Development of Parametric Simulation Models for Structural Analysis of Voting Behaviors in Public Referendum

Abstract

Parametric simulation models are developed for structural analysis of voting behaviors in public referendum. By decomposing the residents into eight groups, a mechanism is established to construct transition probability matrices defined on three states (0: undecided; 1: YES; 2: NO), thereby capturing behavioral patterns of the residents in forming their individual opinions toward the voting date. This approach enables one to devise a strategy concerning how to transform the formation of the eight residential groups so as to achieve a target voting result. The validity is tested through eight real cases of Japan and a mock public referendum.