Abstract
Examination of situations in which bystanders missed intervention opportunities to prevent sexual assault can advance our understanding of how bystanders can prevent sexual assault. The present study utilized an incident-specific approach based on reports from 427 female sexual assault victims (ages 18-25 years) recruited via Mechanical Turk. Results indicate that bystanders had an opportunity to intervene before 23% of sexual assaults, alcohol use in settings with bystanders was widespread (by perpetrators, victims, and bystanders), and several factors were more common in situations involving missed intervention opportunity. Findings provide useful information for continued development of bystander training and new directions in bystander research.
The bystander approach has emerged in recent years as a promising method of addressing the high occurrence of sexual assault (SA), which is perpetrated against approximately one fifth of women (Elliott, Mok, & Briere, 2004). Bystander training programs are informed by both theory and research related to the bystander effect (Banyard, 2015; Latané & Darley, 1970), the well-documented phenomenon in which onlookers often do nothing to help an individual in need. By engaging bystanders, rather than potential victims or perpetrators, all members of a community are entrusted with the task of preventing SA. Even small changes by a large number of individuals in a given system could result in a meaningful shift in rape-supportive attitudes, behaviors, and environments. To date, a number of bystander training programs are documented to result in increased reported bystander behavior (e.g., Banyard, Moynihan, & Plante, 2007; Jouriles et al., 2016) and decreased SA perpetration (Coker et al., 2015; Salazar, Vivolo-Kantor, Hardin, & Berkowitz, 2014).
A major goal of bystander training programs is to teach individuals how to capitalize on opportunities to engage in behaviors to combat SA. McMahon and Banyard (2012) proposed a framework for understanding bystander opportunities, which can be either proactive or reactive. Proactive bystander opportunities arise outside the context of sexual risk but have potential to reduce future risk of SA. These include engaging others in discussions about SA or participating in SA prevention activism. By contrast, reactive bystander opportunities are linked to a specific incident of SA or expression of assault-supportive attitudes. These opportunities encompass situations involving bystanders prior to, during, and after assault. Primary reactive bystander intervention, which occurs before a SA takes place, may be particularly critical to reduce the occurrence of SA, but its impact is dependent upon how often bystanders have opportunities to intervene in a specific incident of SA risk. Undoubtedly, many potential SAs are averted through the behaviors of bystanders already, particularly in cases when individuals have undergone bystander training (Storer, Casey, & Herrenkohl, 2016). However, there are also likely many situations resulting in SA that involved missed opportunities, in which bystanders did not intervene (or intervened ineffectively). Indeed, situations that involve alcohol use by the perpetrator, victim, and/or bystander may present more barriers that thwart intervention and warrant attention (Leone, Haikalis, Parrott, & DiLillo, 2017). Examination of these missed opportunities can be helpful in further improving bystander training programs. The present study is the first to examine the occurrence, and common situational and individual factors, of missed bystander opportunities as reported by women who experienced SA in alcohol-related and non-alcohol-related situations.
Measurement of (Missed) Bystander Opportunities
To date, assessments of bystander opportunity have focused primarily on bystander reports of their own opportunities to address SA in response to specific risk situations. Bystander self-reports are useful in determining how often third-party individuals perceive opportunities to intervene. However, this approach is not without critical limitations. For example, instances of sexual risk can go unnoticed by bystanders (Burn, 2009) and thus lead to an underreporting of opportunities to intervene. In addition, when bystanders do intervene, they may have little knowledge of whether their intervention attempts are ultimately successful in thwarting an assault. Thus, bystanders who assume that their attempts are successful will overestimate the actual efficacy of their intervention.
In contrast to bystanders, victims of SA possess firsthand knowledge of the immediate context surrounding their victimization. As such, victims are in a unique position to identify and describe the situations in which bystanders were present but did not intervene. These reports can help to answer the question of how many additional SAs could be averted through reactive bystander intervention. Victims’ reports of missed opportunities to prevent SA can also shed light on whether and how bystanders attempted to intervene, as well as the situational and individual factors that were present when bystanders had opportunities to intervene.
Despite the potential value of victims’ reports, few studies have assessed the victim’s perspective to examine bystander intervention. Using National Crime Victimization Survey (NCVS) data, Planty (2002) found that victims ages 12 years and older reported third-party witnesses during nearly one third of SAs. However, the NCVS definition of SA encompassed a very broad range of experiences (e.g., grabbing sexual body parts), which could have resulted in a higher proportion of bystanders than a more limited definition of SA. More recently, a study examining victims’ reports of bystander presence found that someone “[saw] what happened” in 18% of cases and did something helpful to intervene in only 27% of those cases (Hamby, Weber, Grych, & Banyard, 2015). Although the purpose of this study was to assess bystander presence during the SA, the phrase “[saw] what happened to you” could be interpreted to include witnesses to the aftermath of the SA, which may have led some participants to overreport bystander presence. Furthermore, in both studies, reports did not reflect bystander presence just prior to the assault, when risk cues may be present and the opportunity for successful intervention is especially high.
