But, please, I'm not quibbling with the data, just trying to analyze the historical precendents. Since you've brought up large vs. small changes, I gathered some more numbers. I classified midterm elections into three kinds based on the House seat change, expressed as a percent, of the then President's party and correlated that to the Presidency changing parties. The categories of change are VU (Very Unfavorable - Pres's party loses 20% or more), SU (Somewhat Unfavorable - Pres's party loses less than 20% but more than 5%), and NU (Not Unfavorable - Pres's party loses nore more than 5% or gains). There are 8, 11 and 8 elections of those types resp. From the data (which I'll gladly share), those seemed reasonable partitions. Here are the numbers. (PC is Presidency Changed Parties and !PC is the complement).
PC !PC
VU 2 6
SU 5 6
NU 3 5
So again, it doesn't look to me that history bears out the claim that strong midterm swings against the President's party result in the President's party losing the Presidency.
I forgot to say, although it's clear from the table, that weak swings against the President's party also don't seem to correlate with the Presidency's changing hands either.
wow, you are really digging. I think you've got me to at least a point of saying that whatever correlation there is, is weak. SU is worse than NU or VU... and you have to throwin incumbent vs non-incumbent elections to see that factor. Still, if you look at historical context, midterm trends can portend the next election ... so ...
can you at least admit that the data doesnt suggest any kind of 'bounce-back' effect like some 'cut-n-run' conservatives suggest? That is, if we let the Dems win in 2006, we'll do better in 2008. Can you see at least that such a claim is nonsense?