If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to: Jodi M. Gatley, M.P.H., Northern Medical Program, University of Northern British Columbia, 3333 University Way, Prince George, British Columbia V2N 4Z9, Canada.
Northern Medical Program, University of Northern British Columbia (UNBC), Prince George, British Columbia, CanadaHuman Brain Laboratory, Centre for Addiction and Mental Health (CAMH), Toronto, Ontario, CanadaDalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, CanadaInstitute for Mental Health Policy Research (IMHPR), Centre for Addiction and Mental Health (CAMH), London, Ontario, CanadaDepartment of Epidemiology and Biostatistics, Western University, London, Ontario, CanadaSchool of Psychology, Deakin University, Geelong, Victoria, Australia
Northern Medical Program, University of Northern British Columbia (UNBC), Prince George, British Columbia, CanadaHuman Brain Laboratory, Centre for Addiction and Mental Health (CAMH), Toronto, Ontario, CanadaDalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
Sexual-assault crimes, primarily perpetrated by males against female victims, impose a substantial burden on societies worldwide, especially on youth. Given that approximately half of all sexual assaults involve alcohol consumption by the perpetrator or victim, it is reasonable to expect that minimum legal drinking age (MLDA) restrictions might have an effect on sexual-assault patterns. Canadian MLDA laws are 18 years in Quebec, Alberta, and Manitoba (MLDA-18), and 19 years in the rest of the country (MLDA-19). The present study assesses whether MLDA laws might have an impact on sexual-assault crimes.
Methods
A regression-discontinuity design was applied to sexual-assault crime data (n = 12,980 incidents) from the national Uniform Crime Reporting survey 2009–2013, a population-level registry of all police-reported crimes in Canada. Uniform Crime Reporting data does not include an explicit alcohol involvement indicator.
Results
Nationally, in comparison to males slightly younger than the MLDA, those just older had significant and immediate increases in sexual-assault perpetration of 31.9% (95% confidence interval: 8.7%–54.5%, p = .007). In MLDA-19 provinces, there was an immediate post-MLDA increase of 56.0% (95% confidence interval: 18.9%–90.8%, p = .004) in sexual-assault crimes by males just older than 19 years, whereas in MLDA-18 provinces no significant effect was found. For females, there was no evidence of MLDA effects on sexual-assault crimes.
Conclusions
Release from Canadian MLDA law restrictions was strongly associated with increases in sexual-assault perpetration by young men. These findings lend support to the potential effectiveness of population-level alcohol control policies for sexual-assault prevention among youth.
This study provides evidence that release from minimum legal drinking age laws is associated with increases in the perpetration of sexual-assault crimes by young men. These findings lend support to the potential effectiveness of population-level alcohol control policies for sexual-assault prevention among youth.
Sexual-assault crimes, primarily perpetrated by males against female victims, impose a substantial burden on societies worldwide, especially on youth [
World Health Organization Global and regional estimates of violence against women: Prevalence and health effects of intimate partner violence and non-partner sexual violence.
]. In Canada, sexual assault of young women, especially postsecondary students, has become a prominent national issue, and many Canadian colleges and universities have been criticized for lacking comprehensive policies in this area [
Most of the literature estimating the prevalence of sexual-assault victimization and perpetration among youth draws on postsecondary student populations in the United States and Canada. Prospective surveys of male college students in the United States have found relatively high prevalence of self-reported engagement in sexually violent or sexually coercive behaviors of 10%–15% annually [
American College Health Association American College Health Association-National College Health Assessment II: Canadian Reference Group Data Report Spring 2013.
American College Health Association,
Hanover, MD2013
American College Health Association American College Health Association-National College Health Assessment II: Canadian Reference Group Data Report Spring 2013.
American College Health Association,
Hanover, MD2013
] reported sexual-assault victimization in the past year. In the United States, a 2011 national survey reported that 19.3% of women had experienced rape in their lifetime, with 78.7% of first instances happening before the age of 25 years [
Prevalence and characteristics of sexual violence, stalking, and intimate partner violence victimization–National Intimate Partner and Sexual Violence Survey, United States, 2011.
