RETURN OF THE CLINCHER: Kerry wins 120 of 120 scenarios

[TruthIsAll]

I posted this months back, but it needs repeating

Assume that 100% of Bush 2000 voters still alive (48.7
million, or 39.82% of 122.26) turned out to vote in 2004.

Assume that the Final National Exit Poll "How Voted in
2000" weighting for Gore (37%) was correct. In fact,
let's assume that ALL Final National Exit Poll (13660
respondents) statistics are correct. Remember, the Final was
the ONLY poll Bush won.

Even assuming all of this, Kerry still wins by 50.22%-48.4%, a
2.23 million vote margin. See Case # 14 and the Final Exit
poll column (Kerry 54% of new voters)

Kerry won each of the 8349, 11027 and 13047 respondent
National Exit Poll time lines (3:39pm, 7:33pm, 12:22am,
respectively) by 51-48%

KERRY IS THE WINNER OF ALL 120 TURNOUT SCENARIOS IN WHICH WE
ASSUME CONSTANT 100% BUSH 2000 VOTER TURNOUT AND DECLINING
GORE VOTER TURNOUT (100% TO 88.6%). HOW COULD THIS BE? 

HERE'S HOW:
ACCORDING TO THE FINAL EXIT POLL (WHICH BUSH WON), KERRY WON
54% OF NEW VOTERS.

EVEN IN THE WORST CASE SCENARIO (CASE 20) IN WHICH WE ASSUME A
MISERABLE 88.6% Gore TURNOUT (35.50% OF 122.26), KERRY STILL
WINS BY 49.66% - 48.94%. THAT'S AN 870,000 VOTE MARGIN. See
Case# 14 and the Final Exit poll column (Kerry 54% of new
voters)

					KERRY NATIONAL VOTE SCENARIOS					
					New Voter Share and Gore 2000 Voter Turnout					
					(Assume 100% Turnout of Bush 2000 voters)					

	Assumed		Prelim.	      Prelim.			Final		
	Turnout	Turnout	7:33pm	      12:22am		       2:05pm	         
                                              
	Bush	Gore	11027	       13047			13660	Bush   Gore

			KERRY SHARE OF NEW VOTERS							
Case			59%	58%	57%	56%	55%	54%		

1	100.0%	100.0%	52.28%	52.11%	51.94%	51.77%	51.59%	51.42%	39.82%	40.25%
2	100.0%	99.4%	52.20%	52.03%	51.85%	51.68%	51.50%	51.33%	39.82%	40.00%
3	100.0%	98.8%	52.12%	51.95%	51.77%	51.59%	51.41%	51.23%	39.82%	39.75%
4	100.0%	98.2%	52.04%	51.86%	51.68%	51.50%	51.32%	51.14%	39.82%	39.50%
5	100.0%	97.6%	51.96%	51.78%	51.60%	51.42%	51.23%	51.05%	39.82%	39.25%
										
6	100.0%	97.0%	51.88%	51.70%	51.51%	51.33%	51.14%	50.96%	39.82%	39.00%
7	100.0%	96.4%	51.80%	51.62%	51.43%	51.24%	51.05%	50.86%	39.82%	38.75%
8	100.0%	95.8%	51.72%	51.53%	51.34%	51.15%	50.96%	50.77%	39.82%	38.50%
9	100.0%	95.2%	51.64%	51.45%	51.26%	51.07%	50.87%	50.68%	39.82%	38.25%
10	100.0%	94.6%	51.56%	51.37%	51.17%	50.98%	50.78%	50.59%	39.82%	38.00%

11	100.0%	94.0%	51.48%	51.29%	51.09%	50.89%	50.69%	50.49%	39.82%	37.75%
12	100.0%	93.4%	51.40%	51.20%	51.00%	50.80%	50.60%	50.40%	39.82%	37.50%
13	100.0%	92.8%	51.32%	51.12%	50.92%	50.72%	50.51%	50.31%	39.82%	37.25%
14	100.0%	92.2%	51.24%	51.04%	50.83%	50.63%	50.42%	50.22%	39.82%	37.00%
15	100.0%	91.6%	51.16%	50.96%	50.75%	50.54%	50.33%	50.12%	39.82%	36.75%

