Bystander-Based Sexual Violence Prevention With College Athletes: A Pilot Randomized Trial

Abstract

The prevention of sexual violence on college campuses is a pressing public health issue. Given recent U.S. federal and state requirements for campus responses to sexual assault, many campuses may plan to implement brief, bystander-based programs to create a violence-free environment. This pilot study evaluates one such program for male undergraduate athletes, Wingman 101. The primary purpose of this study was to determine the feasibility and acceptability of participation in Wingman 101, as well as barriers to program implementation. Data for this project were collected from 80 undergraduate male athletes (M age = 19.99) on three contact sport teams in spring 2012. Participants were randomly assigned to program or a no-program control condition. Implementation data were collected at the end of each session from program participants and facilitators. Outcome data were collected over three waves (pretest, posttest, 2-month follow-up) and assessed bystander attitudes, beliefs, and behaviors. Analysis of implementation data indicated that the program was well received and implemented with high fidelity, with facilitator relatability emerging as a particularly important aspect of implementation. However, participants also listed numerous barriers to potential bystander intervention following the program. The presence of these barriers supports quantitative reductions in positive attitudes about intervention at posttest. Implications of findings for policy and practice on postsecondary campuses are discussed.

Keywords

sexual violence bystander prevention athletes

Sexual violence on college campuses is a documented public health problem, with national data indicating that a substantial minority of women experience some form of unwanted sexual contact during their time at college (Cantor et al., 2015; Fisher, Cullen, & Turner, 2000). Consequences of this victimization are broad, and include physical injury, physical health complaints, poor mental health, and substance use (Basile & Smith, 2011), as well as increased risk of future revictimization (Classen, Palesh, & Aggarwal, 2005). Women who experience victimization while at college are also at risk of poor academic performance (Jordan, Combs, & Smith, 2014), and the economic impacts of sexual violence are high for victims and their institutions (Gray, Hassija, & Steinmetz, 2017). Thus, the prevention of sexual violence on college campuses is a pressing task. In terms of how to approach prevention, recent guidance from the Centers for Disease Control suggests that bystander-based approaches are a key prevention strategy for campus-based sexual violence, in order to change social norms that condone this violence (Dills, Fowler, & Payne, 2016).

As opposed to more traditional prevention approaches, the use of bystander-based programming highlights the role of the whole community in preventing the attitudes, beliefs, and behaviors that support sexual violence (Banyard, Plante, & Moynihan, 2005). By focusing on the role of community, and not on victims and perpetrators exclusively, proactive bystander-based programs may decrease defensiveness to prevention messages (Banyard et al., 2005). Such programming may thus be especially important among groups such as male student athletes. Although some studies indicate that college athletes are at higher risk to perpetrate sexually aggressive behavior compared with their nonathlete peers (McCray, 2015; Murnen & Kohlman, 2007), the majority of athletes are not perpetrators and may exhibit defensiveness to individually targeted messages. Furthermore, although there are several reasons why increased risk among this subgroup may occur (including violence-supportive gender norms and the normative use of violence in contact sports; McCray, 2015), McMahon and Farmer (2009) highlight that it is important to also consider the strengths of athletic communities, and how these strengths can support prevention in this context. One strength present within this setting is team bonding, which may contribute to more willingness to intervene to help fellow members of the team (McMahon & Farmer, 2009). Bystander-based prevention programs are approaches that specifically draw on this strength.

Because of the particular norms and assets that may exist within campus teams, determining how best to implement bystander-based prevention within this context (as opposed to bystander-based prevention in the general student population) is important to achieving whole-campus climates that are violence free (McMahon, 2015). However, work on bystander-based prevention with student athletes is limited (McMahon, 2015; McMahon & Farmer, 2011). To our knowledge, there exists only one controlled evaluation of a bystander-based approach with college athletes. In this study, Moynihan, Banyard, Arnold, Eckstein, and Stapleton (2010) found that athletes’ participation (n = 139, 43% female, 49% in first year of college) in a 4.5-hr, one-session, single-sex version of the Bringing in the Bystander program was associated with greater bystander efficacy and likelihood of helping over a 2-month follow-up period, compared with nonparticipating teammates. There were no differences in rape myth acceptance or bystander behaviors, although both variables showed trends in the expected direction. This study did not present results separately by sex, and we are not aware of any past research that evaluates a bystander-based approach to sexual violence prevention for male college athletes, or that specifically examines the feasibility and acceptability of bystander-based programming with this group.

We also note that the program evaluated by Moynihan et al. (2010) was brief (one session) in length. Although short programs are not prevention best practices (Nation et al., 2003), short programs are the norm on college campuses (DeGue et al., 2014), primarily due to logistical and other constraints. As such, evaluation of brief programs is meaningful, as it is important to understand the feasibility and acceptability of these short interventions, as well as potential impacts on attitudes, beliefs, and behaviors.

The Current Study

Given the limited attention to bystander-based approaches for male collegiate athletes (Dills et al., 2016), this pilot study evaluated Wingman 101, a bystander-based sexual violence prevention program designed for male undergraduate athletes and fraternity members (Appendix A). Unlike most existing bystander-based programs (Exner-Cortens & Wells, 2017), Wingman 101 was designed for all-male, intact groups (athletes, fraternities). Thus, as compared with the two evaluated one-session, in-person, bystander-focused programs for postsecondary students (Bringing in the Bystander, Banyard et al., 2005; and Green Dot, Coker et al., 2015), it is not appropriate for campus-wide distribution, as a critical component of the program is that it is administered within cohesive peer groups. For example, while the program uses scenarios such as the Bringing in the Bystander program (Banyard et al., 2005), scenarios for Wingman 101 were specifically written to describe contexts and situations pertinent to the target audience (e.g., being coerced to drink at a team party, hearing a teammate say a lewd comment in the locker room, attending a team party and seeing a teammate take an intoxicated woman upstairs). Furthermore, given the short program length (as desired by campus administrators), scenarios are also discussion based, as opposed to role-play based as in Bringing in the Bystander. Unlike Green Dot (Coker et al., 2015), the program focuses on engagement with an intact group, as opposed to a general motivational speech (the comparable portion of the Green Dot program in terms of length).

