The literature on sexual violence in a university context has more often reported the risk posed by athletes as perpetrators (McCray, 2019; Bonar et al., 2020) than the risk of sexual violence towards varsity athletes. In a recent review of the literature on risk factors associated with sexual violence in the university setting, Bonar et al. (2020) concluded that being a member of a men’s university athletic team was a risk factor for sexual assault perpetration. Murnen and Kohlman (2007) explained that this could be due to hypermasculinity values and level of aggressiveness in male sport, those problematic values and behaviors being associated with sexual violence. Indeed, it appears that male athletes adhered to more rape myths and held more rape-supportive attitudes than non-athletes (Boeringer, 1999; Sawyer et al., 2002; Young et al., 2017). In contrast, studies that have examined the adoption of sexually violent behaviors are not as straightforward as those linking male athlete status and rape-supportive attitudes (Murnen and Kohlman, 2007; Tharp et al., 2013). Boeringer (1996) found that athletes did not engage in sexual violence behaviors to a greater extent than non-athletes in a university setting. Furthermore, another study identified that membership in a sports team contributed to explaining only 1% of the variance in sexual assault in a university setting (Koss and Gaines, 1993). Conversely, Young et al. (2017) showed that college athletes reported higher rates of sexual coercion than non-athletes.

While these findings are contradictory and deserve further research, we can question whether the university sports environment represents a risk factor for victimization for athletes. In this regard, Bonar et al. (2020) concluded that male peer affiliation such as with an athletic team represents a risk factor for the perpetration of sexual violence in the university environment. Given that some male athletes may more likely be perpetrators, their peer athletes may be at increased risk because they more often interact with them. The Routine Activity Theory (Cohen and Felson, 1979) have been used in the past in sexual violence victimization studies (e.g., Cass, 2007; Snyder, 2015; Davis et al., 2021) and could explain this potentially elevated risk for athletes. This theory explains the occurrence of crime by three contributing factors. In the case of sexual violence, those would be a motivated perpetrator (i.e., male peer athletes), a suitable target (other athletes), and an absence of a capable guardian. For the last factor, we know that the sport context has often been singled out for its institutional tolerance of sexual violence and other forms of maltreatment against athletes (Jacobs et al., 2017; Parent and Fortier, 2018; Roberts et al., 2020). Based on those factors and the literature presented above, there could be more perpetrators in the sport environment, or a lack of capable guardians given the greater adherence to rape myths in the masculine sport environment.

Fasting, Brackenridge, and Sundgot-Borgen first proposed the Sport Protection Hypothesis in 2003. This hypothesis is based on the belief that athletes “develop strength, self-confidence, and a sense of physical adeptness through their sports experiences” (Fasting et al., 2003, p. 428) and that this could provide a form of protection or defense against sexual violence. In 2008, Fasting et al. (2008) tested the Sport Protection Hypothesis, stating that university student-athletes somehow benefited from some form of protection against sexual aggression (penetrative sexual intercourse against their will) compared to other students. The results supported this hypothesis by documenting that university student-athletes (women and men) reported significantly less sexual violence during their late high school and early college years than non-athletes. The authors explained this phenomenon by two main factors or processes: (1) victims of sexual violence may be less likely to join a university sports program, or (2) a sports program or the practice of a sport inherently protect athletes from being exposed to this form of violence (Fasting et al., 2008). These results were corroborated by the work of Parent et al. (2016), who showed that the odds of lifetime prevalence of sexual abuse were 1.32 times higher for adolescents not involved in organized sports than those who were. The results of Parent et al. (2016) and those of Fasting et al. (2008) suggest that participating in sports is a protective factor against the lifetime prevalence of exposure to sexual violence.

