Past Political Violence and Interpersonal Violence Against Children and Youth: Micro-Level Evidence from Africa - Stata Code and Replication Scripts, 2025

Fadare, Olusegun (2026). Past Political Violence and Interpersonal Violence Against Children and Youth: Micro-Level Evidence from Africa - Stata Code and Replication Scripts, 2025. [Data Collection]. Colchester, Essex: UK Data Service. 10.5255/UKDA-SN-858298

There is a more recent version of this item available.

Violence against children (VAC) is a global challenge. The prevalence of VAC is particularly acute in low-income settings, and more so in regions affected by different forms of violent conflict (VC). The picture emerging from existing evidence suggests different types of violent conflict heighten children's exposure to violence, which can be emotional, physical or mental violence. However, the specific mechanisms by which different types of violent conflict lead to different forms of violence against children are still not fully understood - and the awareness about this possible causal link from violent conflict to VAC is also not very well developed. Our research will examine, for the first time, the prevalence and the mechanisms from VC to VAC, using high quality secondary data available for ten countries, seven in sub-Saharan Africa and three countries in Latin America. We will investigate empirically different causal pathways from violent conflict to VAC looking at idiosyncrasies of different conflicts. Our aim is threefold. First, to perform an in-depth study of the prevalence of different forms of VAC across various VC VAC to occur when institutions break down. Second, to test if and how VC impacts on VAC in both the short-term and the long-term (across generations), both across various forms of VC and various types of VAC. Our strategy to identify short-term causality from VC to VACs builds on the observation that violent conflict episodes are exogenous and preceding the episodes of violence against children reported in the survey data. For the estimation of long-term effects, we exploit data on exposure to violence of children's parents to test how their childhood experience of VC may determine how they treat their own children. Third, we aim to identify which policy options may break the causal link between VC and VAC by modelling and simulating the additional 'surplus' effect of VC on VAC based on the short- and long-term estimates of VC on VAC. The simulation of different scenarios under which VC causes VAC will inform prevention and intervention policies to reduce the trigger mechanisms that exacerbate VAC in violent settings. Our research represents an example of how existing knowledge gaps on topics and contexts difficult to access can be addressed innovatively by merging existing datasets. We use the Violence Against Children Surveys (VACS), a series of population-based surveys of girls and boys aged 13-24 years which are nationally representative and designed to capture the magnitude and prevalence of violence against children in each country. We will also use conflict event data like the Armed Conflict Location Events Dataset (ACLED) which has (nearly) universal coverage of conflict event data. We will merge the two types of datasets through geo-spatially reference codes disaggregated temporally and geographically in both datasets. This will create a new international, multi-year dataset at the individual level. We will create an Impact Panel engaging with multiple stakeholders working on VAC. The Impact Panel will include senior representatives from the Armed Conflict Location and Event Data Project (ACLED), Together for Girls (TfG), the partnership that coordinates and publishes the Violence Against Children Surveys (VACS), the U.S. Centers for Disease Control and Prevention (CDC) providing technical support to the VACS, the Sexual Violence Research Initiative (SVRI), the world's largest network on research on violence against women and violence against children, the Child and Adolescent Rights and Empowerment team of the UNICEF Office of Research - Innocenti, one of the most influential policy research hubs on violence against children and women, and the NGO Amnesty International, who advocate for children's rights.

Data description (abstract)

This collection comprises the replication package (code and syntax) for the paper: "Past Political Violence and Interpersonal Violence Against Children and Youth: Micro-Level Evidence from Africa."

Contents: The deposit includes the Stata code (.do files) required to reconstruct the datasets and reproduce every table and figure in the published article. The code is organised into two tiers:

Data Preparation (cleaning and merging VACS and ACLED data).

Analysis and Output (generating figures and tables).

Data Access Note: Due to third-party license restrictions, the raw data from the Violence Against Children and Youth Surveys (VACS) and the Armed Conflict Location and Event Data Project (ACLED) are not included in this deposit. Users must obtain the raw data directly from the original providers (instructions and links are provided in the included README file).

Once raw data is obtained, the supplied 0_master.do script will automatically merge the files and generate the final analysis (Analysis_Data.dta).

Software: The code was developed and tested using Stata 18 MP.

