CNN: Warren Mitofsky, "father of exit polling," dies at 72

Mitofsky, 'father of exit polling,' dies at 72
September 2, 2006
From Keating Holland
CNN

WASHINGT0N (CNN) -- Warren Mitofsky, considered by many to be the "father of exit polling," died of heart failure in New York on Friday. He was 72.

Mitofsky changed the way the media covers elections by pioneering the use of exit polls to project winners in U.S. elections beginning in the 1960s. He also developed many of the telephone polling techniques still in use today.

Mitofsky worked for CBS News for nearly three decades before leaving in 1990 to head Voter Research and Surveys, the first network exit poll consortium.

When Mitofsky joined CBS in the 1960s, political pollsters relied on house-to-house interviews to project winners of elections in the coming days.

Mitofsky developed the technique of canvassing people soon after they voted into a staple of modern news coverage, changing the way elections were covered and "called" by network news....

http://www.cnn.com/2006/US/09/02/obit.mitofsky/index.html

I wonder if my e-mails from him are historical documents now?

:hi:

The fact that he replied directly to some of us says something. I'm not sure what, but I'm inclined to feel that it was, whatever the verdict of history, in his favor.

Maybe he was just as duped as the rest of us. If that's the case, however, that's yet another black mark against the cabal: making the last major election of his life the most embarrassing of his career. Hmm...

at all from this last election and the personal attacks he received.

He had the eagerness to defend his work directly that contrasts so starkly with the * cabal's reliance on secrecy, both offical and institutional.

I think your instincts may be correct.

It can't have helped his blood pressure. He sent Ron Baiman several recent emails which were - to be kind - a bit terse.

Mitofsky did not benefit from this - undeniably this is the case.

The 2002 election saw his data discarded at the last minute - and as far as I am aware never actually released.

The 2004 debacle saw his integrity questioned and attacked. And given the fact that he did nothing for several months after I can only surmise that the NEP - i.e. the networks his clients - told him he was not free to defend himself until it no longer mattered.

When he finally did get to "defend his work" as you put it he defended himself by assuming all responsibility for the screwup on his own poll. He was not defending himself or his work but the official story.

He may have done this willingly beleiving that his flawed methodology was really the cause of the discrepancies, or he may have had his hands tied behind his back.

It seems we will never know now - at least not from him - unless he has left some sort of tell all will in a safe deposit box to be opened on is death. Now that would be interesting :)

One thing is certain about Mr Mitofsky, and that is that he has taken a lot of very important inside information pertaining to the health of American democracy to the grave with him.

Maybe stress due to the official results of those exit polls from 2004, have never been released. Would love for something official to be found in a safe box to be opened in event of his death :)

it is hard to know where to start.

But as I only just heard the news, it's hard for me to start anywhere.

Maybe later. For now, I'll just say that he gave me carte blanche to find what I could in that data that might explain the discrepancy. He wanted to know the answer.

The 2002 election saw his data discarded at the last minute

Not at all. Nothing was discarded, AFAIK.

- and as far as I am aware never actually released.

Yes, the questionnaire data was released, as it always is, in full, and until recently was dowloadable by anyone. What was not released were precinct identifiers, as this would raise a real possibility of respondent identification. Nonetheless "blurred" data for Ohio was released, and my understanding is that other "blurred" datasets may have been commissioned.

The 2004 debacle saw his integrity questioned and attacked.

Indeed.

And given the fact that he did nothing for several months after I can only surmise that the NEP - i.e. the networks his clients - told him he was not free to defend himself until it no longer mattered.

Immediately after the election he instigated an exhaustive investigation into the causes of the discrepancy, and made a report on his findings publicly available within 3 months.

When he finally did get to "defend his work" as you put it he defended himself by assuming all responsibility for the screwup on his own poll. He was not defending himself or his work but the official story.

He neither "defended" himself nor the "official story". He published a report detailing the conclusions of his investigation.

He may have done this willingly beleiving that his flawed methodology was really the cause of the discrepancies, or he may have had his hands tied behind his back.

He was certainly convinced that a biased sampling was the major contributor to the discrepancy between the poll and the count in the raw data, because his investigation provided substantial evidence in favor of that hypothesis. However, he continued to seek more detailed answers - which was why he hired me.

It seems we will never know now - at least not from him - unless he has left some sort of tell all will in a safe deposit box to be opened on is death. Now that would be interesting :)

Well, you could try this:

http://www.exit-poll.net/election-night/EvaluationJan192005.pdf

if you haven't already. Also, there is more information (from my own work for him) here:

http://inside.bard.edu/~lindeman/slides.html

One thing is certain about Mr Mitofsky, and that is that he has taken a lot of very important inside information pertaining to the health of American democracy to the grave with him.

Unlikely.

1. The 2002 exit polls were what were discarded altogether. I wasn't talking about the 2004 data.

I.E. Warren's problems started well before 2004. And between 2002 and 2004 I do not recall any attempts to explain what happened in 2002.

2. If immediately started an investigation means waiting till January 20th and then starting an investigation then perhaps you are right. (no doubt his internal investigations began before Jan 20th but nothing from mitofsky saw the light of day till after the inaugeration by my recollection - the data meanwhile was being repeatedly and keenly requested but not supplied during this period.)

Can you give us the date of your engagement Febble. I would be fairly certain that it was post Jan 20th.

anything about 2002.

But yes, "redshift" problems are apparent in the data for the presidential election years going back to 1988, as given in the E-M report. They have been the subject of considerable research.

No, the investigation, as you must be aware, was completed by 19th January, not started then. It was completed within 3 months of the election.

I was hired to reanalyse the data on the 7th May, 2005.

are on the right lines. Except that my perception was that his eagerness was not to "defend his work" but to pinpoint exactly where the problem lay. That was why he hired me.

I never met him, but obviously we exchanged many emails. My over-riding impression was of a man driven by a passion to know.

There is absolutely no doubt in my mind that Mitofsky believed that the exit poll discrepancy was due to bias in the sample, and that if he hadn't believed it, he wouldn't have said he did. Moreover, his belief was based on actual evidence in his own data. Indeed he was warning that his sample was biased in some states before a single result had been reported, on the basis discrepancy with pre-election polls.

People may (and no doubt will) disagree with Mitofsky's interpretation of his data. But the idea that he came to his conclusions for any reason other than informed, evidence-based conviction is simply false. I found him intelligent, funny, open-minded, and utterly intolerant of sloppy statistical inference.

I don't know what to say...

... until well after his engagement with us in discussion became purely academic. Moreover much of his engagement was cantankerous, combative, dismissive, patronising and generally obstructive.

I am sorry he has died prematurely. However that sympathy does not mean I am inclined to now start to engage in an effort to rehabilitate his conduct in relation to the 2000-2002 & 2004 elections.

If indeed these elections were stolen as many of us beleive they were - then his exit polls were an important part of the puzzle that enabled this to happen. His refusal for so long to engage at all and the poor quality of his engagement once it started has been the opposite of a service to democracy.

That said.

RIP Warren.

They are available via ICPSR.

Mitofsky was indeed famously combative -- one reason among several why his colleagues of my acquaintance didn't consider it remotely plausible that he was in any sense complicit in a Republican cover-up.

Last time (as far as I know) that you addressed the exit polls, you asserted, "They actually said nothing for days weeks infact" -- notwithstanding that Mitofsky was interviewed on PBS three days after the election (transcript here ). I don't know that you are really in a position to speak "for the record."

But if you ever want to get back to those arguments, I can continue to lay out for you why survey professionals generally disagree with your point of view -- which might be of some interest to you, if you actually think that the issues matter, as I certainly do. I don't profess to understand your apparent willingness to Just Know that the experts are wrong, but to make no effort to persuade them.

I was unaware that the 2002 data had been released. Has it been analysed?

