Tuesday, May 29, 2012

Congress: Partisan, not Polarized

Since this is my first post, I want to say a few words about myself, some of the themes I will stress, and then make a quick point about parties in Congress.  Me:  I currently work in Miami, but (like Hans) I am from the Pacific Northwest. I went to Willamette University, worked in the U.S. House of Representatives, and then Ph.D.'d at UCLA. I have written a book on filibustering called, simply enough, Filibustering. Matthew Lebo and I are writing a book on Congressional parties, which provides fodder for this blog. My most widely-read work, however, is on the Death Star

Like Hans and Seth, I am interested in the connections between lobbyists, interest groups, donors, and political parties, and inn the effect of the nomination and election process on representation and governance. I am especially interested in Congressional parties. A lot of the academic work and media coverage of Congress attributes the conflict and gridlock we observe to "polarization"--the notion that Republicans and Democrats represent starkly different world views and the gaps between them are too vast to leap across. I find this unpersuasive. While it is certainly true that the Republican and Democratic parties tend to nominate candidates with different views on a range of "hot" issues, this does not explain the nature and extent of the partisanship we see in Congress. So my posts will tend to emphasize these themes:
  • Partisan conflict in Congress is a strategic choice: to a large extent, Republicans and Democrats fight because they want to.
  • Much of the "polarization" we observe is actually teamwork: legislators suppress their disagreements with their co-partisans so they can compete with the opposing party. 
  • A lot of the "legislating" that goes on in Congress consists of one party seeking to improve its reputation at the opposing party's expense, rather than actually striving to, you know, make the world a better place or the U.S. a stronger nation. This is not news to regular Congress-watchers but it is at odds with a lot of the research that we do on Congress.
  • Parties manipulate legislative rules to gain competitive advantages and, especially, to shield their intra-party differences from public view.
  • Congressional partisanship actually has negative consequences for public policy.
Next:  Congressional Partisanship in Scatterplot Form!

Congressional Partisanship in Scatterplot Form!

There are a lot of statistics used to measure Congressional partisanship, but I think it helps to see what it looks like. One way to do so is in scatterplot displays of the percentage of each party voting "aye" on a roll call. In this form, we can see how often there was bipartisan consensus, partisan disagreement, or one party united while the other split (asymmetric partisanship). Most important for this series, we can see how often both parties were internally divided.  Here's a schematic for "reading" the scatterplots.

I can make these scatterplots for any chamber or period of time, but I thought we could start with the U.S. House of Representatives during the 111th Congress (2009-2010) and some comparison periods of united Democratic control of Congress and the White House:  the 103rd Congress (1993-4), 95th Congress (1977-8), and the 89th Congress (1965-6). Let's start with the 89th, elected in the landslide of 1964 and charged with a historic agenda: the Voting Rights Act, Medicare, Vietnam, and many more issues.

89th Congress (1965-6) (295D, 140R)

Three patterns to note: 1) there is a cluster of consensus at the top right 2) the House held a lot of votes on issues that united about 80% of the Democratic party while splitting off the remaining 20% or so and 3) there is a smattering of votes in the middle of the scatterplot on issues that divide both parties.

95th Congress (1977-8) (292D, 143R)

Patterns: 1) by the late 1970s, the House held a lot more votes overall; 2) these votes were all over the map; there were lots of votes that cleaved each party in two. 

103rd Congress (1993-4) (258D-176R)
In the early 1980s, the House majority party began to restrict the issues that the House votes upon, a pattern continued by both Democrats and Republicans. Combined with shifts in the constituencies of the two parties, this behavior reduces the number of votes in the middle of the scatterplot, with more votes in the partisan corners. Still, many issues come to a vote that cleave the Democratic majority in two. As Hans might point out, we should expect this when we cobble together a political party out of many disparate "factions." It's impossible to be unanimous when this country--and thus the constituency base of any national party--is inherently diverse...right?

111th Congress (2009-2010) (257D-178R)

Wrong. This is how the 111th Congress looked. The middle of the chart is virtually empty. At least 90% of the Democrats stuck together on almost every vote, so all the dots are along the left or right side. The Republicans divided on many of these votes (more than I expected) so there was plenty of asymmetric partisanship. 

One way to read these charts is to take them at face value:  over time, the Democrats became completely unified while the Republicans are a diverse coalition of members willing to join with the Democrats as their consciences and constituents dictated. Another way to understand Congressional partisanship is to treat the behavior we see as part of a larger game in which the set of issues that come up is carefully manipulated (to the extent that the rules allow it) and members of each party coordinate their votes with other members of their partisan teams. I take the latter view. I suspect many Congress-watchers do. And one of my goals on this blog is to try to understand and explain how the game is played.


  1. I may have missed it but do these scatterplots take in all votes in the House or do you weed out the insignificant resolutions that both parties support? Also, how would polarization appear differently in these scatterplots from partisanship.

    1. Rob-both excellent questions. These plots show ALL votes. I thought about refining a bit--e.g. just amendments, passage, & other policy votes--but I wanted to keep it simple on the first pass. I also wanted to do a companion set for the Senate, but tempus fugit.

      As for polarization vs. partisanship, the differences tend to be clear in the specific but hard to tease out with a first cut at aggregate data. The key point above is that the center of the scatterplots is hollowing out. One can explain this pattern in terms of partisan strategy and agenda manipulation, or in terms of distributions of preferences, proposals, and status quos. Personally, I think the latter approach requires some very strong assumptions, but a clean test will have to wait until later posts.

  2. nice posting.. thanks for sharing..