Abstract
This cross-sectional study examined students’ (Campus 1, n = 1,153; Campus 2, n = 1,113) experiences with four situations of direct confrontation of those at risk for sexual assault perpetration. The most reported opportunity was to confront those making false statements about sexual assault; many students reported more than one opportunity to intervene in the past year. Bystanders intervened most of the time across the four situations examined in this study. The most reported consequence of intervening was that no further harm was caused. More nuanced measures can offer practitioners further information to tailor sexual violence prevention programs.
Among institutions of higher education (IHE), bystander-based models for the prevention of sexual violence (SV) are commonly used (Orchowski et al., 2018), and recent meta-analyses demonstrate the promise of these initiatives (Jouriles et al., 2018; Kettery & Marx, 2019). Still, researchers in this field continue to grapple with the methodology for measuring bystander intervention (BI) experiences. As BI scholarship has grown, measurement methodology has evolved to try and capture the nuances of individuals’ experiences as prosocial helpers in a range of situations that are violent or have the potential to become violent.
BI has been described as a primary violence prevention approach; however, measures commonly used to assess the efficacy of such programs (i.e., the Bystander Behavior Scale [BBS]; Banyard et al., 2007) are comprised of intervention behaviors that constitute secondary and tertiary prevention and/or harm reduction, which serve to mitigate harms associated with violence that has already occurred (DeGue et al., 2014). Although BI can reflect any number of prosocial behaviors, and intervention across the spectrum of violence is important for its prevention, our study gathered information on bystanders’ experiences—perceived opportunity, frequency of prosocial action relative to opportunity, and consequences of intervening—in confronting those at risk for perpetration. This focus was inspired by the founding principles of bystander-based prevention models to engage individuals in shifting norms that perpetuate violence (McMahon & Banyard, 2012). While we centered our quantitative measures on those specific situations, our methodology can be adapted for different research objectives.
Review of the Literature
Our work was guided by scholarship demonstrating that BI actions can represent conceptually distinct factors (Cascardi et al., 2021), and that bystanders’ experiences may differ depending on the severity of risk (Zozula et al., 2021) or level of prevention (i.e., pre-, mid-, or post-assault; Hoxmeier et al., 2015). Many BI measures use composite scores, which may muddy our understanding of when, and how, bystanders intervene (Hoxmeier et al., 2020). Examining students’ opportunities to intervene and their actions during those opportunities on a situation-by-situation level can show whether and how BI differs depending on the risk situation. This can help refine programming efforts based on commonly encountered situations (Cascardi et al., 2021). SV is a major public health issue among IHE, as demonstrated by the high incidence of victimization (Fedina et al., 2018; Krebs et al., 2009) and determinantal impact on health (Martin et al., 2011), academic (Banyard, Demers et al., 2020; Potter et al., 2018), and economic (Martin et al., 2011; Potter et al., 2018) outcomes. As such, primary prevention efforts, and methods of investigating the impact of bystanders on confronting those at risk for perpetration, are critical to reduce rates of campus SV (Dills et al., 2016). Furthermore, bystander-based prevention models are predicated on the assumption that intervention is helpful, and while evidence demonstrates that bystanders may experience positive reactions from those who perpetrate harm (Banyard et al., 2019), few studies have examined whether bystanders perceive their actions as preventing further harm.
Building on current research in the field, this study analyzed a methodology used to gather data on students’ experiences. Our work measured students’ perceived opportunities for, and reported frequency of engaging in, preventing violence though direct confrontation of those at risk for perpetrating SV. We prioritized such actions because they may correct potential perpetrators’ problematic behaviors (Reid & Dundes, 2018) and better reflect bystanders’ desire to rectify the social injustices that contribute to SV (Hoxmeier et al., 2019) rather than helping individuals avoid victimization, which has been linked to benevolent sexism (Leone et al., 2020).
Recognizing the complexity of quantitatively capturing BI experiences, our methodology builds upon current research to address several challenges in this field. Given that BI research favors assessing proactive behavior across a continuum of violence with composite scores (Hoxmeier et al., 2020), we analyzed item-level opportunities and actions to illustrate that these vary as a function of the risk contexts in which individuals can intervene. Only intervening prior to an assault has the potential to prevent violence; as such, we focused our examination on direct confrontation of those at risk for perpetrating violence in situations that could lead to an assault.