16	100.0%	91.0%	51.08%	50.87%	50.66%	50.45%	50.24%	50.03%	39.82%	36.50%
17	100.0%	90.4%	51.00%	50.79%	50.58%	50.37%	50.15%	49.94%	39.82%	36.25%
18	100.0%	89.8%	50.92%	50.71%	50.49%	50.28%	50.06%	49.85%	39.82%	36.00%
19	100.0%	89.2%	50.84%	50.63%	50.41%	50.19%	49.97%	49.75%	39.82%	35.75%
20	100.0%	88.6%	50.76%	50.54%	50.32%	50.10%	49.88%	49.66%	39.82%	35.50%

[LightningFlash]

1. Amazing! The teeth are nailed shut on this one.

[jsamuel]

2. vote switching made ez

http://comp.uark.edu/~jsamuel/ohioElectionProblem.jpg

and

http://comp.uark.edu/~jsamuel/ohioElectionHamilton.jpg

[autorank]

3. RIP "nay-sayers, doubters, quibblers...take a hike!

But I'm sure the Republicans will want one more run.

This is it. There are no more excuses.

[kster]

4. kick.nt

[Name removed]

5. Deleted message

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

[LightningFlash]

6. Read the full reports right here.

http://exitpollz.net

Independants breaked for the challenger 60/40, according to the tallies. Independants are not included in this pie chart.

[Name removed]

7. Deleted message

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

[TruthIsAll]

8. Scenarios of Kerry's vote margin of victory

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x350526

[OnTheOtherHand]

9. I'm still waiting for an clarification

of whether I am supposed to take such arguments seriously, or whether they are just intended to rally the troops, and I should leave well enough alone.

As it stands, the post pretty much refutes itself: "...let's assume that ALL Final National Exit Poll (13660 respondents) statistics are correct." Obviously you aren't assuming that, since you are seeking to refute the statistic that says that Bush won. (And of course you and I agree that E/M had to reweight the responses in order to obtain that result.)

So the reader is left with the riddle of figuring out which statistics you are accepting, and which not. I haven't figured it out yet.

You stipulate (for purposes of argument) that 48.7 million Bush 2000 voters turned out again in 2004.

You stipulate that 37% of 2004 voters were Gore 2000 voters. Using the figure of 122.26 million presidential votes in 2004, that's 45.24 million Gore 2000 voters.

You stipulate, I guess, that voters who didn't vote in 2000 (new voters plus lapsed voters?) were 17% of the electorate and that they split for Kerry 54-45. And I guess you also stipulate that the 3% who reported "other" for 2000 vote split for Kerry 71-21. (These numbers from the final weighted national results.)

And I guess you stipulate that the Gore 2000 voters split 90-10 for Kerry, and the Bush 2000 voters split 91-9 for Bush, as in the final weighted results.

So, we have, in millions:

48.7 Bush 2000 voters: 44.32 Bush, 4.38 Kerry
45.2 Gore 2000 voters: 4.52 Bush, 40.71 Kerry
20.78 did not vote 2000: 9.35 Bush, 11.22 Kerry
3.67 voted "other" 2000: 0.77 Bush, 2.60 Kerry

Totals: 58.96 Bush, 58.92 Kerry (K picks up an extra 0.01 from rounding error).

Why these figures don't agree with yours, I cannot tell.

Now, why don't those numbers agree with the official result that Bush won by about 3 million votes? Have you used E/M's assumptions to refute the official results? Assuming that E/M's pocket calculators work, we can safely infer that you are making some other assumption contrary to E/M's. And that assumption, of course, is in the number of people you are putting in the "Bush 2000" category. The exit poll puts 43% of voters in that category. 43% of 122.26 would be 52.57 Bush 2000 voters instead of 48.7 -- and those extra 3.87 million voters, breaking 91/9 for Bush, net him over 3 million votes in margin. Of course this exercise doesn't prove anything, except that the exit poll results are weighted to the official returns, which we knew in the first place.

So we are back to the argument about the 43%/37% recalled vote.