]. Recent large-scale U.S. college surveys have found that up to 18.9% of undergraduate females had been a victim of any sexual assault since starting university [
]. In addition, a recent U.S. study found disturbingly high prevalence of sexual assault among college women—11.4% experienced attempted rape, and 8.5% experienced completed rape during their first year of studies [
] supports that sexual-assault perpetration is influenced by many factors, including traumatic childhood experiences, personality traits, endorsement of violent and/or rape-supportive attitudes toward women, abusive dating relationships, and sexual experiences in adolescence and adulthood; however, alcohol may also be an important contributing factor. Alcohol use by the victim, perpetrator, or both is a feature of approximately half of sexual assaults [
]. Alcohol is thought to interact with numerous individual and situational mediating factors to increase the risk of perpetration of sexual violence among males who were already predisposed to these behaviors [
]. Some such mediating factors from the cross-sectional survey and alcohol administration literature include individual traits of impulsivity, narcissism, reduced empathy, delinquency in adolescence, hostility toward women, engagement in casual sex, and perceived peer support or pressure [
]. In addition, young adult drinkers spend time in risky drinking situations, such as bars and parties, where they go to meet potential sexual partners, and where unwanted sexual advances are common [
]. For example, a recent survey of women recruited in a bar district in a mid-sized Canadian city found that over 50% reported experiencing sexual harassment or sexual aggression on a single night out [
Minimum legal drinking age (MLDA) legislation is a primary population-level alcohol control policy limiting youth access to alcohol. In Canada, an expert group recently endorsed increasing the MLDA nationally to 21 years of age—from 18 years of age in Québec, Manitoba, and Alberta, and 19 years of age in the rest of the country. It is reasonable to expect that release from MLDA restrictions will be associated with increases in sexual assault crime perpetration by youth. According to the microlevel model proposed by Ullman [
], alcohol consumption by the perpetrator does not independently contribute to sexual-assault perpetration, but rather interacts with a complex set of behavioral and situational characteristics (e.g., drinking context) which mediate the impact of drinking on the probability of perpetrating sexual assault. Prior research not only has demonstrated significant and immediate increases in population-level drinking directly after the MLDA [
] but also changes in drinking context, with more frequent drinking at bars and night clubs—venues associated with increased likelihood of alcohol-related violence [
The potential impacts of MLDA laws on population-level patterns of sexual assault have rarely been investigated: one study showed no relation between drinking age laws and criminal perpetration of sexual-assault crimes in California [
], and another indicated that raising the MLDA in New York State in 1985 was associated with increases (rather than expected decreases) in sexual-assault victimization in the newly alcohol-restricted age groups [
]. The present study aimed to assess the potential impacts of Canada's MLDA legislation on patterns of sexual-assault perpetration by men toward women reported to the police. Guided by Ullman's model of sexual-assault perpetration [
], we expected that in comparison to Canadian males slightly younger than the MLDA, those males just older than the MLDA would manifest abrupt and significant increases in their perpetration of sexual-assault crimes.
Methods
This study received ethical approval from the Research Ethics Boards of the Centre for Addiction and Mental Health and the University of Northern British Columbia.
Data source
Police-reported crime data were obtained from the Uniform Crime Reporting (UCR) Incident-based survey, 2009–2013 [
]—a national database held and managed by the Canadian Centre for Justice Statistics (CCJS), Statistics Canada. The UCR is a national population-level registry of police-reported crimes that have been substantiated through investigation from all federal, provincial, and municipal police services in Canada. Standardized reporting of criminal incidents to the UCR from their administrative records is mandatory for all Canadian police services; thus, UCR data capture virtually 100% of police-reported criminal incidents in Canada. The UCR does not contain an indicator variable capturing alcohol or drug use by perpetrators or victims.
