Monday, October 20, 2008

Late Polls

A bunch of late state polls came out today along with three late national polls. Democracy Corps has Obama up by 5 nationally in line with CNN's national poll earlier today. The Economist has Obama up by 6. ABC/WP has Obama up by 9 nationally, basically unchanged from the 10-point lead the same poll gave him 9 days ago. But, believe it or not, that's not the REALLY bad news for McCain in this poll. Obama remains at 53% in this poll. The significance of this is that McCain cannot rely on winning undecideds to win the election. He needs to convert some of the "soft" Obama supporters ... and there are not many of those. We know this by looking at the enthusiasm gap. 64% of Obama voters said they are "very enthusiastic" about voting for him while just 40% of McCain voters said the same of their candidate. To put this in historical perspective, in 2004, 55% of Bush voters were "very enthusiastic" about their choice while 46% of Kerry voters said the same of their candidate. In 2000, the numbers were 44/41 for Bush/Gore respectively.

Here are the state polls from this afternoon:

Ohio (Rasmussen) - McCain +2
Florida (Rasmussen) - McCain +1
North Carolina (Rasmussen) - Obama +3
Colorado (Rasmussen) - Obama +5
Missouri (Rasmussen) - Obama +5
Virginia (SurveyUSA) - Obama +6
Wisconsin (SurveyUSA) - Obama +8
Pennsylvania (Susquehanna) - Obama +8
Oregon (Grove) - Obama +13
Georgia (Democracy Corps) - McCain +2
Pennsylvania (Muhlenberg) - Obama +12

What to make of all this? If it weren't for the Missouri number, I would say the Rasmussen polls strike me as leaning a little towards the GOP as a group. These polls are certainly on the low side of all polls for Obama support but let's assume they're all real. If they are, game over. Obama really only needs to win any one of the 5 states Rasmussen polled here. So pick your poison. The Virginia number from SurveyUSA is good ... but a peek at the internals make it look even better. SurveyUSA estimates African-Americans will make up 19% of voters in Virginia. But African-Americans made up 21% of voters in the 2004 election. Why would they be a smaller portion of the electorate in 2008? Also, SurveyUSA has Obama winning 85% of African-American voters but John Kerry won 87% of African-American voters in 2004. I suspect Obama will outperform Kerry on both these numbers. So, all this means the Rasmussen number (Obama +10) in Virginia is looking pretty accurate to me right now. The Georgia number from Democracy is outright deadly for McCain as well. Here's how the shorthand works: Take Virginia, make it a little less Democratic and you've got North Carolina. Take North Carolina and make it a little less Democratic and you've got Georgia. So Obama down 2 in GA is consistent with Obama up 3-5 in NC and that's consistent with Obama up 6-10 in Virginia. My point is all these numbers from different pollsters are scary, scary consistent from this perspective and that's really, really bad news for McCain.

Nate Silver sees all this just a tiny bit less positively than I do (he sees a bit more of a drift back to McCain in the last week or so). Meanwhile, John King reports that "senior McCain officials" have all but written off Iowa, New Mexico, AND Colorado. So how does McCain get to 270 without those three states that all voted for Bush in 2004? John King from CNN says McCain is planning for a comeback in Pennsylvania. Make sense? No. I don't get it either.

No comments: