Abstract
Prior research documents that perceived peer norms are related to bystanders’ intentions and intervention behaviors in the context of sexual violence. Given the popularity of bystander intervention programming, it is important to know if variables like gender, race, or year in college impact intervention attitudes/behaviors or interact with perceived peer norms. Also relatively unexplored is the question of missed opportunities for intervention. For our final sample of 232 college students (66% female, 36% Black), screened by age, race, and missing data from an initial pool of 315 respondents, perceived peer norms supporting intervention positively predicted willingness to intervene against sexual violence (bystander intentions) but did not independently predict bystander behaviors or missed opportunities for intervention. Although women reported greater bystander intentions than did men, and Black participants reported more bystander behaviors than did White participants, gender, race, and year in college often interacted with peer norms and with each other in complex ways. Specifically, the predicted positive relationship between peer norms and bystander behaviors was observed only among Black students in at least their second year of college, and the predicted negative relationship between peer norms and missed opportunities was observed only for Black men. These nuances in factors that influence bystander actions have important implications for tailoring prevention tools on college campuses.
Over the past few decades, researchers and educators have begun to appreciate that many of the health and safety issues faced by college students not only are individual problems but also have a strong social component. Research has shown that behaviors such as alcohol use and sexual behavior are related to perceived norms for those behaviors. For instance, students drink more and engage in casual sexual behavior more often when they believe that their peers engage in and approve of these behaviors (Borsari & Carey, 2001; Stinson, 2010). These problems are also gendered—with men and women often holding different attitudes about the behaviors and about how to address or prevent problem behaviors as well as different levels of engagement in the behaviors themselves (Bogle, 2008; Suls & Green, 2003).
An appreciation of the social nature of risky behaviors facilitates the development of novel intervention strategies that capitalize on social perceptions and behaviors—strategies that are arguably more effective than those that simply urge students to change their own behaviors. One such strategy involves communicating healthy or prosocial norms—with the result being that students often change their attitudes and behaviors to fit what they believe is normative (Perkins & Craig, 2006). A related strategy for preventing violence, injury, and other harmful attitudes and behaviors among college students takes advantage of the fact that peers are often first responders (Sharkin, Plageman, & Mangold, 2003) as well as witnesses of these incidents. Likewise, peers have opportunities to use “informal social control” to express disapproval about behaviors that violate community norms (Chaurand & Brauer, 2008). Additionally peers often engage in “informal helping” when they are first told about problems that another peer is having and are therefore in a position to provide support or connections to more formal campus services (Sharkin et al., 2003; West & Wandrei, 2002). In turn, these prosocial behaviors may affect the perceptions of others and contribute positively to the prevailing community norms.
Research on perceived peer norms has highlighted the need to educate college students about resources and skills for intervening across a variety of situations (e.g., bias incidents, vandalism, friends’ use of alcohol, peers’ eating concerns, peers’ mental health problems, and interpersonal violence). Of particular concern is mobilizing peers around the problem of sexual violence because individuals aged 16–24 are at the highest risk for victimization (Bureau of Justice Statistics, 2011). Indeed, a number of sexual assault prevention programs for both male and female audiences on college campuses have now begun to focus on training students to be empowered, helpful, prosocial bystanders to prevent sexual violence before it occurs, to intervene when it is occurring, and to be an ally to the victim after sexual violence has occurred (Banyard, Moynihan, & Plante, 2007; Cares et al., in press; Moynihan et al., in press). Even though research and programming on prosocial behavior in situations of sexual violence is increasing, further understanding of how to promote positive helping behaviors in this context is needed and is a focus of the current study. Specifically, our goal is to refine our understanding of the relationship between college students’ perceived social norms about preventing sexual violence and their willingness to intervene against sexual violence, as well as their experience with such intervention, by exploring the possible moderating effects of gender, race, and year in college.
Peer Norms and Bystander Intervention
Social norms, specifically perceptions about peers’ attitudes and behaviors, are relevant in the context of helping because they may affect willingness to intervene as a bystander (Berkowitz, 2002). Several studies have shown that male peer norms supporting the use of coercion in relationships are a key factor in increasing rates of sexual violence (DeKeseredy, Schwartz, & Alvi, 2000; Schwartz & DeKeseredy, 1997; Schwartz, DeKeseredy, Tate, & Alvi, 2001). Additionally, Brown and Messman-Moore (2010) and Banyard and Moynihan (2011) found significant relationships between peer support for sexual violence and research participants’ behavioral intentions to be active or helpful bystanders. At least two studies have specifically shown that men’s willingness to intervene in the prevention of sexual violence was strongly related to their perceptions of their peers’ willingness to intervene (Fabiano, Perkins, Berkowitz, Linkenbach, & Stark, 2003; Stein, 2007). However, two limitations of most of these studies are that (a) the researchers studied men exclusively and (b) they measured only willingness to intervene, not actual intervention behaviors.
