TIA and others: How many more votes did GWB get this time than last?

[davidgmills]

What percentage increase is this?

What did the exit polls predict about this huge percentage increase?

Isn't this supposed to be one of those demographics that was carefully exit polled?

Isn't his one of those demographics for which the exit poll was "designed"?

If so, where in the exit polls can the huge increase in these percentage numbers be explained?

Also, what was the margin of error for this demographic spinning it as best one could to favor Bush? At most 3%?

What is the difference between percentage of predicted increase vs. "actuaL" increase? Outside MOE? If so, by how much?

[The Backlash Cometh]

1. That's a good question.

Someone was holding a sign up at his inauguration that said, "50.8% does not make a mandate" and I wondered if the votes came down drastically, because he was originally claiming a 3% spread.

[Wilms]

2. This may help.

<http://uselectionatlas.org/USPRESIDENT/index.html >

[TruthIsAll]

3. According to the pristine 13,047 NEP....

How many more votes did GWB get this time than last?
What percentage increase is this?

Bush received 11.573 million more votes than in 2000, an increase of 22.94%.

........2000.... .2004....... %...... Change
Bush 50.455 62.028 50.73% 11.573
Dem 50.992 59.027 48.28% 8.035
Nader 2.882 1.212 0.99% -1.67

Total 104.32 22.267 100.0% 17.938

...............................................
What did the exit polls predict about this huge percentage increase?
ACCORDING TO THE NEP OF 13047 VOTERS WHO DID NOT VOTE IN 2000:

1) 20.785 MILLION (17%) did not vote in the 2000 election.
KERRY won 11.73 million (57%) of these voters
BUSH won 8.44 million (41%)
NADER et al won 2%


2) 13.449 MILLION (11%) WERE FIRST-TIME VOTERS.

KERRY won 7.4 million (55%
BUSH won 5.78 million (43%)
NADER et al won 1%


...............................................
Isn't this supposed to be one of those demographics that was carefully exit polled?
Yes
...............................................
Isn't this one of those demographics for which the exit poll was "designed"?
Yes
...............................................
If so, where in the exit polls can the huge increase in these percentage numbers be explained?
It cannot be explained.
...............................................

Also, what was the margin of error for this demographic spinning it as best one could to favor Bush? At most 3%?

No, the MOE was 1.0%, according to Edison/Mitofsky’s own notes at the bottom of the WP/NEP site.
...............................................

What is the difference between percentage of predicted increase vs. "actuaL" increase? Outside MOE? If so, by how much?

According to the National Exit Poll of 13,047 respondents:
1) Bush received 8.44 million votes from those who did not vote in 2000. His final vote total showed an increase of 11.57 million, or 3.1 million ABOVE his expected total. The percentage change was 26.4%

2) Kerry received 11.73 million votes from those who did not vote in 2000. His final vote total increased by 8.04 million votes over Gore, or 3.7 million BELOW his expected total. The percentage change was -47.5%


Based on the weighted exit poll percentages of those who DID NOT VOTE IN 2000:

Kerry won the election by 50.99% to 48.22%

The probability of Bush deviating from 48.22% to 50.73% = 0.0000004342

Odds of this occurrence: 1 in 2,302,876

...............................................

[Mistwell]

4. I disagree

"Isn't this supposed to be one of those demographics that was carefully exit polled?
Yes
...............................................
Isn't this one of those demographics for which the exit poll was "designed"?
Yes
...............................................
If so, where in the exit polls can the huge increase in these percentage numbers be explained?
It cannot be explained."

I'm sorry TIA, but you are incorrect. "Demographics data" is data OBSERVED, and not the data obtained by the exit poll forms themsevles. This is why it is the first to be tabulated...because you do not need to enter the data from the exit poll forms (the questionaires), which takes a lot more time to do. So it includes gender, approximate age, and race. It does not include ANY data about how people voted, why they voted, or what party they belong to. THAT is what the MOE in those early polls that you like is tied to.

[RaulVB]

5. Your argument is circular and does not address the main point

Explain the 23% increase of "*" "votes" what DID NOT TRANSLATE IN AN ELECTORAL LANDSLIDE.

Thanks

[Mistwell]

11. That's easy

Since Kerry ALSO had a massive increase in voters from Gore voters...just not quite as massive as Bush's increase.

Hey, my next comment really is not meant in an insulting manner at all. I'm just asking out of pure curiosity: Is english your first language?

[RaulVB]

14. "not really meant in an insulting manner at all..."

