Prominent Statisticians Refute 2004 Exit Poll "Explanation

[polyb]

BACK FROM THE DEAD: Prominent Statisticians Refute 'Explanation' of 2004 Exit Poll

Uh oh, seems this one is not quite dead yet! Apparently some of those 'crazy liberal elitist' academicians have a problem with the discrepency during our last vote. Gee, why am I not surpised!

http://www.emediawire.com/releases/2005/1/emw203331.htm

(PRWEB) January 31, 2005 -- "There are statistical indications that a systematic, nationwide shift of 5.5% of the vote may have occurred, and that we'll never get to the bottom of this, unless we gather the data we need for mathematical analysis and open, robust scientific debate.", says Bruce O'Dell, USCountVotes' Vice President.


The study, “Response to Edison/Mitofsky Election System 2004 Report”, was co-authored by a diverse group of academicians specializing in statistics and mathematics affiliated with University of Notre Dame, University of Pennsylvania, University of Utah, Cornell University, University of Wisconsin, Southern Methodist University, Case Western Reserve University and Temple University. Their study does not support claims made by Edison Media Research and Mitofsky International that exit poll errors were to blame for the unprecedented 5.5% discrepancy between exit polls and official 2004 election results.

Here is a link to the report:
http://uscountvotes.org/ucvAnalysis/US/USCountVotes_Re_Mitofsky-Edison.pdf

BTW, have a look at some of the credentials from the list of participants:

Josh Mitteldorf, Ph.D. - Temple University Statistics Department
Kathy Dopp, MS in mathematics - USCountVotes, President
Steven F. Freeman, PhD - Visiting Scholar & Affiliated Faculty, 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 - Professor of Numerical Analysis, School of Computing, University of Utah
Richard G. Sheehan, PhD -Professor, Department of Finance, University of Notre Dame
Elizabeth Liddle, MA - (UK) PhD candidate at the University of Nottingham
Paul F. Velleman, Ph.D. - Associate Prof., Department of Statistical Sciences, Cornell University
Victoria Lovegren, Ph.D. - Lecturer, Department of Mathematics, Case Western Reserve University
Campbell B. Read, Ph.D., Professor Emeritus, Department of Statistical Science, Southern Methodist University

Any thoughts?

 

[GENIERE]

A little googling determines several ( I only did 4 before I got bored) of the participants made leftist biased statements months, sometimes years before the study. Obviously the study was a designed to support their bias.

 

[polyb]

A little googling determines several ( I only did 4 before I got bored) of the participants made leftist biased statements months, sometimes years before the study. Obviously the study was a designed to support their bias.

When you get out of lineland maybe you'll see there is more at stake than 'liberal bias'. oh yeah. HINT: make an orthogonal turn to whatever directions you are capable of moving.

 

[wasteofo2]

A little googling determines several ( I only did 4 before I got bored) of the participants made leftist biased statements months, sometimes years before the study. Obviously the study was a designed to support their bias.

So bassically, any Liberal who raises any concern about anything can be discounted, simply because they're left-leaning? Are Liberal Geologists to be dismissed when they tell us that Global Warming is real, and happening, because they are Liberal?

 

[GENIERE]

US university faculties are biased to the left. The less technical the discipline, the more likely the instructor is to be left leaning. I’m going from memory now, but one study indicated less then 10% of poly-sci profs deemed themselves right leaning while those teaching statistics range 30-35%. Those who accept a political statistical study without doing a bio of the authors might as well read the Inquirer. The MIT study completely disemboweled the Berkeley exit poll study. You can satisfy your bias with the Berkeley garbage or educate yourself with the MIT filet mignon.

Global warming is a political issue but fortunately there exists a great many studies wherein the author’s political bias is nowhere evident. I hope Al Gore’s book or the dogma of the Union of Concerned Scientists would not be cited as evidence in these forums. Global warming over the last 150 years is an accepted fact; the warming influence due to human activity is the issue. Is it possible that human endeavor (all countries) can reverse it without draconian measures is the problem. Are you willing to stop using your 300-1000watt PC to help the cause? Did the French stop flying the Concorde because it was bad for the environment or because it blew up?