The few studies that have measured actual bystander intervention behaviors (e.g., Banyard & Moynihan, 2011; Moynihan et al., in press) still have an important weakness: Their work did not determine whether reported failure to intervene reflects an unwillingness to intervene or a lack of opportunity to intervene. Although Moynihan and colleagues (in press) did assess opportunity with a general 1-item measure, they did not assess opportunity for each of the individual bystander behavior items as we have done in the current study. Thus, to date, no known studies have assessed actual opportunities for specific bystander interventions against sexual violence. Thus, more work is needed to clarify how perceived peer norms relate to intended behaviors, as well as to actual behaviors and opportunities to act, and whether the impact of norms on intentions and behaviors is similar for students in various demographic groups.
Bystanders’ Characteristics and Helping
A broad base of empirical studies in social psychology highlights additional variables beyond social norms (both within the individual and the social context) that promote helping. These include knowledge and awareness of the problem, confidence in one’s ability to help, sense of bonds or ties to the community, and the number of and relationship to other bystanders who also witness the situation (see Banyard, 2011, for a review). Social identity variables, especially gender, have been highlighted in this work—with men and women more likely to help in ways that are consistent with prescribed gender roles (Burn, 2009; Dovidio, Piliavin, Schroeder, & Penner, 2006; Eagly & Crowley, 1986). More specific to helping behavior in instances of sexual or relationship violence, a number of researchers have found key gender differences: Women tend to report being more willing to provide support to survivors and to report more often acting to help in risky situations, whereas men report greater barriers to helping (Banyard, 2008; Beeble, Post, Bybee, & Sullivan, 2008; Burn, 2009; McMahon, 2010; Moynihan et al., in press; West & Wandrei, 2002).
Although bystanders’ gender has been extensively studied, other demographic characteristics may also play an important role in helping behavior. Some studies have examined the impact of race on helping, but race has typically been explored in terms of race of the victim being helped (Saucier, Miller, & Doucet, 2005). Fewer studies have examined the impact of race of the helper (Laditka & Laditka, 2001), and those studies that have examined race have failed to find significant relationships between participant race and bystander intervention (Christy & Voigt, 1994; Frye, 2007).
A few studies have looked at the relationship between age and helping and have tended to show decreases in helping as people age. Longitudinal research with adolescents showed overall decreases in prosocial behavior as respondents increased in age from middle to high school (Carlo, Crockett, Randall, & Roesch, 2007), a study of college students observed a negative relationship between age and prosocial bystander behaviors (Banyard & Moynihan, 2011), and a study of adults showed more helping related to crime or violence among younger (under age 55) in contrast to older (over 55) bystanders (Zhong, 2010). One longitudinal study of first-year college students found declines over 1 year in bystander actions related to sexual and relationship abuse (Moynihan et al., in press), suggesting that college experiences are influential. Thus, within a college sample with a restricted age range, a more important variable may be year in college. However, to our knowledge, no research has examined the relationship between year in college and bystander intentions or behaviors.
Bystanders’ Characteristics and Peer Norms
To date, most studies of helping have examined main effects of variables such as gender, race, or social norms. However, gaps in the literature and some mixed results suggest that interaction effects may be important. Even though few studies exist to suggest how gender, race, or year in college might interact with norms in the context of intervening against sexual violence, they are important variables to study. Katz (2006) has argued that race and other cultural factors can influence how men react to sexual aggression and how important social norms are to their reactions. Given that interaction effects on helping in the context sexual violence are understudied, it is useful to look at studies of general helping or studies that have examined moderators of peer norms in contexts such as drinking and smoking behavior. This research may help us predict how perceived peer norms might interact with bystander characteristics in regard to intervening against sexual violence. For example, there is reason to suspect that the relationship between perceived peer norms and bystander intentions or behaviors might be different for men and women. Work by George, Carroll, Kersnick, and Calderon (1998) found that, for men, empathy was significantly related to greater amounts of time spent helping friends, but this variable was not as critical for women. For women, closeness of the relationship to the person in need of help was significantly related to helping. Similar gender differences have been found in research on college drinking norms, such that perceived peer norms have a more powerful influence on the drinking behavior of men than of women (Prentice & Miller, 1993; Suls & Green, 2003). The relationship between peer norms and behavior may also be moderated by race and year in school: Hong, Rice, and Johnson (2012) found that descriptive norms had a stronger impact on the smoking behavior of Caucasian than of African American high school students. They also found that peer norms became more predictive of smoking behavior from 9th to 12th grade. Within the study of bystander intervention against sexual and relationship violence, Banyard and Moynihan (2011) found that the predicted relationship between perceived peer approval of sexual coercion and bystander intervention was more apparent among junior- and senior-level college students than among first-year students.
Opportunity and Intervention Behavior
What we learn from existing research is that bystander intervention is potentially important in the prevention of sexual violence and that peer norms, in combination with individual level characteristics (gender, race, and year in college), likely influence bystander involvement. However, interaction effects are largely uninvestigated in the context of bystander intervention against sexual violence, and more research is needed. What is more, the relatively small number of studies that have measured actual bystander intervention has generally failed to consider opportunities for intervention. That is to say, if members of one category (e.g., men) report fewer intervention behaviors than members of another category (e.g., women), it might be because the former (men) are reluctant to intervene when given the opportunity. However, it could also be that men in this example have fewer opportunities to intervene. Research has shown that women report both greater willingness to intervene and more actual intervention behaviors (Banyard, 2008; Banyard & Moynihan, 2011; Moynihan et al., in press), which suggests that, with regard to gender, lower rates of intervention may be due to lower willingness to intervene rather than to lack of opportunity. Thus, number of opportunities for intervention is a variable that deserves further study.
