538 Update Thread

[Jake]

At this point, I'd put a lot less stock in Nate's numbers than I would've several months ago. His model is likely to be "right" if the turnout numbers are where the aggregate of all the polls place them. His model's problem is it doesn't know anything about the early voting data that's coming in. Meaning that if the early voting partisan turnout numbers (and the resultant overall partisan turnout numbers) are different than what the aggregate of the polls predicted, his model will be "wrong". I'm not at all sure whether that will be the case, but that's an extremely important factor that he's counting on polling companies to be correct about. Now polling companies could be adjusting their turnout projections based off early voting numbers, but they also might not be, especially the less savvy ones.

All in all, I'm much more worried that the pollsters are overestimating Obama's turnout than Romney's, if only because their main data point will be the crazy turnout in 2008.

[Jake]

<>

Crazy turnout meaning very heavily a Democratic electorate. Turnout was 55/45 D/R among those who display a partisan preference in 2008,  50/50 D/R in 2004, 53/47 D/R in 2000, and 53/47 D/R in 1996. Any partisan turnout model based primarily on 2008 (and note I'm not claiming this is actually the case) will over-estimate D turnout. It's pretty likely that we get turnout much closer to '96 and '00.