How Do We Measure When and How Bystanders Intervene?
In quantitative research, BI measurement methods generally include risk situations across the spectrum of violence and a range of actions taken in those situations. McMahon and Banyard (2012) categorized opportunities for prosocial helping in SV situations by when bystanders can intervene: pre-, mid-, and post-assault. Similarly, the widely used BBS (Banyard et al., 2007) includes 51 different helping behaviors that capture a range of intervention actions, including post-assault support for those who have experienced victimization. Other BI scales also include behaviors that can occur across the continuum of violence, from pre- to post-assault (i.e., Sexual Assault Bystander Behavior Questionnaire, Hoxmeier et al., 2015; Bystander Situation Questionnaire, Yule & Grych, 2017). Recent research shows that bystanders are more likely to act in high-risk situations, such as when violence is imminent, and more likely to help those at risk for victimization (Zozula et al., 2021), such that differentiating when, and how, bystanders intervene is critical for understanding the role they play in violence prevention. Scholars have encouraged specifying the nature of the relationship between bystanders and interveners, differentiating between actions taken with friends as opposed to strangers (Banyard et al., 2014). Regarding whether, and how, the relationship between the bystander and those at risk of perpetrating SV impacts intervention, Bennett & Banyard (2016) found that bystanders perceived situations involving their friends (who were at risk of perpetrating) as safer to intervene compared to situations involving strangers, and Burn (2009) found men expressed greater intent to intervene when it was their friend who was at risk of perpetrating SV. Collectively, this scholarship guided our decision to frame our BI items as actions taken with friends who were demonstrating risk for SV perpetration.
Regarding how bystanders intervene, qualitative research shows that bystander action takes various forms (Casey et al., 2018; Edwards et al., 2015; Hoxmeier et al., 2019; Lee et al., 2021; McMahon et al., 2013) and scholars commonly distinguish bystander action as direct or indirect (Berkowitz, 2009). Direct actions may include confronting a person acting aggressively or in harmful ways during or after a situation, whereas indirect actions may include distancing potential perpetrators from potential victims or requesting help from others. The BBS (Banyard et al., 2007) includes both direct (e.g., “If I saw a friend taking a very intoxicated person up the stairs to my friend's room, I would say something and ask what my friend was doing”; Banyard et al., 2005, p. 14) and indirect helping (e.g., “Call 911 if I hear someone calling for help” Banyard et al., 2005, p. 14). Similarly, Hoxmeier et al. (2020) include both direct or indirect intervention actions, based on McMahon et al. (2011 and 2014). In a different approach for capturing BI, some measures only ask if the person took action in a particular situation (i.e., Yule & Grych, 2017) without capturing what kind of action was taken. Scholars have argued that confronting potential perpetrators is key for preventing SV (Reid & Dundes, 2018). Therefore, our study used measures that specifically describe the direct confrontation of those demonstrating risk for perpetration of SV rather than indirect actions.
How Do We Measure Intervention Relative to Perceived Opportunity?
Increasingly, scholars in the field recognize the importance of assessing BI relative to perceived opportunity to distinguish those who failed to intervene versus those who had (or perceived) no opportunity to do so (McMahon et al., 2017). whether measures that ask respondents whether they performed BI behaviors and only offer dichotomized answers of “yes/no” are unable to account for perceived opportunity. A variety of methods have been used to assess perceived opportunity, including the provision of “yes/no/wasn’t in the situation” response options, calculating bystander action/opportunity ratios, and asking how often individuals had the opportunity to intervene in certain situations (see McMahon et al., 2013). Assessing action relative to perceived opportunity also allows researchers to calculate scores for missed opportunities (e.g., Brown et al., 2014; Kania & Cale, 2021).
