Modeling International Negotiation Statistical and Machine Learning Approaches

Abstract

An earlier study by Druckman et al. (1999) showed that a variety of cases of international negotiation can be distinguished in terms of their objectives. A set of 16 features of negotiation effectively distinguished—by multidimensional scaling (MDS)—among the types of international negotiation objectives proposed by Iklé (1964). The features include aspects of the parties, issues, process, negotiating environment, and outcomes. The statistical analyses performed in that study showed a distinct profile for each of the Iklé categories: innovation, re-distribution, extension, normalization, and side effects. In addition, a sixth category was identified as being different than the others. This category was referred to as multilateral regime negotiations, a form that became prevalent twenty years after Iklés book appeared. These results were further supported by discriminant analysis classifications. When only information about the 16 features were known, 78 % (or 21 of 27 cases) were placed in the correct a priori category. These are impressive results. They provide empirical validation for this well-known typology of negotiation. In this chapter, an attempt is made to extend these analyses in several directions with the help of sophisticated methodological approaches not previously used to interpret data on negotiation.