Bystander Presence Versus Opportunity
Clearly, if bystanders are absent from the immediate context prior to an assault, there can be no opportunity for them to intervene. However, bystander presence alone does not guarantee that a successful intervention can occur. For example, cues that signal the presence of sexual risk can be subtle and not always detectable by third parties. In some cases, what may appear to be mutual flirtation may actually reflect unwanted sexual advances by a perpetrator. In other cases, sexual interactions could initially be consensual in public (i.e., when bystanders are present) and only later escalate to SA when the victim and perpetrator are alone. Thus, the measurement of bystander opportunity must include not only bystander presence in a sexual risk situation but also whether bystanders could have taken action to stop the SA. An incident-specific approach that examines victims’ firsthand accounts—and recognizes that victims are the experts on their own experiences—can thus be valuable in understanding missed opportunities for bystander intervention.
Situational Factors Present When Bystanders Have Opportunity
Understanding the situational factors associated with missed opportunities to intervene may help bystander intervention programs improve bystanders’ recognition of the need to intervene. For instance, information about common settings in which bystander opportunities are missed can inform the development of training vignettes. Different settings may demand the use of different bystander behaviors; for example, bystanders at a loud bar might need to employ nonverbal intervention strategies compared with those intervening in a quieter environment.
Alcohol use is a well-established risk factor for SA (e.g., Abbey, 2002; Testa, 2002). Although bystander training programs address drinking by victims and perpetrators as important risk factors for SA, it is unknown whether victim and perpetrator alcohol use differs as a function of bystander presence. Furthermore, the frequency with which bystanders are present during the escalation of alcohol-related SA is unknown. This knowledge is important, in part, because victim and perpetrator intoxication may make it difficult for bystanders to discern whether interactions are consensual. Intoxication may also make victims or perpetrators less receptive to bystander intervention attempts.
Relatedly, despite the high rates of alcohol-involved SA, the possibility that bystanders themselves may be drinking has rarely been considered in the literature or addressed in bystander training programs (Leone et al., 2017). Intoxication on the part of bystanders may impair their abilities to identify and intervene effectively in response to sexual risk (for a review, see Leone et al., 2017). Alcohol myopia theory (AMT; Steele & Josephs, 1990) offers a framework that may explain how alcohol can impair a bystander’s ability to intervene. AMT purports that the pharmacological properties of alcohol narrow attentional focus, restrict internal and external cues individuals perceive, and reduce one’s capacity to process meaning from information they do perceive. Intoxicated individuals allocate their attention such that they perceive and process only the most salient cues of a situation. In many cases, it is likely that environmental cues that inhibit bystander behavior (e.g., social norms) will be more salient than cues that would lead bystanders to intervene (e.g., potential victim’s discomfort).
Individual Factors Present When Bystanders Have Opportunity
Individual factors related to the victim or perpetrator may also indicate risk for SA and thus serve as prompts for bystander intervention. One such risk factor is sexual objectification, which occurs when women are reduced to their bodies or sexual functions for use by others (Fredrickson & Roberts, 1997). Although sexual objectification manifests in a multitude of ways, a common form is body evaluation, which includes behaviors such as ogling or making comments about women’s bodies (Kozee, Tylka, Augustus-Horvath, & Denchik, 2007). More extreme examples of sexual objectification include potentially assaultive behaviors, such as touching or grabbing a woman without her consent. These unwanted sexual advances are consistent with “pre-rape” behaviors (Rozee & Koss, 2001). Although sexual objectification may set the stage for SA (Gervais, DiLillo, & McChargue, 2014; Haikalis, DiLillo, & Gervais, 2017), it is unclear whether perpetrators engage in subtle or more obvious forms of sexual objectification that could be observed by bystanders prior to assault.
Victim discomfort with the perpetrator while in the presence of bystanders might also signal to bystanders that intervention is warranted. In particular, victims may display their discomfort with perpetrators in observable ways that can attract the attention of bystanders (e.g., averted gaze, distancing from perpetrator, unhappy facial expressions). However, when bystanders are intoxicated, they may be more likely to misinterpret these affective cues (e.g., Abbey, Zawacki, & Buck, 2005; Farris, Treat, & Viken, 2010). In other circumstances, victims may even communicate their discomfort directly to bystanders (e.g., “This guy is creeping me out”). Conversely, victims may not recognize their feelings as discomfort in the moment, and only later label them as such. Thus, they may not display any external signs of discomfort. Although victim discomfort can serve as a potential cue to intervene, it is unknown how often this occurs prior to a SA.
