Saturday October 4, 2008
Our former student Sean Theriault got some press:
By one measurement, Congress is the most polarized it has been in a century. Sean Theriault, a scholar at the University of Texas at Austin who just published a book called “Party Polarization in Congress,” analyzed voting patterns to put each two-year session on a scale. In his study, Congress in its Watergate session from 1973-74 was 29 percent polarized. By 2005-6, it was 46 percent, the highest since the most polarized Congress in history, back in 1905-6, when it reached 48 percent on Theriault’s scale.
“The electoral campaign has infiltrated the legislative process,” Theriault told me. “Congressmen used to campaign at home, win elections and then come to Washington” to grapple with the issues of the day. Now, he said, “They’re just looking to gain advantage wherever they can.”
I’m not sure what “29 percent polarized” means; I’m sure its well explained in Sean’s book. But all these indicators tend to say the same thing: more party line voting, bigger distances between, say, the median House Dem and the median House Rep on the scales we recover when fitting ideal point models etc.
Still, the distress over partisanship in the Congress strikes me as odd given my upbringing in a parliamentary system. That is, if you want to see what polarization really looks like, look at the Australian House of Representatives, the British House of Commons, etc. “Crossing the floor” is a rare phenomenon in the Australian House of Reps: looking over 50 years, you’ll see someone cross the floor in 2% of all divisions in the Australian House, and this rate dropped to just 0.3% of all divisions in the 1996-2005 period. Kinda makes roll call analysis and computing polarization measures redundant (there is a little more to look at with voting in the Australian Senate). Sure, the institutional configuration is very different, with the executive being drawn from the parliament, requiring the support of a majority of the parliament, and so on. But hey, you want responsible parties? Well, you got ‘em (or so it would seem).
So this hand-wringing presents something of a puzzle to me, and others. What exactly is the “null model” or “baseline model” here? What is the level of party-line voting we ought to expect, ceteris paribus, whatever the ceteris might be? That is, what patterns of voting in the Congress ought we expect to see if members were simply showing up and voting the positions of the median voter in their respective districts?
These questions have sustained more than one or two Stanford dissertations over the last few years, and probably a few more to come.