Choosing appropriate techniques for socio-political forecasting

Abstract

After defining scientific forecasting, the crucial role of assumptions in such forecasts is explicated. This is followed by a discussion of the representations upon which forecasting systems are based. Six variables are then introduced to capture differences in socio-political forecasting circumstances: level of detail, accuracy, agreement on problem representation, robustness-brittleness, number of variables and interdependencies, and disturbance. A categorization of forecasting approaches - expert based, Bayesian, extremal statistical, and rule based - is offered. These forecasting approaches are then cross-referenced with the forecasting circumstances to produce recommendations for choosing an appropriate forecasting technique in a given policy circumstance. Most examples in the article are drawn from the realm of foreign policy and international politics, and the cross-referencing section concentrates on foreign energy policy examples.