As for this PBS interview

"WARREN MITOFSKY: Well, Kerry was ahead in a number of the -- in a number of the states by margins that looked unreasonable to us. And we suspect that the reason, the main reason, was that the Kerry voters were more anxious to participate in our exit polls than the Bush voters. That wasn't the case in every state. We had a few states that overstated the Republican margin. But for the most part, it was Democratic overstatement for the reason I just gave you. "

This was news to me. But doesn't it seem odd OTOH that the rBr theory was already fully formed 3 days after the poll and well b4 anyone had had a chance to look at the data in any analytical fashion. In light of this new revelation my allegation that the theory and analysis which took place subsequently were fashioned to justify the hypothesis seems all the more likely?

Finally OTOH I find your and febble's continual, persistent and neverending attempts to pour scorn on anyone who seens anything unusual in the exit polls being so spectacularly wrong for 2 elections in a row annoying and more than a little patronising.

Just to remind you in both 2002 and 2004 there was:
- Motive to steal elections
- Opportunity to do so (unaudited election machines which are easily hackable and in the case of the 2000 election tabulation systems which we know that someone attempted to hack - Volusia county.)

The exit polls provided evidence that something was up and yet Warren Mitofsky who conducted the most expensive and detailed exit poll in history in 2004 was able to conclude within 72 hours that his poll was spectacularly wrong because of rBR.

None are so blind as those who refuse to look.

even before a single result was reported, E-M was aware that there was a problem with the poll because of discrepancies between their poll data and pre-election expectations. Not only that, but it wasn't a new problem. They'd had the same problem in previous elections, and they actually had experimental evidence that what you call "rBr" was a hazard for their poll. So no, it doesn't seem odd to me that "rbr theory was already fully formed 3 days after the poll" because it was formed years before the election, and had been experimentally investigated, and far from being "b4 anyone had had a chance to look at the data in any analytical fashion" they'd actually spotted the discrepancy on the day of the election itself. Wonkette, in fact, leaked this information at 5.40 pm on the 2nd November.

http://www.wonkette.com/archives/hot-fresh-polling-024809.php

I refute your allegations regarding my "continual, persistent and neverending attempts to pour scorn on anyone who seens anything unusual in the exit polls being so spectacularly wrong for 2 elections in a row annoying and more than a little patronising." Plenty of scorn has been poured on me for even questioning the received DU wisdom that the discrepancies could have been due to bias in the poll. I do not pour scorn on those who consider it "unusual" - though in the interests of accuracy I have been known to point out that the discrepancy in 1992 was nearly as large as in 2004. Flawed statistical inferences drawn by those who ought to know better have also made me angry on occasions. And I am prepared to accept that people have found me annoying. But the charge of patronising I completely refute. It would be completely hypocritical. The reason I was even interested in the exit poll data was because I suspected fraud might have been responsible for the discrepancies. For precisely the reasons you give.

Within not 72 hours but before the poll was even completed, i.e. before a single vote count, corrupt or not, was available, Mitofsky was aware that his sample was likely to be biased - as it had been on previous occasions. Yet far from refusing to look he conducted an extensive investigation into correlates of the discrepancy, including correlations with voting technology, and published the results within three months of the election.

Like many people, I was not satisfied with the analysis, and I said so, publicly. I thought the analysis was flawed. So in May of 2005 Mitofsky actually hired me to re-analyse the data. Your charge that either Mitofsky or I "refuse to look" is therefore absolutely refuted.

I looked. He asked me to look. I looked bloody hard. I turned the damn data upside down and inside out looking for fraud. And what I found was strong support for the hypothesis of differential participation rates. Plus evidence that actually contra-indicated the hypothesis that the discrepancy was due to fraud.

Sir, once you write, "...you seem determined to undermine exit poll analysis so intently that one cannot but wonder whether you are doing so with clean hands," you have no right to expect my responses to rise to the level of patronizing. If you don't have an argument, then you don't have an argument. Call it "scorn" if you will.

Considering that I had to point out to you a factual error that you could have corrected with a few moment's googling, I think it is pretty funny that you close your post with "None are so blind as those who refuse to look." I will apologize for laughing at my expense when you apologize for questioning my motives. I am not holding my breath, but I am prepared to be happily surprised.

It is facially risible to take Mitofsky's conjecture "that the Kerry voters were more anxious to participate in our exit polls than the Bush voters" and represent it as, "the rBr theory was already fully formed 3 days after the poll." I'm sorry, but that just isn't serious.

Laughing at my expense, indeed. Well, believe it or not, I'm usually willing to do that, too. (I'm sadder about the apostrophe, really.)

But really, as far as I can tell, you aren't actually thinking about the exit polls. Non-response bias is a major concern in survey research, as you can readily verify. Neither Mitofsky or anyone else in survey research would have to retreat to a mountaintop to come up with the idea that Kerry voters may have been more willing to respond than Bush voters. It is pretty frustrating to walk through Survey 101 over and over for people who are determined not to learn anything, or even to look around and see whether maybe there is actually something to learn.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=448047&mesg_id=448063

Thanks skids!
TIA's eulogy?
TruthIsAll
Progressive Independent
Edit on Sat Sep-02-06 06:51 PM

Ultimate Smoking Gun: Change in 2-Party Gender/Race Vote (2000-2004)

Consider the Democratic RECORDED vote in 2000 and 2004. One would expect that female AND male 2-party vote shifts would be nearly equal. But the female vote shifts to Bush were dramatic while there was virtually no change in the male vote.

According to the 12:22am National Exit Poll, Kerry won a solid majority(57%) of those who did not vote in 2000 (mostly newly registered and female) and 70% of Nader 2000 voters, how could:

1) Over 10% (net) of white females who voted for Gore switch to Bush?
2) Over 4% (net) of black females who voted for Gore switch to Bush?
while
3) 1.4% (net) of white males who voted for Bush switch to Kerry?
4) 0.87% (net) of black males who voted for Gore switch to Bush?

These are the RECORDED votes, NOT the exit polls.

Can anyone explain the wide discrepancy between the male and female vote shares between 2000 and 2004?

TO BELIEVE THAT BUSH WON THE 2-PARTY VOTE BY 51.5-48.5%,
WE MUST ALSO BELIEVE THAT HIS 3% MARGIN WAS DUE ENTIRELY TO FEMALES WHO
a) VOTED FOR GORE or
b) DID NOT VOTE IN 2000 or
c) WERE FORMER NADER VOTERS.
...

For more click on the link.

None are so blind as they who refuse to look.

Rather than ask what was wrong with the poll lets ask is the exit poll result credible. Is it internally consistent?

We know for a fact that the exit poll result when applied to the actual results is complete nonsense in so many different ways. But if the exit poll result were looked at as a map of the actual result does it look credible?

NOTE: I don't know the answer to this question. But I do think that Febble and Warren's approach of looking at the WPE has failed to provide a credible explanation to the conundrum. And it seems to me there are several ways to skin a cat. No doubt Febble and the other "professional" investigators will disagree considering that RBR is the only answer required.

None are so blind as those who refuse to look.

that I "refuse to look" is making me quite angry. I have spent over a year finding different ways to skin the cat, precisely because I did NOT assume "that RBR is the only answer required".

It's not even the only answer I found. What I didn't find - and found evidence to contra-indicate - was widespread, vote-switching fraud.

on another thread.

The point that I want to make on this one is that regardless of whether you, or TIA, or anyone else, consider that the exit poll discrepancy indicates fraud, the fact is that a) Mitofsky did not and b) he based his view on actual evidence. Charges that he was party to a cover-up are completely unfounded, IMO, as is the charge that he somehow concocted "rBr" a couple of days after the election because he "refused to look" at an alternative explanation.

He might have been wrong. I don't think he was, but I'm prepared to debate it - indeed I was prepared to discover that he was. But the allegations made on this thread and elsewhere that he somehow invented the explanation on the fly to cover-up fraud are false, as is evidenced by his address to NYAAPOR (link in post 12), and the fact that he was flagging up problems of bias in the poll before a single vote-return was available.

"One would expect that female AND male 2-party vote shifts would be nearly equal."

