Abstract
Among violence prevention educators and researchers, there is growing interest in sexual, dating, and intimate partner violence (SV/DV/IPV) prevention programs for males because of evidence showing that boys and men are more likely than girls and women to perpetrate SV as well as more severe forms of DV/IPV. To date, comprehensive guidance on the content, structure, delivery, and effectiveness of such programs is limited. We reviewed randomized controlled studies that evaluated SV/DV/IPV perpetration prevention programs for boys and men. Searches yielded 5,249 potential documents for review of which 10 met inclusion criteria—representing 9 unique studies of 7 distinct programs. Two reviewers independently reviewed and abstracted data from these studies regarding program setting and target audience; type of violence addressed; number and length of program sessions; program duration, topics, activities, and delivery mode; and implementer details. Study characteristics were also examined (sample size, participant characteristics, recruitment, randomization, comparison/control condition, data collection protocols, attrition, measures of violence perpetration, and perpetration findings). The Cochrane Risk of Bias Tool was used to assess study design quality. Results show considerable heterogeneity among program content and delivery strategies, study designs, and outcome measurement. Study sample size ranged widely, and most used cluster-randomized designs, recruited undergraduate college students, and evaluated a multisession program delivered via group sessions. Only one program reduced men’s self-reported SV perpetration. Accordingly, critical gaps exist around “what works” for SV/DV/IPV perpetration prevention programs for boys and men.
Developing and evaluating sexual, dating, and intimate partner violence (IPV) prevention programs are critical given the high prevalence and deleterious consequences of such violence (Breiding et al., 2014; Centers for Disease Control and Prevention [CDC], 2017; DeGue et al., 2014). IPV includes stalking, sexual violence (SV), psychological aggression, or physical aggression perpetrated by current or former romantic partners (Breiding, Basile, Smith, Black, & Mahendra, 2015; CDC, 2017). Dating violence (DV) is a type of IPV that occurs between current or former unmarried adolescent or young adult romantic partners (Dardis, Dixon, Edwards, & Turchik, 2015). Although SV—any unwanted sexual activity where consent is not received or freely given (Breiding et al., 2014)—can occur within romantic relationships, SV can also occur between acquaintances and strangers.
Although boys and men are not the sole perpetrators of SV/DV/IPV, they are more likely than girls and women to perpetrate SV and severe forms of DV/IPV that result in emotional trauma, physical injury, and death (Archer, 2000; García-Moreno et al., 2013). Such evidence has spurred the development of gender-specific SV/DV/IPV programs that focus on preventing perpetration by boys and men. Despite growing interest in male-focused interventions among prevention educators and researchers, to our knowledge, there have been no literature reviews to examine whether gender-specific programs are effective in reducing SV/DV/IPV perpetration. Likewise, guidance is limited regarding the program content, structure, and delivery. This review sought to address these gaps by identifying male-focused SV/DV/IPV perpetration prevention programs that have been evaluated using randomized designs and have measured changes in perpetration behaviors longitudinally.
Preventing Male-Perpetrated SV/DV/IPV
Strategies are needed to prevent both first incidents of SV/DV/IPV perpetration (i.e., primary prevention) and its recurrence (i.e., secondary prevention; CDC, 2004). Additionally, prevention includes interventions focused on the general population of boys and men (i.e., universal interventions in the context of this review) and on boys and men at increased risk for violence perpetration (i.e., selected interventions; CDC, 2004). Therefore, this review focused on interventions intended to prevent SV/DV/IPV perpetration, included primary and secondary prevention, and focused on universal or selected interventions.
Our research team focused this review on SV/DV/IPV rather than a single form of violence for two reasons. First, boys and men are disproportionately identified as perpetrators of SV/DV/IPV (Archer, 2000; García-Moreno et al., 2013), several existing SV/DV/IPV prevention programs have been developed for delivery to boys and men exclusively (DeGue et al., 2014). Second, SV/DV/IPV perpetration share multiple risk factors (e.g., harmful gender role norms, association with delinquent peers), and the CDC promotes a crosscutting framework to target these shared root causes to prevent multiple violence types simultaneously (Wilkins, Tsao, Hertz, Davis, & Klevens, 2014).
Current Study
Other reviews of SV/DV/IPV prevention program evaluations and intervention research exist (e.g., Cornelius & Resseguie, 2007; DeGue et al., 2014; De La Rue, Polanin, Espelage, & Pigott, 2014; Ricardo, Eads, & Barker, 2011). However, the current study is unique in several ways. First, we focus on studies that used randomized designs to evaluate SV/DV/IPV prevention programs designed specifically for boys and men. Previous reviews include programs for mixed-sex audiences (e.g., Cornelius & Resseguie, 2007; DeGue et al., 2014; De La Rue et al., 2014), focus on a single form of violence (e.g., SV exclusively: DeGue et al., 2014; Ricardo et al., 2011; DV/IPV exclusively: Cornelius & Resseguie, 2007), and/or focus on a particular delivery context (e.g., school-based interventions; De La Rue et al., 2014). In addition, the current review provides a comprehensive overview of male-focused programs across developmental periods and in various contexts (e.g., programs for both adolescents and adults in school- and community-based programs) and exclusively describes programs that assessed change in perpetration.
We had several aims for this review. Specifically, we (a) identified programs for boys and men that intended to prevent SV/DV/IPV perpetration and that have been evaluated using randomized designs; (b) systematically reviewed each of the identified programs to determine the violence types addressed, the target audience, and key components of program delivery, activities, structure, and content; and (c) examined each program’s outcome evaluation, including methods used, program effectiveness, and methodological quality using the Cochrane Risk of Bias Tool (Higgins & Altman, 2008).
