An Ecological Analysis of Alcohol-Outlet Density and Campus-Reported Violence at 32 U.S. Colleges (2024)

Abstract

Attention to the issue of campus violence was sparked by the 1986 rape and murder of Jeanne Clery, a 19-year-old woman at Lehigh University. Jeanne Clery's parents campaigned for greater transparency in campus violence statistics, arguing that, if they had known about violent-crime incidents at Lehigh in recent years, the family could have made a better decision regarding Jeanne's choice of college. The U.S. Congress passed the Crime Awareness and Campus Security Act in 1990, which required all colleges that participate in federal aid programs to provide campus crime statistics (Kassa, 1998). That law was amended in 1998 to strengthen and standardize reporting requirements and was renamed for Jeanne Clery at that time. The U.S. Department of Education now operates an online clearing-house that provides statistics on campus violence (Office of Postsecondary Education, 2000).

The relationship between alcohol and crime is well established in the research literature (Pernanen, 1991). The most recent statistics compiled by the U.S. Department of Justice's Bureau of Justice Statistics determined that, in the United States, 40% of violent crimes and two thirds of intimate-partner violence incidents are alcohol related (Greenfeld, 1998). Hingson and colleagues (2005) estimate that 600,000 college students are assaulted annually by another drinking student. Although the Hingson et al. estimate covers both on- and off-campus assaults, the Jeanne Clery Act requires the reporting of only on-campus crime. However, the Bureau of Justice Statistics report indicates that the role of alcohol in violence is greater for students living on campus and that alcohol-involved violence is more likely to occur on campus than off campus (Greenfeld, 1998).

A key environmental factor associated with violent crime is the physical availability of alcohol, which is typically measured by alcohol-outlet density. This relationship has been documented in several studies (Gorman et al., 2001; Lipton and Gruenewald, 2002; Norstrom, 2000; Reid et al., 2003; Scribner et al., 1995, 1999; Zhu et al., 2004). One explanation is that a higher density of alcohol outlets more often brings potential perpetrators and victims together in high-risk settings. An alternative explanation is that higher alcohol-outlet density leads to greater levels of alcohol consumption, which in turn are associated with more frequent violence.

Identifying a link between the physical availability of alcohol and campus violence is a logical extension of this past research. In this ecological study, we hypothesized that alcohol-outlet density in campus communities is associated with violence rates at the campus level, even after controlling for other independent predictors of campus violence. We also hypothesized that any relationship between alcohol-outlet density and violence rates would be mediated by student alcohol-consumption level.

Method

Results

The overall annual violence rate on campus was 0.57 offenses per 1,000 students (Table 1). Rape was the most common violent offense reported, with almost 0.32 offenses reported annually per 1,000 students. On-campus assaults and robberies were the least prevalent of the reported offenses, with annual rates of 0.16 and 0.09 per 1,000 students, respectively. The overall annual rate for burglary, a nonviolent crime, was 2.26 offenses per 1,000 students.

Alcohol-outlet densities within the 3-mile buffer around campus varied widely. On-premise outlets were by far the most common, ranging from 2.02 to 81.76 outlets per 1,000 students. Other covariates also varied widely between the campuses. The mean number of drinks in the past week ranged from 1.97 to 10.10 drinks, and the mean for the greatest number of drinks in one sitting in the last 2 weeks ranged from 2.19 to 6.88 drinks. On average, 5% of students were involved in fraternity/sorority activities for 6 or more hours per week; this proportion varied by campus from 0% to 15%.

Table 2 reports the results of regression analyses to test the effect of on-premise alcohol-outlet density on reported violence rates. Model 1 includes on-premise outlet density and the proportion of students who are male, who are White, and who participate in fraternity/sorority activities. Model 2 adds an individual drinking variable: the greatest number of drinks at one sitting in the last 2 weeks. Looking at the rates for all violent crimes taken together, on-premise outlet density and the proportion of students who were White were the strongest predictors. In Model 2, the effect of on-premise outlet density was reduced to nonsignificance in the presence of the student-drinking measure, suggesting that student drinking mediated the relationship between alcohol-outlet density and campus violent-crime rates. A similar effect was noted for the proportion of White students on campus. This suggests that the relationship between the proportion of White students and violent-crime rates is also mediated by student drinking.

When the overall campus violence rates were disaggregated into rape, assault, and robbery rates, the models for rape explained the greatest amount of variance, with R² ranging from .57 to .61. Model 1 shows that campus rates for rape are positively associated with on-premise outlet density, the proportion of students who are male, and the proportion of students who are White. Adding the student-drinking measure increased the amount of variance explained and reduced the effect for on-premise outlet density, again suggesting that student drinking mediated the outlet-density/crime-rate relationship.

Assault, robbery, and burglary offenses were not well explained by the models. A significant effect for student drinking on assaults was observed, but no strong effect was observed for robbery. In addition, the assault rate was negatively associated with the proportion of students who are White. Burglary, a property crime, was better explained by the models than assault or robbery (see the adjusted R²s in Table 2). On-premise outlet density and fraternity/sorority participation were strongly associated with the burglary rate, but adding the student-drinking measure did not improve the amount of variance explained.

Findings for off-premise outlets are presented in Table 3. For some offenses, the associations between off-premise outlet density and violence were similar but weaker than those found for on-premise outlet density. Off-premise alcohol-outlet density was not associated with robbery, burglary, or assault rates but did show a positive association with the rape-offense rates, as was seen with on-premise outlet density. Adding the student-drinking variable into the models revealed a strong relationship between drinking and rates for rape, assault, and burglary. Looking at rape, introducing the student-drinking variable reduced the effect of off-premise outlet density, just as in the model for on-premise outlet density.

