Modeling of the parties' vote share distributions

TL;DR

This paper investigates the statistical properties of vote share distributions in Lithuanian elections, proposing that the beta distribution is a suitable model due to its flexibility and empirical fit.

Contribution

It introduces a simple agent-based model and argues for the beta distribution as the best fit for vote share data, filling a gap in sociophysics opinion dynamics.

Findings

Vote share distributions are often well fitted by multiple distributions.

The beta distribution provides a flexible and effective fit for vote share data.

A simple agent-based model supports the beta distribution's suitability.

Abstract

Competition between varying ideas, people and institutions fuels the dynamics of socio-economic systems. Numerous analyses of the empirical data extracted from different financial markets have established a consistent set of stylized facts describing statistical signatures of the competition in the financial markets. Having an established and consistent set of stylized facts helps to set clear goals for theoretical models to achieve. Despite similar abundance of empirical analyses in sociophysics, there is no consistent set of stylized facts describing the opinion dynamics. In this contribution we consider the parties' vote share distributions observed during the Lithuanian parliamentary elections. We show that most of the time empirical vote share distributions could be well fitted by numerous different distributions. While discussing this peculiarity we provide arguments, including a simple agent-based model, on why the beta distribution could be the best choice to fit the parties' vote share distributions.