Wingman 101 was developed in 2007 by the second author, in consultation with campus administrators and key campus stakeholders. Program activities (see Appendix A) were designed to align with Berkowitz’s (2004) recommended best practices for male-focused sexual violence prevention programming (e.g., approaching men as partners and not perpetrators; conducting programming with small, all-male groups; focusing on interactive discussion). The program includes a heavy focus on facilitated scenario-based discussion, to allow participants to identify existing group norms around sexual violence, and then engage in examination, critique, and deconstruction of those norms. The program is offered in a 75-min, one-session format, and is cofacilitated by two to three male student peers. Facilitators interview for this position, with selection based in part on relatability to the target groups (athletes and fraternity members; for example, facilitators often have prior experience on an athletic team or are themselves a fraternity member). Facilitators receive ongoing training on the program manual, and meet weekly as a group during the school semester to receive coaching on facilitation from the second author and discuss emerging issues. The trainings are also always cofacilitated, embedding a peer mentorship model into program facilitation. Additional program information is provided in Appendix A. To date, Wingman 101 has received positive feedback from program participants, but has not been the subject of formal evaluation.

The program theory of change is centered on Banyard et al.’s (2005) work on prosocial bystander intervention, which considers sexual violence a community (and not individual) issue, and where the goal is to create positive community norms about intervention. At the start of Wingman 101, participants are introduced to bystander theory, and facilitators use each scenario to highlight the role of the community (in this case, the athletic team) in preventing sexual violence. Wingman 101 also draws on understandings of the importance of addressing social norms to create community social change. For bystander intervention particularly, social norms “influence the extent to which [the individual] feel[s] that others in their immediate environment share their concerns and will support their efforts” (Berkowitz, 2010, p. 148). In Wingman 101, the scenarios are used to foster social norms-based discussion, for example, by demonstrating that others on the team are also concerned with sexual violence and would support the efforts of a teammate to prevent this violence. Finally, as part of strengths-based practice, the focus of the program is not on individual behaviors associated with perpetration, but rather on emphasizing a community prevention approach that draws on the strengths (e.g., bonding) of these intact groups. As Katz (2011) noted,

One of the key differences in facilitating bystander education sessions with cohesive groups like teams, and with groups composed of people who don’t know each other well, is that few ties bind the latter group . . . With these [cohesive] groups you are always reinforcing the idea that they are responsible to each other—and for each other’s behavior. (pp. 3-4)

Thus, the program structure is intended to encourage those who know each other, trust each other, and socialize together to act together to prevent sexual violence.

Given the paucity of information on the implementation of bystander-based approaches with male athletes in postsecondary settings, the primary purpose of this pilot study was to evaluate the feasibility and acceptability of the one-session Wingman 101 program, as well as barriers to program implementation. As part of feasibility and acceptability, we also examined preliminary bystander-based outcomes of the program using an experimental design to compare male collegiate athletes who participated in the program with those who did not.

Method

Setting and Participants

Participants for this study were drawn from a research-intensive, private Northeastern university. In fall 2011, male varsity team coaches were approached to participate in an evaluation of the Wingman 101 program. The primary recruitment mechanism was through a coaches’ meeting held in December 2011, organized by the athletics department. Coaches of individual teams were also contacted to see whether they would like to participate. Of the 17 men’s varsity teams, three teams elected to participate in the study. Participants from these three Division I, contact sport teams were recruited by going to individual team meetings organized by the coaches. At these meetings, members of the research team explained the study, and handed out consent forms to interested athletes. All members of the three participating teams were invited to participate in the study. Coaches were not in the room during recruitment and consent, and were not informed about their players’ decisions to participate.

From these three teams, the written consent return rate from all eligible players was 47.9% (n = 102). Of these, 78.4% (n = 80) completed the baseline (T0) assessment in February and March 2012. These 80 participants were then randomized to treatment and control groups, using team as the stratification variable. Thirty-two treatment individuals attended a Wingman 101 session. The retention rate at T1 was 92.5% (n = 74) and at T2 was 72.5% (n = 58). This is comparable with the retention rate in Moynihan et al.’s (2010) evaluation of bystander-based education with student athletes. Participant flow through the study is shown in Appendix B, and participant demographics at baseline are shown in Table 1. This study was approved by the university’s institutional review board.

Table 1.

Baseline Characteristics of the Full Sample at T0 (N = 80).

Variable	Total (N = 80)	Treatment Group (n = 40)	Control Group (n = 40)	p
Demographics	M (SD)	M (SD)	M (SD)	p
Age	19.99 (1.32)	19.97 (1.09)	20.00 (1.51)	.92
Social desirability	16.20 (5.52)	15.33 (5.31)	17.03 (5.67)	.19
Demographics	% (n)	% (n)	% (n)	p
Gender: Male	100.0 (80)	100.0 (40)	100.0 (40)	—
Year in school: First/second year	53.8 (43)	55.0 (22)	52.5 (21)	.82
Program of registration: Arts/sciences	32.5 (26)	42.5 (17)	23.1 (9)	.060
Race/ethnicity: White, non-Hispanic	80.0 (64)	75.0 (30)	85.0 (34)	.77
Living arrangements: Campus residence hall	28.8 (23)	17.9 (7)	40.0 (16)	.005
Fraternity member: Yes	56.3 (45)	60.0 (24)	53.8 (21)	.58
Prior participation in sexual assault prevention program: Yes	15.0 (12)	12.5 (5)	17.5 (7)	.53
Know a victim of sexual assault: Yes	38.8 (31)	46.2 (18)	32.5 (13)	.21
Outcomes	M (SD)	M (SD)	M (SD)	p
Pros of intervention	17.33 (3.42)	16.98 (2.99)	17.68 (3.81)	.36
Cons of intervention	15.24 (4.03)	14.52 (4.02)	15.97 (3.97)	.11
Willingness to help	42.58 (4.65)	41.73 (4.86)	43.44 (4.31)	.10
Bystander efficacy	72.78 (14.73)	72.79 (13.53)	72.77 (16.05)	1.00
Rape myth acceptance	35.75 (9.32)	36.03 (7.71)	35.47 (8.97)	.77

