Writing for the Atlantic, three American academics posed a challenge in their article titled: “You Say Pakistanis All Hate the Drone War? Prove It.” I thought I did prove it a few weeks ago. But I welcome the opportunity to elaborate even further.

The American academics are convinced that a large number of Pakistanis are ignorant of drone strikes, and that another sufficiently large number of Pakistanis support the American drone strikes on Pakistani territory. The academics have relied on a data set by Pew Global Attitudes Project to reach these conclusions. I have demonstrated in an earlier submission that their conclusions are not supported by data. I further illustrate here that the data set is deficient in answering questions of such wide implications.

Let me begin by answering the question, “how many Pakistanis support drone strikes?” Using the same data set as the one used by the American academics I found no more than the 20 respondents out of the 2,000 (i.e., 1 per cent) surveyed by the Pew Global Attitudes Project who truly supported the drone war. Let me explain: if we consider only those respondents who were aware of the drone strikes and who considered drones to be a good thing and went ahead with supporting the drone strikes, they add up to only 20, which is a much smaller number than the one estimated by the American academics.

I would also like to mention that Pew’s questionnaire is deficient in several ways. For instance, some key questions about drones are convoluted while other questions force respondents to either support or oppose the drone attacks. The questionnaire does not allow respondents to declare indifference, i.e., neither support nor oppose drone strikes, which may be the reason why so many chose the option ‘Don’t Know’ or refused to answer the question. I am of the view that the ‘Don’t know’ option in the data set is capturing ignorance of, indifference to, and the refusal to answer questions about drone attacks. Thus, the data set imposes restrictions on one’s ability to infer from it.

Let me first make it clear that my opposition to drone attacks is not a disguised support for the Taliban. I consider Taliban to be murderous thugs who pose a clear and present danger to Pakistani people and the State. I am, however, also convinced that the American drone attacks weaken the efforts by the Pakistani State and the civil society to confront the menace head-on.

Pew data limitations

Data analysis is somewhat similar to being interrogated by an intelligence agency in the dungeons of the Lahore Fort. Given enough servings of torture, eventually the suspect, in our case the data, will say whatever the interrogator wants.  The American academics have also acted like Punjab Police. First, they have tortured data enough to ensure it sang their ideological song. Second, just like the Pakistani Police, they have discarded useful evidence that contradicted their assumptions.

The Pew data is not sufficient to support the conclusions drawn by the American academics. In addition to the data deficiencies I have mentioned in my earlier blog, let me explain my additional concerns about the data and the methods used. The Pew Centre’s questionnaire posed a convoluted question about the support for drone attacks, which suggested to the respondents that the drones will be managed by the Pakistani authorities and not the Americans. Even with a highly misleading question, only 23 per cent reported the support for drone attacks whereas another 32 per cent registered outright opposition to the drone attacks (see the table below).

Support or oppose drone strikes	Freq.	Percent
support	454	22.70
oppose	648	32.40
don't know	860	43.00
refused	38	1.90
Total	2,000	100

What is not obvious from the above table is that the option ‘don’t know’ is not read to the respondent. It was up to the interviewer to assign ‘don’t know’ or ‘refused’ as a response to those who neither supported nor opposed drone strikes. I am doubtful of the manner in which the interviewers decided upon ‘don’t know’ as the response, whereas I see this as a simple refusal to answer the question.

Earlier in May 2012, I helped the Pew Global Attitudes Project fix coding errors in the same data set. They were quick to acknowledge the error and fix the problem. I alerted Pew of the above-mentioned concerns weeks ago. They are yet to respond.

Notice also that there is no missing data in the above table, which suggests that all 2000 respondents recorded their answers.  The American academics argue that if we consider responses from only those who either support or oppose drone strikes, only then 41% of the informed Pakistanis support drone attacks.

There are serious problems with this approach. Notice that the above question forces one to either support or oppose drone strikes. The option to be indifferent or neutral on the issue is not available. Thus, most respondents (43 per cent) chose ‘don’t know’ to either indicate their ignorance of, indifference to, or simple refusal to answer the question about drone attacks in Pakistan. The American academics treat ‘don’t knows’ as respondents being ignorant of drone strikes.

Given that drones are the preferred topic for Friday sermons in almost every mosque (radicalized or otherwise) across Pakistan, I find it hard to believe that 43 per cent Pakistanis, irrespective of their education attainment, are ignorant of drone attacks.

The coalition of the unwilling and the ignorant

One of the basic tenets of good detective work or data mining is not to discard evidence that does not support your hunch. The American academics have regrettably ignored evidence that showed limited support for drone attacks in Pakistan. Let me explain. Apart from the question about supporting drone attacks, the questionnaire first asked the respondents to report if they knew of the drone strikes that have taken place in Pakistan. It turned out that a large number of those who reported no knowledge of the drone strikes in Pakistan later registered their support for the same in the future.

The following figure illustrates the dilemma that would have confronted the American academics if they had looked at the data more carefully. Of the 454 who reported support for the drone attacks, the majority (234) did not know of drone strikes in the past. They are more in number than the 220 who knew of drone attacks and registered support for the same. This suggests that the American academics have equated the support of those who are knowledgeable of drone strikes with that of those who are ignorant. To me it appears more of a support of the ignorant than that of highly educated and internet savvy Pakistanis with a taste for English.