Ulric B. and Evelyn L. Bray Social Sciences Seminar

Abstract: The last decade has witnessed a surge of interest in engaging men in programming to reduce intimate partner violence (IPV). However, evidence about what works to both enroll men in IPV programs and succeed in shifting their behavior is limited. These are two distinct problems, both confounded by the fact that IPV is a crime with potential legal and social consequences for men who reveal a propensity for violence. First, how do we identify and encourage participation among men most in need of IPV prevention, and second, how do we structure these programs to best encourage behavior change so that they are not only well targeted but effective?

In this paper, we experimentally evaluate a novel approach to IPV prevention that harnesses social media to recruit and engage men in a group-based IPV prevention program delivered as a virtual support group (VSG) by trained male facilitators exclusively via WhatsApp. The program succeeded in recruiting men at high risk of committing IPV through self-targeting alone: 52% of partners of men who enroll in the program in response to social media ads report experiencing IPV at baseline, more than four times the national average from household survey data and nearly twice the rates observed in men recruited through targeted and untargeted invitations. Moreover, on average, participation in the program reduced the probability that female partners report sexual violence at endline by 20%. Among men who exhibit violence at baseline, the rate is 27% and among those whose wives report that they do not drink alcohol, treatment is associated with a 40% reduction in sexual IPV. These results demonstrate that targeting men through a highly scalable and cost-effective behavior change program can reduce IPV. To date, few IPV programs have proven to be effective, and the vast majority have been targeted to women. While the importance of targeting men is frequently mentioned, there is to date little evidence that this strategy works.

In addition, program effects on violence are highly sensitive to group composition, which was randomly assigned. In particular, men assigned to groups whose members have a relatively high propensity for violence – measured either in terms of incidence of sexual IPV or an index of IPV justification at baseline – experience significantly larger decreases in sexual violence at endline. Moreover, sensitivity to group composition is observed only among men prone to violence. In other words, placing men at high risk of violence among like-minded individuals has a significantly larger impact on their behavior than placing them in a group of more progressively minded men.

By analyzing the message content of group chat transcripts, we establish two key mechanisms through which group homogeneity activates positive and attenuates negative peer effects. First, while the overall amount of chat activity is not affected by group composition, the nature of interactions is: groups with higher concentrations of men who justify violence experience significantly more positive group interactions, in which members reveal personal relationship struggles or facilitators or other group members offer problem solving advice, and significantly fewer negative interactions, in which group members argue or insult one another. Possibly as a result, high-risk men are also more likely to exit the chat group midway through the program as the fraction of non-violent group members increases. These findings provide novel evidence of both positive and negative peer effects in behavior change programs, with implications for the optimal composition of group interventions. Specifically, segregating individuals based on baseline risk appears to magnify program impacts on high-risk individuals – or those with the most to gain from the intervention -, and hence the program impact overall.

Joint work with Christopher Boyer, Rachel Lehrer, Andrew Morrison, and Claudia Piras.