You can always twiddle assumptions, turn cranks, and generate as many "scenarios" as you like, but you can't refute the weighted poll while accepting its assumptions, unless there is an actual calculation error. It is considered good form to specify which assumptions you are challenging and why.

The 43/37 debate was worth having, and might even be worth having again. But burying that debate in a pile of "scenarios" does not seem useful to me. Except perhaps, again, as a means of rallying the troops.

[Land Shark]

10. While "inconsistent", there's nothing wrong with accepting final exit poll

(I'm responding to the first 3 paragraphs of otoh directly above.)

While "inconsistent", there's nothing wrong with accepting final national exit poll numbers as true in order to show that various internal inconsistencies or impossibilities are therefore necessary to support them when accepted as true, and thereafter using those same inconsistencies or impossibilities as evidence that the final national exit poll numbers showing bush victory (which otoh expressly agrees in the above post had to be changed to show a bush victory) are not correct because they can not be a valid reweighting process leading to MORE accuracy when subject to such unlikely internal tensions and inconsistencies.

Lawyers do this "assuming as true" process all the time for purposes of motions to dismiss: the court assumes that all allegations in the complaint are true, then the court can only dismiss the complaint at the motion to dismiss stage of the process if, after being accepted as true, the allegations of the complaint still fail to state a cognizable legal claim (i.e. fails to make legal sense).

General Form:
If X is true, then Y and Z necessarily follow as being true. But Y and Z can't be true, therefore X is likely not true either, (or is not true at all).

You can quibble with details of this style of argument, but you can't allege that one must either accept the exit poll numbers or reject them, and that assuming them as true for purposes of analysis is somehow confusing or illogical. And that you're still trying to figure out (in effect) which side TIA is on.

I think it was Keats who coined the term "negative capability" referring to the ability to hold two opposed ideas in mind at the same time as being one of the hallmarks of intellect. Indeed, interplay between two or more forces in conflict is necessary to understand the world and both of these forces in tension are "true", it is not "one or the other". Here, TIA goes one step further by showing that these two "forces" as it were can not logically exist in the presence of each other, so we need not accept the simultaneous co-existence of final national exit poll numbers showing * victory and the data it is claimed to be derived from.

It doesn't appear to me that TIA in any way needs to choose "one or the other" in terms of either rejecting final national exit polls results, or accepting them provisionally and for purposes of debate only.

[OnTheOtherHand]

11. I don't object to the fact that he uses assumptions

I object to the analytical incoherence of the argument.

If we assume that all the NEP statistics are true, then there is no internal contradiction (barring an undemonstrated calculation error).

I tried to reconstruct the assumptions that TIA actually used in his analysis, and couldn't. So I cannot know specifically which of those assumptions vary from NEP's.

TIA's results are only as good as his assumptions, and if we can't even identify the assumptions -- much less consider whether they are better or worse than NEP's -- then we are nowhere.

I get the impression from your post that TIA has evoked the appearance of an internal contradiction in the NEP figures that (as far as I can tell, and as the rest of my post attempted to illustrate) doesn't actually exist.

I assume that some of NEP's assumptions are indeed measurably wrong. It is my impression that the way in which they have weighted to the official returns will distort some of the crosstabs. Unfortunately, even if I am right, it doesn't shed any light on why the unweighted results differ from the official returns.

[TruthIsAll]

20. Do you object to this graph?

[OnTheOtherHand]

25. a picture buries a thousand assumptions

or perhaps fewer -- but I don't buy pigs in pokes or .PNGs.

I believe this picture depicts the assumptions I have challenged elsewhere in the thread.

[Name removed]

12. Deleted message

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

[Name removed]

13. Deleted message

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

[Name removed]

14. Deleted message

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

[Name removed]

16. Deleted message

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

[autorank]

17. Ahoy!!! Here substantive analysis (from you):

febble on TIA's argument (Hobson's Choice): "This is a good argument. It makes mathematical sense." Full text and link below.


Febble (357 posts) Sat Jun-25-05 05:00 AM
Response to Reply #41
49. OK, to put the record straight:


Edited on Sat Jun-25-05 05:21 AM by Febble

I referred to TIA on DKos:

http://www.dailykos.com/story/2005/6/22/183230/208

so it is only fair that I comment here.

