http://www.fivethirtyeight.com/2008/09/todays-polls-921.html"So ... what's the headline here? Probably the Public Policy Polling survey showing a 46-46 tie there. One wonders if the McCain campaign's internals are telling them something similar, since they are now shifting resources into the state. And Barack Obama was out in Charlotte today.
But yet, my model still does not consider North Carolina to be a plausible tipping point state. Why not? Partly because the polling there has hardly been uniformly favorable to Obama -- it was barely ten days ago when SuvreyUSA released a poll showing him down 20 points in the state. But mostly, it comes down to the parameters of the electoral math that we have discussed here before. It is unlikely that Obama wins North Carolina without winning Virginia -- this parlay occurred just 103 times out of the 10,000 simulations we ran today. And if he's won Virginia, Obama likely won't need North Carolina to win the election -- he'll already be well past 270 eletoral votes.
...Unless, maybe, something goes wrong in the Midwest. For instance, if Obama were to lose Pennsylvania or Michigan (in addition to Ohio), then the Virginia/North Carolina parlay would probably save the election for him. So, of course, would a Virginia/Florida parlay, which seems more probable at the moment, although the Obama campaign has a larger ground game advantage in North Carolina than they do in Florida.
The reason this discussion is interesting, I suppose, is if you believe there is some sort of Bradley Effect, perhaps coupled with a reverse Bradley Effect in the South. The polling badly underestimated Obama's performance in the South during the primaries, while being roughly accurate elsewhere in the country. Suppose that the polls are 3 points low on Obama in the South, but that there is some sort of Bradley Effect in the Midwest, and his polls are 2 points high there. In that instance, North Carolina (in addition to Virginia and Florida) become quite important."