Why? Did I miss the memo that says that the gender gap is a universal constant?

The fact is that both the weighted and the unweighted 2004 exit poll data indicate an appreciably smaller gender gap than in 2000. Did TruthIsAll never take the time to determine this fact, or is he deliberately suppressing it?

people might like to read the address he gave in 2004 to the NY AAPOR on receipt of a lifetime achievement award:

http://www.nyaapor.org/WMitofskySpeech.htm

Interesting to read as a former survey professional.
My personal theory on the 2004 polls is that the methodology failed, but what do I know.

is well-supported by the evidence.

Specifically, the respondent selection methodology.

to suppress the exit poll data in 2004 was a stress factor. He knew he was cheating the voters in order to pander to the media moguls.

No tears here.

did he suppress?

Your assertion has no basis in fact.

Mitofsky suppressed the exit poll data in 2004. Or he was muzzled, whatever. This cheated us of any chance for reviewing the data within a critical time frame for election challenge.

You KNOW the history very well and don't need an education from me. The archives of the DU election Reform Forum contain the details for anyone who's interested. I'm NOT interested in rehashing it. It's not debatable as far as I'm concerned.

Mitofsky did us all a huge disservice and undermined the legitimacy of media-conducted exit-polling forever. This is his unfortunate legacy.

:thumbsdown: Mitofsky

ain't necessarily so.

The data were released in January 2005. They were not, therefore "suppressed", although I suppose you could argue that release was not specifically expedited. More to the point, the answers to the question as to what caused the exit poll discrepancy required, inter alia, data that was collected after the election, namely precinct-level conditions. In other words, to determine the factors that correlated with the exit poll discrepancies, data on those factors was required.

You apparently consider that "everybody knows" the issues, and that "it's not debateable". Mitofsky, to his credit, approached the issue with a more open mind, which was why he initiated an exhaustive investigation into the factors associated with the discrepancy.

you can make excuses and try to put lipstick on a pig.

I'm NOT impressed by CYA "exhaustive investigations" after everybody gets their talking points together.

Yep my mind is closed on Mitofsky. His behavior stank. It was a hard lesson for everyone who cares about election integrity.

just some debunking of the invidious myths that have been spread about Mitofsky.

But by your own admission your mind is closed. Shame.

I made up my own mind after following the whole Mitofsky saga very closely.

Sometimes you have to close the book on a topic. What's done is done. Mitofsky skunked us. Many witnesses to this. We have to accept it and learn from it.

Sometimes it's shameful to cast continual doubt about a subject under the guise of "keeping an open mind."

till you've read to the end. Especially if parts of the beginning turn out to be fiction.

that details how unprofessionally Mitofsky handled the episode and compares it with the 'revisionist historian' perspective such as yours. I'll read that.

"Proprietary information" that can't be shared with independent auditors at the appropriate time is highly suspicious. Mitofsky got caught up in a classic conflict of interest. Classic.

As you know in statistics, even the appearance of conflict of interest is unacceptable.

a "revisionist historian".

The questionnaire data were released, as they have been in past elections. What was not released were precinct identifiers. It is absolutely standard practice in social science research not to release information that could compromise the confidentiality of participants in a study, and, indeed, is mandated by the ethical guidelines of Mitofsky's own professional organisation.

What do you consider should have been shared when, and with whom?

And what do you consider gives the "appearance of conflict of interest"?

that the TV media is a private, not public, entity in this country. Mitofsky's not releasing the data at the expected time for others to review, was "an appearance of conflict of interest" re. media-sponsored exit polling. (You know, like how it's an appearance of conflict of interest that you defend Mitofsky, having worked for him--not that it matters, but I must take your admitted bias into account). Anyway what Mitofsky did was a shocker, and outlandish to the point that I don't think this kind of compromised exit polling will be taken seriously again. That's pretty basic deductive reasoning.

Re. That straw dog argument about "precinct identifiers." This is truly BS, a line that has been put out there only to obscure the truth. Sorry you bought it. Precincts do not attach names or personal identifiers to their exit poll data--just general attributes such as age, political party, etc.--as they always have done. How deceptive to try to give the impression of "protecting privacy" when that kind of data collection happens everyday! Besides, the sampling fraction is too small within a given precinct to identify an individual person within a precinct anyway.

C'mon, the DANGER is that the data would reveal the truth about precincts where problems happened and the impact of e-voting machines on outcomes. There was EVERY motive to suppress the data. I'm not saying Mitofsky masterminded this, but at best he got caught up in a conflict of interest and his hands were tied. Just another weak individual in a world of corruption.

Good luck trying to convince the world of your Revisionist POV.

as you seem to understand, were commissioned by the NEP - the networks. They were Mitofsky's clients. The data belong to them. This is the case with all commissioned surveys - the data belongs to the clients.

Is your complaint that Mitofsky shouldn't have undertaken the survey for the networks? Or that someone other than the networks should have commissioned the exit polls? Or that Edison-Mitofsky should have done them for free as a way of checking the integrity of the election?

What seems to be widely misunderstood is that the exit polls were not designed as an independent check on election integrity. They were commissioned by the networks to do two things: to project the "winner" in each state in advance of the complete count (and therefore to allow networks to "call" each state for one candidate or another) and to provide "rich details of who voted and why" - in other words, the networks commission the polls so as to give them something to do on election night. Nowhere in that brief is there anything about the polls being intended to serve as an independent check on the count. It's not what they are designed to do.

The confidentiality issue is by no means bullshit. The questionnaire data (which is released) gives enormous personal details about each respondent, including age, race, sex, marital status, and of course, the respondents views on all kinds of things. In some cases this would allow individual respondents to be identified. In which case these details, which were collected on the condition of confidentiality (the word CONFIDENTIAL appears in large letters at the head of the questionnaire) would become publicly available information. This would be quite unacceptable in any study involving human participants. In my own research our ethics committee requires us to go to extraordinary lengths to ensure that respondent data cannot be matched to participants. The same is true of the ethics guidelines for AAPOR.

And your point about the sampling fraction is simply wrong. The sampling fraction varied from 10% to 100% depending on the size of the precinct.

I am aware that people do not take this issue seriously, and therefore infer some other motive. It is, nonetheless, an issue that all social scientists take extremely seriously.

the problem is not that Mitofsky undertook the job. The problem is that the data was not released in the timeframe expected, thereby creating the appearance of collusion. Although the data belongs to the clients, it is data of vital public interest. Mitofsky was very well aware of that, but I'm sure there was incredible pressure on him to withold that data. So as we learned the hard way, such information gathering should not be conducted by private entities who can suddenly decide to withhold data that they have made available in past years. Obvious after this disaster, is that the media polls are designed to reflect the perspectives of the media and should never have been trusted in the first place. It's clear now that we need independent polling and independent checks on election integrity.

Confidentiality should not be an issue, if you DO trust Mitofsky to be operating under the ethics of the business. Data can be released very easily with personal identifiers stripped OUT. Personal identifiers do NOT have to become publicly available. What's your point?

Sampling fraction: So HOW MANY precincts had a sampling fraction of 10% vs. 100% (and points between?) You would have to give that information for your argument to carry any weight. What's the ratio? For ex, did 1,000 precincts have 10% sampling-- to every 1 precinct that had 100%? (in which case the possible ID would be very low) Or, if you DID have a large number of precincts with 100% sampling fraction then that information would be even more valuable to everyone. But again--if there is integrity in the whole business, you wouldn't even be worried about this.

No need to reply. This is not going anywhere. :hi:

as this is a public thread, and you keep posting things that are misleading (although I certainly don't accuse you of doing so deliberately).

What is your evidence that "the data was not released in the timeframe expected"? A "fast-track" data set was released within three months of the election. Do you know when it is usually released?

E-M also, in January, took the unusual (unprecedented, as far as I am aware) of releasing a great deal of additional information in the form of their evaluation report (link below), in which they reported the results of their extensive investigation. They did not "suddenly decide to withhold data they have made available in past years". They published exactly what they always publish PLUS a great deal of additional information in the form of this 77 page report.

