Mathematics, science, statistics and the taoist

[taoist]

This post is in response to the following:

davidgmills said, "So, as I have challenged the statisticians, I will challenge you. Show us your numbers. Talk (especially with scientists) is cheap."

__________

Scientists and statisticians deal with numbers. Mathematicians deal with underlying structures. The "numbers" I use most are homomorphic images of algebraic extensions of the rationals. These "numbers" are elements of the finite Galois fields. Hey, you asked.

Their applications, outside academia, are unfortunately quite restricted. Worse, those most interested in their applications are government agencies. I've no idea why the NSA would be interested in the possible existence of a projective plane of order 10, but I know they tossed millions into the problem. Not much to a physicist, perhaps, but in pure mathematics, that's a lot of money.

Formally, my "quals" were in number theory, abstract algebra, universal algebra and computer science.

__________

It's only by happenstance that I became familiar with "polling" problems, and even then, in a far different application. Consulting work shortly after my PhD led me to a position as research director for a Spanish-language television station in Chicago. A station owner had an interest in hiring a big gun to take down an evil polling agency. By necessity, I had to become an "expert" in the television industry's equivalent of E-M, Nielsen Media Research, the "TV ratings people."

Well, the results were mixed. Nielsen obviously has not been slain, but they do have a different president heading their Spanish-television division. It took four years to assemble sufficient evidence of bias to nail them, but the appropriate resignation was given less than a week after I publicized my analysis. The former president immediately moved into an executive position with our competing network for a final QED.

With the acquisition of my network by GE/NBC in 2002, I left the field, but retained the experience, and received a severance package sufficient to allow me to pursue other interests.

__________

As my chosen screen name suggests, I am far more interested in the religious divide in our country than in what has become an associated political divide. My politics are left of center, but that is only because our society has swung too far right, in my opinion.

__________

I am interested in any analysis that might show "nearly incontrovertible" evidence of fraud in the 2004 presidential election sufficient to change the results. The exit poll data so carefully analyzed by TIA says little more than what was common knowledge. Both Kerry and Bush knew that the early exit polls said Kerry would win in 2004. When contradictory vote totals began appearing, the exit polls were "revised" in order to bring them into alignment with the final tally.

There are many ways to read this "revision", but most of them involve that sinking feeling that the bad guys are going to get away with it. It's ridiculously simple to toss in enough fudge factors to realign a statistical model after you know which way it needs to be realigned. Add in the economic interests of the owners of the major networks represented by the NEP, and you can almost read the obvious outcome.

[taoist]

"This kills the vote fraud argument," said Mitofsky in Philadelphia on October 14, 2005, according to Kathy's "ESI-hypothesis-illogical" white paper.

My first question is of course whether this quote is correct. My second is why on earth Mitofsky should be interested in specifically debunking vote fraud. My third is to carefully examine the hypothesis behind this statement. Apparently, this chain of thought is natural.

I've read Kathy's paper now, and I think she missed the target by failing to critically examine the initial clause. In addition, as I have previously stated, the analysis inappropriately conflates logic and statistics. In particular, she fails to assign values from the appropriate set to her logical variables, assuming they must be either identically true of identically false.

Rather than dwelling on this latter point, as it would require a lengthy aside into paraconsistent logics, let me show the more natural semantic analysis of the suspect hypothesis and a better inference that avoids this error.

__________

"This kills the vote fraud argument."

"If systematic fraud or error in vote counting occurred in 2004 but not in 2000, Bush would have done significantly better in those precincts in 2004, and we would see larger differences between the reported vote and exit poll in those precincts."

__________

Assign as follows,

SystematicError2004 = Systematic fraud or error in vote counting occurred in 2004 favoring Bush.

SystematicError2000 = Systematic fraud or error in vote counting occurred in 2000 favoring Bush.

BushShareImproves = Bush's vote share at the precinct level increased in 2004 over the Bush's vote share at the precinct level in 2000.

BushDiscrepancyIncreases = The difference between Bush's vote share and exit poll share at the precinct level increased in 2004 over its value in 2000.

__________

I claim the final clause of Mitofsky's above statement should be read as follows:

BushShareImproves implies BushDiscrepancyIncreases

__________

And thus the full paragraph above reduces to:

SystematicError2004 and not SystematicError2000 implies BushDiscrepancyIncreases

__________

Or equivalently:

not BushDiscrepancyIncreases implies not SystematicError2004 or SystematicError2000

__________

As the counter-fraud quote can be rendered as:

"This" implies not SystematicError2004

__________

Where

This = not BushDiscrepancyIncreases implies not SystematicError2004 or SystematicError2000

__________

We thus see that the full statement, in context, reduces to the following:

not SystematicError2000

__________

Thus, in order for Mitofsky to conclude the vote fraud hypothesis is dead, he must show there was no systematic error in the Bush election of 2000. As far as I know, this has not even been attempted.

