Polling and turnout

A friend writes:

According to RCP, the electoral college has been locked up at Obama 273, McCain 265 for about a month. Although it seems like there have been fits and starts in the media regarding the two campaigns, it suggests to me that people's minds have essentially been made up. Do you see it that way, and if so, does that mean we're now in a full-on turnout battle?

There still seems like there's some movement in the national polls- as we've seen Obama shift from maybe down 1 or 2 to McCain during the latter's convention/Palin bump, to Obama up 3 or 4 points as voters' attention has turned toward the economy, Palin's had unsuccessful interviews with Charlie Gibson and Katie Couric, etc. I think that indicates that we're probably looking at, nationally, maybe 5 to 8 points of "swing voters", barring some major catastrophe for either candidate that causes voters to wholly reconsider them.

This plays out in each of the states. At this point in the race, where many voters' perceptions are hardened, the demographic and partisan makeup of each state give them baseline numbers in comparison to a candidate's national polls. For instance, Pennsylvania seems to be about a +2 Obama state, such that when his national polls show him up 3 against McCain, PA is +5 or so for Obama. North Carolina, on the other hand, is maybe a +4 or 5 for McCain, so that when he's down 3, NC is still a +2. Most of the battleground states have base numbers that are within the 5-8 point "swing" I described above.

There are a couple other factors to keep in mind. The first is that a superior turn-out or get-out-the-vote ("GOTV") effort can be worth maybe 2-3 points for a candidate. From all evidence Obama's voter registration, early voting and GOTV efforts are going to be superior to McCain's in most blue and contested states. The second factor is that, for the first time, pollsters are finding a demonstrable partisan difference between voters who do not have a land-line telephone (and consequently are not polled by most organizations) and those who do, favoring Obama by a couple points. The third factor is the so-called Bradley effect, where voters over-report their intention to vote for a black candidate by some amount. I've read some credible research indicating that the Bradley effect is much diminished since 1996, and was not present in the Democratic primary voting this year.

So, keeping all of the above in mind, there's still a lot of possibility for national events, the debates, and the candidates' messages to drive the national polls in one direction or another, which will affect the state polls in accordance with each state's "base number." The base numbers themselves aren't entirely fixed- if either candidate was caught on tape saying the reason we needed to bail out the auto industry was because of the lazy American auto-workers, his base number would drop precipitously in Michigan. That said, with the polling numbers so close in many of the battleground states, it's quite possible for a superior gotv effort for either candidate to generate a 2-3 point bump which would put him over the top in closely contested states.