You then say that "Data can be released very easily with personal identifiers stripped OUT. Personal identifiers do NOT have to become publicly available." This is true. But if they had stripped out those identifiers they would NOT have published what "they have made available in past years". They would, in fact, have withheld what they normally make available each year, and political and social scientists around the country would have been, rightly, up in arms. So my point is a simple one: given the release, as usual, of full questionnaire responses, releasing additional precinct identifiers would, in some cases, have compromised respondent confidentiality.

As for your question: There were about 1460 precincts (did you think there were more?) in the entire poll, of which 1250 were subjected to further precinct level analysis (the remainder had large proportions of absentee voters - see the E-M report). The interviewing rate ranged from 1 to 10. According to the E-M report

http://www.exit-poll.net/election-night/EvaluationJan192005.pdf

On page 36, you can read yourself that of those 1250 precincts, 142 had an interviewing interval of 1 (100% of voters selected) and 198 had an interviewing interval of 10 (10% of voters selected. Each other interviewing interval was represented at about the same frequency. Of course completion rate was often lower than 100% (the average was 53%) so although in theory, over 10% of precincts could have had 100% participation, in practice 100% would have been rare. Substantial participation rates would have not have been uncommon however, particularly in rural areas. In other words, in certain precincts, a voter with a slightly unusual demographic profile (say the only 60 year old black female voter in a largely white community) could be identified. And by the same token, not only that voter's vote, but that voter's marital historal, sexual orientation, and views on all kinds of issues from security to gay marriage, would be in the public domain.

As I said, it's by no means a trivial issue. It's not insuperable, as the ESI study of "blurred" data from Ohio showed, but that involved a great deal of time-consuming data preparation.

But I think, from your post, that you, like many people, have grossly overestimated the number of precincts in the poll. In large states it was about 50 - fewer in smaller states. Out of about 11,000 precincts in Ohio, 49 precincts were in the poll. The exot poll was not an audit, and not designed as such.

was revised since I responded, but FWIW:

Sure, I am bound to have a bias in Mitofsky's favour, seeing as I did work for him. I don't expect people to agree with my assessment - merely to take it into account. What I do expect is for people to support their own assessments with facts.

As for conflict of interest: it is of course true that anyone who is paid for anything is at risk of being suborned. It is one of the reasons we have professional organisations, and why, in many professions, we can be expelled from those organisations if we violate ethical standards of that organisation. And it is why, if we find ourselves in a position where what we are asked to do in the course of work conflicts with the ethics of our profession, those professional organisations are there to support us.

So it is true that any pollster has a potential conflict of interest between that of his/her client and his/her own professional integrity, in the sense that any researcher (me for instance) has a potential conflict of interest between the principles of ethical research and the interests of their client/funding body.

But that is quite different from the conflict of interest of, say, a lawyer who has investments in a company that he is being asked to sue. Or a Secretary of State for elections who is also chairman of the campaign to elect a particular candidate.

And in that second sense, neither Mitofsky, nor I, had a conflict of interest. Mitofsky contracted to conduct the exit polls for his clients, and conducted that survey in accordance with the ethical guidelines of his own organisation - that is precisely why he took care to preserve the confidentiality of his respondents. Similarly, I was contracted, as a data analyst, to analyse the precinct-level exit poll data. My only other interest in the matter, was to my own ethical standards, to which I adhered - and in the integrity of the election.

If that constitutes "the appearance of a conflict of interest" it is no more so than that of anyone who contracts to do a professional job for a professional fee. Not that voluntary organisations are immune from conflicts of interest. Is there such a thing as a totally disinterested person? Are you?


edited for clarity

well I'm sure I couldn't have any more "facts" than you do, since you worked for Mitofsky. So I'm not going to trade "facts" with you. And puleeze don't give me lectures about professional ethics in a world where there is so little of that. I see the evidence of the failure of that noble 'professional ethics' concept everywhere these days. I'm sure you do too.

Bottom line = Mitofsky colluded in the withholding of exit poll data beyond the point at which it would have been useful for contesting any aspect of the election. It was withheld in a manner that was inconsistent with years past, highly unusual. This was information of vital interest to the voting public. And this displayed integrity? Was this a decision in the true interests of the American voter
or was it a decision based on the interests of the "client" only? There is reason to suspect that data was changed to match the client's interests at the time of the election. But we were not given enough "facts" to KNOW the truth.

LOL "confidentiality" smokescreen. Why not worry about that before the election--oh so, the topic suddenly came up AFTER the election?

Go back to tarting up this pig. You've got a lot of work to do to sell this particular pig, which seems to be your quest. As far as my interest? Well this is just one more reason why our election system stinks to high heaven. We can't even get decent exit polling when the "Father of Exit Polling" does it. My interest is to IMPROVE this sad situation, whatever it takes. Your interest is to shore up a crumbling facade.

"facts" - then you trade assertions unsupported by evidence.

What is your evidence that "Mitofsky colluded in the witholding of the exit poll data beyond the point at which it would have been useful for contesting any aspect of the election"? Or that it was "inconsistent with past years?"

The only difference I am aware of between 2004 and past years was that for 2004 E-M provided additional information. If you have alternative information, please provide it.

And I'm sorry you don't want lectures on professional ethics - it seems to me you just gave me one.

As for your statement that "There is reason to suspect that data was changed to match the client's interests at the time of the election" - again - what is your evidence?

Let me, in fact, suggest what your evidence is. Your evidence is that the early estimates of the vote proportions for the two candidates had Kerry at 51% and Bush at 49%, whereas later estimates had these two figures reversed. This is, of course, quite correct, the reason being that the later estimates were reweighted in line with the incoming vote returns. And the reason for this was because the pollsters assumed, as they always do, that the vote returns were correct, and that their crosstabulations of who had voted for whom would be more accurate if they were reweighted to the count. Moreoever, they based their "projections" on the vote returns, because the counted result was what they were hired to project. They were NOT, repeat NOT, hired to conduct an independent check on the count. The count was assumed to be accurate. Beforehand. And details of how the projections would be made were given on the E-M website before the election, and are still there:

http://www.exit-poll.net/faq.html#a10



from which you will learn that not only the exit poll interviews, but also vote returns would be used to make the projections. The reason for this being, as you will gather if you read the link I gave in post 12, that Mitofsky, like all survey researchers, is acutely aware of the hazards non-response bias in polls. Indeed he used exit polls to research this very phenomenon.

Now, in retrospect, of course, we had cause to suspect that perhaps the reason for the discrepancy in 2004 was not non-response bias but fraud (although the discrepancy had been almost as great in 1992). But there is absolutely no reason to suppose that the reweighting of the crosstabs, and the use of vote-returns to make the state projections was either unprecedented or done to "match the clients' interest" except in the trivial sense that what the clients commissioned was a poll that would project the results. Which it did. As it always does, because it's what the damn things are designed to do.

You ask for "decent exit polling". Exit polling was not designed to monitor elections. Mitofsky was the Father of Exit Polling" because he figured out how to make good projections of the results. The reason he had such a great reputation is that he was famously cautious about making any projections until he was sure of the answer. In other words, in all close races, the projections were not made until a substantial proportion of vote returns had been incorporated into the estimates.

Call it tarting up a pig if you like. I'd put it differently. It was always a pig. It's still a pig. It's only because people thought it was a thoroughbred racehorse that they were aggrieved to discover it was a pig.

But pigs have their uses. The survey data is invaluable. And in fact, it was possible to reverse engineer it a bit to try to find out whether it indicated that there was corruption in the count. That's one of the things I did, and I found that the evidence pointed, if anything, the other way.

You say "exit polling was not designed to monitor elections"... and then you say that the survey data is 'invaluable' and you found this data useful to point away from corruption in the count (!?). So WHY were you doing this at all IF exit polling is not designed to monitor elections? And why were others not allowed to analyse the exit poll data immediately after the election? And why is half the world under the impression that exit polling is useful to monitor elections?

