The Numbers on Party Polarization

Since 2010, the media have constantly harped on Congress’s inability to do anything productive, blaming it on a spike in party polarization due to right-wing movements such as the Tea Party and a lack of cooperation. Data indicate party polarization has gradually increased over the past 20 years with two spikes in 1994 and 2010 and is the responsibility of both the Democratic and the Republican Parties.  However, the numbers indicate that the traditional correlation between party polarization and Congressional gridlock is decreasing in importance, possibly indicating the relevance of other factors in causing today’s dysfunctionality.

Data

The DW-NOMINATE score was used to track the ideological slant of every representative over the past 20 years.[1] This system tracks voting records and assigns a number to each representative based on who they vote with most often; economic conservatives receive positive scores and liberals negative scores. This score ignores subjective factors such as which representative proposed, spoke in favor of or against, or obstructed a particular bill. Only the voting record is considered, because ultimately the voting record determines how the representative’s ideology affects the real world. This score has been proven to predict how a representative will vote in the future.

In order to quantify party polarization, two measures were used: the standard deviation (a measure of the spread of Congress), and the difference between the means of the Republican and Democratic representatives.[2] In theory the standard deviation accounts for increasing polarization within the parties as well as the difference in opinion between the parties, so it summarizes more information about party polarization.[3] However, the standard deviation and the difference between the means had a high linear correlation[4] so by Occam’s razor the difference between the means was used to analyze the data.

Does Party Polarization Exist?

Contrary to popular perceptions, the polarization in Congress today is the result of a gradual increase over the past 20 years, which was exacerbated by a sharp increase in the 2010 midterms. Between the 104^th and 111^th Congresses, or the years 1995 and 2011, the difference between the Republican and Democratic parties’ mean increased from 0.77 to 0.82. Roughly one-third of the increase between 1993 and 2013 can be attributed to very slow movements of the parties away from each other.

The media correctly assert that the 112^th Congress from 2011 to 2013 was significantly more polarized than any Congress that had preceded it in the last 20 years due to the Tea Party movement. The difference between the parties increased by 0.06, more than it had increased in the last 16 years. However, such sharp increases have occurred before; a similar one occurred just prior to the 104^th Congress elected in the 1994 midterms in the Republican Revolution.

The Tea Party movement is commonly blamed for the increase in Congress’s polarization in 2010; however, the data indicate that both parties became significantly more polarized in that election. Both parties had their means change by roughly 0.3 points, so they are equally responsible for Congress’s current polarization. The Republican Revolution in the 1994 midterms caused a similar shift in both parties’ means.

In both 1994 and 2010, an extreme right-wing movement sparked a response from the Democratic Party that resulted in increased polarization. Democratic representatives changed their votes and ideologies to align more closely with the party platform, while many Democratic moderates lost their seats during the Republican wave. As a result, the members that remained in Congress tended to align more closely together as a Democratic bloc. Thus both parties became increasingly polarized due to a right-wing conservative movement.

The Effects of Party Polarization

The media today commonly assert that party polarization is the sole cause of Congressional gridlock. The most cited data-based analysis of this was done by Binder et al. in the early 2000s and also used DW-NOMINATE data to quantify party polarization. The Binder study tested a number of possible causes of Congressional gridlock and found that bicameral divide, i.e. the difference in opinion between the House and the Senate, and party polarization were the primary causes of gridlock, but these causes were not sufficient to explain gridlock.[5]

Here Binder’s methodology is updated to include the most recent data. The method of measuring Congressional gridlock is also changed. Binder’s method uses the human judgments of the reporters at the New York Times to determine what legislation is significant; in order to avoid any subjectivity, the number of bills passed was used to measure Congressional productivity (the opposite of gridlock). A linear fit was then performed with respect to both the bicameral divide and party polarization; very low correlation was observed.[6] Normalizing[7] for just the effects of party polarization did not improve the correlation;[8] this normalization is shown below.

The data indicate that since the time of Binder’s study, party polarization has lost some of its ability to predict Congressional gridlock. Given this pattern, it’s likely that other external factors have come into play, reducing the strength of the relationship between polarization and gridlock. It is Congress’s duty to its electorate to identify and reduce these factors, as they prevent Congress from serving its people appropriately.