I've been running some calculations on the early voting statistics that the great Michael McDonald (not of the Doobie Brothers, but of the US Elections Project) has been meticulously updating for the key swing states of Colorado, Florida, Iowa, Nevada, and North Carolina. (Ohio and Virginia need a different approach, which I'm still working on.)
I've applied the current average percentage of support by party within each state (i.e., based on the most recent five polls in each of the states) to the numbers of early voters. What that allows me to do is to calculate the current "lead/deficit" for the President and perhaps more importantly, what percentage of the vote on Election Day the President would have to get in order to get 50.1% of the vote in each of these states. (A clear limitation on this is that my approach assumes the same number of people will vote in this election as did during 2008.) Obviously, as more people vote early, the numbers needed on election day change.
Here are my current estimates:
CO: 53-47%; 48.3% needed on Election Day
FL: 52-48%; 49.5% needed on Election Day
IA: 57-43%; 47.2% needed on Election Day
NV: 54.5-45.5%; 46.2% needed on Election Day
NC: 55.3-44.7%; 47.1% needed on Election Day (!!)
On the face of it, these are encouraging numbers for the President. I'd dismissed much hope of bringing in FL and had next to no hope for NC, but these calculations make both of them look winnable - indeed perhaps NC even more likely. Of course, the flip side of this is that these voters were already in the President's camp already and they simply "got to the dance early" and that therefore, there'll be much less support to draw from on Election Day.
Time will tell. More updates as the week goes on.