Finally, the nature of a victim’s relationship with the perpetrator and bystanders in situations involving bystander opportunity is unknown. Victims who have ongoing relationships with the perpetrator (e.g., as friends or romantic partners) may report less bystander opportunity to stop the SA, possibly because they are less likely to be around bystanders prior to the assault. In regard to victim–bystander relationships, there has been a particular focus on victims’ friends, because bystanders are more likely to intervene with friends than strangers (Bennett, Banyard, & Garnhart, 2014). In fact, one bystander training program focuses specifically on bystander intervention to take care of friends (Jouriles et al., 2016). Furthermore, individuals who know a SA victim are more likely to say that the individual is their friend as opposed to an acquaintance (Weitzman, Cowan, & Walsh, 2017). Thus, it is expected that situations with bystander opportunity to prevent SA will be more likely to involve the presence of bystanders who are friends to the victim.
Study Questions
Drawing on the above literature, the present study utilized a victim-centered, incident-specific approach to answer key questions relevant to bystander opportunity to prevent SA:
How often are bystanders present prior to and during SAs? Although we are not aware of studies that have examined presence prior to the onset of SA, studies examining bystander presence during SA (e.g., Planty, 2002) suggest that bystanders are likely to be present prior to the occurrence of a sizable minority of SAs.
When bystanders are present prior to SA, how often is there an opportunity for them to intervene to prevent SA? In contrast to prior work that has equated presence with opportunity, we expect that victims will report that bystanders who were present prior to SA had opportunity to intervene in a majority of, but not all, cases.
When bystanders have an opportunity to intervene prior to SA, how often do they engage in bystander behaviors? Because past research has focused almost exclusively on bystander reports, it is difficult to offer a specific hypothesis for this research question; however, the victim reports collected here should yield novel data about the frequency with which bystanders intervene when presented with the opportunity.
When bystanders are present prior to SA, is the pattern of victim and perpetrator alcohol use and alcohol impairment different than when bystanders are not present? Given the highly social nature of alcohol consumption (Kuntcshe, Knibbe, Gmel, & Engels, 2005), we expect there to be more alcohol use by victims and perpetrators when bystanders are present.
How often are bystanders using alcohol, and how impaired are they from alcohol, prior to SA? We expect bystanders to be intoxicated in a majority of cases.
When bystanders have an opportunity to intervene to prevent SA, what situational (i.e., setting, alcohol use) and individual factors (i.e., sexual objectification by perpetrator, unwanted sexual advances by perpetrator, victim discomfort with perpetrator, perpetrator–victim relationship, bystander–victim relationship) are present? We expect that SAs in which bystanders are present and have the opportunity to intervene (relative to not having the opportunity to intervene) will be characterized by more perpetrator sexual objectification and unwanted sexual advances toward the victim, victim discomfort with the perpetrator, close personal bystander–victim relationships as well as less bystander intoxication and less established victim–perpetrator relationships (e.g., boyfriend, spouse).
Method
Participants
Participants were 427 women (Mage = 22.95, SD = 1.73). To be eligible for the study, participants had to identify as female, be between the ages of 18 and 25 years, be a U.S. resident, and report experiencing a SA in the past year. As shown in Table 1, most participants were White, heterosexual, single, and had at least some college education. About half of the participants were college students at the time of their SAs. In general, participants experienced sexual victimization by a single perpetrator who was most often a dating partner.
Demographic Information and Descriptive Statistics.
Note. Relationship with perpetrator only represents participants who reported a single perpetrator because of issues with correct completion of the item for those with multiple perpetrators.
Participants were recruited through Amazon’s Mechanical Turk system (MTurk) to complete an online survey about “women’s life experiences, attitudes, and beliefs.” MTurk is an online marketplace that connects researchers with individuals willing to be research participants. All participants were compensated financially through the MTurk system. Participant samples collected via MTurk have comparable internal consistency and test–retest reliability to those obtained by traditional methods and are slightly more demographically diverse (Buhrmester, Kwang, & Gosling, 2011). The survey link was only available to MTurk workers who were U.S. residents.
Measures
Sexual victimization
Sexual victimization was measured during eligibility screening with three behaviorally specific items that assessed sexual acts perpetrated by men in the past year without the victim’s consent: (a) “had oral sex with you or made you have oral sex with him (a male put his penis in your mouth, or put his mouth on your vagina or butt),” (b) “put his penis into your vagina, or inserted fingers or objects,” or (c) “put his penis into your butt, or inserted fingers or objects.” For each item, participants were also asked whether they had experienced that activity as a result of five perpetrator tactics drawn from Koss et al.’s (2007) Sexual Experiences Survey–Short Form. These tactics were (a) verbal threats and pressure, (b) berating the victim, (c) incapacitation due to alcohol or drugs, (d) threats of force, and (e) force. Response options for all items were “yes” or “no.” Participants who met criteria for SA victimization (i.e., “yes” to one or more item) and who met other eligibility criteria proceeded on to the study. Participants who reported about more than one sexual victimization incident within the past year were then asked to report about the most recent incident for the remainder of the survey.
