The 113th Congress was sworn in this month, installing 47 new Democrats and 32 new Republicans. Democrats gained eight seats in the House, but failed to capture the majority. Because of newly drawn districts, Republicans hold 54% of the House despite having won only 48% of the vote. In the Senate, Democrats gained two seats. With 55 Senators, the Democrats still hold the majority, but again fall short of the 60 votes needed to surpass the ever present threat of filibuster.
So what has the election changed, if party control of Congress remains the same? The ideological composition did.
To place this ideological shift in context, I’ve put together a graph that plots the ideological position of every member of Congress from the 73rd (elected in 1932) to the 113th (elected in 2012). Democrats are shown in blue, and Republicans in red, arrayed from left to right based on their observed voting behavior. (Click to enlarge.)
Looking at it this way, the recent polarization of Congress is starkly evident in the unprecedented, widening void between the parties. Though it was once common for there to be some ideological overlap between the parties, that hasn’t been the case since 2004. The polarization is deeper than just party affiliation, though. The centrists of each chamber have been eroded over the course of the last decade. In the election of 1994, the extreme edge of the House Republicans lurched to the right, eventually bringing along more and more of its members in both chambers.
The other element of the graph is a metric of my own calculation. The shaded boxes indicate the narrowest majority coalition possible in each session, ideologically. Every box represents a majority of members spread across a particular ideological range. In the last four sessions of the Senate, in recognition of the increased use of the filibuster, I’ve also shown the narrowest sixty vote coalition. Conceptually, it is more difficult to pass a bill with a wider coalition, because you need to get more ideologically disparate people to agree.
Considering the coalitions side-by-side, you can see how difficult it must be to craft a bill that can at the same time satisfy a majority of the House, a super majority of the Senate, and the White House. Looking back on the last century, no other Congress was as divided as the 112th. Unless the Senate reforms the filibuster and the House can routinely hold votes that cross the partisan gulf like the fiscal cliff vote did, the 113th Congress looks to be just as divided.
This graph is based on the DW-NOMINATE Common Space data, which has meticulously analyzed the votes cast by each member of Congress, all the way back to the first. Because so many members serve for multiple terms, and sometimes in both chambers, it’s possible to compare the ideological composition of each Congress, arraying every member on a left-right axis. (The data set also includes a second dimension, corresponding roughly to North-South, which only seemed to be important during the Civil War and during the struggle for Civil Rights. However, there have been some interesting/frightening indications that this second axis is making a comeback.)
The calculation of the narrowest coalitions depends only on the first dimension of ideology. Because of the overlapping ideology of the parties in years past, and the onetime importance of the second dimension, the coalitions I have highlighted are not always found solely among the majority party. Also, to accommodate the intra-session changes in membership when calculating coalitions (e.g., due to resignations or death), I weighted each member (and the Vice President) by number of votes cast.
For the 113th Congress, direct observation of ideology is impossible, because they haven’t been casting votes. In many cases, however, we do have past observations: members were not up for reelection (in the Senate), or were reelected, or had served in prior years, or had served in the other chamber. For the newly elected members, it is possible to forecast their ideology based on a regression of past voting behavior by party, sex, self-identification as a member of the Tea Party, representative from a state of the South, and the share of votes for the President in each district or state. Here are the regression results, from the team that maintains the DW-NOMINATE data set, for the House and the Senate.
The Effects of the 2012 Election
- House Democrats expanded the size of their caucus, but seem to have further concentrated its ideological composition.
- House Republicans continued shifting right (although they did lose some of their most conservative members). This, in addition to the loss of some members, has widened their narrowest majority coalition, requiring votes from even more of the far right. The changes to the caucus do not seem to have made Speaker Boehner’s job any easier.
- Senate Republicans have also shifted further right, adding conservatives Jeff Flake (R-AZ), Ted Cruz (R-TX), and Deb Fischer (R-NE). Having lost Olympia Snowe (R-ME), Scott Brown (R-MA), and Ben Nelson (D-NE), the centrists Senators now number just one: Susan Collins (R-ME).
- Senate Democrats expanded their majority. From left to right, they added Tammy Baldwin (D-WI), Mazie Hirono (D-HI), Chris Murphy (D-CT), Angus King (I-ME), Martin Heinrich (D-NM), Tim Kaine (D-VA), Elizabeth Warren (D-MA), Joe Donnelly (D-IN), and Heidi Heitkamp (D-ND). These shifts mean that their narrowest majority coalition has actually grown a little broader, but still has plenty of latitude on either side. Due to the loss of the centrist Republicans, the narrowest sixty vote coalition now reaches further to the right, covering 17 of the 45 Republicans, as far as Mitch McConnell (R-KY). The election does not seem to have removed any of the reasons to reform the filibuster.
I have previously commented on another attempt to display this data by the XKCD web comic. Although I love XKCD’s penchant for data display, I had some problems with his design choices in this case. I chose to represent ideology, which is a continuous variable, on a continuous scale from left to right, unlike XKCD’s categorical approach with six shades of color. The XKCD approach more clearly showed the number of members from each party by stacking them left and right. I tried to represent the same information less precisely by overlaying transparent markers that indicate the density of each caucus, and also by highlighting the members that constitute a majority. I hope that you find this approach a little easier to digest.