The Current Study
The current study aimed to examine demographic correlates of self-reported helping, the association of peer norms and helping, as well as how these demographic variables interact with peer norms to explain self-reported bystander intentions, behaviors, and missed opportunities (failing to intervene when given the opportunity). We put forth two specific hypotheses and four more tentative hypotheses, as well as investigated a number of other relationships without posing directional hypotheses.
For our first specific hypothesis, we predicted that greater perceptions that friends approved of bystander intervention related to sexual violence (peer norms) would be associated with (a) greater self-reported intent to help, (b) self-reports of more actual intervention behavior, and (c) fewer missed opportunities for bystander intervention (Hypothesis 1). Given the findings from the large body of research on gender and helping, with our second specific hypothesis, we predicted that (a) women would self-report greater intent to intervene than would men, (b) women would report a greater number of intervention behaviors than men, and (c) women would report fewer missed opportunities than would men (Hypothesis 2).
Little research exists to allow for the prediction of main effects for race, and the dearth of research examining year in college as a predictor of helping precludes making a strong prediction for main effects for year in college. However, based on research on the relationship between age in relation to helping (Banyard & Moynihan, 2011; Carlo et al., 2007), we tentatively predicted that first-year students, compared to more advanced students, would report (a) greater intent to intervene, (b) more intervention behaviors, and (c) fewer missed opportunities (Hypothesis 3). We also made tentative predictions regarding the interactions of perceived peer norms with gender, race, and year in college (Hypothesis 4). Based on research by George et al. (1998) and work on gender and drinking norms, we tentatively predicted that peer norms would have a stronger association with bystander intentions, behaviors, and missed opportunities for men than for women (Hypothesis 4a). Based on the work by Banyard and Moynihan (2011) and Hong and colleagues (2012), we also tentatively predicted that the impact of peer norms on intentions, behaviors, and missed opportunities would be stronger for White than for Black students (Hypothesis 4b) and for more advanced students than for first-year students (Hypothesis 4c). Because characteristics such as gender and race do not operate in isolation but instead intersect in important ways (Crenshaw, 1991), we also explored two- and three-way interactions among the bystander characteristics of race, age, and year in college without offering specific hypotheses for these potential interaction effects.
Method
Participants
Our initial sample included 315 students enrolled in introductory psychology classes at a medium-sized university in the southern part of the United States. Participants received extra credit in their course for participating. Approximately 500 students were invited to participate, yielding a response rate of about 63%. Because sexual assault is a particularly relevant problem among young adults (Bureau of Justice Statistics, 2011), and because we were interested in the interaction between college year and norms, we chose to limit analyses to participants falling between the ages of 18 and 24 years for whom college norms would likely be more salient. Because almost 94% of the initial sample was either Black (33.8%) or White (59.9%), and because analyses included race as an independent variable and endeavored to explore interactions between race and bystander peer norms, participants identifying as any race other than Black or White were excluded from analyses.
Thus, the original pool of 315 students was reduced to 282 after removing 23 individuals who were younger than 18 or older than 25, 9 individuals who failed to report an age, and 1 individual who failed to report her year in college. Of this sample of 18- to 24-year-olds, 18 individuals who identified their race as other than Black or White were removed, resulting in a sample of 264. Data were further reduced by eliminating 32 individuals who failed to complete two or more key surveys (perceived peer norms, bystander intentions, or bystander behaviors). To assure voluntary participation, answers were not required on any question, and participants could skip entire questionnaires by skipping to the next page; it was not possible to calculate scale scores for individuals who skipped all questions on a measure. Scales were scored using available item analysis (Parent, 2013), so missing data at the item level (which did not appear to be nonrandom) were not considered a problem. Failure to report age or complete any questionnaires was not associated with any study variables. Thus, 56 Black women, 27 Black men, 96 White women, and 53 White men remained in our final sample. This sample reported a mean age of 19.22 years (SD = 1.32). However, primary analyses are based on a somewhat smaller sample because six individuals failed to complete the intentions or behaviors survey but had complete data on other study variables and were thus retained for analyses on which data were complete.
Procedure and Materials
In April 2012, several instructors of introductory psychology courses provided their students with a link to an online survey and offered them bonus points for participating in the survey. After connecting to the survey and indicating their consent to participate, participants completed a number of measures: a demographic questionnaire and measures of bystander intentions, perceived peer norms, and behaviors. The demographic measure always came first; the bystander behaviors measure, last. The bystander intentions and norms measures were presented in random order in the middle of the survey. All responses were anonymous. After completing the bystander behaviors measure, participants were directed to a different survey where they indicated their name, student ID number, and course information; this information was then given to instructors for the purpose of granting extra credit. All procedures were approved by the Institutional Review Board of the university where data were collected.
