I’m not generally a big fan of randomised control trials (oversold, squeeze out other forms of knowledge – more here), but a recent RCT on violence against women in Uganda by researchers at Columbia University got my attention. Here are some excerpts from the summary on the website of Innovations for Poverty Action (IPA).
First the summary of the summary:
‘In Uganda, IPA worked with researchers to evaluate whether videos encouraging communities to speak out about and counter violence against women (VAW) in the household could change behavior, attitudes and norms related to VAW.
In surveys conducted eight months after the intervention, the proportion of women who reported any VAW in their household over the preceding six months was substantially lower in villages where the videos were screened than in villages randomly assigned to the comparison group.
The impact appears to be driven by a reduction in the perception that those who speak out against violence will face social sanctions.’
What seems to be happening here is a shift in social norms, in the categories described by Cristina Bicchieri in Norms in the Wild (see diagnostic). Progress occurs not by changing men and women’s internal beliefs, but by changing what they think other people in their communities consider (un)acceptable. Back to IPA for more detail:
‘Video halls, known as bibanda, are ubiquitous in rural Uganda. They typically hold 10-50 people and are located in the center of a village or trading center. Since few households in rural Uganda own TVs, bibanda are popular places to watch movies and soccer, especially among young men.
Nationally representative opinion polls suggest that some forms of VAW are widely viewed as legitimate in Uganda. However, not all violence is condoned. While 31 percent of respondents in our study said a husband is justified in beating his wife when she disobeys him, only 2 percent would condone violence perceived as more severe than slapping. Eighty-eight percent stated that others should intervene to stop violence.
Nonetheless, almost a third of women reported in 2011 that they had experienced violence such as being punched or threatened with a knife. Communities do not seem able to prevent such violence, partly because witnesses do not speak out: just one quarter of respondents in our study said they would tell authorities if their cousin had been beaten by her husband. Only one in ten would report to police. A common justification for withholding information was fear of being branded a gossip.’
‘Three short anti-VAW videos were produced in collaboration with Peripheral Vision International. Ranging 4-8 minutes each, the videos depicted deadly violence by a husband towards his wife and appealed to viewers to speak out about VAW in order to prevent it from escalating.
Audiences saw the videos via film festivals held in bibanda throughout 112 rural villages. Every village in the study featured a film festival comprising six popular Hollywood films unrelated to VAW that were shown once a week over consecutive weekends from July through September 2016. In 48 randomly selected villages the three short video vignettes on VAW were inserted into the intermission of the Hollywood film. In the other 64 villages, the film festivals featured video vignettes on other social issues (teacher absenteeism or abortion-related stigma), or just the Hollywood films with no video vignettes at all. These 64 villages thus constitute a comparison group that received a “placebo” film festival unrelated to VAW.
As is typical in Uganda, the Hollywood films were narrated by a VJ who added his own commentary to the movie’s storyline. Unlike most entertainment screened in bibanda, however, the anti-VAW videos were produced in the local language (Luganda) using local actors, enabling villagers to identify with the characters in the videos.
Admission to the film festivals was free of charge to encourage the attendance of a broad-based audience, and was notably successful in attracting women (31 percent of all attendees). In total, over 10,000 adults attended 670 film screenings. [31% is pretty good as bibanda tend to be very male spaces – see video from PVI]
The research team interviewed 6,449 individuals across all villages through two waves of surveys conducted 2 and 8 months after the film festivals. Importantly, the surveys were presented as opinion polls unrelated to the video campaign. Questions measured behavior and attitudes among a random sample of the adults living in the catchment area of the video hall, irrespective of their attendance of the festival. Respondents therefore include not only those who attended the screenings, but also their neighbors, community leaders and village health workers.’
Results and Policy Lessons:
‘Impact on violent incidents: Eight months after the anti-VAW vignettes were shown, women in the treatment group were 5 percentage points less likely to indicate that a woman in their household experienced any violence over the previous six months relative to the comparison group. Around 20 percent of women respondents in the comparison group reported that there was at least one case of VAW in their household over the previous six months. The campaign thus reduced the proportion of women respondents who reported any violence in their household by approximately one quarter.
Impact on social norms and attitudes: Among men and women who watched the videos, there is little evidence the anti-VAW videos had an effect on attitudes about the legitimacy of VAW or on perceptions of whether others in the community see VAW as legitimate behavior. Nor is there statistically significant evidence in favor of an increased empathy for VAW victims or a change in their perceptions about whether initial acts of domestic violence are likely to escalate.
The most plausible causal channel for the reduction in VAW appears to be a change in the willingness of victims and bystanders to speak out about violence. Both men and women who watched the anti-VAW videos were more willing to report instances of VAW to village authorities. This effect was particularly pronounced among women: When asked whether they would report a hypothetical incident of VAW across a range of different scenarios, two months after the campaign women in the treatment group were 9 percentage points (22 percent) more likely than those in the comparison group to say they would report violence across all scenarios. Eight months after the campaign, this willingness remained higher in the treatment than the comparison by a margin of 13 percentage points (35 percent). This increased willingness to report VAW may be related to a concurrent change in the perceived social consequences of speaking out, particularly among women. Women who watched the anti-VAW videos were 11 percentage points (18 percent) less likely to believe they would face social repercussions, such as scolding for gossiping, for intervening in a VAW incident.’
Wow. Can watching a few public service ads in the interval really change people’s lives that much? I really hope so, but can’t help being sceptical given the message saturation we all experience. Maybe that saturation is less in rural Uganda, or the novelty of hearing videos in your own language cuts through the noise.