From Elizabeth Liddle ("Febble"): a statement

[Febble]

There has been widespread misunderstanding, some fair criticism, and some unfair criticism of my work here on DU.

I won’t be hanging around here much in future (you may be happy to know), but this is an attempt to put the record straight and my paper in context before I go.

1. Bias

I cannot claim to be free from “bias”. No scientist can. It is the reason we use double-blind studies when we want to find out whether a drug really works. So I will come clean. I was biased. I was biased in favour of a “fraud” explanation for the discrepancy between the early exit poll projections of a Kerry win, and the final count. To me, the evidence of voter suppression in Ohio, and the obstruction of the Ohio made this explanation only too likely. I entered the investigation hoping to find evidence that the exit poll discrepancy was due to fraud. So yes, I was, and am, biased. I have therefore made every effort to ensure that no mathematical technique I have used in my exploration of the data is asymmetrical.

2. Language

I am aware that I have used terminology that has given people the impression that I have “framed” the question on the “assumption” that the cause of discrepancy lies in the polls not the count. I have attempted a number of alternative usages, some of which have still met with criticism. Getting language politically correct is clearly a minefield on DU and I will make only two points here. One is the use of the word “error”. Error is a statistical term with a precise technical meaning. It means unexplained variance. Use of the term involves absolutely no assumptions about the explanation – it simply means it is unexplained. So “Within-precinct error” means the unexplained differences between the poll estimate (from the responses) and the vote count. It is known as WPE to its friends. I am not one of its friends, and the major point of my paper is that it is poor measure. It over-estimates the discrepancy in evenly balanced precincts, and underestimates it in precincts in which the vote-count margin (for whatever reason) is greater. Use of the WPE in any analysis therefore potentially invalidates the analysis, whether it is by E-M or USCV. I therefore devised a fairly simple algebraic transformation of the WPE that removes its “confound” with vote-count margin. I called the resulting measure a “bias index” as it allows true “bias” to be quantified. It makes absolutely no assumptions as to whether the bias is biased polling, or the bias is biased counting. The bias index will simply measure “bias.

3. rBr

My paper does not support rBr over vote count corruption. I have simply demonstrated that a very specific refutation of rBr by USCV is probably not justified. Other refutations remain possible. USCV claimed the pattern of data was “inconsistent” with rBr – because, they claimed, it showed that WPEs were greatest in “high Bush” precincts, which does not jive with the rBr hypothesis. Why would Bush supporters be more “reluctant” in their own strongholds? It was an interesting point. I was tempted by it. However, because of the WPE problem noted above, I dug further, and demonstrated that the pattern is not necessarily inconsistent with widespread differential non-response (i.e. rBr) across all precinct categories. It is, instead, consistent. Consistent does not mean proven. And more importantly, for your purposes, the pattern is also consistent with widespread fraud, although again not proven. I make this point in my final paragraph:

The pattern instead is consistent with the E-M hypothesis of “reluctant Bush responders”, provided we postulate a large degree of variance in the degree and direction of bias across precinct types. Mathematically, the observed pattern could arise from widespread fraud as well as from widespread response bias; differential vote spoilage rates for Kerry votes, or “ballot stuffing” of Bush votes, would produce results indistinguishable from “reluctant Bush responders.” However, this is not the inference currently drawn from the data by USCV in their report.

In other words, I have simply moved the argument back to where we were – that the “bias” in the polls was widespread, and not concentrated, as USCV claim, in Bush strongholds. So that particular refutation of rBr does not hold water. The irony is (if you do irony on DU) that it helps the fraud case. I am saying that whatever was happening in those Bush strongholds was happening everywhere.


4 What next?

I believe E-M need to some more homework. I think they need to apply my algebraic transform to their WPE values and re-do their analyses. I think they need to use multiple regression to find out how much variance in “bias” can be accounted for by sampling protocol variables, and how much “error” (unexplained variance) remains. I think specific hypotheses (that “error” should be greater in swing states, for example) need to be formulated, and tested. I have been working with USCV to develop precisely that kind of hypothesis – hypotheses as to what the fingerprint of “fraud” might look like.

But to be honest – and I am honest – my hope is fading. I do not think we are going to find evidence of fraud in the exit poll data. This does not mean that I think no fraud occurred. It does not even mean I believe that that fraud was not a factor in the exit poll discrepancy. But it is hard to find effects in very noisy data, and one thing that we do know from the E-M report is that it is very noisy data. And the fact that the data is so noisy is starting to convince me that polls are a lot more fallible than I once believed, and than many on DU still believe.


5. Some final thoughts (bias?):


Perhaps the fairest criticism of my work has been the charge that right or wrong, it will be misunderstood and used to argue against investigation of the problem. I have some sympathy with this view, especially having seen how easily it has been misunderstood here. Perhaps you are right, perhaps what is required is not the kind of dispassionate analysis I believe I have carried out, but rather the propagation of any evidence that can be used in favour of the cause, and suppression of any evidence against. Maybe that sounds bitter. Maybe it is.

But I have never believed that any caused is best served by disinformation. In fact I think a small tragedy is that attention from excellent case for investigation into the blatant mismanagement of the election in Ohio by a Secretary of State for Elections who was blatantly partisan and who had a blatant conflict of interest (as did the supplier of the election equipment), may have been fatally diverted by the conviction that the exit polls contained the key to the treasure.

I still think it is possible that if the election had been fair, Kerry might have won Ohio, and the electoral college vote. Far more Kerry voters than Bush voters were disenfranchised in Ohio,

http://electionarchive.org/index.php?option=com_content&task=view&id=65&Itemid=43

and although that number is impossible to quantify reliably, maybe it swung the election. And there was no recount in Ohio in any meaningful sense of the term.

But I think the evidence is mounting that Kerry did not win the popular vote, and that the Great Exit Poll Discrepancy was largely due to poor random sampling. I realise most on this forum disagree. I can only say that I have not come to this view through bias. I was massively biased in favour of finding the opposite.

Thanks for having me. I share your pain.

Lizzie

(Elizabeth Liddle)

[Sancho]

1. Fair enough, but....

I understand your statistical model, but not your assertion that Bush won (final statement). I don't see evidence to validate the election:

First, the WPE does not explain, to me, the strange inconsistencies (correlations) in individual races. i.e.; here in Florida, Senate races etc. were sometimes at odds with the Presidential race with strange differences in the number of votes cast for one election and not the other...with raw data from E-M, this may be evidence of fraud. If poll data reveal that 99% of the presidential voters also voted for a senator, but the actual vote is that only 90% of presidential voters voted for a senator...that's an issue, but we don't have "precinct" data from the polls as far as I can tell (but E-M knows).

Second, TIA makes a number of good points that many seem to avoid...how does one explain the results that more voted for Bush "in 2000" than really exist, or that the polls made magical changes at midnight? Scientists should look carefully at these things as threats to the validity of the poll process - at least once the manipulations appear to occur - so stronger demands that E-M produce raw data and an explanation of this issue would be more useful than a bias index for error variance.

Third, a careful look at the surveys on line reveal that with the raw data, "correlation, regression, etc" where the unit of analysis is the individual person, should reveal any weird profiles (RBR's etc) that are not possible with aggregated data as currently posted by E-M; a "bias index" becomes irrelevant if the raw poll data is available. My experience is that E-M's explanation is fabricated because they should KNOW better what happened, and E-M offers us a rather poor explanation. How strongly have you demanded such an investigation occur? I'd like to see how "swing" voters in the polls answered questions about themselves besides who they voted for...

[Chomp]

2. It seems to me that...

...the main reason why you have been "misunderstood" by some/many on DU is that very many people who are not statisticians or mathematicians respond to your technical posts without the first clue what it is they/you are talking about.

I say to these people - how dare you criticise the work of a scientist without understanding that science? It is shocking, stupid, illinformed, blind and utterly illiberal to do so.

I am not referring to DU's matemeticians of course - these people have a genuine right and responsibility to call it as they see it. But people who don't understand the specific use of the word "error"? People who don't understand "noise"? People who cant tell WPE from PMT? All these people should butt out when the going gets technical.

Finally, some people want to believe so badly that they cannot think straight. There's no mathematical formula that legislates for that.

(Disclosure: I do not understand stats and maths at all.)

[tommcintyre]

6. That's an elitist attitude that just don't fly around here.

You'd be surprised how many people are pretty good at using the "original equipment" (their brain) that they were issued. That's why they're here. They don't believe in "absolutist" ideas like creationism, fundamentalism, and yes, even in forgoing questioning inferred statistical outcomes.

The people around here are pretty savvy with quite a few acronyms. But the one they seem to be particularly adept at seeing through is the "BS". ;)

[FogerRox]

66. Right on Brother--BS it is-- Febble has made comments (erroneous)

that were out of her field of experise--this would not be tolerated

by my dad--retired Math PHD---advisor to dozens of doctorial candidates over the last 3 -- 4 decades.

Most us know the difference 'tween touch pads and Touchscreens---20 years between the 2---------

Most of us know that Ohio had 4 county's w/DRE"s so throwing out the "Touchscreen precincts in Ohio" MAkes me think this gal dont know all that much--and she tends to talk about what she dont know. Or posts up links to things that have nothing to do with her own posts & Comments.

For example--here
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=365850&mesg_id=366335

I still cant get a opps wrong link --or sorry I didnt read the article before posting or any retrenchemnet of any kind.

If you are not that carefull here--than how good is your paper and the work behind it? If your "work here"--as you said---aint up to snuff, then ------------------

[Melissa G]

3. Thanks for your explanatory post. n/t

[tommcintyre]

4. Bias vs. "The Elephant in the room" (excuse the pun) ;)

Thank you for acknowledging the potential for bias when subjective evidence (inferred only) is all that's available in an attempt to evaluate the situation. We are ALL subject to bias, and I propose that it is responsible for much of the pain, suffering, conflict and misunderstanding among people. (A look at the number of critical thinking books testifies to the difficulty that we humans seem to have in just thinking clearly (objectively).

However, I think you miss the point on the possible origins of your possible bias (probable, when ALL the evidence is looked at as a whole). Statistics is a WONDERFUL tool, and, of course, can often even help us avoid bias in our thinking. However, like all tools (no matter how powerful), it still has it's limitations (especially when only inferred information is available).

You see, dear Febble, it doesn't matter what your original bias was. You can be the strongest Kerry supporter there ever was. Still, you must conclude the probability that bias has contaminated your efforts. This is because of the overwhelming amount, and the relatively "harder" (more direct, and less inferred), evidence that TIA and others have uncovered and presented.

So, this becomes "the elephant in the room". ALL the inferred data in the world will NOT overcome the fact that the "elephant" is there. You may do massive inferred permutations that, in theory, the elephant simply can not be where it is - in the room. But, the data notwithstanding, he is simply "there". ;)

So, at that point, you must conclude there is something wrong with the inferred process that was used to arrive at the conclusion that the elephant is not there.

A suggestion. Why not read all the links TIA gave you (and consider all the other information you have been exposed to here), and revisit your premise, processes and conclusions, and ask the question from a different "angle". Why is there such an overwhelming body of evidence that indicates the "elephant" (election fraud), while my inferred data is favoring the rBr hypothesis? Who knows, if you do it, you just might win the Nobel Peace Prize. ;)

Good luck Lizzie!

[Febble]

17. Thanks for the response

I have followed many of TIAs links, and have responded to some of his comments.

He raises some good points, which need to be addressed. The senate races are odd-ish, as is the "rGr", though neither are inexplicable IMO. And there is plenty of anecdotal evidence of machines that seemed to favour Bush. All this is legitimate evidence, and needs to be examined carefully and impartially. However, I believe that TIA has misunderstood at least one aspect of the numbers. In other words, I am not convinced that all he has put forward is indeed "evidence" as he sees it (is he a he?)

I do not dispute, as TIA appears at times to think I dispute, that the poll results were impossibly far outside the margin of error. The precise number of zeros may be debated, but the possibility that the polls were out that far due to "chance" is indeed astronomically small. Statistically impossible, as many have put it. We can agree on that. I've done the math myself, and using the WPEs, I get even more astronomically small odds than TIA gets from the state projections.

However by "chance" what we mean, in statistics, is "due to sampling error" and sampling error also means something very specific. If you were to throw a coin 100 times, you would get approximately 50% heads and 50% tails. However, if you kept repeating the exercise, tossing the coin 100 times and recording your scores each time, fairly frequently you might get 49% and 51%. Less frequently you would get 52% and 48%. Extremely rarely, you might get 10% and 90%. The 95% "margin of error" is the range of scores you would get 95% of the time. In other words, if you did it fifty times, you would stand a chance of getting a score outside the MoE 2 or three times (so if you did it in 50 states....you get my drift). The 99% MoE tells the range of scores you'd get 99% of the time. In fifty goes (or states) you might expect a score outside this MoE maybe once, maybe not at all.

This is "sampling error". If a score falls outside the MoE more often than predicted by sampling error, we know that it is not chance.

We know that the exit polls did not differ from the vote count by chance. What we do NOT know, is what the factor was that pushed the results beyond the boundaries of chance. We know it was a factor that pushed them not only beyond the boundaries of chance, but in a particular direction. Something was pushing them. That is undisputed. The question is: WHAT?

This is what the debate is about. Proving, over and over again, that the discrepancy was not due to sampling error, as TIA has correctly done, does not address the issue as to what was doing the pushing. It is notoriously possible, as anyone in the behavioural sciences knows, for sampling to be inadvertently non-random. I do research into dyslexia. I recruit both volunteers with dyslexia and those without. I find undiagnosed dyslexia far more frequently amongst my non-dyslexic "controls" than I would expect by chance, or sampling error. Why? Probably because when people see an ad about a study into dyslexia, they are more likely to respond if they know something about dyslexia. Maybe someone in their family has it. Maybe they think they might have it. So they volunteer, and because they are themselves "at risk", they stand a chance of scoring positive on my screening. In other words, my studies have "sampling bias"

That is an extreme example. But any sample that is non-random (and any sample with a response rate of only 53% is non-random) runs the risk of having a "sampling bias", because the refusals are not random. This is quite different from "sampling error". Everyone who refused to answer those polls had a reason. We do no know, and can never know, what those reasons were. But we certainly cannot conclude that they were random. The minds of those who voted for George W Bush may contain mysteries that we will never fathom.

All I am saying, therefore, is that sampling bias is a possible explanation. Not that it is the explanation. But ruling it out, a priori, is as foolish, IMO, as to rule out fraud.

Cheers

Lizzie

[Febble]

20. Post script

An answer to TIA's question about the reweighting.

Yes, the numbers were reweighted to match the count after the final sample was added. Whether this was necessary because of sampling bias or vote fraud is the question we are addressing.

Not whether they were reweighted. They were.

[TruthIsAll]

23. I never said you disputed the odds...

"I do not dispute, as TIA appears at times to think I dispute, that the poll results were impossibly far outside the margin of error.

We are way past the question of odds. I posted on this in December.
That is not an issue of debate.

The last minute change in the weightings in going from 13047 respondents to 13660 is an issue.

The fact that the state polls confirmed the national within .06% is an issue. I'm sorry if you think .06% is too close. Is it too close for your comfort?

The IMPOSSIBLE 43% Bush 2000 voter weighting IS an issue.
You can't explain it. Forget sampling error. It is physically impossible. Period.