Method
This review was guided by Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations for systematic reviews (Moher, Liberati, Tetzlaff, & Altman, 2009). Figure 1 details our review process. Literature included in this review met the following eligibility criteria: was a peer-reviewed article or gray literature (i.e., report or doctoral dissertation) published in English between January 1997 and August 2017, reported an evaluation of a universal or selected primary or secondary prevention program designed to prevent SV/DV/IPV perpetration, evaluated a program designed for and delivered to boys and men exclusively, studied program effects on SV/DV/IPV, and used a randomized design.

Flowchart of the review search process. SV = sexual violence; DV = dating violence; IPV = intimate partner violence.
To identify studies meeting our inclusion criteria, we used several strategies. In September 2017, we identified peer-reviewed literature through searches of six electronic databases: PsycINFO, CINAHL Plus with Full Text, PubMed, Web of Science, Social Work Abstracts, and Social Services Abstracts. In July 2018, we supplemented earlier search results with searches of seven gray literature databases and repositories: ProQuest Dissertations and Theses Global, Open Grey, National Resource Center on Domestic Violence publications, CDC SV/DV/IPV publication lists, WorldCat Dissertations, National Institute of Child Health and Human Development publications, and National Institute of Mental Health publications. For all searches, we used the search string (violence OR assault) AND prevention program AND (boys OR men OR male). When a Boolean search string was inappropriate, we used combinations of these search terms. When feasible and appropriate, searches were focused using filters for English language, peer-reviewed article, human subjects, and published 1997–2017. Database and repository searches yielded 5,188 documents (4,238 unduplicated documents). A title review conducted by one member of the research team identified 420 potentially eligible documents. The same team member then examined the document abstracts to assess eligibility for inclusion and retained 36 articles for further review. This team member and one additional team member then independently assessed the Method and Results sections of each of the 36 documents for eligibility and agreed that eight documents met eligibility criteria.
We also reviewed reference lists of eligible documents found in searches of the peer-reviewed and gray literatures, as well as the reference lists of two prior reviews (DeGue et al., 2014; Ricardo et al., 2011). This process of reference harvesting identified another 61 potential documents, which were reviewed using the same protocol described above, yielding two additional documents for inclusion. Our final sample included 10 documents that discussed nine distinct evaluation studies of seven violence prevention programs designed for boys and men.
We developed a data abstraction form that collected information on the type of violence perpetration addressed; program setting, target audience, number and length of sessions, duration, delivery mode, topics, and activities; program implementer and required training; and study methods. One team member abstracted data, and a second member checked for accuracy. We discussed discrepancies until these two team members reached consensus.
The quality of each study design (k = 9) was assessed using the Cochrane Risk of Bias Tool (Higgins & Altman, 2008; Higgins et al., 2011) to evaluate the study’s potential for selection, reporting, performance, detection, and attrition biases. Selection bias refers to the potential for bias in study findings to result from both the process used to randomize study units (i.e., random sequence generation bias) and the process used to mask the randomization protocol from those involved in the study (i.e., allocation concealment). Reporting bias refers to the possibility of study authors having chosen to report certain outcome findings instead of all outcome findings. Performance bias refers to the possibility for biased findings due to procedures used to mask the study condition from study personnel and participants, and detection bias refers to the possibility for biased findings because of procedures used to mask the study condition from outcome assessors. Last, attrition bias refers to potential for biased findings resulting from missing outcome data and methods used for addressing such missingness (Higgins & Altman, 2008). Two team members pilot tested this tool with two randomly selected studies. The same team members then independently assessed the remaining studies and discussed discrepancies to reach consensus.
Results
Programs Evaluated
The 10 reviewed documents (nine peer-reviewed articles and one dissertation) reported on nine evaluation studies of seven distinct programs (see Table 1). Three articles described two studies that assessed the Coaching Boys into Men (CBIM) program (Jaime et al., 2016; Miller et al., 2012, 2013). Two articles reported on evaluations of The Men’s Program (Foubert, 2000; Foubert, Newberry, & Tatum, 2007). The remaining five documents reported on an evaluation study of an individual program: Men’s Discussion Groups (Hossain et al., 2014), RealConsent (Salazar, Vivolo-Kantor, Hardin, & Berkowitz, 2014), Sexual Assault Prevention Program for College Men (Lobo, 2004), the Men’s Project (Gidycz, Orchowski, & Berkowitz, 2011), and an unnamed program, hereafter referred to as the “Video Program” (Stephens & George, 2009).
Descriptions of the Seven Violence Prevention Programs Assessed.
Note: NA = not applicable. SV = sexual violence; DV = dating violence; IPV = intimate partner violence; Language used in this table mirrors language used by study author(s) and reflects information provided in each document reviewed.
CBIM
CBIM focuses on preventing DV perpetration among male high school athletes. This program includes 11–12 brief sessions (10–15 min each) delivered in-person, in a group format. Program topics include abusive behavior, gender attitudes/norms, and ways in which bystanders can prevent DV. The program includes weekly discussions and encourages use of “teachable” moments in which the program implementer models positive behaviors or language. The program may also include supplementary activities such as developing an anti-violence campaign with program participants and hosting a school-wide pledge day to encourage all students to sign pledges not to perpetrate violence. Program implementers are athletic team coaches whose training on the CBIM program includes a 60-min session by a trained violence-prevention advocate (e.g., someone from a local crisis center) and a “coaches kit” (available online) that includes program materials. In addition, a trained advocate is available during program delivery to assist coaches with concerns and violence-related disclosures.