 

[russ_watters]

So bassically, any Liberal who raises any concern about anything can be discounted, simply because they're left-leaning? Are Liberal Geologists to be dismissed when they tell us that Global Warming is real, and happening, because they are Liberal?

There are a great many scientists who you would be hard-pressed to figure out which direction they lean. Why? Because they aren't activists. The people you should be worried about are the activists because their bias is strongest.

Btw, those people are listed as "contributors and supporters" - it doesn't say anywhere that I can see, who actually wrote that paper.

First impression (perhaps more later), it is more of the same allegations as the last paper, but substantially lower quality and higher bias.

apropos: Why America Went Red

 

[selfAdjoint]

So apparently the answer to this:

The statisticians go on to note that precincts with hand-counted paper ballots showed no statistical discrepancy between the exit polls and the official results, but for other voting technologies, the overall discrepancy was far larger than the polls’ margin of error. The pollsters at Edison/Mitofsky agreed that their 2004 exit polls, for whatever reason, had the poorest accuracy in at least twenty years.

is to go on and on about how some of the authors contributed to liberal causes and in general the old wheeze about how university faculties are not sufficiently welcoming to libertarians and other conservatives and so on ad nauseam.

Is all of this misdirection caused by the fact the right wingers really have no valid response to the situation I quoted?

 

[russ_watters]

Is all of this misdirection caused by the fact the right wingers really have no valid response to the situation I quoted?

Hey, polyb is the one who presented the Argument-From-Authority.

The second part of the statement is the easiest to refute (since it doesn't really say anything to refute) -

The pollsters at Edison/Mitofsky agreed that their 2004 exit polls, for whatever reason, had the poorest accuracy in at least twenty years.

The basic problem with this statement is that it implies the exact opposite of what the paper wants it to imply. Yes, the poll data is the poorest accuracy - that fits with the fact that voting patterns and motivations were significantly different from any election in the past 20 years. Voter turnout was unusually high and this is the first election we've had in the middle of a war since Vietnam. These factors combined with a known democratic bias of the polls themselves (they always overstate the democrat vote) make the discrepancy this time around quite unsurprising.

The first part, I'll get back to you on, but IIRC, the pollsters say specifically that the data does not support that assertion.

 

[polyb]

Hey, polyb is the one who presented the Argument-From-Authority.

No russ, I only said take a look at the credentials because I was impressed. Besides, I don't need an authority to back my opinion, I have already co-authored one paper(hey, that makes me an author-ity!). So to clarify for you, I AM OF THE OPINION THAT THE '04 VOTE WAS RIGGED! I am not alone in this opinion and this is one reason why only about half of the eligible voting public ever votes. They do not feel that the system is set up for them and in many respects they are right!

Here is something that you overlooked from the article:

Their paper titled "Response to Edison/Mitofsky Election System 2004 Report" notes that the Edison/Mitofsky report offers no evidence to support their conclusion that Kerry voters “participated in the exit polls at a higher rate than Bush voters”. In fact, the data provided in the Edison/Mitofsky report suggests that the opposite may have been true: Bush strongholds had slightly higher response rates than Kerry strongholds.

You know russ, it seems odd to me that for someone who hates being lied to seems to have so much 'faith' in this particular administration. Especially considering the sheer magnitude of lies since 9/11. First off, all politicians lie, secondly the whole dichotomy you seem to propose and adhere to is a lie as well(you know, left/right, liberal/conservative, democrat/republican;divide and conquer works in politics too!). My concern is about what little semblence of integrity that exists in the system and when will it be time to teach the scumbags a lesson!

