Editor's Note: This guest blog was originally published by Together for Girls, AidData's partner in this project, at https://www.togetherforgirls.org/building-the-evidence-base-aiddatas-deep-dive-analysis-of-zambian-school-related-gender-based-violence/. Together for Girls is a global partnership working to end violence against children and adolescents, particularly sexual violence against girls.
What drives violence?
High-quality, disaggregated data on school-related gender-based violence (SRGBV) is essential to help drive effective policies and programs for prevention and response. Unfortunately, there is a dearth of data available.
At Together for Girls, we believe that governments and stakeholders should support efforts to collect high-quality, disaggregated data on violence against children and women. This data will help us fill the gaps related to our understanding of violence against children and how it can vary, such as by age, sexual orientation, gender, type of violence, perpetrator, and geography.
Additionally, we are always seeking new ways of using the data we already have. This means doing deep-dive analyses with existing data, to find new insights and patterns, as well as doing more data triangulation to get a better understanding of the complex nature of violence against children.
The impact of school-related gender-based violence
School-related gender-based violence (SRGBV) affects millions of children, families and communities. This violence has wide-ranging consequences for children’s physical and emotional well-being, school performance and attendance, and the likelihood of experiencing or perpetrating future violence. Furthermore, SRGBV can impact educational outcomes, with many students avoiding school, achieving below their potential, or dropping out completely as a result of such violence.
Mining the data
The need for deep-dive analysis is why we were so thrilled to partner with AidData, a research lab at William & Mary. In collaboration with Together for Girls and the U.S. Centers for Disease Control and Prevention (CDC), and with support from USAID’S Higher Education Solutions Network, AidData connected high-quality data from the Violence Against Children and Youth Surveys (VACS) to subnational, georeferenced datasets and applied advanced statistical and geospatial techniques to investigate SRGBV in three countries: Côte d’Ivoire, Kenya, and Zambia. For the Zambian analyses, AidData assessed the causal relationship between access to schooling and violence against children – an innovative use of the VACS and school construction data.
Why focus on SRGBV, and what makes the relationship between schooling and childhood and adolescent experiences of gender-based violence difficult to analyze?
The relationship between schooling and childhood and adolescent experiences of gender-based violence is both important to understand and also difficult to analyze. Violent experiences adversely affect girls’ psychological and physical health in a way that likely directly impacts both whether they attend school and also how well they perform in school when attending.
In addition, the availability of schooling for girls may influence investment in girls’ education and impact gender norms in a way that is reflected in the incidence of gender-based violence.
We know you used the 2014 Zambia Violence Against Children Survey (VACS), and that through collaboration with the Zambian government you were able to access confidential geocoded survey respondent locations at the enumeration area level. What other data sources did your team use?
The data sources used throughout this study consist of the VACS data, school construction data, and geoBoundaries Global Database of Political Administrative Boundaries, which is an online, open-source resource of administrative boundaries for every country in the world. Zambia underwent a nationwide school construction expansion program in the late 1990s and early 2000s, representing an increase in school availability for many of the respondents in the VACS survey.
We were able to construct the total number of schools in each district for each year in this time period, based on a dataset of historical school construction, “Bride Price and Female Education,” by Nava Ashraf, Natalie Bau, Nathan Nunn, and Alessandra Voena (2020).
Is there a way to briefly summarize your triangulation process?
The geocoded respondent locations in enumeration areas of the VACS survey were spatially merged to district names using the geoBoundaries district-level shapefiles for Zambia. These district names were then used to merge the VACS survey responses to the school construction data.
From the school construction data merged to districts, we counted the total number of schools constructed in each district per year. School availability was then measured by the total number of schools in a district per year divided by the physical area of the district.
We know that many other factors besides school availability, contribute to school-related gender-based violence. Can you explain how you were able to use school data as a proxy?
In the Zambia study, we are interested in assessing the causal relationship between access to schooling and violence against children. One of the biggest problems we face when considering this relationship is that access to schooling and experiences with violence may be correlated due to the influence of other variables, even if there is no underlying causal relationship between access and violence. For instance, an increase in household income may increase access to schooling and reduce exposure to violence independently.
The problem this causes is that we cannot observe the relationship between schooling access and experiences with violence independent of other factors, and we therefore cannot determine if there is a causal relationship. To solve this problem, we must find a proxy for access to schooling that is not related to the individual characteristics that affect experiences with violence.
We choose the availability of schools in an individual’s district, as this is determined at a district level and is not affected by individual-level characteristics. This allows us to be sure that the relationship between access to schooling and violence we observe is not driven by individual characteristics we cannot account for in our regression.
Although on the surface it may appear that using a proxy weakens the analysis, it actually strengthens the analysis because it allows us to determine the underlying relationship between access to schooling and violence without muddying the analysis with other correlated variables.
What kind of impact do you hope this research will have?
Ultimately, we hope this research informs the way we think about and implement programs targeted at increasing educational attainment and decreasing SRGBV. We know that children throughout the world experience SRGBV and that these experiences influence their performance in school.
In addition, attending school may serve as either a protective factor against certain forms of violence or exacerbate other school-related violence. By utilizing Zambia as a case study, our research highlights that programs and policies should tackle both of these simultaneously because SRGBV and school performance are intertwined.
We hope that as data become available for additional countries, this type of analysis will be replicated to continue unpacking the complex relationship between GBV experiences and educational attainment.
About the project
Together for Girls, AidData (a research lab at William & Mary), and the U.S. Centers for Disease Control and Prevention (CDC), with support from USAID’S Higher Education Solutions Network and Global Affairs Canada, conducted secondary analyses of national Violence Against Children and Youth Surveys (VACS) to better understand the experiences of children and youth who attended school with respect to violence in and around school settings, particularly school-related gender-based violence (SRGBV).
Learn more about our work to address the SRGBV data gap by visiting togetherforgirls.org/schools.