Published on Development Impact


This page in:

This post is coauthored with Manisha Shah and Jennifer Seager

According to the WHO, about 1 in 3 women will experience intimate partner violence (IPV) in their lifetimes. In a recent paper, we look at interventions that try to reduce this problem by working on both sides of the gender divide.

The setting is Tanzania, where we partnered with a group of NGOs. The target population was females in adolescent empowerment clubs run by BRAC. For those girls, facilitators worked with them to set goals around sexual and reproductive health. The idea was to set some behavioral strategies that would enable them to be healthy and free of sexually transmitted diseases over the next year. 

We also worked with males. Here another NGO, Grassroots Soccer ran a program that integrated messages around risky behavior, HIV prevention and intimate partner violence with soccer practice sessions. For example, in one session “Red Card” boys discuss scenarios worthy of a red card including things like older partners pressuring younger women to have sex and gender-based violence. This intervention was targeted at boys who were in the girls’ social and sexual network, although a significant number of other boys from the communities participated as well. In all, the curriculum comprised of 11 sessions in each community, with coaches hanging around after the sessions to talk with the boys one-on-one if they wanted to. 

There was a third arm that provided contraceptive supply, but it didn’t have any impacts due to low takeup, so we’re going to skip that here.

At the time we started the study, BRAC was working across 149 communities in Tanzania with teenage girls (average age was around 16). The soccer intervention was randomly assigned to 50 communities, while the goal setting intervention was randomized at the individual level (stratified by club) across all communities. 

Our analysis uses survey data. The baseline for girls happened toward the end of 2016.  One tweak to business as usual was to ask the girls about their social networks at baseline.  We then use the boys they named as the sample frame for the boys’ baseline.  We did an endline in 2018, 6 months after all interventions had wrapped up. 

So what happened? The goal setting intervention seems to have worked – on average, girls set 1.7 strategies and achieved 1.6 of these. In terms of top-level outcomes, we define two indices for IPV that combine a range of different forms (psychological and physical abuse and forced sex) across different frequencies: IPV often and IPV in last year (at all).  As a result of the goal setting intervention, girls experience less violence: IPV often drops by around 0.3 standard deviations (sd), and the IPV in last year index drops by about 0.2 sd (significant at 10 percent).  In terms of potential mechanisms, this effect is higher for the girls who set more than 1 strategy to get to their goal.  The overall effects are also markedly larger for girls who were sexually active at baseline (0.38 sd for IPV often and 0.44 sd for IPV in last year). 

Soccer also seems to have worked.  There is a big change in attitudes as boys in these communities are 17.4 percentage points less likely to say that a woman should tolerate violence from her partner (with a broader index backing up this result).    Nonetheless, there are a fair number of studies out there that document attitude change but not behavior change.  Not so in this case, especially when we use the girls reports of what is going on.  The IPV often index drops by 0.23 sd and the IPV in last year index drops by about 0.2 sd (significant at 10 percent).  Again, these effects are larger for those girls who were sexually active at baseline for the IPV often index, which drops by 0.48 sd (the IPV in last year index, while showing a larger coefficient, isn’t significant at conventional levels). 

What could be behind these results?  Starting with the soccer intervention, the boys’ attitudinal change towards violence is coupled with a significantly higher perception that peers have sexually transmitted infections. What could be going on in this case is that the boys might have been using violence to get risky sex (keep in mind that forced sex is part of the index). They now think that violence is less acceptable, so its cost has gone up. And they also perceive risky sex as riskier.  And indeed, girls in these communities show significantly lower sexual activity, driven mainly by a lower likelihood of currently having a partner. 

On the other side of the gender divide, the mechanisms are different. Females who were invited to goal setting are equally likely as control girls to have a partner, but they have had more partners in the recent past. Indeed, they are 3.9 percentage points (significant at the 10 percent level) less likely to be with the same partner as baseline. Moreover, their current partners are of better quality. We construct an index of partner education, whether she thinks he uses contraceptives, and his age as a measure of boyfriend quality.  Goal setting girls’ partners score 0.2 sd higher.

Thus, while the soccer intervention appears to have reduced IPV by reducing boys’ inclination to use violence against their partners, the goal setting intervention appears to have achieved the same outcome by empowering girls to leave violent partners (in order to pursue their goals).

These patterns are interesting and some of them are, perhaps, not what you might expect. The paper has a model which lays out the choices in a game theory framework that helps put both of these interventions into the same context – choosing risky sex or not in the context of partners who might choose violence to get it.

The set up of our experiment also allows us to see if there are interaction effects between these two interventions. In this context, we don’t see any indication of this. One reason could be because each intervention alone has pretty sizable effects. Take the case of forced sex, which about 5 percent of our control sample experienced in the last year.  The goal setting intervention alone reduces this pretty much to zero. Hence, in this case, there might not be room for any complementarities to manifest. 

So, both of these are effective interventions, working in different ways, on boys and girls. They also turn out to be fairly similar in cost. The per-female cost of the boys’ effects we observe is $41, and the goal treatment comes in at $38. So fairly cost-effective. 

All in all, these are intriguing results. By aiming at the intersection of reproductive health and IPV, different pathways to improved violence outcomes become evident when we target girls versus boys with different interventions. It will be interesting to see what happens as these adolescents move into adulthood.   


Markus Goldstein

Lead Economist, Africa Gender Innovation Lab and Chief Economists Office

Joao Montalvao

Development Economist

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000