Demographics
The demographics questionnaire included questions about gender (female = 0, male = 1), age, year in college, and racial or ethnic identity. A single item asked about racial or ethnic identity and included the following categories: Black, White, Hispanic, Asian/Asian American, Middle Eastern, Native American, and an additional open-ended “other” category. Participants were instructed to mark as many categories as needed. Participants were considered Black if they selected “Black” and no other categories; participants were considered White if they selected “White” and no other categories. Race was coded as 0 (White) and 1 (Black). Because year in college was highly skewed, with 77.2% of the sample indicating “first-year student,” year in college was dichotomized to distinguish between first-year students and more advanced students and coded 0 (first-year) and 1 (second-year and above).
Bystander intentions
Our measure of bystander intentions was adapted from measures created by Banyard (2008) and Banyard, Moynihan, Cares, and Warner (2013) as a self-report assessment of likelihood of performing different bystander interventions. The current version contained 30 items, all assessing intention to help in the context of potential sexual violence. Items assessed likelihood of engaging in a range of behaviors in various contexts, including proactive/protective behaviors (e.g., “Watching my friends’ drinks at parties”), direct intervention (e.g., “Stopping and checking in with a friend who looks very intoxicated when they are being taken upstairs at a party”), speaking out against sexual violence in general (e.g., “Talking with my friends about sexual violence as an issue for our community”), and confronting (potential) perpetrators (e.g., “Expressing disagreement with a friend who says having sex with someone who is passed out or very intoxicated is okay”). Participants responded to each item on a 5-point scale, ranging from 1 (not at all likely) to 5 (extremely likely). Responses to all items were averaged to create an overall intentions score, wherein higher scores indicated higher intervention intentions. Internal reliability for the scale was high (Cronbach’s α = .97).
Perceived peer norms
Our measure of bystanders’ perceptions of peer norms contained the same items as the bystander intentions scale, but it was worded to assess perceived peer norms regarding bystander intervention against sexual violence. Participants were instructed to consider the extent to which their friends approved or disapproved of engaging in each of the behaviors (e.g., “Watching their friends’ drinks at parties”). Participants responded to each item on a 5-point scale, ranging from 1 (strongly disapprove) to 5 (strongly approve). Responses to all items were averaged to create an overall peer norms score, wherein higher scores indicated stronger perceived peer approval of bystander intervention. Internal reliability for the scale was high (Cronbach’s α = .97).
Reported behaviors
Our measure of bystander behaviors contained the same items as the bystander intentions and peer norms scales but was worded to assess self-reports of actual behaviors performed during the previous semester (Banyard, Moynihan, Cares, & Warner, 2013). Data were collected near the end of the semester, allowing for a reference period of about 3.5 months. For each item (e.g., “Watched my friends’ drinks at parties”), participants responded 1 (yes) or 0 (no), based on whether or not they had performed the behavior during the previous semester. The number of self-reported behaviors was summed across items to create an overall score of bystander behavior; higher numbers indicated more experience with bystander intervention during the semester. Internal reliability for the behaviors scale was high (Kuder-Richardson formula [K-R 20] = .95).
Reported missed opportunities
For each item on the measure of bystander behaviors on which a participant answered “no,” a follow-up question asked whether there was a time during the previous semester when they had the opportunity to engage in the behavior but did not for whatever reason, coded 1 (yes) or 0 (no). Any item for which a participant indicated “yes” to this follow-up question was considered a missed opportunity for intervention. The number of missed opportunities noted across items was summed to create a separate variable for missed opportunities; higher numbers indicated more occasions when a respondent had an opportunity to intervene but did not. Internal reliability for the missed opportunities scale was acceptable (K-R 20 = .71).
Results
Preliminary Analyses
Table 1 includes descriptive statistics and intercorrelations for all study variables. The distribution of scores on the measures of peer norms and bystander outcomes indicates that, overall, students in our sample reported a high degree of perceived peer support for bystander intervention, a high level of their own intention to intervene, and relatively low levels of missed opportunities. Students reported an average of 12.74 bystander actions (range = 0–30) taken in the course of a semester. Perceived peer norms and bystander intentions were highly correlated; the variable of peer norms was also related to reported missed opportunities, but not to reported bystander behaviors. Number of reported bystander behaviors was associated positively with bystander intentions and negatively with number of reported missed opportunities. Gender was related to norms, intentions, and missed opportunities such that men reported lower peer support for intervention and lower bystander intentions, but more missed opportunities, than did women. Race was linked to behaviors and missed opportunities such that Black students reported more behaviors and fewer missed opportunities than did White students. Year in school was not significantly correlated with any study variables.
Descriptive Statistics and Correlations Among Study Variables.
Note. Norms = perceived peer norms, range = 0–5; intentions = bystander intentions, range = 0–5; behaviors = reported bystander behaviors, range = 0–30; missed opportunities, range = 0–12; gender: Female = 0, male = 1; race: 0 = White, 1 = Black; year: 0 = first year student, 1 = second year or above. Correlations for dichotomous demographic variables are point-biserial.
*p < .05. **p < .01. ***p < .001.
