Choice stabilities in the graph model for conflict resolution

Abstract

The graph model for conflict resolution is a comprehensive, adaptable system to represent and analyze a strategic conflict. Though construction of a graph model requires relatively little information about preferences, obtaining that information is often the most difficult step of model calibration. We propose a version of the graph model that avoids preferences, relying instead on direct information about individual choices. After defining appropriate choice functions, we extend the five basic graph model stability definitions to this new context. We then show how the new definitions are related, both to each other and to the standard definitions based on preference relations. We also use a simple model to illustrate how choice stabilities can differ from preference-based stabilities.