Studying perceived opportunity can be beneficial for experimental and non-experimental research alike. It can provide insight into which risky situations are most common on campus, which student populations report the most opportunities to intervene, and can also calculate BI as a measure of action proportionate to the number of perceived opportunities (which could vary over time or as a result of programmatic efforts). Relatedly, BI may be more effectively captured when methods measure the frequency of prosocial action (the proportion of intervening relative to opportunities to do so; see Hoxmeier, 2019). Whether bystanders’ actions vary as a function of the number of opportunities they encounter has been questioned (Hoxmeier et al., 2020), and researchers recognize that BI is not a one-time, linear phenomenon (Banyard, 2015). Banyard and colleagues (2019) asked participants how many times they had observed risky situations and the number of times they took action, with response options ranging from 0 times to 10 or more times. Frequency of performing an intervention behavior has also been measured ordinally, such as 0 = not at all to 3 = 6 or more times (Coker et al., 2011) and 1 = always to 7 = never (Murphy Austin et al., 2016). Capturing perceived opportunity and action with continuous data, Murphy and Gidycz (as cited in McMahon et al., 2017), created a variable of the proportion of intervention by dividing prosocial action by the number of reported opportunities and recommended that items regarding intervention action directly follow those regarding intervention opportunities to prevent errors, such as respondents whose reported number of actions are greater than their reported number of perceived opportunities.
Assessing the timing of BI has meaningful implications for cross-sectional, longitudinal, and experimental designs. Providing respondents with specific time parameters, such as the past 3 months (e.g., Yule & Grych, 2017) or past year (e.g., Banyard et al., 2020), has the advantage of focusing on the recency of opportunities, and thus, behaviors. Collectively, this research guided our use of items to measure intervention relative to opportunity, as well as capturing the number of perceived opportunities to intervene in the previous year.
How Do We Measure the Consequences of BI?
Moschella and colleagues (2018, 2020) examined perpetrators’ reactions to intervention, as perceived by bystanders. In their findings, bystanders commonly reported that interventions were perceived negatively from perpetrators (and positively from victims). In a new measure of BI consequences, Banyard and colleagues (2019) captured outcomes including personal feelings (e.g., feeling embarrassment over the intervention) and consequences (e.g., getting in trouble over the intervention), in addition to negative and positive reactions from both perpetrators and victims (e.g., the intervention harmed a friendship, or the bystander was thanked for intervening). These measures help shed light on the range of potential consequences for bystanders who intervene. Just as we need to measure BI beyond a simple yes or no, we similarly need more nuanced response options to understand how often prosocial bystanders experience various outcomes. Given our focus on BI that prevents violence, we built upon Banyard et al.'s (2020) work to include an item assessing whether bystanders perceived that their action prevented those at risk of perpetration from committing further harm.
How Do We Analyze Intervention? Beyond Composite Scores of Intervention Actions
Despite arguments that BI is multi-dimensional (Banyard, 2015), a common practice in both explanatory and experimental research is to create composite scores of BI. Given the wide range of potential intervention behaviors, composite scores may miss the extent to which individuals intervene along the continuum of violence. It is important to recognize that how and when bystanders intervene is likely context-specific; multiple factors affect intervention decision-making and experiences (Banyard, 2015). The skillset needed to intervene in a low-risk primary prevention situation (e.g., when an individual makes a sexist comment) is very different than that which is required in a high-risk situation where someone is acting aggressively or to support someone seeking help or services for victimization.
The knowledge we currently have about barriers and facilitators of intervention, as well as evidence of programmatic effectiveness, however, typically comes from studies in which BI is examined as a mean or total score of past behaviors across the continuum of violence. Composite scores of BI may prohibit a more nuanced understanding of bystanders’ intervention experiences, which could guide programming efforts (Cascardi et al., 2021). For example, McMahon et al. (2015) found students were more likely to report perceived opportunities to intervene in low-risk situations, yet took action in those situations less frequently than in high-risk situations (McMahon et al., 2015, 2017). Recognizing the multi-dimensional nature of BI, our method isolates four separate risk situations to highlight the ways in which bystanders’ perceived opportunities for, and intervention in, vary depending on the situation.
The Current Study
Collectively, scholars have developed a range of measures to capture the complexities of quantitatively assessing BI experiences (Banyard et al., 2005, 2007; Hoxmeier et al., 2015, 2020; McMahon et al., 2011, 2014, 2017). We sought to build on this body of literature to develop specific measures to gather more in-depth information about the nature, frequency, and consequences of BI. We crafted these quantitative measures in recognition that BI can reflect any number of prosocial behaviors; although the methodology discussed here can be adapted for different research objectives, we sought to quantify BI specifically as direct confrontation of those at risk for perpetration. As such, this study seeks to answer the following research questions:
How much perceived opportunity do students report for intervening with those at risk for perpetration, and how do opportunities vary across four situations? How frequently do students report to take direct action in situations with those demonstrating risk for perpetration? What kinds of consequences do students report when they have taken direct intervention action?