So we should just give up on exit polling entirely then? From what I've seen, maybe that's the best way to go. What do you think? Basically useless for election integrity? Only good for media purposes? (And we all know how much that's worth).

So maybe the end point of this discussion should be--to just forget exit polling, instead of better exit polling. I'm interested in protecting the American voters. Maybe we have to find new ways of ensuring election integrity, since we have been so conditioned to believe that exit polling means something. Can we agree on that?

I am just aware there is a lot of misunderstanding about what exit polls are for, what they can and can't do, and what they can and can't tell us about the integrity of the election.

The exit poll was not designed to monitor the election. Exit polls are very poor tools for monitoring elections, because there is no guarantee that you will get a representative sample of voters, although some poll designs will work better than others. They are mostly used to give an early estimate of what the result is likely to be, because, unlike pre-election polls, the respondents have actually voted. So they can be regarded as the best pre-result estimate of what the result will be. The E-M polls, however, are also designed to do something potentially much more interesting, which is to collect data as to who voted for whom and why. That's why the questionnaire is so long.

So when I say that the survey data is "invaluable" it's that second purpose that makes them so. The questionnaire data archived with the Roper Archive is an amazing record of voting patterns and trends over many elections.

But if you want to use them ascertain whether vote corruption occurred, then you have to reverse-engineer them to do so. And you have to take into account evidence for bias. And that is what I did. I investigated the correlates of discrepancy in the poll, at precinct level. The first thing I wanted to know was the extent to which the discrepancy was correlated with factors likely to be associated with bias. And it was. And the second thing I wanted to know was whether it was correlated with factors likely to be associated with vote corruption. And mostly it wasn't. I did find a correlation between the discrepancy and voting technology in urban precincts (not in rural or suburban precincts) but, surprisingly, the finding was that the discrepancy was greater where non-digital voting technology was used (levers; punchcards) not digital technology. But most importantly, I found absolutely no correlation between the discrepancy and benefit to Bush. In other words there was absolutely no tendency for Bush to do better in precincts where the discrepancy was in his favour than in precincts where the discrepancy was less, or in Kerry's favour.

To address your other questions: the dataset I analysed included data collected from interviewers after the election (and reported on in the Evaluation report), so it wasn't even available, in full, until after the election.

I don't know why so many people think that exit polls are used to monitor elections. They have been used, although one problem is that the more unstable the political system, the less reliable they are likely to be. Yes, I think exit polls are basically useless for ensuring election integrity. I'd much rather see mandatory random audits, transparent counting methods, and a secure chain of custody of the ballots. Yes, they are primarily a media tool. That doesn't mean they are not useful. Knowing the kinds of reasons people voted for Bush (and why they didn't) is important knowledge in a democracy. It tells you, amongst other things, what arguments you have to win, to win.

And yes, we can agree that we need new ways to ensure the integrity of elections - especially with electronic voting methods! I think HR550 is a good start, but it's only a start. And even audits won't help with voter suppression, which remains a huge blight on American democracy.

Thanks for listening!

Anthony S. Reissig, 59, was an election expert

http://news.enquirer.com/apps/pbcs.dll/article?AID=/20060902/NEWS0104/609020373/1060/NEWS01

BY REBECCA GOODMAN | ENQUIRER STAFF WRITER

MOUNT WASHINGTON -...

"Anywhere they would have free elections, he would go," said his friend, Standford Mendelson of Dayton. "He was one of the premier election officials in the world."...

Here at home, Mr. Reissig was called to Washington, D.C., after the 2000 election to work with the National Institute of Standards and Technologies in setting standards for the country's voting equipment...

He was an official observer at the first Florida election using touch-screen voting...

Mr. Reissig, an administrator for the Hamilton County Board of Elections, died of brain cancer at Hospice of Cincinnati in Blue Ash on Aug. 22. The Mount Washington resident was 59...








(just kidding i think)

.. are there any more?

My own trip down the rabbit hole began two days after the 2004 election, when I wrote Mitofsky asking him four direct questions about the discrepancies in his exit polling. I never heard back from him (surprise, surprise).

I have been taught (here in the polite South) that if you can't say anything nice about your enemies, you can alway say they live in a beautiful state. Well, Mitofsky lived in a beautiful country, one with democratic traditions that I hope we can preserve, despite his best efforts to the contrary.

ain't necessarily so."

My very thoughts concerning your own panegyric to your unfortunate former employer, Liz. Your claims concerning his character sound extraordinarily presumptuous - as if your own probity were clearly unimpeachable, nay, sovereign, and should immediately and automatically put all doubts concerning Mitofsky's probity to bed, forthwith! Nanny's here. Mrs Doubtfire indeed.

We know you are extraordinarily impressed by your own scholarly discipline and grasp of the issues, but some of us feel that you and The Great Vacillator may not be in the best position to be aware of your own short-comings in terms of the weightier matters of your own underlying assumptions.

I can't remember now, did he mess up the exit-polls for both elections, 2000 and 2004?


Paul Becke

but whether it is open to debate.

is the truth or falsehood of the arguments.

an excerpt from a statement recently posted on AAPORNET, which I won't attribute since I haven't requested permission to quote it: "With Warren's passing our profession has lost a man of great dignity, great integrity, and of the highest ethical standards. The field of public opinion research has suffered a great loss; his contributions will be missed."

I must caution you that whether you believe it or not, this view is widely held. Human decency and fairness aside, election integrity activists (and/or wannabes) do pay a price in credibility when they casually slander people of such reputation.

It won't be lost on any thinking person that Mitofsky's work on the 2004 exit polls was highly anomalous, nor, perhap more significantly, that it follows a pattern with regard to people whose work represents extreme danger to the power of the incumbemt regime; that he would have been permitted to carry out his work without being subject to the direst threats to himself and/or his family seems virtually inconceivable - particularly after the Ukrainian election.

At least one official investigating possible fraud by the neocons was murdered in quite bizarre circumstances, and the subequent police investigation or lack of it, equally bizarre. It was, I believe simply blamed on organised crime - though what their interest in his work could possibly be - other than as historic friends of the CIA'S Republican-friendly so-called B Team - is impossible to imagine.

Also, a man who manufactured voting machines with paper trails was careered into by a large articulated truck not long before the election.

It is also interesting that the international observers at the election seemed to OK it at the time, yet once they had returned home, stated that there were many gross irregularities they had observed. I can't quote chapter and verse, but I dare say this could be checked by web-savvy individuals here.

Do you consider that compliance with the neocons under such duress is shameful? I don't. But neither do I accept your opinion that Mitofsky's conduct of the exit polls in 2004 represented his normal evidently high level of integrity. (As I recall, his work had been greatly admired by partisans of both parties, and he was himself held in the highest regard).

Perhaps most mysterious is the equally anomalous, one-off, 180 degree volte-face of certain Democratic luminaries (and I'm not using the term ironically here), who also represented a fierce threat to Bushco. One a major figure who suddenly began touting a Democratic candidate of the most dubious credentials, who had hitherto been a staunch Republican supporter, and was quoted by a named Republican politico as admitting that he was standing for the Democrats from pique at the failure of a Republican grandee to return a call he had made to him. He had also been an avowed apologist for the School of the Americas.

Another singular volte face, imo, was performed by an absolutely brilliant analyst of the shenanigans of Bushco, whose blog often figures on DU's home page, when he poo-hooed some manifest collusion of Busco in an outrage against the country other than electoral.

If you want names, you won't get them from me, but if you are really interested in the welfare of the US, then you would be aware of them.

It's been pointed out repeatedly that the 1992 exit poll discrepancies were almost as large. Characterizing the 2004 results as "highly anomalous" is hyperbolic at best.

