Disaggregating the Costs of War: Explaining Resolve in Military Interventions
Download data and study materials
Sample size: 1260
Field period: 04/08/2012-07/18/2012
Why do some segments of the public display remarkable persistence in wartime, while others are more sensitive to the costs of war? This survey experiment explores participants' opinions towards a hypothetical military intervention: first modeling the initial choice to intervene, and second, modeling how long participants want to stay as the costs of war mount. The embedded experiment manipulates the costs of fighting (ex-ante and ex-post) and the reputational costs of backing down. The survey instrument also measures a series of individual difference characteristics (time preferences, risk preferences, and honor orientations). The results suggest that resolve is both situational and dispositional, pointing in particular to the importance of reputation costs, time preferences, and risk preferences.
Against the standard approach to resolve in the IR literature -- which tends to treat "the costs of war" as shorthand for resolve (and view the costs of war as synonymous with casualties) -- I seek to disaggregate the costs of war, and test the hypothesis that actors are sensitive both to the costs of fighting, and to the costs of backing down. That is, actors should display less resolve when the costs of fighting are high, but actors should display more resolve when the reputational costs of backing down are high. I also test a set of dispositional hypotheses that link resolve to a series of characteristics of respondents themselves (time preferences, risk preferences, and honor orientations), testing whether the amount of resolve participants display increases as a function of their time preferences, risk attitudes, and adherence to a "culture of honor."
The study is a 2x2x2x2 between-subjects factorial design, which includes two major components: a dispositional questionnaire, and an embedded 2x2x2 factorial experiment that manipulates the expected costs of fighting (low or high), reputation costs (primed, or not discussed), and the actual cost of fighting (low or high) of a hypothetical military intervention. The design can be understood as a 2x2x2x2 because it includes an order manipulation that randomizes whether the dispositional questionnaire is administered before or after the situational experiment, to minimize treatment spillover effects.
Participants are asked two sets of questions, the first involving conflict onset, and the second concerning the intervention's duration: i) whether the United States should intervene, and ii) how long should American troops remain deployed? This second measure -- how long should the United States remain involved in an ongoing intervention -- is used as a measure of how much resolve participants display, and is thus the key dependent variable in the study. However, since the study models onset and duration as part of the same continuous process, analyses can also be conducted differentiating between participants who were opposed to the intervention from the beginning, and those who initially supported intervening but developed cold feet once the costs of war mounted.
The results of this experiment find relatively little support for the importance of the costs of fighting, but greater support for importance of the reputational costs of backing down: participants reminded of the reputational consequences for failing to stand up to aggression were willing to stay longer than those participants who were not reminded about reputation costs. Similarly, although there was no indication that honor culture is positively associated with resolve, participants with more patient time preferences also displayed more resolve in the intervention scenario, while risk attitudes have a significant curvilinear relationship with resolve.
Book manuscript, "Resolve in International Politics."