I don't really care about your opinions in any topic so, let's not play "nice" here. Is english your first language? (just curious)

Now, to the issues, your bogus "explanation" about "*" "totals" based on KERRY'S TOTALS is rather amusing. That's all you can offer?

Thanks for your highly entertaining...comments?

[ROH]

23. Glass houses?

You wrote: "Is english your first language?"

Shouldn't that be: "Is English your first language?"

[Karenca]

6. I think you're just pulling our leg, aren't you? ;))

Seriously now, what are your qualifications to deem TIA's reports incorrect.

This has just been released----http://uscountvotes.org/ucvAnalysis/US/USCountVotes_Re_...
Contributors and Supporters include:

Josh Mitteldorf, Ph.D. - Temple University Statistics Department
Kathy Dopp, MS in mathematics - USCountVotes, President
Steven F. Freeman, PhD - Center for Organizational Dynamics, University of Pennsylvania
Brian Joiner, PhD - Prof. of Statistics and Director of Statistical Consulting (ret), University of Wisconsin
Frank Stenger, PhD in mathematics - School of Computing, University of Utah
Richard G. Sheehan, PhD - Department of Finance, University of Notre Dame
Elizabeth Liddle, MA - (UK) PhD candidate at the University of Nottingham Paul F. Velleman, Ph.D. - Department of Statistical
Sciences, Cornell University
Victoria Lovegren, Ph.D. - Department of Mathematics, Case Western Reserve University
Campbell B. Read, Ph.D., Professor Emeritus, Department of Statistical Science, Southern Methodist University

Also Peer Reviewed by USCountVotes’ core group of statisticians and independent reviewers.

This fine work has been prepared by NINE PHD'S, not to mention it was reviewed by USCountVotes’ core group of statisticians and independent reviewers.

So, I'm sorry, unless you are God and you know all, your assumptions a hysterically funny

[davidgmills]

9. Great article -- hadn't seen this one yet!

Doesn't this just about say it all:

The Edison/Mitofsky report fails to substantiate their hypothesis that the difference between their exit polls and official election results should be explained by problems with the exit polls. They assert without supporting evidence that (p. 4), “Kerry voters were more likely to participate in the exit polls than Bush voters.” In fact, data included within the report suggest that the opposite might be true.

Their analysis of the potential correlation of exit poll errors with voting machine type is incomplete and inadequate, and their report ignores the alternative hypothesis that the official election results could have been corrupted.

[bleever]

18. Thanks for your great question here, dgm. And isn't this article the

bunker-buster we've been waiting for?

I think the truth just got a huge nudge here.

[Mistwell]

12. It's all based on an assumption that the MOE is correct

All of those contributing to that analysis go in assuming the MOE is correct. Mitofsky, the guy who set the MOE in the first place, says it is not. The studies from Duke say it is not. The studies from Cal Tech & MIT say it is not. And a direct interview between Freeman et. al. found that there was a serious flaw in one of the basic assumptions.

I just give you the links and the summary of their claims. I am not the guy doing the statistical analysis. So please, don't attack the messenger. i don't have to be a diety to tell you that there is a reasonable disagreement on this issue that you should look into further.

[TruthIsAll]

26. Let's make the list easier to read.

1 Josh Mitteldorf, Ph.D. - Temple University Statistics Department
2 Kathy Dopp, MS in mathematics - USCountVotes, President
3 Steven F. Freeman, PhD - Center for Organizational Dynamics, University of Pennsylvania
4 Brian Joiner, PhD - Prof. of Statistics and Director of Statistical Consulting (ret), University of Wisconsin
5 Frank Stenger, PhD in mathematics - School of Computing, University of Utah
6 Richard G. Sheehan, PhD - Department of Finance, University of Notre Dame
7 Elizabeth Liddle, MA - (UK) PhD candidate at the University of Nottingham
8 Paul F. Velleman, Ph.D. - Department of Statistical Sciences, Cornell University
9 Victoria Lovegren, Ph.D. - Department of Mathematics, Case Western Reserve University
10 Campbell B. Read, Ph.D., Professor Emeritus, Department of Statistical Science, Southern Methodist University

[indigoblue]

7. Do you know what you are talking about?

"I'm sorry TIA, but you are incorrect. "Demographics data" is data OBSERVED, and not the data obtained by the exit poll forms themsevles. This is why it is the first to be tabulated...because you do not need to enter the data from the exit poll forms (the questionaires), which takes a lot more time to do. So it includes gender, approximate age, and race. It does not include ANY data about how people voted, why they voted, or what party they belong to. THAT is what the MOE in those early polls that you like is tied to."

