List of Exit Polls - Edison/Mitofsky - NEP

[BeFree]

http://www.exit-poll.net/election-night/stateraces.html

Not every state and not every race.

Was reading a DU page yesterday that claimed there would be no release of estimates before 5pm. That two guys were gonna be locked up in a room and data from around the country would flow in, but nothing let out until 5.

[FloridaPat]

1. They want to make sure the exit polls match the Diebold results

before people get the idea the election was rigged.

[SheilaT]

2. Exit poll results

are not released before close of polls. There was a huge problem back in the 70's, as I recall, when they were released throughout the day. But exit polling is one of the reasons networks can, if they want, announce a winner in a race with just a tiny fraction of the results officially reported -- they've got good, reliable results.

Actual machine results, real polling place results, regardless of voting method or any chicanery occurring, take longer, because the polls generally close at 7pm (exact times vary) and then it does take some time to tally results in any precinct and get them reported to whatever official site there is, and then get those results out to the news media.

So it's more than silly to see not releasing exit poll results until 5pm means anything beyond what I said in the first paragraph.

[BeFree]

3. Yes

You are absolutely correct... I think. Here's why:

Exit-polls can reliably predict winners with just a tiny fraction of the reults reported.

Releasing the prediction before 5pm means nothing.

The only reason not to report the estimates before 5pm - it has been said - is that doing so may keep people from going to vote after 5pm since the race had already been decided.

[SheilaT]

4. Exactly. I seem to think

that happened some time in the 70's. I know for sure that in 1980 Jimmy Carter conceded before the polls had closed in California. Whether that was because Reagan had already gotten enough electoral votes, or if it was from exit polling, I don't know. I was actually out of the country that day, having already voted by advance/absentee ballot.

But exit polling, done right, is phenomenally accurate. It's only in the last few election cycles, when the final "official" tallies don't match the exit polling that the idea that exit polls are not reliable come about. So now there's doubt all over the landscape about polls and actual outcomes, which is not at all good for confidence in our elections.

[EFerrari]

5. I voted for JImmy Carter after they predicted he would lose.

I don't remember if he'd conceded or not. We, as commuters from a mountain community, always voted with the last batch of voters.

[Febble]

7. My understanding

is that while the estimates (the estimates of state results that allow networks to "call" each state) are dynamically reweighted to the vote-returns, the cross-tabulations are not adjusted to the vote returns until 2 or 3 hours after close of poll. So if you want to know what the cross-tabulations look like before that, check out the cross-tabs as polling closes in each state.

In 2004, CNN had them, and while I do not know for sure that they will post them again, I have heard nothing to suggest that they will not.

The embargo, as you say, only refers to release of data before 5.00pm Eastern. Unadjusted crosstabs would normally be posted after that at close of poll. And, as I say, my understanding is that these will not be adjusted to the vote returns for 2 to 3 hours.

But remember, they are not "estimates" - and unless a call made for that state at close of poll, it means they have not reached 99.5 confidence level (whatever back-of-the envelope MoE anyone feels like calculating at home).

[BeFree]

8. Well

Your lack of understanding is palpable.

The first numbers that come out, this time not before 5pm, will NOT have any offical results folded into them. It is what we call *raw* numbers from exit-polling. IOW, single source.

Since you are not American, one can guess that you wouldn't remember when the news called elections as early as 5pm. How did they do that? Using *raw* exit polling numbers, that's how. And they were always proved to have correctly called the winner.

[Febble]

9. Well, yours is too

or at least your understanding of my post.

What I said was that the first cross-tabulations that appear (if they do), at close of poll, will not be adjusted to the vote returns (because at that point there won't be any), although they won't actually be "raw". Raw enough, though. And as I said, my understanding is that these won't be adjusted to the vote returns for 2 to 3 hours. So grab them while you can.

I gather you agree with me.

[snot]

6. Remember this:

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=364&topic_id=2546367

ABC, CBS, NBC, CNN, Fox and AP Defy Conyer's Request For Poll Data

The theft of American elections is a media issue
by Andi Novick, Northeast Citizens for Responsible Media
October 31, 2006

The loss of our right to vote (or have it counted) is a media issue which is why you are all getting this email. I've talked a lot about the effects of media ownership consolidation on the destruction of democracy by permitting a few corporations to fail to report the essential information we need to be a self-governing people. Not only does the mainstream media refuse to investigate and report on the impossible discrepancies between the 'official' count and the exit polls, but they are actively complicit in this fraud. ABC, CBS, NBC, CNN, Fox and the AP own the exit polls and have defied John Conyers' request for the raw data, keeping that data secreted from even qualified independent researchers.

[althecat]

10. They are only polling senate races...

That will make it considerably harder to steal the house.

