In all the annals of media stupid this silly season, we may have a winner. At least in the "usually sane outlet" category (the other one, WND pretty much has a lock).
You want a horse-race narrative? Business insider asked some pundits to predict the electoral college total, and that included a guy who writes about actual horse races.
The Washington Post's Andrew Beyer predicts a Romney win with 284 electoral votes in this pundit-palooza
But that's not remotely the oddest or most unlikely prediction. The prize for the Whiskey Tango Foxtrot viewpoint goes to (who else?) Jim Cramer. Yes, the guy who called Lehman a great bargain buy just a week or so before they went belly up.
Cramer's call is 440 e.v. for - wait for it - Obama! The only reasonable conclusion is that after the election, he can say Obama massively underperformed according to the "conventional wisdom" (and on that score, I'll give Cramer half credit - by CNBC standards he's safely conventional; the other part, he's never been within a light year).
Its been quite a disappointing year for election coverage. In a cycle when poll aggregates and statistical models have become big news (Nate Silver, Sam Wang, et. al.), the punditocracy has not bothered to get in to a serious discussion of the numbers.
Take margins of error. Most people think if two candidates poll, say, 3 points apart, and the margin of error is 4, then they are essentially tied. I've even seen news reports call this situation a "statistical dead heat"
Uh, no. I should probably do a whole diary entry on this, but let's just say if you understand how margins of error work, there's a huge difference between, say, 25% off from the margin of error (i.e. a one point gap in a 4 point MOE) and 75% off (3 point gap in a 4 point MOE). The latter is a better than 80% chance of a win for the one who leads.
But why explain those things, when its far more entertaining to weave anecdotal "evidence" in to a charming narrative.