Oct 09 2010
There has been a serious bout of denial in the media of late, with stories about a resurgent Democrat Party somehow holding the House and saving Nancy Pelosi’s failed Spearkership. Ain’t going to happen. As Nate Silver has noted the GOP has been marching towards a higher and higher probability of a GOP take over – now sitting at 72% in his model.
Our model now estimates that the Republicans have a 72 percent chance of taking over the House, up from 67 percent last week. Moreover, they have nearly even odds of a achieving a net gain of 50 seats; their average gain in a typical simulation run was between 47 and 48 seats. However, the playing field remains very broad and considerably larger [gains] are possible, as are considerably smaller ones.
Nate’s models are very slow to adjust because his reference base are races from 1999 onward. He has no real wave elections in his historical data to rapidly detect massive changes in voter attitude. But as slow as his models are to respond, they are responding.
What caught my eye was a table at the Cook Political Report, which Nate referenced showing 85 Democrat seats in jeopardy. Cook includes the PVI index [I believe this stands for Partisan Voting Index] for each race and I wanted to do a very simple analysis of what differing size waves would do to the House. What I am assessing is the damage to the Dems if the November wave were to consume all or most races below a certain PVI.
So I am going to assess 3 wave heights related to PVI: R+1, D+2 and D+6. I do not know of any direct correlation between the generic ballot and the PVI, but one could argue a generic ballot that is tied between the parties would move R+1 districts to the GOP, a generic ballot gap that has the GOP +3-5% would move the D+2 districts into GOP hands, and a generic ballot greater than +8% for the GOP would take out D+6 districts. Right now the generic ballot is sitting in the GOP +5-6% range, but some pollsters have it well above 10+% in their likely voter models. So this should be interesting.
R+1 seats go GOP: Using the Cook tables we find there are 44 Democrat seats with a PVI of R+1 or greater (R+0 not counted) in the Democrat columns, and 18 Democrat seats in the GOP columns. The total potential damage here is 62 seats. Now there is no scenario where the GOP picks up all 62 seats, so we will look at four levels of success (95%, 90%, 85% & 75%). These are summarized in the first row of the table below.
D+2 seats go GOP: Using the Cook tables we find an additional 13 Democrat seats in the range of D+2 through R+0 in the Democrat columns, and 3 Democrat seats in the GOP columns, for total potential loss of an additional 16 seats. This is very important because if the wave is a D+2 wave, then most of the R+1 seats will be lost, which makes for some interesting permutations of outcome. More on that when we look at the table.
D+6 seats go GOP: The last grouping, for the really high wave scenario, shows 21 additional Democrat seats in the D+6 through D+3 range in the Democrat columns, and 1 Democrat seat in the GOP columns. This gives us an additional 22 seat potential loss if the November wave is at D+6.
Results: The following table shows a range of possible scenarios for the November election, only one of which shows the Dems holding a slim majority (click to enlarge).
If we look at the R+1 wave scenarios, we see the range of Democrat losses (not accounting for any GOP losses) runs from 34-59 seats. Only in a very weak R+1 wave would the Dems hold the House, which is why I doubt the House is even in play given the generic ballot levels we see now at RCP. If the wave is centered at the R+1 level I would say the Dems would lose in the range of 53.
Now what happens if the wave is centered at the D+2 level? We see we have an additional 16 seats in play, but the way the table is laid out is a bit deceptive. If you are winning 80% in the D+2 range, you sure as heck are not just winning 80% of the R+1 range. It is more likely you are winning 95% of R+1 even with 80% of the D+2. Conversely, if the wave is a weak D+2 (75% range) then the R+1 could be 85%. The range of scenarios is as the subtotal notes: Democrat losses in the 43-74 seat range. Again, since these are just rough estimates and there is no attempt to precisely predict (vs bound the probable outcomes) I think we could safely assume a 67 seat pick up if the wave is solidly hitting the D+2 range. Note, this level of 78 seats being in play is LOWER than the number of seats the Cook Political Report says are in play.
Now we come to the historic tsunami level – where 100 Democrat seats could be in play. If the recent Gallup likely voter models are even 70% close (0.7 x 15% = 10.5%) then we would expect every seat with a D+6 and lower (towards the GOP) would be in trouble. If the tsunami is centered at this level, the range of Democrat losses (sub total D+6 row) is 61-95 seats, with a solid expectation of around 87 seats gone. This would be truly historic, and it is not unreasonable given what the polls are showing. As with the previous level, numerous scenarios exist in between the D+6 and D+2 lines.
With GOP enthusiasm skyrocketing and Dems wavering (or bolting) the R+1 and D+2 scenarios seem reasonable. Add in Independents moving strongly to the GOP and I think you hit the D+6 scenario. Check out this table from Jay Cost yesterday:
If Indies are both trending +20% to the GOP and also ready to go out and vote to send a message to DC, then the D+6 level could be too conservative. I don’t know where the wave will hit come November 2nd, but the range of 34-96 seats is not as crazy as some Dems wish it was. I see no reason not to settle in on the middle and say the Dems could easily lose 70 seats this election in the House.