They crunched some numbers, put together an equation or two, sacrificed three white cockerels to Moloch and came up with... the Democrats picking up three seats in the House, and losing seven in the Senate. Which will lead people all over the spectrum to write posts and articles all using a variant of the concept Yes, probably. (pause) But...
The actual equations themselves are a lot less impressive than they look, of course. I haven't done statistics since grad school, but I can't help but notice that the post in question doesn't tell us when said equations successfully predicted the outcome of the 2006, 2008, and/or 2010 elections. If they were doing so at the end of March in those years, that's one thing; it they were doing so two days before the elections in question, well, that's another thing entirely. And if in March they were merely saying Well, the Democrats/Republicans are going to be in trouble this year then that is yet a third thing. In the first case we have an actually useful prediction; in the second we have a somewhat superfluous prediction; and in the third we simply have conventional wisdom all gussied up with some math.
Which is not to say that Sabato (actually, Alan Abramowitz) is wrong, here. I am actually expecting small losses in the House, myself: in 2010 we had a wave election that generally overturned the results of the two wave elections preceding it. The Republican-leaning seats that went Democratic in 2006 and 2008 have largely reverted back to the GOP, and Republican gains were mostly in fertile ideological territory anyway. A Democratic resurgence in 2012 is also hampered by what the kind would call 'bad luck' and the cruelly accurate would call 'a karmic backlash:' after the 2010 elections the redistricting process was mostly controlled by various state Republican parties. Between district reform and the need to respect Section V of the Voting Rights Act, the Democratic party were given relatively few opportunities to redraw the maps in ways that would fuel a fourth-in-a-row wave election. So I for one expect the usual churn; and net low single-digit loss sounds about right. Truth be told, freshmen Congressmen are traditionally considered vulnerable.
Likewise, gaining seven in the Senate sounds a bit high, but very possible. This is not only that the Democrats are defending twenty-three seats to the Republicans' ten; this is also because there are either twelve (11D, 1R) or eighteen (13D, 5R) first term Senators (depending on how you score Massachusetts, Mississippi, Nevada, New York, West Virginia, and Wyoming) up for their first re-election fight; and there are ten Senators retiring (7D, 3R). That represents a large amount of churn for a Senate cycle, and it's largely on the Democratic side: on the Republican side Maine is vulnerable, Massachusetts is theoretically vulnerable, Arizona is a long-shot for the Democrats... and that's pretty much it. If I had to guess I would say that net +5 GOP would be fairly reasonable, given that the Senate is designed to minimize the reach of a wave election anyway. +7 GOP is not unreasonable; then again, +3 GOP wouldn't surprise me either.
But all of that isn't really based on numbers; it's based on a feel for the races. That it fairly closely dovetails with an outside statistical analysis of favorability ratings and generic Congressional ballot races does not convince me that you can reduce the whole thing to a predictive model. Particularly one this far out...
Moe Lane (crosspost)
PS: Here are the things I look for:
- Who is retiring? What positions do they have in Congress?
- Who is suddenly retiring?
- Who is getting recruited?
- How crowded is the opposing party's field?
- Who is self-funding? (Self-funding candidates worry me slightly, when they're on my side.)
- And who just made a major howler in front of a video camera? - But this last one is mostly because I like to have fun while I'm doing this.