Men’s Discussion Groups
This program aims to prevent IPV perpetration among male community members 15 years and older. Men’s Discussion Groups was developed for implementation in communities affected by war or conflict. The program consists of 16 weekly sessions, but the duration of each session is not specified. Sessions are delivered in-person using a group format. The primary activity is the weekly discussion of topics such as gender violence against women and girls, the effects of violence, and healthy relationships. Hossain et al. (2014) did not specify program implementers (e.g., advocate, peer), but the implementers in the study participated in a 6-week training during which each trainee practiced delivery of the program.
The Men’s Program
This program targets male undergraduate college students, including fraternity members. The program consists of a single 1-hr group session delivered in person. Topics presented include definitions of rape, common experiences of rape survivors, skills development to help female rape survivors in recovery, enhancing communication in sexual situations, such as gaining sexual consent, and statistics on prevalence of SV. Program activities center on a video depicting the rape of a male police officer by two men. The video is followed by discussion on the range of topics previously mentioned. The session concludes with a question–answer period and statistics on SV prevalence. The program is implemented by peer educators, such as undergraduate men, who have received 20 hr of peer education training.
RealConsent
This program for male undergraduate students consists of six, 30-min modules delivered online to individual participants who can access and complete the modules at their own pace. Module topics include information on bystander knowledge and skills to prevent SV; myths about SV; attitudes regarding date rape, sexual consent, masculinity/gender roles, and communication; and empathy for sexual assault survivors. Modules incorporate various activities, including interactive activities, didactic presentations, and a “serial drama” in which video clips provide information and model strategies for violence prevention (Salazar et al., 2014, “Interventions,” para. 2). Participants can post messages and questions or contact RealConsent program staff via e-mail.
Sexual Assault Prevention Program for College Men
This program for male undergraduates consists of a single 2-hr, in-person group session. The program delivers content through “a combination of didactic presentation, group discussion, and a small group activity” (Lobo, 2004, p. 57), covering topics such as sexual assault facts and definitions, consent and sexual communication, how to approach risky sexual situations, including those involving alcohol, attitudes toward rape, and confronting other men’s sexually aggressive behavior. Program activities include vignettes of hypothetical situations for discussion. Lobo (2004) evaluated the program as implemented by a male graduate student with “professional training” (p. 57) and did not describe the details of that training.
The Men’s Project
This program for male undergraduates is delivered through an initial 1.5-hr session and a follow-up 1-hr booster session 4 months later. Program content includes topics such as masculinity/male gender roles, sexual consent, rape myths, social norms regarding sexual aggression, the bystander role, data on sexual assault, and men’s norms for appropriate attitudes and behaviors in sexual encounters. The initial session includes discussion of SV’s effects on women, rape myths, and sexual consent; operationalization of consent in example scenarios; presentation of program participant and campus-wide data on men’s discomfort with witnessing inappropriate behavior and language of other men; and an adapted version of the Small Group Norms Correction model (Far & Miller, 2003) to challenge misperceptions of other men’s attitudes and behaviors regarding sexual assault. The booster session reviews initial session content and includes small group discussions.
Video Program
This program for male undergraduates is a single 50-min session that uses a group setting; the sole program activity is watching a prerecorded video that covers topics such as developing empathy for sexual assault survivors, how to help survivors in recovery, and the links between alcohol and sexual assault. The video has four segments (see Table 1). Stephens and George’s (2009) evaluation of this program did not describe implementers.
Findings on Program Effectiveness
Table 2 presents the characteristics of the nine evaluation studies (reported in 10 documents) and findings regarding program effects on SV/DV/IPV perpetration.
Study Characteristics of Eight Studies Reported in 10 Documents.
Note: RCT = Randomized controlled trial. CTS2= Revised Conflict Tactics Scale. SV = sexual violence. DV = dating violence. IPV = intimate partner violence.
Language used in this table mirrors language used by study author(s) and reflects information in each document. As such, some information is available for some studies and not others. We report Cohen’s d when available. This table includes information on any significant differences in study condition attrition rates and between study completers and noncompleters when reported. All studies were U.S.-based except Hossain et al. (2014), which was conducted in Côte d’Ivoire.
aSample size reported indicates the number of participants who completed all data collection time points included in analyses.
bMiller et al. (2012, 2013) report on data collected at two distinct time points (i.e., 12-week follow-up and 12-month follow-up, respectively) in the same evaluation study.
c Miller et al. (2013) reported “White n = 18,” which appears to be a typographical error; we calculated the value as 518.
dWe retained this study because of the limited number of studies available; however, the study design for analyses of program effects on perpetration is quasi-experimental (i.e., people could not be randomized to join or not join a fraternity).
Effectiveness of CBIM
Although the two articles by Miller et al. (2012, 2013) reported on the same evaluation study, each article reported on a different follow-up period. Miller et al. (2012) included 1,798 male athletes in Grades 9–12. The 12-month follow-up by Miller et al. (2013) included 1,194 of the athletes who participated in the 12-week study, all of whom were in Grades 9–11 at baseline. The racial/ethnic makeup of the study was diverse.
