Sexual assault on college campuses is a public health issue. However varying research methodologies (e.g., different sexual assault definitions, measures, assessment timeframes) and low response rates hamper efforts to define the scope of the problem. To illuminate the complexity of campus sexual assault, we collected survey data from a large population-based random sample of undergraduate students from Columbia University and Barnard College in New York City, using evidence based methods to maximize response rates and sample representativeness, and behaviorally specific measures of sexual assault to accurately capture victimization rates. This paper focuses on student experiences of different types of sexual assault victimization, as well as sociodemographic, social, and risk environment correlates. Descriptive statistics, chi-square tests, and logistic regression were used to estimate prevalences and test associations. Since college entry, 22% of students reported experiencing at least one incident of sexual assault (defined as sexualized touching, attempted penetration [oral, anal, vaginal, other], or completed penetration). Women and gender nonconforming students reported the highest rates (28% and 38%, respectively), although men also reported sexual assault (12.5%). Across types of assault and gender groups, incapacitation due to alcohol and drug use and/or other factors was the perpetration method reported most frequently (> 50%); physical force (particularly for completed penetration in women) and verbal coercion were also commonly reported. Factors associated with increased risk for sexual assault included non-heterosexual identity, difficulty paying for basic necessities, fraternity/sorority membership, participation in more casual sexual encounters (“hook ups”) vs. exclusive/monogamous or no sexual relationships, binge drinking, and experiencing sexual assault before college. High rates of re-victimization during college were reported across gender groups. Our study is consistent with prevalence findings previously reported. Variation in types of assault and methods of perpetration experienced across gender groups highlight the need to develop prevention strategies tailored to specific risk groups.
Citation: Mellins CA, Walsh K, Sarvet AL, Wall M, Gilbert L, Santelli JS, et al. (2017) Sexual assault incidents among college undergraduates: Prevalence and factors associated with risk. PLoS ONE 12(11): e0186471. https://doi.org/10.1371/journal.pone.0186471
Editor: Hafiz T. A. Khan, University of West London, UNITED KINGDOM
Received: July 28, 2017; Accepted: October 2, 2017; Published: November 8, 2017
Copyright: © 2017 Mellins et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying the study cannot be made available, beyond the aggregated data that are included in the paper, because of concerns related to participant confidentiality. Sharing the individual-level survey data would violate the terms of our agreement with research participants, and the Columbia University Medical Center IRB has confirmed that the potential for deductive identification and the risk of loss of confidentiality is too great to share the data, even if de-identified.
Funding: This research was funded by Columbia University through a donation from the Levine Family. The funder (Levine Family) had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Recent estimates of sexual assault victimization among college students in the United States (US) are as high as 20–25% [1–3], prompting universities to enhance or develop policies and programs to prevent sexual assault. However, a 2016 review  highlights the variation in sexual assault prevalence estimates (1.8% to 34%) which likely can be attributed to methodological differences across studies, including varying sexual assault definitions, sampling methods, assessment timeframes, and target populations . Such differences can hamper efforts to understand the scope of the problem. Moreover, while accurate estimates of prevalence are crucial for calling attention to the population-health burden of sexual assault, knowing more about risk factors is critical for determining resource allocation and developing effective programs and policies for prevention.
Reasons for the variation in prevalence estimates include different definitions of sexual assault and assessment methods. Under the rubric of sexual assault, researchers have investigated experiences ranging from sexual harassment at school or work, to unwanted touching, including fondling on the street or dance floor, to either unwanted/non-consensual attempts at oral, anal or vaginal sexual intercourse (attempted penetrative sex), or completed penetrative sex [3,5–7]. Some studies have focused on a composite variable of multiple forms of unwanted/non-consensual sexual contact [8,9] while others focus on a single behavior, such as completed rape . Some studies focus on acts perpetrated by a single method (e.g. incapacitation due to alcohol and drug use or other factors) , while others include a range of methods (e.g., physical force, verbal coercion, and incapacitation) [12–15]. In general, studies that ask about a wide range of acts and use behaviorally specific questions about types of sexual assault and methods of perpetration have yielded more accurate estimates . Behavioral specificity avoids the pitfall of participants using their own sexual assault definitions and does not require the respondent to identify as a victim or survivor, which may lead to underreporting [10,17–19].
Although an increasing number of studies have used behaviorally specific methods and examined prevalence and predictors of sexual assault [20,21], they typically have used convenience samples. Only a few published studies have used population-based surveys and achieved response rates sufficient to mitigate some of the concerns of sample response bias . US federal agencies have urged universities to implement standardized “campus climate surveys” to assess the prevalence and reporting of sexual violence . Although these surveys have emphasized behavioral specificity, many have yielded low response rates (e.g., 25%) , particularly among men , creating potential for response bias in the obtained data. Population-based probability samples with behavioral specificity, good response rates, sufficiently large samples to examine risk for specific subgroups (e.g., sexual minority students), and detailed information on personal, social, or contextual risk factors (e.g., alcohol use) [22,23] are needed to more accurately define prevalence and inform evidence-based sexual assault prevention programs.
Existing evidence suggests that most sexual assault incidents are perpetrated against women ; however, few studies have examined college men as survivors of assault [26–28]. Furthermore, our understanding of how sexual orientation and gender identity relate to risk for sexual assault is limited, despite indications that lesbian, gay, bisexual (LGB), and gender non-conforming (GNC) students are at high risk [29–31]. It is unclear if these groups are at higher risk for all types of sexual assault or if prevention programming should be tailored to address particular types of assault within these groups. Also, although women appear to be at highest risk for assault during freshman year [32,33], the dearth of studies with men or GNC students have limited conclusions about whether freshman year is also a risky period for them.
Additional factors associated with experiencing sexual assault in college students include being a racial/ethnic minority student (although there are mixed findings on race/ethnicity) [34,35], low financial status, and prior history of sexual assault [3,33,36]. Other risk factors include variables related to student social life, including being a freshman , participating in fraternities and sororities [19,37,38], binge drinking [1,39] and participating in “hook-up” culture [40–42]. Whether sexual assault is happening in the context of more casual, typically non-committal sexual relationships (“hook-ups”)  vs. steady intimate or monogamous relationships has important implications for prevention efforts.
To fill some of these knowledge gaps, we examined survey data collected from a large population-based random sample of undergraduate women, men, and GNC students at Columbia University (CU) and Barnard College (BC). The aims of this paper are to:
- Estimate the prevalence of types of sexual assault incidents involving a) sexualized touching, b) attempted penetrative (oral, anal or vaginal) sex, and c) completed penetrative sex since starting at CU/BC;
- Describe the methods of perpetration (e.g., incapacitation, physical force, verbal coercion) used; and
- Examine associations between key sociodemographic, social and romantic/sexual relationship factors and different types of sexual assault victimization, and how these associations differ by gender.