Hypothesis Testing
Analysis plan
Regression analyses were used to test all hypotheses. One model tested bystander intentions as the dependent variable (see Table 2), another model tested bystander behaviors as the dependent variable (see Table 3), and the third model tested missed opportunities as the dependent variable (see Table 4). For the model predicting bystander intentions, perceived peer norms, gender, race, and year in college were entered on the first step to examine main effects; two-way interactions were entered on the second step; and three-way interactions were entered on the third step. For models predicting behaviors and missed opportunities, bystander intentions was added as a control variable and entered prior to other predictors, but interactions involving intentions were not examined. Intentions was included as a predictor of behaviors and missed opportunities because previous studies have shown intentions to be an important correlate of behaviors, but that the two variables often have different predictors (Banyard & Moynihan, 2011). In our study, we wanted to remove the variance associated with intentions and examine the unique variance accounted for by other predictors of behaviors and missed opportunities. We also ran these analyses without including bystander intentions as a covariate in these models, and the pattern of results was similar to what is reported in the following.
Model Statistics for Regression Analysis Predicting Bystander Intentions.
Note. SE = standard error. Model based on a sample of 230 participants that includes everyone with complete data on age, gender, race, norms, and intentions. Only variables added at each step are included in the table.
Model Statistics for Regression Analysis Predicting Bystander Behaviors.
Note. SE = standard error. Model based on a sample of 227 participants that includes everyone with complete data on age, gender, race, norms, intentions, and behaviors. Only variables added at each step are included in the table.
Model Statistics for Regression Analysis Predicting Bystander Missed Opportunities.
Note. SE = standard error. Model based on a sample of 227 participants that includes everyone with complete data on age, gender, race, norms, intentions, and behaviors. Only variables added at each step are included in the table.
In all models, continuous variables were mean centered, and dichotomous variables were coded as follows: gender (female = 0, male = 1), race (White = 0, Black = 1), and year (first year = 0, second year and above = 1). All significant interaction effects were further analyzed using simple slope and cell mean calculations using a web calculator developed by Preacher, Curran, and Bauer (2006; http://www.quantpsy.org/interact/).
Peer norms and bystander intervention
We predicted that college students who believed their peers supported bystander intervention would report higher bystander intentions, more bystander behaviors, and fewer bystander missed opportunities (Hypothesis 1). This hypothesis was supported for bystander intentions (see Table 2) but was not supported for bystander behaviors (see Table 3) or missed opportunities (see Table 4). The more students believed their peers supported bystander intervention, the more willingness they reported to intervene against sexual violence.
Bystander characteristics and intervention
Hypothesis 2 predicted that women would report higher bystander intentions to intervene, more bystander behaviors, and fewer bystander missed opportunities than would men. Gender emerged as a significant predictor of bystander intentions (see Table 2) and missed opportunities (see Table 4) but not of reported bystander behaviors (see Table 3). Men reported weaker bystander intentions and reported more missed opportunities than did women. We did not form a specific hypothesis for race but rather explored the main effects of this bystander characteristic. Race was unrelated to bystander intentions (see Table 2) but did show significant main effects in the models predicting bystander behaviors (see Table 3) and missed opportunities (see Table 4). Black participants reported more bystander behaviors and fewer missed opportunities than did White participants. Hypothesis 3 suggested that younger students would report higher bystander intentions to intervene, more self-reported bystander behaviors, and fewer bystander missed opportunities than would more advanced students. Hypothesis 3 was not supported: Year in college did not significantly predict intentions, behaviors, or missed opportunities.
Moderation by bystanders’ characteristics
To test Hypothesis 4, that gender, race, and year in college would moderate the relationship between peer norms and bystander outcomes, interaction terms were added to each of the regression analyses described above. Hypothesis 4a predicted that the main effect of peer norms on intentions, behaviors, and missed opportunities would be stronger for men than for women. The interaction term for gender and peer norms was not significant in the model predicting bystander intentions (see Table 2) but was significant in the models predicting bystander behaviors (see Table 3) and missed opportunities (Table 4). However, Hypothesis 4a was not exactly supported for bystander behaviors: An analysis of simple slopes showed that although the relationship between perceived peer norms and bystander behaviors was in a different direction for women (b = −1.918, standard error [SE] = 1.204, t = −1.593, p = .113) than it was for men (b = 1.623, SE = 1.605, t = 1.011, p = .313), it was not statistically significant for either gender. In contrast, Hypothesis 4a was supported for missed opportunities: An analysis of simple slopes showed that there was no relationship between peer norms and missed opportunities for women (b = −.102, SE = 0.281, t = −.363, p = .717), but there was a significant negative relationship between peer norms and missed opportunities for men (b = −1.295, SE = 0.376, t = −3.449, p < .001). In other words, the more men (but not women) believed their peers supported bystander intervention, the fewer missed opportunities they reported.
Hypothesis 4b predicted that the relationships between peer norms and bystander intentions, behaviors, and opportunities would be stronger among White than among Black participants. Hypothesis 4b was not supported. Peer norms and race did not interact when predicting bystander intentions (see Table 2) or missed opportunities (see Table 4). Although the interaction term of Norms with Race was a significant predictor of bystander behaviors, an analysis of simple slopes showed a nonsignificant negative relationship between norms and behaviors for White participants (b = −1.918, SE = 1.204, t = −1.593, p = .113) and a nonsignificant positive relationship for Black participants (b = 1.185, SE = 1.305, t = .908, p = .365).
