Modeling the evolution of ideological landscapes through opinion dynamics

Abstract

This paper explores the possibilities to explain the stylized facts of empirically observed ideological landscapes through the bounded confidence model of opinion dynamics. Empirically, left-right self-placements are often not normally distributed but have multiple peaks (e.g., extreme-left-center-right-extreme). Some stylized facts are extracted from histograms from the European Social Survey. In the bounded confidence model, agents repeatedly adjust their ideological position in their ideological neighborhood. As an extension of the classical model, agents sometimes completely reassess their opinion depending on their ideological openness and their propensity for reassessment, respectively. Simulations show that this leads to the emergence of clustered ideological landscapes similar to the ones observed empirically. However, not all stylized facts of real world ideological landscapes can be reproduced with the model. Changes in the model parameters show that the ideological landscapes are susceptible to interesting slow and abrupt changes. A long term goal is to integrate models of opinion dynamics into the classical spatial model of electoral competition as a dynamic element taking into account that voters themselves shape the political landscape by adjusting their positions and preferences through interaction.