12
3. Research Design
In thinking about how to test the argument presented in the preceding section, it is
important to keep in mind that my argument does not conceive of impartiality as either a
necessary or a sufficient condition for intervention success (or failure), but as only one
among several variables that influence the chances that a peaceful outcome can be
achieved.
18
Methodologically, this implies that merely identifying a number of cases in which
the outcome of a given intervention effort matches my theoretical predictions would do as
little to prove the argument correct as finding occasional anomalies would do to refute it,
as both seemingly confirming and seemingly disconfirming observations could in fact be
driven by factors other than the intervener’s impartiality or bias. To avoid this problem,
we ideally would want to examine a relatively large number of cases and explore whether,
controlling for the influence of other likely determinants of intervention success, a
significant relationship exists between the impartiality or bias of interveners and their
performance as peacemakers that follows the hypothesized empirical pattern.
Unfortunately, this “ideal” research design is much easier to conceptualize than to
implement. Two challenges in particular stand out: First, although commitment problems
are easy to separate analytically from other obstacles to efficient bargaining, such as
informational asymmetries or perceptual biases, empirically they rarely occur in total
isolation from these other causes of conflict.
19
Even in hindsight, determining which of
18
As I show more formally in the game-theoretic model from which the above propositions are derived,
other important determinants of intervention success include the intensity of the intervener’s preference for a
negotiated outcome, the cost of intervention, as well as the intervener’s audience costs and its ability to
reduce the gains that disputants can reap from cheating.
19
Theoretically, too, there are reasons to believe war often results from the combination or interaction of
several different bargaining problems rather than a single cause. Cf. Fearon 1995.
Convention