I claim in passing that my above argument is independent of value assignments to the assigned logical variables and is thus a valid semantic analysis of Mitofsky's statistical inferences.

Critiques and questions are welcome.

[stealermachine]

I think or *know* from experience there was sampling error some of it on purpose, but the WPD error line is fairly flat....

the only reason anyone would seek to distort this line, is to hide something.

and that is precisely what mitofsky did....I think warren mitofsky, being an unapologetic shill, has hidden plausible fraud from the masses and that he is using behavioral sciences to discredit his own exit polls.

and the reason? because NEP, and certainly economics play a part yet again....and if the exit polls are done away with, fraud becomes undetectable.....

now of course, I don't know exactly what level of understanding the so called "expert" mrs liddle has with this....I suspect she may not even realize this is a mitofsky regression line, and a mitofsky argument designed to cover for the real story....

but the fact they receive financial compensation from this source, and from mitofsky says all that needs to be said. they are paid to do something, whether they are aware the argument invalidates suspicion of fraud or not.....and she does it everywhere she goes....

there are other auditing methods including secondary voting, where you vote in a seperate area to compare your/their vote total to the county board office's total...personally I think that should be the new way to measure fraud or tricks especially after election day.

[taoist]

Mitofsky has a very large financial incentive to support his exit polls as a valid measuring device in relationship to the final vote tallies. The networks have a very large financial incentive to support the interests of their owners. The owners have a very large financial incentive to support the winners of the election. To support the exit polls as evidence of election fraud would be to precipitate a constitutional crisis. This is anathema to corporate interests.

Before Disney and GE went into the television business, there was some chance that our network news media would carefully examine election results in comparison to exit polls. This is no longer the case. I remember the time when television news divisions were not considered "profit centers" ... and I have seen first-hand how this change has effected reporting standards.

__________

"... but the fact they receive financial compensation from this source, and from mitofsky says all that needs to be said"

On your further point, I must object, on two grounds.

First, on logical grounds.

Because someone is paid does not imply they have been paid to lie. It does imply that one should examine their statements carefully, but that is all that it implies.

Second, on ethical grounds.

It is my understanding that Ms. Liddle is not allowed to directly rebut criticisms here. I cannot in good conscience cheer on the thrashing of a helpless enemy, independent of whether it is deserved. It seems to me that those who wish to question her motives are obliged to extend an invitation to her to directly defend herself.

Though Ms. Liddle has been in contact with me privately in an effort to protect her reputation, I am no more willing to defend her in proxy than I was willing to see my own words stand as proxy for her position. I've read her paper as well, but her absence from this board prevents me from saying any more about it. While her position with regard to Mitofsky invites criticism of her paper, criticism of her paper should be based on mathematical analysis alone.

_________

I repeat myself, yet again, there is no place for "expert" opinion in mathematical analysis. A theorem is not "more true" because it's presented by a credentialed mathematician. A calculation is not "more valid" because it's presented by a credentialed scientist.

A high school student is just as capable of presenting a paper to a journal as the most prestigious scientist or mathematician. If it presents novel information accurately, it will be published, following the normal course of peer review. I'm imagining a reviewer objecting to a paper because it was authored by someone lacking credentials. No, that's a fantasy. It just plain wouldn't happen.

Please, if we can't bury it entirely, let's put the "expert" argument to bed. It has no place in a discussion among grown-ups.

__________

Now, in a scientific setting, there is a further issue involved when examining an analysis. Are the data good? Normally, this can be checked by rerunning an experiment. Alternatively, if the experiment cannot be repeated, the raw data can be examined and independently analyzed. If the raw data are denied, however, be it in the form of laboratory notes or "proprietary" datasets, there is no reason to accept any analysis.

Mitofsky denies us the raw data needed to check for discrepancies in the exit polling. This is not unusual. I commonly came up against this barrier when dealing with Nielsen. I don't know how one might get around it in this instance, but I know how I managed it with Nielsen. I went around the client reps and made friends with the researchers directly. Eventually, I found a fellow traveler.

It worked for me.

[stealermachine]

"I went around the client reps and made friends with the researchers directly. Eventually, I found a fellow traveler.

