The Start of Prediction Season Is In Full Swing...

Well, I guess it's been in full swing for a while.

A note on my absence - life's been a bit insane. Lots of speechifying, a couple of trips up to the Tri-State Area, all sorts of stuff.

In any case, we've talked about the '06 races already here on the Liberty Blog, and I'll offer more of the perspective from Professor Horace Cooper later this week (I offer the Carnac photo in his honor). Today, however, I want to offer the learned opinion of my old friend, Dr. Patrick Brandt, who teaches political science at UT Dallas. Pat and I went to William and Mary together, and he remains one of the best "go-to" guys that I know.

Anyhow, I've been consistently saying that when it comes to the mid-terms, my predictions are that the GOP is going to hold onto both the House and the Senate, with a net loss of 9 seats in the House and 2 in the Senate. I haven't gotten into specifics, but as the weeks go by I will. Admittedly, there have been times that I've been less certain of those numbers, but a few factors are making me more certain:

- The Presidents numbers are up. Yes, this could change, but for right now, they are where they are;
- Gas prices are down. I'm someone who follows gas prices religiously - not for any political purpose, but because that's the kind of thing I think about. And I know that other people do, too. Gas prices are going to continue to drop;
- The numbers for the GOP base are up, substantially. This is the important number. Though the core GOP numbers were down as low as 70%, they're now back up in the mid-80s.

Now, fast-forward to what Dr. Brandt's been up to. The American Political Science Association does a contest for political science professors wherein they predict the House outcome. Pat and his partner on this, Dr. Thomas Brunell, have their's up on a website. They're predicting a net GOP loss of 12, but hedge their bets with the President's approval numbers - Bush's number go up, and they theorize that the GOP net-loss will go down.

What's more, their data are current thru June 2006, so we've got several more months to noodle through.

The paper can be found here:

http://www.utdallas.edu/~pxb054000/USHouseElectionForecast2006/HouseForecasts2006.html

And their summary is as follows:

We estimate a set of forecasts for the House seats won by the president's party. Using monthly data from January 1948 to June 2006, a Bayesian state-space model is employed to generate the out-of-sample forecasts. Our point forecast shows that the Republicans should win about 220 seats in the November 2006 election -- a net loss of 12 seats from their November 2004 total. We also quantify the probability of this outcome and the associated uncertainty of this estimate. Further, we demonstrate that inter-election increases in presidential popularity have a sizable effect on the number of House seats won by the Administration party. For November 2006, we predict that a 10 point swing in the president's popularity between now and Election Day could result in 4 fewer seats lost by the GOP.
---end quoted material---

So, I feel confident that I'm right in Drs Brandt and Brunell's ballpark.

- Andrew Langer