Data collection and data processing: Uniform Crime Reporting survey
Initial and ongoing training and support is provided by the CCJS to each police department in Canada to ensure the correct completion of UCR survey forms [
]. UCR survey data are submitted to the CCJS central data processing system by trained police personnel following a standardized submission process, where more than 300 intensive checks and edits of all responses from police department respondents are performed. Police departments are contacted by the CCJS to resolve all edit failures or data anomalies. Each police service is required to approve the annual department-specific UCR summary data tables as accurate and complete upon completion of the year end verification process and before final annual release by the CCJS. The final, annual master Statistics Canada UCR data file is considered the gold-standard crime-data resource in Canada, and it is used in national governmental reports on crime [
], the aggregated count data in the present study were based on the most serious offense in each incident. For incidents involving multiple violations, the most serious violation is determined by the reporting police department according to the standardized scoring guidelines [
]. Our major sexual-assault crime outcome variable aggregated all major sexual-assault crimes incidents, including sexual assault level I, level II, and level III (see Table 1) [
Coding categories of UCR codes [25] were constructed based on the Criminal Code and prior research using CCJS criminal incident database [27] and represent an integration of both categorization schemes.
Subcategories
Violation code in UCR 2 database
Sexual assault
Sexual assault
1310—Aggravated sexual assault—level 3: Sexual assault resulting in serious physical injury of the victim or endangering their life. 1320—Sexual assault with a weapon, threats to a third party, or causing bodily harm—level 2: Sexual assault with a weapon, threats, or causing injury. 1330—Sexual assault—level 1: An assault of a sexual nature resulting in violation of the sexual integrity of the victim, minor, or no physical injury of the victim.
Cannabis-related offenses (control crime outcome)
Possessing and trafficking of cannabis (e.g., possession schedule II cannabis or cannabis derivative, trafficking schedule II cannabis, and intent to traffic schedule II cannabis, and so forth).
4140—Possession schedule II (cannabis); possession schedule II (cannabis derivative); possession schedule II (cannabis ≤ schedule VIII) 4240—Trafficking schedule II (cannabis); intent to traffic schedule II (cannabis); trafficking schedule II (cannabis ≤ schedule VII); intent to traffic schedule II (cannabis ≤ schedule VII)
Given that our statistical analyses employed the quasi-experimental regression-discontinuity (RD) approach, it was important to demonstrate, if possible, that MLDA laws might show no evidence of impacts on a crime outcome unrelated to alcohol consumption. We chose cannabis-related crimes as a “quasi-control” crime series, given that cannabis-related crimes are unlikely to be affected directly by alcohol consumption and that prior research using a RD approach has demonstrated no evidence of a significant relation between MLDA laws and cannabis consumption [
We chose to analyze counts—rather than rates (which would be based on counts in the numerator and population statistics in the denominator)—for two reasons: the denominator for our primary age in weeks running variable cannot be calculated (or imputed accurately) from population statistics; and the use of census-based data for the calculation of rates might bias the estimate of the “jump” at the MLDA. We limited the analysis to 208 weeks (approximately 4 years) before and after the MLDA (14–21 years or 15–22 years), primarily because the incident counts become very small for ages earlier than 14 years.
Perpetrator's age
We calculated the age of the perpetrator at the time of the offense in terms of age in weeks. Such a finely graded age variable allowed us to estimate relatively precise regression models of criminal incident count patterns leading up to and following the MLDA.
]. The RD design takes advantage of the sharp discontinuity in the legality of alcohol purchasing and consumption appearing at the MLDA: our approach assigns individuals younger than the MLDA to the “alcohol-restricted” group and young adults no longer subject to the MLDA to the “alcohol-accessible” group. The primary, intuitive idea of the RD approach is that individuals slightly older than the MLDA and those slightly younger than the MLDA will be similar on observed (and unobserved) characteristics—except for the influence of the removal of the MLDA in the alcohol-accessible group. The RD design assumes that all observed and unobserved variables (which might influence the alcohol-related outcomes) are smoothly distributed across the age cutoff [
]. All RD analyses were stratified by gender, provincial MLDA of 18 years of age (MLDA-18), 19 years of age (MLDA-19), as well as at the aggregated national level.