Although innovative, these results remain fragmentary and merit further research to counteract some limitations. For example, because they tested the Sport Protection Hypothesis using lifetime prevalence rates of sexual abuse and current athlete status, there is a risk of confusing cause and effect or the timing of events. Since childhood sexual abuse can occur long before becoming an athlete, sport participation may not be protective. Instead, it may be that those with past abuse are less likely to become athletes. To test the Sport Protection Hypothesis, we should thus strive to document sexual violence exposure while being an athlete. Clarifying this antecedent would improve current designs. Another important limitation of these studies is the lack of control for known risk factors for sexual violence victimization in the university context to account for potential confounding variables. These should include belonging to an ethnic, sexual or gender minority (Martin-Storey et al., 2018; Eisenberg et al., 2021; Klein and Martin, 2021), having a disability (Bergeron et al., 2016), being an international or Indigenous student (Dion et al., 2021; Fethi et al., under review)¹ or having been sexually abused as a child (CSA) (Bergeron et al., 2016).

Studies on sexual violence and sport in university settings have mainly focused on the risk of athletes committing sexual violence rather than being exposed to it. Therefore, the current body of knowledge does not allow a clear understanding of the role that sport participation plays in sexual violence in the university context. Studies of sexual violence against varsity athletes are also limited by their failure to include control groups and various known risk factors and their sole focus on lifetime exposure to sexual violence. In addition, they sometimes document childhood abuse and sexual violence in the university context indiscriminately. The purpose of this study is to fill the gaps from previous studies and determine whether varsity sport participation is a risk or a protective factor for sexual violence victimization in university. We will test these hypotheses while controlling for other known risk factors for sexual violence victimization to ensure that if a risk exists it is not better accounted for by confounding factors. A secondary objective is to determine whether the context of violence differs between varsity athletes and non-athletes.

The data for this article come from the ESSIMU study (Enquête sur la Sexualité, la Sécurité et les Interactions en Milieu Universitaire) conducted in Québec, a broad survey of students, teaching staff and other employees at six francophone universities regarding sexual violence on university campuses (Bergeron et al., 2016). Varied strategies were used to recruit the sample. The principal one was a massive email invitation to answer the online questionnaire sent to the entire university community, using the institutional email lists. The sole criterion for participating in the survey was to be employed or studying at one of the six universities during data collection (January to May 2016). The survey obtained the approval of human research ethics committees at the universities in question. The overall sample consisted of 9,284 participants. A total of 6,485 students with complete data on sexual violence, athlete status, and gender were included in the present study. From this total number of students, 267 identified themselves as varsity athletes. Descriptive statistics about the final sample of participants are presented in Table 1 in the Results section.

The complete questionnaire was composed of 13 sections and took an average of 15–20 minutes to answer. The two sections of interest for the present study were the sociodemographic measures described below and the sexual violence in the university context survey.

Descriptive analyses were first conducted on all study variables to determine whether they distinguished varsity athletes from non-athletes (chi-square). The Sport Protection Hypothesis was tested using a series of logistic regressions on each dichotomous measure of violence. We included athlete status as a predictor and all sociodemographic characteristics associated with sexual violence victimization or distinguishing between varsity athletes and non-athletes as confounding variables. There were 0.3% of missing values for the overall data set. Less than 2% of participants had missing values for each variable, and 4% of respondents had at least one missing value. Such low missingness suggests imputation would bring negligible benefits (Schafer, 1999), specifically for logistic regressions, which are less susceptible to bias from missing data (Bartlett et al., 2015).

Results of the four logistic regression analyses on the yearly incidence rates of the three dimensions of sexual violence and their combination using varsity athlete status as a predictor are presented in Table 2. Known predictors of sexual violence victimization on campus and participant characteristics that distinguish varsity athletes from non-athletes were included as confounding variables (age, graduate level, gender identity, sexual identity, disability status, visible minority status, international student status, Indigenous status, childhood sexual abuse). The four logistic regression models, including all predictors against a constant-only model, were statistically significant. For the combination of the three dimensions of sexual violence, the model explained 12.4% of the risk of exposure to sexual violence over the past year. The other three models explained 11.3% of the risk for sexual harassment, 8.7% of unwanted sexual behavior, and 6.9% of sexual coercion.

