Drudge is displaying a phenomenal statistic.

Preliminary: Senate Ballots cast: 31,591,495 (D) 25,054,569 (R)...

If that is true, it means that the Democrats won the two-party vote by a whopping 12%: 56% to 44%.

If the House went anything like this, it means that the Republicans should count themselves very, very lucky. It could have been worse. A lot worse.

Over the weekend, I offered a projection of 19 seats based upon (a) Gallup's generic ballot prediction of 54-46, and (b) an OLS regression equation of votes to seats for 1996 to 2006.

If these 56/44 numbers are roughly the same for the House as they are for the Senate -- and my intuition is that they might actually be higher because uncontested Democratic seats outnumbered uncontested Republican seats by about 5:1 -- then the model would have predicted Republican losses of 25 or so seats, a figure statistically consistent with the final result of 29-ish (assuming that the Democrats hold CT 02 and GA 12). In fact, inserting the 56/44 popular vote and 29 seat swing into the post-1994 model greatly improves its predictive accuracy. Whereas it explained only 40% of the variation prior, it now explains 91% of the variation.

In the pre-1996 era, a 56-44 result would have produced a 73-seat gain for the Democrats in the House of Representatives. That would have been a 276 D to 159 R House.

56/44 would also mark a 7.4% decrease in the Republican share of the two-party vote. This would be the greatest drop in share of the two-party vote since, I believe, 1938. All in all, these vote numbers -- 73-seat loss and 7.4% vote loss -- most closely represent the 1946 election in which Harry Truman and the Democrats lost 54 seats and 6.4% of the vote. They went from a large majority to a tiny minority in the course of two years.

Of course, by seat comparisons, the 29-ish seat loss mimics the much-less-impressive-but-still-mighty 1982 midterm, when the GOP lost 27 seats and 4.9% of the vote.

What mitigated Republican losses? Why is it that the GOP lost seats akin to 1982 and votes akin to 1946?

I believe that the reason boils down to the structural issues I have been discussing all year. My election eve mistake was perhaps not so much an overestimation of the GOP's structural advantages, as I speculated yesterday, but rather an underestimation of the anti-Republican mood. Read: More Fox/Newsweek/Time/RT-Cook, Less Pew/ABC News/Gallup/Democracy Corps.

When I say "structural basis," I mean two things.

Incumbency advantage is a major part of it. Many would-have-been top-tier Democratic challengers are probably kicking themselves this morning for not having challenged their local Republican, especially with upsets like MN 01, NH 01, KS 02 and PA 04.

Also, the tight alignment of the electorate, which, I think, boils down to the fact that a large proportion of the Republican delegation is in the South. By my count, the Republicans lost 11 seats in the Northeast, 10 seats in the Midwest, 4 in the West and 4 in the South. The West's numbers are not terribly impressive for the GOP, considering that many of the Republican seats in the west are in gerrymandered-into-stasis California, which only saw Abramoff-tainted Richard Pombo go down. However, their success here might also have meant that immigration helped them.

The South's numbers are truly stunning to me. It was in the South that, despite a wide and deep anti-Republican national mood, the party still managed to hold all but 4 seats and almost won 2 Democratic seats. In 2 of these Republican-held seats, the Republican candidate was not even on the ballot, and in 1 of these seats -- he nearly won! The Republican's capacity to hold the South despite the pro-Democratic national mood is a stunning feature of our contemporary politics. Just as urban centers are solid Democratic bastions in the House -- so also is the South.