A first take on Election 2010

There's no question that the GOP outperformed expectations for the House last night by picking up more than sixty seats. The Douglas Hibbs model, which doesn't include contemporaneous political factors, predicted a 45-seat pickup (PDF); the median pre-Labor Day forecast among political scientists compiled by John Sides was 43 seats; and the median pre-Election Day forecast among 538, Stochastic Democracy, and Sam Wang was 54 seats. Personally, I expected substantial Democratic losses but not a 1994-style wave in the fall 2009-spring 2010 period before revising my expectations downward in April and again in September.

Senate forecasts are more difficult due to the small sample sizes, but going into last night the GOP was predicted to gain approximately seven seats; they appear likely to pick up six.

To put the historic nature of the House seat change in perspective, here are post-WWII seat losses by the president's party in midterm elections (I'm using the 538 estimate of a 65-seat loss when all the votes are counted):

The number of seat losses is a record in the postwar era.

However, it's important to keep in mind that the seat losses suffered by the president's party are partly a function of the number of seats held by the party entering the election. In this case, the relatively large Democratic caucus in the current Congress means that the proportion of seats lost, while still a record, is roughly comparable to 1946, 1958, 1974, and 1994:

While it's still very early, exit poll results suggest that the pro-GOP swing was relatively widely shared across demographic groups -- here are two New York Times visualizations from the slideshow currently running on their homepage:

As I emphasized yesterday, Obama's House majority was especially vulnerable due to the number of seats held by Democrats that had supported John McCain in 2008 (48 in all). The proportion of seats held by the president's party that supported the opposition's presidential candidate was the highest since Bill Clinton in 1994 (when Southern districts that supported Republicans for president started electing Republicans to Congress):

As expected, Democrats performed extremely poorly in these vulnerable seats and only performed moderately well in seats that Obama narrowly won (I'm using CNN's projected winners as of 10 AM EST - races that were still rated too close to call are excluded):

The political scientists Seth Masket and Steven Greene found that Democrats who supported health care reform in relatively conservative districts were underperforming those who had opposed it. To assess the initial support for this hypothesis, I disaggregated incumbent Democrats by their final vote on health care reform and Obama's performance in their district:

Democrats who supported health care reform did indeed seem to perform worse in the districts that supported McCain, though further analysis is required to determine whether the health care reform was responsible for that difference. (It might be, for instance, that anti-HCR Democrats compiled an overall legislative record that was a better fit for their districts.)

Finally, given claims that the Tea Party cost the GOP Senate seats in Nevada and Delaware, it's worth briefly following up on my post on Tea Party-endorsed candidates for the House. I previously found that Tea Party groups backed relatively experienced candidates in competitive House races. When we disaggregate Tea Party-endorsed candidates by race type (challenger or open seat) and Obama's vote in the district, we see a similar pattern -- Tea Party-endorsed GOP candidates won at a similar rate to other Republicans:

As I argued yesterday, it's important to keep these results in perspective. While Republicans face a relatively favorable path to retaining control of the House and gaining control of the Senate going forward, this election is not a realignment and will not create a permanent Republican majority any more than the 2006 and 2008 elections did for Democrats.

[Cross-posted to HuffPost Pollster]

Comments

[Rob]

I agree that the 2010 election doesn't necessarily signal a realignment, and I agree that it's not possible to isolate opposition to the Democrats' health care legislation as the reason for the variance between the projected outcome and the actual outcome. However, that inquiry is a step in the right direction, because even though it may not be possible to isolate one policy difference or another as the reason for the variance, it stands to reason that voter objections to Administration policies accounted for a large part of the variance.

And of course that's just what we should hope would happen. If in a democratic society people are opposed to legislation that's enacted, it would be very disappointing if the offending legislators escaped any consequences. That's exactly why we have elections.

So most of the Democratic losses in the House were probably the result of structural factors and relatively little of the losses beyond the structural factor projections can be attributed to Obama's failure to communicate. (Brendan's convinced me on both points.) That leaves disagreements over policy as the logical major factor in the variance--except for that relatively small number who voted Republican this year because they think Obama is a Keynesian.

[JP]

Rob, thank you for posting that video. It made my day.

[Jim]

You say, "Democrats who supported health care reform did indeed seem to perform worse in the districts that supported McCain," using a crude "win rate." Wouldn't a better analysis be by percentage of the vote received vs expected percentage of the vote based on past history of their district and national trends? You treat someone losing with 49% the same as someone losing with with 30% of the vote, which doesn't make sense.

Further, you fail to say anything about your own finding that in districts where Obama won 50-60% of the vote, Democrats who voted for HCR had a higher win rate. Perhaps HCR played a role there in that higher win rate, motivating base Democratic voters. Many of these districts would also be considered competitive districts.

Further, I believe you should wait on these analyses until all the votes have been counted and winners declared. Of the remaining uncalled races, most are leaning to the Democrat. Further, the West Coast was much more favorable to Democrats even in competitive districts, so your analysis is missing something if it does not have good data from those states where the vote counting was not as progressed.

[bnyhan]

Jim -- see my new post which uses vote totals rather than win rate here: http://www.brendan-nyhan.com/blog/2010/11/beware-context-free-election-analysis.html