Dead Spin has a fascinating interview with polling guru Nate Silver. The whole thing is worth your time to read, but of particular interest is Silver’s response to a question about his call in the North Dakota Senate race where he predicted a Rick Berg win with a 92% probability.
Obviously, he got that wrong, and a questioner asked Silver what went wrong. Silver’s response? Heitkamp’s win fell in that 8% probability range:
Berg’s loss shocked a lot of political observers, both inside and outside of North Dakota. The paid political operatives and the die-hard partisans on the left are gloating now, but I spoke with several Democrat insiders before election day and most of them were resigned to a Berg win.
Heitkamp winning is proof that even the best predictive models aren’t right 100% of the time.
The stars seemed to be aligning for Berg down the stretch. The movement in the independent public polling was toward him (though it’s clear now that Forum Communications ought to be indicted for polling malpractice), and most expected a much bigger oil country turnout.
Turns out the polling wasn’t going his way, and the oil patch turnout didn’t manifest, and a significant number of voters in the state’s electorate seem to care a whole lot more about the personality of the candidates than they do the policies.
Thus, Heitkamp is North Dakota’s Senator for the next six years.