It worked for me."

this seems to be another level of the problem. how might one actually do this and gain access to the "data" without cozying up to the corporate executives who "own" and justify that/their data?

this is exactly what happened with the year 2000 data from voter news service. and still to this day, they have objected to release all the precinct samples they claimed (with or without names redacted) so how could we know if they told the truth?

or if they lied?

"That does not mean one is paid to lie."

I never said mrs. liddle is technically paid to lie however, I said her arguments are based on mitofsky's ideas and his constructs. he is having her repeat the byline of these theories everywhere, indiscriminantly.

whether the person realizes it or not then, they are simply advancing coverup by repeating this company line and not providing their argument, a counter argument, and going forward until the precinct data is released.

this as I understand it goes on with exit polls, but other types of data analysis as well. the person does not need to have anything other than an agenda, and financial compensation or closely guarded financial "contributors" is more than enough to be suspicious of the arguments that come forward....which happen to be included in all of their press releases.

[davidgmills]

"There are many ways to read this "revision", but most of them involve that sinking feeling that the bad guys are going to get away with it. It's ridiculously simple to toss in enough fudge factors to realign a statistical model after you know which way it needs to be realigned. Add in the economic interests of the owners of the major networks represented by the NEP, and you can almost read the obvious outcome."

For those of us that want our country back, proof of this inference means a great deal.

It is my feeling that each and every American who has talent of any kind needs to do everything he can to ensure that we have valid elections. Without them this democracy will not stand.

I brought my own lawsuit to attempt to bring back paper ballots in my state, to get partisan corporations out of the voting process in my state, and to get polling duty instituted to help out overworked voting officials.

I would implore you to use your math skills in any way you can to help ensure the validity of our election process.

[stealermachine]

taoist, we can't just sit and debate a tv network's motives. everyone is pretty clear they are corrupt, and lie or twist the truth for their 'own' reasons.


that doesn't help us to debate exit poll owners or executives.

the point is, if they tried to hide the fraud we need "real mathematical analysis done on the miscounts" and to uncover every facet of the fraud that one can do.

this is what I mean...

http://www.bbvdocs.org/FL/volusia/Vo...2-polltape.pdf

these are polling tape discrepancies, areas all around florida that had vote miscounts or more than one "memory card reprogrammed" during more than just the general election.

mathmeticians, need to match up their analysis against the very discrepancies that have been found, in a severe and resolute fashion that the likes of NEDA have come up with. and they need to confirm how many miscounted votes there were, and take it as far as it can possibly go.

we're dealing with the integrity of our elections system, and that is all we should be concerned with. or talking about.

we shouldn't be concerned with the exit poll owners, corporate executives, or their "debunking". we should be concerned with how many votes were "miscounted", in the year 2004, 2002, and 2000. david who speaks as an attorney with experience, has the idea.

[taoist]

The only way to get raw data is to get close to someone with physical access to the raw data. How would I go about it in this case? Well, I can certainly think of ways, but I'm not likely to do so publicly ... still, there's an obvious link ...

__________

The real issue when you're considering outside analysis of proprietary data is whether the data released to the outsider has been "handled" ... I remember a similar situation when I was engaged in debate with a fundy homeschooler.

Understand that I am adamantly opposed to homeschooling for the purpose of dissociating children from exposure to scientific theories such as evolution and modern geology that tend to contradict the religious views of young earth creationists. Indoctrinating children with beliefs in a world wide flood 4500 years ago and an earth less than 10,000 years old is indistinguishable from child abuse in our modern world.

I was linked to a study that tried to claim fundy homeschoolers scored in the 80th percentile in comparison to the general population.

Turned out the study was put together by Bob Jones University. Are the bells ringing yet? To prevent allegations of bias, they brought in a third-party analyst. Follow the links, follow the links, examine the methodology and what do you find? Right there, bold as brass, in the web-published study:

[link:epaa.asu.edu/epaa/v7n8/|Scholastic Achievement and Demographic Characteristics of Home School Students in 1998]

"After we formed the dataset with 20,760 students, we asked for the remainder of the 39,607 achievement test scores. We were informed that it would not be possible to disaggregate the remaining home school students from students in private schools also contracting testing services."

Translation: BJU cherry-picked the data for analysis, dumping half the data in the process.

How much do you want to bet that only the top half of the scores were presented to the analyst for review? I think my point is clear. There is no point in acknowledging an analysis based on data which is otherwise undisclosed. I recommend this protocol.

___________

It could be that third-party analyses commissioned by Mitofsky were based on unbiased data. But an honest reviewer cannot begin with that assumption.

[wli]

Probably the quickest way to get the gist of what's happening is to look at the z-scores. I've run the numbers and they were much worse than this map. I'm not sure what was used to generate this map, but I suspect they were using some sort of sample variance calculation that overestimated the variance.