Here is where I agree with TIA:

I agree that his table showing the "12.22" exit poll responses shows a) Kerry winning and b) proportions of respondents who say they voted for Bush and Gore respectively that match the proportions that actually voted for Bush and Gore in 2000.

One possible interpretation therefore, is that the poll responses accurately reflected both the proportions of votes in 2000 and the proportions of votes in 2004.

I also agree with TIA that his table showing the "1.25" responses shows a) Kerry losing and b) proportions of respondents who say they voted for Bush and Gore that do not match the proportions who actually voted for Bush and Gore in 2000. Moreover, I agree with TIA that not only are the proportions wrong for the 2000 vote in the second table, the proportions are impossible, as it that implies more repeat Bush voters in 2004 than actually voted for Bush in 2000.

One possible interpretation of this, therefore, is that the responses have been reweighted to match the results (we know this was done), but that the fact that this reweighting makes the plausible number of Bush repeat voters implausible (and in fact, impossible) suggests that the vote count used for the reweighting must have been fraudulent.

This is a good argument. It makes mathematical sense. And I agree that it is suggestive that the Gore/Bush proportions in the early table actually match the vote-count Gore/Bush proportions, a point in favour of their veracity. So far so good.

Here is where I diverge from TIA's thinking:

Suppose there is a discrepancy between the count and the poll, as there was. And that this was either due to the fact that Kerry voters had been polled at, say 1.12 times the rate of Bush voters, or that a proportion of Kerry votes had been switched to Bush (I haven't worked out the proportion, but I know TIA has).

When the precinct results start to come in, the E-M computer program will start to discover this, and reweight the projections in line with the precinct results. I do not know how this is done, only that it is done, and I know this from the E-M FAQ:

[Febble]

18. And to quote what I said further on in that post:

One reason for this nonsense, could, I agree, be fraud. However, I see two alternative explanations that I find plausible, though I respect your right to disagree.

One is that no-one knows whether the weightings were correctly applied - my guess is that they would be optimized to constraints more binding (like gender) than past vote, as past vote is a notoriously unreliable measure. However, there are almost certainly multiple solutions to the possible weights that could be applied to match the poll data to the count data (and I agree - who could not? - that they were matched: they were designed to be matched) and some may give a more veridical match to the Gore/Bush proportions. So the final Gore/Bush proportions may simply be wrong, even though the estimate of the vote count now matches the vote count. Weighting solutions may exist that do a better job of approximating the actual Gore/Bush proportion.

But secondly, I do not believe the Gore/Bush raw responses are necessarily accurate. I find it plausible that the initial proportions of Gore/Bush respondents only coincidentally matched reality. I am not saying this is so, just that it could be so. In other words it is possible that the final table tells you something real about what people who voted for Bush in 2004 wanted to think about they way that they had voted in 2004. And that some Bush voters in 2004 wanted to think they had voted for Bush in 2000, whether they had in fact voted for him, or whether in fact they had voted at all. We know, from good research, that people do this.

However, I do not think this is necessarily the case. I think it may simply be an artefact of the way the weighting was done.

But it means that although I follow your mathematical reasoning, I cannot agree that Gore/Bush proportions in the final table are incontrovertible proof of fraud, particularly in the light of evidence that bias was greater where random sampling protocol was most likely to be compromised, for example, where interviewing rate - the rate at which voters are selected for interview - was low.

And I think to assert that it is incontrovertible proof is misleading. It is contingent on a notoriously unreliable measure of past voting behaviour from a minority subsample of the participants in the poll, and moreover, the final Bush/Gore proportions diverge from what is possible in exactly the way that one would expect - that people like to report have previously voting for the person they are currently voting for. I think it is known as the "consistency bias".

However, these are behavioural arguments, not mathematical. I do not dispute your math.

[autorank]

22. "... behavioural arguments, not mathematical. I do not dispute your math."

That's why I included the link. That's my favorite febble line. The behavioral arguments are up for grabs but there are inferences that can be made.

One of the strongest inferences is that "they've got something to hide." Land Shark will understand this from his court room experience but it's axiomatic for the masses.