Bystander presence and opportunity to intervene
Bystander presence before the assault was assessed with the item: “Just prior to the incident (when the unwanted sexual experience occurred), were you and the man in a situation where other people were present?” Bystander presence during the assault was assessed with an item modified from Hamby and colleagues (2015): “To your knowledge, did any person see (or hear) what happened to you during the unwanted sexual experience, besides you and the man involved?” Response options for both items were “yes” or “no.” Those who indicated bystanders were present before the SA responded to a subsequent item that assessed bystander opportunity: “Do you think there is something someone (other than you or the man involved) could have done to stop the unwanted sexual experience from happening?” using a 5-point scale (1 = not at all, 5 = very much). This item was dichotomized in analyses to represent situations in which bystanders did (i.e., response of 2-5) or did not (i.e., response of 1) have opportunity to intervene.
Bystander behaviors
Reactive bystander opportunities elicit several potential bystander responses, including direct, indirect, passive, aggressive, and engagement intervention behaviors. To maximize the likelihood of capturing the full range of strategies that bystanders may employ, bystander behaviors were assessed using three different pools of items. First, 18-items from the Bystander Behavior Questionnaire (BBQ; Leone et al., 2017) were administered. 1 Some items from the BBQ were modified to make sense in the context of the present study (e.g., changed “asked the female if she is okay” to “asked if you were okay”). Second, an additional eight-items were modified from the Bystander Behaviors Scale–Revised (McMahon et al., 2014) to assess more types of direct intervention by bystanders. Finally, the authors developed five items to assess additional indirect bystander behaviors. In total, 31 items were used to assess prosocial bystander behaviors (direct intervention, indirect intervention, aggressive intervention behaviors), engagement bystander behaviors, and passive bystander behaviors. Participants responded to all items using a 5-point scale (1 = not at all, 5 = very much). Three items measured passive behavior by bystanders (e.g., “did nothing”; α = .76), 10 items measured indirect intervention (e.g., “tried to divert the man’s attention”; α = .94), 11 items measured direct intervention behaviors (e.g., “asked if you were okay”; α = .94), five items measured aggressive intervention behaviors (e.g., “pushed or shoved the man”; α = .96), and four items measured engagement behaviors, in which the bystanders allied with or had positive interactions with the perpetrator (e.g., “encouraged the man”; α = .82).
Setting
Participants who reported that bystanders were present prior to their SA answered an item to identify the setting that the victim, the perpetrator, and bystanders were in together (see Table 2 for response options).
Situational and Individual Factors as a Function of Bystander Opportunity When Bystanders Present.
Note. Numbers that are not percentages represent means and standard deviations.
Percentages exceed 100 because many situations involved multiple types of bystanders.
p = .052. *p < .05. **p < .01.
Alcohol use
Victims indicated whether they, the perpetrator, or bystanders were using alcohol prior to their SA. Response categories included yes, no, and unsure. If participants reported that any of these groups was drinking, they were asked to assess their level intoxication using an item from the Subjective Intoxication Scale (Zeichner & Phil, 1979). This measure uses a 12-point scale (0 = no impairment, 11 = strong impairment). If the victim reported that she or the perpetrator drank alcohol prior to the SA, she was asked whether (“yes” or “no”) alcohol consumption occurred in the presence of bystanders.
Sexual objectification
Sexual objectification behaviors by perpetrators while in the presence of bystanders were measured using 10 items from the Interpersonal Sexual Objectification Scale (ISOS), Body Evaluation subscale (Kozee et al., 2007). One item from the ISOS Body Evaluation subscale was excluded because it did not make sense in the context of the present study (i.e., “honked at”). Participants used a 7-point scale (1 = never, 7 = very frequently) to report about perpetrators’ body evaluation sexual objectification (e.g., “stared at your breasts/chest when you were talking to him”; α = .91)
Unwanted sexual advances
Unwanted sexual advances while in the presence of bystanders were assessed with six items. Two of these items were drawn from the Unwanted Explicit Sexual Advances Subscale of the ISOS (i.e., “touched or fondled,” “grabbed or pinched”). The remaining four items reflected actions by the perpetrator that involved unwanted physical contact, consistent with “pre-rape” behaviors (Rozee & Koss, 2001). Specifically, these items assessed acts in which the perpetrator “danced with,” “tried to kiss,” “forced you to touch (him),” and “removed some of your clothes” against the victim’s will or without her consent. Consistent with the ISOS, participants responded using a 7-point scale (1 = never, 7 = very frequently). Coefficient alpha for the 7-item scale was .88.
Victim discomfort
Victim discomfort with perpetrators was assessed with three items: (a) “Just before the unwanted sexual experience occurred, when you were still in the presence of others, was there a point when you felt uncomfortable with something that the man did?” (b) “Did you tell anyone around you that you were uncomfortable with the man before the unwanted sexual experience?” and (c) “Did you think that other people noticed that you were uncomfortable with man before the unwanted sexual experience?” For each, victims responded using a 5-point scale (1 = not at all, 5 = very much).