Methods
Sample and Recruitment
Data for the current study were collected as a part of a larger campus climate study designed to assess experiences and perceptions related to sexual and dating violence conducted at two college campuses in the mid-Atlantic. All students, graduate, and undergraduate, enrolled in courses on the two campuses during the fall 2020 semester were invited to participate. To reduce the burden on participants, students were randomly assigned to either SV module or a dating violence module. The final survey tool included a total of 49 items for students, regardless of assigned module, who did not have a reported experience of dating or SV since coming to campus due to the survey's skip logic. It included 67 items for students who did report an experience of either sexual or dating violence since coming to campus (regardless of assigned module). The survey was administered to students online through the survey platform Qualtrics. The survey, which was open for 6 weeks, was publicized through a broad outreach campaign tailored to each campus (including print materials, social media, and direct communications). Participants were recruited over email and were entered into raffles to receive several cash and electronic prizes. All procedures were approved by the university Institutional Review Board.
A total of 2,424 students completed the survey. On Campus 1, 17.5% of eligible students accessed the survey; 9% of eligible students accessed the survey on Campus 2. Students were removed from the sample if they had declined informed consent (n = 148) and/or if they failed to answer any questions on the survey (n = 10), for a final analytic sample of 2,266 students (Campus 1, n = 1,153; Campus 2, n = 1,113). Because this study focuses on methodology for measuring BI experiences, as well as discussing opportunities and challenges therein (including sampling), we retained all participants despite missing data on any survey items, including the variables of study. On both campuses, the sample racial/ethnic identity (data and terms used therein were provided by the university's Office of Institutional Effectiveness) and student status (undergraduate or graduate) were similar to those at their respective campus. On both campuses, women were overrepresented in the survey sample. The mean age for Campus 1 and Campus 2 was 24.08 years (SD = 7.55) and 23.98 years (SD = 7.19), respectively. See Table 1 for sample characteristics.
Sample Characteristics (N = 2,266).
Note. Minoritized gender identities include participants who self-identified as agender, gender nonbinary, genderqueer, trans men and women, or another identity. Minoritized sexual identities include participants who self-identified as gay, lesbian, bisexual, fluid/pansexual, queer, asexual, questioning or unsure, same-gender loving, or another identity.
BI Experiences Measures
The research team reviewed prior research to select situations that centered bystander experiences in the prevention of SV by examining four BI behaviors in which bystanders directly confront those at risk for perpetration, including: (a) speaking up when someone is making false statements about sexual assault, (b) confronting a friend who plans to use alcohol to intoxicate someone to have sex with, (c) intervening when a friend is persistently trying to get another person to come back to their room with them, and (d) saying something to a friend who is taking an intoxicated person back to their room (McMahon & Banyard, 2012; McMahon & Farmer, 2011). Participants were presented with one BI item at a time. Each BI item was presented to participants as a brief scenario in narrative form such as, “Have you intervened with an intoxicated friend who is persistently trying to convince someone to come back to their room even though it is clear they do not want to?” When participants endorsed “I’ve been in this situation at least once in the last year,” a follow-up asked, “About how many times have you had the opportunity to take this action in the last year?” (“Action” here refers to the direct confrontation described in each measure of BI, that is, “intervened with an intoxicated friend.”) Response options ranged from 1 to 10+ on a drop-down menu. Only those participants who reported opportunities received another follow-up question, asking, “How often have you taken this action?” Response options were measured on a 5-point scale (0 = I have been in this situation, but I have never taken this action; 1 = When I have been in this situation, I have taken this action some of the time; 2 = When I have been in this situation, I have taken this action half of the time; 3 = When I have been in this situation, I have taken this action most of the time; and 4 = I have taken this action every time I have been in this situation). Based on the experience of Murphy & Gidycz (as cited in McMahon et al., 2015), in which some participants reported greater intervention actions than perceived opportunities, we measured frequency of action with ordinal responses, which also eliminated the need to create an additional variable of proportion of intervention action relative to perceived opportunity.
