Well nobody had us gaining less than 55 seats. That was a given. He said we had only a 25% chance of getting 60 or more seats and we got 67. He was flat out off and wrong in 2010.
“Well nobody had us gaining less than 55 seats. That was a given. He said we had only a 25% chance of getting 60 or more seats and we got 67. He was flat out off and wrong in 2010.”
Have you ever had college level statistics? I’m not being facetious or condescending here, I just don’t think you are getting what I am saying. Poll aggregators work in probabilities and their final projects are the peak of their bell curves.
You keep saying Nate Silver gave us a 25% chance of getting 60 or more seats and see that as a bad thing. You don’t realize how great that guess is until you understand his entire bell curve. There was a 25% chance, according to silver, that we would get less than 30 seats and a 25% chance that we would get more than 60 seats and a 50% chance that it would fall somewhere between.
Now based on that final projection curve, his best guess was 59 seats but since he works in the realm of probabilities, he was forced to hedge it. He specifically said:
“If we allocate all 435 seats to the leader projected by our model — no matter how slim the margin — Republicans would net a gain of 59 seats. In 15 of these 59 seats, however, the Republican is projected to win by fewer than 2 points. It is likely that Republicans will lose at least some of these — which is why the model forecasts an average gain of 54-55 seats, rather than 59”
He goes on to explain the right side of the curve like this:
“Moreover, given the exceptionally large number of seats in play, the Republicans’ gains could be significantly higher; they have better than a one-in-three chance of winning at least 60 seats, a one-in-six chance of winning at least 70 seats, and have some realistic chance of a gain exceeding 80 seats, according to the model.”
So if you see those odds at those ends, you can conversely call similar probabilities at the other end, i.e., 1 in 3 chance of only 30 seats, 1 in 6 changes of only 20 seats, etc.
In midterm elections, you have two components, the generic ballot and the house seats. The generic ballot is the only component that is comparable to a president election since it is binary, the republican or the democrat. In the generic, Nate Silver called it 6.8% as the post above showed. The final 2010 generic vote was EXACTLY 6.8%.
The seats are a completely different animal since you are dealing with 435 elections all with very different polling strengths. His probability curve was not dead on but really good especially since he ignored his liberal friends and said it would be waaay worse than they thought...kind of the converse of what happened here earlier this week.
My point is that if you can’t see that he is probably the best sports and politics statistician alive, then your ideology is blinding you or you aren’t understanding his work. Trust me, I sure as HELL would like to have that title belong to a neutral or conservative mathematician but it is what it is. Talent is talent and results are results, he doesn’t have many peers in his league.