Let me see if I have this straight. The exit polls are adjusted

[truckin]

throughout election day to reflect the actual results so they are not a good tool to indicate possible vote discrepencies. The final polling data is a much better indicator of possible problems since these numbers do not get adjusted. There were big differences between the final polling numbers and the actual results for Clinton (see link below) but none of the other Democratic candidates. There are explanations for these differences other than problems with the scanners but these facts would indicate that the recount that will take place is a good idea.

Is this accurate? Hillary supporters, please do not attack, I'm not accusing her of anything, just trying to present the facts as I understand them. If someone did tamper with the machines I would be very surprised if Hillary even knew about it.

http://matthewyglesias.theatlantic.com/archives/2008/01/how_wrong_were_the_polls.php

[midnight]

1. Seems very strange indeed.

[sunonmars]

2. I think it means

they constantly adjust information as they get it during the day to explain voting trends in areas.

from what i saw the final exit polls really put them about neck and neck with a variance of about 3%, so i think in reality, they didnt do that bad. Exit polls can never be 100% due to the fact that some people do lie their asses off the the exit pollsters, i also think a lot of people lied about voting intentions in the last few days up to the vote basically because they were so fecked off with pollsters.

Its not conceivable that Obama had no real bounce out of Iowa other than to where he ended up and they vastly under estimated Clintons ability to get the vote out and put her where she was really going into NH.

That 10% that they had him over Clinton never sat well with me, i always thought it as far too much.

TBH I think the msm had set Obama and Clinton both up for a fall. Clinton before Iowa and Obama after it. They put the expectation bar too high and it saw both of them get whalloped.

[truckin]

4. Do they adjust the exit polls as they get more exit poll data or

do they adjust it to the actual results? I do not know the answer to that; does anyone here?

[OnTheOtherHand]

3. lessee

The exit poll tabulations initially are weighted to a mix of the interview data and pre-election data, and subsequently are weighted to vote counts. (Of course they can't be weighted to vote counts before vote counts are available.) Also, the exit polls are notoriously inaccurate in New Hampshire. So, they are not a good tool.

It's true that Obama's vote share was close to his poll average and Clinton's was not -- as if Clinton won the entire undecided vote. (That's a simplification: in real life, "undecided" isn't an identity, it's just a response someone might give on a survey. And of course people don't even know for sure whether they will vote.) It was impossible for every vote share to match every poll average, because it's impossible to vote for undecided.

Some people regard the results as suspicious given the pre-election polls. Others (including me) do not.

I think people who advocate a recount should do it based on general principle -- 'why not check the scanners?' -- and say as little as possible about the suspiciousness of the results (unless they can come up with some much stronger suspicions, at least). That maximizes the benefits of a recount and minimizes the risks.

[truckin]

5. Thank you for that input. Do the networks adjust the exit polls to

give them the best available tool to predict the outcome and to look at trends in sub-groups? If that is the case the exit polls are of no value to indicate possible problems with the vote count. If the unadjusted exit poll data was released, would that give indications of vote counting problems? This information should be made public, although since the networks pay for the information they have no legal obligation to release it.

Final polling data that is way off from the actual results would indicate possible problems and that is the situation in NH. While there are other explanations other than problems with the count, it does raise a red flag. It will be interesting to see how the recount proceeds.

[OnTheOtherHand]

8. a few answers

As for predicting the outcome, the exit poll data isn't so much "adjusted" as it is replaced with vote counts as those arrive. What we mostly are focused on is the tabulations, and yes, those are updated in order to look at 'trends' in subgroups.

Unadjusted exit poll data might give indications of vote counting problems. Febble probably knows more than anyone in the world about that subject (and I'm sorry to say that some people here dislike her for it). In the 2004 general election, the balance of evidence indicates that the NH exit poll was just way off. It's hard to learn much from exit poll data, especially within a state, because the samples are small and the data are noisy.

The exit polls are routinely archived* some time later, but without precinct identifiers. The questionnaires ask for a lot of demographic information, which could compromise the confidentiality of the respondents. That's a big ethical (and potentially legal) issue for the researchers. The Election Science Institute arranged to inspect some 'blurred' data from Ohio 2004 to see what they could find, which was basically nothing. That doesn't mean that Ohio was copacetic, just that the exit poll data didn't provide much insight into what went wrong. (ESI's analysis does tend to indicate that the vote count in Ohio may have been 'fairly' accurate.)

Yes, it's always interesting to see how manual counts of op-scan results turn out. I'd like to make it routine.

*ETA: The data are archived with an outfit called ICPSR which is basically a research consortium. The 2004 data were also made publicly available for about a year.

[truckin]

9. Thanks again. You're right, the audits with sample hand counts

should be automatic. We shouldn't need any evidence to double check the machines.

[HughMoran]

12. Good explanation

People forget that an exit poll can't possibly get a representative distribution of the demographic of the population of the state. Also, there is bias based on those who would volunteer to answer the exit poll - for instance, when polled on Tuesday, I did not want to admit who I voted for (for various reasons.) The results must be weighed against existing data available at the time. As you stated, before the actual votes come in, the exit polls are weighed against the existing polling data to adjust between demographic differences of the exit polls and the actual demographic of the state and the existing voting trends of each demographic. Once the actual results come in, the exit polls then have to be adjusted to consider the demographic of who actually voted & how each demographic voted. When things are in a state of flux on election day (as they were on Tuesday), the polling data tends to be worth very little. Once the exit polls were adjusted to the actual demographic voting trends (for instance that woman voted 10% higher for Hillary than expected), then the exit polls were nearly identical to the actual voting totals (39-37 versus 39-36.)

