������Ƶ

These databases reflect the aggregation of credible news reports about U.S. counterterrorism strikes in Pakistan, Yemen, Somalia, and Libya. The media outlets that ������Ƶ relies upon are the three major international wire services (Agence France Presse, Associated Press, Reuters), the leading regional newspapers (Dawn, Express Times, The News, Yemen Observer, Yemen Post), leading South Asian and Middle Eastern TV networks (Geo TV and Al Jazeera), and Western media outlets with extensive in-country reporting capabilities (BBC, CNN, the Guardian, Los Angeles Times, New York Times, Telegraph, Washington Post).

Other sources, including U.S. military press releases and non-governmental organization reports may be included where they shed light on the data. In the case of Libya, ������Ƶ and its project partner Airwars include reporting from international and local news agencies; non-governmental organizations; and social media sites that corroborate evidence, such as local residents’ groups, Facebook pages, YouTube footage of incidents, and tweets relating to specific events. ������Ƶ makes no independent claims about the veracity of casualty reports provided by these organizations.

������Ƶ would like to thank the Bureau of Investigative Journalism, the Stanford International Human Rights and Conflict Resolution Clinic, the Global Justice Clinic at NYU School of Law, and the Columbia University Law School for their valuable work on this subject, from which we have benefited and improved our own research.

• There must be at least two credible media sources verifying a strike in order for it to be added to our databases. In Yemen, where reporting on strikes is often more sparse, we include strikes with one major news outlet reporting that it occurred.
• While we will only add a strike if there are two credible media sources verifying it (acknowledging the caveat for Yemen), we endeavor to find at least four credible media sources per strike.
• Strikes that occur on the same day but in different locations or at different times are counted as separate strikes.
• Multiple drone strikes that occur in short succession (two hours or less) in one area are counted as one strike.
• In the case of Libya, ������Ƶ and Airwars track multiple international belligerents (countries that conduct air, drone, and artillery strikes) and state belligerents (GNA/LNA, for example). There are times when strikes are deemed contested because of multiple competing claims of responsibility (we list GNA/LNA as a contested category if there are competing claims for both, as an example). If we are unable to corroborate evidence to support one belligerent or another, we list the strike(s) as contested.

• The death toll for each strike is expressed as a range, created using the lowest and highest number of deaths reported, as the different media outlets upon which we rely for our data often report different casualty counts. Each strike has a total death toll (or range) and then specific sub-counts for “militants,” “civilians,” and “unknowns.” We endeavor to ensure that the total of the sub-count ranges equals the total death toll range. Thus the sum of the low sub-counts should equal the low total, and the same applies for the high sub-counts and high total.
  • This proved challenging in certain instances where different sources provide conflicting casualty ranges. In those instances, the difference between reported “civilians” and “militants” is usually reflected in the “unknown” category, or we have erred on the side of reporting more “civilians.”
• If two or more news reports label the dead as “militants,” while others call them “people” or some other neutral term, we have labeled them as militants.
• If two or more media outlets explicitly refer to the dead as “civilians,” “women,” or “children,” then we have listed them as civilians.
• If a majority of reports do not refer to the dead as “civilians,” “women,” or “children,” but one media outlet does, then we have labeled them “unknowns.”
• If the various media reports are so contradictory that we are not comfortable drawing a conclusion, then we label all of the dead as “unknowns.”
• If a media report cites “some” “civilians/women/children” but does not specify how many, and no other media sources provide a specific total, we have reported one third of the total victims referenced in that source as “civilians” or “unknowns.”
• In the case of Libya, ������Ƶ and Airwars have three categories: "civilians" (non-combatant totals that include men, women, and children taking no active part in hostilities), "combatants," and "unknowns."

• Officials and local villagers in Pakistan’s tribal regions often refer to militants from outside Pakistan simply as “foreigners.” When this term is used to describe drone victims, we have assumed that the dead are militants, unless the word “civilian” is used explicitly.
• Similarly, the term “local tribesman” does not preclude the possibility that a tribesman was also a local militant. Because news outlets tend to use the word “civilian” or “bystander” to refer to non-militant casualties, we have interpreted “tribesman” as a neutral term, like “people,” and not a claim that the casualties were “civilians” or that they were “militants.” When multiple news reports identify victims as “militants” and some reports identify them as ‘tribesmen,” we identify them as “militants.”

• There is a wide range between our low estimate of deaths from drone strikes and our high estimate. The wide range reflects the large numbers of drone strikes since 2004 and also reflects the differing death tolls that are consistently reported in credible news accounts.
• Normally, we list the range exactly as it was reported in the media accounts. In certain instances the reported death tolls increased rapidly over time and offered new detailed estimates of militant vs. civilian deaths. In these cases we sometimes elected to exclude the lowest casualty ranges, on the assumption that they were out of date and denoted this with an explanatory asterisk.