I'm looking mostly at Larry Sabato (21-26), Stu Rothenberg (18-28), and Charlie Cook (20-35) where the average median seat gain by Democrats currently sits at 24. So I have it at 20-30.

Interesting quote from Larry Sabato for Repub naysayers and pessimists on our side: ''In 2004, I'm proud to say the Crystal Ball's predictions were accurate in 475 of the 479 Governor, Senate and House races on the ballot (99.2%) on election night. Of course, looked at more critically, we missed one House, one Senate and two Governor contests.''

it's not going to be as big as '94, because since that time, due to the '90 and '00 polarized redistricting, the dynamic has changed:

The best estimates put the number of seats gained by Democrats at  around 20 from this conservative, and 10-24 from the Republicans at RCP, at 8-26 by CQ's current projection,  and anywhere from 16 to 40 from Pollster.com, and from 24-29 by Chris Bowers here.

Did it never occur to you that the conventional wisdom can be wrong?

I ask you to consider two facts:

- All projections from statistical analysis of polling are inherently flawed.  The statistics rely on ludicrous assumptions that every voter can be classified accurately and lumped in a neat box with all other like-minded voters.  Then we have the usual allowance for error.  That sometimes leads to the hilarity of double digit leads for the Democrat within a 3-5% margin of error and the same for the Republican in another poll.  All that is without considering all the guesswork that goes into estimating actual voters as compared to registered voters.

- Computer-directed districting to maximize election victories could be a double-edged sword in a wave election.  The idea was to dump as many Democrats as possible into one surrendered district and then thin out the Democratic voters in Republican districts and spread them evenly around to dilute their value to worthlessness.  In a wave election, those "thins" could become the majority.

The future is known only to prophets and fools.

I suspect there are many more fools than prophets. The odd faith of many in the infallibility of statistics is offered as powerful evidence.

Best,  Terry

I've come to the conclusion that none of us have much of an idea what's going to happen based on individual House polls. There's such a variance in the polls; it's ridiculous that districts as disparate as NV-02 and NY-24 have single-digit races. There has to be a bigger difference in those races. Either the polls are over-stating how close some of the GOP-held districts are, or they are missing the wave in many swing districts.

My optimistic side says that NY-24 looks like a far easier win for Arcuri than the national pundits seem to think, so I'm excited for the results next Tuesday nationally. My pessimistic side says ... well, pessimistic things.

There's a lovely piece in the print edition of the latest Mother Jones which describes the teenaged Karl Rove showing up for high-school debates with ten shoeboxes' worth of note cards while his opponents only had one.  This scared the crap out of them, but what they didn't realize is that nine of those ten boxes were stuffed with blank note cards.  (Personally, the people that scared me at the debates were the ones who showed up with NO note cards; they were either total idiots or so damned good -- like Patrick Fitzgerald good -- that they didn't NEED notes.)

As always, a nice piece, Jerome.  

The historical note on 1974 was that a Republican driven reform in theHouse rules made it a Use It or Lose It decision of a different sort.  House members had traditionally been able to pocket all unspent campaign funds when they retired.  This would no longer be true for any congress person in office after January 1995.  The number of open seats that year went bananas, particularly for Democrats in the south who were fighting against a long, hard regional trend (with a lot of success).  Maybe Marty meehan would retie from the House if he could convert that $4.9 million to a one-time cash bonus. Of course, racially based gerrymanders in some states and just gerrymanders in others also favored (and continue to favor) Republicans.

FWIW, my take on the uncertainty factor is that a seat by seat prediction should give Democrats about 25 seats with an outside chance of a few more (30 max).  Some of the macro trends are worse for Republicans than they were for Democrats in 1994 and more closely resemble 1974.  That's a fairy tale as there is no way Democrats win 291 seats this time (plus 88 or 89 depending on how one accounts for Sanders).  For the past two weeks my thoughts have been that there is some chance of a big-wave or tsunami election.  Today I'd put it at 1 in 3 but I have no real hard evidence for this. In five of the last six big-waves, the winners picked up from 45 to 55 House Seats.  So, 25 with a 33% chance of 50.

Part of the story in off year elections hs been Democratic turnout.  We peaked 20 years ago getting about 35 million votes and have slid a small margin from this (2 to 5 million votes) ever since.

