«Review of Prevalence and Risk factors related to Sexual Assault: Prepared for the Harvard Sexual Assault Task Force Lisa F. Berkman, PhD Amy ...»
Review of Prevalence and Risk factors related to Sexual Assault:
Prepared for the Harvard Sexual Assault Task Force
Lisa F. Berkman, PhD
Amy Ehntholt, MA, SM
And members of the Research Report Subcommittee of the Harvard Sexual Task Force
In this report, we have aimed to review the literature on the prevalence and risk factors related to
sexual assault, limited for the most part to college campuses. We examine issues related both to victims
of assault and to perpetrators, and identify “proximate” as well as “distal” risk factors. Building on frameworks related to characteristics not only of individuals but also of the environment or social context, we review the evidence linking these conditions to a range of outcomes associated with sexual assault. This report does not review the evidence on programmatic efforts to reduce sexual assault on college campuses.
Our review suggests that there are several major limitations in most of the literature to date. Perhaps most important are the low response rates to surveys (with a few exceptions), which limit our ability to generalize results to a broader population base of students on college campuses. The response rates make it difficult to identify sources of bias, which would allow us to have more confidence that the associations we have found would be the same in a more representative sample of college students.
We alert the reader to actual response rates for each of the studies mentioned in this report.
Prevalence of Sexual Assault The true prevalence of sexual assault is still difficult to state with confidence because of frequently low response rates to surveys, framing of questions using variable definitions, potential self-report bias (which can result in both under- and over-reporting), and varying timeframes used in assessments. Here we describe the prevalence for some of the most commonly studied outcomes that generally fall under the umbrella of sexual assault, and we evaluate the quality of the evidence.
In our review, we identify types of assault, prevalence and estimates by type of assault, and problematic issues with methods and study design. In the next section, using a parallel approach we discuss the same issues related to perpetrators.
I. Types of Sexual Assault Sexual assault can range from unwanted sexual touching only (usually labeled “battery,” as in the Campus Sexual Assault study) to rape. These outcomes can be classified by: relationship between perpetrator and victim (stranger vs acquaintance); whether the act was completed or just attempted;
and how the act was achieved (e.g., forcibly, by drugging, verbal coercion…). Rape can be broken down further into the following types (this breakdown and these definitions are taken from the Medical University of South Carolina (MUSC) study, 2007 (Kilpatrick et al.’s Drug-facilitated, Incapacitated, and
Forcible Rape: A National Study):
Drug and alcohol facilitated rape (DFR): The perpetrator deliberately gives the victim drugs without her permission or tries to get her drunk, and then commits an unwanted sexual act against her involving oral, anal, or vaginal penetration. The victim is passed out or awake but too drunk or high to know what she is doing or to control her behavior.
Incapacitated rape (IR): Unwanted sexual act involving oral, anal or vaginal penetration that occurs after the victim voluntarily uses drugs or alcohol. The victim is passed out or awake but too drunk or high to know what she is doing or to control her behavior.
Forcible rape (FR): Unwanted sexual act involving oral, anal or vaginal penetration. The victim also experiences force, threat of force, or sustains an injury during the assault. In cases where FR includes elements of DFR, we categorized the incident as DFR.
