Political Instability Task Force Worldwide Atrocities Dataset
For the past few years we have been collecting the data set described below for the U.S. government-sponsored Political Instability Task Force. The current version is in three files: first two cover January 1995 to December 2012 (7,775 events) and January 2013 to December 2015 (4000 events). The most recent file covers January 2016 to the most recent collected data, which is usually posted around the 15th of the month following the end of the data.
These data were coded using conventional human coding rather than automated methods, although we do employ a couple of customized filter programs to reduce the number of stories we need to look through. For the past four years we have been using external NGO sources for data on civilian casualties in Syria. The codebook contains extensive detail on the contents of the various data fields and points of ambiguity.
Data are manually geocoded, primarily using the geonames.org database, sometimes supplemented with maps provided by IGOs and other resources. Locations are resolved to either the city level or other levels readily available in geonames.org, for example local government areas in Nigeria, but village names are provided in text fields even if their coordinates cannot be determined. In cases where location information is vague, approximate coordinates are provided either based on provincial centroids or arbitrary locations based on the available information (e.g. "eastern DRC near the Ugandan border").
Note:There is a shift in January-2016 to a new source for the Syria data: this link provides information on this change.
Data are in zipped MS-Excel files; compressed 1995-2012 file size is about 3.5 Mb; it decompresses to about 12 Mb.
The following narrative describing the data is included at the request of the sponsor:
“The Political Instability Task Force (PITF) Worldwide Atrocities Dataset is a global dataset that describes, in quantitative terms, the deliberate killing of non-combatant civilians in the context of a wider political conflict. This data collection project, which is still ongoing, is intended to advance efforts to understand and anticipate atrocities, i.e., the deliberate use of lethal violence against non-combatant civilians by actors engaged in a wider political or military conflict. The practical objective of this project is to create a dataset representing a reasonably systematic sample of atrocities occurring worldwide in recent decades in order to: (1) enable the development of statistical models that might be used to identify countries vulnerable to the occurrence of atrocities or, if atrocities are already occurring, to an escalation in their rate or intensity; and (2) create a descriptive record that might be used by researchers with an interest in particular countries or conflicts. The effective date of data in this dataset is 1 January 1995 to the present date. Data are updated monthly.
“This data-collection is sponsored by the Political Instability Task Force (PITF). The PITF is funded by the Central Intelligence Agency. The data set is the responsibility of the authors’ alone and does not represent the views of the US Government.”
Political Instability Task Force Worldwide Atrocities Dataset coding manual (.pdf) [Version 1.1B1: extensively updated]
Note: There is currently an “issue” in the Excel files wherein some of the ranges for the deaths and injuries are converted to dates. We've tried several work-arounds for this, thus far unsuccessfully. Sorry.
LAST UPDATED: 6 March 2019