This repository provides a comprehensive analysis of major violent incidents directed against humanitarian aid workers. The analysis is based on data colelcted by the Aid Worker Security Database (AWSD), available at https://aidworkersecurity.org/. The dataset contains more than 2000 incidents that have been reported since 1997 until today. This quantitative research aims to shed light on the crucial issue of the security of civilian aid workers in fragile contexts. Because high insecurity for aid workers may significantly reduce access to a needy population and limit both the amount and quality of aid provided. assdad

Key Outcome:

• Risks to humanitarian aid workers are power law distributed. A small number of insecure locations (Afghanistan, Sudan, Somalia, Syria and Pakistan) account for most of the incidents.

• Over the last fifteen years the number of serious incidents has continously increased and is now more than 4 times higher than 15 years ago.

• The increase of attacks does primarily affected national humanitarian aid workers!

• The types of violence directed at aid workers is dependent on the regional contexts. Looking into data from 2015 shows that in Afghanistan kidnapping is the prevalent mean of attack. However, in Sudan and Somalia, in contrast, shootings and (sexual) assualt are the most signifiant type of violence. Incidents caused by explosives and heavy weapons are with more than 50% only the majority of events in Syria.

For each incident, the database records the:

• Date
• Country and specific location, including geocodes
• Number of aid workers affected (victims)
• Sex of victims
• Institutional affiliation of victims (UN/Red Cross/NGO/other)
• Type of staff (national or international)*
• Outcome of the incident (victims killed/wounded/kidnapped)
• Means of violence (e.g. shooting, IED, aerial bombardment)
• Context of attack (ambush, armed incursion, etc.)
• Summary of incident (public details)

Link to web display of iPython Notebook

The next steps of this analysis will be to create a spatial regression model to understand the relationship between local context, incidents and consequences on aid provision (such as suspensions of programmes and withdrawals).

The MIT License (MIT) Copyright (c) 2016 Philipp Schwarz