Discussion

Given the established association between college student alcohol use and interpersonal violence (Abbey et al., 2001; Harford et al., 2003; Ullman and Brecklin, 2000), it was not surprising that, at the campus level, mean student alcohol consumption was strongly associated with the overall violence rate. Our findings extend this literature by demonstrating that a campus-level measure of the physical availability of alcohol—namely, the number of on-premise or off-premise alcohol outlets within 3 miles of campus—also was associated with the overall violence rate. Also noteworthy is that introducing the student-drinking measure into the models substantially reduced the association between outlet density and the overall violence rate. This finding suggests that campuses with higher alcohol-outlet density have higher drinking levels, which in turn explains their higher rates of violence. It should be noted that this is an ecological study, and these individual-level inferences are not intended to suggest a causal mechanism. After controlling for individual drinking, the strength of the bar-density/ violence association was reduced in this analysis, consistent with the simple explanation that more bars per capita lead to more violence through the mechanism of increased drinking. Individual drinking is then an intermediate variable in the causal pathway. Another way of conceptualizing these relations, however, is as a cross-level interaction, where the campus context operates by regulating or modifying risk (Glass and McAtee, 2006). Higher outlet density might lead to more drinking among those who do drink, thus increasing the likelihood of violence for a given proportion of drinkers or presenting more venues in which a drunken perpetrator commits a crime or a victim falls prey to violence.

Also of note was the varying profile of study variables associated with the individual violence outcomes. Rape-offense rates were positively associated with student drinking, alcohol-outlet density, the proportion of male students, and the proportion of White students. Higher levels of student drinking are associated with alcohol-outlet density (Scribner et al., 2008; Weitzman et al., 2003), and White and male students tend to be the heaviest drinkers (Task Force of the National Advisory Council on Alcohol Abuse and Alcoholism, 2002). Because alcohol use is involved in a high percentage of campus sexual assaults (Abbey et al., 2001; Baum and Klaus, 2005), the association of these variables with rape-offense rates is not surprising. Note that, when the alcohol-consumption measure was added to the model, the effect of outlet density was no longer significant, suggesting that alcohol outlets are contributing to higher levels of alcohol consumption and do not serve as a significant independent situational risk factor for sexual assault. In contrast, studies of alcohol-outlet density in the community context have suggested that the association of alcohol-outlet density with violence may be, in part, the result of the situational risk that alcohol outlets pose by bringing victims and perpetrators together in a high-risk setting, rather than by increasing drinking levels (Gruenewald and Remer, 2006; Scribner et al., 1995).

Assault-offense rates were positively associated with student drinking but negatively associated with the proportion of White students and not at all associated with alcohol-outlet density. This pattern of results was unexpected. Contrary to studies of alcohol-outlet density in the community context (Felson, 1997; Scribner et al., 1995; Zhu et al., 2004), we found no association between outlet density and assault rates. It should be noted, however, that the reported campus assault rates were surprisingly low, with the average rape-offense rate being twice the average assault rate. Given this apparent underreporting, the study design may not have been sensitive enough to detect the expected relationship between outlet density and the assault rate.

Surprisingly, the proportion of White students was negatively associated with campus assault rates while being positively associated with the rape-offense rates. In a follow-up analysis (not shown), we found a negative relationship between the proportion of White students and overall crime rate for the metropolitan statistical area where each campus was located. This finding suggests that the negative association between the proportion of White students and campus assault rates may be explained by overall community crime rates.

There are several study limitations that should be considered when interpreting these results. First, the study is ecological, and therefore caution should be taken in drawing inferences from the campus to the individual level. Consequently, the relationships described here can be considered only suggestive of an environmental effect on individual drinking and violent behavior. Validation of these findings will require larger studies using multilevel and longitudinal designs to adequately model the person-in-environment fit and to establish the directionality of the relationships.

Second, several characteristics of the neighborhoods surrounding the college campuses were not considered (e.g., the socioeconomic status of the campus neighborhood). This is largely a result of the restricted number of variables that can be studied using a sample of only 32 campuses. Consequently, larger studies involving more campuses will be required to assess the effects of additional environmental variables.

In addition, the student data were aggregated over multiple years, yet the determination of alcohol-outlet densities at each campus was based on 2004 data. Clearly, there is the potential for bias if the number of alcohol outlets changed during the study period before or after the 2004 determination of outlet densities. Furthermore, student drinking is likely to be underreported, and therefore the full impact of potential mediation by student drinking may have been reduced by any underreporting.

Finally, although the campus violence data were obtained from the same source, the reliability of those data depends on whether individual campus authorities are following the same reporting procedures. A study evaluating the possibility of reporting bias in the U.S. Department of Education data set has not been conducted.

Colleges and universities around the United States have implemented a wide range of campus-focused programs and policies to reduce student drinking and thereby decrease alcohol-related outcomes such as violence. This study adds to the growing evidence that community-level efforts are equally important (Holder, 2000, 2002; Toomey et al., 2007). Zoning regulations that limit the number of bars, mandate responsible beverage service, and mandate strict adherence to underage drinking laws have all been demonstrated effective in community-level interventions and should be embraced by campus officials (Toomey and Wagenaar, 2002).

Footnotes

^*This research was supported by National Institute on Alcohol Abuse and Alcoholism grants R01 AA014679 and R01 AA12471.

References