The CBIM study used a cluster-randomized controlled design in which 16 high schools were randomly assigned to two study conditions—8 clusters assigned to the CBIM intervention and eight assigned to a coaching-as-usual comparison condition. Data were collected at three time points: baseline (n = 2,006 male athletes; Miller et al., 2012), 12 weeks after baseline (n = 1,798; Miller et al., 2012), and 12 months after baseline (n = 1,194; Miller et al., 2013).
A modified version of the Revised Conflict Tactics Scale (CTS2) was used to capture self-report data on DV perpetration (i.e., physical, sexual, and emotional/psychological) of participants who had ever dated a girl or woman. The CTS2 was used at both follow-up time points to measure DV perpetration in the past 3 months. Participants were classified as DV perpetrators if they affirmed any of the 10 items describing abusive behaviors. Analyses controlled for baseline differences in school grade, immigration status, race/ethnicity, and parental education. Findings from the 12-week follow-up (Miller et al., 2012) showed no statistically significant difference in DV perpetration between the intervention and comparison groups. However, findings at the 12-month follow-up (Miller et al., 2013) showed significantly less physical/sexual/psychological DV perpetration reported by the intervention group relative to the comparison group; past 3-month DV perpetration was .15 units lower in the intervention group than the comparison group (95% confidence interval [CI] = [−0.27, −0.03], p < .05). A third article reported an evaluation of CBIM conducted by Jaime et al. (2016) using a sample of 148 athletes from Grades 7 to 12 drawn from two high schools. The sample was somewhat diverse racially/ethnically.
Jaime et al. (2016) used a cluster-randomized approach in which all participants received the CBIM program but compared the program as implemented by either a coach or a community-based violence prevention advocate. One high school was assigned to the coach-led condition and the other to the violence prevention advocate-led comparison condition. Data were collected at baseline and at 3 months after baseline.
Jaime et al. (2016) used the same perpetration measures as Miller et al.’s (2012, 2013) study, allowing comparisons across studies. Of note, this study differed from other studies included in this review because its primary focus was program implementation, testing implementer type (coach versus advocate), rather than program effects on violence perpetration. Analyses were adjusted for school grade and race/ethnicity. Jaime et al. found no statistically significant differences in DV perpetration over time for either condition and found no significant differences between the study conditions for changes in pre/posttest scores for DV perpetration.
Effectiveness of men’s discussion groups
Hossain et al. (2014) assessed the effectiveness of Men’s Discussion Groups in a sample of 300 men from 12 rural conflict-affected communities in Côte d’Ivoire. The mean age of participants was 40 years (range not reported). The sample of 346 men represented 10 ethnic groups, with Baoulé (25.7%) the largest group.
A cluster-randomized design was used with six communities assigned to the intervention group and six to the comparison group. All 12 communities received a similar community-level gender-based violence (GBV) program to increase awareness of the consequences of GBV, raise awareness of the rights of women, and create a committee trained to support survivors of GBV. The intervention group received this community-level intervention plus Men’s Discussion Groups. The study team collected data at baseline and 12 months after the program ended.
The study measured physical and sexual IPV based on the reports made by the female partners of study participants regarding violence perpetrated by the study participant in the 12 months prior to baseline and again at the 12-month follow-up. The study “…questionnaire drew from violence and health outcome modules including the World Health Organization Multi-country Study on Women’s Health & Domestic Violence against Women, the LSHTM violence and health among women asylum seekers study, and a trial on intimate partner violence and HIV prevention in Uganda. (Hossain et al., 2014, p. 344)
Effectiveness of the Men’s Program
Foubert (2000) examined the impact of the Men’s Program on 145 college fraternity men (M = 20.33 years; range not reported). Participant race/ethnicity was overwhelmingly White.
A cluster-randomized approach was used with four fraternities assigned to the intervention group and four to the no-program control group. Participants were assessed at baseline, posttest immediately after the program, and 7 months after the program.
The self-report Sexual Experiences Survey (SES) was used at posttest and 7-month follow-up to assess likelihood of sexual coercion perpetration (continuous measure). Analyses examining intervention effects, pretesting, and a Treatment × Pretesting interaction did not find any significant differences between the intervention and control groups on sexual coercion.
Foubert, Newberry, and Tatum (2007) conducted a second evaluation of The Men’s Program with a sample of 565 male undergraduate students. Participant race/ethnicity was not reported. The randomized trial used a Solomon four-square design that varied pretesting among four conditions: program with pretest, program without pretest, comparison with pretest, and comparison without pretest. The programming used with the comparison condition was not intended to promote change in participants’ attitudes or behaviors regarding violence. Participants were assessed at three time points: baseline, posttest immediately after the program, and 7 months after the program.
The SES was used to assess self-reported sexual coercion perpetration. At baseline, the survey measured lifetime sexual coercion perpetration. At 7-month follow-up, the survey measured sexual coercion perpetration since baseline. Among study participants who joined fraternities, program participants (M = .05, SD = 0.22) reported perpetrating significantly fewer sexually coercive behaviors than those in the comparison condition (M = .40, SD = 1.3) at the 7-month follow-up, F(1, 109) = 4.06, p < .05.
Effectiveness of RealConsent
Salazar, Vivolo-Kantor, Hardin, and Berkowitz (2014) evaluated the RealConsent program with a sample of 215 male undergraduate students who self-identified as either heterosexual or bisexual. The sample average age was 20.38 years old (SD = 1.67; range not reported) and was somewhat racially/ethnically diverse with more White participants than other groups.
Participants were randomly assigned to the RealConsent intervention group or the comparison group that received the Health Connection program, which is a web-based health promotion program that does not address SV prevention. Data were collected at three time points: baseline, posttest immediately after the program, and at 6-month follow-up.
