How to do social science

There’s a terrific article by Jack Snyder and Erica Borghard in the latest issue of the American Political Science Review, entitled "The Costs of Empty Threats: A Penny, not a Pound." Apart from its substantive contributions, it’s also a nice illustration of how social science knowledge accumulates and progresses. Indeed, the article is also an excellent reminder of why we need a diverse array of scholars in our business, employing a variety of methods and theoretical perspectives.

In this piece, Snyder and Borghard challenge a well-known argument in the field of IR and foreign policy: namely, the idea that "domestic audience costs" give democratic states certain bargaining advantages in international disputes. The idea originated in a brief comment in one of Thomas Schelling’s books, but the seminal treatment is a widely cited 1994 article by Jim Fearon (then at Chicago, now at Stanford). Fearon argued that democratic leaders who issued threats toward an adversary and then backed down risked paying "audience costs" (i.e., their publics would punish them for making the threat and then retreating). By contrast, authoritarian leaders did not face similar audience costs (because they were not accountable to public opinion), so they could retreat without fear of domestic electoral punishment. Paradoxically, this situation could give democratic leaders a bargaining advantage in crises: Because democratic leaders would worry in advance about the dangers of bluffing and then being forced to retreat, they would only issue public threats if they were serious and not going to back down.

Fearon’s argument had considerable prima facie plausibility, and the basic idea has been used to explain a variety of international phenomena, including the so-called democratic peace. But Fearon did not provide systematic evidence to support his argument, and subsequent attempts to conduct empirical tests of the idea have yielded mixed results.

The Snyder/Borghard piece is the most serious attempt to test this conjecture to date. They conducted detailed historical investigations of a series of post-1945 international crises, and in their words, they find "hardly any evidence" that audience costs operate in the manner depicted. Instead:

Audience cost mechanisms are rare because (1) leaders see unambiguously committing threats as imprudent, (2) domestic audiences care more about policy substance than about consistency between the leader’s words and deeds, (3) domestic audiences care about their country’s reputation for resolve and national honor independent of whether the leader has issued an explicit threat, and (4) authoritarian targets of democratic threats do not perceive audience costs dynamics in the same way that audience costs theorists do.

And for your would-be policymakers out there, their bottom line is well worth emphasizing:

Future leaders of democracies should not come away from their political science classes having gained the impression that democracies can safely get their way in a crisis by publically committing themselves to fight for otherwise unpersuasive objectives.

There’s also a broader lesson for the social sciences too. Like all fields of study, social science advances through a process of conjecture, debate, argument, and refinement. Fearon’s original article was an important contribution that stimulated lots of creative thinking, and the fact that it may not stand up to careful empirical scrutiny is nothing to regret. We would not know what we now think we know had he not written the original piece (assuming, of course, that Snyder and Borghard’s critique stands up to subsequent scrutiny too).

Note further that there was a genuine division of labor involved here, operating over more than a decade. Fearon’s original piece was in the rational choice tradition, based on a formal mathematical model and buttressed by some suggestive supporting anecdotes. Snyder and Borghard’s article, by contrast, is mostly a careful and sophisticated empirical test using qualitative, "process-tracing" methods, designed to tease out whether "audience costs" played a significant role in key decisions or not. But Snyder and Borghard also identify possible flaws in the original causal logic and identify alternative causal paths to account for the observed results. Thus, neither type of work is either purely "theoretical" or purely "empirical."

For the academy, the moral of this story is that study of international relations would be greatly impoverished if one approach, method, or theoretical perspective came to dominate the field. This would produce an intellectual monoculture where scholars with different strengths were less likely to engage in a productive if competitive interchange, and one where research agendas were set largely by what questions could be studied by the reigning method du jour. In addition to real-world relevance, academic departments ought to prize intellectual, theoretical, and methodological diversity, both because they will be better able to deal with new topics and because different approaches have different strengths and limitations. This is not an argument for an "anything goes" approach to methods; rather, it is an argument against the repeated efforts to impose a single template for what is "good scholarship" on the field.