occurring on the property of religious minorities were motivated more by bias than economic desire or opportunity.
Distribution of crime types by religious tradition.
Interestingly, we discovered a temporal pattern among our incidents. Specifically, incidents involving Muslim mosques seemed concentrated in the months of Ramadan, the month-long fasting period that is a key practice of Muslims. As shown in Figure 1, of the 17 incidents involving Muslim places of worship, nine (53%) occurred in the two months that Ramadan spanned in 2012. It is likely that Ramadan receives its own news coverage, which then motivates those with an anti-Muslim bias to target these properties during this period.
Distribution of criminal incidents involving Muslim congregations by month in 2012.
Discussion
Places of worship are one of the most ubiquitous types of property found in communities, whether those communities are rural, suburban, or urban. While there has been some limited consideration of places of worship as the site of criminal activity, more expansive and systematic analyses have been absent from the literature. Based on our analysis, we found that crimes occurring on the property of religious congregations are a relatively rare but not insignificant event.
We acknowledge the potential issues with using newspaper reports of crime, but the data actually did not show some of the biases that we might have anticipated. For instance, we initially thought that newspapers might tend to disproportionately report exceptionally violent crimes or those involving the property of religious minorities, as these would be seen as more newsworthy. This does not appear to be the case just simply in terms of any mention of a crime. However, it is true that more serious and violent crimes did appear to receive more extensive coverage in terms of how much was written about them. For example, 60% of our incidents were reported as full newspaper articles versus 40% that were reported as short police blotter or log entries. However, 89% of incidents involving arson were reported as full articles.
Despite this difference in coverage type, the inclusion of police blotters appeared to balance out the overall frequency of crime across religious traditions and crime types. In fact, the frequency of crimes across religious traditions was almost exactly proportional to the US population of places of worship. About 93% of crimes in our newspaper data occurred at a Christian church, and about 93% of places of worship in the US are Christian. These overall percentages hide some significant differences in the types of crime experienced by different religious traditions, though. Our data showed that religious minority buildings appeared to be more likely to experience vandalism and arson, while Christian churches were more typically experiencing thefts.
The next step in this area of research would ideally be the inclusion of questions looking at experiences with crime on a representative national survey of places of worship. Such data would allow us to assess how much crime is unreported to police and/or in newspapers. Furthermore, such data would allow us to see how the characteristics of the congregation (i.e., size, wealth, crime prevention strategies) and the characteristics of its neighborhood shape its likelihood of experiencing some crime on its property.
Footnotes
Acknowledgements
The author would like to thank Tom Lano, Sara Kircher, Kia Her, Nicole Cornell, and Michelle Hansmann for their assistance and feedback on this research. The author was supported by a National Science Foundation grant while conducting this research (Law and Social Science Program-Award #1349738).