The perpetration outcome of interest was change in sexual coercion, assessed with the Sexual Coercion subscale of the CTS2. Baseline assessments asked participants about lifetime sexual coercion, and the 6-month follow-up asked about sexual coercion in the past 6 months. Analyses, adjusted for baseline scores for sexual coercion, prosocial intervening behaviors, and sociodemographic variables, showed that at the 6-month follow-up, program participants reported perpetrating significantly less sexual coercion than those in the control group (Cohen’s d = .29). As compared with the control group, program participants had 73% lower odds of perpetrating sexual coercion.
Effectiveness of the Sexual Assault Prevention Program for College Men
Lobo (2004) evaluated this program with 237 male undergraduate students (age range:18–21+ years). Racial/ethnic makeup of the sample was primarily non-Hispanic Caucasian. Study participants were randomly assigned to the intervention or the waitlist control group. Data were collected at 3 times: baseline, 3 months postprogram, and 7 months postprogram.
For each of the data waves, participant sexual aggression was measured using the self-report SES. At baseline, participants were asked to report on their behavior between age 14 and 17 years. Each of the two follow-up surveys asked participants to report on their behavior during the interval since completing the prior questionnaire. Analyses controlled for history of perpetration (none, moderate, or severe) and found group membership (i.e., intervention vs. control) was not significantly related to sexual aggression perpetration at the 3-month or the 7-month follow-ups.
Effectiveness of the Men’s Project
Gidycz, Orchowski, and Berkowitz (2011) evaluated The Men’s Project with a sample of 460 male college students, aged 18–19 years old, from 12 randomly selected first-year residence halls. The racial/ethnic makeup of the sample was overwhelmingly White. The study used a cluster-randomized design with six residence halls assigned to receive the intervention and six assigned to the waitlist control condition. Study participants were assessed at three time points: baseline, before the booster session (4 months after the initial session), and 7-month follow-up (unclear whether follow-up occurred 7 months from baseline or another time point).
The study measured self-reported sexual aggression with the SES as a dichotomous variable (no sexual aggression/sexual aggression). At baseline, the SES assessed sexual coercion from age 14 years. At the 4- and 7-month follow-ups, SES assessed sexual coercion since the last data collection. Results from the 4-month follow-up showed the intervention group (1.5%, n = 3) had significantly lower rates of sexual aggression than the control group, 6.7%, n = 17; χ2(1, n = 437) = 7.33, p < .01; p < .05, Fisher’s exact test. However, at the 7-month follow-up, the results showed no significant between-group differences on reports of sexual aggression.
Effectiveness of the Video Program
Stephens and George (2009) examined the impact of the Video Program on 65 White males. Participants aged 18–29 years (M = 19.3 years, SD = 1.8). The study used a randomized controlled trial with a 2 (high risk vs. low risk) × 2 (intervention vs. comparison) between-subjects design. Participants were randomized to the intervention or comparison condition. Participants in the intervention condition watched a 50-min video addressing SV topics, whereas participants in the comparison condition watched a 50-min video that did not include information on SV. Participants were assessed 3 times: baseline before randomization, posttest immediately after the video, and 5-week follow-up.
The study measured self-reported sexual coercion using the modified-SES (M-SES). The M-SES was used to classify participants at baseline as high risk versus low risk for perpetrating sexual coercion. This tool was readministered at the 5-week follow-up to measure sexual coercion perpetrated since the posttest. Findings for the full sample at the 5-week follow-up showed that, as compared with the comparison group participants, intervention participants had an increase in self-reported sexually coercive behavior; however, this effect was not significant (p = .053). Also, as compared with comparison group participants (47%, n = 9) categorized at high risk for perpetrating sexual coercion, a higher percentage of high-risk intervention participants (82%, n = 9) reported perpetrating sexual coercion since the baseline measure (Cohen’s d = −.74, 95% CI [−.97,−.50], p = .030). No significant effects were found for the low-risk group. Covariates included outcome pretest scores and a social desirability measure.
Bias Assessment of the Studies
We used the Cochrane Risk for Bias Tool (Higgins & Altman, 2008) to assess the potential for SV/DV/IPV perpetration findings to be biased by selection, reporting, performance, detection, and attrition biases (see Table 3). For the majority of reviewed studies (k = 7), we could not clearly evaluate the potential for both forms of selection bias (i.e., random sequence generation and allocation concealment) because the study reports did not provide sufficient information. We rated the remaining two studies (Miller et al., 2012, 2013; Salazar et al., 2014) as low risk for selection bias based on each study’s explicit description of a randomization protocol. Only three of the seven studies were registered and listed on the National Institutes of Health Clinical Trials website (clinicaltrials.gov). However, none of the listings included the original study protocol. As such, the studies included in this review could not be evaluated for risk of reporting bias because our team did not have access to original study protocols.
Quality Assessment of Reviewed Studies Based on Cochrane Risk of Bias Tool Ratings and Support (k = 9)
Note: Miller et al. (2012, 2013) are treated as a single study. All ratings are based solely on the information provided in the articles reviewed. As such, studies may have accounted for potential biases without reporting such methods in the articles for a variety of reasons (e.g., publication word limits).
a Selection bias is bias in study findings resulting from the randomization process and the process used to mask the randomization protocol from those involved in the study.
b Reporting bias refers to the possibility that the authors chose to report certain outcome findings rather than all outcome findings.
c Performance bias describes the possibility of findings being biased due to procedures used to mask the study condition from study personnel and participants.
d Detection bias refers to the possibility for findings to be biased because of procedures used to mask the study condition from outcome assessors.
e Attrition bias refers to at study’s potential for being biased by missingness in the outcome data and methods used for addressing missing data. For more information on the forms of bias, see Higgins and Altman (2008).