When considering all confounding variables in the four yearly incidence rates of sexual violence, the results revealed that being a varsity athlete did not significantly affect the risk of being exposed to sexual violence at university. All considered, other known risk factors were more significant predictors of the past year’s risk of sexual violence victimization than athlete status was. Overall, age and child sexual abuse were significant risk factors in all four models. Graduate level, gender identity, sexual identity, disability status, and international student status were risk factors in at least one model. Visible minority status and Indigenous status, like athlete status, did not reach significance levels in any models.

Many variables were associated with an increased risk for the past year’s combined forms of sexual victimization regardless of one’s varsity athlete status. Younger students, both for the 18–25 (3.06 times) and the 26–35 (2.02 times) categories were more at risk than those in the 36++++ category. Female students were 1.66 times more at risk than males. Participants who identified with the sexual minorities category were 1.83 times more at risk, while those who were uncertain or questioning of their sexual identity were at 1.90 times more at risk than heterosexual participants. International students were 1.45 times more at risk than domestic students. Students with a disability were 1.35 times more at risk than students who did not have disability. Finally, participants who were sexually abused as a child were 5.78 times more at risk of victimization and those who were uncertain or did not want to answer this question were 1.76 times more at risk than those who had not experienced CSA.

Analyses were conducted using only the sample of participants who reported at least one form of sexual violence against them (n = 2,307). Table 3 shows the results of the chi-squared analyses comparing the contextual characteristics of sexual violence of varsity athletes and non-athletes. Most students, including varsity athletes, reported that the perpetrator was another student. Whereas the reported proportion of other perpetrators was similar, varsity athletes were seven times more likely to have reported a sports coach than non-athletes (Cramér’s V = 0.130). About half of students, including varsity athletes, reported sexual violence during parties or social activities, excluding initiations or in a teaching context (Cramér’s V = 0.102). Varsity athletes were three times more likely to report sexual violence against them in the sporting context and ten times more likely during a sports initiation than non-athletes (Cramér’s V = 0.185). While these three contextual factors were significantly different between varsity athletes and non-athletes, the effect sizes are small.

Our results are not concordant with the Sport Protection Hypothesis (Fasting et al., 2003) and previous work showing less sexual victimization in athletes (Fasting et al., 2008; Parent et al., 2016). In addition, the first explanation offered by Fasting et al. (2003) for the sport protection effect, namely that victims of sexual violence may be less likely to join a university sports program, is not supported by our results. We did not observe significant differences between varsity athletes and non-athletes regarding CSA victimization. The second explanation is related to the fact that sport in itself protects athletes from sexual victimization. Based on this second possibility, sport could be a protective factor against revictimization later in life, such as in the university context. However, our results do not support this explanation either. Indeed, while both athletes and non-athletes reported similar CSA rates, athletes were not less at risk of experiencing sexual violence in a university context than non-athletes.

If we cannot demonstrate support for the Sport Protection Hypothesis based on our results, we cannot support the hypothesis based on the Routine Activity Theory (Cohen and Felson, 1979). Following this theory, varsity athletes could be considered more at risk than other students on campus because they are in frequent contact with other male peer athletes, considered more probable perpetrators, and because of the institutional tolerance of sexual violence in sports (Jacobs et al., 2017; Parent and Fortier, 2018; Roberts et al., 2020). Despite previous evidence showing that being a member of a men’s university athletic team was a risk factor for sexual assault perpetration (Bonar et al., 2020), our results showed that varsity athletes were not more at risk of experiencing sexual violence when controlling for other important confounding variables. Based on this observation, we can postulate that interactions with potential perpetrators (i.e., male peer athletes) do not necessarily mean that other peer athletes become targets of those perpetrators. Previous findings were contradictory, with some showing that male athletes adhered to more rape myths and held more rape-supportive attitudes than non-athletes (Boeringer, 1999; Sawyer et al., 2002; Young et al., 2017), while not engaging in more sexual aggression behaviors than other groups in a university setting (Boeringer, 1996). Another study identified that membership in a sports team contributed to explaining only 1% of the variance in sexual assault in a university setting (Koss and Gaines, 1993). More research that would consider both the risk of victimization and of perpetration of varsity athletes could shed light on this issue.

