Jon Chait and Nate Cohn look at the latest polls and both approach the issue of race and turnout from different perspectives. First Chait:
The thing I’ve been harping on for months is racial composition. Obama has held very steady support among non-white voters, so the key factor is his share of the white vote. Where polls have tended to differ is how many white people will turn up at the polls. The ABC/Post poll shows Obama leading by six points among registered voters — a nice bounce, in line with what other polls show — but only a single point among likely voters. Why is that? Because the Post’s likely voter screen produces an electorate that’s about (according to political scientist Alan Abramowitz, via email) 80 perecent white — even whiter than the 2010 off-year elections, and off-year elections are always whiter than presidential elections. The fact that the Post’s likely voter screen is screening out massive numbers of non-white voters suggests that Obama may have work to do to turn them out, but it isn’t showing us a very plausible portrait of the electorate.
Both campaigns seem to be assuming an electorate that is about as diverse as the one four years ago — non-white voters will probably be a little less jazzed to vote, but more of them will be eligible. That assumption, combined with Obama’s steady share of the non-white vote, produces a break-even point of around 60 percent of the white vote, or perhaps a little higher, for Romney. In the Post’s likely voter sample, he only leads by 55 to 42 percent among white voters. That thirteen-point lead is nowhere close to the twenty-point-plus margin he probably needs.
Despite that, Chait concludes on a very cautiously optimistic note:
The conventions have tested the candidate’s ceilings — they show the candidate in the best light and then see who might support them then. Obama’s ceiling appears higher than Romney’s. But it’s not very high. There’s not going to be an Obama landslide. His path looks a lot like George W. Bush’s in 2004 — that is, 51 percent of the vote. That happens to be right about where he stands in Nate Silver’s forecast.
Though, of course GWB was re-elected. Nate Cohn seems to take a somewhat more optimistic take for Obama:
In the survey [latest Gallup], there were hints of an important shift:
The number of non-white voters who said they would “definitely vote” increased to record highs. 81 percent of African Americans said they would “definitely vote” compared to 82 percent of whites—the smallest deficit to date. In total, 75 percent of non-white voters said they would definitely vote compared to a low of 68 in July. Only one quarter of the sample reflects the DNC, so it will be interesting to see if these numbers increase further over the coming weeks…
Even so, the initial returns suggest the Obama campaign may have made up some ground in its effort to narrow the gap between registered and likely voters, even if it still have more work to do.
Romney has been doing much better in polls of “likely” rather than “registered” voters. Those “likely voter” screens are as much art as science and if they are under-predicting minority votes they are over-stating Romney’s strength. I have to say that 81% of Black voters saying they will definitely vote is as good an internal polling statistic as Obama could hope for lately. There’s many simple ways of looking at the election, but it’s safe to say that if minority voters show up in 2012 at a proportion just reasonably close to what they did n 2008, Obama becomes very, very hard to beat.