To assess the consequences of BI action, we reviewed the existing literature and ultimately created items based on Banyard et al.'s (2019) measures. Following the items which assessed frequency of intervention behavior, we asked participants about the frequency of experiencing consequences associated with the behavior (“How often did the following occur when you took action?”), including, for example, “the person who was engaging in the behavior got mad or upset with me” and “the person who was engaging in this behavior didn’t cause further harm because of my actions.” Response options were measured on a 5-point scale (0 = never, 1 = some of the time, 2 = about half of the time, 3 = most of the time, and 4 = always). Only participants who endorsed having acted some of the time, half of the time, most of the time, or every time they were in that particular situation received these follow-up items.
Results
Amount, or Frequency, of Intervention Opportunities Across Risk Situations
For the situations examined here, the proportions of participants’ perceived intervention opportunities were low. The most reported opportunity was the first (heard someone making false statements about sexual assault) with 9.08% (n = 97) and 10.19% (n = 102) students reporting to have at least one perceived opportunity for intervention in this situation at Campuses 1 and 2, respectively. Regarding the amount of opportunity, or the frequency of such in the last year, most students reported to have one opportunity, though 8.33% (n = 8) and 10.89% (n = 11) of students at Campuses 1 and 2 respectively, reported to have 10 or more opportunities for the first situation.
Frequency of Taking Intervention Action
Frequency of action varied between the type of risk behavior exhibited. However, the mean scores for each campus were above 3 on the 5-point scale (0 = never to 4 = always), indicating frequent intervention (see Table 2 for mean frequency scores). The proportion of students who reported to have “taken action every time” was lowest (Campus 1 = 57.73%; Campus 2 = 54.9%) in the most commonly reported situation (heard someone make false statements about sexual assault) and highest for higher-risk situations, for which a smaller proportion of students reported to have opportunities. For example, 3.2% (n = 32) of students at Campus 2 reported to have the opportunity to “intervene with an intoxicated friend who is persistently trying to convince someone to come back to their room even though it is clear they don’t want to,” Two-thirds of those students (67.74%, n = 21) reported one opportunity to intervene in the past 12 months; among those who reported to have the opportunity to take action in this situation, 81.25% (n = 26) reported to do so “every time.”
Bystander Intervention Opportunity, Amount, and Frequency of Action.
Note. Those with missing data were excluded when calculating the proportion of students reporting perceived opportunities. There were no missing data on frequency of action. Frequency of action was measured on a 5-point scale (0 = I have been in this situation, but I have never taken this action; 1 = When I have been in this situation, I have taken this action some of the time; 2 = When I have been in this situation, I have taken this action half of the time; 3 = When I have been in this situation, I have taken this action most of the time; and 4 = I have taken this action every time I have been in this situation).
Frequency of Experiencing Intervention Consequences
Results indicate that perceived consequences varied for bystanders (see Table 3 for mean frequency scores). For example, for those bystanders that took action, more than half reported that those at risk for perpetration “never” or “sometimes” got mad or upset when they intervened (Campus 1 = 25.93% and 24.44%, respectively; Campus 2 = 28.91% and 28.91%, respectively). Nearly half of bystanders reported that those at risk of perpetration did not feel “relieved” or “better” due to their intervention (Campus 1 = 49.25%; Campus 2 = 45.53%). Similarly, few bystanders reported that those at risk for perpetration “indicated that [the bystander] was helpful” or “thanked [the bystander] for stepping in” (less than 25% chose “most of the time” or “always” for this response). Mean scores for frequency were highest for bystanders who perceived that their action prevented those at risk of perpetration from causing further harm (Campus 1:
Mean Frequency Scores for Experiencing Consequences of Bystander Intervention Actions.
Note. Frequency of experiencing outcomes was measured on a 5-point scale (0 = never, 1 = some of the time, 2 = about half of the time, 3 = most of the time, and 4 = always).