Change the weights back to the earlier 13047 respondents. Even the UNREALISTIC 41% weight shows Kerry a winner by 51-48%, or 4 million votes. Change it to a REALISTIC 39% and Kerry wins by 7 million votes.

This evidence is far more compelling then your Febble Function, which is evidence of nothing.

[farmbo]

5. Please stay..This board is the closest thing to peer review you will find.

I am an Ohioan who witnessed first hand the tragic, calculated effort to suppress and steal votes on Nov. 2nd and the merciless exercise of political power by the GOP to prevent any meaningful legal investigation in its aftermath.

Statistical evaluations by citizen-patriots like TIA and yourself (well, OK...Scottish patriots :) ) remain our only method of keeping this epic political story alive. Like the second rate burglary in June 1972 which eventually brought down a president in 1974, this story needs time; time for cover stories to unravel, for witnesses to come forward, for journalists to gain backbone.

I am not a statistician, nor do I play one on TV, but I enjoy reading your posts...as well as the technical responses you inspire from others on this unique DU discussion board. (Kos is great, but I truly believe the best work in this area is being showcased here).

As you have iterated, your study does not rule out fraud. TIA's work suggests-- accurately in my view-- that fraud was more likely than not in play. I believe you can and should continue to post on this board and help to reconcile these views and perhaps move on to more ground-breaking work; perhaps applying your models to Warren, Butler and Clermont counties-- where Bush enjoyed his miraculous, late- night turn around.

The vast majority of us enjoy your posts and, as for the disagreeable minority who give you grief; *uck 'em if they can't take a joke... that's why we have 'ignore' buttons.

Stay here. We love ya baby!

[TruthIsAll]

7. By saying Kerry did not win the popular vote, you ignore all the evidence.

Febble, I agree with everything you said up to this:

"But I think the evidence is mounting that Kerry did not
win the popular vote, and that the Great Exit Poll Discrepancy
was largely due to poor random sampling. I realise most on
this forum disagree. I can only say that I have not come to
this view through bias. I was massively biased in favour of
finding the opposite."

On the contrary, the evidence is overwhelming that Kerry WON
the popular vote, by over 4 million votes. For you to say
otherwise is to IMPLICITLY STATE THAT YOU DO NOT BELIEVE THE
EXIT POLL DISCREPANCIES WERE DUE TO FRAUD. THAT IS A FLAMING
CONTRADICTION AND SHOWS A PRE-SET BIAS. 

SO NOW IT'S "POOR RANDOM SAMPLING"?

YOU HAVE COME THIS CONCLUSION ON THE BASIS OF WHAT? SURELY NOT
THE EXIT POLLS.

THAT SINGLE STATEMENT IS MOST REVEALING; IN FACT IT IS A DEAD
GIVEAWAY. IT FLIES IN THE FACE OF ALL LOGIC. YOU CONTRADICT
YOURSELF IN SPADES. 

NOT ONCE HAVE FACTUAL NUMBERS BEEN CONTRADICTED; THAT IS
BECAUSE THEY CANNOT BE - NOT BY YOU OR ANYONE ELSE.

BY AVOIDING THESE ESSENTIAL FACTS, YOU ARE NO DIFFERENT FROM
THE MAIN STREAM MEDIA SUCH AS THE TRIBUNE WHICH REFUSES TO
PRINT THE ROBERT KOEHLER ARTICLES. 

YOU REFUSE TO LISTEN TO THE SILENT SCREAM OF THE REAL NUMBERS.

THE FACT IS THAT THE NUMBERS CONFIRM THE VOTING MACHINE
ANOMALIES WHICH TURNED 99% OF KERRY VOTES TO BUSH VOTES. AND
IN SPITE OF THIS DIRECT EVIDENCE YOU STILL BELIEVE THAT THE
DISCREPANCIES WERE DUE TO "POOR SAMPLING"? THATS
POOR LOGIC. THAT'S NO LOGIC.

INSTEAD YOU ARE INVENTING YOUR OWN REALITY TO KEEP THEM
SILENT. THE NUMBERS CRY OUT TO BE HEARD, BUT YOU STIFLE THEIR
SCREAMS BY IGNORING THEM. IS THAT THE LEGACY YOU WANT TO
LEAVE? NOW WE FINALLY KNOW WHERE YOU STAND.

Febble, with all due respect, your bias is showing. 

FEBBLE, HERE IS THE DIRECT EVIDENCE WHICH YOU CHOOSE TO
DISBELIEVE.
TELL US ABOUT THE MYSTERY OF THE FINAL 613 RESPONDENTS.  WAS
THAT AN ATTEMPT TO MATCH THE VOTES AND CORRECT THE "POOR
RANDOM SAMPLING"?

NATIONAL EXIT POLL TIMELINE				
VOTED IN 2000 FOR				

3:59pm 8349 
		KERRY	BUSH	NADER
No	15%	62%	37%	1%
Gore	39%	91%	8%	0%
Bush	42%	9%	90%	0%
Other	4%	61%	12%	16%
Total	100%	51.01%	46.95%	0.79%
Votes 	122.26	62.36	57.40	0.97
				

7:33pm 11027 				
No	17%	59%	39%	1%
Gore	38%	91%	8%	1%
Bush	41%	9%	90%	0%
Other	3%	65%	13%	16%
Total	100%	50.25%	46.96%	1.03%
Votes 	122.26	61.44	57.41	1.26


12:28am 13047				
No	17%	57%	41%	2%
Gore	39%	91%	8%	1%
Bush	41%	10%	90%	0%
Other	3%	71%	21%	8%
Total	100%	51.41%	47.62%	0.97%
Votes 	122.26	62.85	58.22	1.19

				
1:25pm 13660				
No	17%	54%	45%	1%
Gore	37%	90%	10%	0%
Bush	43%	9%	91%	0%
Other	3%	71%	21%	8%
Total	100%	48.48%	51.11%	0.41%
Votes 	122.26	59.27	62.49	0.50
				
				

[Febble]

8. DON'T SHOUT!

It wasn't a pre-set bias. It is my reading of the evidence, having started with the opposite bias. If you like it is a post-set bias.

I could as easily argue that your assertion that the evidence is over-whelming that Kerry won the popular vote is a dead giveaway that you are biased against the opposite view.

But I won't say that, because I assume, as a mathematician, you have, like me, simply come to that view on your reading of the evidence. You also have a "post-set" bias, which is an entirely reasonable thing to have.

Presumably we have both come to our current views on our reading of the evidence. Mine is not set in stone, however, so it is not even "post set" - a "post setting" bias, maybe.

The difference between us is in our reading of the evidence.

[TruthIsAll]

9. I SHOUT WHEN I AM ANGRY. YOU MAKE ME ANGRY.

.

[Febble]

10. OK,

I realise that ;). I guess you make me angry too. But it seems unreasonable of you to assume that I am biased because I consider both possibilities (fraud and response bias) and for you to assume you are not biased because you only assume one.

It's usually the other way round.

[pgh_dem]

13. everybody play nice! (hopeful thinking)

In another thread, you both agreed that the raw data with precincts directly identified could be used to accurately settle your dispute over the reasonability of rBr. Also, I understood you to agree that absent rBr, the only remaining reasonable explanation is fraud.

TIA didn't think it was totally necessary, but agreed it would be 'icing on the cake' to have that data. Febble suggested it would be possible to strip the sensitive information from the data (which is one of two reasons E-M has withheld it, the other being a specious argument that the data 'belongs' to the network clients who paid E-M to do the poll).

Question is: Is there anything that fellow DU'ers could do to help shake that information loose? If repeated emails and letters to E-M(which I know have been sent), letters from participants in the Ohio recount lawsuits, and letters from Rep John Conyers and others in Congress, haven't had any effect, is there any avenue that interested people might pursue to try to settle this matter.