The number, and maybe the quality of House polls makes the whole thing more uncertain.  Two out of three or four out of five polls look encouraging.  Some small percentage, 5 or 10% are mind bogglingly bad and cause me to doubt all the others,  If X, who was neck and neck two weeks ago is suddenly down by eight, ten or even twelve maybe the whole house of cards is tumbling down.

Reading polls is mostly science with a little art.  I've gotten so I can pick out most of the outliers as a matter of course.  So, thanks again Jerome for filling out the picture.

...by something like 15 points in 2004. Didn't matter, becaus ethe GOP got more of tehir voters to the polls.

As I have ssaid, over and over again, Macro beats mico. The reason the GOP GOTV operation works is because the 2004 election was close.

OIf we have a big lead going into these races on election day, tehir GOTV operation cannot win it for them. They can't turn out enough voters to overcome the MACRO effect.

That's why the strategry of the GOP, particularly in these senate races, if to keep them within the marhin of error in the polls. They hope they can overcome small defecits or tossups with their GOTV operation.

Henmce the massive negative advertising via independant expenditures.

http://news.yahoo.com/s/ap/20061030/ap_o n_el_gu/oklahoma_governor

If Ishtook wins, I'll send Karl a nice bread-and-butter note.

NO!!! NO!!! NO!!!! Please don't make any more predictions, Jerome. Your past record sucks, and never fails to disappoint on election night. In the future, at least please put a disclaimer in the first sentence...

"Warning! My past record of predicting Democratic takeovers has an accuracy rate of zero."

Now I'm completely discouraged after reading your optimism.

I say - why let them have any hope at all?

Seriously tho. I was trying to make sense of the "skew "  on the generic ballot and could not quite understand.

Jerome, can you please explain the "skew" diagram?

I do not want this post (or blog for that matter) to leave the reality-based community for the darkness of conspiracy theory, however, given "reality-based" analysis from groups like Princeton regarding vote tampering (http://itpolicy.princeton.edu/voting/), one must question the possibility that the GOP's desire to stay in power knows no limit.  If the GOP does have 60+ polls/week, they would likely know exactly which machines to hit and by what percent to squeak by with a victory and avoid suspicion.  How do we know this isn't happening?  Maybe malicious vote-counting software is "THE math" to which Rove refers?

You're supposed to be smart. Don't you understand gerrymander math? http://www.danablankenhorn.com/2006/10/g errymander_mat.html

Short version. If the goal is to lock-in as many seats for your guys as possible, then they're all taking an average margin of 60-40 (assuming the state is 55-45). The "more gerrymandered," the narrower the average margins for "your guys."

So in a "wave" election, they all go down. The gerrymander actually magnifies the results. It only protects incumbents in states that were gerrymandered by the other side -- because all that side's voters were put inside those districts.

This is the only way such math can work!

I'm interested in this Gallup poll image you link at the top. It's good news for the Democrats, certainly, but what I find interesting is that the massive approval that the Democrats held in the poll immediately after the Mark Foley resignation had returned to early September levels by the very next poll. Does this indicate the effects of Mark Foley have worn off already? Or should we take it as important that in the post-Foley period, the Republicans' approval numbers did not rebound nearly as far toward "normal" for this year as the Democrats' approval numbers did?

I wish. Porter is absent but not his spots. The commercials run day and night. I know, because I click away from them as soon as the familiar opening scenes show up.

There was a local article the other day detailing the TV buys for the final week among candidates in Nevada. Porter and Ensign are dominant. Hafen is well funded and her spots are excellent, but not as frequent as Porter's. I read the GOP will be putting money into that race so their polling must be nervous level. I was hoping for a better net than the new Majority Watch poll, which gives Porter a 51-44 lead over Tessa Hafen. But that feels about right. My estimate is 4-6.

Someone asked how NV-2 can be competitive. We have an exceptional candidate for that in Jill Derby, a rural Nevadan who has ideal themes for a Democrat in that sprawling district. I'm even hearing her radio spots in Las Vegas, unheard of for a Democrat in Las Vegas since the district is primarily determined by the northern Nevada vote even though it extends throughout the state.

Derby would have won this damn thing if Club for Growth wingnut Sharron Angle had pulled out the GOP primary. She failed by a few hundred votes. Make not mistake, moderate Republican Dean Heller is the favorite over Derby. I think the new poll with Heller ahead by 8 is probably closer to accurate than the ones showing it virtually even.