In 2010, Silver gave the GOP only a 25% chance of flipping 60 House seats. It ended up being 67. Some vaunted model...
“In 2010, Silver gave the GOP only a 25% chance of flipping 60 House seats. It ended up being 67. Some vaunted model...”
Yes, but he gave us a 75% chance of getting 55 seats. He mapped out the probabilities district by district, no one else did that at that level of precision. They just gave generics. Newsweek projected a 23 seat gain for us and Gallup projected a 77 seat gain. Everyone else gave some guess in between those extreme parameters...he hit it better than any other group.
Look at paragraph 17 where he sums up his model probability. He projected a 50 to 60 seat gain and got more specific districts right than anyone.
It is all a bell curve of probabilities. If you cherry pick the 25% of the right side of the curve and say he is off, then you don’t understand statistics because there was also a 25% chance of 30 or less seats at the left end of the curve. If you chose the peak of his bell curve where the best guess lies, you would see that 67 was right at the right edge of that projection and 55 was at the peak.