Aleksejus Kononovicius

While each vote counts, the probability that any single vote would be significant is extremely small. Thus from the rational agent’s perspective there are few reasons to vote, as the others will either way make decision for you. Would all agents be this rational, then holding an election and competing for votes would be impossible. Few early game theoretic approaches [1] seemed promising, but later works showed that game theoretic model might be impossible [2] unless certain specific conditions are met [3]. But people are seldom rational in this game theoretic sense.

In the past 30 years physicists have used tools and models from statistical physics to analyze socio-economic systems giving birth to an interdisciplinary field of sociophysics [4]. Today opinion dynamics, and voting dynamics as its proxy, remain one of the major research topics in sociophysics [5]. We contribute to this research trend by analyzing publicly available Lithuanian parliamentary election data. We consider votes cast for the parties and disregard voting for the individuals. We find that party vote share distribution, across polling stations, is very well fitted by the Beta distribution. We propose a multi-state agent-based herding model, derived from earlier proposed financial market model [6], to reproduce this observation. We draw comparisons between the proposed model and well known Voter model [7].

D. Black, The theory of committees and elections, Cambridge University Press, 1958.
R. D. McKelvey, Journal of Economic Theory 12: 472 – 482, 1976.
S. Ansolabehere and J. M. Snyder, Jr., Public Choice 103: 327 – 336, 2000.
D. Stauffer, Journal of Statistical Physics 151: 9 – 20, 2013.
Econophysics and Sociophysics: Recent Progress and Future Directions, Springer, 2017.
V. Gontis and A. Kononovicius, PLoS ONE 9: e102201, 2014.
P. Clifford and A. Sudbury, Biometrika 60: 581 – 588, 1973.