Materials and methods
This study used data from a population-representative survey that formed one component of the Sexual Health Initiative to Foster Transformation (SHIFT) study. SHIFT used mixed methods to examine risk and protective factors affecting sexual health and sexual violence among college undergraduates from two inter-related institutions, CU’s undergraduate schools (co-educational) and BC (women only), both located in New York City. SHIFT featured ethnographic research, the survey, and a daily diary study. Additionally, SHIFT focused on internal policy-translation work to inform institutionally-appropriate, multi-level approaches to prevention.
Survey participants were selected via stratified random sampling from the March 2016 population of 9,616 CU/BC undergraduate students ages 18–29 years. We utilized evidence-based methods to enhance response rates and sample representativeness [22,43]. Using administrative records of enrolled students, 2,500 students (2,000 from CU and 500 from BC) were invited via email to participate in a web-based survey. Of these 2,500 students, 1,671 (67%) consented to participate (see Procedures). Among those who consented to participate, 80.5% were from CU and 19.5% were from BC (see Table 1 below for demographic data on the CU/BC student population, the random sample of students contacted, the survey responders, and the current analytic sample).
SHIFT employed multiple procedures to assure protection of students involved in our study; these procedures also improve scientific rigor. The study was approved by the Columbia University Medical Center Institutional Review Board and we obtained a federal Certificate of Confidentiality to legally protect our data from subpoena. SHIFT also obtained a University waiver from reporting on individual sexual assaults, as reporting would obviate student privacy and willingness to participate. Students were offered information about referrals to health and mental health resources during the consent process and at the end of the survey, and such information was available from SHIFT via other communication channels. Finally, in reporting data we suppressed data from tables where there were less than 3 subjects in any cell to avoid the possibility of deductive identification of an individual student .
SHIFT used principles of Community Based Participatory Research regarding ongoing dialogue with University stakeholders on study development and implementation to maximize the quality of data and impact of research findings . This included weekly meetings between SHIFT investigators and an Undergraduate Advisory Board, consisting of 13–18 students, reflecting the undergraduate student body’s diversity in terms of gender, race/ethnicity, sexual orientation, year in school, and activities (e.g., fraternity/sorority membership). It also included regular meetings with an Institutional Advisory Board comprised of senior administrators, including CU’s Office of General Counsel, facilities, sexual violence response, student conduct, officials involved in gender-based misconduct concerns, athletics, a chaplain, mental health and counseling, residential life, student health, and student life.
Following both the Undergraduate Advisory Board’s recommendations and Dillman’s Tailored Design Method for maximizing survey response rates , multiple methods were used to advertise and recruit students. These included: a) email messages, both to generate interest and remind students who had been selected to participate, crafted to resonate with diverse student motives for participation (e.g., interest in sexual assault, compensation, community spirit, and achieving higher response rates than surveys at peer institutions), b) posting flyers, c) holding “study breaks,” in which students were given snacks and drinks, and d) tabling in public areas on campus.
Participants used a unique link to access the survey either at our on-campus research office where computers and snacks were provided (16% of participants) or at a location of their choosing (84% of participants) from March-May, 2016. Before beginning the survey, participants were asked to provide informed consent on an electronic form describing the study, confidentiality, compensation for time and effort, data handling procedures, and the right to refuse to answer any question. Students who completed the survey received $40 in compensation, given in cash to those who completed the survey in our on-campus research office or as an electronic gift card if completed elsewhere. Students were also entered into a lottery to win additional $200 electronic gift cards. This compensation was established based on feedback from student and institutional advisors and reviewed by our Institutional Review Board. It was judged to be sufficient to promote participation, and help ensure that we captured a representative sample, including students who might otherwise have to choose between paid opportunities and participating in our survey, but not great enough to feel coercive for low resource students. This amount of compensation is in line with other similar studies . On average, the survey took 35–40 minutes to complete.
The SHIFT survey included behaviorally-specific measures of different types of sexual assault, perpetrated by different methods, as well as measures of key sociodemographic, social and sexual relationship factors, and risk environment characteristics. The majority of instruments had been validated previously with college- age students. The survey was administered in English using Qualtrics (www.qualtrics.com), providing a secure platform for online data collection.
Sexual assault was assessed with a slightly modified version of the revised Sexual Experiences Survey , the most widely used measure of sexual assault victimization with very good psychometric properties including internal consistency and validity previously published [17,47]. The Sexual Experiences Survey employs behaviorally specific questions to improve accuracy . The scale includes questions on type of assault, including sexualized touching without penetration (touching, kissing, fondling, grabbing in a sexual way), attempted but not completed penetrative assault (oral, vaginal, anal or other type of penetration; herein referred to as attempted penetrative assault) and completed penetrative assault (herein referred to as penetrative assault). We used most of the Sexual Experiences Survey as is. However, with strong urging from our Undergraduate Advisory Board, we made a modification, combining the questions about different types of penetration (oral, vaginal, etc.) rather than asking about each kind separately. In the Sexual Experiences Survey, for each type of assault there are six methods of perpetration. Two of the types reflect verbal coercion: 1) “Telling lies, threatening to end the relationship, threatening to spread rumors about me, making promises I knew were untrue, or continually verbally pressuring me after I said I didn’t want to” (herein referred to as “lying/threats”), and 2) “Showing displeasure, criticizing my sexuality or attractiveness, getting angry but not using physical force, after I said I didn’t want to” (herein referred to as “criticism”). The remaining types included use of physical force, threats of physical harm, or incapacitation (“Taking advantage when I couldn’t say no because I was either too drunk, passed out, asleep or otherwise incapacitated”), and other. For each incident of sexual assault, participants could endorse multiple methods of perpetration. Participants were also asked to report whether these experiences occurred: a) during the current academic year (this was a second modification to the Sexual Experiences Survey) and/or b) since enrollment but prior to the current academic year. For this paper, data for the two time periods were combined, reflecting the entire period since starting CU/BC. See Fig 1 for a replica of the questionnaire.
Demographics included gender identity (male, female, trans-male/trans-female, gender queer/gender-non-conforming, other) , year in school (e.g., freshman, sophomore, junior, senior), age, US born (yes/no), lived in US less than five years (yes/no; proxy for recent international student status), transfer student (yes/no), low socioeconomic status (receipt of Pell grant-yes/no [need-based grants for low-income students, with eligibility dependent on family income]); how often participant has trouble paying for basic necessities (never, rarely, sometimes, often, all of the time), and race/ethnicity (non-Hispanic white, non-Hispanic-Asian, non-Hispanic black, Hispanic/Latin-x, other [other included: American Indian or Alaska Native, Native Hawaiian or Pacific Islander, More than one Race/Ethnicity, Other]). Gender was categorized as follows: female, male and GNC (students who responded to gender identity question as anything other than male or female).
Fraternity/sorority membership (ever participated) was assessed with one question from a school activities checklist (yes/no). We report on Greek life participation here to engage with the substantial attention this has received as a risk factor.
