Diane Rehm on Exit Polling NOW! Internet Broadcast (10-11AM Th)

[On the Road]

Here's the link:

http://www.wamu.org/programs/dr/

Good discussion. I didn't see other posts:

Guests
Karlyn Bowman, resident fellow at the American Enterprise Institute

Steve Coll, managing editor of "The Washington Post"

Andrew Kohut, director of the Pew Research Center for the People and the Press

[snot]

1. Evidence to Pew

Andy Cohut just said he hasn't looked at the evidence comparing paperless to papered. We need to get this to him.

[farmbo]

2. And he's an "expert"?

The anomalies between exit polling at electronic precincts vs paper ballot precincts are THE SINGLE MOST RELEVANT VARIABLE in the entire exit poll controversy.

And this so-called expert is unaware of this? This man is apparently a patsy brought on stay on message for the GOP: "...nothing here but a bunch of kooky conspiracy theorists on the internet(s)."

:grr:

[Eloriel]

3. Are archives or transcripts of her shows available later? n/t

[On the Road]

4. Either Coll or Kohut Said

that the exit pollsters are supposed to record gender and guess at other demographics (age, ethnicity, etc) for REFUSALS and incorporate those in their reports. That's the first I've heard of this. That gives another basis for reweighting later in the day -- the only available information I had been aware of was turnout by precinct.

[democraticinsurgent]

9. that's correct

i was an exit pollster in this election and had to note the gender, race and age range of each person i was supposed to interview and couldn't because of refusal or because they slipped past me (known as a "miss".

This data was reported each time I called in (three times total).

[On the Road]

10. Awake, I Would Love to Learn About the Process, Especially:

What do you estimate your response rate was? That is, your respondents vs. the people you would have polled if everyone had been cooperative? Along with everything else, 30-question survey is a big hurdle.

Also, do you know what data the organization used in order to reweight? I was guessing that it was only geographical, since that's the only official data. But maybe there are more possibilities.

[democraticinsurgent]

11. Over 50%

That includes both refusals and misses. In my three calling periods I averaged 2 or 3 refusals and a couple of misses.

The voter cooperation rate was probably about 70%, if you discount the "misses". That's when someone slips past you when you're not looking, or busy with someone else. I was supposed to interview every 7th person and if i missed them, or if they refused, i had to start my count over. That way the demographics and total voter count would jibe with the actual vote count. Or at least it was supposed to.

When someone wouldn't cooperate, it was usually that they were on the run to work or get kids. A few indicated suspicion or confusion about what I was doing and passed.

I do not have any info on how the data was used, unfortunately, I was just a minion. However, I've read elsewhere that the reweighting occurred late at night when ACTUAL VOTE tabulations were intermingled with exit polling. That seems to be why the "final exit polls" snapped into place with the vote totals. Seems weird to me and not very scientific.

For me, the third wave of exit polls, the ones that were out around 6pm, should be considered the final ones. Unless someone educates me to the contrary.

By the way, the company running the exit polling was very professional and efficient from my standpoint. They gave me plenty of info, training, attention, and calling in was a snap. I wouldn't have been surprised if I had trouble getting thru, but there was always someone on the line to take my data pronto.

[On the Road]

13. I Don't Remember the Spread in Indiana

Was there 5% or more difference between the initial exit polls and the official vote count? If all your refusals had been Bush supporters, was it enough to cause the divergence between the vote and the exit polls all by itself?

Up to now, I've been assuming that the pollsters reweighted the sample based on official turnout, which may have been included phantom Bush votes and not included lost Kerry votes. This seems like a legitimate check to make sure that the geography was proportionate, and would make the exit polls agree with the official count. It would NOT exclude fraud.

But this comment of yours interested me:

However, I've read elsewhere that the reweighting occurred late at night when ACTUAL VOTE tabulations were intermingled with exit polling.

Are you saying that the exit polls were forced to agree with the official vote? That would be illegitimate. I've tried to find details on the web, but have had difficulty. Would appreciate any insight.

[democraticinsurgent]

15. not sure about the spread here

but my precinct was 95% African-American. I think all the refusals were African-American. And every African-American in my polling said they voted for Kerry.

I think the "Bush supporters refused" is smelly BS. If Bush supporters were as emphatic as they would have us believe, they would be standing proudly by their man. Anything else is just wrong if you ask me.