If EM and the Network consortium have "nothing to hide" they would release the EM data in a verifiable way (to let us know it's really what was collected). They would do this in a way that protected any confidentiality concerns (which happens all the time in other fields and is done well). People would examine the record, which really belongs to us just like a picture belongs to us when someone takes it without our invitation.

Then we would know the truth. But the networks and EM will not release and open up the data to full, intense inspection. It's our election after all, if it hadn't occurred, they would have no product.

Open the files, free the data. You can support that as can OTOH (actually I think he has).

Why don't the two of you write a public letter to EM and the Network consortium, publish it in the NY Times, and we'll settle this?

[Febble]

23. Well, I'm glad you liked it

I'm a behavioural scientist after all.

And behaviour is what is being measured here. Moreover, because we are not analysing the results of an experiment (no random assignment to a manipulated variable) we cannot infer causality. Our only clues to causality lie in what we know about behaviour.

Regarding release of data: the data is released. The analyses that can be done on the data are limited because it really is the "raw" data. But it certainly it tells us what was actually collected - every response is logged. Anyone can download it and check the numbers. What is not there are precinct identifiers, as you know. But you can check both the raw responses and the final weights (they are given).

This is because of the confidentiality issues you mention. And yes, there are ways it could be made available without compromising confidentiality, as we know, as it was done for Ohio. However, it is not just a matter of "releasing" it - to prepare "blurred" data that is usable for statistical analysis is an intricate process - and therefore takes money. Raising money for further blurred data sets to be prepared might be worth doing, and I would certainly support such a venture.

Regarding your photograph analogy: I think there is a somewhat ironic flaw. The "photograph" was NOT taken without permission - people were free to respond or refuse the poll. Those that responded did so on the understanding that the information would be treated in confidence - as all survey data is. It is precisely because the exit poll responders responded on the understanding that their response would be kept secret that the ethical guidelines do not permit simple "release".

I agree that in your lousy electoral system the only audit you have is the exit poll, and it is therefore of great public importance. But it is still a lousy auditing tool, and frankly, even paper audits are lousy if people like Kenneth Blackwell prevent a legitimate recount. So you not only need an auditable voting system, you need a some kind of mandatory audit, as Bill Bored and others have been discussing.

The sad thing to me, as a Brit, is that we have exit polls, and no-one takes them seriously. They are essentially entertainment. And the reason we regard them as entertainment is that no-one in the UK doubts the veracity of the count. So if the exit polls are off, it's a hoot. And yes, they are recalibrated when the first results come in (much faster than yours, despite the hand counting), and watching the "Swingometer" veer wildly when the first correction is made is one of the highlights of election night. So we know exit polls can be off. (They were, tragically, in 1992, when we thought Neil Kinnock had won, and John Major ended up with a majority of 23. November 2004 was like re-living that nightmare.)

And I understand that you guys HAVE to take them seriously, because you can't rely on your count. Dammit, it was why I got involved in the first place, and why I have spent the last year trying to figure out what went wrong. But I can't share the confidence of some on DU that exit polls are anything other than a very blunt instrument with which to dissect an election. Any measure of human behaviour is fraught with error that has nothing to do with the kind of error you get in nice mathematical binomial models. And one of the sources of error is they way people report their own behaviour.

[autorank]

33. Free the data immediately. It belongs to the people!

I'm a big fan of the behavioral sciences for what they do for people. I'm not a fan of the research that is produced, heuristic at best (and that's about all, but often enough and certainly necessary).

If money is all it takes to release the data then I've got the source: HAVA. That's aka the Congressional Vendor Extortion Act (CVEA "see-va") that gives vendors money and said vendors then turn around and give it back to politicians. There's plenty of money around so then you favor this. That's progress.

My photograph analogy related to the permission to use. That's the point, CM and the network consortium do now have permission to use and then withhold public data. There is enough evidence (ask Land Shark directly if you don't believe me) to get this into court. This isn't about a "reporters" right to withhold data. They don't have permission to keep it under wraps. They've not yet been compelled so their voluntary withholding damns them for poor ethics and a disregard for their duties as citizens. The data must be freed despite the absence of one scintilla of public duty on the part of the EM-network group is just the first step; the next will be to compel them. These with holders can have a nice chat with Judith Miller in the Alexandria lock up if they defy an order on this one.