Problematic drinking during the last year was assessed with the Alcohol Use Disorders Identification Test (AUDIT) , a widely used, well-validated standardized 10-item screening tool developed by the World Health Organization. Psychometrics have been established in numerous studies [50–52]. The AUDIT assesses alcohol consumption, drinking behaviors, and alcohol-related problems. Participants rate each question on a 5-point scale from 0 (never) to 4 (daily or almost daily) for possible scores ranging from 0 to 40. The range of AUDIT scores represents varying levels of risk: 0–7 (low), 8–15 (risky or hazardous), 16–19 (high-risk or harmful), and 20 or greater (high-risk). We also examined one AUDIT item on binge drinking, defined as having 6+ drinks on one occasion at least monthly .
Sexual orientation was assessed with one question with the following response options (students could select all that applied): asexual, pansexual, bisexual, queer, heterosexual and homosexual, as well as other [53,54]. Students were categorized into four mutually exclusive groups for analyses: heterosexual, bisexual, homosexual, and other which included asexual, pansexual, queer, or another identity not listed. Non-heterosexual students who indicated more than one orientation were assigned hierarchically to bisexual, homosexual, then other.
Romantic/sexual relationships since enrollment at CU/BC were assessed with one question. Response choices included: none, steady or serious relationship, exclusive or monogamous relationship, hook-up-one time, and ongoing hook-up or friends with benefits. Students defined “hookup” for themselves. Students could check all that applied. This variable was trichotomized: at least one hook-up, only steady or exclusive/monogamous relationships, and no romantic/sexual relationships.
To assess the representativeness of the sample, the distribution of demographic variables based on administrative records from CU and BC for the total University undergraduate population were compared to the random sample of students contacted, the survey responders, and the current analytic sample, which consists of students that responded to the questions about sexual assault. Demographics for survey responders are based on self-report from the survey. Cramer’s V effect size was used to assess the magnitude of the differences in demographic distributions between the CU/BC population and respondent sample where smaller values (i.e. Cramer’s V <0.10) indicate strong similarity .
Analyses were performed on each type of sexual assault as well as a combined “Any type of sexual assault” variable: yes/no experienced sexualized touching, attempted penetrative assault, and/or penetrative assault since CU/BC. Prevalence of each type of sexual assault was calculated by gender and year in school, with chi-square tests of difference used to compare prevalence between genders across each year in school versus freshman year. The total number of incidents of assault and the mean, median and standard deviation for number of incidents of assault per person reporting at least one assault were summarized. Among individuals who experienced any type of sexual assault, the proportions that experienced a particular method of perpetration (e.g. incapacitation, physical force) were calculated by type of sexual assault. Chi-square tests compared proportions between males and females for each perpetration method. The associations of each key correlate with the odds of experiencing any sexual assault were calculated and tested using logistic regression stratified by male/female gender. In addition, a multinomial regression with hierarchical categories (no assault, sexualized touching only, attempted penetrative assault [not completed], and penetrative assault [completed]) as the outcome was performed to examine if associations differed by type of sexual assault. To adjust for the fact that the sample comes from a finite population (i.e. CU/BC N = 5,765 women; N = 3,851 men), a standard finite population correction was implemented for standard error estimation using SAS Proc Surveylogistic. Given the low sample size of GNC students, they were excluded from some analyses. All analyses were conducted using SAS (v. 9.4).
Table 1 presents demographic data on the full University, the randomly selected sample, the respondents and the analytic sample for this paper. Among students who consented to the survey (n = 1,671), 46 stopped the survey before the sexual assault questions and 33 refused to answer them resulting in an analytic sample of n = 1,592 (95% completion among responders). Demographic characteristics (i.e. gender [male, female], age, race/ethnicity, year in school, international status, and economic need [Pell grant status]) of the respondent sample were very similar (Cramer’s V effect size differences all <0.10 ) to the full CU/BC population (Table 1) indicating that the responder and final analytic samples were representative of the student body population.
The analytic sample included 58% women, 40% men, and 2% GNC students (4 students refused to identify their gender) and was distributed evenly by year in school with most (92%) between18-23 years of age. Self-reported race/ethnicity was 43% white non-Hispanic, 23% Asian, 15% Hispanic/Latino, and 8% black non-Hispanic; 13% were transfer students, and the majority of the sample was born in the US (76%). Twenty-three percent of participants received Pell grants and 51% of students acknowledged at least sometimes having difficulty paying for basic necessities.
The majority of women (79%) and men (85%) identified as heterosexual. In terms of romantic/sexual relationships since starting CU/BC, 30.0% of women and 21.6% of men reported no relationships, 21.0% of women and 22.6% of men reported only steady/exclusive relationships with no hookups, and 49.0% of women and 55.7% of men reported at least one hook-up. Finally, 25.5% of women, 9.4% of men, and 47.0% of GNC students reported pre-college sexual assault.
Aim 1: Prevalence of sexual assault victimization at CU/BC
Overall rates by gender and school year.
Since starting CU/BC, 22.0% (350/1,592) of students reported experiencing at least one incident of any sexual assault across the three types (sexualized touching, attempted penetrative assault, and penetrative assault). Table 2 presents data on types of assault by gender and year in school. Women were over twice as likely as men to report any sexual assault (28.1% vs 12.5%). There was evidence of cumulative risk for experiencing sexual assault among women over four years of college, so that by junior and senior year, respectively, 29.7% and 36.4% of women reported experiencing any sexual assault, compared to 21.0% of freshman women who had only one year of possible exposure (p < .05). However, one-fifth (21.0%) of women who took the survey as freshman had experienced unwanted sexual contact, compared to 36.4% over 3+ years (seniors), suggesting that as others have found, the risk of assault is highest in freshman year.
Among men, one in eight indicated that they had been sexually assaulted since starting CU. Similar to women, the risk for sexual assault among men accumulated over the four years of college, with 15.6% of seniors vs 9.9% of freshman reporting a sexual assault since entering CU, although this difference was not statistically significant.
Although the numbers were small, GNC students reported the highest prevalence of sexual assault since starting CU/BC (38.5%; 10/26). Numbers were too small (n<3) to present stratified by year in school (see Table 2).
Types of sexual assault by gender (Table 2).
The most prevalent form of sexual assault was sexualized touching; rates for women (23.6%) and GNC students (38.5%) were significantly higher than rates for men (11.0%; p < .05). Prevalence of attempted penetrative assault and penetrative assault were about half that of sexualized touching. Compared to men, women were three times as likely to report attempted penetrative assault (11.1% vs 3.8%) and over twice as likely to experience penetrative assault (13.6% vs 5.2%). Among GNC students, the majority reporting sexualized touching, with rates of the other two types too small to report.
Students could report multiple types of sexual assault incidents (i.e. sexualized touching, attempted penetrative, and penetrative assault) as well as multiple incidents experienced of each type. Overall, students reported a total of 1,007 incidents of sexual assault experienced since starting CU/BC. For the 350 students who indicated any sexual assault, the median number of incidents experienced was 3.
