Dark Star

Mark Steyn:

I'm kind of beyond polls at this stage — although I found this analysis by Jay Cost very interesting. As he points out, the standard deviation this year is significantly higher than usual - which means that the polls that turn out to be wrong will be wronger than usual. Also, they're not a conventional central cluster — a pyramid with a few outliers at either end. The McCain graph has almost no cluster anywhere, suggesting that the number you get depends as much on the "turnout model" cooked up in the pollsters' office as on anything Mr and Mrs America tell 'em down the phone line.

[More]