Bits and Pieces, Four Days Out

self | 11/3/06 | LS

[LS]

This is kind of a jumble---so many small pieces of information and/or thoughts, so let's get to it:

1. All the polls are off, part umpteen. Earlier this election season, I thought all the polls were off because they were oversampling Dems. Recently, I've been seeing some of the panic-meisters at NRO claim knowledge of GOP "nightlies" or "internals." I don't doubt they are getting information from somewhere, but I think one of two things is happening.

A. Their sources either aren't as good as they should be, or are RINOs deliberately feeding them crapola. Before you dismiss the latter, remember NRO bought into the "exit poll" hysteria in 2004, and I had to come on this board after doing ON THE GROUND POLL FLUSHING that showed Bush would win OH and calm some of you down. My point is, just because they have an "inside source" doesn't mean squat.

B. More likely, however, is a revision of my "polls are sucking" view. They are still tilted Dem, but the methodologies---even those being inadvertently accepted by the GOP pollsters at times--- are badly flawed, and still tilted Dem. Consider that when polling was first used extensively for elections, it was a face-to-face business with an 80% response rate. Telephone polling drove the response rate down further, and now, after bad experiences with pollsters in 2004, the response rate is under 20%. A decent sample size of 1,000 respondents then requires an unimaginable 5,000 nightly contacts!!! Folks, you know that ain't happenin'. Moreover, that still wouldn't do it, because you would need to get your "quotas" of Dems, Indies, and Republicans.

To conduct a poll that was anywhere near accurate, you would have to make upwards of 7,000 contacts!!!

And we still aren't even talking "likely" vs. "registered" voters. For each category you add, you have to geometrically increase your calls. One method pollsters use to "determine" whether you are a "likely" voter is to ask you and take your word for it. That's highly unreliable, because people like to be thought of as good citizens, so they either lie or have intentions to vote, but don't. A more reliable method involves asking if the person voted in 2004, then ask a bunch of unrelated questions, then say, "did you vote in 2002?" then ask more unrelated questions, then say, "did you vote in 2000?"

Now, when we canvass for Blackwell and do lit drops, we only drop at houses where the person has voted the last four elections. We know that from their record . . . not what they SAY!

So here is what I think has been happening: the pollsters are making a couple of thousand calls a night and taking people's word on their party affiliation and on them being a "likely" voter. You can chalk up 1-2% error right there, in the GOP's favor.

A second source of error, however, involves the technology. Somehow---and I haven't quite figured it all out yet---the cell phone and caller ID technology works against polling Republicans. Now, that's strange, given that Democrats (especially blacks) seem more wedded to their cell phones, but I'm convinced it's a factor. It's like obscenity: I can't quite define it, but I know it when I see it . . . again, and again, and again. But I digress.

I ran this theory by the head of the Warren Co. Blackwell effort, and he agreed 100%. This fellow is a Ph.D. in criminal justice/stats. Last week I ran it by a poly sci prof at Hillsdale who worked on many campaigns in MI, and he agreed as well. None of us can identify exactly how the methodological bias works, but its clear it exists.

2. Even if the polls weren't off, they simply don't begin to measure turnout. This kind of goes back to the "likely voter" issue, but we now have EVIDENCE from early voting and absentees that GOP voting is substantially higher than in 2002. Off year participation levels in OH for Republicans is (thanks to Common Tator) 58% of a presidential year. I'm betting in OH, for ex., we see something close to 60% or even a little more. More important still, I'm sensing from the ground here a massive apathy on the part of the Dems. Canvassing Dem areas, you never see bumperstickers, or yard signs; there have yet to be, anywhere, any DEM ground troops in the Dayton area!! My assessment? Whatever you ADD to the Republicans, you need to also subtract a point or two from the Dems' 2002 turnout levels. They won't get there this year.