Primary regression-discontinuity analyses: Nonparametric local linear regression
In our primary nonparametric local linear regression analyses, we used a recent approach [
]—to fit the robust RD models with bias correction. An important strength of local regression is that the approach is relatively robust to nonlinearities and allows for data modeling without making assumptions about the functional form of the model. This is a key advantage of local regression as compared with traditional parametric regression, where one usually needs to test higher order polynomial terms, a practice which recently has been criticized [
], the primary local linear regression models were rerun without the MLDA birthday week values. This approach accounts the effect of MLDA birthday celebration in the construction of the post-MLDA regression line and the calculation of the potential “jump” in crime immediately following the MLDA.
Gender-specific analyses
Based on prior literature showing strong gender-specific patterns of sexual-assault perpetration and victimization, including alcohol-related sexual assault, models were constructed by perpetrator gender.
Sensitivity analyses
To assess the reliability of our primary local linear regression models with triangular-kernel weighting, we undertook a series of local regression models with varying weighting approaches (i.e., triangular kernel vs. uniform kernel), functional forms (i.e., linear vs. quadratic), and bandwidth selection strategies, as recommended in the recent best practices guide for conducting RD analyses [
]. In addition, to assess the reliability of our primary results, we used linear parametric regression models with a restricted range of 26 points (age in weeks) before and after the MLDA to model the impact of the MLDA on the major crime outcome categories.
Results
Descriptive results
The descriptive results included a total of 13,018 sexual assault incidents and 183,832 (adjusted method) cannabis-related incidents (see Table 2). The relative proportions of sexual incidents by province of occurrence and perpetrator gender are presented in Table 3. The majority of both crime outcomes across Canada were committed by male perpetrators: 96.7% of sexual assaults and 85.5% of cannabis-related offenses. Table 2 presents the proportions of sexual-assault incidents in the data set by province.
Table 2The number and proportions of sexual assault incidents in Canada by province and perpetrator gender, 2009–2013
Table 3Regression-discontinuity results for local linear regression models with triangular-kernel weighting across categories of crime in Canada, 2009–2013: sexual assault, other sexual violations, and cannabis-related violations
All estimates are based on the results from the local linear regression modeling with triangular-kernel weighting, using data-driven selection of the optimal bandwidth.
Analyses with the birthday week data removed were also conducted to isolate the impacts of MLDA laws on crime outcomes from potential effects of heavy birthday drinking [21]. Given that the running variable as perpetrator age in weeks, the sample sizes given here are standardized values for week 20 values. To account for the additional 1 or 2 days each year beyond 52 weeks, these extra events were added to the total incidents for week 20 of each year of age and multiplied by 7/8.25.
a All estimates are based on the results from the local linear regression modeling with triangular-kernel weighting, using data-driven selection of the optimal bandwidth.
b Data were aggregated across provinces with a MLDA of 18 years old (i.e., Quebec, Manitoba, and Alberta).
c Data were aggregated across provinces with an MLDA of 19 years old (the remaining 7 provinces).
d Data were aggregated across all 10 Canadian provinces.
e The sample size n is the number of drivers used in the analysis.
f Analyses with the birthday week data removed were also conducted to isolate the impacts of MLDA laws on crime outcomes from potential effects of heavy birthday drinking
. Given that the running variable as perpetrator age in weeks, the sample sizes given here are standardized values for week 20 values. To account for the additional 1 or 2 days each year beyond 52 weeks, these extra events were added to the total incidents for week 20 of each year of age and multiplied by 7/8.25.