We rated eight of the nine studies as high risk for performance bias based on the high likelihood of study personnel, participants, and outcome assessors being aware of the study condition to which participants had been assigned. Salazar et al. (2014) was rated as low risk for performance bias because (a) the article provided a clear description of the process used for masking study conditions and (b) the intervention was delivered online to individual participants, making it unlikely that study participants would be aware of their own or others’ assigned study condition.
Findings regarding detection bias were more varied, although we found too little information to discern the risk for detection bias in five studies (Foubert et al., 2007; Gidycz et al., 2011; Jaime et al., 2016; Miller et al., 2012, 2013; Stephens & George, 2009). We found three studies to be at high risk for detection bias (Foubert, 2000; Hossain et al., 2014; Lobo, 2004). In the Foubert (2000) study, one researcher implemented all study steps, including the administration of the study questionnaire, making it likely that the outcome assessor was aware of which participants were in each study condition. Hossain et al. (2014) did not mask study conditions and orally administered questionnaires. Lobo (2004) used a waitlist control group and did not administer a program to the control group. Last, Salazar et al. (2014) was rated as low risk for detection bias because the outcome assessment was administered online.
Last, we found eight of nine studies to have unclear risk for attrition bias. These studies received this rating because each was missing data related to the perpetration outcome and did not adequately discuss how, or if, missing data were addressed. Salazar et al. was the exception. This article described missing data and methods used to address it (i.e., imputation).
Discussion
To investigate whether SV/DV/IPV prevention programs for boys and men are effective and to compile available guidance regarding the content, structure, and delivery of such programs, we rigorously reviewed male-targeted programs focused on preventing SV/DV/IPV perpetration that have been evaluated using randomized designs and have measured perpetration behaviors.
Program Characteristics
The format of the seven programs tested varied widely across number of sessions, session length, and program duration. Given these diverse approaches and variation in program dosage, it is not surprising that this review provided little insight on what might constitute an effective dose of a male-targeted perpetration prevention program. However, four of seven programs involved delivery of more than one session, an approach that is consistent with the literature suggesting that single-session programs might be less likely to change complex behaviors such as SV/DV/IPV perpetration (DeGue et al., 2014; Vladutiu, Martin, & Macy, 2011).
Programs also varied in delivery approach, topics, types of activities, and implementer type and training. Although most programs delivered content through in-person groups, a few programs used video recordings or online platforms for program delivery. In terms of topics and activities, all programs addressed definitions of violent behaviors. Most programs addressed gender norms in some manner, and most programs included discussion-based activities delivered to individual participants. Five programs used trained implementers to deliver the program, including peer educators, athletic coaches, or violence prevention advocates. However, the literature provided limited information about who implemented these programs (e.g., professional role, education) and their training, despite research indicating that programs’ effectiveness may be related to program implementer characteristics and training (Mihalic, Irwin, Fagan, Ballard, & Elliott, 2004; Nation et al., 2003). Jaime et al. (2016) was an exception with its focus on comparing CBIM implementation by a coach versus a violence prevention advocate.
Differences Among Study Designs and Approaches to Perpetration Measurement
This review also found considerable heterogeneity in study designs, particularly in terms of study sample sizes and time to final follow-up. Diversity in these two areas is notable because both sample size and follow-up data collection time points influence a study’s capacity to detect significant effects in perpetration behaviors. As the key research outcome, SV/DV/IPV perpetration is a relatively rare event that requires time and opportunity, studies with relatively small samples and shorter follow-up times might not have had sufficient participants and/or time to detect significant program effects (i.e., power). As such, some of the programs in this review may have shown promise in impacting perpetration with larger samples and longer follow-up.
There was also significant diversity in the measurement of violence perpetration. Across studies, researchers used several different tools and methods for measuring SV/DV/IPV perpetration (e.g., CTS2, SES, M-SES). Such diversity is reasonable, given that the programs and studies varied in their focal outcomes. Nonetheless, the use of varied measurement approaches makes it challenging to make direct comparisons of program effects.
Potential Risk for Bias in Perpetration Findings
Given the limited amount and types of information reported in the documents we reviewed, we were unable to assess the potential for multiple types of bias, particularly selection, reporting, and attrition biases. Based on our assessment of studies using the Cochrane Risk for Bias Tool (Higgins & Altman, 2008), we found that most studies were at high risk for performance bias.
Effects of the Programs on Male-Perpetrated SV/DV/IPV
Significant effects on SV/DV/IPV perpetration were found for five programs: RealConsent (Salazar et al., 2014), CBIM (Miller et al., 2013), The Men’s Program (Foubert et al., 2007), The Men’s Project (Gidycz et al., 2011), and the Video Program (Stephens & George, 2009). RealConsent was the only program found to significantly decrease SV perpetration among a universal population (i.e., college men broadly). CBIM had significant effects on overall DV perpetration (i.e., not reported individually for physical, psychological, or sexual DV) at 12-month follow-up specifically among male high-school athletes, a high-risk population. One of the two studies of The Men’s Program (Foubert et al., 2007) found significant program effects on SV perpetration only among participants who joined a fraternity, whereas the other study (Foubert, 2000) did not find effects on SV perpetration. Gidycz et al.’s (2011) evaluation of The Men’s Project indicated a lack of persistent effects on SV perpetration, and the evaluation of the Video Program showed that as compared with high-risk comparison group participants, a higher percentage of high-risk program participants reported perpetrating SV after watching the video. This mix of findings, paired with the heterogeneity in intervention approaches, suggests that there is insufficient evidence available to describe with certainty what works to prevent SV/DV/IPV perpetration in male-focused programs.