As far as the mixing of actual votes into the exit polls, i read that on DU, but i can't remember where. I'll try to find it and if I do, I'll start a new thread.

[On the Road]

16. Well, When I Was Canvassing in Virginia

people who refused to talk were invariably Bush supporters. That's my experience. But yours was different, which is why I asked.

The outspoken Bush supporters blinded me to another group which I think really exists -- embarrassed Bush supporters who voted on the gay marriage issue but really don't care to defend him personally.

The issue of reweighting is a big one. I could support geographic reweighting based on official turnout, but if the vote was substituted for polling, that's completely illegitimate. I can't believe they would do that.

[toodistracted]

5. exit polling accuracy

Hi--I've been lurking obssessively the last week or so but this is my first post--I'm new.

This morning my brother sent me a link to an article by Prof. Steve Freeman at U Penn, in which Freeman argues quite forcefully that a) exit polls have high rates of accuracy, b) the odds against 3 states having such high discrepancies between polls & votes is 250 million to one and c) the "explanations" that have been offered to explain the discrepancies do not hold water.

Please forgive if this is a duplicate posting.

Here's a link. It will download a .pdf file to your computer.

[PATRICK]

6. Well, if true

it unfortunately SOUNDS ridiculous. I don't listen to NPR anymore. Look at that stacked panel! And Pew hasn't a clue?

This is the soft depression of the story they fear(for various reasons).
Muckraking by the softie station that depends on government handouts? OUR present government? I'm sorry they are not braver than the DNC. Suicide will not be conducted in the open.

I get suspicious of these scientific figures when the results seem an exaggerated absurdity. The same when one of THEIR scientist salesmen that on the swiss chess voting software, the chances of fraud were a million to one. Wild.

And when reality hits, whose exaggeration will be remembered? Only the people with just cause running up against the protective wall of Corporate Pravda(apologies to Pravda of today, this is a Stalinist reference).

The heat we are generating here will not be enough. I believe that. The evidence is clearly ignored then respun. It is simply our honest duty to try. To know it is not crazy to believe the truth and painful not to be able to know more.

[txindy]

7. Welcome, toodistracted!

You've come to the right place! :hi:

[On the Road]

8. Welcome to DU, TooDistracted!

:toast:

[peabody71]

12. Welcome aboard!!

This is where it all began. We are the Media!!

[Straight Shooter]

14. That's a zip file

Cut and paste the contents of the article here, or send a link to Freeman at U Penn.

No one is going to download and extract a file in this manner. Anyway, I hope not.

welcome to DU

[crickets]

17. Why not? I did.

Of course, I checked the zip file with Norton AV before extracting, just to be on the safe side. Nothing is wrong with the zip file, or the three files I extracted. There's a 10 page pdf and two WordPad files of e-mail correspondence.

The author's note on page one of the pdf ("The Unexplained Exit Poll Discrepancy") specifically requests the paper not be reproduced in full on web pages or electronic bulletin boards and the like because it is a draft of a paper intended to be eventually published.

I've only skimmed it so far, but to my layman's eye, the information and analysis definitely deserves a look by anyone hunting down more data. The only thing that makes me uncomfortable about the file is the shorter "do not circulate, reproduce, post," etc. note on subsequent pages. I don't recall where I originally found the link on DU yesterday. I'm now wishing I had bothered to look at the file immediately and ask the poster where the link came from, whether the author is aware it is circulating, and whether it is appropriate to post. I'm going to send an e-mail to Mike Lewis (the person providing the zip file) and ask.

:shrug:

The e-mail correspondence also contains a great deal of data and one very interesting exchange between Mike Lewis and Suzeth Pimentel of the Election Reform Information Project. Go Mike!

Again, I've only skimmed to get a general idea, so...

[Roger_Otip]

18. exit poll analysis great!

that's the best explanation i've seen of exit polls which people should take a look at and get out to the media.

i've put a copy of the pdf on here - http://www.legjoints.com/AmericanElectionFraud/ - a page where i'm trying to get together links to things i think are good - that's a wiki page so anyone can edit it and put things up there if you want.

[newyawker99]

20. Hi Roger_Otip!!

Welcome to DU!! :toast:

[newyawker99]

19. Hi toodistracted!!

Welcome to DU!! :toast: