The Slip

You may be worried about (some) polls showing the Democratic lead in the generic congressional ballot slipping. Certainly, I'm at least a bit worried. Democracy Corps, however, tells me not to worry. They've been running a poll of the 50 most competitive congressional districts where they use the candidates' names. They say the with-names questions has consistently shown an aggregate Democratic lead in these races, but that it's been consistently smaller than the generic ballot lead. Their most recent named poll, meanwhile, shows no change in the Democratic lead in the with-names question. The Democratic lead in the generic question, meanwhile, has declined so as to converge with the Democratic lead in the with-names question.

This, according to DCorps, is all that's happening. As you go down to the wire the difference between the with-names and without-names version of the questions goes away. But the without-names questions was always worthless. The with-names number is the real number, and it continues to show the same lead it's been showing for a while -- one that's good enough for big Democratic pickups in the House. I don't really have the chops to assess this argument in an expert manner. In its favor, I'll say that it sounds convincing to me. Against it, I'll say that DCorps proved overoptimistic about the 2004 election.

Comments

Personally, I'm inclined to believe people who have something to lose when making these sorts of predictions, like the gamblers over at Tradesports. According to them, the GOP has a 24% percent chance of holding the House and at 74% chance of holding the Senate.

Posted by djLicious | November 6, 2006 3:38 PM

ugh...this is all starting to read like Ruy Teixeira's endless speculation in 2004 that the real number to watch was the likely voter v. the registered voter (or was it the other way around?) and that one of them showed a definite Kerry lead. Through that I learned even more important than not trusting polling trends, it's important not to trust those who explain away polling trends using some kind of statistic sieve that happens to confirm their hopes.

Posted by festus | November 6, 2006 3:39 PM

How big is the market? Knowing there are people like you, and wanting to create a positive buzz, how hard would it be to move the market?

Posted by SomeCallMeTim | November 6, 2006 3:41 PM

I don't think you need to start panicking and hyper-analyzing just yet. TPM has a list of 7 endgame polls and the two most recent show a return to a rather large D lead.

http://www.talkingpointsmemo.com/archives/010859.php

Posted by Ben | November 6, 2006 3:41 PM

I've heard almost everything. The races are getting closer. The Democrats are expanding their lead. The individual races are getting closer. Your pals have the individual races staying the same. The generic advantage is widening. The generic advantage is shrinking. Rove and Bush say that everything is hunky dory for the Republicans. Other Republicans have already rented the abattoir for the sacrifice tomorrow night.

So, the aggregate sez, "Nobody knows anything."

As for me, I've never been happy with a single election since I came to my majority mumblemumble years ago. I've never voted in the fall for a candidate for national office that I liked and respected. Had I been 2 years older, I could and would have voted for John Sherman Cooper, a Republican, for Senate. Happily.

Posted by Jeffrey Davis | November 6, 2006 3:45 PM

Okay MY, fair enough.

But what I wonder is all those people (the 7-10%) that find "my ideal Democrat" better than the "ideal Republican" what kind of "ideal Democrat" do they actually have in mind if they actually keep coming back to the disastrous Republican regime?

I don't think Dems should satisfy those people, but as a curiosity: What the hell do those people want?

Posted by MNPundit | November 6, 2006 3:48 PM

For what it's worth, the Post today had an unnamed Republican stragist making roughly the same, or a congruent, point, with the focus on where Republicans can gain:

A senior GOP strategist said party officials anticipated that the generic vote would tighten, but they do not consider the shift significant enough to change the contours of this election. More than 20 GOP incumbents are tied with their opponents heading into the final days. "It is the 50-50 districts that turnout can help," said the strategist, who like others spoke on the condition of anonymity in order to talk about strategy.

Posted by Jeff | November 6, 2006 4:06 PM

several weeks ago or so, during the brief was it larry craig rumpus?, prof instanitwit announced that he had been considering voting democratic (meaning, presumably, for ford) but he just couldn't do it after that awful democratic smear campaign aimed at was it craig?

i believe that prof instanitwit had exactly as much intention of voting for a democratic candidate as george bush did of becoming a uniter, not a divider.

which is to say that i didn't really believe the generic numbers, because as push came to shove, people tend to vote their established patterns.

i've felt all along that the dems had a 50-50 chance to win the house and a zero chance to win the senate, and i still think that way....

Posted by howard | November 6, 2006 4:20 PM

If one believes that the generic/without-names question "was always worthless," then I'm not sure why one would be worried about a couple of polls showing a tightening of the numbers on...that very question.

There's been some decent news for the GOP in a few of the Senate races, but I can't think of a single House race in which the Republicans have had any suddenly good news to point to in the last week or so.

Posted by Haggai | November 6, 2006 4:21 PM

Looks more like 80-90% in the House and 50-50 in the Senate, howard. We'll find out tomorrow night!

Posted by Haggai | November 6, 2006 4:23 PM

Personally, I'm inclined to believe people who have something to lose when making these sorts of predictions, like the gamblers over at Tradesports. According to them, the GOP has a 24% percent chance of holding the House and at 74% chance of holding the Senate.

But these punters don't have any sources of information on which to make these guesses besides the public polling, right?

Posted by Matthew Yglesias | November 6, 2006 4:51 PM

Yeah, that Tradesports business isn't based on any polling or reporting info that isn't already available in public to the rest of us. One can have faith in the unicorn-like powers of the market to transform all of that into an unimpeachably awesome predictor, but what comes out of it is pretty much whatever the CW of the day is.

Posted by Haggai | November 6, 2006 4:58 PM

MY said:

"But these punters don't have any sources of information on which to make these guesses besides the public polling, right?"

But since they are putting up money they are more likely to be paying attention and less likely to be spinning (lying) than the average blogger.

Posted by James B. Shearer | November 6, 2006 5:19 PM

The gamblers arguably have an incentive to study polling and figure out which poll has the best likely voter screens and predictive modeling. If you can accurately predict turnout in this election based on past polling, you can probably get a good idea of who is really in the lead in some of these districts/states. They could also purchase the internal numbers from the pollsters and have more info than is made public.

There's lots of info on this stuff at www.pollster.com

Posted by Bob | November 6, 2006 5:31 PM

But these punters don't have any sources of information on which to make these guesses besides the public polling, right?

As to whether or not the gamblers have inside info, I wouldn’t know, but they would certainly have a strong incentive to find it. It would also seem to me that they would be paying a lot more attention to all the ins and outs of polling, like what James and Bob are saying. This is not to say, of course, that pundits aren’t, but pundits don’t lose money when they’re wrong. (In fact, it seems like there’s a whole cottage industry for being wrong.)

Posted by djlicious | November 6, 2006 8:41 PM

"But the without-names questions was always worthless. The with-names number is the real number"

Meh. What makes the DC methodology this cycle really interesting isn't the names, it's the fact that they're polling the 50 weak GOP CD's that are this year's playing field.

Posted by Petey | November 6, 2006 9:09 PM

As to TradeSports, it's obviously not predictive. But it is a very nice summation of the available CW quantified into an easy to read number.

Posted by Petey | November 6, 2006 9:11 PM

In what has to be one of his last major later ad buys, Schwarzenegger uses the same voiceover guy Reagan used for the "morning in America" ads.

Schwarzenegger will (not suprisingly) win tomorrow, but the voiceover guy sounds like he is about to drop dead.

Make of that what you will about Reagainism.

Posted by Linus | November 6, 2006 10:15 PM