In other words, is there anything short of a subpoena (which we obviously can't issue) that might induce E-M to tip their hand?
(Edit for crafty thought: is there anyone else who *already* has it?)

[TruthIsAll]

28. It's not your claiming RbR is a possible cause, but the fact

that you believed it to be true.

You let the cat out of the bag with that whopper.

What brought you to that belief?
Where is your factual evidence?

Or is your belief faith-based?

I believe there was massive fraud.
This is based on factual data - amazing polling discrepancies and documented incidents.

You said that you believed it was due to FAULTY POLLING.

So now are you prepared to say that E-M did not factor and account for refuseniks?

That is pure BS and you know it.
Tell it to Mitofsky.

No, Febble, it was FAULTY COUNTING.

[TruthIsAll]

11. My bias is the truth. And what exactly is yours?

Don't lecture me about bias.
What bias? Those are REAL exit poll numbers which were meant to be hidden from view.

I showed you my evidence as to why the odds are astronomical that Kerry won the popular vote by over 4 million votes.

Now you show me your data.

[Febble]

15. I want the truth

as you do.

[TruthIsAll]

16. Once again. Prove it. Show us your data - not the money, the data.

.

[TruthIsAll]

38. Febble, what is your reading of th evidence based on?

"The difference between us is in our reading of the evidence".

Mine is based on statistical analysis of the data.
No pre-set assumptions whatsoever.

Please be specific.
What evidence are you reading?

[Verve]

12. You need to stay Feeble.

Despite the controversy you have caused, the more people around the world that look into this, the better. We need the world to help us Americans find out what went wrong with our elections.

However, your academic arguments are bittersweet. To understand this you need to understand this forum's mindset. Many people like Truthasall have been working nonstop to uncover fraud and doing a superb job I might say. In a different circumstance, I assume he and many others would welcome a debate with you and your material.

However, he and many others on DU have been screaming at the top of their lungs for months that something went wrong in November. Despite their hard work and extreme persistence, the word here in America is NOT getting out! It is beyond frustrating to be told even by fellow democrats that nothing happened in November!

We have to give Bushco credit as they continue to muffle any hint of fraud that is brought up. Our administration is the greatest spinmaster on earth! With that in mind, having an esteemed individual, like yourself, question all their hard work and then have your info used as spin to again squash any fraud concerns is another blow to our cause of getting the word out.

With that said, I do hope you continue to challenge and work here at DU. We need millions of people like you from other countries questioning our democratic process. Screaming from the inside of U.S. is not working yet. Maybe screaming from the outside will.

[TruthIsAll]

14. Febble, stay... with all your contradictions, careless wording and biases

exposed for all the world to see.

Up until today, I believed we were essentially in agreement. No longer.Today you showed us your true colors.

No, Febble, don't leave.
Like all the naysayers who came before, your specious logic just strengthens the case.

Stick around.
You want to debate?
Let's get it on.

[Verve]

18. TruthisAll, please look at the big picture!

I may be wrong, but it sounded like she is open minded. Give her the benefit of the doubt. Maybe she hasn't digested your extensive research yet? And if she has, what's wrong with a healthy academic debate?

If you don't like what she's saying, use your persuasive skills and your great evidence to get her to see where you are coming from. You convinced many of us. If you continue to disagree, than find some middle ground on another issue. It appears to me she is not ruling out fraud.

IMO, DU would be extremely boring and unproductive if we all had the same opinions on EVERY issue! Yet, we share the same opinion that fraud occurred or we wouldn't be here.

I am one of your biggest fans here at DU, TruthisAll. It pains me that I offended you. That definitely was not my intent. My intent was to keep as many people looking at this issue as possible. If other scientists from around the world are starting to look into what we've been screaming about for months, why should we push them away so fast?

[Name removed]

19. Deleted message

Message removed by moderator. Click here to review the message board rules.

[Verve]

22. Good point!

I apologize for butting in. I'll leave it to you two statisticians to battle it out.

[TruthIsAll]

21. I don't deal in opinions, just facts. I showed her facts. She showed zip.

Until she has found a numerical rock to stand on, she has nothing.

I leave it to USCV to prove it.

[Name removed]

26. Deleted message

Message removed by moderator. Click here to review the message board rules.

[TruthIsAll]

27. Your philosophical musings are pathetic. Show me the numbers.

You call me a hack?

What have you ever brought to this forum except for expertise in doublespeak, circular nonsense.

You are incapable of rational analysis.

[mgr]

30. I believe that term was used elsewhere

to describe your efforts to Febble prior to her involvement on this site. You might check Daily Kos or Mystery Pollster for the source. That appears to be your reputation in the wider world.

Let me see if I have this straight--the only thing that matters are numbers, not what they represent, nor how they were measured. All you want to do is pop them into a spreadsheet. I would say that pretty much meets the definition.

There is no reason to evaluate the manner that calculation was performed if the rationale for doing so was wrong. And it is the rationale you have to have right before going on to the number crunching.

Your arguments are circular. Prove to me that they are not. Do so in a manner that a non mathematician can appreciate. If you can do this, I will leave you in peace.

Mike

[TruthIsAll]

36. Where do I make ONE assumption in my analysis. It's all numbers.

You have been around a long time and you keep repeating the same old same old.

Febble is the one who is making the implicit assumption that the exit polls were wrong. She said so today.

I make no such assumptions.
I used the exit poll data as presented.
I have analyzed the data every which way from Sunday.

No assumptions, just permutations of the same data.

State time zones.
National Exit poll time lines.
Optiscan vs. touchscreens in Fl.
Exit poll data from Ohio.
Historical national exit poll data vs. the vote.
Weighted state exit poll matched to the national.
MoE calculations based on standard formulas.
Probability calculations based on standard formulas.

I could go on and on.

Show me where Febble has done likewise.
She has presented a hypothesis which is at VARIANCE with exit poll data (see Kathy Dopp and Bruce O'Dell)

In time, the truth shall unfold.
In spite of your eloquent diversions.

[Name removed]

40. Deleted message

Message removed by moderator. Click here to review the message board rules.

[TruthIsAll]

49. I would gladly disappear from DU like Kerry's votes lost in the ether.

" I think that you were afraid of Febble stealing your thunder". Right. As if I enjoy this. As if I enjoy this ongoing banter with someone as apparently close-minded as your=are.

mgr, you enjoy listening to yourself talk. I would rather if I had never once posted on this farce of an election. I don't need the glory. You are getting personal again.

I want my vote to count. I'm funny that way. I care about nothing else.

What the hell do you want?
What do you stand for?
Transparent, open clean elections?

I will repeat a question I have posed to you before but which you never responded to. I hope this time you will.

You have said you believe there was fraud.
What is the basis of your belief?

If you believe there was fraud, why do you keep nay-saying my posts which just confirming YOUR belief using very straightforward mathematics from Probability and Statistics 101.

Yet you agree with Febble's function which is an attempt to prove that RbR not fraud, was the cause of the discrepancies.

You see, you are a world-class contradiction.
I call it hypocrisy.

[TruthIsAll]

63. Correction.

I said:
Yet you agree with Febble's function which is an attempt to prove that RbR not fraud, was the cause of the discrepancies.

It should be:
Yet you agree with the thrust of Febble's analysis. Febble's assumptions, according to Bruce O'Dell and Kathy Dopp at USCV, is contrary to actual exit poll data.

[Febble]

65. Except that neither of us have seen the exit poll data.

My analysis makes larger assumptions regarding variance, and as a result my modelled absolute WPE values are closer to those given in the E-M report.

[mgr]

68. Reply 2: The term is teleological

It is arguing that the consequent causes the antecedant, its Aristotle's term for circular reasoning.

Assumption 1: ~13000 reported poll responses are true, the next 660 are not. Well, the ~13000 talley to a Kerry win, when the 660 are included, they talley to a Bush win . Kerry must be winning, therefore portions of the tally are wrong--a teleological argument.

