When will states talk? Predicting the initiation of conflict management in interstate crises

This research addresses the question of why some crises between states are resolvedthrough negotiated agreements while others result in continued conflict or escalate to war.The model deviates from previous approaches to the study of conflict management in fourkey ways: 1) management is treated as a conflict strategy rather than an outcome; 2) costs,rather than calculation of the relative benefits of conflict over management, motivate theinitiation of conflict management; 3) the conceptualization of costs is broadened toincorporate subjective factors; and 4) issue salience is proposed to determine the thresholdat which an actor’s preference for conflict over management changes.The central question this conceptualization raises, therefore, is what factorsinfluence actors’ strategy choices during a crisis. The theory proposes that, when it comes tothe initiation of conflict management, it is costs that dominate the decision process. Or asJackman (1993) so succinctly puts it; “for those confronted with a very restricted range ofavailable alternatives extending from horrendous to merely awful, minimizing pain is thesame as maximizing utility”.Both experimental and statistical methodologies are used to test the hypothesesderived from the theory. Original experimental data were collected from experiments run on undergraduate students at Texas A&M University. For the statistical analysis a data set ofinterstate crises and negotiation behavior was compiled using data from the SHERFACSand International Crisis Behavior data sets and data collected specifically for this research. Thismulti-method approach was chosen because of the nature of the questions being examinedand in order to minimize the limitations of the individual methodologies. The experimentaltests demonstrate that the expectations of the model are supported in the controlledenvironment of the experiment. The results from the empirical analysis were, within therestrictions of the data, consistent with both theoretical expectations and the experimentalresults.