Perpetrator–victim and bystander–victim relationship
Participants also reported about the nature of the perpetrator–victim and bystander–victim relationship. Regarding victim–perpetrator relationship, participants selected the category that best described their perpetrator. Bystander–victim relationship was assessed with a similar item, which allowed participants to select multiple response categories to account for different bystanders present prior to their assault (see Tables 1 and 2 for response options).
Demographic information
A demographic questionnaire was used to obtain participant age, education, ethnicity/race, marital status, and sexual orientation.
Procedure
The description of the survey on MTurk informed participants that only U.S. women ages 18-25 years would be eligible to participate. Interested individuals who were eligible completed the informed consent process before proceeding to the survey. Of the 2,091 individuals who were screened, 578 individuals (27.6%) met eligibility through the gender, age, and SA screening items. Of these eligible women, 427 (73.8%) provided complete responses and passed the validity checks (e.g., “If you are paying attention, select 2”) and made up the final sample. All procedures were approved by the institutional review board (IRB) of the host institution.
Results
How Often Are Bystanders Present Prior to and During SAs?
Victims reported that bystanders were present or not present before their SAs in 34% (n = 147) and 57% of cases (n = 244), respectively. Victims reportedly could not recall whether bystanders were present before their assaults in 8% of cases (n = 36). Victims also reported that bystanders were present during the SAs in 3% of cases (n = 13). Furthermore, there was overlap in these cases, such that bystanders were present both before and during the SA in 2% of cases (n = 8).
How Often Is There Opportunity for Bystanders to Intervene to Prevent SA?
According to victims, bystanders had at least some opportunity to intervene to prevent the SA from occurring in 67% (n = 98) of the cases when bystanders were present (n = 147; see Figure 1). This corresponds to 23% of the total number of SAs reported (N = 427).

Bystander presence and opportunity among reported sexual assaults (N = 427).
When Bystanders Have Opportunity, How Often Do They Engage in Bystander Behaviors?
When victims reported that bystanders had an opportunity to prevent SA, the most common response of bystanders was to remain passive (86% of cases; M = 2.95, SD = 1.24). Bystanders engaged in indirect bystander behavior 42% of the time (M = 1.32, SD = 0.68), direct bystander behavior 58% of the time (M = 1.44, SD = 0.72), and aggressive bystander behavior 15% of the time (M = 1.23, SD = 0.71). Bystanders also participated in engagement behavior (problematic behavior in which they allied with the perpetrator) in 54% of cases (M = 1.60, SD = 0.86). Participants responded to bystander behavior questions independently and, as such, the percentages exceed 100%.
Does the Pattern of Victim and Perpetrator Alcohol Use Differ Based on Bystander Presence?
Victim alcohol use
Victims reported consuming alcohol prior to the SAs about half of the time (53%, n = 225). Chi-square analysis revealed that victim alcohol use differed based on the presence of others prior to the SA, χ2(1, n = 389) = 53.78, p < .001. Specifically, among the victims who reported bystander presence prior to their SAs, 76% (n = 111) reported consuming alcohol in the presence of bystanders. However, only 39% (n = 95) of victims who said no bystanders were present consumed alcohol before their SAs (see Figure 2). Victims also reported being moderately to severely impaired from alcohol (M = 7.56, SD = 2.71) while in the presence of others.

Victim, perpetrator, and bystander alcohol use based on bystander presence.
Perpetrator alcohol use
Participants reported that perpetrators consumed alcohol prior to the SA about half of the time (53%, n = 223). Victims were unsure of perpetrator’s alcohol use in 12% of cases (n = 51), which were not included in subsequent analyses related to perpetrator alcohol use. The pattern of perpetrator alcohol use was consistent with that of victim alcohol use. Specifically, chi-square analysis indicated that perpetrator alcohol use differed based on the presence of others prior to the SA, χ2(1, n = 343) = 41.05, p < .001. Among perpetrators who were around bystanders before they committed the SAs, 78% (n = 98) drank alcohol in front of those bystanders; of the perpetrators who did not have the pre-assault presence of bystanders, 47% (n = 101) drank alcohol (see Figure 2). According to victims, perpetrators were moderately impaired from alcohol (M = 7.02, SD = 2.28), prior to the assault while in the presence of others.
Co-occurring alcohol use
Among the cases that either the victim or perpetrator consumed alcohol in the presence of bystanders (n = 120), 74% involved alcohol consumption by both individuals (n = 89).
How Often Are Bystanders Using Alcohol?
When bystanders were present prior to a SA, they reportedly consumed alcohol in 88% of cases (n = 130; see Figure 1). Victims estimated that bystanders were moderately impaired from alcohol (M = 6.43, SD = 2.32).
Co-occurring alcohol use
Among the cases that either the victim or perpetrator consumed alcohol in the presence of bystanders (n = 120), 97% involved alcohol consumption by bystanders (n = 117). More specifically, 18% of cases involved alcohol consumption by only the victim and bystanders (n = 21), 8% involved alcohol consumption by only the perpetrator and bystanders (n = 9), and 73% involved alcohol consumption by the victim, perpetrator, and bystanders (n = 87).