Among the 350 students reporting any sexual assault, Fig 2 presents different combinations of sexual assault experienced by students since CU/BC. Most prevalent, 38.0% reported experiencing only sexualized touching; 19.0% reported both sexualized touching and penetrative assault incidents; 17.0% experienced all three types of assault; and 12.0% sexualized touching and attempted penetrative assault.
Aim 2: Methods of perpetration (lying/threats, criticism, incapacitation, physical force, threats of harm, and other) by gender (Table 3)
Across types of assault, incapacitation was the method of perpetration reported most frequently (> 50%) in both men and women. For both women and men, approximately two-thirds of all penetrative assaults and about half of sexualized touching and attempted penetrative assaults involved incapacitation.
Physical force was reported significantly more frequently by women than men (34.6% vs 12.7%) for any sexual assault. More specifically, compared to men, women were three times more likely to experience sexualized touching via physical force (32.1% vs. 10.0%), and six times more likely to experience penetrative assaults via physical force (33.3% vs 6.1%).
Lastly, a sizeable number of respondents reported verbal coercion (ranging from 21.0% to over 40.0% depending on type of assault). Criticism was cited by women at rates similar to physical force for both sexualized touching and penetrative assaults. Among men, both verbal coercion methods were cited most frequently after incapacitation for all three types of assault.
For GNC students, we examined rates of each perpetration method for only the composite variable any sexual assault (due to small numbers in any specific type of assault). Among those who experienced an assault, incapacitation was the most frequently mentioned method (50.0%), followed by criticism (40.0%).
Aim 3: Identify factors associated with sexual assault experiences
We examined the association between sexual assault (both any sexual assault [Table 4] and each type of sexual assault [Table 5]) and key demographic, sexual history and social activity factors. Results are stratified by gender (women/men).
For both women and men, the prevalence of any sexual assault was similar for all race/ethnicity groups compared to non-Hispanic White students with one exception. Asian students (women and men) were less likely to experience any sexual assault than non-Hispanic White students. For women only, differences emerged by type of assault. Asian women compared to non-Hispanic White women were less likely to experience penetrative assault (OR = 0.35, CI: 0.19–0.62), but not attempted penetrative assault (OR = 0.56, CI: 0.25–1.26), nor sexualized touching only (OR = 1.00, CI: 0.59–1.69). Black women were found to have increased odds of touching only incidents compared to non-Hispanic White women (OR = 1.99, CI: 1.05–3.74). There were no other significant racial or ethnic differences.
Women who often or always had difficulty paying for basic necessities had increased odds of any sexual assault; for men the trend was similar but it did not reach statistical significance. Considering penetrative assault specifically, both men and women who often or always had difficulty paying for basic necessities had increased risk (women OR = 2.24, CI: 1.23–4.09; men OR = 3.07, CI: 1.04–9.07) compared to those who never had difficulty.
Women transfer students were less likely to experience any sexual assault than non-transfer students. Closer inspection of type of assault revealed that this protective effect was seen for sexualized touching only (OR = 0.34, CI: 0.15–0.80), but not for penetrative (OR = 0.60, CI: 0.34–1.08), nor attempted penetrative (OR = 1.03, CI: 0.48–2.21) assault. There were no significant differences between men who were transfer students and those who were not.
For women, those who identified as bisexual and those who identified as some other sexual identity besides heterosexual, homosexual, or bisexual (includes people endorsing exclusively one or a combination of: Asexual, Pansexual, Queer, or a sexual orientation not listed), were more likely to experience any sexual assault than heterosexual students. For penetrative assault specifically, this increased risk was only present for individuals with some other sexual identity (OR = 2.11, CI: 1.20–3.73). For men, those who identified as homosexual were more likely to experience any sexual assault than heterosexual male students. For penetrative assault specifically, those who identified as homosexual, bisexual, or some other sexual identity all had substantially increased risk compared to those with a heterosexual identity (OR = 4.74, CI: 2.10–10.71; OR = 3.39, CI: 1.03–11.16; OR = 4.74, CI:1.10–20.48, respectively).
Information about the gender of the perpetrator for different gender and sexual orientation groups was available for a subset of incidents (336/997). Among these events, 98.4% (3/184) of the heterosexual women indicated the perpetrator was a man, while 97.1% (33/34) of the bisexual women, 75% (3/4) of the homosexual women, and 88.9% (24/27) of the other sexual identity women indicated it was a man. For men who were assaulted, 84.9% (45/53) of the heterosexual men reported the perpetrator was a woman, while 0 of the homosexual men said the perpetrator was a woman. Numbers for bisexual men and other sexual identity men were too small to report separately, but combined showed that 5/8 (63.0%) of bisexual and other sexual identity men said the perpetrator was a woman. Of the GNC students reporting on a most-significant event, 77.8% (7/9) reported that they were assaulted by a male perpetrator (the numbers are too small to further examine by sexual orientation).
Lived in US less than 5 years.
There was no association found between living in the US for less than 5 years and any sexual assault, nor any specific type of sexual assault.
Among both women and men, students who had at least one hook-up were more likely to have experienced any sexual assault than students who were in only steady/exclusive relationships since starting college. Among women who had engaged in at least one hook-up, this increased risk held for each type of sexual assault (penetrative: OR = 5.03, CI = 2.91–8.68, attempted penetrative: OR = 4.43, CI = 1.83–10.8, sexualized touching only: OR = 3.26, CI = 1.74–6.09), while among men the increased risk was found for sexualized touching only (OR = 13.33, CI = 2.09–85.08), but could not be estimated (due to small numbers) for completed penetrative assault. Women who did not have any romantic or sexual relationship since CU/BC were found to be less likely to experience penetrative assault than women who had a steady/exclusive relationships only (OR = 0.05, CI: 0.01–0.31).
Although a relative minority of students participated in fraternities (24.1%) or sororities (18.2%), for both men and women, those who participated were more likely to experience any sexual assault than those who did not. Examination of type of assault revealed that the effect is driven primarily by sexualized touching only which is significant in both women (OR = 1.63, CI: 1.00–2.67) and men (OR = 2.40, CI: 1.25–4.63) and not significantly increased for penetrative nor attempted penetrative assault.
Risky or hazardous drinking.
For both men and women, individuals who met criteria on the AUDIT for risky or hazardous drinking were more likely to experience any sexual assault than those who did not. When examining each type of assault separately, for men this increased risk was only significant for penetrative assault (OR = 4.07, CI: 2.01–8.21). For women, the increased risk of assault held for each type of assault—penetrative (OR = 6.04, CI: 4.10–8.90), attempted (OR = 3.38, CI: 1.84–6.19) and touching (OR = 2.33, CI: 1.42–3.81). We also looked at one AUDIT item specifically on binge drinking (6 or more drinks on a single occasion). Individuals who reported binge drinking at least monthly were more likely to experience any sexual assault than those who did not. When examining each type of assault separately, for men this increased risk was only significant for penetrative assault (OR = 2.15, CI: 1.12–4.15). For women, this increased risk was significant for penetrative assault (OR = 3.12, CI: 2.09–4.65), attempted assault (OR = 2.28, CI: 1.20–4.33), and touching (OR = 2.42, CI:1.50–3.91).