In other words, whatever your polls say (unless in an overwhelmingly Dem state like NJ or RI), you can figure on 3% more GOP and 1-2% less Dems actually voting. (In a red, red state like Montana, I think you can increase the GOP %, in a blue state like PA, you have to temper it some.)

Now, what do we know for sure: In just Warren Co., my next door neighbor which is deep red, the Blackwell people have made 9,800 calls in ONE WEEK. My team in Dayton has by itself hit 2000 homes in three weeks. We all go out tomorrow again. Even the rural GOPers are getting drivers coming by and putting stuff in their doors.

I haven't seen any "internals," but one poll had DeWine down 8, one had him down 2. Split the difference and figure the polls have him down 5. That is VERY winnable in OH. That's right at "a turnout victory." Blackwell is apparently close to this same spot, except he's had some fantastic ads with Rudy Guliani. I can't imagine those great ads won't make a little difference.

Moving on to the House: we are starting to see polling (again, beware) showing two of the three IN seats coming home. Chacoba, once "dead," is within 3; Sodrel, always trailing, now leads. In NC, Taylor, again considered "a goner" according to NRO, is now tied. Negron is now figured to win the Foley seat; Sekula-Gibbs tied in a deeply red district, and will win that. Wilson now "safe."

There is concern over the CT seats, where right now only one of three GOPers leads, but again, this is "polling" and I think, even in CT, these are GOP wins. Drake in VA is now moving into safe territory. I never did think Steve Chabot was in trouble in OH, and I'm hearing that Pryce and Padgett are in good shape. Roskam now ahead of Duckworth in IL.

In AZ, it's simply bogus to suggest that J.D. Hayworth will lose. Randy Graf, however, can't break into single digits. The AZ papers say he simply is a one-note samba, and can't speak to health care or any issue but immigration. It's clear immigration is the #1 issue in a district like this, but it will not be the ONLY issue. He has about one day of recovery time. If he trails by double digits on Sunday, he's finished, even in red AZ. Most people now think the CA seats (Pombo, Doolittle) are safe. We still could lose one in IA, one in IN, Curt Weldon (PA), the open seat in NY, probably one of the three in CT, Graf, and O'Donnell (CO). There may be another two I'm missing. That's nine. I don't have a read on Gerlach.

But there are now four very vulnerable Dems: Barrow and Marshall (GA), Carson (IN), Bean (IL), plus an OR seat that is somewhat beatable.

In the Senate, I have Talent, Allen, Corker and Burns (yes, Burns) in the "safe" column; Steele "ahead and nearly safe," Kean "slightly ahead," DeWine, McGavick, and Bouchard "slightly behind," and Santorum and Chafee behind outside the "turnout margin." But finally Santorum has moved a little, and by tomorrow could be within the turnout margin. Chafee is in such a blue state, he is my most vulnerable candidate right now---I know that makes some of you very sad (sarcasm). In other words, I think we'll at least pick up one in the Senate; and if the close ones break for us, three. Chafee could even stage a comeback and the number would be four. That's right, four. Right now, my best guess is +1 in the Senate for the GOP.

In the house, I'm reluctantly abandoning my prediction of +1. All the IN and CT races would have to go to the Republicans, then we'd need a comeback in either CO or AZ. My guess, now, is that we lose fewer than five, picking up three of the four Dem seats.

But stay tuned. I'll know more after I walk tomorrow and talk to the on-the-ground peeps.

[ConservativeGreek]

You have one more prediction coming out Monday night?
I'll be eager to see that.

[veronica]

The answer is : (b) groupthink...

[PGalt]

Test marketing 1 precinct only (anywhere USA) and explaining to the participants that they are part of an historic new polling company.???? (Let's go)

[Miss Marple]

I beleive it isn't that the polling companies don't know how to phrase a question, as much as they are doing the bidding of those who hire them (usually the media). The whole thing is one huge flim-flam.

Someone knows how to get accurate answers. I live near a company that does markey research. Before I can even qualify for a survey, I am asked a bunch of questions that detail my age, income group, prior purchases, etc. Then I am put on a call list for certain products.