What Situational and Individual Factors Are Present When Bystanders Have Opportunity?
Setting
The most common setting when bystanders were present before a SA and when they had opportunity to intervene was a house party. However, the setting did not vary as a function of bystander opportunity (see Table 2).
Alcohol use
Neither victim, χ2(1) = 1.76, p = .18, nor perpetrator, χ2(1) = 2.76, p = .10, alcohol use was associated with bystander opportunity (i.e., present with opportunity vs. present without opportunity). However, level of perceived bystander impairment from alcohol did differ based on bystander opportunity (see Table 2). Specifically, bystanders were seen as more impaired when there was opportunity to prevent SA.
Risk behaviors by perpetrators
According to victims, perpetrators engaged in sexual objectification in 98% of cases and unwanted sexual advances in 83% of cases, all prior to the SA while in the presence of bystanders. Furthermore, the frequency of these behaviors was greater in situations when there was an opportunity for bystanders to intervene compared with situations with no opportunity (see Table 2). Victims perceived that others were more likely to notice sexual objectification by perpetrators in situations involving opportunity to intervene (see Table 2).
Victim discomfort with perpetrator
Overall, most victims experienced at least some discomfort with the perpetrator while still in the presence of others (73%; that is, response greater than 1 [not at all]). Furthermore, those in situations with bystander opportunity to intervene were more likely to report experiencing at least some discomfort (82%) compared with those in situations with bystander presence but without bystander opportunity (55%), χ2(1) = 11.61, p < .01. Those in situations with bystander opportunity also reported more severe discomfort (see Table 2). Of those who felt any discomfort, 27% reported that they had disclosed their discomfort to a bystander prior to the SA (i.e., response greater than 1 [not at all]). Again, those in situations with bystander opportunity were more likely to disclose their discomfort to a bystander (34%) than those in situations with bystander presence but no reported bystander opportunity (7%), χ2(1) = 1.07, p < .01. Finally, victims who experienced discomfort in the presence of bystanders were much more likely to perceive that bystanders noticed their discomfort in situations with bystander opportunity (78%; that is, response greater than 1 [not at all]) than in situations without bystander opportunity to intervene (30%), χ2(1) = 20.45, p < .001. Victims were also more certain that bystanders noticed their discomfort in situations with bystander opportunity to prevent the SA (see Table 2).
Perpetrator–victim relationship
The victim’s relationship to the perpetrator did not vary as a function of bystander opportunity within situations where bystanders were present, χ2(10) = 15.22, p = .124. However, the relationship to the perpetrator did vary based on bystander presence, χ2(11) = 57.12, p < .001. Specifically, z tests with Bonferroni corrections (ps < .05) revealed that perpetrators were less likely to be a spouse/ex-spouse or current/ex-dating partner when bystanders were present (4% and 25%, respectively) than when bystanders were not present (16% and 45%, respectively). When bystanders were present before the SAs, perpetrators were more likely to be casual dates (24%), new hook up partners (7%), or acquaintances (13%) than when bystanders were not present (14%, 2%, and 3%, respectively).
Bystander–victim relationship
The rates of the types of bystanders present prior to the occurrence of a SA are shown in Table 2. In situations where victims reported opportunity for bystanders to have prevented the SA, the victim’s female and male friends were more likely to be present than in situations where there was bystander presence but no opportunity.
Discussion
The present study draws on victim reports to offer novel incident-level data related to bystander opportunity to intervene in SA. Among the most significant findings are that (a) approximately one fourth of SAs involve bystander opportunity for prosocial intervention; (b) alcohol use is highly prevalent among victims, perpetrators, and bystanders prior to SA; and (c) victim–bystander relationship, sexual objectification and unwanted sexual advances by perpetrator, and victim discomfort were related to bystander opportunity to intervene. These and other notable findings are discussed below.
Results indicated that bystanders were present during (vs. prior to) very few SAs (3%). This is far less than reports from samples using a broader age range and more inclusive SA definitions (Hamby et al., 2015; Planty, 2002). Conceivably, the more specific item within the present study that made it clear to report only about bystanders present during the assault was a factor in the smaller proportion of bystanders reported. Furthermore, many victims reported that the perpetrators engaged in unwanted sexual advances while in the presence of bystanders. Together, these results suggest that while bystanders may often have opportunity to witness and intervene in low-severity assaultive behaviors (e.g., unwanted sexual advances), bystanders will rarely have opportunity to intervene during more severe sexually assaultive encounters (i.e., unwanted oral, vaginal, or anal penetration). These results stress the importance of intervention at the first signs of sexual risk, rather than waiting for the escalation to more severe assaultive behaviors.