Data creators:

Creator Name	Affiliation	ORCID (as URL)
Fadare Olusegun	University of Greenwich	https://orcid.org/0000-0002-0647-7438

Sponsors:

Economic and Social Research Council

Grant reference:

ES/X00192X/1

Topic classification:

Social welfare policy and systems
Politics
Health
Economics

Keywords:

CHILD ABUSE, DOMESTIC VIOLENCE, POLITICS, HEALTH, SOCIAL WELFARE, ECONOMICS

Project title:

Causal pathways from violent conflict to violence against children: evidence from multi-country secondary data

Grant holders:

Natural Resources Institute, University of Greenwich

Project dates:

From	To
1 May 2023	1 November 2025

Date published:

03 Feb 2026 15:23

Last modified:

16 Feb 2026 13:29

Coverage and Methodology

Temporal coverage:

From	To
1 May 1998	31 August 2018

Collection period:

Date from:	Date to:
1 September 2013	30 September 2019

Geographical area:

Africa

Country:

Côte d’Ivoire, Kenya, Malawi, Mozambique, Namibia, Nigeria, Uganda, Zambia, Zimbabwe

Spatial unit:

Administrative > Council Areas
Administrative > Counties (Metropolitan)
Administrative > Districts (Metropolitan)
Administrative > Regions

Data collection method:

This research integrates secondary data from two primary sources:

Violence Against Children and Youth Surveys (VACS): Nationally representative household surveys focusing on adolescents and youth aged 13–24 years. These are conducted by national governments with technical assistance from the US Centers for Disease Control and Prevention (CDC) and the Together for Girls partnership.

ACLED: Political violence event data aggregated from over 5,000 public sources.

Sampling Strategy The VACS utilize a three-stage cluster split-sample design, selecting one respondent per household. Uniquely, male and female interviews are assigned to different enumeration areas to ensure boys and girls are never interviewed in the same geographical locations. The final sample comprises 35,439 individuals across nine sub-Saharan African countries (Côte d’Ivoire, Kenya, Malawi, Mozambique, Namibia, Nigeria, Uganda, Zambia, and Zimbabwe), representing approximately 30% of Africa’s population.

Weighting and Bias Correction While VACS aim for national representation, some country samples are skewed toward females (e.g., 90% female in Zimbabwe, 80% in Namibia). To address this, the analysis utilizes the official VACS weights, which adjust for split-sample probabilities, differential non-response, and gender imbalances.

Observation unit:

Individual

Kind of data:

Numeric, Geospatial, Software, Other

Type of data:

Cross-national survey data, Geospatial data

Resource language:

English

Access and Administration

Data sourcing, processing and preparation:

VACS: Raw survey data was obtained via formal request from the Together for Girls partnership/CDC. ACLED: Conflict event data was downloaded from the Armed Conflict Location & Event Data Project (open access). Population Data: Administrative area population estimates were extracted from citypopulation.de. The analysis uses Stata 18 to merge the individual-level VACS data with ACLED conflict data based on the lowest common administrative unit (e.g., District, Region, or County). Geospatial Matching: The merge covers 723 sub-national administrative units across the nine countries. Temporal Aggregation: For every administrative unit, the code aggregates ACLED political violence events for the 15-year period immediately preceding the specific VACS data collection date for that country. Cleaning and Variable Construction The Stata code performs several critical cleaning steps: Skip Logic Correction: For Malawi, Nigeria, and Zambia, the code corrects survey skip logic discrepancies where "last 12-month" questions were omitted if the "ever experienced" question was negative. These are systematically recoded to "No" to ensure consistency with other countries. Construction of VAC Indicators: Composite variables are generated for Sexual Violence (SV), Emotional Violence (EV), and Physical Violence (PV) by aggregating responses regarding violence from partners, family, peers, and community members. Construction of Political Violence Measures: The merging process involved standardizing administrative unit names across datasets to ensure accurate matching. The code then computes the political violence rate per 100,000 population for each unit and standardizes these values (Z-scores) to capture relative conflict intensity.

Rights owners:

Name	Affiliation	ORCID (as URL)
Iversen Vegard	University of Greenwich	https://orcid.org/0000-0002-0425-2286
Brück Tilman	Humboldt-University of Berlin	https://orcid.org/0000-0002-8344-8948