 

[Moonbear]

Okay, when you say the polls had the poorest accuracy in 20 years, you realize that's only 5 presidential elections, right? That's a pretty small sample. Does that mean 24 years ago the accuracy was as bad or worse than this year? Was there any reason to question it then? Out of 5 presidential elections, ONE of them has to have the worst polling accuracy.

You'd also have to look at the voting patterns and timing of the polls in precincts that had apparent discrepancies. Are the Democratic university professors all voting in the morning because they don't have to worry about what time they get to work, while the Republican engineers had to wait until they got done with work at 5 PM to go out and vote, while the pollsters were out between 9 AM and 5 PM? With such a heated and close election, hearing early results and exit poll data on the news may have motivated Bush supporters to finally get off their butts and hit the polls later in the day when they realized it wasn't going to be the clear victory they thought it was going to be.

The thing is, IT'S OVER! Not only have the votes been certified, he's already been inaugurated. Some people need to learn to get a life and move onward. Let's just work on getting some better candidates for 2008! Wouldn't it be nice to have a close election because we like BOTH candidates so much we can't decide rather than a close election because nobody likes either candidate enough to decide?

 

[Bystander]

--- and, the exit polls picked up how much of the early vote, absentee ballots, mail-in ballots? --- and, these were how much of the total vote?

 

[russ_watters]

No russ, I only said take a look at the credentials because I was impressed.

The credentials might be impressive if they had written the paper, but it doesn't say they did. And I say "might" because it would be surprising if they didn't have people with credentials like that read/write the paper.

So to clarify for you, I AM OF THE OPINION THAT THE '04 VOTE WAS RIGGED!

No clarification needed - I knew that.

...this is one reason why only about half of the eligible voting public ever votes. They do not feel that the system is set up for them and in many respects they are right!

People choose not to vote because of apathy, not because they think votes are rigged. No, sorry, you're part of a very small fringe there.

Here is something that you overlooked from the article:

No, I didn't miss that - its another logical fallacy (of a great many) in that article. Lack of evidence for one theory does not constitute evidence for theirs. Sorry, Mr. Watson, you can't use Sherlock's process of elimination here, especially when your hypothesis is an extrordinary claim. edit: and even then, they really did provide evidence that the problem was sysemic, ie. that it wasn't specific to the presidential election, but was actually a data collection problem.

You know russ, it seems odd to me that for someone who hates being lied to seems to have so much 'faith' in this particular administration.

You misunderstand: this administration had nothing to do with the carrying-out of the election.

Ok, I guess I'm going to have to go through this article because though a few people now have picked up on some flaws, there's a lot that is just being taken with a knee-jerk 'hey, that sounds good!' when the reality is that this article was a terribly written propaganda piece. Its not a study.

First off, speaking of being lied to, polyb, where can I find substantiation for this:

The study, “Response to Edison/Mitofsky Election System 2004 Report”, was co-authored by a diverse group of academicians...

The names are listed under "Contributors and Supporters" - you can bet that if they really were "co-authors" it would say so. The way I read that, none of those "academicians" had anything to do with writing the paper.

Ok, now starting from the beginning (this may take multiple posts...):

The first part is background. It starts by listing "irregularities" in the voting process. These problems have been highly publicised and are well-known. The question/implied thesis of the paper and the methodology follow:

The crucial question is whether these problems were part of a larger pattern. Were these issues collectively of sufficient magnitude ot reverse the outcome of the election, or were they isolated incidents, procedurally distubring but of little overall consequence?

Under such circumstances we must rely on indirect evidence, such as exit polls, to acertain the overall integrity of the official election results.

