We find that outcomes improve in the entire study area, across both intervention and control villages, over the study period – likely reflecting broader social trends and positive impacts from the holistic Tushinde program. Between the two survey waves, we find a drop in IPV, non-Partner GBV, IPB and GBV perpetration as well as increases in many of our mental health indicators including depression, anxiety and PTSD. There is a 14-percentage point decrease of women reported having experienced IPV and a 9-percentage point decrease in non-partner GBV. These changes are impressive, as they occurred despite the global COVID-19 pandemic. When comparing villages assigned to CBTH programming and control, we see some promising changes. Non-partner sexual violence was 50% lower in villages assigned to CBTH programming compared to those assigned to control villages (a drop in reported incidence from 5% to 2.6%). This study is among the first interventions to show an effective reduction in the experience of non-partner sexual violence among female participants of an intervention. Self-efficacy and community resilience also improved in villages assigned to CBTH programming. While the differences are small, these findings show some promise, and indicate that CBTH has helped improve the outcomes it aimed to target—feelings of agency and ability to improve one’s own life and a greater sense of connectedness with, and trust in, one’s community as a whole. However, these results should be seen as suggestive and not conclusive evidence that CBTH improved GBV related outcomes. When we adjust our analysis for multiple hypothesis testing (a statistical procedure to correct p-values), none of the reported results are significantly different from zero. Other mental health outcomes also moved in the right direction, but are not significantly different between villages assigned to CBTH programming and those assigned to control (even before correcting for multiple hypothesis tests). One explanation for finding over-time changes in all villages but no experimental effect of CBTH is that spillover between treatment and control villages tampered the differences between CBTH and control villages. Indeed, nearly identical proportions of respondents across treatment and control villages expressed awareness of and exposure to CBTH programming in our survey. Combined with the positive overall progression in GBV outcomes between baseline and endline, this suggests that the program may have had an overall positive effect on both types of villages.
