Science: Republicans Keep Senate, Lose House
Rusty Shackleford looks at the latest sophisticated modeling forecasts produced by political scientists to predict the results of next month’s elections. Amazingly, they’re very close to what amateur poll watchers are guessing.
I’ll wait for election day to cheer or be depressed. Or maybe until after the final court ruling if 2006 produces a spate of election lawsuits.
If the party in power loses the House but keeps the Senate, it will be a first in at least a very long time, if not ever.
Therefore I am inclined to think rumors of a Democrat House after this election, are greatly exaggerated.
McGehee: Yeah, I blogged about the fact that no party has ever taken control of the House without also taking the Senate back in April. The analysis there, building off Jay Cost’s, still stands. Still, the numbers aren’t looking good for the GOP at the moment.
“It hasn’t happened yet” isn’t a very convincing argument — think how boring sports would be if it were. Didn’t 2002 also buck convention by having the President’s party gain seats in the house?
The argument put forth in James’ April blog is weak. Start with A pattern that holds over 46 observations without exception is probably not random.
False, Mathematically speaking. One can not make any assessment about how likely the observed pattern is due to chance until one estimates the “variance” (ie. how much the numbers jumps around). OTB attracts a pretty bright readership so I’ll not belabor this point (Unless you want to. How did you parametrize the joint distribution)?
Further down James makes a really good point:Senate races are increasingly national, with the parties getting involved in recruiting and financing candidates. House races, by contrast, are still largely based upon local idiosyncrasies and the incumbent’s reputation for constituent service.
However, this may not hold in 2006. The only other by-years I can remember being this nationalized werere 1974 and 1994. I think we can all agree that this is not 1974 — take that one off the table. But echoes of 1994 are clear. 1994 saw a new grassroots force, led by Newt Gingrich, effectively nationalize dozens of house races via the “Contract with American.” The Internet has definitely nationalized numerous House elections: DKos alone is a million-dollar fire house spraying money and media attention directly into obscure corners like NE-3 and ID-1. Apparently many Rebuplican candidates agrees, becuase many seem to be runing against “Speaker Pelosi” as much than their local Democratic opponent. While Pelosi’s “First hundred hours” is but a pale echo of the Contract, the widespread discontent with Bush’s War may form a suitable replacement.
To be clear, I am not saying the pattern of “house can’t change if the Senate doesn’t” won’t hold this year. I am saying the arguments given thus far are weak.
None of us are arguing the determinism of the past. The argument, though, is that Senate elections, for institutional reasons, are more likely to be bellweathers of national trends than House ones, which are more localized. Further, while close observers of politics who are not professional pollsters often have a good grip on what’s going on in Senate races–because of their relative celebrity status–they seldom have a clue about House races.
I do agree that the 1994 parallels continue to get stronger as more scandals erupt. The difference, though, is that Gingrich and Co. not only ran against the corruption of the Democrats but on a positive message. There’s not much of the latter this year.
My take looking at the polling data is that this year will the first time the House goes and the Senate doesn’t. That’s what the race-by-race polling shows us. Still, there’s plenty of room for error that could mean the GOP narrowly holds onto the House or, conversely, lose an additional seat or two in the Senate beyond current projections and lose both houses.