Editor’s Note: The Pakistani people’s deep opposition to U.S. drone strikes against targets in Pakistan is common wisdom, and some observers see the associated anger as a major source of the country’s rampant anti-Americanism. C. Christine Fair, Karl Kaltenthaler, and William J. Miller challenge this conventional wisdom, pointing out the (many) flaws in polling data from which this conventional wisdom is drawn and suggesting that drone strikes, while not exactly popular, are not high on the list of concerns for most Pakistanis.
Since 2004, the United States has employed weaponized unmanned aerial vehicles (UAVs), popularly known as “drones,” to kill alleged terrorists in Pakistan’s Federally Administered Tribal Areas (FATA). The program is controversial in Pakistan and has further strained U.S.-Pakistan ties. Although drone strikes irritate an already sensitive U.S.-Pakistan relationship, commentators often depict the strikes as the single most significant aggravation. This view is buttressed by the belief—which has become a truism in Western and even Pakistani media—that not only do most Pakistanis know about the program, they overwhelmingly oppose it.
In a recent article published in Political Science Quarterly, we report that this conventional wisdom about Pakistanis’ universal opposition to the drones is not empirically supported. Pew data from 2010 demonstrate that nearly two-thirds of Pakistanis have never heard of the drone program, despite the extensive media coverage it has received in Pakistan and beyond. (Since drafting this paper, Pew has released newer data. However, at the time of writing and analysis, this was the most recent data available.) We conclude that Pakistani public opinion is less informed and much less unanimous than is often presumed.
We examined Pakistani media coverage of the drone program to draw out the arguments advanced in what we call an “elite discourse” about drones. We argue that this discourse is elite because most Pakistanis do not obtain information from newspapers (in any language) due to illiteracy: Only 55% of Pakistanis over the age of fifteen years of age can read and write. We augment this Pakistan-focused research with the scholarly literature on public opinion formation, which contends that societal and political elites play a large role in shaping what the public thinks about policy issues, particularly policy issues they do not understand very well. Our analysis suggests that the Pakistani debate over drones is waged among elites, who nonetheless differ in key ways, such as level of education, literacy in English, and access to non-Urdu media.
Drawing from our examination of Pakistani media coverage and the scholarly literature on public opinion formation, we put forward several hypotheses about who will oppose the program and why. We tested these hypotheses using data from the 2010 Pew Global Attitudes survey, which included questions about drone strikes in Pakistan. To determine the factors that predict support or opposition to the program, we leverage several questions in the Pew data, which we use variously for our dependent and explanatory variables in a Heckman probit model. We use Heckman to account for the selection effects described below.
The first question we examine is: How much, if anything, have you heard about the drone attacks that target leaders of extremist groups - a lot, little, or nothing at all? Of the responses, the largest category was nothing at all, with 43%. Don’t know/refused was the second largest response category with 22%. Of those who responded that they knew something about the drones, 21% said they knew a little and 14% said they knew a lot. Thus, in 2010, only 35% of the sample claimed that they knew something about the drone program, whereas 43% stated they knew nothing about it. Clearly, a minority of Pakistanis in these data are familiar with the drone strikes. This is the gateway question: if a respondent answers nothing at all or does not answer the question they are not asked the subsequent questions about the drone strikes. If they answer, a lot or a little, they are asked what they think about the drone strikes.
To gauge opposition to U.S. drone strikes, Pew asked respondents: Please tell me whether you support or oppose the United States conducting drone attacks in conjunction with the Pakistani government against the leaders of extremist groups. Among these respondents 23% support the drone strikes, 32% oppose the drone strikes, and 45% said they do not know or refused to answer the question.
Modeling Opposition to U.S. Drone Strikes in Pakistan
We use survey data from the 2010 Pew Global Attitudes Survey, which collected the views of 2,000 Pakistanis on a range of issues. Unfortunately, for several reasons, the sample of persons who gave an opinion about drones (1,681) was not representative of Pakistanis generally. First, Pew’s data on Pakistan are collected from mostly urban areas whereas Pakistan is still a mostly rural country. Second, because nearly two-thirds of the sample indicated that they had never ever heard of the drone program (or declined to answer the gateway question), most of Pew’s respondents must be excluded in the analysis of what drives attitudes about drones. It turns out that well-educated males with access to the Internet were most likely to be in the final sample (see Model 1 in Table 1).
To overcome the bias in our sample produced by these selection processes, we first had to discern what factors predict ignorance of this program, despite the enormous publicity it receives. To do so, we used a standard statistical tool called the Heckman Selection model, which allows us to control for the characteristics of those who are not familiar with the program as well as for other explanatory variables that may predict attitudes about the program among those who were familiar with it and expressed an opinion about it.
We next identified those factors that best account for Pakistani public opinion about drones using a using a statistical method that accounts for the unrepresentativeness of our sample. Our outcome variable (“what Pakistanis think”) comes from a question that asks: Now I’m going to ask you a list of things that the United States might do to combat extremist groups in Pakistan. For each one, please tell me whether you would support or oppose it. The respondent is then offered: Conducting drone attacks in conjunction with the Pakistani government against leaders of extremist groups. The respondent is then offered the choice of: support, oppose, don’t know, or refuse to answer.
We derived a set of potential explanatory variables (i.e., variables that may explain the variation in support and opposition to the strikes). These include education, gender, anti-Americanism, predilections towards political Islam, support for democratic norms, and fear of Al Qaeda. Information about respondent education comes from a question about the respondent’s highest level of education attained. We contend the more educated one is, the more likely one is to support drone strikes.