What makes this especially poor reasoning is that you do not consider that if underhanded actions were to occur with the NEP, it would be with the earlier report polling as well. Scoop's data are all weighted, and if Mitofski is in on fraud, and in giving Bush a mandate, why do it at the end of the day, when you can hide it in the weighted data? After all your analysis of the pre election polls showed a Kerry landslide, how would these conspirators not know?

If the 660 is bad, then it is likely that the ~13000 are as well. If the 13000 is good, then the 660 is just as good. If there is criminality or incompetence it should pervade throughout the exit polls, not just the portions where you want it to be.

On the one hand you expect NEP to be perfect when your position requires it, and on the other hand to be incompetent, again when your position requires it. That's called cherry picking, and it is another way of making assumptions.

Mike

[Chi]

69. emcguffie has a question, I thought her best bet for an answer was you.

If your interested.
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=365850&mesg_id=366107&page=
Post #17

(gonna take on the 613 huh, this should get interesting)
:popcorn:

[mgr]

71. Boy are you the devilish imp.

Chi: I think I accumulated my quota of messages deleted by moderator for a while given how upset I was with the reception given Febble. Thank you for showing her some hospitality, that was class that was sorely missing.

I gave Mcguffie a partial answer. Its actually a very good question.

The 613 is a non issue, let the fool keep fighting over it, we're just spinning our wheels. I mean what we have to do is get these assholes at the state level, not with the national poll, and we have got to do it soon. Show that they fucked us over in the electoral college, and not just in Ohio. Its been six months, and look at what we have to show for the effort since December.

Mike

[Chi]

72. No sir, no chumming the water from me. 8)

Really though, I wasn't trying to instigate.

I went looking for you for mcguffie, cause she really seemed to be wanting an answer.
And I sent her astray with a joke, accidentally.

I too was....disturbed by the level of hostility shown toward febble.
It reminded me of my time at FreepLand (before I knew about DU) where any differing idea is looked upon like you are a Major Troll, hell bent on pissing them off.
I knew to expect it there, I'm disappointed to see it here.

I'll skip on the 613 debate, you've had enough deletes today ;-)

[LightningFlash]

75. Hmmm...End of the day?

Just where are you getting these assumptions from, in fact may I ask?

http://www.exitpollz.org

http://www.exitpollz.org/cnn2004epolls/Pres_epolls/US/P/00/epolls.0.html

CNN themselves, apparently say that the exit poll results went online at midnight. That wasn't until basically the next day.....Certainly not 4:30 PM....and it was an hour later at 1:41 AM where they altered....

I have to say your arguments ignore the wider issue at hand. The exit poll re-weighting was NOT random. There was an actual switching place of the numbers, and confirmed subtracted tallies. There was nothing under any rule of thumb, that was considered random about the exit poll re-weighting nor the entire exit poll itself.

That is if we are to go by the official raw United States Cencus data of the government, instead of assumptions by non- PHD's trying to simulate other tests.

http://www.census.gov/Press-Release/www/releases/archives/voting/004986.html

[mgr]

76. It is an argument, based upon other's assumptions

NEP reports reweighing the exit polls until 4 pm the following day. (1/19/05). What we have after 12pm EST is a partial talley, evident by the discrepant numbers in TIA's tables.

Reweighting is supposed to do precisely what you describe.

And, it is the US Census, not Cencus; and the release is not a complete count by the census, but a sampling, with its own inherent bias and sampling errors.

Explain what you mean by random.

What I am describing is that if there was conspiracy, then why do it so half assedly?

Mike

[LightningFlash]

80. You just answered your own question.

"then why do it so half-assedly?"

http://www.vheadline.com/readnews.asp?id=34681

Because the people who did the conspiracy, knew they could stack the courts. It's that simple, time to stop living in denial.

mg, the election was stolen. The demographic weightings prove Kerry was elected but lost votes. However, what we don't know is by how many votes and still are no where close to that answer.

He either won barely, or won by and by the popular vote as well which is still not possible to prove with the cited charts you "point" to.

We have to be doing a full regressive analysis of the eastern states, and how many deviations were actually bias and were actually fraud. There was clearly more fraud than bias so far, with what you can demonstrate physically in the states where vote fraud has been most popular.

But there was also sampling bias, and it needs to be regressed to determine how much sampling bias and fraud happened in those 9 states. How much democrat based fraud also, was used in this election because the conspiracy to "steal" it had nothing to do with the individual races where both democrat/republican were conducting fraud.

[mgr]

81. I really wish your writing was more precise.

The Keefer article sums up the positions as of five to six months ago. Significant to this was the refutation on this board of the random generator argument. I was active on the site then.

Is what you are saying is that Mitofski and company were not part of the conspiracy, or were? What does owning the courts have to do with anything, as to why such a sorry job was done covering it up? The courts do not affect public opinion.

Care to explain how you are going to regress data on the eastern states, and discriminate bias from fraudulent activity within what context? The exit polls or the actual voting? What does regress mean?

I'm very tired of folks telling me that I need to understand that the election was stolen, when my position is that it was. I just do not agree with the vocal majority that seem to see it in the exit polling data; and since no one seems to be able to refute the position I have staked out, I'm inclined to think that I have it right.

Mike

[TruthIsAll]

82. mgr, are you aware that you have actually mellowed?

I'm even starting to like you.

You say that the election was stolen.
I know you said it a while back, even when you were one of the most persistent naysayers.

Good to have you aboard.

P.S.
You must admit, we have done a good job in showing that the exit poll deviations (state and national) are extremely powerful circumstantial evidence.

Once again, if you believe the election was stolen, then you must also believe that the final 13660 Exit Poll which has Bush a 51-48 winner is bogus. Therefore, you must at least concede that the 13047 Exit Poll, which had Kerry winning, is much closer to the truth.

And that is all I've been trying to show over the last six months.

[mgr]

83. You know

I really have not forgiven you for flaming me as a newcomer, and tarring me as a republican, when it was obvious then as now that I considered fraud a reality, and was anything but. It is not that I am mellow, it is that your position is growing untenable with what is going on with USCV--I recall you stated that you and some of its members worked together (see this http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x375366#375497, and this http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x372464#374499). I think everyone now can see given these comments, that that was never the case. I think you hurt our efforts by promoting the red shift as widespread, when that response was the responsiblity of NEP, and took away much of their thunder when the movement needed it most. I told you then that you were in the way, and it was time to stand down. Now you take O'Dell's assumptions and behave as if those were your own. It would be really helpful if you showed some original and critical thought instead of bashing those around here that do. I would almost conject that you might be the repuke, not me; but I just think that you let the limelight blind you.

The fact is is that you never, until this moment, referred to the evidence that the exit polls present as circumstantial. But circumstantial evidence of what? Fraud, or that exit polls are useless when the electorate is so polarized?

The problem with the 13047 tally is that is represents an incomplete tally, since NEP compiled from Los Angeles, and the screen save is 2400 EST, which would be 2000 PST, just when the final totals for the west coast were being obtained (the polls were closing in the 48, except select precincts in Ohio). It is reasonable to presume that the 13660 represents the final numeric total, since all that would be missing would be 613 respondants, mostly from the west, midwest, and south (more conservative regions than the east). The only thing missing would be Hawaii and Alaska, and they would have to come in the next day. I think it is safe to presume that NEP suspected how they needed to reweight the data before H & A were in, and that is what you have with the 'corrupted sample'. The official reweighting occured the next afternoon. That is the plausible explanation. The thing is that we have probably alienated NEP to where I don't think they will dignify us with any response, so we are unlikely to ever know.

The thing that you have taught me is that it is as appropriate to hate progressives as I would conservatives for their truncated world views and ignorance. I wish you enlightenment when you enter that bolgia reserved for the schismatics.

