Normally, I use this space to run down various news items related to the campaign, but today I am going to deviate from that practice to offer a snapshot of where I think the midterm battle stands.
The place to start is with independent voters, as they are the ones who swing elections in contemporary America. Indeed, as the parties have become more polarized, independents are now more important politically than ever before. The following is an average of the independent vote in the generic ballot polls in the RealClearPolitics average:
This is a major problem for the Democratic party. It is averaging less than one out of three independent/unaffiliated voters. Importantly, all of these polls indicate that party voters have generally sorted themselves out – Republican voters have already broken to Republicans, Democrats to Democrats. Most of the undecided voters at this point are independents.
It’s hard to say how the remaining independents will finally break, though I think the best case scenario for the Democrats is that they break evenly between the two parties. The worst case would be if they break 9:1 or thereabouts for the Republicans. My guess is that the remaining independents break roughly as the decided ones have broken, on about a 4:3 basis toward the GOP.
This would create a result among independent voters that is basically indistinguishable from 1994 – about 56 percent to the Republicans, 40 percent to the Democrats, and a smattering to third-party candidates. By the way, this is also about the inverse of how independents broke in 2006. What's more, Gallup has generally found President Obama's job approval with independents around 40 percent, so it stands to reason that this is how well the Democrats will do with independents next month.
But where will the final results stand? The polls are all over the place – Newsweek says the Dems have a 5 percent lead, while CNN says the GOP is running away with it. These differences are not really due to variations in the independent vote, which as you can see above is pretty stable across polls. It is mostly due to variations in the relative strength of both parties in any given sample – how many Republicans are in the sample, how many Democrats, and how well both sides perform with their partisans.
This is where polls this cycle can really lead us astray. In particular, that Newsweek poll, which finds the GOP down 5 overall but up 17 with independents, should raise some alarms. The only way to square a D+5 national result with an R+17 independent result is to predict a heavily Democratic electorate, which is exactly what Newsweek did. But in reality the partisan composition of the midterm electorate has been quite stable for the last four midterms, with both sides being roughly equally proportioned.
Of course, small changes in the partisan composition of the electorate could have a big impact on the House. While the independent vote has been pretty stable for some time, Democrats are working furiously to drive up enthusiasm among their own partisans. What kind of result could their efforts produce, if we assume that independents remain unchanged?
To get an answer to this question, let’s map out a couple scenarios for the final 2010 House result and assume that the independents break like they did in 1994 – 56 percent Republican to 40 percent Democratic.
To start, let’s use the party breakdowns from 2006, which was a year when Democratic enthusiasm outpaced Republican enthusiasm. The GOP still got out its base vote well enough, but the Democrats had the Big Mo, and it showed up as a slight edge in party identification. What if we give the Democrats a similar party ID edge, but assume independents go Republican? We’d get this result:
To me, this is a best-case scenario for the Democrats. In such a situation, they could indeed hold the House, as Republicans might “waste” their national edge in the popular vote by running up the score in heavily Republican districts (or turning heavily Democratic districts from D+15 to D+5). This would, however, be a Pyrrhic victory, as the liberals would have responsibility without power. A coalition of Republicans and moderate-to-conservative Democrats would put an end to the Obama agenda while the liberals would get all the blame for the inactivity.
But this fails to account for how agitated the Republican base is. So, let’s run another simulation keeping the independent vote breaking heavily towards the GOP, but the party spreads reflecting the 2004 House results. That’s one in which the Democrats and Republicans were amped:
This is where we start to approach a 1994-style result. In this scenario, we really see the effect of the independent vote. The GOP gets its vote out, the Democratic party gets its vote out, and independents swing the popular vote toward the Republicans. A 6-point victory would be just a little bit off the 6.7-point edge the GOP pulled in 1994, and it is all but certain to deliver a House majority to the Republicans.
But even a 2004-style party spread does not fully take into account the current enthusiasm gap. How can we factor that in?
It’s hard to go back in time to find a corresponding election for comparison. The number of self-identified Republicans in the electorate has increased in the last 30 years as the South has realigned: when Ronald Reagan defeated Jimmy Carter in 1980, the Democrats outnumbered Republicans by nearly 13 percent on Election Day; a quarter century later, when George W. Bush defeated John Kerry, the two sides were at parity. The trouble is that we have not seen a big GOP blowout midterm since this party realignment was completed. Not even 1994 works for this purpose, as the Democrats still outnumbered Republicans in the electorate. The closest midterm we have is probably 2002, but the exit polls were all screwed up that year.
There is another way to approach the question. In 2008, Virginia’s statewide party breakdown was basically a mirror image of the national breakdown. This is in keeping with the longstanding expectations of the emerging Democratic majority proponents, who predicted (accurately, in this case) that Virginia was turning from red to purple. But in the 2009 off-off-year gubernatorial election in the Old Dominion, we saw the enthusiasm gap in action, and the GOP opened up a small party identification lead over the Democrats. So, let’s suppose that the shift in party strength this year mimics the shift seen in Virginia in 2009. We’ll also continue to use our 56-40 GOP advantage among independents:
The only people who can recall a result so pro-Republican are nonagenarians, as nothing like this has happened since 1928. How many seats this would produce is very hard to say, but I’ll put it this way: Republicans picked up 52 seats in 1994 with a +6.7 win; a +10 win should produce more than 52 seats, perhaps many more.
There are of course difficulties with using the 2009 off-off-year election in Virginia to compare to the 2010 off-year election nationwide. The biggest is that second “off.” Virginia turnout last year was a bit lower than it was in 2006, which might have favored the Republicans in 2009. On the other hand, the partisan composition of the 2006 electorate in Virginia was roughly identical to the 2009 electorate. So, it’s hard to say which party was favored by the reduced turnout in 2009. Also, the independent vote in Virginia 2009 broke much more heavily to the GOP than I’ve modeled here (ditto in New Jersey), and it’s possible that this could happen nationwide as well.
If there were a clean comparison between 2010 and some previous national midterm, I would use it over the Virginia 2009 results. However, this hypothetical still has merit. Its point is to show that it doesn’t take much of a change in the party spreads to get to a history-making Republican victory when you assume that the GOP will have a healthy lead among independents. And in evaluating the reasonableness of Hypothetical #3, consider it this way: 2004 was a year in which both parties were about equally motivated to cast ballots; 2006 was a year in which the playing field was slightly, but not overwhelmingly, tilted toward the Democrats; Hypothetical #3 is a little more pro-Republican than 2006 was pro-Democratic. What's more, Gallup has recently found that self-identified Republicans and Republican leaners outnumber Democrats by 6 points, the largest number since immediately after the 2004 November election. An R +3 electorate akin to Hypothetical #3 is not an unreasonable proposition at this point.
I’m not predicting this is going to happen. My point here is much more modest: this kind of +10 result is a distinct possibility, not just a flight of fancy. The assumptions it requires are based on hard data - from the polls, previous results in a bellwether state, and recent midterm returns - not on pie-in-the-sky. I think it helps explain why, with just a month to go, Democrats seem to have given up on appealing to non-partisans and instead are working furiously to gin up the party base with lots of red meat. It explains why they are shifting money from Kentucky to Connecticut. It explains why Gallup is talking about a double digit GOP win.