Procedures

Treatment group participants were offered the group-based, one-session Wingman 101 program in February and March 2012. The primary goals of this program are to teach men to recognize potentially risky situations that can lead to sexual assault, identify obstacles to intervention, and build intervention skills through group discussion. As described above, the program is offered to preexisting groups. Of the 40 individuals from three teams assigned to the treatment condition in this study, 32 (80%) attended a Wingman 101 program session with their teammates. Substantial efforts were made to get the remaining eight players to attend a session, but we were unsuccessful in our attempts to have these players attend. Participants attended their session with other members of their team. The sessions were facilitated by two or three experienced male peer facilitators. The control group did not receive any programming. All research participants were entered into a draw for gift cards for completing assessments, but were not paid to attend treatment sessions.

Measures: Implementation

Facilitators

At the conclusion of each session, facilitators independently completed a fidelity checklist indicating which program activities were accomplished during the session. Facilitators also completed a paper-and-pencil survey gathering open-ended feedback on any adaptations made, participant engagement, and general impressions.

Participants

At the conclusion of each session, treatment participants filled out a brief, paper-and-pencil survey to indicate their satisfaction level with the program (1 = completely dissatisfied, 6 = completely satisfied), how well they felt they could relate to program facilitators (1 = definitely no, 4 = definitely yes), and how realistic they found the scenarios (1 = completely unrealistic, 4 = completely realistic). Participants also provided open-ended feedback on their satisfaction level, facilitator relatability, and scenario realism. Participants were also asked whether they had enough time during the session to think about the topic (0 = strongly disagree, 6 = strongly agree), and for their open-ended feedback on barriers they felt would keep them from using bystander techniques in everyday life, barriers encountered during the session discussion, any topic during the session that made them feel uncomfortable/defensive, and general feedback on the session. Finally, participants were asked to respond to the prompt “As a result of this program, I will . . . ” (Banyard et al., 2005).

For analyses, we collapsed quantitative implementation data on overall satisfaction, 1 = high (mostly or completely satisfied), 0 = low (completely dissatisfied, mostly dissatisfied, somewhat dissatisfied, or somewhat satisfied); facilitator relatability, 1 = high (definitely yes), 0 = low (definitely no, somewhat no, or somewhat yes); scenario realism, 1 = high (completely realistic), 0 = low (completely unrealistic, somewhat unrealistic, or somewhat realistic); and time to think about the topic, 1 = disagree/somewhat agree, 2 = mostly agree, 3 = strongly agree, due to small sample size in individual cells. Cut points for collapsing were chosen based on the distribution of responses within the sample.

Measures: Outcomes

Outcome measures for this study were taken from prior evaluations of a bystander-based sexual violence prevention program (Banyard et al., 2005; Moynihan et al., 2010). Outcome data were collected at three waves: pretest (T0), 1-week posttest (T1), and 2-month follow-up (T2). All outcome data were collected online using Qualtrics. Due to the small sample size, we note that outcome variables are included in this article as indicators of program feasibility and acceptability, and not of program efficacy.

Pros and cons of intervention

Beliefs about the pros and cons of intervention were assessed using the 11-item Decisional Balance Scale (Banyard et al., 2005). On this scale, pros of intervention are measured using five items (e.g., “If I intervene regularly, I can prevent someone from being hurt”), and cons of intervention are measured using six items (e.g., “Intervening might cost me friendships”). Responses to both subscales were given on a 5-point Likert-type scale, in terms of how important the statement would be if the participant was considering intervening in a situation where he thought someone might be hurt or was at risk of being hurt (1 = not important at all, 5 = extremely important). Scores for each subscale were created by summing individual items (α, Pros, range, T0-T2 = .62-72; α, Cons, range, T0-T2 = .76-.83). Higher scores on the Pros subscale indicate more belief in the pros of intervention; higher scores on the Cons subscale indicate greater belief in the cons of intervention. This scale was included at all waves. Missing data on the Pros scale ranged from 1.2% to 4.1% across waves, and on the Cons scale from 3.5% to 5.4%.

Willingness to help

Attitudes toward willingness to help friends, strangers, and acquaintances were assessed using an adapted version of the 12-item Bystander Intention to Help Scale–Short Form (Moynihan et al., 2010). An example item from this scale is “Express disagreement with a friend who says forcing someone to have sex with them is okay.” Responses were assessed on a 5-point Likert-type scale, in terms of how likely the participant was to engage in each behavior (1 = extremely unlikely, 5 = extremely likely), and scale scores were created by summing across the 12 items (α, range, T0-T2 = .62-.79). Higher scores indicate greater willingness to help. This scale was included at all waves. Missing data ranged from 3.8% to 15.5% across waves.

Bystander efficacy

Beliefs about ability to intervene in a potentially risky situation were assessed using the 18-item Bystander Efficacy Scale (Banyard et al., 2005). After each item, participants were asked to indicate their confidence related to doing the behavior, on a scale from 0% to 100%. An example item from this scale is “Ask a stranger if they need to be walked home from a party.” A mean score was created across the 18 items (α, range, T0-T2 = .91-.95), with higher scores indicating greater confidence. This scale was included at all waves. Missing data ranged from 0% to 2.7% across waves.

Rape myth acceptance

Rape myth acceptance was assessed using the 17-item Illinois Rape Myth Acceptance Scale–Short Form (Payne, Lonsway, & Fitzgerald, 1999). An example item from this scale is “Rape accusations are often used as a way of getting back at men.” Responses were assessed on a 5-point Likert-type scale, assessing agreement with each statement (1 = very much disagree, 5 = very much agree). Scores across all items were summed to create the total score (α, range, T0-T2 = .88-.91). Higher scores indicate greater rape myth acceptance. This scale was included at all waves. Missing data ranged from 3.8% to 8.1% across waves.