Implications for Research
This study determined limited experimental research exists on male-focused SV/DV/IPV prevention programs that measured effects on SV/DV/IPV perpetration behavior. Further, the limited available research on this topic was conducted primarily in the United States in school settings, which were primarily colleges and universities with predominantly White samples. Only one study (Hossain et al., 2014) was conducted outside of the United States in Côte d’Ivoire, and this same study was also the only study conducted in a community setting. As such, our key recommendation is for researchers to conduct additional evaluation of such programs, using experimental designs, measuring SV/DV/IPV perpetration behaviors, and including more diverse settings and participants. Whenever possible, we recommend that such studies have large samples and sufficient follow-up time to ensure that the research is designed to detect effects if they exist. Relatedly, studies need to be resourced to recruit and retain adequate numbers of participants in longitudinal studies. Additionally, given the dearth of replication studies determined in this review, we call for interventions to be tested in new settings and by investigators who were not the original program developers.
Consistent with research showing that SV/DV/IPV perpetration and other forms of violence (e.g., bullying, youth violence) share risk factors (Wilkins et al., 2014), we recommend that future studies of violence prevention programs, including but not limited to those on SV/DV/IPV, investigate multiple perpetration outcomes across violence types. We also call on researchers to determine common measures for such outcomes that could be used across studies to help to compare outcomes among prevention strategies.
Only three of the documents representing two distinct studies (Jaime et al., 2016; Miller et al., 2012, 2013), all of which reported on CBIM, examined program effects on an adolescent population. Six studies were conducted with college men, and one study was conducted with boys and men aged 15 and older. These findings demonstrate both a lack of diversity of developmental stages across our sample of studies and the need for additional research with younger populations. The objective of primary prevention programs is to stop violent behavior before it starts, therefore administering such programs to young adults likely misses this goal for some of the participants (DeGue et al., 2014; Nation et al., 2003). Thus, future research should focus on testing programs for boys in early adolescence and possibly even younger, when feasible. Nonetheless, we would also like to acknowledge the potential challenges of developing, implementing, and rigorously investigating the sensitive topics of SV/DV/IPV with minors. Although there is a clear need for primary prevention programs aimed at young people, the practicalities of accessing male minors in the settings in which they are engaged (e.g., community-based programs, faith-based organizations, schools) are likely to be challenging. Likewise, obtaining both minors’ assent and parental consent for such research might not be feasible.
Given the complex and sensitive nature of research on SV/DV/IPV, avoiding all risk for sources of bias is extremely challenging. For example, it might not be feasible for researchers to prevent study staff from being aware of participants’ study condition assignment. Likewise, research participants not only need to be fully informed about study goals and procedures but also need to be free to withdraw from participation at any time. We encourage future research to address potential sources of bias (i.e., threats to study validity) in study findings whenever possible and to do so in ethical ways. Such methods may require creative and novel approaches. We also encourage researchers to consider which sources of bias are most critical in the conduct of prevention research studies. If not all sources of bias can be fully addressed, violence researchers could collectively determine which sources of bias are most important to address in order to ensure high-quality study findings.
Of equal importance, recent research has underscored the potential for both mixed-gender and gender-specific SV prevention programs to produce a boomerang effect (i.e., increase SV perpetration among men at high risk for SV perpetration; Malamuth, Huppin, & Linz, 2018). Stephens and George’s (2009) evaluation of the Video Program (reviewed here) is consistent with this research—a significantly higher percentage of high-risk intervention participants reported perpetrating sexual coercion since the baseline measure compared with comparison group participants categorized at high risk for perpetrating sexual coercion. We echo Malamuth, Huppin, and Linz’s (2018) call for researchers to attend to the potential harmful effects of SV/DV/IPV perpetration prevention programs delivered to audiences with men at high risk for perpetrating such violence to determine which prevention programs should be offered to boys and men universally and which should be targeted to specific subgroups.
Last, of the studies included in this review, the heterogeneity in program activities, content, delivery, length, and strategies precluded a clear, comprehensive summary of recommended program elements of SV/DV/IPV prevention. Research is needed to determine what key program elements and program dosage are necessary for effective SV/DV/IPV perpetration prevention among boys and men. Therefore, we recommend that future research seeks to investigate the elements (e.g., activities, topics, delivery method, implementer) of such male-focused programs that are most likely to ensure effective prevention of SV/DV/IPV perpetration behaviors. We also encourage researchers to detail and document program approaches, activities, and implementation strategies in published reports and/or supplementary materials that can be accessed by other researchers and practitioners.
Implications for Practice and Policy
Although no clear consensus regarding a “gold standard” approach for male-focused SV/DV/IPV prevention programs emerged from this review, the studies we reviewed present some promising approaches that can inform practitioners who are considering implementing prevention strategies in their communities and organizations. For example, RealConsent, the only program with significant positive effects on SV perpetration behavior with a broad group of men, was also the only program delivered both online and to individual participants (i.e., not in a group setting). Using technology such as online modules and social media platforms might be a promising path forward for SV/DV/IPV prevention programs to allow efficient individual implementation to a large audience. Furthermore, employing online and social media platforms in violence prevention might especially resonate with younger audiences in countries with high Internet usage. For example, an estimated 95% of U.S. adolescents aged 13–17 years report having “…access to a smartphone,” and 45% report being online nearly all the time (Anderson & Jiang, 2018, p. 2).