I cannot imagine that the polliing companies are less savvy than the market researchers for breakfast cereal. I therefore have to assume that they aren't interested in providing an accurate picture, but rather are interested in providing a point of view.

Oh...and while I will take the time to answer surveys, my husband will NEVER do it.

[Sal]

They are all using, I think, bad data. No different than in 1994 when all the then "all stars" had the Dems winning big.

I'll never forget '94 when all of the MSM were claiming the whole country was terrified of Gingrich's bomb throwers and they were going down to certain and humiliating defeat.

I especially remember Peter Jennings crowing triumphantly that these guys had to be wishing they'd never signed that contract or even seen it...

That was the first time I realized how deeply in the tank Jennings was and I had always watched him because the other two were worse.

IMO they absolutely believed what they were saying. They were too ecstatic and it looked like they were really feeling it, not just thinking it.

Then that stupid, untrustworthy electorate had a temper tantrum...

[libravoter]

Last night, my wife got a polling call. The first thing they asked "Is there a male that we can talk too?".

I worked for a survey place in high school (1986-90). I got hired right before they started asking everyone for ID to prove they were 16 and it was great job for a teenager. Paid $2 more than minimum wage and you never got burned by a fry-o-later.

What you say is true -- women tend to answer the phone. So the first day of a particular survey, we'd ask to speak to the "head of the household". Sometimes the woman would pass the phone to the man. Sometimes the woman would say she was the head of the household (because she was single or because she was ornery). And that was fine. But the second day, we'd have already reached our quota on women and were just asking the men.

No one was happy.

[AuH2ORepublican]

The OH-13 is completely off the radar screen, since people figure that if OH Republicans are trailing in GOP districts that a district that only gave Bush 44% won't be competitive. I have no idea whether Foltin can pull off the upset, but have a feeling that it will be closer than people assume. Foltin is our best possible candidate, while Sutton has some problems, including how she moved into the district to run.

[AuH2ORepublican]

I hope you're right about Clay Shaw. A couple of days ago a Palm Beach County GOP insider (the FL-22 takes in almost all of the Palm Beach County coast) told me he was worried about Shaw's seat. We need to have a big GOP turnout on Tuesday.

[cgbg]

It is amazing that the pollsters are claiming that Nancy Johnson (Republican incumbent) is losing in such Republican strongholds as Avon and Simsbury. I am convinced that these busy Yuppie households with children are heavily using cell phones and answering machines and are unreachable by the pollsters. When I talk to these folks they are so busy running their kids around from one sporting event to another that there is no way they would have time to talk to a pollster.

They will press the lever for Johnson on election day while the pollsters gaze at their navels while they figure out what happened.

[axes_of_weezles]

http://tks.nationalreview.com/post/?q=OWFhNDhjOTNmYWNlMzcyOWVjNGY0OTI0NWExM2NlYmE=


A quick thought on Arizona 8th District and illegal immigration
11/03 01:43 PM
The Corner has a good collection of links to news stories covering the House race in the Arizona 8th District — Democrat Gabrielle Giffords vs. Republican Randy Graf. It's Kolbe's old seat, one of the races every observer is chalking up as an automatic Democratic pickup.

I've indicated that Republican Randy Graf is my Super Mega-Shocker Special What-The-Hell-Are-You-Smoking-Jim Upset Pick for 2006; I just figure the illegal immigration issue is huge, stirs passions, motivates voters, and that a lot of people may support a Minuteman, a hardline view, but don't want to tell it to a pollster or others for fear of being called racist, nativist, etc.

And as a candidate the state GOP wasn't eager to see to win the primary, Graf was seen as unelectable and expected to get blown out by 20-30 points. So the fact that he's not getting blown out, and is in fact down 12, 10, 8 suggested to me that maybe there's something going on out there.