The distinction between bystander presence and opportunity to intervene was highlighted by findings showing that victims do not always perceive bystanders as being able to intervene prior to a SA. Specifically, whereas bystanders were present before approximately one third of assault cases, victims only reported there was an opportunity to intervene in 66% of those cases (23% of the total cases of SA). When bystanders had the opportunity to intervene, the most common response was to remain passive. Moreover, many victims reported that bystanders took some prosocial action involving indirect, direct, or aggressive action. Ultimately, because reports were drawn from SA victims, none of the prosocial bystander behaviors reported in this study prevented a SA. Although several victims reported that bystanders engaged in some prosocial behaviors, mean ratings of these behaviors suggest that bystanders were engaging in very low levels of pre-assault intervention attempts. Collectively, these findings demonstrate wider dissemination of bystander training programs or improvement of existing ones can have a significant public health impact, particularly in cases when bystanders are both present and have opportunity to intervene in situations with risk of escalating to SA. Prevention could be increased through training to reduce missed bystander opportunities and—when risk is noticed—by encouraging bystanders to persist in their intervention attempts, increase the intensity of their intervention, or employ different strategies if their initial attempts fail.
Although the overall rates of alcohol use reported by victims is consistent with prior research (Abbey, 2002; Testa, 2002), both victims and perpetrators were more likely to consume alcohol when bystanders were present than when they were alone. These results suggest that alcohol intoxication may be an even more important risk factor for bystanders to be aware of than previously thought. Victim or perpetrator alcohol use likely creates unique obstacles to effective bystander intervention. For example, determining whether interactions between potential perpetrators and victims are consensual may be more difficult when one or both parties are intoxicated. Although existing bystander training programs underscore the importance of recognizing that people cannot consent to sex if they are intoxicated, in practice, identifying the point at which someone is too intoxicated to consent can be challenging for a bystander.
Victims reported that bystanders were intoxicated in over three fourths of SA instances and in almost all cases where either the perpetrator or victim was also drinking. Thus, the present findings indicate that bystander alcohol use is exceedingly common in situations that lead to SA. The pervasiveness of bystander alcohol use has critical implications for bystanders’ abilities to intervene effectively (Leone et al., 2017). For example, bystander intoxication could create a barrier to recognizing the sometimes-subtle risk behaviors in which perpetrators engage due to the pharmacological effects of alcohol, which narrow attentional focus (Steele & Josephs, 1990). Even if intoxicated bystanders notice sexual risk, their alcohol-related impairment may make it more difficult to engage in effective intervention behaviors. Finally, bystander intoxication might further compound any challenges associated with victim and perpetrator intoxication (e.g., ambiguity about consensual interactions), making it even less likely that bystander interventions will succeed. Unfortunately, alcohol use by bystanders is rarely addressed in research and is sorely needed to strengthen existing bystander programming.
General characteristics of situations where bystanders were present included setting and victim-perpetrator relationship. The most common setting when bystanders were present was house parties. The victim-perpetrator relationship varied as a function of bystander presence prior to SA. Although dating partner was the most common category regardless of bystander presence, perpetrators were less likely to be spouses or dating partners and more likely to be casual dates, new hook up partners, or acquaintances, when bystanders were present prior to SA. These results will be helpful in the development of training scenarios that represent the most common opportunities for bystander intervention.
Situations where victims reported that bystanders had an opportunity to intervene (vs. merely being present) prior to assault were associated with (a) greater bystander intoxication, (b) greater likelihood of the victims’ friends as bystanders, (c) more frequent sexual objectification and unwanted sexual advances by perpetrators, (d) increased victim discomfort with the perpetrators, (e) greater effort by victims to communicate their discomfort, and (f) greater perception by victims that bystanders noticed their discomfort. Victims’ reports that their friends were more often present when there was opportunity to intervene complement findings that bystanders report being more likely to engage in intervention behaviors to protect friends (Burn, 2009). It is notable that even in situations with bystander opportunity, when most victims reported experiencing discomfort with the perpetrator, only 34% of victims communicated their discomfort to bystanders. These data suggest that bystanders cannot assume that potential victims will initiate communication of discomfort; rather, bystanders must be proactive in checking in with friends and others they notice at potential risk.
Limitations
While the present study advances our understanding of missed bystander opportunities to prevent SA, there are several limitations. First, despite the novel perspective that victim reports can bring to the bystander literature, these reports are subject to the usual limitations that arise from retrospective reports of complex events. Although participants reported on events within the past year, they may nevertheless have had difficulty recalling some aspects of their experiences, particularly when alcohol consumption was involved. Furthermore, victims may hold particular beliefs related to their assault that could shape their responses to survey questions. For instance, a victim who harbors self-blame for her assault may be less likely to report that bystanders had an opportunity to intervene. Future work might address this limitation through the use of event-level data collection strategies such as daily diaries or ecological momentary assessment. A second limitation is the relatively narrow age range of the sample (18-25 years). Although young adulthood is the period of highest risk for SA (Sinozich & Langton, 2014), the current results cannot be generalized to other age groups, such as adolescents or older populations. Third, the present work focused on male-to-female SA perpetrated. While this represents the most common form of SA, it is important for researchers to examine bystander opportunity for other forms of SA (e.g., female-to-male, same-sex, gang rape).
