UncleSam
Sr. Member
Posts: 2,513
|
|
« on: September 15, 2017, 12:10:09 PM » |
|
I think there's an issue with your second chart Mondale - the swings are not consistent in terms of which number is being subtracted from the other. From looking at it it looks to me like Republicans gained on average 2.5 points though from special election performance to midterm performance, if I'm reading that correctly? So saying that specials 'predict' house margin is a bit overstated - the margin of error on that is fairly large (standard deviation of 2.9 points, if I'm calculating it correctly), which would make the 95% confidence interval on the swing from special elections to national midterm house vote range from (-3.1,+8.1), where negative indicates a swing towards Democrats.
Of course the normal approximation to such a data set is not likely to be particularly accurate, and in fact the larger special margins tended to normalize (+9 for Republicans shifted left 2 points, +15 for Democrats shifted right 7 points), rather than shift with the average. I guess Dems basically hope that 2018 will mimic 2006, but I'm having a hard time seeing anything beyond the 30-seat pickup of 2006 based on the data (and there are good arguments why even a 2006 comparison is overstating things a bit).
|