Bystander behaviors (T2 only)

Bystander behaviors toward friends, acquaintances, and strangers were assessed using 15 items created for this study, adapted from Banyard et al. (2005). On this scale, each item was repeated three times, changing whether the behavior was directed toward a friend, an acquaintance, or a stranger (e.g., “Indicate my displeasure when I hear catcalls made by a friend/acquaintance/stranger”). Thus, each subscale had five items. Acquaintances were defined as people you know a little but not enough to consider them friends. For example, you have been in class with them or are members of the same organization and strangers were defined as people you may recognize by sight from campus but may not have met or had any formal contact with before. At T2, participants were asked whether or not they engaged in these behaviors in the past 2 months (1 = yes, 0 = no). Due to small cell sizes, sum scores on each subscale were collapsed into no behaviors, one behavior, or two or more behaviors for analyses. Cut points for collapsing were chosen based on the distribution of responses within the sample. Missing data ranged from 5.3% to 8.8% across sub-scales.

Measures: Demographics

Included at T0 were age, year in school (1 = first/second year, 0 = third/fourth year), program of registration (1 = arts/sciences, 2 = agriculture/business/life sciences, 3 = other), race/ethnicity (seven categories), living arrangements (five categories), fraternity membership (1 = yes, 0 = no), prior participation in sexual assault prevention programming (1 = yes, 0 = no), whether the participant knew a victim of sexual assault (1 = yes, 0 = no), and social desirability (Crowne & Marlowe, 1960). Missing data on demographic variables ranged from 0% to 7.5%.

Analysis

Descriptive and bivariate (t tests, one-way ANOVA) statistics were used to explore quantitative implementation data. Open-ended implementation data were themed by an undergraduate research assistant (RA) and the first author. Given the small sample size, bivariate differences between the control and treatment group on outcome measures (as a marker of feasibility) were calculated using t tests or chi-square tests, as appropriate. We also calculated effect size for each outcome variable using Cohen’s d for continuous variables and Cramer’s V for categorical variables. Effect sizes were interpreted per Cohen (1977), where for d, .20 is small, .50 is medium, and .80 is large; and for V, .10 is small, .30 is medium, and .50 is large. Because missing data on the bystander scales were usually the result of one to two missed items, these data were handled using person mean substitution (Shrive, Stuart, Quan, & Ghali, 2006). On imputed bystander scales, missing data ranged from 0% to 2.7%. Results were evaluated at the p < .05 level. All outcome analyses were conducted using intent-to-treat procedures.

Results

Demographics for the full sample at T0 are shown in Table 2. In the overall sample, participants were primarily White, non-Hispanic (80%), and in their first, second, or third year of university (90%). Approximately 57% of participants belonged to a fraternity. The treatment and control group were fairly equivalent on demographics at T0 (Table 2), but there was a trending difference between treatment and control participants on program of registration (p = .060). Also, although treatment participants were not more likely to be a fraternity member, they were more likely to live in fraternity housing (p = .005). All other measures were equivalent at T0.

Table 2.

Bivariate Analyses and Effect Sizes for Main Effect of Treatment Group on Outcomes.

Variable	M (SD), Treatment	M (SD), Control	Cohen’s d	p
Pros of intervention
T1	15.65 (3.04)	17.10 (2.82)	−0.49	.039
T2	15.54 (3.43)	16.52 (3.10)	−0.30	.27
Cons of intervention
T1	16.32 (2.98)	15.44 (3.92)	0.25	.29
T2	15.63 (3.60)	15.43 (4.66)	0.048	.87
Willingness to help
T1	42.56 (5.41)	43.13 (5.39)	−0.11	.65
T2	44.53 (5.32)	43.04 (5.68)	0.27	.32
Bystander efficacy
T1	67.85 (17.99)	72.83 (15.73)	−0.29	.21
T2	75.60 (13.08)	72.95 (13.04)	0.20	.45
Rape myths
T1	36.03 (9.21)	35.49 (11.09)	0.053	.82
T2	30.54 (7.64)	31.63 (10.68)	−0.12	.67
Variable	% (n), Treatment	% (n), Control	Cramer’s V	p
Behaviors, friends, T2
None	12.5% (3)	27.3% (9)	0.33	.045
One	20.8% (5)	39.4% (13)
Two or more	66.7% (16)	33.3% (11)
Behaviors, acquaintances, T2
None	45.8% (11)	57.6% (19)	0.12	.66
One	16.7% (4)	15.2% (5)
Two or more	37.5% (9)	27.3% (9)
Behaviors, strangers, T2
None	66.7% (16)	63.6% (21)	0.040	.96
One	12.5% (3)	15.2% (5)
Two or more	20.8% (5)	21.2% (7)

Feasibility and Acceptability of Wingman 101

Facilitators

Facilitators in all sessions implemented all the program content. In terms of adaptations, in most sessions, facilitators reported that there were no changes made. If changes were made, it was to the duration of the session or discussion (shorter); however, in these cases, facilitators reported that conversation was still robust, and this was also reflected in participant feedback (see below). In terms of engagement, facilitators reported that the discussion became deeper as the scenarios progressed (from Scenario 1 to Scenario 3; Appendix A), especially with regard to Scenario 3.

Participants

In terms of satisfaction, the majority of the 32 individuals who participated in the program were relatively satisfied (M = 4.33, SD = 1.38). Participants’ primary explanation for this high rating was that they felt they were able to talk openly and honestly about a traditionally uncomfortable topic. As one participant reported, he was highly satisfied because “we talked about uncomfortable things and I felt comfortable” (Team 2). Many participants also commented on the accuracy of the scenarios as a reason for their satisfaction level.