(DFR and IR are, by definition, mutually exclusive.) II. Prevalence Estimates by Type Sexual Assault (Overall) Recent studies’ estimates for completed and attempted SA (combined) generally range from 20% to 25%. Most often cited (and the source for the White House’s “1 in 5” stat), the 2007 Campus Sexual Assault (CSA) study (Krebs et al.) revealed that 19.7% of its 5,446 female survey respondents reported completed or attempted sexual assault since the start of their freshman year. When restricted to seniors only (much smaller sample size), the figure was slightly higher: 19.8%. Nearly 14% of respondents reported a completed sexual assault since starting college; 12.6% reported attempted assault (and 7.2% reported both). The majority of these SA cases met legal definition of rape in most states. Problems with CSA numbers include: generalizability (based on samples from 2 large public universities, 1 in South, 1 in Midwest); and relatively low response rate (42%). The initial focus of the CSA study was not to attain a national estimate, but rather to assess the prevalence of drug-facilitated SA. However, an older study with nearly twice the response rate—the U.S. Department of Justice’s Sexual Victimization of College Women study (Fisher et al., published in 2000)—had similar findings: One in 36 (1.7%) women reported a completed rape since the start of the school year (survey administered between March and May of 1996), so that "over the course of a college career—which now lasts an average of 5 years — the percentage of completed or attempted rape victimization among women in higher educational institutions might climb to between one-fifth and one-quarter." In another study by Fisher et al. (1998) of a nationally representative sample of undergraduate and graduate students enrolled full- or part-time (n=3,472), the investigators found a reporting rate of 16.4 per 1,000 for completed SA, and 5.2 per 1,000 for attempted SA. Among a national random sample (n=1,835) of Canadian college and university students (DeKeseredy and Kelly, 1993), prevalence of reported sexual abuse in the last year was 27.4%, while sexual abuse since leaving high school was even higher: 44.5%. In a study focusing on sexual assault at historically black colleges, Krebs et al. (2011) found that 9.7% of undergrad women surveyed in a web-based questionnaire reported experiencing completed sexual assault since entering college, "…considerably lower than the comparable rate obtained from undergraduate women at non- HBCUs (13.7%). This difference,” the authors note, “seems to be associated with differences in alcohol-use frequency."
While rape has been the main focus of most research, the CSA did also investigate unwanted touching:
1.4% of women reported physically forced sexual battery, 2.6% incapacitated sexual battery. In the slightly older (Fisher et al., 2000) Sexual Victimization of College Women study (with a very high response rate of 85%) of 4,446 college women surveyed, nearly 9% reported having experienced unwanted sexual contact within the past year.
Rape (non-specific) Probably the strongest recent overall national (not limited to university settings) data on rape come from the Centers for Disease Control’s National Intimate Partner and Sexual Violence Survey (“NISVS,” Black et al., 2011), which reported that almost 1 in 5 women (18%) are raped during their lifetime (and 44.6% are subjected to sexual violence other than rape). The same survey found evidence that 1 in 7 men experience attempted or completed rape during their lifetime. As stated above, close to 14% of female participants in the CSA reported completed SA since starting college (the majority of these were rapes). In Kilpatrick 2007 Dept. of Justice/MUSC report, 5.14% of undergraduate female respondents reported past year rape. The U.S. Department of Justice’s NCWSV, published in 2000, found that 2.8 percent of college females had experienced either a completed (1.7 percent) or an attempted (1.1 percent) rape within a 7-month timeframe, the equivalent of a rate of about 27.7 rapes per 1,000 female students. Extrapolating from these numbers over this time period, the authors point out in the study, the figure would grow to 5% over a 1-year period. And multiplying this number over a 4 or 5-year period brings the estimate up to 20-25%. Again, this last study’s very high response rate (of roughly 85%) lends it a certain power. Nine in 10 victims (of both completed and attempted rape) reported knowing their offenders. Nearly 13% of completed rapes, 35% of attempted rapes, and 22.9% of threatened rapes took place on a date, according to respondents’ answers. A comparison component of NCWSV using different methodology mirroring that of the federally funded National Crime Victimization Survey (NCVS) (sample size of 4,432 college women conducted during 1996-97 academic year; very high response rate of 91.6%), with less emphasis on behaviorally specific, graphic questions, more on labeling, resulted in far lower percentages of reported rape within its sample for that same year.
Proportions were roughly 11 times smaller for completed rape: 0.16%, and six times smaller for attempted rape (0.18%). In a cross-sectional study of women at a large urban commuter university, Abbey et al. (1996) found that 8% of 1,160 women surveyed said that they had experienced an attempted rape; 23% reported having been raped. Copenhaver and Grauerholz (1991) limited their examination of sexual assault to 140 randomly selected sorority members at a large public Midwestern university, and found that 24% reported attempted rape, and 17% completed rape.