The main RD results are presented in Table 3. Nationally, in comparison to males just younger than the MLDA, those just older had an increase in perpetration of sexual-assault crimes of 31.9% (95% confidence interval: 8.7%–54.5%, p = .007). There were no changes in sexual-assault crimes by female perpetrators. Among males in MLDA-19 provinces, there was an immediate increase in sexual-assault perpetration by those just older than 19 years of 56.0% (95% confidence interval: 18.9%–90.8%, p = .004). There were no significant effects for males in MLDA-18 provinces, likely due to lack of statistical power. Also, there were no significant associations for females in MLDA-18/19 provinces. In addition, no significant changes were seen in cannabis-related offenses, the quasi-control condition, for males or females.
The analyses replicating the primary local regression results—but excluding the MLDA birthday week values—demonstrated somewhat attenuated effect sizes but similar patterns of statistical significance as those indicated in our primary local regression models.
The number of sexual-assault incidents across the MLDA nationally for males and females are presented in Figure 1. This figure visually demonstrates significant “jumps” immediately following the MLDA among males for sexual-assault crimes. The results for the quasi-control category, cannabis-related violations, are presented in Figure 2.
Figure 1Number of sexual assault crime incidents in Canada (2009–2013) by age in weeks leading up to and following the minimum legal drinking age: males and females.
Figure 2Number of cannabis-related offenses in Canada (2009–2013) by age in weeks leading up to and following the minimum legal drinking age: males and females.
In comparison to the primary local regression results, the estimated MLDA effects at the national level in the sensitivity analyses for males and females were generally similar in terms of size and statistical significance for local regression models with linear or quadratic forms and either uniform- or triangular-kernel weighting schemes and linear parametric models using 26 data points before and after the MLDA.
Discussion
Our findings demonstrate that MLDA laws in Canada may have a substantial impact on the commission of sexual-assault crimes by young men. In comparison to males slightly younger than the MLDA, those just older than the MLDA had significant and abrupt increases in incidents of police-reported sexual-assault perpetration both nationally and in MLDA-19 provinces.
The present study provides the first estimates of the impact of Canadian MLDA laws on perpetration of sexual-assault crimes. At least in our understanding, only one prior US-based study has assessed the specific effect of MLDA laws on sexual offending, and the results showed no evidence of an association between MLDA laws and sexual offending [
]. However, the project's null findings may have been influenced by the lack of gender-specific analyses, a narrower definition of sexual-assault crimes (i.e., rape), and a potential lack of statistical power to detect an effect [
There is compelling evidence to support the expectation that population-level alcohol control strategies reduce the burden of alcohol-related sexual aggression. Experimental and survey studies of sexual assault have established alcohol consumption by male perpetrators and female victims as one of a number of consistent, contributing factors in sexual-assault events [
]. Our findings contribute to this experimental and survey-based literature by demonstrating the potential utility of MLDA laws and other alcohol policies to reduce alcohol-related sexual assault, a possibility that has rarely been discussed in the literature. We are only aware of one recent study suggesting that college-wide alcohol restrictions are associated with reductions in institutional sexual-assault reports [
] have identified sexual-assault victimization of youth as a significant issue, likely not only due to its high prevalence among young people but also because of the severe psychiatric and physical sequelae associated with sexual victimization [
World Health Organization Global and regional estimates of violence against women: Prevalence and health effects of intimate partner violence and non-partner sexual violence.
]. Policy changes at the level of postsecondary institutions may have the potential to influence youth sexual assault, given that a large proportion of young people in Canada enroll in postsecondary education (74.7% had accessed either college or university by the age of 21 years in 2006) [
]. Postsecondary students may also be more likely to binge drink than their peers who are not students and consume alcohol in contexts associated with heavier drinking [
]. The current results suggest that population-level alcohol control strategies may have an impact on sexual-assault perpetration. At this time, it may be important for policymakers to incorporate alcohol control strategies in the recent calls for sexual-assault laws and policies in Canadian universities and colleges [
]. Few Canadian universities or colleges have incorporated alcohol control issues into their sexual-assault policies and doing so may prove to be an effective means to reduce sexual-assault perpetration victimization among the large proportion of Canadian young people attending these institutions.
