Using the Scenario Method for Predicting Electoral Results: The Case of Castilla y León

Abstract

This study uses the scenario methodology for predicting electoral results. The most common models for estimating opinion polls in Spain have been systematised on the basis of two parameters, the effect of time and the incorporation of uncertainty. The combination of the two parameters shapes a series of scenarios that provide insight into existing electoral structures and patterns in public opinion. It also makes it possible to conduct a sensitivity and robustness analysis of the estimates of electoral support for political parties. This was empirically applied by the CIS to the regional election in Castilla y León for diagnostic purposes and showed the existence of singular model fits by province. The estimates were very sensitive to the interaction between the parameters defined by the dynamics of change and uncertainty.