Are you saying that these exit poll results are based on what pollsters guessed on gender, age, race, and religion, etc.? I need to see the reference if you have one.

[Karenca]

8. You are too funny!! :))

read post #7.

I mean post #6,,
KERRY
KERRY
KERRY
KERRY
KERRY

[Mistwell]

13. Yes

I am saying that the EARLY exit polls, the ones that were not supposed to be released, were fixed for DEMOGRAPHIC DATA ONLY (that's gender, race, age - NOT voting). Mitofsky put it IN those early poll releases, but TIA has ignored it and prefers a screen shot that removed that disclaimer. TIA admits that it WAS fixed to demographics data, he just disputes what is included in "demographics", and believes you can infer how people voted from the demographics data if you include some of the "how you voted" data as well. Both TIA and I agree, however, that the non-demographics data was not expressly tied to the MOE he is using.

[TruthIsAll]

24. I do not agree with ANYTHING you say. Are you referring to this?

I said in another thread:
As any naysayer will tell you, exit polls are not designed to predict the vote. They are only used to show how various demographic groups voted, not how people voted.
Like men and women.

And you call this agreeing with you?
Why is it you guys are always saying I agree with you, when I never have.

Oh, I forgot. This one was true:
TIA is not an expert. He said so himself.
Just because I am not an expert does not mean that I an not right.

I can think of at least one polling "expert" who has been proven wrong time and time again.

Mistwell, not only do you torture the facts, you have no sense of irony. Let me explain.

Gender is, in Mitofsky's words, a "characteristic", a very strong one. Men and women total 100% of the population and are nearly equal in voting power.

But not quite.

You see:
The majority of voters (54%) were WOMEN.
And 54% of the WOMEN voted for Kerry.

Just 46% of the voters were MEN.
And only 52% of the MEN voted for Bush.

Mitofsky claims that any characteristic which occurs in 50% of the exit poll sample, has a 1.0% MOE.
Like GENDER.

So....taking him at his word,
Kerry = .54 * .54 + .46 * .47
Bush = .54 * .46 + .46 * .52

Do the math.
Logical, right?

Q.E.D.

[Mistwell]

28. You have me confused with someone else

I never said a word about your expertise on this matter. You must have me confused with someone else...again.

I said you and I agree that the MOE is concerning demographic data, but disagree about what "demographics" are included.

Are you saying I am wrong in that?

Mr. Mitofsky says that his count of the number of women who voted is correct...BUT NOT HOW THEY VOTED. ALL DATA ON HOW PEOPLE VOTED IS *NOT* INCLUDED IN "DEMOGRAPHICS".

It is not logical for you to keep pretending that you don't get this. You have intentionally misconstrued a single out of context sentence from Mitofsky to fit your own blatantly subjective agenda. You know damn well that how people voted cannot be obtained from the observers, but just from the poll questioners, and the poll questioners have a totally different MOE than the poll observers. Unless you think Mitofsky was able to hire pollserts who could TELL FROM SIGHT how people voted without even asking them...in which case I think even you would think the MOE was in error.

Do you really think you are persuading people with this game you are playing, misrepresenting the margin of error to force the numbers to fit your preconception that the poll proves fraud?

TIA, why don't you show me specifically where Mitofsky defines "demographics data" as including ANY element of how people voted...beyond your warping of his general statement to fit your personal definition of what the word "demographics" means. The methodology is published....show me where he says it.

[davidgmills]

10. I've known for weeks now the MOE was 1.0%

But I guess my question is whether it is arguably possible for anyone of integrity to spin it up to 3 and how it might be done.

Please note the word integrity.

[Mistwell]

15. Your assuming the MOE was tied to how people voted

However, the early exit polls EXPRESSLY state it is for Demographics data only. You only think it is 1% for all the data, and not just the demographics data, because TIA always posts the screen shot that left that disclaimer off the poll (for space reasons, I assume). It was and still is always on Mitofsky's site.

You can read a detailed analysis of how the MOE changes when you pin it to the "how people voted" data, and it's a pretty drastic change. See, for example:

http://www.mysterypollster.com/main/2004/11/the_freeman...

which provides some background to the Freeman debatem and the follow-up (which talks more about the MOE):

http://www.mysterypollster.com/main/2004/12/what_is_the...