On the other hand, if Graf does get blown out as expected, staunch opponents of illegal immigration are going to have a challenge persuading the GOP that their position is an electoral winner. It's easy to see the mentality taking root, "If opposing immigration couldn't help stave off a blowout in Tucson, Arizona, it's not going to help much anywhere else..."

[MNJohnnie]

I think I can help you with the technology aspect.

People who are less technologically savvy tend to be older. For example, my older siblings are not nearly as technologically savvy as I am and I am not any were near as tech savvy as my nieces and nephews. So the people who are likely to still be answering their phone instead of using technology to screen out random callers will be rather towards the elderly end of the voter spectrum.

In additions, there is usually a cost in using technology to restrict access to your phone line. Therefore those on the poor end of the economic spectrum are less likely to be accessing technlogy due to the costs

So you are going to have a build in bias toward older poorer respondenets. Since the average Boomer tend to be towards the Left of the political specturm so they would tend to bais the polls to the Left.

Question is HOW much are they baising the polls? I guess we will find out on Tuesday.

[conservative blonde]

Somewhere between "group think and scare psychology." I remember election night 2004 when the whole bunch except Fred Barnes I think, bought into the skewed exit polls showing a big win for the Rats.

[Seattle Conservative]

"They wanted to speak to the male voter in the house because they already had more women than their quota."

FNC said earlier that across the country 33-38% were undecided and a larger % of them were women (obviously, some areas are much lower than that, they quoted one area as only 5%).

Mort said on Special Report last night that the Dims said that normally the undecideds break 2/1 for the incumbent, but it's more than that this year. I'm not buying this last one, I think it's more Dim spin. Also, IIRC there is a large percentage of undecideds that don't vote.

[1035rep]

 Michael Barone says in 2004, the electorate that went to the polls or voted absentee was, according to the adjusted NEP exit poll, 37 percent Democratic and 37 percent Republican.  The recent national polls show Democrats with an advantage in party identification in the vicinity of 5 percent to 12 percent.

"If you could go back in history and conduct polls, I don't think you'd find any, and certainly not many, two-year periods when the balance in party identification shifted from even to having one party 12 percent ahead of the other."

[shield]

Thanks for the heads up. ;o) Interesting....makes my thought process less anxious...The Lord keeps letting me know there will be no problem for the Republicans this election. You know that quiet knowing in your spirit...coming from Our Creator...

[EDINVA]

your fingertips to God's ears

[ClaireSolt]

I studied statistics and worked for both Gallop and Yankelovitch and have been in telmarketing. The sample description is based on the pollsters analysis and projections. They tell the callers the demographics they want. There are huge rooms that work contacts and I do not think 7,000 contacts per night would be any big deal. Depends on the quality of the list for the job, and there are many special lists. By now, they probably have millions of people they know will answer questions.

Obviously, pollsters only stay in business if they show a track record of accuracy. They are probably finding ways to compensate for technology issues you bring up and keeping them secret. Of course, there is a huge difference in professional polls for the campaigns and those designed to create news.

But is really all rests on the analysis that weights issues and factors to draw up the sample. That's why they credit Rove with finding the exurban vote in 04 and say the exit polls in 2000 in FL were wrong because of new movers they didn't characterize correctly. That could easily be a factor in OH were people have been leaving in drove. Who's left?

[Toidylop]

If I relied in polls and "so-called" pundits, we'd have President Kerry in the WH now!!! nuff said.

[rwfromkansas]

The Dems here in KS seem motivated....lots of signs and not very many for the GOP.

But, that always is the case, and they still lose.

[LS]

Whether I believe a poll has no relation to whether I vote. Unlike some sour-grapes, single-issue whiners, I vote R when I like the candidate and when I don't.

And you don't have a clue what "sources" the people I know use. I suggest you stop making assumptions. Try presenting an argument with just a shred of real evidence that Graf is anywhere near 5%, and I'll be happy to trumpet it to the heavens.