Access the Dataset at: armedgroupdataset.org
Principal Component Analysis of Dataset
Conventional wisdom might suggest there are certain `types’ of armed groups. I explore whether there are different types of groups using Principal Component Analysis of the dataset.
The PCA results suggest there are some features – like ideology – are easy to help discriminate between groups. Other features – like insurgency status – are not useful to discriminate between the groups.
The dataset primarily draws on the population introduced by the Terrorist Organization (TORG) Crosswalk Project, hosted by the Project for Violent Conflict at the University of Albany.
The full list of armed groups is available for download here.
The dataset is compatible for merger with existing conflict and insurgency datasets, like the Uppsala/PRIO Armed Conflict Dataset, the Global Terrorism Database, and the Ethnic Power Relations dataset . It also sources new information from archival sources, declassified documents, and online resources.
For each of the different armed groups in the dataset, a 1-2 page historical narrative is written about the group. This set of narratives offers scholars, policy-makers, and the community an opportunity to learn more about terrorism and insurgency. We plan to release the full set of narratives in the future.
The narratives center around five different topics:
- Group Formation: Information about the group’s founding date, its initial goals, ideology, and date when it first came to attention as a violent group.
- Geography: Information about the group’s operational environment including the names of areas where they operate from, the name of any external sanctuaries, the name of any cities, towns, or neighborhoods where they conduct attacks.
- Organizational Structure: Information about the group’s organizational structure including its leadership, membership, source of funding, and different wings.
- External Ties: Information about the group’s ties to other actors including both state and non-state entities.
- Group Outcome: Information about the state’s response to the group, if any, and how this affects the group. This also identifies whether the group is still active, when it stopped using violence, and what happened to the group to cause it to stop using violence.
For each of the different armed groups in the dataset, a 1-2 page historical narrative is written about the group. This set of narratives offers scholars, policy-makers, and the community an opportunity to learn more about terrorism and insurgency. A sample set of narratives can be found here. We plan to release the full set of narratives in the future.
The dataset includes the following variables:
- Politicized Opposition to State
- Sustained Violence
- Electoral Violence
- Date of Formation
- Date of First Violent Attack
- Campaign Duration
- Type of Formation
- Political Aim
- External Base
- Transnational Attacks
- Minimum Distance to Border
- Ethnic Membership
- Ethnic Group
- Social Membership Base
- Political Wing
- State Support
- Type of State Support: Material, Funding, Territory, Training, Endorsement, Other
- Non-State Support
- Type of Non-State Support: Material, Funding, Training, Endorsement, Other
- Date of Last Known Violent Attack
- Reason for Ending Violence
The dataset acknowledges financial support, in part, from the Stanford University Department of Political Science, the Tobin Project, and a research grant from Milt Lauenstein.
Research assistance, in part, comes from the following individuals: Lauren Black, Alexa Corse, Sarah Gao, John Ghantous, Katherine Irajpanah, Amisha Kapur, Tanya Menezes, Michael Rover, Cristina Serrano, Siya Sharma, Ellie Stanton, Lauren Tankeh, and Rahul Verma.