Collaborative Research on the Expected Utility Theory of War

The researchers significantly revise the data-generating and management software program EUGene (Expected Utility and Data Generation), originally developed under an NSF award, in order to 1) add functions that will improve its usefulness as a data management tool, and 2) address new theoretical research questions concerning the expected utility theory of war and comparative testing of international conflict theories.

EUGene is a software package that generates data for variables used to test a version of an expected utility theory of war and dispute initiation. Specifically, EUGene is the first publicly-available program that generates tau-b alliance similarity scores, risk attitude scores, expected utility data, and international interaction game equilibria for all states (or dyads) and years from 1816 to 1984. In addition, EUGene serves as a data management tool for creating data sets to be used in the quantitative analysis international relations, and is particularly useful for creating data sets with the directed-dyad-year as the unit of analysis. Until now, creating such data sets has been cumbersome and difficult. The researchers have used EUGene in conducting the largest analyses of the expected utility theory of war to date (analyzing nearly 700,000 directed dyad-years of data. They also have used it to build a data set to analyze 11 competing theories of international conflict on the population of all directed interstate dyads.

EUGene is expanded in a number of ways that benefits the international relations community. First, the researchers implement a new measure of alliance similarity, the 'S' score which has been argued to be a better measure for analyzing alliance similarity than tau-b, on which current expected utility data are based. Next, the software are updated to allow users to add any data set (of a suitable format) to those provide with the program, making EUGene useful for a wide variety of data set development projects. Finally, the researchers implement a number of options that have been requested by users, including the ability to generate non-directed dyad data sets, to update expected utility data as new alliance data is made available, and to track versions of data for replication.

With the updated data software, the researchers build new data sets that facilitate significant new research projects. Three main substantive projects result from this research. First, using newly generated data the researchers expand on comparative tests of theories of international conflict conducted previously. In particular, with added data the researchers test a three-stage econometric model of conflict that integrates trade, political institution, and non-governmental organization. Second, the researchers explore the nature of preference stability across the international system, developing an evolutionary model of preferences that helps to explain earlier findings of an inconsistent fit of the expected utility theory of war across regions and over time. Third, by fully implementing the 'S' score measure of alliance similarity and using it to develop new risk scores and expected utility data, the researchers fully evaluate the effect of this new measure for prior and future empirical research.

This research enhances substantially our understanding of this important topic.