Testing the Spatial Voting Theories on a Computational Model: A Demonstration of the Polarizing Influence of Directional Voting

Abstract

This study proposes a Gaussian utility function for the proximity theory so that it only describes the variation of party affinity of the voter with the ideological distance from the party. Eliminating the repulsive portion of its currently adopted utility function distinguishes the proximity theory from the directional theory. Instead of trying to fit actual voting behavior to either utility function, the functions are tested on an idealized computer model of the voters’ opinion space to test their suitability. By testing varying linear combination of the utility functions of the two theories, optimum locations of parties are found that will maximize the average value of the combined voter utility function. Results indicate that catch-all parties with large voter bases choose locations near the neutral center, but increasing the contribution from the directional utility function pushes the parties further away from the neutral center. These findings support the observed relationship between ideological polarization and the directional voting tendency. The conclusion is that a linear combination of the two utility functions appear to be more suitable to explain idealized voter behavior.

Keywords: spatial voting, proximity theory, directional theory, computational model, optimization

DOI: 10.7176/JSTR/6-02-04