If Nate Silver is Wrong

In the last few days Joe Scarborough and Nate Silver have been commenting back and forth to each other about the likelihood of Silver's forecast for the election coming true. It has been fairly quiet and not venomous. Scarborough being a politician does not doubt Silver's past success but openly wonders if it is possible to put a political race inside a formula and be sure of the forecasts.

Silver, correctly in my opinion, has held true to his models and the fact that they have him as an outlier. An outlier not necessarily in the fact that Obama will win but in the likelihood of an Obama victory. The most recent number given by Silver is that as of this moment there is a seventy eight percent chance of a victory in the electoral college by the President.

Interestingly at the same time Silver gives Governor Romney a three percent chance of winning without an Ohio victory. While Republicans have traditionally needed Ohio to win the election this might not be a great shock, however Silver has the President's victory percentage at eight percent if he does not win Ohio. This would seem to imply that statistically speaking Ohio will determine the next President. While Obama seems to be holding a steady lead in the polls I think what gets Silver into trouble, not with just the nutty right, but even to some extent with the rational right, such as Scarborough, is his apparent certainty of victory for Obama.

I believe Silver would say I am not certain, the numbers are what the numbers are, and of course in terms of a formula he is correct. For Silver however saying that the race is statistically very likely to come down to Ohio and then with the race appearing close giving Obama an almost eighty percent of chance of victory in the electoral college is going to earn him some heat. For me what seems hardest to be sure of, especially when it comes to Ohio are the factors that could well affect this race that perhaps do not show up in statistical markers. What are the chances of voting machine trouble in Cleveland which Obama needs to win and win big. What are the chances of voter suppression efforts having an effect in certain areas, what if it rains, what if the race is close enough that the Republican Secretary of State can affect the outcome in some way. In short even for someone like me who wants to believe the numbers his level of certainly reflect can be unsettling.

I saw a comment yesterday by some commentator defending Silver stating that if he is wrong his credibility will not be hurt, after all the best hitters in baseball strike out three times in a game sometime.

I could not disagree more. It would be one thing if Silver was like most pollsters and had Obama winning but claiming it was very close and that the polls showed the outcome within the margin of error. This is not what Silver has done. He has said that Obama has a better than 75, that is seventy five percent chance of winning. He has said this even though it is likely that the popular vote will likely be split by less than 2 percent of the vote.

No, if Silver is wrong, and I hope he is not, he will be harmed a great deal. He cannot get great praise by calling the race in the way he has, and then if wrong get a mulligan. Because of the level of absoluteness he called this race, a wrong outcome for Silver will diminish not just his formulas and tables but him. You cannot have it both ways.