Mike

[LightningFlash]

84. Problems with the final weighting.

Like stated, the final weighting is conclusively impossible because at least 1.5 million of those who voted in 2000 had died(All Bush voters--nonetheless) and the gender weighting split up by reduction.

What that proves is not any oversampling of women, but instead lost unaccounted for votes that are not included in the final weighted average.

Using a composite analysis of each individual state's exit poll data, is how to regress and determine which regions are responsible. It seems indirectly high kerry regions received the most deviations based on fraud, and not undersampling.

[Blue Shark]

44. TIA...

...you have some special skills that you have shared with us all...too bad graciousness is not among them.

[progressivejazz]

24. Febble, please stay.

We need at least one person who has both your expertise and intellectual honesty working on this issue.

Thank you for your work.

[Cocoa]

25. I disagree that "most on this forum disagree"

my personal view is that there is a vocal group of "activists", whose methods of discussion exaggerate their numbers and exaggerate how much they represent DU as a whole.

I would agree that they've probably been successful in creating the perception you express, that "most on this forum agree."

[Name removed]

29. Deleted message

Message removed by moderator. Click here to review the message board rules.

[Chomp]

31. Just a reminder

lib·er·al

adj.

1. Not limited to or by established, traditional, orthodox, or authoritarian attitudes, views, or dogmas; free from bigotry.
2. Favoring proposals for reform, open to new ideas for progress, and tolerant of the ideas and behavior of others; broad-minded.
3. Of, relating to, or characteristic of liberalism.
4. Liberal Of, designating, or characteristic of a political party founded on or associated with principles of social and political liberalism, especially in Great Britain, Canada, and the United States.

n.

1. A person with liberal ideas or opinions.
2. Liberal A member of a Liberal political party.

[mgr]

32. As were her papers with USCV?

As I recall, she independently addressed the same phenomena as Hout's paper, with same conclusions. They were all posted here.

Mike

[progressivejazz]

33. For slow people like me...

would you please show me explicitly where the rule you have in boldface disqualifies Febble from being here. I just don't see it.

She claims to be a progressive. She claims to have supported Kerry in the last election. What disqualifies her?

[autorank]

58. Delete

[glitch]

34. A scientist should appreciate when their presumptions are challenged.

It helps them hone their analyses.

Upthread someone recommended you do your analysis again with the presumption of fraud in the count, not flaws in the exit polls.
You may still come to your same conclusion, but wouldn't you find it an interesting experiment on your own cognitive behavior?

[mgr]

35. Not how the null hypothesis works. n/t

[davidgmills]

37. Apparently it works like

A+B=C but two variables are unknown.

On edit:

Say C is 7

The null hypothesis is that you can't rule out A from being any number between 0 and 7.

That helps things alot.

On edit again:

Oops.

That means you can't rule out A from being any number in the universe of numbers since A or B could be negative numbers.

[mgr]

43. here is a simple example:

David, I really expect better of you. What you are characterizing is a universal.

In the first place, the logic is not that of an addition problem, that is more appropriate for and 'and' statement or an 'or' statement.

Experimentation uses the 'if-then', a simple minded one is "if it is cloudy, then it will rain" What actual conditions would make this 'if-then' statement false? Use your own experience and observations as I run through this.

S1.) It is cloudy, and there is rain. That makes the statement true

S2.) It is not cloudy and there is no rain. That does not refute the statement

S3.)It is cloudy, but there is no rain. Does this condition refute the statement? No it does not, all it does is suggest that other conditions related to being cloudy may be necessary for the statement to be true, it could be that there are no condensation nuclei or the precipitation is coming down as virga. In this case, we conclude only that the condition cloudy is necessary for explaining rain, but not sufficient.

S4.) It is not cloudy, but there is rain. This refutes the statement.

Now look at what has happened, our hypothesis has three conditions where it may be true, and only one where it may be false. S3 makes it impossible to know with assurance that the causal relationship is
true.

How do we make a null hypothesis that may show a connection between clouds and rain-- it would run-- When it is not cloudy, then it will rain. What happens is, if you live in Los Angeles or the Atacama Desert, it may take many days to get the event that refutes the null hypothesis, each of those days up until it rains is similar to the condition S3 (what Febble referred to as the innocent until proven guilty condition).

Don't feel bad, this concept was the most difficult I had in getting across to students in my introductory science courses. It reflects poor introductions in basic science by secondary school teachers that did not get it when they were taught it.

Mike

[davidgmills]

46. mgr I've taken a cat scan of your brain and shown it to some of my MD

friends. They tell me they are almost certain that you have Alzheimer's and that Alzheimer's is the reason for your premature senile dementia. Of course, they can't know for sure unless they do an autopsy on your brain.

They also say that they cannot rule out your premature senile dementia as being due to your high blood pressure which is off the charts and is totally uncontrollable.

Go ahead and insist if you will that it is your blood pressure causing your dementia because it can not be ruled out. Go ahead if it makes you feel better.

[Name removed]

51. Deleted message

Message removed by moderator. Click here to review the message board rules.

[davidgmills]

56. I'm sick of it mgr

Stop.

[progressivejazz]

53. Terminology.

The way I learned if-then statements was that the "if" part was sufficient to produce the "then". That's the way it's used in the physical sciences and mathematics. You're using it to mean the "if" part is merely necessary to produce the "then". More of an "only if".

"If you are a pregnant human, then you are a female." Now THAT'S an if-then statement.

No wonder your students have trouble.

[mgr]

62. There was a point to the example

Your point is valid and correct. The point of my example was to provide an 'if then' statement that was operationally equivalent to a scientific 'if then' but not really.

The distinction I was making between sufficiency and necessity, was from a perspective that only S3 was observed, and why one would not reject the hypothesis.

Science's purpose is to predict the future, and to explain the past. The association between clouds and rain is one of the earliest scientific observations made, and is one, that when you rise in the morning, and see that it is cloudy, you will take an umbrella.

However, the concern with explantory power also drives science. You certainly don't see much explanatory power in your tautology, since the antecedent anticipates the consequent by definition; but correlation is also a problem. Biology, in attempting to develop testable hypotheses related to evolution has to run this gauntlet; and unfortunately, so do the other physical sciences.

Although the relationship of clouds and rain may be adequate for you to take an umbrella, it may be inadequate from a scientific perspective that desires to obtain better quantitative data. So, the antecedant would be better configured--water vapor, uplifted beyond the elevation of the condensation point, and the presence of condensation nuclei, and a normal lapse rate, will produce conditions that will yield rain. With this model, one can better characterize conditions that will give or not give rain.

Mike

[gagarux]

59. True Always Implies True

You have S3 and S4 below exactly backwards, mgr. Trot out the old truth table. Your premise actually states that being cloudy is a necessary condition for rain. A truth value of False can only be obtained if it is cloudy but does not rain (S3). May be because your example was based on a false premise, it confused you (it doesn't always rain when it is cloudy!).

Gary


>S3.)It is cloudy, but there is no rain. Does this condition refute the statement? No it does not, all it does is suggest that other conditions related to being cloudy may be necessary for the statement to be true, it could be that there are no condensation nuclei or the precipitation is coming down as virga. In this case, we conclude only that the condition cloudy is necessary for explaining rain, but not sufficient.

>S4.) It is not cloudy, but there is rain. This refutes the statement.

[gagarux]

60. Change that to

"Your premise actually states that being cloudy is a sufficient condition for rain". Big difference.

[tommcintyre]

61. I hope you don't use this example. It DOES rain without clouds

Well... at least according to NASA and Dr. Livingstone (I presume):

http://astrobiology.arc.nasa.gov/palebluedot/abstracts/toon.html
"...it is quite possible to have rain without clouds."

http://etext.library.adelaide.edu.au/l/livingstone/david/mission/chapter29.html
David Livingstone
Missionary Travels and Researches in South Africa - Chapter 29.
"Humboldt has seen rain without clouds..."