Drug and alcohol facilitated rape (DFR) This particular breakdown has received less attention in the past, but CSA data suggest that DFR has a much lower prevalence (just 0.6% reported certain DFR, and 1.7% suspected) than does incapacitated rape via voluntary alcohol or drug use. In the MUSC/Kilpatrick 2007, 3.58% reported for previous year;
DFR/IR was nine times more prevalent among college women than among general population women.
In their survey of the US general population, Black et al. (NISVS, 2011) found that 8% of women reported experiencing completed alcohol or drug-facilitated penetration over their lifetimes; 0.6% of men reported the same.
Incapacitated rape (IR) In the CSA, 11% of women reported being sexually assaulted when they were incapacitated after voluntarily consuming drugs and/or alcohol (i.e., they were victims of alcohol and/or other drug- [AOD] enabled sexual assault); 8.5% reported incapacitated rape specifically. It is important to note that these numbers—for incapacitated rape, meaning that voluntary intoxication had occurred—are quite a bit higher than those for both forcible rape and drug and alcohol-facilitated rape (in which intoxication is not voluntary). An even more recent study by Carey et al. (2015) followed 483 female first-years, administering surveys to them at four time points: (1) arrival on campus; (2) at the end of the fall semester; (3) at the end of the spring semester; (4) at the end of the summer following their freshman year. Reported rates of both attempted and completed incapacitated rape were higher in the fall. For attempted IR: for academic year: 10.1%; precollege: 16.2%; fall semester: 7.7%; spring semester: 4.8%;
summer: 4.4%. For completed IR: for academic year: 7.1%; precollege: 9.0%; fall semester: 4.5%; spring semester: 4.0%; summer: 4.7%.
Forcible rape (FR) The CSA found that 3.4% of its sample reported experiencing physically forced rape. In MUSC/Kilpatrick et al.’s (2007) sample population, 6.4% reported experiencing forcible rape within the previous year (compared with a much higher 14.6% found in their general population sample). Carey et al. (2015) found that, among 483 women surveyed, attempted FR was experienced by 6% during the academic year; precollege: 13.3%; fall semester: 4.3%; spring semester: 3.3%; summer: 2.8%. Prevalence of completed FR among this sample: for academic year: 5.2%; precollege: 6%; fall semester: 3.4%; spring semester: 3.6%; summer: 3.3%.
III. Problematic Issues in Prevalence Estimates Self-Report vs Admin Records A major limitation and potential problem with all of the above statistics that we should acknowledge and keep in mind: All of these numbers are based on survey and self-report; they are not based on crime statistics. As mentioned above, under-reporting is rampant; even when actions meet legal definitions of SA, victims themselves do not always perceive or label the encounter as SA (Orchowski, Untied, & Gidycz, 2013). Evidence suggests that even when women do believe they have been sexually assaulted, they often do not report the experience to law enforcement (Kilpatrick, Edmonds, & Seymour, 1992).
The FBI’s statistics on crime would indicate a much lower prevalence than the numbers cited above, but their numbers only capture SA that is actually reported to law enforcement officials (Koss, 1993).
Nonresponse Rate Another, possibly more serious, issue with the numbers that we currently have is that they are based not just on self-report, but on self-report among a small proportion of the people targeted by survey researchers. Just as a recent example in the news, the response rate for the Oct. 2014 survey by MIT was only 35%, making generalizability of results very risky. Response rates of the major studies mentioned here are somewhat better, but even the much-cited Campus Sexual Assault study had a response rate of just 42%. When less than 50% of the population of interest actually provides answers to the questions asked, we can not say for sure whether the data we get from this responsive sample reflect the experience of the whole population. Those choosing to respond to a survey could be less affected by sexual assault than those choosing not to respond (and might consequently, for example, find replying to such a survey less painful), resulting in underestimates of prevalence. Or the opposite scenario could be true: Those responding might include more victims of sexual assault, in which case prevalence numbers would be overestimated. The researchers involved in the CSA study did an analysis of nonresponse bias, but they were only comparing the responding vs overall population with respect to demographic variables (age, race, socioeconomic, etc.), not with respect to sexual assault victimization (though it is worth noting that their nonresponse bias analysis showed no real evidence of a differential response with respect to those variables used).