Hypothesis 4c predicted that the relationship between peer norms and bystander intentions, behaviors, and opportunities would be stronger among more advanced students than among first-year students. Hypothesis 4c was supported for bystander intentions (see Table 2). An analysis of simple slopes revealed that although bystander intentions increased as peer norms increased for both first-year (b = .491, SE = .076, t = 6.339, p < .001) and more advanced students (b = .787, SE = .141, t = 5.565, p < .001), the strength of the relationship between perceived peer norms and bystander intentions was stronger for more advanced than for first-year students. The interaction effects for norms with year in college did not reach conventional levels of significance for reported behaviors (see Table 3) or missed opportunities (see Table 4).
Intersectionality effects
Two-way interaction effects among bystander characteristics were also explored. Across regression models, only one such interaction effect reached statistical significance: The interaction between gender and race was a significant predictor of missed opportunities (see Table 4). However, this two-way interaction was qualified by a significant three-way interaction involving norms, gender, and race. An analysis of simple slopes revealed that there was no significant relationship between norms and missed opportunities for White women (b = −.246, SE = .297, t = −.829, p = .408), for White men (b = −.896, SE = .472, t = −1.897, p = .059), or for Black women (b = .31, SE = .332, t = .935, p = .351), but there was a strong negative relationship between norms and missed opportunities for Black men (b = −2.033, SE = .514, t = −3.957, p < .001), such that the more strongly Black men believed their peers supported bystander intervention, the fewer missed bystander opportunities they reported.
No three-way interaction terms emerged as significant predictors of bystander intentions (see Table 2). In the model predicting bystander behavior, the interaction among norms, race, and year emerged as a significant predictor. This effect qualified the two-way interaction observed between norms and race. An analysis of simple slopes revealed that the relationship between norms and behaviors was not significant for White first-year students (b = −1.815, SE = 1.269, t = −1.43, p = .154), for Black first-year students (b = .349, SE = 1.418, t = .246, p = .806), or for more advanced White students (b = −3.816, SE = 3.443, t = −1.108, p = .269). However, there was a significant relationship between norms and behaviors for more advanced Black students (b = 10.241, SE = 4.866, t = 2.105, p = .036) such that the more these students believed their peers supported bystander intervention, the more intervention behaviors they reported.
Discussion
The results of the current study are both consistent with previous research and also extend our understanding of bystander attitudes and behaviors among college students. The more students believed their peers supported bystander intervention, the more willingness they reported to intervene against sexual violence. Although the relationship between perceived peer norms and willingness to intervene against sexual violence has been demonstrated in prior studies with men (Fabiano et al., 2003; Stein, 2007), the current study showed that this relationship also exists for women.
Moderating variables were also shown to be important. Although there was no overall relationship between perceived peer norms and reported bystander behaviors or missed opportunities, the predicted relationships were observed for some groups, highlighting the need for bystander research to do more to explore moderating variables. Specifically, Black students in their second year or higher reported very few bystander behaviors when they believed their peers did not support bystander intervention, but they reported many bystander behaviors when they believed their peers were highly supportive of bystander intervention. Black men who believed their peers did not support bystander intervention reported many bystander missed opportunities, but Black men who believed their peers were highly supportive of bystander intervention reported very few missed opportunities.
Our data show some consistency with prior research by Hong et al. (2012) in that we found that norms were more predictive of intentions in more advanced than in first-year students and were more predictive of behaviors in more advanced Black students than in first-year Black students. However, our data are inconsistent with their data in that they found that norms were more influential for Caucasian than for African American adolescents. Of course, the two studies examined very different behaviors (smoking vs. helping) and different developmental contexts (high school vs. college), so the findings may not be comparable.
Consistent with previous research, we did find some patterns of gender differences. Women reported greater intentions to help and fewer missed opportunities than men overall. This pattern matches previous research (Banyard, 2008; McMahon, 2010) and theoretical work suggesting that helping attitudes and behaviors tend to be consistent with gender norms (Dovidio et al., 2006; Eagly & Crowley, 1986). However, contrary to previous work (Banyard & Moynihan, 2011) and to our predictions, men and women did not differ from one another in their self-reported bystander behaviors. Even more intriguing, when behaviors were analyzed at the item level, men and women only differed significantly from each other on 3 of the 30 items: Women had more experience than men with making sure that they (first item) and their friends (a second item) left parties together and had more experience challenging catcalls (the third item). As such, the similar reports of bystander behaviors between men and women cannot be attributed to men and women endorsing different subsets of behaviors—for the most part, men and women were similarly likely to engage in the same behaviors. One possibility that we were not able to analyze was the frequency of various behaviors. On our measure, participants simply indicated whether or not they had engaged in each behavior over the course of the semester. It is possible that frequencies of engaging in certain behaviors differ between men and women but that overall experience does not. More work will be needed to better understand the intervention behaviors of male and female college students.
Although we had predicted relationships between year in college and bystander intentions, behaviors, and missed opportunities, this prediction was not supported. This prediction was largely based on studies showing a relationship between age and helping. Although a negative relationship between age and helping has been demonstrated across various developmental periods (Banyard & Moynihan, 2011; Carlo et al., 2007; Zhong, 2010), there may be multiple conflicting influences on college students that could account for no overall relationship between year in college and helping. Although students generally age as they progress through college, they may also experience greater feelings of connection to their college communities. If helping decreases with age but increases with feelings of connection to one’s community (Ahmed, 2008; Levine, Cassidy, Brazier, & Reicher, 2002), a null relationship between college year and helping would be understandable.