Discussion
As the literature grows, and research and intervention initiatives refine, methods of measuring BI behavior should evolve to capitalize on our understanding of the nuances of these experiences. Building on current literature, we sought to describe a methodology for measuring BI experiences that center opportunities for, and frequency of engaging in, direct confrontation of risk behavior among those at risk for perpetration. We chose this focus because only those actions taken prior to an assault can prevent its occurrence, and there is potential for such interventions to correct problematic behavior.
We looked to scholars’ work in the field to understand and adapt approaches to BI measurement for intervention behavior relative to opportunity (McMahon et al., 2015), frequency of perceived opportunity (Murphy, 2014), intervention frequency (Banyard et al., 2020; Hoxmeier, 2019) and consequences of intervention (Banyard et al., 2019; Krauss et al., 2021; Moschella et al., 2018, 2020). Furthermore, our focus on direct confrontation of those demonstrating risk for perpetration, and consideration of risk situations as separate and distinct, was inspired by the conceptual work and arguments for such made by Berkowitz (2009), Cascardi et al. (2021), Hoxmeier et al. (2020), McMahon et al. (2013), and Reid and Dundes (2018). The following outlines implications of the findings, as well as describes how intentional consideration of both process and content of BI measures can benefit from future research in this area. Although we measured bystanders’ use of direct confrontation of those demonstrating risk for perpetration of SV, our methodology can be adapted to align with other objectives.
Perceived Opportunities, Frequency of Taking Action, and Intervention Consequences
Our results demonstrate that some students often have more than one opportunity to intervene in the four situations described earlier in this paper. The most reported intervention opportunity was to confront those making false statements about sexual assault, with a greater proportion of students (about 10%) reporting this opportunity, as well as higher mean scores indicating having more than one opportunity (about 3) to do so in the past 12 months. Confronting these beliefs can help shift the violence-perpetuating norms necessary to realize its end (McMahon & Banyard, 2012). Although false beliefs about SV may be conceptualized as “low-risk” on the continuum of violence (McMahon & Banyard, 2012) due to the absence of imminent harm, rape myth acceptance is a well-established antecedent of SV perpetration (Trottier et al., 2021). Bystanders have the potential to shift cultural norms that support the use of violence by confronting those who make false claims about SV and such actions should be emphasized as part of BI programming.
However, a low proportion of the sample generally reported having opportunities to intervene in the other situations presented in this study. Less than 5% of the samples reported opportunities to intervene in the other three situations, and mean scores of the number of opportunities equaled less than two for the previous 12 months. This highlights the importance of inquiring about opportunities. Investigations that examine behavior without opportunities would likely find a considerable number of participants reporting “0 times” or “never” when asked if they have taken action. Although we restricted participants to think about only four distinct risk situations, these methods demonstrate that the number of perceived opportunities to intervene varies as a function of the situation (a finding similar to that in McMahon et al., 2013). Such methods could be useful for needs assessments designed to understand which situations are most common, whereby subsequent research—and programmatic efforts—can focus on those most reported. In longitudinal designs, inquiring into the number of opportunities can provide benchmark data for whether the frequency of risk situations changes over time, and, when combined with robust sampling/strategic designs, can provide meaningful data on whether the incidences of risk situations are decreasing.
Like perceived opportunity, frequency of direct confrontation of perpetration-related risk behavior similarly varied between the four actions. When using the mean score of frequency, we observed that most students reported to have intervened at least some of the time. Like McMahon and colleagues (2013) who found that students reported more opportunity to intervene in low-risk situations but took less action in those situations, we found that mean scores of intervention frequency were lowest for the most common opportunity: confronting someone making false statements about sexual assault. Given the theory that bystander action is predicated upon identifying a situation as risky and thus, warranting intervention (Latane & Darley, 1970), the normalization of false statements about sexual assault, as evidenced by the frequency of which students report hearing them, may result in diminished risk detection in these situations. That is, there is potential for students’ intervention to be less likely in situations that are most common.
Mean scores of BI frequencies could be generated for research objectives seeking to understand groups at risk for low engagement (in cross-sectional research) and how intervention behavior changes over time (in longitudinal or experimental research). For example, scholars have investigated differences in BI based on age, racial identity, gender, and/or sexual identity (see e.g., Brown et al., 2014; Hoxmeier et al., 2020, 2021, 2022) using measures that capture a single intervention opportunity or action; however, given the complexity of accounting for perceived opportunities and prosocial action, measuring frequency of BI may be a better way to investigate whether and how demographic characteristics relate to intervention. Additionally, using measures that assess intervention frequency could be useful for program evaluation efforts such that analyses examining pre- and post-test scores of BI frequency can illustrate how behaviors change over time and as a result of training.