If you're speculating that Mitofsky was killed to silence him, well, out of respect for his family I will just leave it there. (If you think that the 2004 U.S. election was after the Ukrainian election, I will just leave that there, too.)

first language? Someone else suggested he feared Mitofsky might have been murdered, and it did set me thinking about it some time later.

If someone can prompt you with your reading, you'll find I suggested it seemed to me highly probable that he and other powerful Democratic critics of the Republicans had been very heavily leaned on, threatened. And since elements in the Republican party and the CIA appear to have a track record that makes Pol Pot seem small beer, perhaps they wouldn't dismiss the direst threats out of hand. There would be no shame in deferring under duress to the wishes of such people in his work.

For a moment, I thought it might have been because you were prepared to cynically twist the meaning of my words to score a cheap, unctuously pompous and self-righteous point at my expense, re respect for his family's grief. Take each sentence, no, each phrase at a time. There's no shame in wanting to be able to read properly.

Well done for your indulgence re my anachronistic reference to the Ukrainian election, though. It's so important to distinguish between the fraud and intimidation of the Republicans and their minions in the elections in 2000 and 2004, isn't it?

PS: Don't start me on the fathomless cynicism and hypocrisy of the capitalist West in ensuring that Pol Pot died in his bed.

Why would anyone familiar with survey research suppose that the exit polls evinced* fraud? and, therefore, why would there be any need to lean on Mitofsky? and how many hundreds of other people have they had to lean on?

*EDIT TO ADD: evince: "to show clearly; make evident or manifest; prove." Certainly the exit poll discrepancies could have been at least partially attributable to fraud, but one couldn't use the exit poll interviews to determine the actual votes. I don't think that the exit poll discrepancies can be attributed to fraud, although that isn't to rule out the possibility of fraud.

your English. It will pay in the long run, you'll see. And you'll thank me for it, later.

Poor Tweedle-Dum, she's so bereft of ideas, she's just chided me on my English and recommended a English primer to me!

that you have not, as usual, addressed my question.

Bereft of ideas?

with the novel idea of mocking their opponent's somewhat limited grasp of English!

But why would I address any of your question? You know the saying about making two fools. And that is not a glib joke. There has to be shared assumptions or there can be no intelligent engagement. You could point it at me, but you seem to think people will be taken in by your technical ramblings, while you altogether miss the bigger picture.

Anyway, I've just addressed one of them in another response, but that's it now. What we all require from you two, is the simple acknowledgement of what we all know, i.e. that the Rebublicans played merry hell with those machines, before, during and after the election. Nor have we heard a plausible reason from you as to why M's predictive polling suddenly went down the toobs.

I know you're going to come back with technical guff that will impress the life out of you, but simply doesn't address the issues above a microscopic level.

"You appear to have difficulty with your reading. Is English your first language?"

You (2):

"That's a good one coming from someone who's just come up with the novel idea of mocking their opponent's somewhat limited grasp of English!"

that I was ridiculing your pot/kettle syndrome. I ridicule TGV's English comprehension, then you, mine. So original!

have little idea what you were trying to say, if you weren't trying to say what OTOH thought you were trying to say. Perhaps you should take a look at this organisation:

http://www.plainenglishawards.com/

then state (in plain English), what it was about "Mitofsky's conduct of the exit polls in 2004" that you consider did not represent "his normal evidently high level of integrity", and what form of duress you consider he was under.

But please note: if your answer implies that you think that the fact that he based his projections of the winner in each state on vote-returns as well as on exit poll data represented "compliance with the neo-cons", then you might want to know that this is how the projections have been made for years, and the methodology is what he is famous for developing.

the other way round.

In any case, to make an argument worthwhile, there has to be shared assumptions, and yours and Tweedle Dee's are so obstinately and frivolously pedantic that, rightly or wrongly, I harbour the gravest doubts concerning your good faith. We must agree to differ. But I don't believe the passage of time is going to be kind to either of you, in terms of your position on this issue. Time will tell, as it always does.

I'll give you just one example of what I consider to be your myopic mindset, both of you:

You adduce protection of voter privacy as the justification for Mitofski's refusal to release the data that would settle or go a long way towards settling the issue, but the resolution of that issue, the revelation of the truth, one way or the other, is of infintely more importance to the common weal. People are routinely executed by the State, though it is known and acknowledged that the justice of the legal process is anything but safe in that area. I'm not talking about divine justice, but human justice, in the case in point particularly with regard to the poor and marginalised.

Yet you have the gall to maunder (and keep a straight face) over infringement of the voters' right to privacy, when it has been shown that hundreds of thousands of citizens fully entitled to vote were criminally prevented from doing so; when there is a mountain of evidence to suggest that allowing Mitofsky's poll data to be examined could have contributed significantly towards establishing the fraudulency of the elections, as evidenced (not 'evinced', you'll note) in so many other ways. Remember, it has been established that in 2000, Gore clearly won the vote for presidency, but in their demented lawlessness the Republicans procured the good offices of the Supreme Court to anoint Bush as President!!!!! Once the elections are cleaned up, voter privacy can be available for all.

I have not even seen a single word of acknowledgement by you or Tweedle Dum (it's easy to get you two muddled up) of your even being aware of the outrageous shenanigans of the connections of the voting-machine companies prior to and during the elections; for all we know, you are unfamiliar with the name Wally O'Dell, or with the video clip of the drunken neocon troglodyte boasting, "Don't worry. It's all over bar the counting, and we'll take care of that."

Absence of evidence is not evidence of absence, and given that the evidence is there, your ignorance of it looks like myopia to me.

But let me take your points:

It might be possible to make the argument that in the case of the 2004 election, the benefits of releasing data that would compromise the confidentiality of the exit poll respondents might be trumped by the greater public good. What I am arguing is that the argument that the pollsters owe a duty of confidentiality to their respondents is not a trumped up argument to hide the truth, as some have alleged. It is a serious issue, and is, quite simply, the reason why precinct identifiers are not released with the data. The data is, in all its detail, released, and publicly accessible, for free (or was until a couple of months ago).

So the next question is: what could the additional data, if released, tell you about the 2004 election? I don't know, what do you think it could tell you? I've been asking the data what it can tell me for a year, and so far all it's told me is that a number of methodological factors were associated with "redshift" in the poll, and that "redshift" was not correlated with benefit to Bush. Do you have a hypothesis you think should be tested? What do you think it can prove - or even indicate, that the data currently released cannot? In other words, what argument are you making that would trump the confidentiality of the respondents? What is this "mountain of evidence to suggest that allowing Mitofsky's poll data to be examined could have contributed significantly towards establishing the fraudulency of the elections"? Or is it simply that you do not believe the findings?

And I completely fail to see the connection between protecting the confidentiality of voters and voter suppression. Voter suppression, whether due to malice or negligence, in the US is outrageous. If you are interested you may even like to see my work on one instance of it here:

http://uscountvotes.org/ucvAnalysis/OH/FranklinCountyReport_v2.pdf

Does it have anything to do with respondent confidentiality? Not a thing.

As for your final comment - well, as I said, it is evidence of your own lack of omniscience only. You might be interested to read another piece of work of mine on voting machine problems in NM here:

http://uscountvotes.org/ucvAnalysis/NM/NMAnalysis_EL_JM.pdf

OTOH has worked on both these issues, as well as on analyses of other data in connection with the conduct of the 2004 election in Ohio.

I have to ask - what have you done?


edited for clarity

discussion on the shocking gallimaufrey of malfeasance in the preparation, manufacture, and use and abuse of the voting machines during and after the 2004 election perpetrated by Republican operatives, and indeed sought to refute it. This, oddly enough, at a time when the Republicans seemed to be becoming particularly nervous about their chicanery with the machines.

You are still rabbiting on frivolously about red shifts and gizmeters, without addressing the fundamental questions, some of which I cited for you in my last post.

That remorselessly pedantic tunnel vision you show in asking me the following question in your last post, makes any kind of argument with you utterly futile:

"And I completely fail to see the connection between protecting the confidentiality of voters and voter suppression. Voter suppression, whether due to malice or negligence, in the US is outrageous. If you are interested you may even like to see my work on one instance of it here".

