This report examines the impact of social network over voting behavior.
THEORETICAL ARGUMENT AND HYPOTHESES
Thomas Hobbes said, “To have friends is power: for they are strengths united.” (Hobbes (1651), chapter 10, as cited by Arnold et al. (2000))
Dyad-level variables
The literature of voting behavior has increasingly adopted social network properties as explanatory variables, at both dyad- and network- levels. Amongst the first scholars who examine the correlations between dyad-level property and voting behavior, Arnold et al. (2000) posit friendship ties affect roll-call voting behavior, while shared ideological orientation, shared gender, and distance between districts do not, based on evidence from Ohio House of Representatives; Nieuwbeerta et al. (2000) argue intimate contact affects voting behavior, while similarity between ego and alter do not, based on empirical findings from the Netherland. In brief, tie formation affects voting behavior, while shared attributes across nodes do not.
Despite friendship being one of the critical factors to voting, Arnold et al. (2000) state that party has the most effect. This correlation has been confirmed by social network researchers using roll-call voting ties in various parliament sessions. For one, Porter et al. (2005) posit eigenvectors predict party affiliation, based on evidence from committees in the US House of Representatives. For the other, Dal Maso et al. (2015) argue degree of closeness predicts party affliction, based on evidence from the Chamber of Deputies of the Italian Parliament. Furthermore, Dalege et al. (2017) postulate a causality between voter’s attitude network property and his voting decision. Using data from the US presidential elections from 1980 to 2012, they demonstrate Average Shortest Path Length (ASPL) and closeness (i.e., shortest path between lengths between nodes) impact voting behavior. In a few words, a node’s closeness affects its preference (i.e., attitude and party affiliation) which shapes its voting behavior. Therefore, we present the Hypothesis 1 as:
Hypothesis 1: The higher a node’s degree of closeness is, the more likely it wins election.
Network-level variables
Recent literature on voting behavior adopts network-level properties as explanatory variables. For instance, Derr et al. (2018) argue that network structure of length 4 cycle (e.g., butterfly) is “the simplest cohesive higher-order structure and also a complete biclique” according to US congressional vote analysis. Also, Rinscheid (2020) argues businesses’ discourse connections to political parties lead to the rejection of denuclearization initiative in Switzerland. To put it briefly, the cohesiveness of a network is positively related to its homogeneity in voting behaviors. Hence, we present the Hypothesis 2 as:
Hypothesis 2: The more cohesive a node’s network is, the more likely it wins the election.
REFERENCE
1. Arnold LW, Deen RE, Patterson SC. 2000. Friendship and votes: The impact of interpersonal ties on legislative decision making. State and Local Government Review 32(2): 142-147.
2. Dal Maso C, Pompa G, Puliga M, Riotta G, Chessa A. 2015. Voting Behavior, Coalitions and Government Strength through a Complex Network Analysis. PLOS ONE 9(12): e116046.
3. Dalege J, Borsboom D, van Harreveld F, Waldorp LJ, van der Maas HL. 2017. Network structure explains the impact of attitudes on voting decisions. Scientific reports 7(1): 1-11.
4. Derr T, Tang J. 2018. Congressional vote analysis using signed networks. In Proceedings of the 2018 IEEE International Conference on Data Mining Workshops (ICDMW).
5. Nieuwbeerta P, Flap H. 2000. Crosscutting social circles and political choice: Effects of personal network composition on voting behavior in The Netherlands. Social Networks 22(4): 313-335.
6. Porter MA, Mucha PJ, Newman MEJ, Warmbrand CM. 2005. A network analysis of committees in the U.S. House of Representatives. Proceedings of the National Academy of Sciences of the United States of America 102(20): 7057-7062.
7. Rinscheid A. 2020. Business power in noisy politics: An exploration based on discourse network analysis and survey data. Politics and Governance 8(2): 286-297.