New Study Suggests Drone Strikes Don’t Kill as Many Pakistani Civilians as Claimed

It’s the most controversial counterterrorism program there is. The CIA’s remotely piloted aircraft, operating with the tacit consent of the Pakistani government, fire missiles at suspected militants in the Pakistani tribal areas where U.S. ground troops are prohibited from operating and where the Pakistani military is often hesitant to tread. The United Nations’ special rapporteur on extrajudicial killings plans to formally request the Obama administration stop the program out of fears that civilians inevitably die in the strikes. Recent research from the New America Foundation finds that 30 percent of drone strike fatalities are Pakistani civilians. It’s an enormous issue in bilateral relations with a major non-NATO ally, and experienced counterinsurgents like David Kilcullen and Andrew Exum have warned that the incendiary attacks may create more militants than they kill. Even John Brennan, President Obama’s counterterrorism adviser, indicated on Wednesday that he shares Kilcullen and Exum’s fears and gives scrutiny to ensure that the much-valued program doesn’t become “a tactical success but a strategic failure.”

But a forthcoming study, led by Brian Glyn Williams, an associate professor at the University of Massachusetts, finds that the civilian death toll from the drones is lower than most media accounts present. “We came to the conclusion that the drones have a unique capability for targeting militants, as opposed to civilians,” Williams said in an interview.

Williams’ study, which he provided to The Washington Independent, has yet to be published. A writer for a blog affiliated with the International Herald Tribune, Farhat Taj, blogged some of the key details of his research today, but prematurely stated that the Combatting Terrorism Center at West Point will be publishing Williams’ work. Erich Marquardt, the editor of the center’s journal, said that he hasn’t even begun to review Williams’ submission yet.

Much like the New America Foundation study, Williams’ team relied on English-language media accounts of the drone strikes in Pakistan to compile a data base of how many civilians and militants were reported to be killed. He conceded from the start that such a reliance is a “serious limitation” of the study — news reports can, after all, be incorrect — but the tribal areas of Pakistan where the strikes occur are often off limits to Western researchers, and even their Pakistani counterparts. (Still, Williams plans on traveling to the tribal areas on June 10 to attempt a poll of local attitudes about the strikes.) His team took measures to mitigate that limitation: they only considered strikes that had been reported by multiple independent outlets and they erred on the side of treating the deaths of people in disputed militant status as either civilians or “unknown.”

Williams’ results, which he said have been peer-reviewed, are as follows:

That leaves just over 19 percent of reported deaths out of either category, as their status as civilians or combatants can’t be rigorously determined under Williams’ methodology. But he writes that “even if every single ‘unknown’ is assumed to in fact be a civilian, the vast majority of fatalities would remain suspected militants rather than civilians – indeed, by approximately a 3.4:1 ratio.”

Williams insists that he went into the study with an open mind. “We didn’t know what to think” about the drone program, he said, and he considers his research agnostic on the wisdom of the drone strikes (to say nothing of their legality). “We’re not necessarily trying to alter policy on this,” he said.

Both of the principle authors of New America’s drone strike survey, Peter Bergen and Katherine Tiedemann, are on vacation, but they both still (generously) addressed my questions. All three researchers — Bergen, Tiedemann and Williams — appeared to agree that New America was more methodologically aggressive than Williams in counting as civilians all who could not be clearly identified as militants, which perhaps accounts for the variance in their results.

Bergen observed in a Blackberried message that although his civilian death tallies are higher than Williams’, he has observed that the drone program has increased its accuracy over time, “so the later the the date that the study begins the lower the rate [of civilian deaths] will be.” That’s in line with Brennan’s intimation (he never actually uses the word “drones”) that the drone strikes “are more precise and more accurate than ever before.”

Accordingly, Bergen now pegs the civilian death rate from the drone strikes at 20 percent. Williams pegs it at 3.53 percent. What no one knows, however, is how many outraged Pakistanis take up arms against the U.S. or its allies as a result. There are media reports suggesting that Faisal Shahzad, the naturalized U.S. citizen of Pakistani origin accused of attempting to detonate a car bomb in Times Square, claimed to investigators that his attempted terrorist act was vengeance for civilians killed by the drones. Leaving aside the question of the legality of the drones — which the State Department’s legal adviser claims to result from a September 2001 act of Congress that doesn’t mention the program — only policymakers can determine if the benefits of the drones outweigh the risks of blowback.