Identifying all the elements of malfeasance perpetrated by the Republicans in the Bushco coup should trump any delicacy about nullifying the confidentiality of voters, when far worse examples of not merely nullifying the confidentiality of voters, but nullifying their very vote, abound. There... for all the good it will do, I've connected the dots for you.

I don't think any of us are any the wiser as to why Mitfsky's data began to change the data his polling people had obtained, in order to match the 'actual' (the word is used here very loosely, just the same) polling data, but I'm sure you and TGV are evidently impressed beyond belief with your own explanation, and the technical guff you adduce in support.

However, it is not a question that lends itself in the least degree to a technical justification. It's a bit like arguing with a Holocaust denier. I'm sure Irvine would and does adduce all sorts of data, mostly no doubt spurious, but some perhaps valid however ultimately irrelevant.

Fellow historians have, I believe, grotesquely accorded him with the status of a historian, but the average Joe and Jane knows that it's not really about figures, about accuracy, about truth, it's about MINIMISING, ideally, UTTERLY DISCREDITING the whole historical memory of the Holocaust and the Jewish people.

If I must explain why you two incense me, that's the closest I can get to it. I'm sure you would dispute my conviction, indeed have been doing so, insofar as I've tried to convey it without expressly analogisng it with Holocaust denial. Either that or you are the one-eyed king in the land of the blind. And that, I do not believe.

Ah, so two wrongs make a right, then.



And by Godwin's law, the thread ends here.



After all, why let facts get in the way of a good rant.

If I must explain why you incense me, it would be the willful ignorance and the unsupported, bizarre personal attacks on people who don't share it.

If you are calling for a release of exit poll precinct identifiers, surely it is relevant to consider the technical question(s) of what one could possibly do with the information were it released. If you consider the exit polls irrelevant to the fraud debate -- which, ironically, is not so far from my own position -- then what on earth are you actually upset about?

pollster would risk the wrath of presumably the supervisory body (an acronym), and what do we read on DU today, a polling company have been convicted of cooking polls in support of Bush, Lieberman, etc. I'll bet it's the tip of the iceberg.

You create your own reality and expect to bamboozle people who have more sense than you. Dream on! This is just the beginning.

When you can give a plausible explanation for the catastrophic deterioration of Mitofsky's polling expertise in identifying fraud during the 2004 election, in principle at least, you'll have a better claim to have DUers take you more seriously.

Frankly, you should be concerned about having yourself taken seriously. I find it regrettable that you muster so much passion but so little disciplined work.

"what do we read on DU today, a polling company have been convicted of cooking polls in support of Bush, Lieberman, etc."

Case in point: you may have read that on DU, but it's not what the news stories actually said. (It might conceivably be true, but I've seen no attempt to demonstrate that it is.)

"the catastrophic deterioration of Mitofsky's polling expertise"

Second case in point: the factual premise has been repeatedly refuted. Ah, but that is pedantic, technical and pompous.

Kennedy - about how, after the Wall Street crash at the end of the twenties, someone asked him why he had been selling all his stocks, getting out of the market, when everyone else or nearly everyone else had been caught up in the excitement of the price boom. To which question, Kennedy simply replied, that when his shoeshine boy told him what shares to buy, he knew there was something very wrong with the market.

Now, do you suppose that there were not a lot of highly-trained of specialists in such matters, with much higher academic accreditations than himself, who were not as wise, lost their shirt in the debacle? I doubt it - or why the story?

An ability to prioritise first principles, rather than formal accreditations obtained by academic subservience and drudgery, are the mark of high intelligence. Even the Newton, a Colossus of mechanistic science, recognised the primacy of inductive reasoning, and of course, Einstein is a byword for it, specifically stating that his criterion for selecting his hypotheses was aesthetic.

We have a lot of young atheist noggin-heads on here who rubbish anything they don't consider "scientific", as non-knowledge. Yet that other paragon of deductive reasoning, the physician, Conan Doyle, was passionately interested in the psychic world.

It seems to me that Sean Connery, with one simple remark, proved his intelligence to be of a higher order than yours. All three of you - though you're not all friends/colleagues - when he wrily commented on the wisdom of the person who told him that the difference between a rich man and a very rich man was a good lawyer. Again, first principles.

You need to check all your conclusions against what has been the clearly established modus operandi of the movers and shakers of this Administration. Quite apart from how they have consistently suborned technical specialists, in intelligence, science, you name it. What is more, the use of polls for opinion-forming purposes is already a well-established truth. The premature release of voting precinct figures is a particularly flagrant example of it - whether to influence people in later-closing precincts to vote for the winning side or to stay at home.

I'm sorry if my tone has sometimes been snidey. I haven't seen enough of your posts yet to excuse it.

KCDMIII:



Associated Press



KCDMIII:



Edison-Mitofsky



I'll bet it's the tip of the iceberg.

I assure you that I am not 'pals' with anyone on the thread. Like some others here I write from a position of extensive knowledge and experience in the industry. What relevant knowledge or experience do you bring to the discussion?

rebut your arguments. I'm sorry if my remark sounded as if your association with them was closer than it is. But you evidently suffer from the same myopic self-esteem regarding your own specialism and its relevance to the larger issues.

What excuse have you got for making such a sweeping claim? This latest item shows the foolishness of minimising the threat of bent polls - particularly where this administration is concerned.

Are you unaware of the desperately unrepresentative catchment of interviewees that were able to be polled on the phone by land-line prior to the election (not to speak of the times during the day, I believe), yet were touted in the media as in some way meaningful?

Or might do so. Control of Congress is on question, rhetoric is being ramped up, facts about the 2004 are out that are terrifying constitutionally are out, and the guy who invented modern exit polling dies of "heart failure".

Isn't all all manner of death untimately caused by heart failure?

I checked my scalp. I don't SEE any tinfoil hat...

God, I wish Malloy was on the air. Maybe Randi will mention it Tuesday, pull some history and facts together for the world to listen too.

apparently, and exit polls are planned as usual for November.

... while obviously not precise, exit polls provided
some information as to just WTF was going on. Now, alas,
exit polls are likely doomed or, at least, can now be
conveniently be explained away with a wave of the hand
and a reference to 2004.

Example:
"All of this led the authors of the internal CNN report -- Joan Konner,
James Risser, and Ben Wattenberg - to conclude :
Exit polling is extremely valuable as a source of post-election information about
the electorate. But it has lost much of the value it had for projecting election
results in close elections... Cease the use of exit
polling to project or call winners of states. The 2000 election demonstrates the
faults and dangers in exit polling. Even if exit polling is made more accurate,
it will never be as accurate as a properly conducted actual vote count."

(The last sentence here makes me chuckle every time.)

Anyway, as to the discussions going on here I will say that given all
the other evidence of malfeasance in 2000, 2002, and 2004, I must look
at all arguments by people paid by the executors of any polls with an
extremely cautious eye. Not to say I won't look and listen but it's funny
how the people paid to come up with the (right) answer always seem to
do just that.

Also, given the (at least perceieved) irregularities, clinging to the
so-called "ethical guidelines" of not releasing precinct data is suspect.
Let's for a change use "truth" as the ethic and not cloud the issues.

"Not to say I won't look and listen but it's funny how the people paid to come up with the (right) answer always seem to
do just that."

Do you have any evidence that anyone was paid "paid to come up with the (right) answer," by which I assume you mean some preconceived answer? It's a serious charge, apparently lodged against another DU member.

Analytically, extreme caution certainly is warranted, personal attributions aside. As for releasing the precinct data, it might be useful to demonstrate some familiarity with the professional discourse on this issue before you commence with the rhetoric about "cloud(ing) the issues."

Suffice it to say that I can add.

Anyway, you know what they say about "assume".

evidence that you can add, but evidence that you know something about the ethical protocols that surround any study in the social sciences. Here are the AAPOR ethical guidelines:

http://www.aapor.org/pdfs/AAPOR_Code_2005.pdf

Note in particular:



And do you confirm or deny that your post was accusing me of professional misconduct?