The shades of red (or pink or orange) represent Bush-ward redshift. The diagonal lines on a white background indicate either blueshift or non-bias (my numbers on redshift vs. blueshift agreed with this map IIRC).

[http://www.edwardsdavid.com/images/p...t_cont-01.jpg]

From there move on to the rest of [http://www.truthisall.net/]. The raw data is floating around there in various forms (usually in conjunction with reluctant Bush responders etc. modelling data, though some earlier archived posts/etc. show it in standalone format).

[mhkeefer]

It seems to me that there's a disjunction between statistical/mathematical and empirical studies of the evidence for electoral fraud. Do you see any way of crossing this divide?

Let me explain what I mean. In the (in my opinion) generally contemptible DNC report on the 2004 elelction in Ohio, Walter Mebane offered a very interesting study of the consequences of shorting the numbers of voting machines in Franklin County. His statistical analysis concluded that in predominantly African-American precincts the result was (conservatively interpreted) a four percent reduction in the Kerry vote. Results of this kind can be quanitified.

Is it possible to produce similar results from such evidence as (a) Werner Lange's evidence of 'ghost-voting fraud in 106 precinct poll-books in Trumbull County, Ohio; (b) James Q. Jacobs' (and others') studies of 'precinct cross-voting' in Ohio's Cuyahoga County; (c) the combined evidence of vote-tally fraud in many Ohio counties and of disparities between presidential ticket and down-ballot results? One can think of other bodies of empirical evidence, in this and other states, that might lend themselves to quantititaive analysis.

Statistical/mathematical work seems for the most part to have been confined to comparisons of exit-poll/vote tally data. To what extent do you think other data permits serious quantitative analysis?

[taoist]

The inference I make is political, not mathematical.

There is an economic incentive to those corporations which own the networks that form the NEP to avoid the implied constitutional crisis that would result from a successful challenge to the results of 2004. There would normally be an incentive for Mitofsky to back the accuracy of his numbers. I cannot read his refusal to do so as anything other than a reflection of the interests of his clients.

There would normally be an incentive for the networks to hold E-M financially responsible for the errors in their exit polls. I estimate the total bill for the exit polling for election 2004 as in the range of $50 million. If they are not pushing to recover this bad investment, it can only be because they estimate indirect costs to be greater than that which could be recovered.

It is a mistake to consider their actions as "evil" ... a proper analysis must begin by putting such judgments aside and following the money. These people can be had. All it requires is knowledge of the levers. Find a way to make the costs balance out on your side and you can move them.

__________

The "revision" of which I speak was the adjustment to the exit-polling data that resulted in the unbiased data being weighted to conform to recorded vote tallies. First, I need to be clear that I have not checked the accuracy of these claims and am only proceeding on the assumption that TIA has properly presented the shift in exit-poll projections.

A real analysis must begin by checking these sources. (Oh, I should mention that I've received independent corroboration of the Mitofsky quote I cited in my [link:progressiveindependent.com/dc/dcboard.php?az=show_mesg&forum=120&topic_id=1980&mesg_id=1983|first response] above from Liddle.)

ETA: Liddle informs me that she cannot independently confirm the quotation in Philadelphia. Rather she confirms that the same Mitofsky statement was made according to a private email.

I've seen how demographic weighting works. Nielsen typically weights on over a dozen factors. These factors are available to station subscribers and it was a standard part of my book breakouts to check the factors for consistency.

The argument that unweighted exit poll data is more accurate than weighted exit poll data is fallacious. While weighting increases the margin of error, it is easily demonstrated that it increases the accuracy of the projection. The first step in analysis of the exit poll "revision" must be to beg, borrow, steal or legally enjoin E-M to release the raw data and weighting factors used in Ohio 2004 in order to independently reproduce their exit poll projections.

We know the rBr hypothesis is both artificial and post hoc. It was developed in an attempt to rationalize the discrepancies, and from what I've seen, it does so successfully, in the sense that it brings the results into range of the MoE. Again, without reference to the raw data, we've no way to tell if we're dealing with GIGO errors.

I've suggested another analysis of the rBr hypothesis and discussed the necessary data I'd require with Liddle. I'm told this data is also unavailable. I am not satisifed with this answer. It's clear, however, that she does have access to data which she is willing to release. I believe it's a strategic error to demonize her on this site.

__________

In principle, electronic voting machines should be able to minimize voting errors. But they require an independent audit trail. I am sorry to say that I can not agree with your proposed solution of bringing back paper ballots. I assume you argue that they were more accurate than what we have now in the electronic machines.

