House
Democrats 230 (+27) - Republicans 205

Senate
Democrats 50 (+5) - Republicans 50

As has been the conventional wisdom for at least 10 days now, I predict that the Democrats will take the House -- but with only a wave and not a tsunami -- while the Republicans will maintain control of the Senate, if only by Dick Cheney's fingernails.

I'll have more to say below about this prediction, but first our regular look at the numbers.

    2006 ELECTION PROJECTIONS SURVEY


[click on the chart for a larger view]

Analysis: Well, everything continues pretty much as before. The Senate long-term trendlines finally cross after kissing/not kissing for a couple of days, but it's too close to the election to be a harbinger of a Democratic advantage -- it's just an indication that we're pretty much on the same track we have been for the past 10 days. What it mostly means is that the Senate is a dead-even tie, which is also what the automated projection and the averages trendlines say. Every indication here is that the Senate is so darn close that a tie is the most reasonable prediction.

In the House, trendlines project to 228-203-(4) (average) or 226-200-(9) (long-term), the averages themseves agree to 227-204-(4) and the automated projection gives 229-206. Given the number of unassigned seats still kicking around the various projections, this seems like pretty good agreement.

Throughout the previous reports of the survey, I attempted to be reasonably conservative, so that when an analyst gave a range of numbers for a Democratic pick-up in the House, for instance, I would use only the lowest number in the range, or the one the analyst said was most likely. For this last report, I wanted to plug in the highest numbers in these ranges to see what happened. The result was this:

    2006 ELECTION PROJECTIONS SURVEY
TOP OF RANGE RESULTS

So here we see the upper range of the projections, and it's not all that different from the bottom range: one additional seat for the Democrats in the House, and one in the Senate. (Of course that single seat in the House is relatively meaningless, while the seat in the Senate gives them control.)

I could use this as a reason to bump up my projection to 51-49 in the Senate, but I just don't think it's going to happen. Although the Democrats seems to be getting their wind back after a couple of days of tepid Republican "momentum", it still doesn't look like enough to me to put them over the top. It could happen, but I'm not convinced it will.

The House is a different matter. Let's take a last look at the generic House polling graphs:

These don't look too different from the last time I looked at generic polling, and that's because they aren't. There's been a lot of chatter about the race tightening up, but there's really no indication of that here -- we're still looking at the Democrats holding about a 14 point lead on the Republicans, which is enormous. That means that if I was inclined to predict a Democratic tsunami (which I'm not) there would be justification for doing so.

However, the entire point of doing the survey was to provide a different kind of data for projecting the result of tomorrow's election, so ignoring the results in favor of emotion or my personal desires seems silly. Looking at all the numbers the Survey has generated all together, it seems to me there's more than enough to support my predictions of 230-205 and 50-50. (I could have gone with 229-206, but I couldn't leave emotion out of it entirely -- the extra seat represents my optimism.)

Prediction:

House
Democrats 230 (+27) - Republicans 205

Senate
Democrats 50 (+5) - Republicans 50

Update: Since so many of the sites in the Survey rely heavily on poll-reading to make their assignments, this Pollster.com post by Charles Franklin on the efficacy of polls predicting election outcomes is quite interesting.

Basically, a poll that shows one candidate ahead by 10 points will correctly indicate the winner of the election 90% of the time, one that shows a 5 point lead will be correct only 60-65% of the time, and polls that show a 1 or 2 point lead are only correct about 50% of the time. Polls that show ties are good at predicting extremely close outcomes.

Update (11/9): It's been pointed out to me, by John Mangino, that I have the Rothenberg Senate numbers wrong. I saw his range of 4-7 Democratic pickups, and automatically took the lowest (4) for his entry, but somehow missed that he said that 6 was the most probable. That would make Rothenberg's Senate prediction 51-49, not 49-51 as I have it here.

It's interesting to note that if I had had the correct numbers, the unfutz automated projection for the Senate on 11/7 would have been 51-49 and not 50-50. However, I'm fairly certain that I would still have taken as my own prediction 50-50.