Pre-college assault (Table 5).
Among both women and men, those who experienced pre-college assault were more likely to experience any sexual assault while at CU/BC. The increased risk held for penetrative assault in both women (OR = 3.01, CI: 2.07–4.37) and men (OR = 2.44, CI: 1.03–5.76). In women, the increased risk also held for attempted penetrative, but not touching only, whereas in men, the increased risk held for touching only, but not attempted penetrative sex.
The SHIFT survey, with a population-representative sample, good response rate and behaviorally-specific questions, found that 22.0% of students reported a sexual assault since starting college, which confirms previous studies of 1 in 4 or 1 in 5 prevalence estimates with national samples and a range of types of schools [23,24]. However, a key finding is that focusing only on the “1 in 4/ 1 in 5” rate of any sexual assault obscures much of the nuance concerning types of sexual assault as well as the differential group risk, as prevalence rates were unevenly distributed across gender and several other social and demographic factors.
Similar to other studies [4,24], women had much higher rates of experiencing any type of sexual assault compared to men (28.0% vs 12.0%). Moreover, our data suggest a cumulative risk for sexual assault experiences over four years of college with over one in three women experiencing an assault by senior year. However, our data also suggest that freshman year, particularly for women, is when the greatest percentage experience an assault. This supports other work on freshman year as a particularly critical time for prevention efforts, otherwise known as the “red zone” effect for women .
Importantly, our study confirms that GNC students are at heightened risk for sexual assault . They had the highest proportion of sexual assaults, with 38.0% reporting at least one incident, the majority of which involved unwanted/non-consensual sexualized touching. These data should be interpreted very cautiously given the small number of GNC students. However, increasingly studies suggest that transgender and other GNC students have sexual health needs that may not be targeted by traditional programming ; thus, a better understanding of pathways to vulnerability among these students is of high importance.
Similarly, students who identified as a sexual orientation other than heterosexual were at increased risk for experiencing any sexual assault, with bisexual women or women who identified as “other” and men who identified as any non-heterosexual category at increased risk. Similar to GNC students, understanding the specific social and sexual health needs of LGB students, particularly as it relates to reducing sexual assault risk is critical to prevention efforts . Factors such as stigma and discrimination, lack of communication, substance use, as well as a potential lack of tailored prevention programs may play a role. To our knowledge, there are no evidence-based college sexual assault prevention programs targeting LGB and GNC students. Our data suggest that the LGB and GNC experiences are not uniform; more research should be done within each of these groups to understand the mechanisms behind their potentially unique risk factors.
Our data also suggest that the 20–25% rate of any sexual assault obscures variation in assault experiences. Sexualized touching accounted for the highest percentage of acts across gender groups, with over one-third of participants reporting only sexualized touching incidents. Rates of attempted and completed penetrative sexual assault were about half the rate of sexualized touching. This finding does not minimize the importance of addressing unacceptably high rates of attempted penetrative and penetrative assault (14%-15%), but it does suggest the importance of specificity in prevention efforts. For GNC students, for example, the risk of assault was primarily for sexualized touching with very few reporting attempted penetrative assault or penetrative assault during their time at CU/BC. These elevated rates of unwanted sexual touching may be a combination of GNC students’ focus on their gendered sexual boundaries–and thus potentially greater awareness of when advances are unwanted–at a developmental moment when they are building non-traditional gender identities, as well as these students’ social vulnerability. Further investigation is warranted.
Moreover, there was variation in methods of perpetration reported by survivors of sexual assault. Incapacitation was the most common method reported across all gender groups for each type of assault, and female and male students who reported risky or hazardous drinking were at increased risk for experiencing any sexual assault, particularly penetrative assault. Across campuses in the US, hazardous drinking is a national problem with substantive negative health outcomes, risk for sexual assault being one of them [2,39,59]. Our data underline the potential of programs and policies to reduce substance use and limit its harms as one element of comprehensive sexual assault prevention; we found few evidence-based interventions that address both binge drinking and sexual assault prevention. Of course, any work addressing substance use as a driver of vulnerability must do so in a way that does not replicate victim-blaming.
However, similar to other studies with broad foci, incapacitation was not the only method of perpetration reported. For women, physical force, particularly for penetrative sex, was the second most frequently endorsed method. Verbal coercion, including criticism, lying and threats to end the relationship or spread rumors, was also employed at rates similar to physical force for women, and was the second most frequently endorsed category for men and GNC students. Prevention programs, such as the bystander interventions which are the focus of efforts on many campuses , often focus on incapacitation or physical force. These interventions tend to highlight situations where survivors (typically women) are vulnerable because they are under the influence of substances. In SHIFT, verbal coercion is also shown to be a powerful driver of assault; however, it typically does not receive as much attention as rape, which is legally defined as penetration due to physical force or incapacitation. If a survivor is verbally coerced into providing affirmative consent, the incident could be considered within consent guidelines of “yes means yes” but it may have been unwanted by the survivor [61,62]. Assertiveness interventions and those that focus on verbal consent practices may be useful for addressing this form of assault.
We also found high rates of re-victimization. As others have found, pre-college sexual assault was a key predictor for experiencing assault at CU/BC [33,36]. However, we also found high rates of repeat victimization since starting at CU/BC with a median of 3 incidents per person reporting any sexual assault since starting CU/BC, and the highest risk of repeat victimization in women and GNC students. These data underline the importance of prevention efforts that include care for survivors to reduce the enhanced vulnerability that has been shown in other populations of assault survivors . Future studies should also seek to disaggregate the relationship between type of victimization (sexualized touching, attempted penetrative assault, penetrative assault) and repeat victimization.
This study also identified a number of variables associated with sexual assault, some similar to previous studies and others different. As noted, gender was a key correlate. While prevention efforts should respond to the population-level burden by focusing on the needs of women and GNC students, it is important to note that men were also at risk of sexual assault. In our study, nearly 1 in 8 men reported a sexual assault experience, a rate also found in the Online College Social Life survey , but higher than other studies [63,64]. Few programs target men, and issues around masculinity and gender roles may make it difficult for men to consider or report what has happened to them as sexual assault. Importantly, this study found that men who were members of fraternities were at higher risk for experiencing assault (specifically unwanted/nonconsensual sexualized touching) than those who were not members. This is consistent with previous findings, including the Online College Social Life survey , but is of particular note because research has identified men in fraternities as more likely to be perpetrators , but few, if any, studies have looked at fraternity members’ vulnerability to sexual assault. Our data suggest a need for further examination of the cultural and organizational dimensions of Greek life that produce this heightened risk of being assaulted for both men and women. However, it is important to note that we did not examine a range of other social and extracurricular groups which may have produced risk as well and thus a more full examination of student undergraduate life is needed.