Consistent with prior research (Banyard et al., 2020), our study shows that bystanders experience, to some extent, negative consequences because of their interventions. Evidence suggests that bystanders may perceive barriers to confronting potential perpetrators, such as negative reactions, that can be circumvented by helping potential survivors, who may be more appreciative of such assistance (Moschella et al., 2018; Reid & Dundes, 2018). Our results show that, although bystanders reported that those they intervened with “got mad or upset” with them, at least some of the time, they were also “thanked” for stepping in and that those at risk for perpetrating violence were “relieved or felt better” because of the intervention. For bystander-based models to encourage primary prevention, it is imperative that we support bystanders’ confrontation of risk behavior by providing evidence of its positive outcomes.
As IHE increasingly implement campus climate surveys, which are intended to produce actionable data, it is important to have meaningful BI measures. Our findings can also help guide the development of prevention programming by highlighting students’ experiences (McMahon et al., 2019), such as the types of risky situations students commonly encounter and the actions less frequently reported, as well as the consequences of those actions. By sharing data with developers of campus SV prevention, awareness, and education programming, it may be easier to determine the specific bystander opportunities and risk situations that are relevant for research and practical objectives. That said, this methodology also illustrates the importance of sample sizes. Although it was not the goal of this study to analyze the demographic and/or psychosocial correlates of BI consequences in-depth, the small number of students reporting BI opportunities here would make it prohibitive to do so.
Research Implications, Challenges, and Future Directions
Our focus was on measures of direct intervention with those at risk of perpetrating SV, to better assess bystander action as it relates to primary prevention. This builds on previous scholarship calling for SV prevention efforts to prioritize intervention with students who are at risk of engaging in harmful actions toward others (DeGue et al., 2014), as well as informed by discussions of how these tactics have the potential to correct problematic behavior that contributes to violence (Hoxmeier et al., 2019; Reid & Dundes, 2018), as opposed to conceptualizing intervention as harm reduction, or helping individuals avoid victimization, which has been linked to benevolent sexism (Leone et al., 2020). With bystander-based models proliferating as a prevention strategy across the country, research about how these efforts specifically impact those at risk for perpetration is needed to understand how this model aligns with primary prevention of violence. Scholars must recognize the multidimensional nature of BI, and, in discussing this model as one for the prevention of violence, we must consider the broad evidence base, and the measures used therein, that support such conclusions. The methodology presented can help researchers better understand students’ attempts to intervene in situations leading to SV in a more nuanced manner, and thus can provide rich information for practitioners designing prevention programs.
Further research in this area is critical. While mean scores of the frequency that “the person didn’t cause further harm” were highest relative to other consequences, a considerable proportion of participants reported that this “never” occurred, indicating that direct confrontation does not necessarily result in the prevention of violence. Scholarship and practice often assume that BI is a desirable outcome; yet recent research indicates that from the perspective of survivors, bystander action is not always helpful and can sometimes actually be harmful (Hoxmeier & McMahon, 2021; McMahon, 2022). Coupled with the findings here that direct confrontation does not always prevent further harm, continued examination of the consequences of BI is necessary. Little is known about whether BI has a positive impact on perpetration risk behavior; further work is needed to understand how intervention impacts those who engage in harmful behavior, particularly about how risk behavior changes over time as a result of intervention. Because research currently relies almost exclusively on the self-reports of bystanders, alternative methods should be considered, such as observational and/or lab simulations (Parrot et al., 2012). Dyad research on those who intervene and those who received the intervention would be beneficial, particularly as it relates to how those at risk for perpetration interpret the intervention action. BI is just one component of primary prevention and thus, measuring how it may work in conjunction with other efforts (such as gender transformative approaches that focus on men) to impact potential perpetrators—and perpetration—is critical (Casey et al., 2018).
Although the structure for BI measures here focused on intervening with potential perpetrators, it could be adapted for settings and communities where intervention experiences may be different. Researchers and practitioners can investigate intervention opportunities that are relevant for their objectives and campus and community settings. Formative research with students (and community members in non-campus settings) and practitioners can provide information on common risk situations to tailor assessments to align with programmatic initiatives. More work is needed to incorporate a social justice approach to SV prevention, including bystander-based models. This includes ensuring the BI situations and actions used in research and practice are appropriate for, and reflective of, the diverse communities in which this work occurs. As BI research evolves, it is essential to investigate how students with various identities experience this prevention strategy. Students who are marginalized by racial identity and/or sexual identity, for example, may experience additional burden and risk as bystanders, particularly if the larger campus climate is not perceived as welcoming or inclusive (Klein et al., 2020; McMahon et al., 2020). Future work must incorporate questions about how individuals’ identities and the larger campus or community climate impact BI.
A major challenge of both developing and utilizing a measure of BI that captures the wide breadth of opportunities and actions respondents may encounter is the need to administer a survey that is a reasonable length. Some best practices to reduce survey fatigue are the use of rotating modules to assess for multiple domains, as well as collaborating with other offices to coordinate large-scale surveys around other survey initiatives. Related to survey fatigue, missing data can be an issue; in our study, less than 8% of Campus 1 and just over 10% of Campus 2 did not provide data on perceived opportunities; in subsequent follow-up items, missing data were rare on the amount, or frequency, of opportunities and frequency of experiencing intervention consequences.
Limitations and Conclusions
The potential merits of this study must be discussed in the context of several limitations. We limited our examination to four intervention behaviors, and students likely have other opportunities for, and experiences with, prosocial actions. Relatedly, our measures referred to a “friend” who was at risk for perpetrating SV, and results may have differed for measures not specifying the nature of the relationship between bystander and the person at risk for perpetration. Although Bennett & Banyard (2016) found that bystanders perceived greater safety for intervention with a friend who was at risk of perpetrating SV, they were also less likely to identify that risk behavior as problematic. Participants also may not have wanted to disclose that it was a friend of theirs demonstrating risk for SV perpetration. Whereas we focused on pre-assault situations, or primary preventative BI, others may wish to examine post-assault support or disclosure, dating violence, potential or actual victims, or a variety of other situations and samples to make comparisons between students’ experiences. Relatedly, we focused on direct action; however, bystanders may have taken indirect action (e.g., asking for others for help) that we did not capture. Our measure of consequences was not implemented for each BI situation, such that direct confrontation of those at risk for perpetration in these four situations may result in different consequences. As measures continue to evolve, researchers must balance the importance of inquiry with the avoidance of respondent fatigue.
Our study was cross-sectional; as experts discuss (e.g., Banyard, 2015), BI is not linear, nor should we assume that the engagement captured here is stagnant and unmodifiable. These measures lend themselves to experimental and non-experimental longitudinal designs to assess how opportunities, frequency of action, and consequences change over time, as well as the relationship between the occurrence of negative and positive consequences and continued engagement as prosocial bystanders. Finally, though results were similar across the campuses, the response rate for Campus 2 was low and participation for both campuses may reflect students’ interest and engagement in interpersonal violence-related topics. This study was also conducted during the COVID-19 pandemic; campus and community restrictions may have impacted both students’ opportunities for intervention as well as their decisions whether to intervene as bystanders. Finally, we emphasize that the findings here reflect bystander perceptions of their experiences, particularly with respect to opportunities to intervene and consequences thereof. Although advances in BI research have given way to observational/lab studies, additional innovative approaches, such as dyad studies with bystanders and those for which they intervene, would help parse out how bystanders identify risk for SV, as well as the impact of their actions.
BI scholarship has evolved over the last several decades, and the field continues to grapple with questions about methodology for measuring these experiences. Building upon and integrating current research in the field, this study describes more nuanced measures that focus on bystanders’ opportunities for intervention, frequency of acting, and consequences associated with confronting those demonstrating risk for perpetration. As measurement options grow, we hope the methodology discussed here can help address challenges researchers face in their work for a specific investigative intent to understand pre-assault, direct intervention.
Footnotes
Acknowledgments
The authors would like to thank Jennifer Perillo at the Rutgers School of Social Work for her copyediting assistance.
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) received no financial support for the research, authorship, and/or publication of this article.
