Share this:

Jay Lyall 80x108Graeme Blair 80x108Kosuke Imai 80x108As the withdrawal of U.S. and NATO forces from Afghanistan continues in 2014, many are now concerned about the attitudes of local civilians towards the Taliban compared with the International Security Assistance Force (ISAF). Jason Lyall, Graeme Blair and Kosuke Imai have conducted research on the ground in Afghanistan which aims to measure how civilian attitudes are effected by violence from both groups. They find that ISAF victimization is associated with a large increase in support for the Taliban, but that harm by the Taliban only leads to a modest drop in support.They also find however, that small, targeted assistance programs among those harmed by ISAF can help to reverse much of the outflow of support to the Taliban.

The arrival of 2014 promises to open the flood gates of prognostication about Afghanistan’s future as the long-planned withdrawal of US and NATO forces nears completion. Much stock has been placed in discerning Afghan attitudes toward their government and the Taliban as clues for anticipating future events. And with good reason. Counterinsurgency theorists (well, most of them) have argued that winning “hearts and minds” is a key, if not the key, to victory—or at least what passes for victory in these settings.

Yet the deeper question—just how do we measure “hearts and minds” accurately?—is often lost in the revolving shuffle of Powerpoint decks and (endless) debates about metrics. Clearly the obstacles are formidable. Dumping billions of dollars into a country is likely to skew attitudes, if only because it generates incentives for recipients to shade their answers in ways that guarantee future assistance. The shadow of violence also looms over respondents and enumerators alike: speaking honestly, or simply entering a village to solicit opinions, can be risky endeavors.

Credit: DVIDSHUB (Creative Commons BY)

Credit: DVIDSHUB (Creative Commons BY)

In 2010-11, we took a stab at measuring Afghan attitudes toward the International Security Assistance Force (ISAF) and the Taliban using an indirect survey method (“endorsement experiments”) designed to minimize these issues. We were particularly interested in figuring out how support for ISAF and the Taliban were (not) affected by violence. What happened to support for ISAF once an individual (or his family) was harmed by ISAF? Did the same hold true for the Taliban, or did Taliban violence mean something different to its victims?

Very briefly, we surveyed nearly 3,000 Afghan males in 204 villages in 5 predominantly Pashtun provinces. We posed a series of questions about an individual’s exposure to violence by both sides. We then used four indirect questions which we pooled together to measure support for each combatant.

The mechanics of a survey endorsement experiment are straightforward. Randomly selected respondents are assigned to a treatment group and asked to express their opinion toward a policy endorsed by a specific actor whose support level we wish to measure (here, ISAF and the Taliban).

These responses are then contrasted with those from a control group of respondents that answered an identical question without the endorsement. Higher levels of enthusiasm for a policy with an endorsement relative to those without it are viewed as evidence of support for the endorsing actor. Since each respondent is assigned only one condition for any endorsement experiment, it is impossible for enumerators or others to compare support levels across different conditions for any individual respondent.

What do we find? Put simply, the effects of combatant violence on civilian attitudes are highly asymmetric. Harm by ISAF, as outlined in Figure 1 below, is associated with a sharp decrease in support for ISAF (column 1, left side) and a marked increase in support for the Taliban (column 2, left side). Harm by the Taliban, however, is associated with almost no transfer of support to ISAF (column 1, right side) and has only a very modest negative effect on support for the Taliban (column 2, right side).

Figure 1 – Support for ISAF and Taliban and harm

Lyall Fig 1

The final panel in the figure drives this point home: ISAF victimization is associated with a large increase in support for the Taliban; harm by the Taliban, only a modest downturn in support for the Taliban. While it would be inaccurate to conclude that the Taliban can harm civilians without repercussions, it is apparent that they enjoy (if that is the right word) far more latitude than ISAF. 

These findings carry several implications for understanding the dynamics of violence in Afghanistan today. Hoping that Afghans will turn away from the Taliban in disgust at civilian casualties, for example, is unlikely to be a viable strategy, at least among Taliban supporters. According to UNAMA’s data, the Taliban have been responsible for at least 80 percent of civilian casualties since 2008. Yet this victimization is unlikely to have the same meaning, or political impact, as (much rarer) ISAF civilian casualties.

Abandoning efforts to influence attitudes would be equally mistaken, however. We found suggestive evidence, for example, that small, targeted assistance programs among those harmed by ISAF managed to reverse much, though not all, of the outflow of support to the Taliban. Moreover, evidence points to attitudes as an important predictor of future insurgent attacks, suggesting a link between attitudes and (future) behavior that should not be ignored.

That, in turn, requires bringing increasingly sophisticated methods to bear to measure attitudes in violent settings. Glamorous stuff, no. But if we can’t even measure “hearts and minds,” why would we ever believe that we could “win” them?

This article is based on the paper, ‘Explaining Support for Combatants during Wartime: A Survey Experiment in Afghanistan’ in the November 2013 issue of American Political Science Review. The authors would like to thank Cambridge University Press for ungating the article.

This article has also been posted at the Washington Post’s Monkey Cage blog.

Please read our comments policy before commenting. 

Note:  This article gives the views of the authors, and not the position of USApp– American Politics and Policy, nor of the London School of Economics.

Shortened URL for this post:


About the authors

Jay Lyall 80x108Jason Lyall Yale University
Jason Lyall is an Associate Professor in the Department of Political Science at Yale University.  His research focuses on violence and its effects in both conventional and counterinsurgency wars, with special emphasis on Afghanistan. His website is Follow him on Twitter (@jaylyall_red5)


Graeme Blair 80x108Graeme BlairPrinceton University
Graeme Blair is a Ph.D. candidate in the Politics Department at Princeton University studying comparative politics with a focus on West Africa. His website is Follow him on Twitter (@graemedblair)


Kosuke Imai 80x108Kosuke Imai Princeton University
Kosuke Imai is Professor in the Politics Department at Princeton University, where he is also an executive committee member of the Committee for Statistical Studies and the Program for Quantitative and Analytical Political Science (Q-APS). His website is