Before the midterm, Professor Coppock introduced Keith Poole and Harold Rosenthal’s issue-space analysis of the US Congress. We learned that Poole and Rosenthal were able to model the voting behavior of congressman, who decide on a vast spectrum of issues, with just two dimensions. Their analysis pinned down what one might consider a boundless issue space to just two dimensions because votes on one issue are highly correlated with votes on others and can be assigned to two groups (for our class example, social conservatism and economic conservatism). These groupings for congressional voting seem reasonable, especially within our two-party system, because Democratic congressmen who vote for the socially liberal position of gay marriage probably vote for abortion rights. Furthermore, Poole and Rosenthal suggest that an equilibrium decision may lie between the Democratic and Republican clusters that emerge within the two axes at the median voter’s position.
This discussion immediately reminded me of a New York Times article my father sent me in 2003 regarding the work of his friend and mathematician, Dr. Lawrence Sirovich. Using singular value decomposition and information theory, Dr. Sirovich looked for patterns in the voting behavior of the Supreme Court and found that “the decision space of the Rehnquist court requires only two dimensions for its description.” In fact, he found that each justice’s vote could be considered a combination of two vectors: unanimous decision and the predominant ruling of 5-to-4. As with Poole and Rosenthal, Sirovich’s findings “point up the high degree of correlation” in voting patterns. Sirovich suggests the cause may be “a sameness in the overall quality for cases percolating up to the court through the judicial substructure,” such as controversial issues that yield a unanimous vote like Brown vs. Board of Education or a split vote like the 2000 Presidential Election. Either way, Sirovich’s findings corroborate common observations of the Court moving toward unanimity or 5-to-4 rulings with a decisive swing, also known as median, vote (usually Justice Kennedy or O’Connor in the Rehnquist court).
I was astonished at how both Poole and Rosenthal and Dr. Sirovich could model the voting behavior of two of our branches of government in two-dimensional space. It is fascinating that both branches face numerous issues, but yield highly correlated decisions in two vectors where a median voter may be decisive. Since I think it isn’t a stretch to suggest that the court is approximately comprised of the two parties that decide its appointments, perhaps Congress and the Supreme Court’s similar voting behavior stems from an exhibition of two assumptions of the Median Voter Theorem: simple majority and two parties.