Participants also felt they could relate to the facilitators (M = 3.56, SD = 0.62) and that the scenarios were realistic (M = 3.52, SD = 0.68). Regarding facilitator relatability, participants overwhelmingly said the facilitators seemed like typical guys of a similar age who did similar activities. For example, one participant noted that “they go out and socialize the same way we do” (Team 3), while another stated that the facilitators “were cool guys who knew what goes on at parties and they asked about realistic situations” (Team 1). Finally, most participants reported that they had adequate time to think about the topic during the session (M = 5.03, SD = 0.97).

Barriers

Implementation

Within the session itself, the majority of participants did not perceive any barriers to a quality discussion, and reported that they did not feel defensive about any of the topics discussed at their sessions. However, among the relative few that did mention barriers, more than one response involved the possibility of fellow teammates negatively viewing the participant’s openness. For instance, one participant described a “reluctance to appear more uptight about social situations than your friends” (Team 2) as a barrier encountered during the session. Similarly, a participant from another team stated that they might have been “afraid to look bad in front of friends” during the session (Team 1).

Bystander intervention

In open-ended feedback, many participants described barriers that might prevent the use of bystander techniques in everyday life. Main themes about barriers included the opinion of others (e.g., “what people might think of you if you intervene”; Team 1) and the relationship with the people involved (e.g., “how well I know the people”; Team 3). A few participants also mentioned potential power differentials (e.g., age differences; differences in status on team) as barriers to intervention.

Participants’ open-ended implementation data around barriers to intervention support findings of preliminary outcome analyses. In bivariate analyses, we found a main effect for treatment group on pros of intervention at T1 (Table 2), indicating that individuals who participated in the program reported significantly fewer pros of intervention than individuals who did not participate in the program at posttest. The effect size for this finding was medium in magnitude (Table 2). However, this reduction in pros was no longer present at 2-month follow-up (Table 2). Thus, it is possible that an increased awareness of potential barriers to intervention was associated with a temporary reduction in perceived pros of intervention, suggesting issues with the feasibility of the current delivery format.

Future Actions

Participants were also asked to name one specific action they would take as a result of participating in the program. A majority of participants expressed some variation of the idea that they would make a more concerted effort to “think before they act” in the future. Many participants also discussed being increasingly conscious and aware of the things they learned during the session, and applying that knowledge to future situations where they may have to intervene as a bystander. Finally, about one third of respondents mentioned that they would look for opportunities to act (e.g., “Look out for my friends more and hopefully strangers too”; Team 1) as a result of program participation. However, outcome data suggest this increase in intentions to act may only have extended to friends (Table 2). Specifically, in preliminary outcome analyses, there was a medium-sized effect of program participation on helping behaviors toward friends 2 months following the program, but only a small effect for helping behaviors toward acquaintances, and a negligible effect for helping behaviors toward strangers (Table 2). Furthermore, the effect for friends appeared primarily driven by the behavior “I stopped and checked in with my friend who looked very intoxicated when they were being taken upstairs at a party,” χ²(1, N = 56) = 5.63, p = .018. This finding aligns well with the focus of the program on helping teammates (i.e., friends), and on the amount of discussion of Scenario 3 (which describes a teammate taking an intoxicated woman upstairs), but suggests that the program may need to expand its scenarios to better affect potential behaviors toward acquaintances and strangers.

Exploratory Analysis: Relationship Between Implementation and Outcomes for Treatment Group Participants

To link our implementation and preliminary outcome evaluation data, we conducted exploratory bivariate analyses that looked at how participants’ feelings about the program (i.e., in terms of overall satisfaction, facilitator relatability, scenario realism, and thinking time) related to outcome variables. We found no associations between our outcome variables (rape myths, pros and cons of intervention, willingness to help, bystander efficacy, bystander behaviors) and scenario realism, thinking time, or program satisfaction at either T1 or T2. However, there were several associations with facilitator relatability. We note that due to the small cell sizes in all these analyses, findings in this section should be interpreted as preliminary trends.

Facilitator relatability

Treatment group individuals who reported that they found facilitators highly relatable immediately following the program also reported lower rape myths at 1-week posttest (M = 33.01, SD = 9.25), compared with those who found facilitators only somewhat or not relatable (M = 40.28, SD = 7.90); t(28) = 2.12, p = .043. However, this difference did not remain at 2-month follow-up (T2); t(20) = 1.80, p = .087. There were no statistically significant differences in rape myths at baseline (T0) based on facilitator relatability level (i.e., there was not an association between the rape myth attitudes participants entered the program with and their feelings about facilitator relatability), suggesting that differences at T1 were not the result of preexisting differences in attitudes at baseline.

Treatment individuals who found facilitators highly relatable also reported greater willingness to help at 1-week posttest (M = 44.32, SD = 5.74), as compared with those who found facilitators only somewhat or not relatable (M = 39.91, SD = 3.65); t(29) = −2.30, p = .029. This difference in willingness to help remained at T2 (M_{high relatability} = 46.67, SD = 5.05, vs. M_{low relatability} = 41.70, SD = 3.75); t(20) = −2.31, p = .032. As with rape myths, there was no association between willingness to help at T0 and perceived facilitator relatability level.

When we also included control group individuals in this exploratory analysis (i.e., comparing high-relatability treatment individuals, low-relatability treatment individuals, and control group individuals using one-way ANOVA), we found no differences between control and treatment individuals on rape myths or willingness to help at the .05 level (just as we found no main effects for these outcome variables, Table 2).

Discussion

The purpose of this study was to understand the feasibility and acceptability of a bystander-based program designed for and implemented with male collegiate athletes. We also explored preliminary program outcomes as a marker of feasibility, and the potential relationship between acceptability and outcomes. In this study, we found the program was well regarded by participants and implemented with high fidelity by facilitators, but that there were several challenges to feasibility. Particularly, open-ended implementation data gathered from Wingman 101 participants immediately following the program indicated that the program was successful in getting them to think about bystander intervention, but few participants reported that the program gave them the skills needed to intervene. This lack of skills seemed to be reflected in preliminary outcome analyses, where we found that athletes who participated in Wingman 101 reported fewer pros of intervention compared with nonparticipating athletes at 1-week posttest. Given that the majority of these individuals had not participated in prior sexual assault programming (Table 2), it is plausible that this short-term reduction represents an evolution in understanding of what bystander intervention involves. Specifically, the focus on thinking may have led to reductions in positive beliefs about intervention as participants realized what was involved in taking action (e.g., realized they were less prepared than they believed themselves to be at baseline), without the time needed to reflect on concerns or build new skills (due to the short length of the program). Thus, this suggests that additional program time is needed to increase feasibility.

