Following are the top ten states where I think we could use some fresh polling data.
10. Illinois. All the other big states get polled -- even the ones that aren't expected to be especially competitive. But Illinois seems to be taken for granted and hasn't been sampled since February. While that's understandable in Obama's case, the regression model thinks there's an outside chance that McCain could compete against Clinton in Illinois if there were a sufficient amount of backlash from Obama supporters. It would be nice to be able to test that hypothesis.
9. Connecticut. Another state that's more interesting in the Clinton scenario than the Obama scenario, but if Clinton is making progress with the white working class, it would be helpful to know if she's losing ground with the investor class. There's no better place to evaluate that than Connecticut, and previous polling had shown the Clinton-McCain matchup relatively close in this state.
7. North Dakota and 8. South Dakota. Every day that goes by, our model gives less weight to the tantalizing SurveyUSA poll from February that showed Obama four points ahead in North Dakota, and therefore his win percentage goes down there. But the Dakotas have decidedly moderate electorates, and with Obama polling extremely well in neighboring Minnesota recently, we're overdue for a check-up.
6. Massachusetts. SurveyUSA data has sometimes shown Massachusetts to be surprisingly close in the McCain-Obama matchup, whereas the only Rasmussen poll put Obama safely ahead. Getting a third opinion from a Suffolk or a University of New Hampshire would be helpful.
5. Montana. While Alaska has been polled surprisingly heavily, Montana really has not been. And yet, it might be the more plausible of the Big Sky states as it has more history of electing Democrats -- including darkhorse VP candidate Brian Schweitzer as well as John Tester and Bill Clinton (in 1992).
4. South Carolina. The Palmetto state has been polled just once, and that poll put Obama just three points behind. While we can certainly make some inferences about South Carolina based on the polling in North Carolina and Georgia, this state had a somewhat special relationship in securing the nominations of both McCain and Obama, and it would be interesting to examine the localized effects.
3. West Virginia. I was disappointed that we had a meaningful primary in West Virginia and yet didn't get a single general election poll out of it. While I don't expect the state to be competitive for Obama, it is so idiosyncratic demographically that it would help us to calibrate our estimates in other areas. And it certainly is an interesting state for Clinton on the chance that she gets the nomination.
2. Nevada. It's received considerably less polling attention than Colorado or New Mexico, and yet is just as important as a part of Obama's potential Western strategy. Moreover, it's a difficult state to get a hold on demographically, as it combines some extremely wealthy demographics with some extremely poor ones.
1. Michigan. Although Michigan has been polled some, it lags way behind Ohio and Pennsylvania in its depth of data and yet ranks second on both Obama's and Clinton's swing state lists. Particularly in light of Obama's recent visit to the state and the continuing debate over the disposition of its Democratic delegates, it would be interesting to have a better idea of where we stand.