Monday, November 03, 2008

Nate, don't fail me now

On Election Eve, Silver at has it up to 98.1 percent.

This will make it easier to sleep tonight.

I'm planning to vote by 6:30 a.m., but just the once.

EARLY AFTERNOON ELECTION DAY UPDATE: If Silver's estimates are treated as reliable (and they assume lack of overall systematic bias in the polling data), Obama gets 264 electoral votes from states that the model treats him as having a 100 percent chance of winning. (Pennsylvania, with 21 electoral votes, is among those states.) He would go over 270 if he got Colorado's 9 EVs (98%), Virginia's 13 (97%), Ohio's 20 (88%), or Florida's 27 (73%). Nevada's 5 (95%) would get him to 269, presumably good enough for the win given (a) Democratic control in Congress plus (b) the persuasive significance of his popular vote edge, assuming it holds.

Obviously, these various probabilities, even if we take them at face value, are unlikely to be entirely uncorrelated.

While some might take comfort from these numbers, they also provide a panic guide if Virginia doesn't fall briskly into place in the hour after 7 pm (as this might undermine confidence in the entire projection).

No comments: