MODELING CONFLICT PROCESSES

For some conflict processes, it is possible to specify relationships between variables with precision. These specifications may take the form of models that are often evaluated with experiments.

The most popular solution concept for two- person games is the Nash equilibrium. The simplicity and elegance of this concept derives from the property that it minimizes the losses incurred by game players. It is a social norm in the sense that all players prefer this outcome to any alternative. It is calculated as the outcome that maximizes the product of the preferences of the two parties. Along with other solution concepts (see Avenhaus and Zartman, 2007; Guner and Druckman, 2000), this idea covers a wide range of choice situations where a person must decide between making an offer or remaining silent, voting for one candidate or another, or intervening versus standing aside. Ithas the feature of providing a parsimonious explanation for social behavior such as bystander apathy. (See Osborne, 2004, for this and other examples.) These concepts come to life in the sequence of players' choices in different types of game structures.

An old problem for students of politics is the decision to form a party coalition. And, perhaps the oldest idea about how to do this is the minimum winning coalition (Riker, 1962): this is a coalition consisting of only those parties needed to win. Further work extended this idea by adding policy positions to the model. Referred to as a minimum connected winning coalition, this model has been shown to be a better predictor of coalitions in some parliamentary systems (Axelrod, 1970).

Another type of model was developed for capturing the exchanges that occur during a negotiation leading toward an agreement or impasse. The model consists of three parameters, the tendency to reciprocate, the tendency to make unilateral concessions or to initiate reciprocation, and the level of friendly feelings. It is evaluated by varying the values of these parameters and observing the timing and type of agreement reached. Using computer simulation, Bartos (1995) evaluated the relative merits of a concession exchange (distributive) and an informative search (integrative) approach to bargaining. This was done by having the computer generate a series or path of demands that each type of party (a distributive or integrative bargainer) would make. He concluded that the concession exchange process is faster, but the information search process may be more productive because it can increase the chances of getting an agreement.

Many conflicts and negotiations pass through phases of antagonistic and coopera­tive behavior. A key challenge is to identify the transitions that occur between these phases. This challenge has been addressed by a form of modeling that captures the dynamics of conflict, referred to as stochastic processes.

Anumber of other formal models of conflict processes have been shown to be useful. These include negotiation support systems, game and decision-theoretic formulations, and system dynamics. For summaries and examples of application, the reader is referred to the book edited by Avenhaus and Zartman (2007). Vallacher and Nowak (2007) offer an interesting new approach to modeling social- psychological processes and mechanisms with implications for resolving intractable conflicts. Referred to as a dynamical systems model, this approach is useful for capturing both gradual and sudden changes that are likely to occur in the course of an unfolding conflict (Coleman et al., 2005). Arelated form of modeling is developed by Gabbay (2007). His non-linear approach to modeling captures the internal dynamics of small-group decision making with implications for the occurrence of turning points in complex international negotiations.

Another recent modeling approach has been found to provide insights into factors that drive conflict-resolving processes. Referred to as machine learning, this approach uses decision-tree algorithms to sort through a variety of coded variables hypothesized to influence outcomes. It was shown to provide added value to statistical approaches, particu­larly with regard to contingent relationships among variables. The path-like features of the decision trees and the “if-then” feature of the association rules can reveal hidden structures in the data. For example, Druckman et al. (2006) found that enlarged joint benefits occur when friendly parties discuss only a few issues or when they discuss many issues in an asymmetrical power structure. The authors considered this finding to be a refinement over those obtained with multidimensional scaling techniques. For discussions of a variety of other computer-aided methods for conflict analysis, see the chapters in Trappl (2006).