On the other hand, partisan corporations in the voting process are a fine target. I hope you have not neglected to consider mass media outlets as well.

__________

On a last practical note, I hope the participants in this process have not neglected the need to properly publicize their issue. If you do not have a committee entrusted with formulating your public relations and press coordination, you should.

I'd like to point out that the "Intelligent Design" movement was organized and run by a group of less than a dozen participants. With minimal infrastructure and a tight message, they managed to create a national juggernaut.

If the fundies can do it ...

[stealermachine]

http://www.progressiveindependent.co...0&mesg_id=1992

[TruthIsAll]

In April 2005, I challenged interested parties to refute the following
odds based on Z-scores for exit poll vote discrepancies.
So far, no one has.
-----------------------------------------------------------

http://www.democraticunderground.com...ess=203x361819

This is for the Mystery Pollster and other interested parties.

Go through the analysis of probabilities in this post based on Z-Scores
of state exit poll discrepancies.
Red Shift: Z-scores and Probabilities. Take a deep breath.
Edited on Sat Apr-23-05 03:26 AM
http://www.democraticunderground.com...mesg_id=361617

Go here for state Z-scores and corresponding probabilities:
http://www.democraticunderground.com...ess=203x361736

For Bush, Z=1 (standard deviations from the mean) was exceeded in 35 states.
For Kerry, Z=1 stdev was exceeded in FIVE states.

For Bush, Z = 2.20 was exceeded in 16 states.
For Kerry, Z = 2.20 was exceeded in ZERO states.

I have computed probabilities of exceeding both Z-score thresholds.
The standard deviations are based on the formula:
StDev = sqrt (KP*BP/n)
KP = Kerry 2-party exit poll percentage.
BP = 1-KP = Bush exit poll percentage
n = state exit poll sample size

If you can refute these results, please do so.
I look forward to all replies, whether at DU or elsewhere.
________________________________________________________

Prior analysis has focused on the probability that 16 states
would exceed the margin of error for Bush, using a 97.5%
(one-tail) confidence level.

This analysis focuses on the Z-score, which is the number of
standard deviations that the actual vote is displaced from
the polling sample mean, based on the normal distribution.

The Z-score exceeded one (1) standard deviation in 35 states for Bush,
at the 84.13% one-tail confidence level per state.
Z exceeded one in 5 states for Kerry.

Calculating the probabilities for each occurrence:
Prob (Bush Z >1 in 35 states): 1 in 4500 trillion;
Prob (Kerry Z >1 in 5 states): 91%

To calculate the probability of the state exit poll discrepancy
based on the Z-score, we use the Excel function:
P = 1-NORMSDIST(Z),
where Z = the number of standard deviations, as above.

To calculate the probability of the Z-score exceeding 1 in at
least N states, we use the Excel function:
Pr= Prob(N>Z) = 1 - BINOMDIST(N-1, 50, P, TRUE)

Probability of N states exceeding actual Z-score thresholds:
Favor; N; Z; Prob; Pr(N>=Z)
Bush; 35; 1.00; 1 in 6; 1/4500 trillion
Kerry; 5; 1.00; 1 in 6; 90.9%
Bush; 16; 2.20; 1/72; 1/ 160,000 trillion
Bush; 10; 2.60; 1/215; 1/ 23 trillion
Bush; 4; 3.35; 1/2475; 1/ 165 million
Bush; 3; 3.75; 1/11307; 1/ 74 million
Bush; 2; 4.10; 1/48382; 1/ 1.9 million

Using one tail confidence levels of 97.5% and 99.5%:
Conf Lev; Favor; N; Z; Prob; Pr(N>=Z)
97.5%; Bush; 16; 1.96; 1/40; 1/ 19 trillion
99.5%; Bush; 10; 2.58; 1/202; 1/ 13 trillion

________________________________________________________

[davidgmills]

First I made a post here

http://www.progressiveindependent.co...2&mesg_id=1987

that you have may not have seen yet. I would be interested in your take on it.

Secondly, paper ballots. Theoretically, as you say, DRE's should produce much more accurate voting results. However, dre's are subject to many programming errors and intentional hacking. Several members of my voting rights group are computer software people and they despise the things, because the possibility for undetectible error or fraud is just too great, especially when source codes are trade secrets and software are trade secrets. If elections are to be open and transparent, you can not do it with trade secrets. The two are simply incompatible concepts.

