Understanding dynamic behavior of swing voters during election campaign period using agent-based simulation

摘要

Election is the ultimate event in which people express themselves to get the most appropriate leader based on majority interest. Among the campaign period, the interactions between swing voters make the election’s result unpredictable. This paper aims to measure the dynamic behavior among swing voters which is affected by the influence from people around the individual, the electability level of a candidate, and the loyalty threshold of a candidate’s voters. The purpose of this paper is to provide a simulation that accommodates all changes in continuous time, so the election result can be predicted dynamically. The influence factors and several additional settings will be analyzed carefully to provide a comprehensive simulation. Furthermore, the proposed setting will be used to test two hypotheses which are commonly known among election analysts. Moreover, the result will give a dynamic simulation which brings a new approach to understand the impact of applied campaign strategy. In other hand, this study also throws brighter understanding to the dynamic behavior of voters and gives feedback for campaign teams to develop a better strategy for increasing their candidates’ competitiveness.