Moral of the story: Absolutist thinking yields the same answer over and over again. Relativist thinking has a much higher probability of revealing the truth - whatever it may be.

Absolutist thinking is a good explanation of why there are so many Bush supporters, in spite of the facts.

Your post #45:
"Actually, it <the null hypothesis> is everything."
<This sounds pretty "absolutist" to me.>

I'm glad to see you responded positively to the post about alternatives to the null hypothesis - maybe there IS hope for you yet. ;)

[mgr]

64. Actually I do, and it works

Alexander Von Humboldt's observation, its been years for me to remember the exact provenience, but I believe it was either the Andes or the Amazon Basin, was reported in Erdekunde I believe, is a well known observation of the phenomena of supersaturation. The air is inordinately clean and lacks condensation nuclei, so the air parcel continues its ascent until the water vapor molecules themselves become the nuclei, and precipitate prior to any observable manifestation as clouds. (By the way, I wrote the paragraph before reviewing the astrophysics citation)

The reason I use the example is to show how in our everyday lives we behave as scientists. The association between rain and clouds goes back to Thales, and is one of the earliest scientific observations (at least in European tradition--it is possible the observation was a borrowing from other cultures). However, all you did was show that the hypothesis is refutable, which makes it a scientific hypothesis.

I think you may be overly enamored of the concept of scientific revolutions. The article, as I read it, is more cautionary against applying tests of significance to control and experimental groups, without considering the effect may be beyond the means of the F test to resolve, and tries to address other methods or approaches to resolve the problem. I would not see it as a recommendation against the null hypothesis, but only the null hypothesis as originally conceived by Fischer, and an attempt to rehabilitate one methodology among the panoply employed using the hypothetico-deductive method. As I may have pointed out, I am unsure whether biologists have already wrestled with the issue or not, and I suspect that we have if I am recalling Sewall Wright's work.

Mike

[glitch]

39. Irrelevant in this case. n/t

[mgr]

45. Actually, it is everything. See post #43 for example.

[glitch]

48. testing against null hypothesis is not the only statistical test available

and not the optimal in this case. I was hoping for a break out of that mindset. Since you are responsible for educating others, you might look into what tools might be more effective than NHST in examining the problem at hand.

Maybe you'll find this interesting, I googled it especially for you:
http://theoryandscience.icaap.org/content/vol4.1/02_denis.html

[Febble]

50. This is a good article

and outlines an approach that was part of my statistical training. It is very relevant to this issue, and in fact is an approach that informed my thinking. If you read my paper you will see a reference to effect size (re response rates).

The conventional way of expressing my objection to the USCV position is that "the null is retained". In other words I do not believe that they have demonstrated that "vote corruption is greater in Bush strongholds". The better way of putting it is that the effect size they postulate is too small to be an explanation of how fraud could have been the major factor in the Great Exit Poll discrepancy, even it it is a real effect.

The irony, as I have said before, is that the effect size I postulate is much larger. The snag is my analysis cannote distinguish between fraud and response bias as being the cause of the effect. Other analyses, however, might.

[glitch]

52. Excellent, which is why we need you to stay on board.

One mountaintop, many paths. If I haven't said this already, welcome to DU :hi:

[mgr]

54. Interesting article

I need to look closer at it again, since I cannot tell if the take off point is in the Model I and Model II ANOVAs, or these are elements of an alternative. Unfortunately, as a biologist, I think we have been taking into account treatment effects for some time--I am not quite sure if the same terminology crosses over exactly to psychology--it was these ANOVAs that I thought could potentially address exit polling at the precinct level when recombining.

My own temperment suggests I would favor combining the two approaches into one, and agree treatment effects are significant even if the population is too small to show significance. One should never presume randomness or drift just because the sample size is too small.

One of my favorite books is Dialectical Biology (Levin and Lewontin)that also addresses an alternative to the null hypothesis model. Give it a gander.

Mike

[Febble]

55. One of my stats tutors

at UBC Vancouver actually deducted marks if we gave a p value. He was very eccentric but he made us think.

One of his other little sayings was:

"And does not God also love p=.06?"

It always strikes me as nuts that students will get terribly excited at a result that is significant at p=.049, and will blithely accept as non-significant p=.051, especially if they are checking that, say, their experimental groups were not different at the outset of an experiment.

P values are a terrible proxy for effect size.

But you do need p values for peer review still!

[Blue Shark]

41. Dear Elizabeth...

...Well-reasoned discourse is the hallmark of being a part of the "Reality-Based Community". You qualify in spades.

...Please stick around as there are some very enlightened arguments made here every day that don't get floated on MyDD, dKos, and other sites.

...Perhaps your superb analytical skills will catch on one of the "other" pieces of fraud and you may bust the 2004 election wide open after all. And wouldn't it be nice!

[Febble]

42. Well thanks!

I do think I've sharpened the tool we need. I just doubt it is sharp enough. But we'll see.

Lizzie

[Blue Shark]

47. Some of us are so...

...frustrated at the lack of vision (seeing what is before our very noses), the lack of outrage, and the lack of coverage of the biggest electoral theft in the history of the Republic that we forget that politness is a real virtue.

...We fear for our freedom.

...Clarence Darrow, Martin Luther King, and Mr. Ghandi did not need to shout to be heard by millions.

...Today's discussion between you and TIA and others is the most substantive and important dialog to be addressed on this (or any other) board in months (IMHO).

...We must come to a concensus among ourselves before we try to convince the world. We must honestly look at all the evidence, weigh it judiciously, discard what is unworthy and shine the light of truth on what is left.

...Both TIA and you (and numerous others) are very powerful voices in our struggle to save our very democracy from the thugs. Thank you for braving out the (at times) rude reception and continue to seek the truth.

...Wouldn't it be nice to know who REALLY won?

[deacon2]

77. Actually, they did shout

"Clarence Darrow, Martin Luther King, and Mr. Ghandi did not need to shout to be heard by millions."

What they did was to challenge the status quo and the powers that held it in place. Their "shout" was so loud that two of them were assassinated. I admire all three of these people, but I'm not them. It's okay to shout, it's okay to be passionate. It's okay to tell people they're full of shit if you think so.

I personally find much of the discussion on this particular thread to be about as mannered, stuck up and tight ass as I have ever read. But it's also intelligent and god knows it's necessary.

But all this ivory tower crap won't get it to the street - that's where hot-blooded activists come in. So let us know when you have it figured out and we'll carry the water from there. In other words, there's a place for everyone at this table. Those who shout and those who don't like the shouting.

Who said (and I misquote): I disagree with your position and I will defend to the death your right to hold it.

Something like that...

[bleever]

57. I'd like to second Land Shark's sentiment.

The passion in this forum regarding this topic is, as you've seen, deep, intense, and unrelenting.

So is the anger and frustration at what we see as the rigging of our democracy, with the complicity (witting or otherwise) of the media that has always acted as a safeguard against the malfeasance of politicians.

So, some of us shout, some of us get abrupt, and most of us are suspicious of those whose work may inadvertantly support those who would suppress the truth. Lord knows we've seen impressive phalanxes of naysayers and disruptors over the days since November 2.

But please don't doubt that the reasoned, detailed, and fair-minded work you've shared here has been deeply appreciated, and that like a hot tub, once you get used to the water here, it's an excellent place to be.

Thanks.

[Land Shark]

67. What in the world is "evidence" of fraud to Febble?

I think I have a pretty clear idea of what evidence of fraud is, since i litigate fraud cases.

Febble posits "we may not find evidence of fraud" in the exit polls. This is an outrageously false statement as a legal matter.