So what my group proposes as a trade off is pre-printed paper ballots which can be scanned by optiscanners. At least these ballots can be counted by hand, can be checked by random statistical counts, and can be used in court for election contests. A dre makes an election contest a legal farce. If you have a thousand of them, do you bring them all to court? How do you verify the results when you face trade secret defenses? Even if you have no trade secrets, you still may have a battle between experts that a jury can not undertand. Much simpler for election contests to have paper ballots that can be scanned and hand counted. Election contests are over in days not weeks or years.

Third. Getting Mytofsky's data by court proceeding. What a legal nightmare. It is his legal property. Could be more difficult than trying to get his house. Plus, the data was collected for his clients and under his contracts with them, he may be obligated not to share the data. I believe that his privacy arguments are legally bogus. But to get his property and to set aside any contracts which require him to keep his data private, those are monumental legal hurdles to overcome.

Which is why I mentioned the census in the thread I cited above. Much more easy to get that data. Freedom of Information request. But of course, the census data is missing one critical question: "Who did you vote for?"


On edit.
Here is something I just thought of that I have never thought before or seen in print. For the future, how about a consortium of about 100-200 colleges and universities to do exit polls for elections? No problem with getting their data if all members of the consortium contracually agree to make their data public. What do you think? Is such an idea a practicality?

Even a serious threat of doing something like this might make Mytofsky give up his data.

[davidgmills]

First I made a post here

http://www.progressiveindependent.co...2&mesg_id=1987

that you have may not have seen yet. I would be interested in your take on it.

Secondly, paper ballots. Theoretically, as you say, DRE's should produce much more accurate voting results. However, dre's are subject to many programming errors and intentional hacking. Several members of my voting rights group are computer software people and they despise the things, because the possibility for undetectible error or fraud is just too great, especially when source codes are trade secrets and software are trade secrets. If elections are to be open and transparent, you can not do it with trade secrets. The two are simply incompatible concepts.

So what my group proposes as a trade off is pre-printed paper ballots which can be scanned by optiscanners. At least these ballots can be counted by hand, can be checked by random statistical counts, and can be used in court for election contests. A dre makes an election contest a legal farce. If you have a thousand of them, do you bring them all to court? How do you verify the results when you face trade secret defenses? Even if you have no trade secrets, you still may have a battle between experts that a jury can not undertand. Much simpler for election contests to have paper ballots that can be scanned and hand counted. Election contests are over in days not weeks or years.

Third. Getting Mytofsky's data by court proceeding. What a legal nightmare. It is his legal property. Could be more difficult than trying to get his house. Plus, the data was collected for his clients and under his contracts with them, he may be obligated not to share the data. I believe that his privacy arguments are legally bogus. But to get his property and to set aside any contracts which require him to keep his data private, those are monumental legal hurdles to overcome.

Which is why I mentioned the census in the thread I cited above. Much more easy to get that data. Freedom of Information request. But of course, the census data is missing one critical question: "Who did you vote for?"


On edit.
Here is something I just thought of that I have never thought before or seen in print. For the future, how about a consortium of about 100-200 colleges and universities to do exit polls for elections? No problem with getting their data if all members of the consortium contracually agree to make their data public. What do you think? Is such an idea a practicality?

Even a serious threat of doing something like this might make Mytofsky give up his data.

[TruthIsAll]

Loud enough to get me banned, in fact.
I got kicked off DU immediately - during the "Game".

You MUST read the "Game" thread.
This is the one which clinched it.
http://www.democraticunderground.com...ess=203x390193

The "Game" thread was one that we were advocating for months on DU.
To no avail.
OTOH and Febble had scores of verbal posts, not one numerical.

Finally, after 4 months, OTOH and Febble were forced to "put up or shut up".
They had to show us a plausible Bush win scenario.
OTOH did back flips and somersaults to come up with one.
Of course, as expected, it was NOT CLOSE to being plausible.
It surely appealed to faith-healers and intelligent designers who represent the Bush base.

This is the post in the "Game" in which I referred to a "fundie".
Febble objected to this reference to his "religious beliefs".
As if my use of the term "fundie" implied religious bigotry.
Classic character assassination.
In true RW fashion, she cast this UNFORGIVABLE ASPERSION.
She ran to the mods.

It was a Golden Opportunity to get rid of me once and for all.
Febble revealed herself right there.
So it was bye-bye TIA.

But she had no idea, nor did I, that I would resurface at PI.
She forgets that her words are etched in stone and will cry out forever.
The rest, as they say, is history.

Revenge is a dish best served cold.

________________________________________________________

An observation from a very biased poster on this thread..
Edited on Wed Aug-24-05 09:58 AM by TruthIsAll

Posters on this thread apparently fall into one of two groups:

1) Fraudsters
2) Naysayers

Group I: 12 individuals
DU members who have worked diligently since Nov. 2, using publicly available but media-ignored data related to Election 2004. Each member has analyzed the data using his/her particular area of expertise.