Even though bystander intentions were strongly and positively related to self-reported bystander behavior, by and large, the pattern of correlates of the two dependent variables was very different. Additionally, although we had predicted similar patterns of results for intentions and behaviors, finding different correlates is consistent with prior research (Banyard & Moynihan, 2011). Perhaps the most dramatic difference in predictors of intentions and behaviors was perceived peer norms, which showed a positive relationship with intentions but inconsistent and nonsignificant relationships with behaviors. One explanation for this difference concerns the nature of the variables. Both perceived peer norms and bystander intentions are attitude variables. It makes sense that students’ own attitudes are influenced by the perceived attitudes of their friends, and thus students who believe their friends are supportive of intervening also report that they themselves are likely to take action. But as we know (Glasman & Albarracin, 2006), attitudes do not always predict behaviors. Actual behaviors are influenced by many variables, including opportunity and other situational variables (Darley & Batson, 1973). Perceptions of what you think your friends would do may make you likely to say you would do the same thing but might not lead you to actually do the same thing in the moment. What is more, perceptions of peer norms may also be markers for other aspects of peer social contexts, and these third variables may impact behavior. It may be the case that peer networks that approve of bystander intervention also practice safer and more prosocial behaviors. Peer networks that disapprove of bystander intervention might engage in riskier behavior. Thus, it is possible that perceived peer norms simultaneously influenced willingness to intervene and opportunity to intervene, and these factors influenced intervention behavior in opposite directions.
On their own, neither the behavior data nor the missed opportunity data fully speak to the issue of opportunity. Reporting few intervention behaviors may reflect an unwillingness to intervene or it may reflect a lack of opportunity. Reporting few missed opportunities could indicate a lack of opportunity or it could indicate that intervention opportunities are largely acted upon. Crossing all combinations of high and low reported behaviors and high and low missed opportunities clarifies the matter: Low levels of both behaviors and missed opportunities suggest an overall lack of opportunity to intervene; high levels of both behaviors and missed opportunities suggest many opportunities to intervene; high levels of behaviors and low levels of missed opportunities suggest a propensity to intervene when given the opportunity; and low levels of behaviors and high levels of missed opportunities suggest an unwillingness to intervene.
Main effects and/or interactions involving gender and race for both behaviors and missed opportunities suggest that opportunity may be important for people in certain demographic groups. It may be informative to compare reports of bystander behaviors and missed opportunities for Black and White men and women. Recall (see Table 1) that the overall mean number of behaviors reported was 12.74, and the mean number of missed opportunities reported was 1.59. Black men reported the highest number of bystander behaviors (15.48) and a relatively high number of missed opportunities (2.07). This suggests that Black men face a number of opportunities to intervene and, although they often act on those opportunities, they sometimes do not. Black women reported a high number of behaviors (15.28) but reported a relatively very low number of missed opportunities (.61). This suggests that Black women also face a high number of opportunities for intervention and usually act on those opportunities. White men reported the lowest number of bystander behaviors (10.81) and the highest number of missed opportunities (2.19), whereas White women reported moderate numbers of both behaviors (11.60) and missed opportunities (1.66). Overall, these patterns conclude that, of these four racial/gender groups, White men are the most reluctant to intervene when given the opportunity.
Ours is one of the first known studies of bystander intervention against sexual violence to contain a large enough sample of Black participants to make comparisons between Blacks and Whites. Our data suggest that race and gender intersect in interesting but as yet not fully understood ways. It may be useful to conduct qualitative studies in which Black and White men and women are interviewed about their experiences and thoughts about intervening in the context of sexual violence. It is likely that Black and White students have different ideas about, and levels of comfort with, intervening to help those of the same race versus other races and that these thoughts may differ between men and women. Some previous research has shown that White students were more reluctant to help Black students than White students in a high emergency situation, but that Black students were equally willing to help Blacks and Whites (Kunstman & Plant, 2008). It is possible that this dynamic could partially explain the different rates of helping behaviors between Black and White participants. In racially integrated environments, if Black students intervene regardless of the race of the person needing help but White students primarily help only other White students, we would expect to see more overall intervention behaviors reported by Black participants and more missed opportunities reported by Whites.
We have discussed opportunity as one situational variable that may play a role in bystander intervention, but there are certainly other important factors at play that contribute to students’ actual intervention behaviors that we did not measure. Our set of predictors was able to explain more than 50% of the variance in bystander intentions, but even after including intentions as an additional predictor, we could explain less than 20% of the variance in bystander behaviors. Our research highlights the importance of not equating bystander intentions with bystander behaviors and points out the need to better investigate the variables that most fully explain bystander behavior in the context of sexual violence.
Limitations
Our research should be evaluated in light of a few limitations. Our sample comprised only students, most of whom were in their first or second years of college and all of whom were taking introductory psychology courses at one university. Although the sample contained a fairly large proportion of African American students, it is not representative of all college students. Because we were only able to define race as Black and White, it is still unknown how members of other groups such as Hispanic/Latino, Asian, or Native American would respond to similar measures. Furthermore, we examined only gender, race, and year in college as demographic predictors of bystander intentions and behaviors; it is likely that other factors may also play a role, including sexual identity, religious values, and social characteristics such as living arrangements, fraternity/sorority membership, and sports affiliation. Future research should expand the factors examined as predictors of bystander intentions and behaviors as well as explore predictors of bystander attitudes and behaviors in noncollege populations.