It may be weak evidence, it may be ambiguous evidence in that it lends itself to more than one interpretation (I'm not suggesting either is the case) but the exit polls are SURELY, and beyond doubt EVIDENCE of either fraud or irregularity in the election.

yes, one can try to use the evidence sometimes to attack the exit polls, but that doesn't prevent it from being evidence of fraud or irregularity, it only means that the evidence can be argued more than one way (is arguably ambiguous)

I am not trying to suggest the exit polls are weak evidence. I am only trying to say that, as evidence is a legal term, Febble's statements regarding evidence imply an entirely bogus concept of what "evidence" is.

They import into the conversation the beginnings of the even-more-bogus idea that "proof" is necessary in order to commence or continue an investigation into this. If anything like this were really the standard, we would never have heard about OJ's criminal case or virtually anything about Whitewater, since nothing was ever proved in those cases, nor would there be any Michael Jackson coverage these days.

No, usually allegations that have an apparent substantive base to them are sufficient to make the news and be fit for public debate. Under that standard, it's news, and the under the legal standards, there's lots of evidence.

As they say in court: "the objection goes to the WEIGHT of the evidence, the evidence will be allowed...."

[mgr]

70. This Manichean thinking is for the birds.

As much as I respect your efforts, the term evidence does not have an exclusive legal import. I think implying that Febble only referred to the legal import of the term is over-reaching, as there is an equally valid concept of 'scientific evidence'. The national exit poll may have utility in a legal arena, since there exists far stronger evidence that it fits a pattern; but have none at all in the scientific arena where it may have to stand on its own alone.

The only resolution to the entire matter is to understand that two possible methods remain to discern if fraud is evident from the exit polls--USCV and Febble, and it may be possible that both can resolve the fraud statistically, or neither. You do not have to pick either position.

Her point is both sobering and appropriate, and should be the clarion call that we are running out of time to obtain redress. But all it means is that statistically, her method may not be sufficient to assist.

Mike

[Land Shark]

73. Febble is leading people to believe something that's not true

namely, that the exit polls are not "evidence" of fraud and therefore don't support a fraud case really at all. But they are legal evidence that would be used in court in favor of Febble's detractors. There is no doubt about that. Of course, on cross examination, Febble would attempt to show otherwise....

Now, maybe Febble doesn't know that there's both a legal sense of the word evidence and her scientific/statistical sense, but I really doubt that. Instead, it seems far more likely that she is fully aware that the word evidence is subject to two materially different definitions.

When a merchant seller knows that their words or conduct are capable of two different interpretations and that the consumer is likely to choose one definition that is not the one the merchant seller intends the agreement to mean or that a court is likely to say it means, then that states a legal case of outright FRAUD, not just consumer deception (which is a lesser and easier claim).

Now, I'm not saying fraud is Febble's intent, but I am most definitely saying that Febble (and anyone else that says this) is irresponsible and deceptive if and when making they are statements of the sort that the exit polls are "no evidence" of fraud, or statements like that.

These are CRITICAL CRITICAL statements that Febble is (for whatever reason) shading in a way that is likely to lead readers to adopt a much stronger "there was nothing wrong" with the election position than Febble's real data and opinions justify.

Febble should instead say, "the exit polls provide some evidence of possible fraud, tempered by contrary inferences..." and usually this kind of paragraph ends up with the conclusion of "more study is needed".

[Name removed]

78. Deleted message

Message removed by moderator. Click here to review the message board rules.

[mgr]

79. I think her position is that they are insufficient evidence of fraud. n/t

[Bill Bored]

74. Stay!

It's always useful to question assumptions.

When it comes to the popular vote though, I'd like to know why the polls were so far off that they had to adjust party ID weighting to give Dems and Repubs equal turnout, and in the Western region, they actually had to reverse the party turnouts so that Repubs outnumbered Dems. This was the only way Bush could have been in the running at all in the popular vote. Since, at least historically, there are more Dems then Repubs, and there were a number of well known and ordinary Republicans who supported Kerry, this raises an important question which I have not seen answered yet. Maybe you will be the one who answers it!

Don't let the true believers get you down.

We have junk voting machines; maybe we have junk exit polls too. The two are not inconsistent and neither is the possibility of election fraud. I just think people get hung up on these polls because maybe it's the best evidence they can come up with personally, and we do use them to spot rigged elections in other countries.

It is somewhat compelling to me that in order for Bush to win in these polls, they had to be adjusted to the point where they say a fair number of Democrats didn't even bother to vote! I find that hard to swallow.

[Peace Patriot]

85. This discussion is infuriating for several reasons:

1. The Democrats blew the Republicans away in new voter registration in 2004, nearly 60/40, indicating great enthusiasm for Kerry among new voters (who in fact voted overwhelmingly for Kerry) and great enthusiasm on the part of Gore 2000 voters, who obviously were pushing all their non-voting household members, family, friends and co-workers, etc., to register and vote in this all-important election. In addition, most Independents voted for Kerry, and most Nader voters voted for Kerry. So...where did Bush's margin of victory COME FROM? Cheney and Rove say they had an "invisible" get out the vote campaign. Uh-huh. Much like all those phantom Bush voters that the UC Berkeley statistical team found in Florida Democratic counties (130,000 to 260,000 phantom votes for Bush in the electronic voting portion of the vote). "Invisible," my ass. All is "invisible." All is spectral, all is phantom, all is shadow--just like this nonsense about "reluctant" Bush exit poll responders!

2. The FACT that both national and state exit polls said Kerry won WAS KEPT FROM THE AMERICAN PEOPLE! Deliberately, fraudulently, lyingly! This election CRIED OUT for verification--and instead these news monopolies and their hired consultant Edison-Mitofsky fiddled the numbers on everybody's TV screens on election night, "adjusting" them to fit the official tally, and creating the ILLUSION that Bush won the exit polls! Damnable journalistic crime, that! Damnable and diabolical!

3. The BASIC DATA that shows a Kerry win of the exit polls HAS BEEN KEPT FROM THE AMERICAN PEOPLE and from ALL BUT A FEW SELECT RESEARCHERS--despite specific requests by numerous credentialed researchers AND by a senior Congressman--John Conyers--who is investigating the election.

4. This person with the handle Febble has been colluding with these bastards--who LIED to the American people, and who are withholding vital information to this day!

5. In addition to all this crime and collusion, the Bush Cartel-controlled electronic voting companies ARE ALSO saying that we, the American people, the voters, CANNOT SEE the SECRET, PROPRIETARY programming code by which they counted--or didn't count--all our votes. We are arguing about this crap in California right now, this week. SECRET, PROPRIETARY PROGRAMMING CODE!!!

And more Iraqis and US soldiers are dying in Iraq! And more people are being tortured in Guantanamo Bay and in secret prisons around the world. And our government is being driven right off the cliff of bankruptcy.

Meanwhile, 60% to 70% of Americans oppose every major Bush policy, foreign and domestic--including 63% who are opposed to torturing prisoners UNDER ANY CIRCUMSTANCES--and Bush's approval rating, low all this year and last, has sunk to barely above 40%!

Where is the Bush Cartel's support? Where is it? It's as phantom as their delusionary margin of victory. And our job is to bring that fact home to our fellow citizens! This regime does not represent the majority of Americans, and does not reflect the "consent of the governed." The facts point overwhelmingly in that direction. And they would do so even if the exit polls didn't show a clear Kerry win--which they do!

Febble has pulled us off course--to argue how many angels fit on the head of a pin! And on that stunning point--the phantom rBr--is apparently trying to derail the vitally important work of USCV, TIA and others, instead of helping to rip the mask off this terrible crime against democracy and humanity, and find some way to get our country back.

That's how I see it. I don't believe in censorship as a rule. So "Febble" can come, or she can go--I don't care. What I DO care about is that this discussion take place IN CONTEXT. IN REALITY. IN THIS DARK NIGHT OF AMERICAN DEMOCRACY, with fascist rule descending upon us.