Group II: 2 individuals
Both worked at USCV and left early last year, immediately proceeding to bash the statistical-based study they helped write which debunked the rBr hypothesis. At least one individual immediately engaged in a "consulting" assignment for Mitofsky, the originator of rBr. Both are currently collaborating with a well-known fundie, banned from DU, who throws stones at any argument pertaining to fraud. Each of them often refers to MP as an polling expert, although his "expertise" appears to derive from his claim of being an unbiased Democrat. Each has apparently deserted the discredited rBr hypothesis and cite as the basis of their case an NES study in order to promote a new theory to explain the exit poll discrepancies - the Forgetful Gore Responder (fGr). To justify the numbers, they have made assumptions regarding Gore/Bush 2000 voter percentages in 2004. The percentage deviations from the Final Exit Poll percentages are nearly triple the margin of error for this voting characteristic, considering that it is a 10/90% split.

Both groups are obviously biased.
___________________________________

This is Febble's reply.
I point you to the last paragraph.
______________________________________________________________


Febble (1000+ posts) Wed Aug-24-05 10:00 AM
Response to Reply #171

177. Two small corrections:
Edited on Wed Aug-24-05 10:51 AM by Febble

First:
Re Mystery Pollster, aka Mark Blumenthal, is a partner in the polling firm Bennett, Petts and Blumenthal.

http://mysterypollster.typepad.com/about.html

A person who is a partner in a polling firm is generally considered to have polling expertise.

Second: I don't think either OTOH or I have "deserted rBr", although rBr is not my favored description of the hypothesis. My own view is that some form of sampling/response bias was probably a substantial contributor to the exit poll discrepancy, although I do not rule out a contribution from fraud.

The theory put forward in this thread by OTOH (and by myself, earlier) is simply a hypothesis regarding a mechanism by which the "final table" with its apparently "impossible" Gore/Bush proportions could be reconciled with a red shift that was largely due to sampling/response bias.

Otherwise, for myself, I would confirm your timeline, although not the implied causal sequence. And because I dispute the causal sequence I dispute the allegation of bias. I changed my views because I was unbiased (i.e. not committed to a fraud hypothesis come-what-may) not because I was biased. And certainly not because I did work for Mitofsky. The timeline was, as you correctly report, the other way round. My paper resulted in my leaving USCV; my paper subsequently induced Mitofsky firstly to reanalyse the data, and later to commission further analyses from me.

So yes, in your words (elsewhere), I am a NAYSAYER to the case that the exit poll evidence is unambiguous evidence of massive electoral fraud. However I am not by any means a naysayer to efforts to investigate the inequities and illegalities of the 2004 election or to bring about fair, secure, auditable elections in the US. It is why I got involved in the first place.

On edit: just noticed your reference to the religious beliefs of one of my colleagues. This is quite irrelevant and quite unacceptable. I suggest you edit this if there is time.
________________________________________________________________

Taoist, don't worry.
Febble can't get you banned here.





_______________________________________

[TruthIsAll]

Loud enough to get me banned, in fact.
I got kicked off DU immediately - during the "Game".

You MUST read the "Game" thread.
This is the one which clinched it.
http://www.democraticunderground.com...ess=203x390193

The "Game" thread was one that we were advocating for months on DU.
To no avail.
OTOH and Febble had scores of verbal posts, not one numerical.

Finally, after 4 months, OTOH and Febble were forced to "put up or shut up".
They had to show us a plausible Bush win scenario.
OTOH did back flips and somersaults to come up with one.
Of course, as expected, it was NOT CLOSE to being plausible.
It surely appealed to faith-healers and intelligent designers who represent the Bush base.

This is the post in the "Game" in which I referred to a "fundie".
Febble objected to this reference to his "religious beliefs".
As if my use of the term "fundie" implied religious bigotry.
Classic character assassination.
In true RW fashion, she cast this UNFORGIVABLE ASPERSION.
She ran to the mods.

It was a Golden Opportunity to get rid of me once and for all.
Febble revealed herself right there.
So it was bye-bye TIA.

But she had no idea, nor did I, that I would resurface at PI.
She forgets that her words are etched in stone and will cry out forever.
The rest, as they say, is history.

Revenge is a dish best served cold.

________________________________________________________

An observation from a very biased poster on this thread..
Edited on Wed Aug-24-05 09:58 AM by TruthIsAll

Posters on this thread apparently fall into one of two groups:

1) Fraudsters
2) Naysayers

Group I: 12 individuals
DU members who have worked diligently since Nov. 2, using publicly available but media-ignored data related to Election 2004. Each member has analyzed the data using his/her particular area of expertise.

