Jacob Levy and I agreed to predict the congressional elections. Jacob's prediction is here, Larry Sabato's Crystal Ball page is our expert to beat, and an extra-special treat if we can beat the Iowa Electronic Market.

I'm doing this with some trepidation, for three reasons:

1) I am not now, nor have I ever been, a member of the American politics fraternity. In other words, American electoral politics are not my area of specialty.

2) Off-year elections rarely reflect aggregate trends, and it's rare that all of the local idiosyncratic factors (Toricelli for Lautenberg, Mondale for Wellstone) perfectly cancel each other out.

3) Predictive models of election outcomes in political science stink. I mean, they suck eggs. Every political science model worth its salt predicted Gore clearing 55% in the popular vote in 2000. The one thing everyone could agree on after the 2000 election was that these models were patently, obviously, wrong.

So, with those caveats, here's what I'm thinking:

A) The news on the economy is decidedly mixed. Unemployment just ticked up a tenth of a point, but productivity growth is high and overall economic growth remains positive. Politically, none of this matters as much as the fact that consumer confidence took a nosedive in October. That may have been due to jitters over a war with Iraq, but it's still the dominant number. In the end, the president's party takes a hit when the economy is perceived to go south, so that cuts in favor of the Dems.

B) It's an off-year election, which traditionally favors the party out of the presidency. 1998 was an aberration, largely because the Republicans foolishly caved to Clinton on substance and made the election a referendum on impeachment

C) Last election, the polls seemed to have a rightward bias of 1-2 percentage points. This was likely due to elevated African-American turnout. Post-Florida, I suspect that turnout for this constituency will remain high.

D) Redistricting from the 2000 census favors Republicans. Of course, this only affects the House races.

So,

House: GOP +1, which will leave Dick Gephardt way, way out in the cold.
Senate: Dems +2, which will leave Jim Jeffords way, way out in the cold.

Because Jacob is also predicting individual Senate races, here are my picks:

NH: D (Shaheen)
MN: D (Mondale)
NJ: D (Lautenberg)
MO: R (Talent)
NC: R (Dole)
SD: D (Johnson)
CO: D (Strickland)
AK: D (Pryor)
LA: D (Landrieu)
GA: D (Cleland)

Similar to Jacob, except that I think Shaheen wins in New Hampshire and Johnson holds on in South Dakota.

One final prediction: because all media outlets have predicted Florida-style legal challenges in multiple races across the country, I come to the inescapable conclusion that no Congressional election will be close enough to prompt a legal challenge. This will leave David Boies way, way out in the cold.

UPDATE: You might notice that there was no mention of Iraq anywhere in that passage. None. The reason is Bush's decision to work through the U.N. Security Council, which deflates the issue like a balloon. Even the New York Times editorial page acknowledges that the Bush administration has made a serious effort to negotiate, to the point where France is looking like the obstructionist. This will reduce any anti-war turnout that would obviously trend Democratic. However, the haggling at the U.N. makes Iraq seem more like "normal politics" and less like the imminent outbreak of hostilities, which also blunts any rally-round-the-flag effect for Bush.