One other key factor associated with assault was participation in “hook ups”. Both male and female students who reported hooking up were more likely to report experiencing sexual assault, compared to students who only had exclusive or monogamous relationships and those who had no sexual relationships. The role of hooking up on college campuses has received much attention in the popular press and in a number of books [65,66], but little has been written about its connection to sexual assault, although several recent studies are in line with ours about its role as a risk factor for experiencing sexual assault on college campuses [40,41]. Multiple mechanisms may be at work: students who participate in hookups may be having sex with more people, and thus face greater risk of assault due to greater exposure to sex with a potential perpetrator, but students who participate in hookups may also face increased vulnerability because many hookups involve “drunk” sex, or because hookups by definition involve sexual interactions between people who are not in a long-term intimate relationship, and thus whose bodies and social cues maybe unfamiliar to each other. Alternatively some aspects of hook-ups may be more or less risky than others and therefore continued study of different dimensions of these more casual relationships that can refer to a wide-range of behaviors is necessary.
Several demographic characteristics were not for the most part associated with sexual assault. We did not find racial or ethnic differences in sexual assault risk with primarily one exception, Asian male and female students were at less risk overall compared to white students. We also did not find transfer students to be at greater risk; female transfer students were actually at lower risk, potentially due to less exposure time, particularly during freshman year. International student status as indicated by having been in the US<5 years was also not associated with increased risk. However, this study highlights the role of economic factors that have received limited attention in the literature. Little is known about how economic insecurity may drive vulnerability, but issues of power, privilege, and control of alcohol and space all require further examination.
There are several limitations to this study. Participants came from only two private schools that are interconnected in one city, and thus findings may not generalize to the rest of the US. There is a continued need for more national studies with different types of colleges and universities in urban and rural environments with more varied economic backgrounds in order to fully understand institutional and contextual differences. Although we had a response rate that was higher than many prior studies and our rates of sexual assault are consistent with prior studies , we cannot assess the extent to which selection bias may have occurred and therefore, our rates could be an underrepresentation or overrepresentation depending on who chose to participate. Although this concern is somewhat mitigated by findings that basic demographic data between respondents and the total population of students at two colleges suggest no significant differences, there may be some bias in factors we did not consider. Our present analysis has focused only on bivariate associations between risk factors and assault. While this analysis provides a valuable description of which groups are at elevated risk or not, future work will consider how combinations of risk factors at different levels may interact to increase risk. Critically, the analysis presented here reflects a focus on those who experience being assaulted, but in other work we look at the characteristics of perpetrators, both from those who reported perpetrating and from a subset of incidents that survey respondents described in depth, which provided more information about the perpetrator. A greater understanding of the characteristics and contexts of perpetration is without question vital for effective prevention. Finally, our data are cross sectional. Longitudinal studies with a comprehensive range of predictors are critical for identifying pathways of causality and targets for interventions.
Despite these limitations, this study confirms the unacceptably high rates of sexual assault and suggests diversity in experiences and methods of perpetration. A key conclusion is that a”one size fits all” approach that characterizes the extant literature on evidence-based prevention programs  may need to be altered to more effectively prevent sexual assault in college. Clearly different groups had differential risk for assault and may require much more targeted prevention efforts. Bystander interventions have shown promise in addressing risk in social situations, including fraternity parties and other settings with high alcohol use [68,69]. However, bystander interventions may not be sufficient for incidents occurring in non-party contexts where verbal coercion methods or physical force may be used without others around.
Creating effective and sustainable changes to campus culture requires engaging with a broad range of institutional stakeholders. SHIFT investigators are in the process of sharing selected findings with both student and institutional advisory boards, and an intensive collaborative process allows us to explore the implications of our results for a broad range of policies and programs, including both elements commonly considered as sexual assault prevention (consent education, bystander trainings), more general topics related to sexual orientation and verbal discussions of sex, and aspects of the institutional context across diverse domains including alcohol policy, mental health services, residential life policies, orientation planning, and the allocation of space across campus.
Overall, our findings argue for the potential of a systems-based  public health approach–one that recognizes the multiple interrelated factors that produce adverse outcomes, and perhaps particularly emphasizes gender and economic disparities and resulting power dynamics, widespread use of alcohol, attitudes about sexuality, and conversations about sex–to make inroads on an issue that stubbornly persists.
The authors thank our research participants; the Undergraduate Advisory Board; Columbia University’s Office of the President and Office of University Life, and the entire SHIFT team who contributed to the development and implementation of this ambitious effort.
- 1. Fisher BS, Cullen FT, Turner MG. The Sexual Victimization of College Women. Research Report. 2000; Available: http://eric.ed.gov/?id=ED449712
- 2. Krebs CP, Lindquist CH, Warner TD. The Campus Sexual Assault Study (CSA) Final Report: Performance Period: January 2005 Through December 2007. 2007.
- 3. Kilpatrick D, Resnick H, Ruggiero KJ, Conoscenti LM, McCauley J. Drug-facilitated, Incapacitated, and Forcible Rape: A National Study [Internet]. 2007 Jul. Report No.: 219181. Available: http://www.antoniocasella.eu/archila/Kilpatrick_drug_forcible_rape_2007.pdf
- 4. Fedina L, Holmes JL, Backes BL. Campus Sexual Assault: A Systematic Review of Prevalence Research From 2000 to 2015. Trauma Violence Abuse. 2016;Epub ahead of print: 1–18.
- 5. Black MC, Basile KC, Breiding MJ, Smith SG, Walters ML, Merrick MT, et al. The national intimate partner and sexual violence survey (NISVS): 2010 summary report. 2011 Nov.
- 6. Catalano S, Harmon M, Beck A, Cantor D. BJS Activities on Measuring Rape and Sexual Assault [Internet]. Poster presented at; 2005. Available: https://www.bjs.gov/content/pub/pdf/bjs_amrsa_poster.pdf
- 7. Calhoun K, Mouilso E, Edwards K. Sexual assault among college students. Sex in College. Richard D. McAnulty. Santa Barbara, CA: Greenwood Publishing Group; 2012. pp. 263–288.