The numerous barriers to intervention described by treatment participants also support a temporary reduction in positive beliefs and feelings about bystander intervention. Most of the barriers reported were comprised of concerns about the relationship with the people involved (e.g., “how well I know the people”) and fears around the negative evaluation of others (e.g., “what people might think of you if you intervene”). In her review of Latane and Darley’s (1970) five-step situational model of bystander intervention (i.e., noticing an event, identifying that intervention is needed, taking responsibility for intervening, deciding how to help, and acting to intervene), Burn (2009) suggested that these two beliefs relate to the barriers of “failing to take intervention responsibility” and “failure to intervene due to audience inhibition,” respectively (p. 781). In turn, these two barriers impede the steps of taking responsibility to intervene and actually intervening. Burn also noted that both these barriers tend to be more common in males than females. Thus, it is possible that participation in Wingman 101 got participants thinking more deeply about what actions would be required for bystander intervention, and potential barriers to those actions. In turn, this may have resulted in less positive feelings about intervention. This finding speaks to the need to ensure programs are of an adequate length to work through barriers and build skills with program participants. This finding also suggests that two important barriers to work on with male participants may be around responsibility and audience inhibition (including how expectations of heteronormative masculinity relate to fears of negative evaluation from male peers; Corboz, Flood, & Dyson, 2016).

Studying barriers to intervention in a sample of players from the Australian Football League (ages 19-32), Corboz and colleagues (2016) also noted that barriers particular to the athletic setting include team hierarchies (e.g., captain vs. new player), age (i.e., older vs. younger players), potential negative repercussions, and intrateam codes of silence. As most of these barriers also came up in our evaluation, it is possible they too affected participant’s positive feelings postintervention (e.g., after having the discussion with their team, participants may have realized that a number of within-team barriers to intervention existed). Although literature on bystander training with athletic teams is limited, the findings of our study, as well as Corboz et al.’s (2016) study, suggest that these barriers may be particularly important to attend to when designing and implementing bystander intervention programs for male athletes.

Although they perceived fewer pros to intervention immediately following the program, program participants did report more helping behaviors toward friends in the 2 months after the program. This finding appeared driven by increased help toward intoxicated friends. The inclusion of a large amount of program content on alcohol-facilitated sexual assault may, thus, serve to explain this finding. The strengths-based focus discussed in the Introduction also suggests that this effect may be the result of increased feelings of responsibility for fellow team members (Katz, 2011). As noted in the “Results” section, this may also be why we did not find an effect of intervention for helping acquaintances or strangers, and suggests a need to expand content covered in program scenarios.

Implementation data from facilitators indicated that, from their perspective, the program was feasible to implement. Program participants were also satisfied with the program, in large part because it allowed them to discuss uncomfortable topics in a comfortable setting. Participants also found facilitators relatable, scenarios realistic, and thinking time appropriate. Exploratory subgroup analyses demonstrated that facilitator relatability was a potentially important feature of program implementation. Prior work on bystander-based approaches with male audiences has also described the importance of being around “like-minded men” as it relates to positive feelings toward intervention (McMahon & Dick, 2011, p. 3). In part, then, the importance of relatability may have to do with the sense of solidarity that is created by being in a group of like-others (McMahon & Dick, 2011). Thus, facilitator relatability appears to be an important factor to consider when planning bystander intervention programming for male participants.

Future Directions

In sum, although Wingman 101 was generally well received, findings suggest that a one-session intervention was not enough to address barriers to potential future intervention, and, thus, that the current format is not a highly feasible way to implement this program. Indeed, we feel the declines in positive attitudes about bystander intervention at posttest were likely a direct result of insufficient time to explore barriers to bystander intervention, and to develop skills needed to address those barriers. Although this may not be surprising to those familiar with prevention program best practices (e.g., Nation et al., 2003), it is important as many bystander solutions in the postsecondary environment rely on very brief interventions (e.g., a 1-hr presentation during orientation week) due to time and resource limitations (DeGue et al., 2014). In our case, the short length of the program was based on discussions with key campus stakeholders (e.g., students were accustomed to 1-hr, one-time programs; coaches preferred a 1-hr program). However, although certainly requiring increased resourcing and buy-in, we believe the findings of this study provide additional evidence to support campus policy that prioritizes multisession programming with opportunities for skills practice.

Based on our findings, several directions for program improvement were identified. First, to let participants work through barriers identified during program participation, additional time should be allocated to Wingman 101 program sessions. At a minimum, a booster session should be offered a few weeks after the initial program session, so that participants can return and discuss thoughts that may have arisen during the first program session, and work with facilitators to develop strategies to overcoming barriers. Longer programs have also previously been identified as more effective in improving rape attitudes, including rape myth acceptance, than shorter programs (Anderson & Whiston, 2005). Second, to allow participants to acquire specific bystander-based skills, supervised practice would be a useful addition (Conley, Durlak, & Kirsch, 2015), as practice is necessary to skill formation (Salas & Cannon-Bowers, 2001), and bystander intervention may require unfamiliar skills. Furthermore, in Bennett, Banyard, and Garnhart’s (2014) study of barriers to bystander intervention, barriers related to skills deficits (e.g., “lack of knowing what to say”) were associated with significantly lower helping behaviors toward strangers, but not friends. Thus, skills practice may address the discrepancy found in this study between helping behaviors toward friends versus acquaintances and strangers. Finally, although facilitators were trained to challenge and deconstruct violence-supportive norms (e.g., as they relate to heteronormative masculinity), these topics were not an integrated focus of the program; given the literature reviewed above, they are an important addition to future iterations.

