This is my first attempt in a while to post all by my lonesome. The last post I made (“She Blinded Me With Science”) got a bit munged in the translation from my html through the blogging software, so I’m hoping by doing this directly I avoid formatting problems as I had before.
September 29, 2006
Congressman Mark Foley resigns his seat
How do we tell when an election turns from one thing into another? What are the predictors of a “wave” election? (Or, for that matter, a tsunami election?) A few weeks ago, Nate Silver and I exchanged a few Tweets about the generic ballot and he indicated he was much more pessimistic than I about the Democrat’s chances at that point. We’ve certainly seen the results of that in his NY Times blog . Other sites are even more sanguine about Republican chances; some are sanguine about Democratic chances, particularly in House elections. I feel pretty good about the predictions I’m seeing on Nate’s blog and other sites for Senate elections. I think the numbers can be predicted pretty well for the Senate, but I suspect we’re flying blind when it comes to the House elections.
My personal rough-and-ready test for seeing the boundaries of a wave election: if a party wins all (or almost all) “lean” seats in their column and almost all of the “tossups”, then I call that a wave election.
For example, Cook Political Report has 204 lean, likely or solid Democratic House races, 181 lean, likely or solid Republican, and 50 tossups. Using my rough measure, that means a Democratic “wave” would render a 254D/181R split (very unlikely, I’d say); a “normal” midterm would split the tossups evenly and you’d get 229D/206R (pretty close to my personal prediction); and a Republican “wave” would result in 204D/231R and Speaker John Boehner.
Eyeballing Nate Silver’s prediction, the mean is 211D/224R with a standard deviation of perhaps 20 seats, so a “wave” of 1 s.d. (about 15% probability) would be 191D/244R one way or 231D/204R the other way.
Sabato’s Crystal Ball has a prediction of +47R or 210D/225R. This is a Republican “wave” by my definition: if all his lean and tossups go D, then it’s 238D/197R while if all his lean and tossups go R, it’s 213D/222R. My “normal midterm” model (splitting Sabato’s tossups evenly) gives 224D/211R.
What “leading indicators” would the 538refugees commentariat suggest for this election? What factors point to a wave election, or to a normal midterm election?
I personally put very little stock in the generic ballot, preferring real polling data (where available) and Cook Political Report and/or Sabato Crystal Ball rankings. However, it’s my impression that real polling data has been hopelessly corrupted by so-called “likely voter” screens plus deliberate manipulation of the news cycle plus an increasing blind spot where the wireless denizens reside. (If I were making this into a bumper sticker, it would say: I’m Wireless and I Vote).