- 8. Banyard VL, Ward S, Cohn ES, Plante EG, Moorhead C, Walsh W. Unwanted sexual contact on campus: A comparison of women’s and men’s experiences. Violence Vict. 2007;22: 52–70. pmid:17390563
- 9. Baum K, Klaus P. National Crime Victimization Survey: Violent Victimization of College Students, 1995–2002 [Internet]. 2005 pp. 1–7. Report No.: NCJ 206836. Available: https://www.bjs.gov/content/pub/pdf/vvcs02.pdf
- 10. Swartout K, Koss M, White J, Thompson MP, Abbey A, Bellis AL. Trajectory Analysis of the Campus Serial Rapist Assumption. JAMA Pediatr. 2015;169: 1148–1154. pmid:26168230
- 11. Kaysen D, Neighbors C, Martell J, Fossos N, Larimer ME. Incapacitated rape and alcohol use: A prospective analysis. Addict Behav. 2006;31: 1820–1832. pmid:16446044
- 12. Humphrey JA, White JW. Women’s vulnerability to sexual assault from adolescence to young adulthood. J Adolesc Health. 2000;27: 419–424. pmid:11090744
- 13. Thompson M, Kingree JB. Sexual victimization, negative cognitions, and adjustment in college women. Am J Health Behav. 2010;34: 55–59.
- 14. Swartout K, Swartout A, White J. A person-centered, longitudinal approach to sexual victimization. 2011;1: 29–40.
- 15. Testa M, Hoffman JH, Livingston JA. Alcohol and sexual risk behaviors as mediators of the sexual victimization—revictimization relationship. J Consult Clin Psychol. 2010;78: 249–259. pmid:20350035
- 16. Koss M, Abbey A, Campbell R, Cook S, Norris J, Testa M, et al. Revising the SES: A Collaborative Process to Improve Assessment of Sexual Aggression and Victimization. Psychol Women Q. 2007;31: 357–370.
- 17. Koss MP, Gidycz CA, Wisniewski N. The scope of rape: incidence and prevalence of sexual aggression and victimization in a national sample of higher education students. J Consult Clin Psychol. 1987;55: 162–170. pmid:3494755
- 18. Fisher BS. The Effects of Survey Question Wording on Rape Estimates Evidence From a Quasi-Experimental Design. Violence Women. 2009;15: 133–147.
- 19. Franklin CA. Sorority Affiliation and Sexual Assault Victimization: Assessing Vulnerability Using Path Analysis. Violence Against Women. Violence Women. 2015;22: 895–922.
- 20. Messman-Moore TL, Coates AA, Gaffey KJ, Johnson CF. Sexuality, Substance Use, and Susceptibility to Victimization: Risk for Rape and Sexual Coercion in a Prospective Study of College Women. J Interpers Violence. 2008;23: 1730–1746. pmid:18349349
- 21. Orchowski LM, Creech SK, Reddy MK, Capezza NM, Ratcliff T. College women’s perceived risk to experience sexual victimization: a prospective analysis. Violence Vict. 2012;27: 194–214. pmid:22594216
- 22. Wood L, Sulley C, Kammer-Kerwick M, Follingstad D, Busch-Armendariz N. Climate Surveys: An Inventory of Understanding Sexual Assault and Other Crimes of Interpersonal Violence at Institutions of Higher Education. Violence Women. 2016;23:1249–1267.
- 23. Cantor D, Fisher B, Chibnall S, Bruce C, Townsend R, Thomas G, et al. Report on the AAU Campus Climate Survey on Sexual Assault and Sexual Misconduct. Rockville, MD: Westat; 2015 Sep.
- 24. Krebs C, Lindquist C, Berzofsky M, Shook-Sa B, Peterson K. Campus Climate Survey Validation Study Final Technical Report [Internet]. Bureau of Justice Statistics Research and Development Series; 2016. Report No.: NCJ 249545. Available: https://www.bjs.gov/content/pub/pdf/ccsvsftr.pdf
- 25. Breiding M, Smith S. Prevalence and Characteristics of Sexual Violence, Stalking, and Intimate Partner Violence Victimization—National Intimate Partner and Sexual Violence Survey, United States, 2011. Morb Mortal Wkly Rep Surveill Summ Wash DC 2002. 2014;63: 1–18.
- 26. Krahé B, Berger A. Men and Women as Perpetrators and Victims of SexualAggression in Heterosexual and Same‐Sex Encounters: A Study of First‐Year College Students in Germany. Aggress Behav. 2013;39: 391–404. pmid:23629691
- 27. Anderson R, Cahill S, Delahanty DL. The Psychometric Properties of the Sexual Experiences Survey-Short Form Victimization (SES-SFV) and Characteristics of Sexual Victimization Experiences in College Men. Psychol Men Masculinity. 2016; Advance online publication.
- 28. Turchik J. Sexual Victimization Among Male College Students: Assault Severity, Sexual Functioning, and Health Risk Behaviors. Psychol Men Masculinity. 2012;13: 243–255.
- 29. Cantor D, Fisher B, Chibnall S, Townsend R, Lee H, Bruce C, et al. Report on the AAU Campus Climate Survey on Sexual Assault and Sexual Misconduct. Rockville, MD: Westat; 2015 Sep.
- 30. Hines D, Armstrong J, Reed K, Cameron A. Gender Differences in Sexual Assault Victimization Among College Students. Violence Vict. 2012;27: 922–940. pmid:23393954
- 31. Rothman E, Exner D, Baughman A. The prevalence of sexual assault against people who identify as Gay, Lesbian or Bisexual in the United States: A systematic review. Trauma Violence Abuse. 2011;12: 55–66. pmid:21247983
- 32. Flack WF Jr. “The Red Zone” Temporal Risk for Unwanted Sex Among College Students. J Interpers Violence. 2008;23: 1177–1196. pmid:18319370
- 33. Carey K, Durney S, Shepardson R, Carey M. Precollege Predictors of Incapacitated Rape Among Female Students in Their First Year of College. J Stud Alcohol Drugs. 2015;76: 829–837. pmid:26562590
- 34. Coulter R, Mair C, Miller E, Blosnich J, Matthews D, McCauley H. Prevalence of Past-Year Sexual Assault Victimization Among Undergraduate Students: Exploring Differences by and Intersections of Gender Identity, Sexual Identity, and Race/Ethnicity. Prev Sci. 2017;Epub ahead of print.
- 35. Gross AM, Winslett A, Roberts M, Gohm CL. An examination of sexual violence against college women. Violence Women. 2006;12: 288–300.
- 36. Walsh K, Danielson CK, McCauley JL, Saunders BE, Kilpatrick DG, Resnick HS. National Prevalence of PTSD Among Sexually Revictimized Adolescent, College, and Adult Household-Residing Women. Arch Gen Psychiatry. 2012;69: 935–942. pmid:22566561
- 37. Humphrey SE, Kahn AS. Fraternities, Athletic Teams, and Rape: Importance of Identification With a Risky Group. J Interpers Violence. 2000;15: 1313–1322.
- 38. Minow JC, Einolf CJ. Sorority participation and sexual assault risk. Violence Women. 2009;15: 835–851.
- 39. Abbey A. Alcohol-related sexual assault: A common problem among college students. J Stud Alcohol Drugs. 2002;14: 118–128.