Moreover, this review, coupled with other research, begs the question of whether gender-specific perpetration prevention programming is needed. Prior research has found prevention programs for mixed-gender audiences to have positive effects on SV/DV/IPV perpetration behaviors (Coker et al., 2017; DeGue et al., 2014), whereas findings from this review demonstrate that, though effort has been placed on developing and evaluating male-focused programs over the past 20 years, little is known about whether such programs are actually effective in preventing SV/DV/IPV perpetration. Arguments in favor of male-focused programs acknowledge that (a) boys and men are more likely to perpetrate SV and severe forms of physical DV/IPV than girls and women (Archer, 2000; García-Moreno et al., 2013), (b) this gender disparity might be explained by the fact that there are gender-specific risk factors (e.g., hostile attitudes toward women; male rape myth acceptance; DeGue et al., 2014; Tharp et al., 2013) that lead to these types of perpetration behaviors, and (c) these gender-specific risk factors are best targeted by male-focused programs. A counterargument to this thinking is that programs for mixed-gender audiences can also effectively change these risk factors while working to change factors that make people of all genders vulnerable to victimization. Furthermore, male-focused programs, which are often heteronormative, may leave some participants feeling stigmatized or excluded (e.g., participants who do not identify along the gender binary, male survivors of SV/DV/IPV perpetrated by men or women) when they cannot identify with the program content. SV/DV/IPV prevention as a field likely needs a mix of effective programs given the various forms of violence and the diverse audiences for prevention programs. As such, practitioners should have a clear and practical rationale for using a specific approach, whether it is gender-specific or mixed-gender or whether it is universal or selected.
For policy makers and funders, the results of this review show that considerable room exists for innovation and advancement in the prevention of male-perpetrated SV/DV/IPV. Given the clear need for additional evaluation of SV/DV/IPV prevention programs, funding and policy should encourage program innovation. In addition, we urge funders to provide sufficient resources for rigorous evaluation of prevention interventions (i.e., support for large samples, follow-up data collection, and outreach to diverse communities).
Strengths and Limitations
Although our team used rigorous methods for this review, this research has some limitations. This review represents a comprehensive search and synthesis, including review of both peer-reviewed and gray literature. However, our searches might have failed to identify all studies with randomized designs that could have informed this review. Additionally, despite our mindful use of multiple strategies to enhance the rigor of our review methodology (e.g., independent coders, multiple search strategies), we might have missed or misinterpreted some details presented in the reviewed articles. Last, our assessment of risk for bias was based on information presented in the reviewed articles. Our team did not have access to original study protocols or any other information. However, to enhance the assessment process, we used a standardized tool and two independent coders in this bias assessment.
Conclusion
This review identified only nine studies that used randomized designs to evaluate male-focused programs designed for SV/DV/IPV perpetration prevention. This small body of research showed substantial heterogeneity across intervention approaches, study design, measurement of perpetration outcomes, and findings regarding intervention effectiveness. Therefore, no firm conclusions can yet be drawn regarding “what works” in SV/DV/IPV perpetration prevention programs designed for boys and men.
Implications for Research
This review highlights the need for future SV/DV/IPV perpetration prevention research to Use experimental designs when possible Assess prevention program effects on perpetration behaviors Focus on young boys and adolescents whenever possible Address potential sources of bias in study findings to enhance the rigor of evaluations of prevention programs Determine which prevention programs should be offered to boys and men universally and which should be targeted to specific subgroups Determine the key program elements necessary for effective prevention (i.e., program activities, content, delivery, length, and strategies)
Implications Practice and Policy
Although the small body of literature available for this review did little to clarify best practices for SV/DV/IPV prevention programs designed for boys and men, we provide initial recommendations for prevention practice and policy: Practitioners considering implementing prevention strategies in their own communities and organization will likely benefit from reviewing the studies identified in this review for guidance on promising SV/DV/IPV prevention programs; in particular, we recommend that practitioners review RealConsent (Salazar et al., 2014), Coaching Boys Into Men (Miller et al., 2013), The Men’s Program (Foubert et al., 2007), and The Men’s Project (Gidycz et al., 2011) SV/DV/IPV prevention as a field likely needs a mix of effective prevention programs given the various forms of violence and the diverse audiences for violence prevention programs. Thus, practitioners should have a clear and practical rationale for using a specific prevention approach, whether it is gender-specific or mixed-gender or whether it is universal or selected Given the prevalence of SV/DV/IPV in U.S. society and the lack of effective prevention programs, more research in this area is urgently needed. Therefore, policy makers should encourage and fund evaluation of SV/DV/IPV programs designed for boys and men as a public health issue
Footnotes
Authors’ Note
Points of view in this document are those of the authors and do not necessarily represent the official position or policies of the Centers for Disease Control and Prevention.
Acknowledgments
We also thank the Caroline H. and Thomas S. Royster Fellowship from the University of North Carolina at Chapel Hill which supported, in part, Laurie Graham’s time and effort for this research.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The National Center for Injury Prevention and Control of the Centers for Disease Control and Prevention Grant # 5U01CE002842-02 supported this research.