Group II: 2 individuals
Both worked at USCV and left early last year, immediately proceeding to bash the statistical-based study they helped write which debunked the rBr hypothesis. At least one individual immediately engaged in a "consulting" assignment for Mitofsky, the originator of rBr. Both are currently collaborating with a well-known fundie, banned from DU, who throws stones at any argument pertaining to fraud. Each of them often refers to MP as an polling expert, although his "expertise" appears to derive from his claim of being an unbiased Democrat. Each has apparently deserted the discredited rBr hypothesis and cite as the basis of their case an NES study in order to promote a new theory to explain the exit poll discrepancies - the Forgetful Gore Responder (fGr). To justify the numbers, they have made assumptions regarding Gore/Bush 2000 voter percentages in 2004. The percentage deviations from the Final Exit Poll percentages are nearly triple the margin of error for this voting characteristic, considering that it is a 10/90% split.

Both groups are obviously biased.
___________________________________

This is Febble's reply.
I point you to the last paragraph.
______________________________________________________________


Febble (1000+ posts) Wed Aug-24-05 10:00 AM
Response to Reply #171

177. Two small corrections:
Edited on Wed Aug-24-05 10:51 AM by Febble

First:
Re Mystery Pollster, aka Mark Blumenthal, is a partner in the polling firm Bennett, Petts and Blumenthal.

http://mysterypollster.typepad.com/about.html

A person who is a partner in a polling firm is generally considered to have polling expertise.

Second: I don't think either OTOH or I have "deserted rBr", although rBr is not my favored description of the hypothesis. My own view is that some form of sampling/response bias was probably a substantial contributor to the exit poll discrepancy, although I do not rule out a contribution from fraud.

The theory put forward in this thread by OTOH (and by myself, earlier) is simply a hypothesis regarding a mechanism by which the "final table" with its apparently "impossible" Gore/Bush proportions could be reconciled with a red shift that was largely due to sampling/response bias.

Otherwise, for myself, I would confirm your timeline, although not the implied causal sequence. And because I dispute the causal sequence I dispute the allegation of bias. I changed my views because I was unbiased (i.e. not committed to a fraud hypothesis come-what-may) not because I was biased. And certainly not because I did work for Mitofsky. The timeline was, as you correctly report, the other way round. My paper resulted in my leaving USCV; my paper subsequently induced Mitofsky firstly to reanalyse the data, and later to commission further analyses from me.

So yes, in your words (elsewhere), I am a NAYSAYER to the case that the exit poll evidence is unambiguous evidence of massive electoral fraud. However I am not by any means a naysayer to efforts to investigate the inequities and illegalities of the 2004 election or to bring about fair, secure, auditable elections in the US. It is why I got involved in the first place.

On edit: just noticed your reference to the religious beliefs of one of my colleagues. This is quite irrelevant and quite unacceptable. I suggest you edit this if there is time.
________________________________________________________________

Taoist, don't worry.
Febble can't get you banned here.





_______________________________________

[Username]

"For the future, how about a consortium of about 100-200 colleges and universities to do exit polls for elections? No problem with getting their data if all members of the consortium contracually agree to make their data public. What do you think? Is such an idea a practicality?"

This might need it's own thread here in GD. I can't think of why you can't hash something out very quickly. I'm thinking 1000-2000 or 10,000-20,000 schools. Is it possible to collect too much data? My initial reaction is to get moving on this right away! How can I help?


<edit> for taoist

[link:www.progressiveindependent.com/dc/dcboard.php?az=show_topics&forum=101|the link to the bar & grill, in case you need it, navigating new boards and all]

[davidgmills]

I was thinking of a couple of major schools per state but the more the merrier.

I really have no idea of what is possible. I just think of college level academia as the one place in the US right now that could do this project for the public good and be willing to make all necessary disclosures for adequate peer review. I also think that such a project would scare the beejesus out of the private polling establishment. It might scare them enough to do right.

I could be totally wrong about this idea. Perhaps conservative institutions and liberal institutions would fight so bitterly over it that it would not work. I have no idea.

But I think the project could be called something like the Exit Polling Integrity Project and concentrate on being a check on two things: a check on the final vote count and a check to see how many people go to the polls and give up or get turned away. It should not be used to project early winners. It should be used and designed to validate the integrity of elections and to ensure that those entitled to vote get to vote.

[davidgmills]

What if the consortium data indicated massive fraud. Would the corporate media publish their findings?