The experience of conducting this study also highlighted the critical role of buy-in across multiple systems (e.g., administration, judicial, health, athletics) for campus-based programming to be successful. However, although the support of each of these systems is key, successful postsecondary programs rely on systematic integration into and coordination with the other institutions and organizations that educate students. In the case of athletes, their uptake and retention of program messages will, in particular, be affected by the reinforcement of those messages by their coach (Jaime et al., 2015), and so strategies to facilitate coach buy-in are critical.

Limitations

A key limitation of this study was sample size. Although we had initially planned to recruit between 175 and 200 athletes from multiple teams, despite best efforts, we were only able to recruit 102 players from three teams. Given our final enrolled sample size (40 per group), in two-sided t tests, we had 80% power to detect a Cohen’s d of 0.63 at the p = .05 level (i.e., a medium-sized effect). Thus, sample size likely resulted in limited ability to detect effects, and where we did find significant effects, results should be interpreted with caution in light of the small sample size and regarded as preliminary only. The pilot nature of the study also meant that within-team randomization was used (as we did not have enough teams to conduct a cluster-randomized trial), which is an additional potential limitation. Second, our data are all self-reported, including fidelity data, which were self-reported by facilitators. Although facilitators were instructed to complete fidelity checklists independently, and although we did find agreement across all facilitators in a given session, there is still the possibility for reporting bias, and in the future, independent observers should also be used. There was also some nonequivalency between the control and treatment group at T0. We also did not include a specific N/A option for behaviors at T2, and so those not reporting a behavior at T2 may not have had the opportunity. Also, because programs occurred over the span of a month, and posttests (T1) were sent 1 week following the program, there was a large range of time between when individuals completed the pretest (T0) and the posttest (T1; range = 11-80 days, M = 35 days), and the posttest (T1) and 2-month follow-up (T2; range = 13-78 days, M = 59 days). Finally, we note that the length of the program is a limitation: Best practice prevention program recommendations highlight the need for longer programs (e.g., Nation et al., 2003), and we urge administrators to consider this need when planning sexual violence prevention programming on their campuses.

We also wish to note limitations of our study with respect to diversity. Reflecting the larger population of the campus where this study was conducted, our data were collected from a primarily White sample who attended a research-intensive university. Because of the focus of our study, all our participants were also cisgender males. Thus, our findings are likely not generalizable to more diverse samples of students, which is a limit of our work, and indeed of most research on bystander-based sexual violence prevention in postsecondary settings. This lack of diversity should be considered when thinking about the broader applicability of our findings, and work looking at the utility of bystander-based prevention strategies in more diverse samples (including in terms of racial/ethnic diversity, sexual diversity, and gender diversity) is needed.

Conclusion

Although bystander-based programming is a theoretically promising approach to reduce sexual violence on college campuses, it requires adequate time to discuss specific strategies, work through barriers, and build skills. The findings of this pilot study suggest that programs should carefully consider facilitator selection, and provide further evidence that the amount of time allocated to the program is a critical component of feasibility.

Footnotes

Appendix A

Wingman 101 Summary of Session.

Activity	Description
Part 1: Introduction to the program	In the first part of the program, facilitators introduce the program to participants. Facilitators also introduce bystander theory, which is the basis for the program.
Part 2: Scenarios	In the second part of the program, facilitators go through three scenarios with participants. These scenarios focus on common bystander-relevant situations that participants may face (Scenario 1: being coerced to drink at a team party; Scenario 2: Hearing a lewd comment from a teammate; Scenario 3: Seeing a teammate take an intoxicated girl at a party upstairs). After hearing the scenario, participants use a colored card to indicate if they would intervene or not. Through the use of colored cards, the facilitators can quickly assess each individual’s choice and then direct questions to individuals or the group as a whole. For each scenario, facilitators lead a discussion about each situation’s risks, as well as barriers participants feel would keep them from intervening.
Part 3: Concluding discussion	In the third part of the program, a summary of bystander theory is reviewed, followed by discussion about its relevance to the scenarios. Participant questions are also answered.

Appendix B

Acknowledgements

The authors thank the participating coaches and teams. Thanks also to Lydia Gill and Kaitlin Kellner for their assistance with data collection, and to Sade Famakinwa for her assistance with data collection and themeing.

Authors’ Note

The data reported in this article were presented in part at the 142nd American Public Health Association Annual Meeting & Expo, New Orleans, Louisiana, November 15 to 19, 2014.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The second author was part of a team of developers of the program evaluated in this paper, which may be a perceived conflict of interest. However, the first author conducted the study and ran all analyses. The first author is not affiliated with the program.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported in part by Doctoral Foreign Study Award 113296 from the Canadian Institutes of Health Research, Ottawa, Ontario.

Author Biographies

Deinera Exner-Cortens, PhD, MPH, is an assistant professor in the Faculty of Social Work, University of Calgary, Canada. Her research focuses on adolescent dating violence, as well as on sexual violence prevention and healthy relationships promotion program evaluation.

Nina Cummings, MS, is the Program Director for Sexual Violence Prevention and Victim Advocacy in the Skorton Center for Health Initiatives at Cornell Health, Cornell University, and has dedicated her career to the prevention of sexual violence and victim advocacy.

References

Anderson

L. A.

Whiston

S. C.

(2005). Sexual assault education programs: A meta-analytic examination of their effectiveness. Psychology of Women Quarterly, 29, 375-388.

Banyard

V. L.

Plante

E. G.

Moynihan

M. M.

(2005). Rape prevention through bystander education: Bringing a broader community perspective to sexual violence prevention (Final report). U.S. Department of Justice. Retrieved from https://www.ncjrs.gov/pdffiles1/nij/grants/208701.pdf

Basile

K. C.

Smith

S. G.

(2011). Sexual violence victimization of women: Prevalence, characteristics, and the role of public health and prevention. American Journal of Lifestyle Medicine, 5, 407-417.