Implications for Research
Our findings indicate that bystanders were usually using alcohol in situations where they could have potentially intervened to prevent SA. More research is needed to examine how intoxication may influence bystanders’ abilities to enact intervention behaviors. It is possible that the myopic effects of alcohol could either promote or reduce effective intervention, depending on the prevalence of environmental cues that promote or inhibit bystander actions. Relatedly, future evaluations of bystander training programs should assess whether bystanders are able to implement intervention techniques successfully when they are under the influence of alcohol (for a discussion of bystander alcohol use, see Leone et al., 2017).
In nearly all instances, victims reported experiencing sexual objectification and unwanted sexual advances from perpetrators prior to their assaults. These behaviors occurred in the presence of bystanders who victims perceived could have helped but did not. Sexual objectification is commonly experienced by women and occurs on a continuum from minor to more extreme behaviors (Szymanski, Carr, & Moffitt, 2011). The relatively low rates of reactive bystander behaviors found here may result, in part, from a failure by bystanders to recognize more subtle forms of objectification that portend SA. Future research (e.g., using trained confederates or virtual reality) might examine the specific types of behaviors bystanders tend to notice, or could be trained to notice, as a way to inform bystander training efforts.
The present findings quantify missed opportunities for bystander intervention and identify barriers that may interfere with those efforts. Qualitative studies could build on these findings by providing useful descriptive details, not revealed through quantitative approaches, about how bystanders may prevent, or fail to prevent, SA from a victim’s perspective. Such research could also enhance our knowledge of who is likely to witness SA and how they may perceive such deleterious interactions while intoxicated. Moreover, future research is needed to elucidate barriers to intervention, and the perspectives of victims, bystanders, and perpetrators may each provide unique insight to better inform prevention programming.
Finally, the present study only examined missed opportunities, not instances of prevented SA. Examination of successful bystander intervention to prevent SA poses a unique challenge, because there is not always certainty that a SA would have occurred if not for bystanders. However, stronger conclusions can be made about the success of bystander intervention in response to situations involving imminent risk, or when an assault is ongoing. Future research should examine successful bystander behavior to identify effective intervention strategies.
Implications for Bystander Training
The incident-specific data provided here will be vital for efforts to improve bystander training programs. It is important to note that evaluations of bystander training programs to date have largely focused on outcomes such as bystander attitudes, intentions, or behaviors, rather than examining reductions in the occurrence of SA (Newlands & O’Donohue, 2016). As such, future work is needed to determine conclusively whether bystander training programs result in fewer cases of SA and recommendations from the present study are given with the expectation that such programs are capable of doing so.
The current findings indicate that alcohol may be an even more important risk factor for bystander intervention than previously thought and bystanders should be particularly vigilant in situations when a potential victim or perpetrator is consuming alcohol. Furthermore, given the high rates of missed opportunities while bystanders were intoxicated, bystander trainings should aim to reduce heavy drinking. However, given the likelihood that drinking among bystanders will persist, bystander programming should also consider intervention strategies that can be executed effectively under conditions of intoxication. In addition, programming efforts that aim to shift social norms that perpetuate SA and encourage heavy drinking would complement bystander approaches and have the greatest potential for reducing victimization (Newlands & O’Donohue, 2016). Indeed, the most effective prevention programs are comprehensive (Nation et al., 2003) and more multipronged approaches that incorporate a focus on changing social norms as well as specific bystander intervention behaviors in response to sexual risk are needed (e.g., Gidycz, Orchowski, & Berkowitz, 2011).
The present results elucidate situational and individual factors present when bystanders have opportunity to intervene. These factors may be useful for the refinement of existing bystander training programs. For example, the ecological validity of vignettes used to train bystanders could be improved by incorporating the most common factors associated with missed bystander opportunities identified here. Bystanders should practice responding to the commonly reported examples of perpetrator risk behaviors including sexual objectification (e.g., ogling) and unwanted sexual advances (e.g., grabbing body parts). Discussion with bystander trainees about the risk associated with such behaviors may encourage them to intervene, even when SA does not seem imminent. Bystanders must also be trained to initiate conversations with friends and others whom they think might be at risk, given our findings that few victims communicate their discomfort with the perpetrators to bystanders prior to SA.
The present study provides novel information about bystander opportunity to intervene in SA. Findings related to the occurrence of opportunity underscore the importance of bystander trainings continuing to emphasize both proactive and reactive bystander behaviors to reduce instances of SA. Results deliver a new level of detail about situations in which bystanders miss opportunities to intervene prior to SA. This work will be useful to inform continued development and refinement of bystander training programs to prevent SA and yield new directions for research related to bystander intervention.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Preparation of this article was supported in part by National Institute on Alcohol Abuse and Alcoholism grants F31AA024692 awarded to Michelle Haikalis and F31AA024369 awarded to Ruschelle M. Leone.
