Stats geeks, rejoice: The newest DW-NOMINATE figures are out! DW-NOMINATE, devised by political scientists Keith Poole (now at the University of Georgia) and Howard Rosenthal (now at NYU), is the industry standard system for measuring how members of the House and Senate compare to each other ideologically.
The approach uses roll-call votes to plot members across two left-to-right axes: one for economic issues, and one for social/racial/regional issues. The latter is primarily of interest for analyzing civil rights politics and the movement of segregationist Democrats into the Republican party in the 1960s and '70s, so for looking at contemporary politics, analysts tend to focus on the first, or economic, dimension.
Yesterday, the DW-NOMINATE team released scores for the 112th Congress, which started in January 2011 and just wrapped up at the start of this month. They confirm what the team has found for years: the parties are moving further and further apart. The most straightforward way to measure polarization using DW-NOMINATE is to calculate the average score of each party in each chamber, and then calculate the difference between the two parties' means. The further apart the means are, the more polarized the body. And the means are moving further and further away from each other:
But, as ever, the polarization process was far from symmetrical. For example, look at how the parties' average scores in the House have changed since Reconstruction:
The Democratic caucus got more liberal this past Congress as a lot of Blue Dogs and conservative Southern Dems lost their seats in the 2010 elections, and in general the party has been getting gradually more liberal since the 1930s. And from the '30s to the '70s, the Republican caucus was slowly getting more liberal too. But around 1976, it suddenly shifted and grew rapidly more conservative, and has continued to do so ever since. The Tea Party movement doesn't appear to have sped that process up much, but then again many of its successes came in primaries to determine challengers for Democratic seats, in which those challengers (like Sharron Angle or Christine O'Donnell) went on to lose. That wouldn't show up in these figures.
This also means that there's a lot less party overlap at the middle, which becomes clear if you compare how the 10th percentiles of each party (i.e. the most liberal Republicans and most conservative Democrats) have changed. Here, for example, is how the Senate has shifted:
That's an important thing to remember when you hear politicians bemoaning that the parties can no longer sit down and hash out bipartisan deals. It used to be that you had Republicans who really did want to increase welfare spending and expand social programs, and Democrats who were passionate about reducing the size of government. That made it easy to make deals that were bipartisan on the surface but in practice backed by people of the same ideological stripes. Now that the parties are more unified, people who would have once been liberal Republicans are Democrats, and people who would have once been conservative Democrats are Republicans, so those same deals don't happen to be bipartisan even when they have the same types of folks backing them.