Bennett

Banyard

V. L.

Garnhart

(2014). To act or not to act, that is the question? Barriers and facilitators of bystander intervention. Journal of Interpersonal Violence, 29, 476-496.

Berkowitz

A. D.

(2004). Working with men to prevent violence against women: An overview (part one). Harrisburg, PA: VAWnet.

Berkowitz

A. D.

(2010). Fostering healthy norms to prevent violence and abuse: The social norms approach. In Kaufman

K. L.

(Ed.), The prevention of sexual violence: A practitioner’s sourcebook (pp. 147-172). Fitchburg, MA: NEARI Press.

Burn

S. M.

(2009). A situational model of sexual assault prevention through bystander intervention. Sex Roles, 60, 779-792.

Cantor

Fisher

Chibball

Townsend

Lee

Bruce

Thomas

(2015, September). Report on the AAU Campus Climate Survey on Sexual Assault and Sexual Misconduct. Rockville, MD: Westat.

Classen

C. C.

Palesh

O. G.

Aggarwal

(2005). Sexual re-victimization: A review of the empirical literature. Trauma, Violence, & Abuse, 6, 103-129.

10.

Cohen

(1977). Statistical power analysis for the behavioral sciences. New York, NY: Academic Press.

11.

Coker

A. L.

Fisher

B. S.

Bush

H. M.

Swan

S. C.

Williams

C. M.

Clear

E. R.

DeGue

(2015). Evaluation of the Green Dot bystander intervention to reduce interpersonal violence among college students across three campuses. Violence Against Women, 21, 1507-1527.

12.

Conley

C. S.

Durlak

J. A.

Kirsch

A. C.

(2015). A meta-analysis of universal mental health prevention programs for higher education students. Prevention Science, 16, 487-507.

13.

Corboz

Flood

Dyson

(2016). Challenges of bystander-intervention in male-dominated professional sport: Lessons from the Australian Football League. Violence Against Women, 22, 324-343.

14.

Crowne

D. P.

Marlowe

(1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354.

15.

DeGue

Valle

L. A.

Holt

M. K.

Massetti

G. M.

Matjasko

J. L.

Tharp

A. T.

(2014). A systematic review of primary prevention strategies for sexual violence perpetration. Aggression and Violent Behavior, 19, 346-362.

16.

Dills

Fowler

Payne

(2016). Sexual violence on campus: Strategies for prevention. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

17.

Exner-Cortens

Wells

(2017). State of the science brief: Programmatic approaches to sexual violence prevention and risk reduction in post-secondary settings. Calgary, AB: The University of Calgary, Shift: The Project to End Domestic Violence.

18.

Fisher

B. S.

Cullen

F. T.

Turner

M. G.

(2000). The sexual victimization of college women. Washington, DC: National Institute of Justice.

19.

Gray

M. J.

Hassija

C. M.

Steinmetz

S. E.

(2017). Prevalence and institutional and economic costs of campus sexual assaults. In Sexual assault prevention on college campuses (pp. 7-19). New York, NY: Routledge.

20.

Jaime

M. C.

McCauley

H. L.

Tancredi

D. J.

Nettiksimmons

Decker

M. R.

Silverman

. . . GMiller

(2015). Athletic coaches as violence prevention advocates. Journal of Interpersonal Violence, 30, 1090-1111.

21.

Jordan

C. E.

Combs

J. L.

Smith

G. T.

(2014). An exploration of sexual victimization and academic performance among college women. Trauma, Violence, & Abuse, 15, 191-200.

22.

Katz

(2011, December). Penn State & the bystander approach: Laying bare the dynamics in male peer culture. Harrisburg, PA: National Sexual Violence Resource Center.

23.

Latane

Darley

J. M.

(1970). The unresponsive bystander: Why doesn’t he help. New York, NY: Appleton-Century-Crofts.

24.

McCray

K. L.

(2015). Intercollegiate athletes and sexual violence: A review of the literature and recommendations for future study. Trauma, Violence, & Abuse, 16, 438-443.

25.

McMahon

(2015). Call for research on bystander intervention to prevent sexual violence: The role of campus environments. American Journal of Community Psychology, 55, 472-489.

26.

McMahon

Dick

(2011). “Being in a room with like-minded men”: An exploratory study of men’s participation in a bystander intervention program to prevent intimate partner violence. The Journal of Men’s Studies, 19, 3-18.

27.

McMahon

Farmer

G. L.

(2009). The bystander approach: Strengths-based sexual assault prevention with at-risk groups. Journal of Human Behavior in the Social Environment, 19, 1042-1065.

28.

Moynihan

M. M.

Banyard

V. L.

Arnold

J. S.

Eckstein

R. P.

Stapleton

J. G.

(2010). Engaging intercollegiate athletes in preventing and intervening in sexual and intimate partner violence. Journal of American College Health, 59, 197-204.

29.

Murnen

S. K.

Kohlman

M. H.

(2007). Athletic participation, fraternity membership, and sexual aggression among college men: A meta-analytic review. Sex Roles, 57, 145-157.

30.

Nation

Crusto

Wandersman

Kumpfer

K. L.

Seybolt

Morrissey-Kane

Davino

(2003). What works in prevention: Principles of effective prevention programs. American Psychologist, 58, 449-456.

31.

Payne

D. L.

Lonsway

K. A.

Fitzgerald

L. F.

(1999). Rape myth acceptance: Exploration of its structure and its measurement using the Illinois Rape Myth Acceptance Scale. Journal of Research in Personality, 33, 27-68.

32.

Salas

Cannon-Bowers

J. A.

(2001). The science of training: A decade of progress. Annual Review of Psychology, 52, 471-499.

33.

Shrive

F. M.

Stuart

Quan

Ghali

W. A.

(2006). Dealing with missing data in a multi-question depression scale: A comparison of imputation methods. BMC Medical Research Methodology, 6, Article 57.