[dailykoff]

6. Yes, as of 2004, the exit polling consortium adjusts their polls

to reflect actual vote tallies as they come in. So if the machines are rigged and inaccurately reflect actual voting, such discrepancies will not show up in the adjusted exit polls released to networks once the tallies start rolling in.

This is according to a late 2004 interview with Warren Mitofsky I saw on the PBS Newshour.

[truckin]

7. Thanks. It would be nice if the networks released the unadjusted

exit poll results to the public.

[HughMoran]

14. @ 8PM the exit polls showed Obama up by 1%

How much could those polls have been adjusted since results had only started rolling in 15 minutes before and were no more than 7-9% of the total and mostly the smaller towns whose polls closed before 8PM?

Consider that many of the smaller towns which closed & reported earlier favored Obama as compared with the large towns that closed @ 8PM which favored Hillary. Could the 8PM sample have been that skewed? I tend to think not.

[Bitwit1234]

10. If polls are adjusted to SHOW results how come

John Kerry was ahead in every exit poll in 2004 and lost. It seems to me those exit polls were not adjusted. Is this a new thing?

[OnTheOtherHand]

11. the tabs were updated after midnight

It's not new at all, except that back in the day people weren't sitting around staring at exit poll tabs, because they weren't posted on the Internet.

[Peace Patriot]

13. No, I won't blame Hillary, but it just seems like, when people cozy up to the Bush

Cartel, they get this sticky green slime all over them, that won't wash off, and after a while it really starts to smell.

It's not Hillary's fault, exactly. Bushes and stolen elections just seem to go together.

But I'll tell you something interesting. When the $1.3 billion e-voting boondoggle was before the Anthrax Congress, in Oct 02 (same month as the IWR) (heh, heh), there were ONLY TWO votes against it in the U.S. Senate. Hillary Clinton and Charles Schumer.

Now, I've thought a lot of about this. Why would pro-corporate Hillary have any problem with rightwing Bushite corporations "counting" all our votes in machines run on 'TRADE SECRET, 'PROPRIETARY programming code, with virtually no audit/recount controls? None of the other Democrats in the Senate cared a goddamn. Why should she?

At first, I thought it was so as not offend NY voters, who are very attached to their old, reliable and virtually unriggable lever voting machines. The thing (fascist coup) was going to pass. It didn't need the NY Senators' votes. They could safely vote against it (even if they were for it). And perhaps it meant that they had at least that much regard for public opinion and democracy--to respect NY voters' wishes. And it gave me a twinge of hope that, if Hillary gets Diebolded into the White House, once she is in power, she won't object to us restoring transparent vote counting in the rest of the country for future elections.

But then I began to think that the purpose of her and Schumer--and her and Schumer alone--voting against this egregiously non-transparent election system for the rest of the country, was to put the matter off the radar of NY voters, who live at the center of media/PR power in the country, and might cry foul, say, to the NY Times, and bring pressure for some coverage of the subject, some warning to the rest of us, if THEIR voting system were fucked over the way the rest of ours has been. I favor this explanation today. The Bushite federal election commission also went easy on NY at first, and let them get away with non-compliance (until recently--now they're suing NY). Clinton and Schumer are as collusive with fascist/corporate power as the rest of the Democratic Party establishment, if not more so. It's easy to believe that they colluded on getting this last corporate control over our democracy in place, and on keeping NY voters away from the issue, while it got entrenched all over the country. And now they'll do NY and the coup will be complete.

But take your pick of explanations, or think up your own. It IS weird fact. Of all the Democrats in the Anthrax Congress, ONLY THESE TWO voted against the "Help America Vote Act." Maybe, to you, it means they still believe in democracy. And maybe it means that Hillary will be the first to object if there was a mess-up in her favor in New Hampshire. (Well, not the first--Kucinich was the first--but you know what I mean.) It will give all her detractors the lie. And I'm not being 100% sarcastic. Who knows, these days, what's really going on--with all the media delusion and distraction? It's hard to get a real read on any politician, in this putrid atmosphere. Maybe the rightwing vileness toward her is truly meaningful, and she will slay dragons for us. Maybe she will come out and blast private, secret, corporate vote counting, and disavow any false win, and say she only wants to win on the basis of transparent, honest, aboveboard counting of every vote.

I won't rule it out. I can't. And we really can't make political judgements on the smell of green slime, can we? On nebulous feelings? On media-spun illusions? On electrons whirling around in "black boxes" and coming out with a "winner"? We need to be more hard-headed and not quite so prone to feelings of hysteria induced by deliberately-created, Rumsfeldian-type uncertainties and chaos and disorder. So, let's see what NH election officials actually do. Let's see if they do a real recount with no shenanigans. And let's see how Hillary behaves in this circumstance.

The election is just under a year away. That's a lot of time. Maybe something real will happen, to shed some light on our political system and our future. And, being a progressive, I believe that people can change. I really do. Maybe not Bush, Cheney, Rumsfeld and that lot. (Let's be realistic.) But most others can change. Can learn. Can grow. I think John Edwards has. I think Bobby Kennedy did. And JFK did. Maybe Hillary can, too.

[dailykoff]

15. Plausible deniability would be my pick.

If she knew it would pass, she knew she could safely vote against it and provide cover for any future vote rigging on her own behalf. I've seen politicians do this many times on other issues.

But didn't she co-sponsor this bill? Or was it a later one?