We include gender, as we expect that men may be more informed about political issues in Pakistan than women due to the fact that men are much more likely to be literate than women and have had more education. The most recent Pakistan Social and Living Standards Measurement Survey finds that overall, 70% of males over the age of ten are literate compared to 47% of females. According to the same survey, while 73% of males report having attended school, only 49% of females do so.
We include anti-Americanism as a potentially important determinant of attitudes about drones, and we measure this using a question about favorability toward the United States: Overall, do you think of the US as more of a partner of Pakistan, more of an enemy of Pakistan, or neither? We anticipated that respondents who say that they view the United States as more of an enemy of Pakistan will be more likely to oppose the drone strikes. We hypothesize that those who support democratic norms will be supportive of drone strikes because many drone opponents view the program as a violation of basic standards of rule of law. We identify those who support democracy using the question that asks respondents to agree or disagree to the assertion that Democracy is preferable to any other kind of government.
One of the major arguments made in favor of drone strikes in Pakistan is that they kill foreign terrorists, such as members of Al Qaeda. Therefore, we conjectured that those who fear Al Qaeda would be more in favor of drone strikes. For this reason, we use a question that asks: How serious of a threat is Al Qaeda to our country? Is it a very serious threat, a somewhat serious threat, a minor threat, or not a threat at all? We also expected that those who do not think Al Qaeda is a threat to Pakistan will be more opposed to the drone program.
Finally, drawing upon conventional fears about political Islam in Pakistan, we included a question that asks: How much of a role do you think Islam plays in the political life of our country - a very large role, a fairly large role, a fairly small role, or a very small role? If the respondent answered the question, they were then asked: In your opinion - is this good or bad for the country? If the respondent answered that Islam played a fairly small or very small role in the political life of the country and that this was bad, we expect that he or she would be more likely to oppose drone strikes. Likewise, if the respondent said that Islam played a fairly large or very large role in the political life of the country and that this was good, we would expect him or her to be more likely to oppose drone strikes. Finally, as is standard, we accounted for income in our model, even though we have no theoretical priors about how income could influence beliefs about drones.
So, which factors best explain support for or opposition to drone strikes? (For those interested in the regression results, see Model 2 in Table 1.) It turns out that there is no significant relationship between attitudes toward drone strikes and attitudes towards democracy, attitudes toward Al Qaeda, attitudes toward political Islam, or income. Several factors do explain respondent opinions about drones. First, unsurprisingly, individuals who believe the United States is an enemy are more likely to oppose drone strikes. Second, males and persons with more education are more likely to support the use of drones in Pakistan than others.
The coefficients we present in Table 1 do not portray the marginal effects of the independent variables on the dependent variable. To help present a fuller picture, we report what are called the “marginal effects” of our significant variables in Table 1. (Marginal effects measure the probability that a respondent will oppose drone strikes when all independent variables are held at their mean except for the variable of interest, which is moved from its minimum to maximum value.) Calculating marginal effects permits us to assess the substantive effect of each independent variable in explaining variation in the dependent variable. When we examine our model, the most substantively significant variable appears to be gender, with a marginal effect of .157. This means that solely by moving the variable from female to male, while holding all other variables at their mean, we see a 15.7% increase in the likelihood that a respondent will support drone strikes. Perceptions of the United States as an enemy (4.2%) and education (3.8%) both have smaller substantive effects.
Generally, we find that broad swathes of the Pakistani public are simply unaware of the drone program. However, those who are aware tend to oppose it by a margin of 20%. We find that the most educated Pakistanis are most likely to support the drone strikes, likely because they tend to have access to more varied sources of information (some of them in English) and thus are exposed to occasional—but important—pro-drone arguments presented in more sophisticated Pakistani media sources, as well as in foreign media. Pakistanis who have little education are most likely to oppose the drone strikes. Pakistani women, who are generally poorly educated and excluded from political discussions, tend to be more negative about the drones than men. Additionally, the more negative the respondent was about the United States in general, the more likely he or she was to oppose drone strikes. In contrast, respondents’ views on political Islam and fear of or support for Al Qaeda do not influence Pakistanis’ thinking about drones. These findings should encourage public commentators be more careful in their efforts to describe what is clearly a complicated landscape of Pakistani opinion towards this tendentious program.
So what does this study mean for the U.S. drone program in Pakistan? First, it indicates that this is not a conversation that has seized Pakistanis generally, as is usually reported. Instead, it is a discourse that is held among elites, most of whom are male and educated. This suggests that in the realm of “winning (or losing) hearts and minds,” there is a considerable population that is up for grabs by drone foes and drone proponents alike. (More recent data by Pew demonstrates that with each year, more people are becoming aware of the program, and that those who do know largely oppose it, with persistent minorities who approve of it in some measure.) Second, it appears that drone strikes per se are not driving the controversy. Instead, views about drone strikes mostly appear to be driven in some measure by anti-Americanism, which has a long and expansive history in Pakistan. In fact, Pakistanis have not had a positive opinion of the United States since the mid-1950s. Fewer drone strikes are not going to change this significantly.
TABLE 1: Heckman Logistic Regression Results
Dependent Variable: Oppose Drone Strikes
Note: Figures are unstandardized coefficients shown alongside standard errors. *p<.1; **p<.05; ***p<.01.