An Application of Networks In Understanding Political Science
https://www.cse.msu.edu/~derrtyle/papers/icdmw218-congressional_analysis_signed_networks.pdf
In the study above, a network analysis of signed edges and balance theory related to votes of the US Congress was conducted. The study analyzes the votes of a yea or a nay, to gain insights into congressional voting behavior. The study mentions balance theory and explains the significance of a structurally balanced network, as one that would be expected to be balanced through social tensions, and describes unbalanced networks as having links that would normally be unexpected. It mentions that usually, like we have learned in class, the structure of a network is analyzed through the lens of triangles, a group of three connected nodes. However, the study uses the slightly more complex shape of four connected nodes, called butterflies because of their shape, to analyze the congressional data through an additional node. Like we learned about balance in triangles, there are certain criteria that can be met to assure a butterfly is balanced. In the butterflies, an even number of negatives signed edges indicates a balanced butterfly. Through the analysis, the study concludes that their findings could be used to aid in predicting voting behavior, using what could be expected in a network to predict the future. This analysis inspired my blog topic, a proposition for a simpler version of the study, a network analysis of the Senate using similar concepts we have learned in class.
With two independent, 48 democratic, and 50 republican senators, the 117th Congressional Senate has a tough time passing legislation with a majority vote, often making use of the vice president’s tiebreaker. How often do senators from the two parties agree, though? I find that this could be answered particularly well through the lens of networks, like seen in the article. If a connection for each pair of senators was denoted by a strong tie between senators who often vote similarly and a weak tie between senators who seldom vote similarly, would the network be structurally balanced? Voting similarity could simply be expressed as two senators agreeing on a majority of votes, i.e., voting both yea or both nay on more than 50% of bills, and independent senators could be grouped with democrats in a preliminary analysis due to their democratic voting tendencies.
As a complete graph, this network would have connections from each senator to every other senator. So, there are two possible scenarios where the structure could be considered balanced, both of which we have covered in lecture. First, the structure would be balanced if all connections were positive, requiring all senators to have similar voting tendencies. On the other hand, the second and more probably way the network could be balanced is by having two groups that have only positive connections internally and only negative connections with members of the other group. Therefore, in order for the structure of the network to be stable by this definition, it must be proven that no republican and democratic senators votes similarly more than half of the time and senators agree on votes a majority of the time within their party, which could easily be analyzed with senate voting data. Proving the division between democrats and republicans through networks could aid in the study of political science to learn more about the voting behavior in such important political systems.