Gender-based violence (GBV) remains a pervasive threat to adolescent girls worldwide. In schools, the very spaces meant to foster learning can become sites of fear, silence, and harm—often at the hands of those in authority. Yet, rigorous evidence on how to effectively reduce GBV in educational settings remains scarce or inconclusive.
A recent working paper by Amaral, Garcia-Ramos, Gulesci, Oré, Ramos, and Sviatschi breaks important ground on this front. The study evaluates an intervention in Mozambique designed to both deter violence by school personnel and empower students to recognize and report abuse. It is one of the few large-scale experimental studies that tackles school-based GBV head-on while credibly estimating both the impact of the intervention and the mechanisms through which change happens.
In this blog, I want to highlight two key innovations in this study. First, the paper leverages a clever experimental design that distinguishes between top-down and bottom-up channels of change, allowing the authors to unpack how and why the intervention worked. Second, it devotes serious attention to the challenges of measuring GBV, particularly in the face of potential reporting bias, stigma, and fear of retaliation, providing valuable lessons for future research in sensitive topics.
A Dual Approach: Changing Behavior from the Top and Bottom
The intervention took place in 326 primary schools across Sofala Province in Mozambique, where GBV by teachers and other school staff is distressingly common. The authors argue that in contexts marked by asymmetrical power dynamics, such as relationships between teachers and students, change must happen on both sides of the equation. Considering this, and in collaboration with the Ministry of Education and UNICEF, the researchers implemented a two-pronged program called Está na Hora de Agir (“It’s Time to Act”). The first component, a top-down approach, involved training teachers previously designated as Gender Focal Points (GFPs) to recognize, prevent, and respond to GBV. These GFPs were trained on GBV laws, appropriate response protocols, and the use of a national helpline through which formal complaints could be lodged. The second component, a bottom-up approach, provided a structured curriculum for students that included short videos, group discussions, and reflection exercises focused on recognizing and responding to GBV.
One feature of the study’s design I really like is its cross-randomization of the student training. While all treated schools received the GFP training, the student training was delivered to different groups across three experimental arms: only girls in some schools, only boys in others, and both girls and boys in a third group. This allowed the authors to disentangle the relative contributions of deterring perpetrators through teacher training versus empowering students, particularly girls, to act as informed victims or bystanders.
The results are compelling. One year after the intervention, girls in treated schools were 67% less likely to report having experienced violence from teachers or staff in the last month. Importantly, this effect was consistent across all treatment arms, regardless of whether students received the training, pointing to the pivotal role of the GFP training. GFPs in treated schools were substantially more active in discussing GBV with students and colleagues, reporting suspected cases to school authorities, and referring students to formal complaint mechanisms.
However, the intervention only translated into improvements in educational outcomes when girls themselves received the student training. In schools where only girls were trained, girls were 10% more likely to be enrolled in school at follow-up, compared to those in control schools. The researchers discuss that these gains probably stemmed from increased reporting and greater engagement with school authorities among girls. Furthermore, teachers in these schools also demonstrated greater knowledge of GBV laws and perceived sanctions, suggesting that victim reporting helped activate accountability mechanisms.
Measuring GBV: Getting Beyond the Biases
Another feature of the study I think is very novel is its thoughtful and multi-layered approach to measuring GBV, an area where standard survey methods often fall short. GBV is notoriously difficult to measure due to stigma, fear of retaliation, and social desirability bias. The authors anticipated these challenges and designed the data collection process in a way that they can show that in contexts where underreporting is a serious concern, it is possible to collect credible, meaningful data with careful design and methodological triangulation.
To begin with, they used a combination of self-reports and third-party reports to capture the prevalence of violence. Students were asked not only whether they themselves had experienced different forms of GBV, but also whether they had witnessed or heard of such violence happening to other girls in their class. This “indirect” measure offers an important check on the direct reports, especially for respondents who may be reluctant to disclose personal experiences. Remarkably, both self-reported and bystander-reported data pointed to large reductions in violence by teachers in treated schools, lending credibility to the findings.
In addition, the authors collected baseline data on social desirability tendencies using a well-established scale. This allowed them to test whether students with a stronger inclination to please others responded differently to the intervention. They found no significant heterogeneity in treatment effects by social desirability score, suggesting that the reductions in reported GBV were not driven by attempts to respond in a socially acceptable way.
They also took a creative step in assessing whether the intervention affected students’ willingness to disclose past experiences of violence. At both baseline and endline, students were asked to retrospectively report any incidents that had occurred prior to the start of the study. If the intervention had merely made students more or less likely to report violence, one would expect to see changes in these retrospective reports. But the data showed no such shifts, reinforcing the idea that the main findings reflect real behavioral changes, not just changes in reporting norms.
Further, the study validated its survey findings using administrative data from Linha Fala Criança, the national helpline where GBV cases can be formally reported. The authors found a sharp increase in helpline calls in districts with treated schools after the intervention began, particularly calls that were escalated to formal investigations. This external data source offers a rare and powerful validation of survey-based evidence in a sensitive domain.
The policy implications are significant. The study shows that gender-based violence in schools is not an intractable problem. It can be prevented at relatively low cost by building the capacity of trusted school personnel and empowering girls to report abuse. The intervention cost just under $20 per student, with marginal costs around $10 per student. In a country like Mozambique, where the risks of early dropout, pregnancy, and child marriage are high, reducing GBV in schools may be one of the most cost-effective ways to improve long-term outcomes for girls.
Join the Conversation