This is called "picture painting" - the "irregularities" are listed in order to paint a picture in the mind of the reader of a flawed voting process. And indeed, the question says that the purpose of the paper is to find out if these problems were significant. Now, the astute democrat would look at those "irregularities", start nodding his/her head, agree with the implied thesis (that the "irregularities" were significant) and then go looking for discussion of those "irregularities." But you won't find any. Why, with this direct evidence of "irregularities," do we need indirect evidence, such as exit polls, to acertain the integrity of the results? The answer is that those irregularities do not support the thesis. In fact, the election went relatively smoothly and those "irregularities" are isolated and did not affect the outcome of the election. The http://www.pbs.org/newshour/bb/science/july-dec04/election1_12-02.html (I know you like credentials)

Viewed dispassionately, the national elections ran much more smoothly than in 2000

Now, obviously, we could assume that there were more problems than were reported. But remember - you're alleging actual fraud: if there is actual fraud and these "irregularities" are the evidence, then these "irregularities" should be able to be linked together and stand on their own to prove it.

Also, I know I shouldn't have to, but I'll reiterate from previous discussions: this study is misusing exit poll data. The approach quoted above is not something exit polls were designed to or are capable of supporting.

With such massive flaws in the introduction, it was tough to read further. But I did [more later]...

 

[GENIERE]

Speaking of activists, the first on the list is Josh Mitteldorf.

From his home page-http://mathforum.org/~josh/

“I'm a peace activist and an environmentalist…”

It seems Josh may have authored the “The study, “Response to Edison/Mitofsky Election System 2004 Report...

Again from his page:

“…USCountVotes.org has recently released my analysis of the Edison/Mitofsky pollsters' disavowal of their own exit poll."

Amusingly it was “peer reviewed” by the organization that sponsored it. Again from his home page:

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

The core group consists of two statisticians, one being Josh himself.

..

 

[franznietzsche]

You'd also have to look at the voting patterns and timing of the polls in precincts that had apparent discrepancies. Are the Democratic university professors all voting in the morning because they don't have to worry about what time they get to work, while the Republican engineers had to wait until they got done with work at 5 PM to go out and vote, while the pollsters were out between 9 AM and 5 PM? With such a heated and close election, hearing early results and exit poll data on the news may have motivated Bush supporters to finally get off their butts and hit the polls later in the day when they realized it wasn't going to be the clear victory they thought it was going to be.

BUt the democrats are too stupid to ever think that people voting by economic status (something they assume so readily ) would mean that they would vote at similar times. I mean, imagine a group of people with similar schedules all voting at the same time in the afternoon. Egads!

Population groups are not randomly distributed. Exit polls were coming out early in the afternoon. You cannot take a sample from only one part of the day and call it representative of the whole. It does not work that way, it is not a valid assumption. PERIOD. Get that through your heads.

The exit polls are NOT meant to be independent verification of election results. Stop treating them like it.

 

[russ_watters]

Missed this before

You'd also have to look at the voting patterns and timing of the polls in precincts that had apparent discrepancies. Are the Democratic university professors all voting in the morning because they don't have to worry about what time they get to work, while the Republican engineers had to wait until they got done with work at 5 PM to go out and vote, while the pollsters were out between 9 AM and 5 PM?...

Actually, its women who are the problem (aren't they always? ) Women constitute a higher fraction of the overall voters than men and and even higher fraction during the daytime. One of the flaws in the exit poll (adjusted at 7:30 in the election evening) was that the total number of female voters estimated was 2 percentage points higher than it should have been.

 

[franznietzsche]

Missed this before Actually, its women who are the problem (aren't they always? ) Women constitute a higher fraction of the overall voters than men and and even higher fraction during the daytime. One of the flaws in the exit poll (adjusted at 7:30 in the election evening) was that the total number of female voters estimated was 2 percentage points higher than it should have been.

They must have lied about their genders when they voted! The exit poll has to be right!

Oh...

 

[Moonbear]

Missed this before Actually, its women who are the problem (aren't they always? ) Women constitute a higher fraction of the overall voters than men and and even higher fraction during the daytime. One of the flaws in the exit poll (adjusted at 7:30 in the election evening) was that the total number of female voters estimated was 2 percentage points higher than it should have been.

Okay, I deserved that one after all the cracks about the Republican engineer this week. Well, gosh, the exit polls might have been spot-on if there were only more househusbands voting. It's probably not PC to acknowledge that, is it?