how to create your own site for free


The Drone Wars database aggregates information that has been collected by the Bureau of Investigative Journalism on United States drone strikes in Afghanistan, Pakistan, Somalia, and Yemen.


The Bureau of Investigative Journalism claims to be an independent, non-profit organization that seeks to help the public better understand the world through in-depth, factual, politically-unaffiliated, investigative journalism. However, it should be noted that the areas of their research address left-leaning social justice issues.

The database is comprised of information from news reports, statements, documents, press releases, images, and videos, which are included in the data when possible, from both national and international outlets. The majority of this data is from news reporting.

DRONE WARS uses four discrete databases from The Bureau; if you would like to see these datasets individually, the coutries' data is listed here: 


Before we could create an argument with data, we needed to critically analyze the politics of data collection processes & outcomes.

How was the data generated?

What incentive is there for The Bureau to investigate the drone campaign? What are the stakes & who are the stakeholders?

What are the original sources?

What sources are deemed worth of reporting? Whose information is valid or credible?

What information is left out?

What questions can our dataset not answer? Are there any actors or outfits that benefit from data silences?

Who does the dataset represent?

What ideologies are reflected in the way data is recorded, organized, represented, & published?


the Drone Wars dataset focuses on the countries of Afghanistan, Pakistan, Somalia, and Yemen even though the US has launched drone campaigns against many other nations. For example, data on US drone strikes in countries such as Iraq, Iran, Libya, and Algeria failed to make the dataset.[1] 

The lack of representation may be due to an inherent barrier of information availability for drone strikes since it is a topic known for being highly classified & lacking in public transparency. Information gaps also affect the data within the datasets. Within the data provided for the four countries, the purpose of the drone strikes and the justifications for their targets lack detail. 

The data given for Afghanistan, Pakistan, Somalia, and Yemen also differs in terms of variables, specificity, and quantity. Some datasets have more entries while others might have more details in the form of additional columns of information per strike. Pakistan is noted for having the least amount of columns but the most strike entries.


The U.S.’s first drone-enabled airstrike was in Afghanistan in 2001, however, the Bureau of Investigative Journalism did not begin recording strikes in Afghanistan until 2015.[2] The first strike recorded by the Bureau of Investigative Journalism is in Yemen in 2002, the first known US strike outside of Afghanistan.[3] The U.S. began carrying out strikes in Somalia in 2001, however the Bureau of Investigative Journalism does not begin recording strikes in Somalia until 2007.[4]

The Bureau of Investigative Journalism updates the dataset as needed to include drone strikes authorized by the Trump administration. The data used in this project reflects the dataset as of November 30, 2018. It includes data from 1,696 strikes between November 3, 2002 and November 25, 2018.


Looking at metadata reveals what information is valuable to the publisher. From all four datasets, we gathered three main focus areas the data tries to highlight.



Every entry across all datasets are marked by a single strike. "Strikes" are considered "a missile or missiles fired at one location in a short time period," & are given a unique identifying code.[5] There is also an effort to archive these strikes' locations, with some datasets reaching the specificity of village name, yet some are noted to have "unknown locations."



The dataset is interested in the public relations regarding the drone campaign. Strikes are reported as “possible” unless they are confirmed by a US government source, named international sources, or by three local sources, in which case they are reported as “U.S. Confirmed."



The maximum count of people who have been killed includes updated counts as rescue efforts continue or people succumb to their injuries. It may be presented as a range to reflect vaguely reported casualties or if there are inconsistencies in casualty figures in reports covering the same strike.


There are major discrepancies between how the U.S. government classifies people versus how The Bureau of Investigative Jounrnalism constructs categories for casualties.


“Civilian” casualties are defined as those in which the dead were reported as “people,” “tribesmen,” or unknown identities. Children are reported as those between the ages of 0-17. The majority of deaths are considered to be “militant,” “organised, named groups that bear arms and that are not part of Pakistani, Somali, or Yemeni military, police, paramilitary or militia force” of which a specific count is not kept.[6] 


Drone strikes, as the data and our narrative demonstrate, regularly injure or kill innocent civilians and children "indirectly." Thus, there is intense linguistic debate concerning who is considered an enemy of the state. Obama endorsed a disputed method of counting civilian casualties, stating that “all military-age males in a strike zone count as combatants […] unless there is explicit intelligence posthumously proving them innocent.”[7][8] 


While there are some nuances & inconsistencies in metadata categories across all four countries’ datasets, there is a distinct interest in the individual drone strikes as opposed to their proximal impacts, such as their effects on individual provinces or people. This is apparent in the fact that every entry in the dataset accounts for a single attack.

Additionally, the four databases share a mutual interest in time, location, and how many were injured or killed per attack, further demonstrating that the dataset’s categories are oriented around individualized drone strikes. It is through the dataset’s ontology that the systemic effects of categories can be observed.

Categories inevitably produce bias and bring light to some things while silencing others. For example, the database does not break down casualties by personal identity (ex. by gender) or communities, suggesting an objective emphasis on the drones and their direct, quantitative effects. Though the data can help us make inferences, they cannot truly account for personal experiences, sociopolitical contexts, or geopolitical agendas. This is not inherently good nor bad, but it is important that we recognize that the dataset provider either had an interest in very specific data or that the absent data is difficult to acquire.

[1]    Scahill, Jeremy, et al. The Assassination Complex inside the Government’s Secret Drone Warfare Program. Simon & Schuster Paperbacks, 2017. 155-211. Print.

[2]    Kaplan, Fred. “The First Drone Strike: How a Cold War Idea Became the Dominant Weapon in the War on Terror.” Slate Magazine, Slate, 14 Sept. 2016,

[3]   Hudson, Leila, et al. “Drone Warfare in Yemen: Fostering Emirates Through Counterterrorism?” Middle East Policy, vol. 19, no. 3, 2012, pp. 142–156., doi:10.1111/j.1475-4967.2012.00554.x.

[4]    "Our Methodology." The Bureau of Investigative Journalism, 2018,

[5]    Ibid.

[6]    Ibid.

[7]    Becker, Jo, and Scott Shane. “Secret 'Kill List' Proves a Test Of Obama's Principles and Will.” The New York Times, 29 May 2012,

[8]    Bolland, Thomas, and Jan Andre Lee Ludvigsen. “‘No Boots on the Ground’: the Effectiveness of US Drones against Al Qaeda in the Arabian Peninsula.” Defense & Security Analysis, vol. 34, no. 2, Mar. 2018, pp. 127–143., doi:10.1080/14751798.2018.1478184.