Given the study findings, it is plausible to expect that raising Canada's drinking age laws would result in decreases in male sexual-assault perpetration crimes in newly restricted age groups. According to the Ullman model [
], it seems reasonable to predict that raising the MLDA would reduce sexual-assault crimes in the population by attenuating the risk factors of hazardous alcohol consumption at the population level and drinking in risky settings (such as bars and night clubs) where sexual-assault crimes were much more likely to occur. However, there are likely to be significant political challenges and resistance to raising the MLDA in Canada [
], and it may be important for policymakers to employ other strategies to implement our findings, which indicate that a primary youth alcohol control strategy (i.e., the MLDA laws) can have an impact on sexual-assault perpetration.
This study had several possible limitations. Criminal incidents where perpetrators and/or victims had consumed alcohol could not be identified as this information was not recorded in the UCR database. The main outcome measure—police-reported sexual-assault incidents—likely under-represents actual counts of sexual-assault incidents; in Canada, it is estimated that only about 5% of sexual assaults are reported to police [
]. In addition, the RD approach assumes that potentially confounding variables are smoothly distributed across the MLDA cutoff. However, youth who have reached the MLDA, while gaining access to alcohol, also simultaneously gain access to commercial drinking locations (e.g., bars and nightclubs), which are key locations for unwanted sexual advances, sexual harassment, and aggression [
]. Sexual aggression in these locations can be attributed to several factors, including availability of victims, lowered perception of consequences, and concerns with masculinity and self-identity [
]. Crime setting could not be examined in the present study as it was not included in the data set requested. Thus, the present findings may at least partially reflect the effects of exposure to licensed establishments on sexual assault as a result of reaching the MLDA. Some male perpetrators use alcohol as a tool or means to intoxicate their victim, but the present study does not have any information about that process in the criminal incident records. It was also not possible to determine the perpetrator-victim relationship since this variable is not part of the UCR crime file. Due to the need to maintain confidentiality, data were aggregated and could not be examined for trends in the number of incidents over the study span. It is important to note that it is unlikely that year-over-year trends would affect the discontinuity observed at the MLDA, unless there were factors which affected differently those just younger and older than the MLDA. We are not aware of any evidence for such factors during the study span. In addition, the UCR does not collect detailed information about demographics or socioeconomic characteristics such as ethnicity, race, income, or level of education and, as a result, these variables could not be included in our modeling. The present study findings directly relate to the Canadian context, and the results may not extend to other regions or countries with different cultural norms, expectations, or laws regarding sexual-assault perpetration or alcohol consumption.
In sum, the present study demonstrated that release from Canadian drinking age restrictions was associated with sharp and immediate increases in sexual-assault perpetration among men. These findings demonstrate the potential utility of population-level alcohol control strategies to limit sexual-assault violence among young people, and the results can also contribute to current debates about the effectiveness of drinking age laws on reducing alcohol-related harms among young people in Canada.
Acknowledgments
All the authors contributed significantly to the preparation of the current manuscript. The corresponding author affirms that all individuals who have contributed significantly to the preparation of this manuscript have been mentioned in the Acknowledgments. An initial version of this research has also been presented by the corresponding author at the 27th Annual Association for Psychological Science Convention, May 21–24, 2015 in New York.
Funding Sources
The research described in this study was funded by a Canadian Institutes of Health Research (CIHR) fellowship to the first author (J.M.G.) and a research grant (MOP # 133699) from CIHR to the senior author (R.C.C.).
References
World Health Organization
Global and regional estimates of violence against women: Prevalence and health effects of intimate partner violence and non-partner sexual violence.
Prevalence and characteristics of sexual violence, stalking, and intimate partner violence victimization–National Intimate Partner and Sexual Violence Survey, United States, 2011.
30146-5&id=gr1.jpg){kind=link}
30146-5&id=gr2.jpg){kind=link}