- 40. Flack WF Jr, Hansen B, Hopper A, Bryant L, Lang K, Massa A, et al. Some Types of Hookups May Be Riskier Than Others for Campus Sexual Assault. Psychol Trauma Theory Res Pract Policy. 2016;8: 413–420.
- 41. Flack WF Jr, Daubman KA, Caron ML, Asadorian JA, D’Aureli NR, Gigliotti SN, et al. Risk factors and consequences of unwanted sex among university students: hooking up, alcohol, and stress response. J Interpers Violence. 2007;22: 139–157. pmid:17202573
- 42. Tomsich EA, Schaible LM, Rennison CM, Gover A. Violent victimization and hooking up among strangers and acquaintances on an urban campus: an exploratory study. Crim Justice Stud. 2013;26: 433–454.
- 43. Dillman DA. Internet, Phone, Mail, and Mixed-Mode Surveys: The Tailored Design Method. 4th ed. Wiley; 2014.
- 44. Statistical Policy Working Paper 22 (Second version, 2005): Report on Statistical Disclosure Limitation Methodology. Federal Committee on Statistical Methodology, Statistical and Science Policy; Office of Information and Regulatory Affairs; Office of Management and Budget; 2005 Dec.
- 45. Israel BA, Schulz AJ, Parker EA, Becker AB. Community-based participatory research: policy recommendations for promoting a partnership approach in health research. Educ Health Abingdon Engl. 2001;14: 182–197. pmid:14742017
- 46. Senn C, Eliasziw M, Barata P, Thurston W, Newby‑Clark I, Radtke L, et al. Efficacy of a Sexual Assault Resistance Program for University Women. N Engl J Med. 2015;372: 2326–35. pmid:26061837
- 47. Johnson SM, Murphy MJ, Gidycz CA. Reliability and Validity of the Sexual Experiences Survey-Short Forms Victimization and Perpetration. Violence Vict. 2017;32: 78–92. pmid:28234199
- 48. The GenIUSS Group. Best Practices for Asking Questions to Identify Transgender and Other Gender Minority Respondents on Population-Based Surveys. [Internet]. Los Angeles, CA: The Williams Institute.; 2014. Available: https://williamsinstitute.law.ucla.edu/wp-content/uploads/geniuss-report-sep-2014.pdf
- 49. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption—II. Addict Abingdon Engl. 1993;88: 791–804.
- 50. Reinert DF, Allen JP. The alcohol use disorders identification test: an update of research findings. Alcohol Clin Exp Res. 2007;31: 185–199. pmid:17250609
- 51. Peng C-Z, Wilsnack R, Kristjanson A, Benson P, Wilsnack S. Gender Differences in the Factor Structure of the Alcohol Use Disorders Identification Test in Multinational General Population Surveys. Drug Alcohol Depend. 2012;124: 50–6. pmid:22236536
- 52. Bergman H, Källmén H. Alcohol use among Swedes and a psychometric evaluation of the alcohol use disorders identification test. Alcohol Alcohol Oxf Oxfs. 2002;37: 245–251.
- 53. Redford J, Van Wagenen A. Measuring Sexual Orientation and Gender Identity in a Self-Administered Survey: Results from Cognitive Research with Older Adults [Internet]. San Francisco, CA: American Institutes for Research; 2012. Available: http://paa2012.princeton.edu/papers/122975
- 54. Sell R. Measures. In: LGBTData.com [Internet]. Available: http://www.lgbtdata.com/measures.html
- 55. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988.
- 56. Ford J, Soto-Marquez J. Sexual Assault Victimization Among Straight, Gay/Lesbian, and Bisexual College Students. Violence Gend. 2016;3: 107–115.
- 57. Rankin S, Weber G, Blumenfeld W, Frazer M. 2010 State of Higher Education For Lesbian, Gay, Bisexual & Transgender People [Internet]. Charlotte, NC; 2010. Report No.: 978-0-9830176-0–8. Available: https://www.campuspride.org/wp-content/uploads/campuspride2010lgbtreportssummary.pdf
- 58. Mustanski B, Greene G, Ryan D, Whitton S. Feasibility, Acceptability, and Initial Efficacy of an Online Sexual Health Promotion Program for LGBT Youth: The Queer Sex Ed Intervention. J Sex Res. 2015;52: 220–230. pmid:24588408
- 59. Wahesh E, Lewis TF. Psychosocial Correlates of AUDIT-C Hazardous Drinking Risk Status: Implications for Screening and Brief Intervention in College Settings. J Drug Educ Subst Abuse Res Prev. 2015;45: 17–36.
- 60. Coker AL, Cook-Craig PG, Williams CM, Fisher BS, Clear ER, Garcia LS, et al. Evaluation of Green Dot: An Active Bystander Intervention to Reduce Sexual Violence on College Campuses. Violence Women. 2011;17: 777–796.
- 61. Jozkowski KN, Peterson ZD. College students and sexual consent: Unique insights. J Sex Res. 2013;50: 517–523. pmid:23039912
- 62. Jozkowski KN, Peterson ZD, Sanders SA, Dennis B, Reece M. M. Gender differences in heterosexual college students’ conceptualizations and indicators of sexual consent: Implications for contemporary sexual assault prevention education. J Sex Res. 2014;51: 904–916 pmid:23919322
- 63. Kingree JB, Thompson M. Fraternity Membership and Sexual Aggression: An Examination of Mediators of the Association. J Am Coll Health. 2013;61: 213–221. pmid:23663125
- 64. Murnen SK, Kohlman MH. Athletic Participation, Fraternity Membership, and Sexual Aggression Among College Men: A Meta-analytic Review. Sex Roles. 2007;57: 145–157.
- 65. Wade L. American Hookup: The new culture of sex on campus. 1st ed. W. W. Norton & Company; 2017.
- 66. Cohen R, Klahr R, Vedantam S, Penman M, Boyle T, Schmidt J, et al. Hookup Culture: The Unspoken Rules Of Sex On College Campuses [Internet]. Available: http://www.npr.org/2017/02/14/514578429/hookup-culture-the-unspoken-rules-of-sex-on-college-campuses
- 67. DeGue S, Valle LA, Holt MK, Massetti GM, Matjasko JL, Tharp AT. A systematic review of primary prevention strategies for sexual violence perpetration. Aggress Violent Behav. 2014;19: 346–362.
- 68. Banyard VL, Moynihan MM, Crossman MT. Reducing sexual violence on campus: The role of student leaders as empowered bystanders. J Coll Stud Dev. 2009;50: 446–457.
- 69. Coker A, Fisher B, Bush H, Swan S, Williams C, Clear E, et al. Evaluation of the Green Dot Bystander Intervention to Reduce Interpersonal Violence Among College Students Across Three Campuses. Violence Women. 2014; 1–21.
- 70. Silbaugh K. Reactive to proactive: Title IX’s unrealized capacity to prevent campus sexual assault. Boston Univ Law Rev. 2015;95: 1049–1075.