[althecat]

11. there is a limit of 15 individual state subscriptions for any news organization

3) State Surveys deliver individual state polling results for the senatorial and gubernatorial poll results where statewide elections for these offices occur. Please note that there is a limit of 15 individual state subscriptions for any news organization.

WHY????

[democraticinsurgent]

12. I am an Exit Pollster for Edison-Mitofsky

...as I was in 2004.

Remember when one of the late Warren Mitofsky's reasons for the discrepancy in 2004 was lack of proper training?

I spoke up then on DU. I thought the training was excellent and the job very straightforward and hard to screw up unless you tried to screw it up.

And I can tell you that the training this year was virtually identical to what I received in 2004, so if there was/is an internal agreement that pollsters weren't well trained, they didn't do ANYTHING to address it.

I can also tell you that communication between my contacts there since I agreed to poll again this year has been excellent. Emails, phone calls, overnight packs. Everything is done very professionally.

What happens after all the data comes in, who the hell knows.

But I was, and am impressed with their work from my end of things.

Rick

PS Obviously I will be in the field all day election day, but ask me anything. Should be interesting!

[althecat]

13. So they changed nothing at all? That is really weird...

Mitofsky in his public reports seemed to blame the youth/clothing/attitude of the pollsters?

No instructions on clothing, attitude, etc...

[Awsi Dooger]

14. According to Lenski they made some changes

This was on Pollster.com a couple of days ago. He mentions more training and apparently fewer young interviewers. Rick certainly has first hand knowledge, and perhaps the training and prelims didn't vary as much as Lenski implied.

Also, on the bottom of this link Lenski describes exit polling and how it would differ if designed solely to validate election results, as opposed to providing networks with information on election night. I have a feeling that section won't go over well here so I won't paste it, but it's a quick scroll and description, for anyone interested.

http://www.pollster.com/guest_pollsters_corner/joe_lenski_interview_part_2.php

"Well a lot of it has already been done. We sat down with all of the NEP members after that report came out and did a thorough review of all the recruiting and training procedures for the exit poll interviewers and we got a lot of input from all of the professionals that work at the news organizations that do their own surveys. They looked at the materials we were using. We had discussions and we came up with an improved training manual. We also prepared and filmed a training video that all interviewers are required to watch. We developed a new more rigorous training script and a quiz or evaluation at the end of that script to make sure that the interviewers understand the important facets of their job.

In addition to all that we have the input rehearsals that we have done every year, where we have two days in which we act like its Election Day. People call in with test results using the same phone numbers, the same questionnaires they will use on Election Day just to make sure they understand how the process works. So all of that has already gone into effect and I think our interviewers are much better trained this year than they were in 2004."

Another factor that came into play is that we found the error rates tended to be higher on average in precincts where there were younger interviewers, especially interviewers under the age of 25. This isn't to malign the abilities of interviewers under the age of 25, there just seems to be an interaction between older voters and younger interviewers that make older voters less likely to fill out surveys that are presented to them by younger interviewers, so we've also made a concerted effort to increase the average age of our interviewers.

[Febble]

15. Well, it might be worth pointing out

that in their investigation of 2004, one of the questions asked of the interviewers was whether they felt "well-trained". Rick clearly did, which is excellent. Some clearly didn't. Whether those who didn't were actually not as well trained is a different issue of course. My impression is that Rick both felt well-trained and was well-trained.

But there is actual evidence for variance in the degree to which interviewers felt well-trained. And some didn't. And Rick himself should know, as a trained interviewer, that you cannot generalise from an N of 1!

[democraticinsurgent]

17. Some good points here

I will compare the manuals, I think have 2004 yet. I think there was a video in 04 too, but that was awhile ago.

The rehearsal calls, though, were virtually the same as 2004. So the live one-on-one training seemed identical to me.

[Febble]

16. Mitofsky did not "blame"

anyone. He took great pains not to "blame" anyone. What he did was note clear associations in the data between the magnitude of the discrepancy and factors present at each precinct. By far the clearest finding was that where the interviewing interval was large, bias tended to be greater. I do not know how the interviewing intervals are being allocated this year. But reducing it would certainly be a good move.

[democraticinsurgent]

18. Interval rate may have changed

I asked but no one would tell me. In 2004, I interviewed every 9th person. This year, EVERY OTHER. That is a significant shift and well worth doing.

Also, the questionnaire was far more detailed AND there were two versions so that more data could be collected.

I did not mean to imply that everything was exactly the same. I'm glad to see some improvements from the consumer's end and I hope that all of the interviewers felt better trained.

[Febble]

19. Well, it won't necessarily

eliminate the problem if there is an underlying difference in willingness to participate (as in fact was indicated in a recent poll). So it will be interesting to see the magnitude of the adjustment. Mark Lindeman grabbed the close-of-poll crosstabs for each state, and I think is in the process of figuring out the size of the adjustments.

Were you in the same precinct as last time?