Another important limitation is that all data were collected concurrently. Although we conceptualized peer norms as a predictor of intentions and behaviors, it is not appropriate to assume a causal, or even a temporal, relationship between these variables. This is especially problematic when analyzing bystander intentions as a predictor of bystander behaviors. It is probably more appropriate to imagine prior behaviors as predicting future intentions; however, such a temporal ordering is less consistent with most theories of bystander behavior. As such, we have avoided making a strong case for intentions as a predictor of either behaviors or missed opportunities, and instead we conceptualized intentions as a covariate in our models. There is a strong need for longitudinal studies that examine predictors of bystander intentions and behaviors over time.
The measures used in our study continue to be revised and refined. Previous versions of these measures have been evaluated and used in published research (e.g., Banyard, 2008; Banyard & Moynihan, 2011; Banyard et al., 2013), but the current versions added, removed, or reworded several items and have not been stringently psychometrically evaluated in their current form. One particular potential limitation of the peer norms measure is that the gender of one’s peers was not specified or assessed. Respondents were asked about their “friends,” and we assumed that people would most likely report on same-sex friends, but it remains a likelihood that at least some people were thinking of other-sex peers or a variety of peers of both genders. It is also possible that respondents were simply unaware of their peers’ attitudes regarding bystander intervention. Another limitation related to the measures is that participants responded to the same items three different times in the intentions, norms, and behaviors measures. Although the intentions and norms scales were presented in random order, the behaviors measure was always presented last. The repetitiveness of item presentation could have created response fatigue in participants and contributed to the high rates of attrition we observed. In addition, the order of items across measures remained consistent, which could have primed responses in some unknown way. A final limitation is that behaviors were self-reported and could not be independently verified. We did not control for social desirability and cannot be certain that participants were completely accurate in their reporting. Further refinement of measures of bystander attitudes and behaviors is called for, and intense psychometric evaluation of these measures is necessary (also see Banyard et al., 2013).
Practice Implications
Our findings have important implications for practice. First, sexual assault prevention programs that aim to increase intentions to intervene as a prosocial bystander might be well advised to include a social norms component (see Gidycz, Orchowski, & Berkowitz, 2011, for an example); however, our study offers the caveat that focusing on norms alone may not have much impact on actual behaviors. Multipronged prevention strategies that do not rely on any one aspect of change are likely to be more successful (Banyard, 2013). Social norms approaches should be combined with skill-building workshops, individual-level knowledge and awareness building, and policy changes that incentivize and support bystander actions while providing clear consequences for sexual violence (Taylor, Stein, Mumford, & Woods, 2013).
The current findings also remind us that one size does not fit all in prevention messages (Moynihan et al., in press). Patterns of gender similarity in the current study support prevention work that uses similar curricula with men and women (although having single-gender groups for curricula administration is still important; Banyard et al., 2007). The interesting interactions obtained in the current study, however, also highlight that educational programs need to be sensitive to different barriers and facilitators to helping for different groups of students, including students of color. Furthermore, variables like race may vary in importance by specific campus context. On campuses where White students are overrepresented, individual bystanders who are not White may not feel safe to intervene without a great deal of peer support for their actions. In other contexts, it may be safer to be a lone bystander. Our observed effects of year in college suggest that prevention programs directed at first-year students may need to be different from those aimed at more advanced students. Prevention educators need to be sensitive to these context effects when translating prevention tools across student groups or from one campus to another.
Conclusion
Our study is unique in its consideration of missed opportunities for bystander intervention in addition to bystander behaviors. Previous work on bystander intervention in relation to sexual violence has been unable to determine whether reported bystander behaviors reflect willingness to intervene or intervention opportunities. Banyard and Moynihan (2011) observed an unexpected positive relationship between peer norms supporting sexual coercion and bystander behavior; we contend that this relationship could be explained by assuming that people with coercive peers have more opportunities to intervene. The current study cannot definitely conclude how much of people’s behavior is due to attitudes versus opportunity, but it does open the door for future research aimed at disentangling these variables.
In addition to measuring missed opportunities for intervention, ours is one of the few studies known to assess actual self-reported behaviors in addition to intentions to intervene. It is also one of the few to assess perceived peer norms as a predictor of intervention intentions and behavior amongst both men and women. Previous studies (Brown & Messman-Moore, 2010; Fabiano et al., 2003) have tended to focus on male-only samples. Finally, ours is one of only a few studies known to include race as a predictor of bystander intentions or behaviors, and the only known study in the domain of bystander intervention in the context of sexual violence to do so—despite arguments that race impacts how people think about sexual violence (Katz, 2006). Our data indicate that race and gender are important variables to consider. The current study reminds us that understanding bystander attitudes and behaviors in the context of sexual violence requires complex models that permit assessment of mediating and moderating effects. Future research must outline and test these pathways.
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) received no financial support for the research, authorship, and/or publication of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
