I've been fighting a bit of a cold on and off for the past couple of
weeks, and so I took most of the day to rest and recover and to check a
couple of assumptions in our model. But before we get ahead of
ourselves, let's look at what we have to work with on a relatively
light polling day:
Colorado actually hadn't been polled all that much over the past two weeks or so, so the new PPP poll, giving Obama a 10-point lead there, is a pretty big deal. Note particularly the distinctly large sample size in this poll -- more than 1300 people.
As has been true for most of the post-convention period, Barack Obama appears to have a structural edge in the Electoral College. Colorado is a big part of that. Presently, we are showing a lead for Obama of about 6.5 points in the national popular vote (our model assumes the race will tighten a bit, so we're expecting that number to fall to 5.4 points by Election Day). But if you look at our current estimates in the individual states, you'll see that we have Obama ahead by at least 8 points in all of the Kerry states, plus the Gore add-ons of Iowa and New Mexico, plus Colorado, plus Virginia. Collectively, those states are worth 286 electoral votes -- well more than Obama needs to win. So while McCain has something like a 6-7 point deficit to make up in the popular vote, the gap is more like 8 points in any set of states that would give him a winning electoral combination.
McCain has a somewhat gentler hill to climb in Nevada, where Mason-Dixon now has him trailing Obama by 2, but this too is a pretty good result for Obama, as Mason-Dixon had shown McCain with a 7-point lead in August.
If you try really hard -- and Matt Drudge is doing his best -- I suppose you can perceive some good news for McCain in the tracking polls, as McCain gained ground in Gallup, Rasmussen and Hotline/Diageo. He lost ground, however, in the Zogby and Research 2000 polls, and there may be a bit of reversion to the mean at work, as Gallup and Hotline had been toward the high end of Obama's range before.
Finally, as for the work I did on the model, what I looked at today is the relationship between state and national polling thus far in this election. It turns out that the state polls have hewed a bit closer to the national polls than I had been assuming before; the battleground states remain in roughly the same relationship to one another as the national polls bob up and down. More detail on this later, but the upshot is that the margins of error in individual states are smaller than I had been assuming; that's why a lot of the blue states look a little bluer, and the red states a little redder, than they did yesterday. This change is slightly favorable to Barack Obama on balance, as it means that the electoral college advantage that I described a moment ago is a bit more potent than I'd thought.