When looking strictly at the incidence of sexual violence in the university context between varsity athletes and non-athletes, we observed that a higher proportion of varsity athletes reported sexual harassment (32 vs. 25%) and coercion (4 vs. 2%) over the past year, and more reported at least one form of sexual violence over the past year (35 vs. 27%). However, this increased yearly incidence rate seems to be primarily explained by other known risk factors. More specifically, as varsity athletes were younger, more likely to be undergraduates, Indigenous and international students, these differences may explain the higher uncorrected yearly rates. It does not appear to be because they were varsity athletes that they reported more sexual violence, but because they were more likely to report other known risk factors. This result means that sport participation is not a protective or a risk factor in our sample. The risk of experiencing sexual violence in a university context for varsity athletes might be interacting with individual and contextual factors unrelated to sport. Our results are in line with new research findings indicating that, in comparison to physical or psychological violence, sexual violence in sports is not very well explained by sport-specific factors, such as culture, norms, type of sport, level of sport, or hours of practice per week. Indeed, Vertommen et al. (submitted)² recently showed that the main sport-factor predictors they used explained only 1% of sexual violence compared to 8% for physical violence, 10% for psychological violence and neglect, and 12% for instrumental violence. The results of Demers et al. (2021) also demonstrated that only 1.4% of sexual violence experienced by athletes (from peers) and 5.9% (from coaches) was explained by sport-specific social norms. In their work, (see text footnote 2) pointed out that more sexual violence-specific variables, such as social isolation or health problems, need to be included in future research looking at sexual violence in sports.

The second objective of our study was to determine whether the context and the profiles of perpetrators of sexual violence differed between varsity athletes and non-athletes. Our results showed that only three variables distinguished the two groups as varsity athletes were more likely than non-athletes to report: (1) sport coaches as perpetrators, (2) violence occurring in a sport context, and (3) a sport initiation context. Those results are not surprising as varsity athletes are more exposed to those contexts and perpetrators. However, our results show that only 7% of varsity athletes reported a coach as the perpetrator while most reported another student (90%). These proportions are similar to those of non-athletes in our study. Our data did not document whether student perpetrators were athletes and thus if varsity athletes were more likely than non-athletes to be victimized by another athlete. Previous studies have suggested this using varsity athlete samples where the most frequent perpetrators of sexual violence against them were peer athletes (Elendu and Umeakuka, 2011; Rintaugu et al., 2014). Further research could deepen our understanding of this question.

For both varsity athletes and non-athletes, most participants reported that the perpetrator’s gender was male, at 92 and 91% respectively. This is in line with previous research in sport (Bjørnseth and Szabo, 2018) and outside of sport (Dartnall and Jewkes, 2013; Tharp et al., 2013; Smith et al., 2017). A more surprising finding is that 30% of varsity athletes and 28% of non-athletes reported that the perpetrator gender was female. Research on sexual violence in sport involving female perpetrators is very limited. The study of Vertommen et al. (2017) pointed out that although male perpetrators were predominant, there was a notable proportion of female perpetrators in sport. Their results showed that 9% of athletes identified female perpetrators and 15% identified perpetrators of both sexes. This is supported by qualitative studies showing that female coaches or peer athletes could also be perpetrators (Johansson, 2018; Sanderson and Weathers, 2020). In another study, sport participation was identified as a risk factor for sexual harassment and sexual coercion in female high school students (Cheever and Eisenberg, 2022). The authors underlined that male students were proportionally more likely to be perpetrators but also pointed to the gap in the literature concerning athlete female-perpetrated sexual violence.

Similarly, results presented in our study show that Routine Activity Theory (RAT) was limited in explaining the risk of sexual violence for varsity athletes in university settings. Based on this theory we postulated that varsity athletes would be at increased risk of sexual victimization because they were potentially more in contact with male athletes and that male athletes’ status in university was associated with sexual violence beliefs and behaviors. However, we found that varsity athletes were victims of sexual violence not only by males but also by females, albeit in a lower proportion. This finding challenges the expectations set by RAT. The data collection did not include various contextual characteristics that would have been useful in furthering our understanding. For example, while participants identified the gender of the perpetrator, we did not ask their perpetrator’s athlete status. We also did not have details about the participants’ own athletic involvement, such as the type of sport practiced (team or individual, single or mixed sex). Those characteristics could help better understand the real risk posed by male varsity athletes towards other athletes for sexual violence in university settings. Finally, as Johansson (2018) stated, heteronormativity and predominant male perpetrator – female victim stereotypes may prevent looking closely at same-sex violence in sport. Results of some prevalence data in sport shows that female and male athletes are equally at risk of experiencing sexual violence (Parent et al., 2016; Parent and Vaillancourt-Morel, 2021), meaning that in sport, male to male sexual violence (or female to male) could be more frequent than expected. In sum, further studies should include more detailed contextual information that could allow comparing female and male perpetrators in terms of types of sexual violence committed and the related contextual characteristics.

In terms of context, our results show that varsity athletes were three times more likely to report sexual violence occurring in the sporting context and ten times more likely during a sports initiation than non-athletes. However, about half of students, including varsity athletes, reported being sexually victimized during parties or social activities. It is possible that varsity athletes also experienced sexual violence in sport-related parties or social activities. However, our study did not document this level of contextual detail as it was not specifically designed for athletes.

The current study has several strengths. Most importantly, it is the first to our knowledge that compares varsity athletes with non-athletes’ risk of sexual violence on campus. Moreover, these comparisons were conducted while controlling known risk factors for sexual violence in a university context to account for potential confounding variables. We were also able to document in a more clearly delineated manner sexual violence before becoming a varsity athlete (childhood sexual abuse) and sexual violence while at university (since entering university and last year). This improves on previous tests of the Sport Protection Hypothesis. However, some limitations should also be noted. Although the convenience sample was large and diverse, it did not represent all university students. Moreover, it is possible that students who had previously been abused had abandoned their participation in sports. This would contribute to decreasing the overall prevalence rates for varsity athletes as leaving sport is one of the consequences of sexual violence reported by athletes (Fasting et al., 2002; Fogel and Quinlan, 2021). Another limitation of our study is that our questionnaire did not provide information on the type of sport practiced by varsity athletes, which prevents us from carrying out analyses according to the types of sports practiced. Finally, the study prevented the inclusion of students who might be athletes outside of university context. It would have been pertinent to include in our analyses these athletes who are not part of varsity teams but might still have experienced sexual violence at their university.

In conclusion, while the current study is helpful to deepen our understanding of varsity athletes’ risk of sexual violence in the university context, much remains to be explored. The analyses revealed that the context of varsity sport might not confer protection against or represent a risk factor for all athletes. As such, the study’s findings have cast doubt on the Sport Protection Hypothesis and that the varsity sport environment is conducive to sexual violence, which we hypothesized based on prior research results and the Routine Activity Theory (Cohen and Felson, 1979). Further investigations should be conducted to determine if subgroups of varsity athletes are more at risk of sexual violence victimization. Our findings indicate we need more information on varsity athlete subgroups such as undergraduates, Indigenous, and international students to determine if they are more at risk and, if so, why. Further analyses of potential interactions among these intersecting risk factors and varsity athlete status need to be conducted. This could help develop adequate and specific preventative measures. A longitudinal study that would track participants from their teenage years onward would also allow for more robust evidence of the impact of sport participation. Such study would indicate whether sport participation represents a protective factor against sexual violence later in life or a context conducive to developing rape-supportive attitudes and behaviors.