2006 Random Statistics

I've been working on a paper evaluating the 2006 election results for Senate and Governor, and I wanted to pass along a few tidbits, since I'm not sure they've been reported on with any detail to this point.

First off, Democrats won 24 of the 33 Senate contests, and they captured about 55.4% of the national 2-party vote. If the U.S. electoral system was based on proportional representation, then, the Democrats would have been allocated 18 seats, and the Republicans would have taken 15 seats. Thus, the quirks of Senate apportionment gave the Dems an extra 6 seats in the 2006 cycle. This follows a well-established trend: Senate apportionment has given one party or the other extra seats in all but two elections since 1900. And more often than not, Senate elections tend to have a counter-majoritarian trend: the party that wins a larger proportion of the vote usually underperforms in Senate races. The opposite was true in 2006.

My gut sense has long been that Senate apportionment hurts Democrats; for example, if Republicans were able to hold both Senate seats in each of the "red" states won by Bush in 2004, they would hold filibuster-proof majorities, even with the popular vote nearly split down the middle. But, for some reason, Democratic Senate candidates have managed to be competitive in hostile terrain. They have been able to win in deep-red states like Nebraska, Montana, North and South Dakota, West Virginia, Indiana, and so forth. Looking over the big picture, it's easy to see why the Rove polarization strategy is so effective: when local elections are based on nationally divisive issues, Republicans in Senate races should be able to defeat the moderate Democrats that have, to this point, been able to win in hostile territory.

The structural Republican advantage is less evident in House races and Presidential races, though even in these cases, Republican voters seem to be more strategically distributed than Democratic voters.

Easily the most surprising finding I've uncovered is this: in Senate races, states that have endured higher per capita battle deaths in Iraq had a much higher Democratic vote than states with lower per capita Iraq deaths. I haven't been able to disaggregate the data down to individual states on this yet, but "per capita Iraq battle deaths" is one of two highly statistically significant variables in predicting state voting patterns this year. The other variable, not surprisingly, is the amount of money spent by challengers: those who spent more money had a better chance of winning (poor, sad, unfortunate Ricketts and Lamont notwithstanding).

The point here is that antipathy to the Iraq war may be based less on national media reporting than it is based on the localized impact of casualties. That's not what I would have expected.

In races for governor, Democrats won 20 of 36 seats in 2006, garnering more than 53 percent of the national vote. Here, a PR system would have yielded just over 19 seats to Democrats; thus, the gubernatorial elections did not seem to have much impact from apportionment by state.

Add it all up: a weird election, in which Democrats took advantage of a medium-sized wave, and won every close race they needed to win to take the Senate. Iraq was the predominant issue in the Senate campaigns, and an anti-Republican mood percolated even into Governor's races.

I am now done holding this blog hostage.