[davidgmills]

16. Please find another source

I stopped reading Mystery Pollster some time ago. I read those when they came out.

[Mistwell]

17. They source other folks

You can bury your head in the sand if you want, but MP is sourcing other (named and credentialed) people in their analysis. Why don't you read it, and then judge it, rather than the other way around?

[davidgmills]

20. Always the polls are wrong; never the count

And Mystery Pollster's PhD is in what? Websiting, maybe.

I really want to know what credentialed academics think.

Someone who will put his tenure on the line.

[Mistwell]

27. It IS highly credentialed people: Look

The studies include the combined CalTech/MIT voting project (all highly credentialed):

http://www.scoop.co.nz/stories/pdfs/VotingMachines3.pdf

And the Duke University study (which is written by someone highly credentialed, and sources dozens of highly credentialed folks):

http://www.duke.edu/~mms16/non_response2000.pdf

If you REALLY want to know what credentialed academics think, go ahead and read the links.

And, as for the Mystery Pollster...there is no real mystery. Mr. Blumenthal is a well know DEMOCRATIC pollster. Here is his bio:

About Mystery Pollster

Mark Blumenthal

Phone: [phone]

Biography

I have been in the political polling business for more than 18 years, conducting and analyzing political polls and focus groups for Democratic candidates and market research surveys for major corporations. My experience includes work with pollsters Harrison Hickman, Paul Maslin, Kirk Brown, Celinda Lake, Stan Greenberg and the last 12 with my partners David Petts and Anna Bennett in the firm Bennett, Petts and Blumenthal.

My academic background includes a Political Science degree from the University of Michigan and course work towards a Masters degree at the Joint Program in Survey Methodology (JPSM) at the University of Maryland. I have also served as a guest lecturer at the Communications School at American University and at training seminars sponsored by EMILY’s List, the Democratic National Committee and the National Democratic Institute for International Affairs.

Disclosure: I am a Democratic pollster, but my firm is not conducting surveys for either the Kerry campaign, the DNC nor any of the independent expenditure campaigns working on his behalf. On this blog, I will refrain from commenting on polling in any race in which we are also polling. The views expressed on this blog are mine alone and may not reflect those of my business partners or our clients.

Why "Mystery Pollster?" That name was the anonymous alias blogger Mickey Kaus gave me when he printed excerpts from emails I sent him. Kaus always knew my name; he just chose to keep it off the record. When I decided to start blogging it seemed natural to keep the name. I liked the ring of it, especially the play on "demystifying,” and the subtle homage to Kaus, who kindly alerted the blogosphere to the existence of this blog.

[davidgmills]

31. Is this the best you can do?

The Cal/Tech article was quickly debunked by Freeman and others for using the corrupted data of actual votes. Egregious mistake and they have been silent since Nov. 9.

The Duke study involved Mormons in Utah. Mormons are not representative of the rest of the white population. If Mormonism survives long enough (it seems to be growing) it will be another offshoot of Christianity just like Islam and it's major text will become the Book of Mormon not the Bible. I think any conclusions from this study would be seriously flawed for that reason alone.

Mystery Pollster ... working toward a master's degree. Give me a break. Give me someone who has demonstrated that he has completed something of serious academic rigor. Please.

[Mistwell]

32. Freman himself was debunked

Freeman himself is very discredited in my eyes, given it was proven, TWICE, that he used faulty data and was ingnoring anything that did not further his contention that there was fraud (and intentionally exagerated every number he could as well).

The Duke study was about why people might not take an exit poll. It's independant, and not even about this particular election. Your response that it involved too many mormons is simply illogical and, quite frankly, silly. It was perhaps the most pathetic defensive response I've seen at DU *EVER*, and that includes TIA's claim that ALL elections since the invention of the exit polls were fixed.

If you don't care to get to the truth, then just say so. But holding your hands to your ears and chanting "I'm Not Listening" is juvenile. The exit poll process is flawed - you know that. It doesn't EVER produce reliable results - and you know that as well. If you want to pretend that it works, despite all the evidence to the contrary, then fair enough. Your closed mindedness must serve you well in some circumstances, lets hope it works for this one as well.

[davidgmills]

38. You are in such a minority here

Why should I listen to you.

I also know something about Mormons; maybe you do not.

I am simply saying that I would not use a study involving a high percentage of Mormons to represent the average Caucasian American.