I am happy to accept a denial - your post was sufficiently oblique that I was not sure. If you confirm then I refer you to my response Gormy Cuss.

... unless you want to pay me to do something.

I don't even know what is your so-called profession
so I can hardly be accused of accusing you of
professional misconduct now can I?

Do you have to go to school for this profession or
just join up and start counting?

Anyway, what I said is this: I don't believe you just
because you say it is so. And I will give you less of the
benefit of the doubt due to your vested interest.

See what I mean, jellybean?

That's just common sense and that's not going to change
no matter how blustery you get and no matter how many
"professional" manifestos you wave about.

For one thing, and this is not directed specifically
at you since you appear to be hyper-sensitive, codes
of professional ethics are hardly going to give pause
to anybody aiming to aid or abet the theft of an
election now are they?

Now standing by for request for my IQ or SAT scores
so I can continue to participate.

edit for clarity.

Sorry, but I don't like being smeared with innuendo. If you think I produced a result to please the person who hired me, then say so. If I did, I would be guilty of professional misconduct. If you think that is more likely because I was hired, then, frankly, you are not making sense. Does a doctor have a "vested interest" in producing a favorable diagnosis patient because the patient is paying her? Possibly, but in that case she would be guilty of professional misconduct.

I'm a scientist. I don't fudge results.

Your IQ or SAT score is of no interest to me whatsover.

This just popped up in another thread. See end of this post.
Apparently the manifesto of ethics for pollsters is not
strictly adhered to in all cases. Wot a shocker.

By and large, I think it is safe to say many reasonable
people view polls as a useful tool for one who may want
to perpetrate a little bit of propaganda.

"polling has become a kind of shorthand for everything
people dislike about Washington politics. "Pollsters
have developed a reputation as Machiavellian plotters
whose job it is to think up ways to exploit the public,"
says Andrew Kohut, director of the Pew Research Center
for the People and the Press.

I wonder why they have developed that reputation?

Therefore, a reasonable person may look at a poll which
seems to give counterintuitive information and wonder what
is going on. That same person, when told by another pollster
paid by the first pollster to bolster his claims, may want
to further consider what is going on.

Just talking in the general case here for those looking
to put together a lawsuit or whatever. :)

===============================
BRIDGEPORT — A polling company owner admitted participating in a conspiracy to falsify data in order to meet deadlines for clients, which included the campaigns of President Bush, U.S. Sen. Joe Lieberman, U.S. Rep. Rosa DeLauro and Mayor John M. Fabrizi.

Tracy Costin, 46, of Madison, admitted to U.S. District Judge Janet C. Hall that she participated in a conspiracy to commit wire fraud.

Costin, who owned and operated DataUSA, a survey and polling firm with offices in West Haven and Guilford, faces up to five years in prison when she is sentenced Nov. 30.

<SNIP>

FBI Special Agent Jeff Rovelli said 50 percent of information compiled by DataUSA and transmitted to Bush's campaign was falsified.


more
http://www.connpost.com/news/ci_4298956

and in this instance a criminal case was made.

This woman was running a small local survey house. She may or may not have had the professional training or expertise to conduct said work professionally. DataUSA, now called ViewpointUSA, has 50 CATI stations and conducts only telephone surveys and executive interviews. This is a classic description of the typical mom-and-pop vendor for this industry.

I have worked for several premier nonpartisan policy research companies. I do know how the business is run. This CT outfit is a petty fly by night operator and not in the same league as E-M or anything with Mitofsky's name attached in the last 30 years. The fact that an unprofessional one woman show was corrupt is really not relevant to this discussion, at least not to anyone who understands the business.

edited to add info from CT company web site.

and it's a serious breach of ethics for people in survey research to cook the data to please the client. Serious enough that if you have been found out as someone who finds to the client's liking rather than letting the data speak for itself you'll will be vilified by the rest of the profession. As a former survey professional I've seen it happen. Being drummed out of AAPOR is only the beginning of your problems.

So to offer a different opinion, based on my professional experience, the chances that our fellow DUer cooked the data for Mitofsky, or that Mitofsky cooked the data for his client, is slim to none. Whether either erred in the analysis I can not say.

If the allegation is that I cooked the books for Mitofsky let's have it.

I didn't. Quite apart from violating my own standards of moral and intellectual integrity, I would have had absolutely nothing to gain. I was never going to be applying for a full-time job with him - I don't even live in America (commuting from Nottingham to New York would be quite a pain). And I had to put my PhD on hold while I did the analysis, so if anything it reduced my career chances rather than the other way round. In fact, my sole reason for taking on the contract was that I wanted to know the answer. I wanted to know whether there was evidence of fraud in the data. And Mitofsky knew that.

But sure, I could have erred in the analysis. I couldn't tell you how many times I've woken up in the night thinking there might have been something I'd missed - some interaction I should have tested, some variable I should have considered - and crept downstairs in the wee hours to run another regression.

I did as good a job as I could. And I really appreciate this post - thanks.

And now it powers the cemetery.

I don't believe Mitofsky was "in on it" or "cooked the books" or benefited from the results. But, it is also inconcievable that he missed how important his work turned out to be for millions. He had an opportunity to rise above his profession, his business interest, and his personal inclinations. He chose instead to sink into the very same. What the election controversy transformed into the "right of the people", he chose instead to regard merely as his own personal domain. All who would ask questions became poachers. In the process, he developed a certainty about that which certainty is impossible and came to sneer at those whose pain was obviously a hundred times more sincere than his own.

Goethe wrote: "The Philistine not only ignores all conditions of life which are not his own but also demands that the rest of mankind should fashion its mode of existence after his own,".

This is not a crime in professional America today... it is a way of life.

But... what a horrible eulogy.

Goethe, again: "What is a philistine? A hollow gut, full of fear and hope that God will have mercy!"

I never met him, but I've met many now who knew him well, and I got to know him a bit through my work for him. The sense in which he was certain is that he knew the limits of certainty. It was caution he was famous - notorious - for, not certainty. His view of what had led to the discrepancy between the exit polls and the result in 2004 was based on good evidence and vast experience of similar discrepancies in previous polls, particularly 1992, and the methodological factors that were associated with those discrepancies. So experienced was he that he was warning that there was a substantial pro-Kerry bias in the poll before a single result had been reported.

Where you are also wrong is in the implication that his views were intractable. Sure, he was famous for the vehemence of his opinions. However, he was also famous for being able to listen to other points of view. As I said, he knew the limitations of certainty. He also knew the limits of any one person's expertise, including his own:



You wrote:

"All who would ask questions became poachers." That is not true. I asked a pretty stiff question. Far from regarding me as a poacher, he reanalysed his own data in response to my question, then contracted me to find out more. He wanted to know the answer.

To have invented an entire research methodology, and, at 71, to be so passionate about what flaws it might have, and how it might be improved, and how else the data might be analysed, that he was prepared to listen - and respond - to views buzzing round the internet, strikes me as the mark of someone blessed with the gift of eternal curiousity, not afflicted with delusions of certainty.

But sure, he was cantankerous. He had no time for ignorance.

... how about the Bard?

"A wretched soul, bruised with adversity,
We bid be quiet when we hear it cry;
But were we burdened with like weight of pain,
As much or more we should ourselves complain."

Or maybe, this one is better:

"The lady doth protest too much, methinks."

is things that aren't true:

"In the process, he developed a certainty about that which certainty is impossible and came to sneer at those whose pain was obviously a hundred times more sincere than his own."

What he was certain - and cantankerous - about - were ignorant statistical claims make by people who hadn't bothered to find out anything about the nature of polling in general and exit polling in particular.

Mercifully, I don't think he paid much attention to the kinds of scurrilous allegations you rightly reject. But I don't suppose they improved his temper.

No longer LBN and thread has gone off topic.