I admire their work ethic. But their views are not anything like mainstream on many subjects. Maybe they do not know that at Duke. NC is a pretty good distance from Utah.

[TruthIsAll]

39. MP: Just course work toward an M.S.? There it is.

And MP is the best they can do?
OK, he has a BS in Political Science.

But what is his mathematical background?
How is he going to argue the mathematics with all of Freeman's Ph D's?

I remember that MP asked for help when I first used the Normal and Binomial distributions to calculate the probabilities.
That told me something.

And then he wrote about 99.5% confidence levels to call elections, which were totally irrelevant to the discussion of exit poll deviations.
That told me something else.

Then there was the exit poll MOE brouhaha.
First he claimed a cluster factor of 30%. He raised it to 50% or more. Now that Mitofsky said the National Exit Poll MOE was 1.0%, he doesn't broach the issue.
That told me something, also.

Now, he is getting an education from mathematical posters on his site which debunk the Reluctant Republican Responder Theory.
That told me something.

But that was BEFORE the latest Freeman paper which was peer-reviewed and supported by 10 Ph D's. An analysis of Mitofsky's own numbers showed the OPPOSITE- Republican Responders actually outnumbered the Democrats.

How is he going to spin that one?

No offense, but I believe MP is over-matched.
Everyone keeps saying he's a democrat.
As if that justifies anything he says.

MP is up against big-league pitching now.
It's as though he is batting against Sandy Koufax.

[darthdemocrat]

29. mistwell - you made me curious

I read the mystery pollster's discussion of MOE. His discussion of Freeman is strongly based on information from Mitofsky. (Actually secondhand information provided by an intermediary.)

I'm not placing my faith in explanations from the guy given after the fact. Mitofsky has not been a forthcoming or transparent pollster. How long did it take after the election for these people to cleanly admit their numbers were recalibrated with the official tallies? How evasive has Mitofsky been about the early numbers he provided (trying to confuse leaked numbers with legitimate results in the media, etc.) ?

If you believe Mitofsky, the error was accountable to Kerry voters being more likely to respond. He provided no evidence to back up that opinion, only evidence of the effect it would have if true. Of course, response rate was actually highest in Republican strongholds... just ask Mitofsky. He left that relevant data out and provided an opinion based on no data instead. Greeeeeeat!

"Because Mitofsky says so" isn't good enough. The best he has to offer is that a still fairly amazing improbability is what happened. With neither an honest system (whether you look at disenfranchisement or cheat sheets to avoid actual recounts) nor a transparent pollster no one can say who actually won the election. The exit polls by themselves may not be all that convincing. The big picture? Just look.

[Mistwell]

33. The Duke Study...

...was done before this election, and not based on Mitofsky. It shows why Democrats tend to answer exit polls much more often than Republicans.

You can't blame all this on Mitofsky. There is real, legit data showing the exit poll system is broken.

[davidgmills]

37. Legit for who -- Mormons that's who

[bleever]

19. After you're done with Mystery Pollster, I have a game I'd like to

play with you, maybe for a little money just to make it interesting.

[davidgmills]

21. And that game would be?

[bleever]

22. I'll start with 16 coins, laid out in four rows...

MP and others think they can rig the game by always playing this "on the other hand" type of game, which of course isn't science but politics, so it's a bait and switch. They pretend it's science, but their motives are political, so engaging in the conversation (past a certain point, which you already passed), is, as you know, pointless and a distraction.

And I'm going to get the rules to that game of yours figured out any day now.

Thanks for your posts.

[davidgmills]

30. Are you still trying to figure that game out?

One guy did. Proved it to me. Played me one night. Took 50 minutes on the net to play a three minute game. He wanted to play best of seven but at that rate ...

Anyway, I posted instructions on how to play the game. If you want to know how, just ask.

[bleever]

36. Thanks, but I have to figure it out myself. Just haven't applied my whole

brain for a sufficient uninterrupted span.

[Igel]

25. I've always wondered how you can have a 1.0 MOE

on some of those numbers when 47% of the voters weren't actually asked anything.

[Mistwell]

34. The answer...

...is what Mitofskey said before he even released the first exit poll...the MOE of 1% is for demographics data only. It tells you what the apparent gender and race and broad age range is of the people exiting the polls. That's it. There is a different MOE for how people voted...which made it into some newspapers and TV shows and not into others (and, I think not coincidentally, it was left off of the screen shot that TIA always posts).

[Igel]

35. I was ragging on TIA. "Some of" the #s.

Observable stuff, agreed.