Re: our "lousy electoral system"--I like GB and have close family who are citizens of the realm. I wish that there was fraud in your election. It's the only redeeming explanation for re-electing the war criminal, Tony bLiar.

Exit polls were good enough for the Ukraine, they're good enough for the USA.

[Febble]

39. "voluntary withholding"

means conforming to AAPOR ethical guidelines. They don't have permission NOT to keep it under wraps. But like I said, the raw data is released. There are files for each state so you can do state-level analysis. Only the precinct identifiers are missing. You can download it yourself, here

ftp://ftp.icpsr.umich.edu/pub/FastTrack/General_Election_Exit_Polls2004/

When I get permission to use confidential data for a study, what I can do with it is strictly limited. I can publish my analyses, but must not do so in such away that any respondent can be identified, and I have to carefully secure all material that could identify the respondent. If I did not agree to this I would not get permission from my ethics committee to conduct the study. Same applies for the exit polls.

And I voted Labour at the last election. I have my reasons.

[autorank]

49. Raw data, entirely down to precinct level. The data released argues FRAUD

That's what's been done here--a hard look at the state and national exit polls,including those not meant to be seen. Release it at the precinct level.

Since you have not agreed to write your letter to the NY Times demanding a release and a donor (how about the networks, they use the public airwaves for free), why not come up with a figure to get the data ready. Oh and include the cost of an independent auditing firm to make sure it really is the data recorded (that can be done through forensic accounting, auditing techniques and is vital). You get the figure and we'll circulate it. In all good faith, the networks should pay for this.

As for this "And I voted Labour at the last election. I have my reasons." Tony bLiar is stilll vile little man with blood all over him; sorry to say, whether you had your reasons or not. He should be removed as leader. But enough on that, your position is clear.

[Febble]

52. I voted for my Labour Party candidate

and for the party that had the most anti-war MPs (my own candidate voted against the war). It happened to be Tony Blair's party. As a Labour Party member I'm doing what I can to get rid of him. We do not vote for prime ministers in the UK, we vote for parties. If the Labour Party gets rid of Tony Blair as leader we will have a new prime minister. It's a struggle, but we are working on it.

The raw data is released at precinct level. It's released at voter level. Take a look. Demographics, everything. Even precinct numbers, so you know which voters belong together. The only thing missing is the actual precinct ID, so you can't match the responses to the precinct vote totals. But you can match the whole thing to the state totals.

Not sure what you mean about "those not meant to be seen". If you mean the screenshots, similar data is in the E-M report, making it clear that there was a discrepancy between the projections and the count.

But no, I will not write to the NY Times demanding "a release and a donor". I have given you reasons why I do not support release of un-blurred data. And I am not in a position to do the budgeting you suggest for preparation of a blurred data-set (I don't even know what a dollar is worth). Ohio was the important one, and that's been done.

But why don't you write yourself?

[autorank]

56. I have written the NY but YOUR letter would get published.

Precinct ID -- there will always be a question if things are held back (expert personal identifier). The Precinct ID allows a serious look at election fraud -- guess that's why it's being held back. The Networks and EM don't seem to give a damn about election fraud. They can prove us all wrong and release this.

My request to you was give an estimate of what it would cost to "blur" the data to meet ethical, etc. standards...give in pounds, I'll do the conversion for you. I wrote the Times. They didn't publish the letter. Since you've been in the media, go ahead and do it along with the cost and the methodology to get it released.

It's all about transparency and open and honest elections.

[Febble]

85. You misoverestimate me

I'm just a British psychologist (resident in Britain) who cares about the way the leader of the most powerful nation on earth is elected. I have no idea what it would involve or cost to blur the data, only that it would be a fairly intricate process if the data was to remain usable.

And I repeat: there is nothing inherently suspicious about the unavailability of the precinct IDs. What is available is quite extraordinarily detailed demographic data on the respondents. This makes it a very valuable dataset in itself. But it means that if the precincts where to be identified, many voters could also be identified. Therefore, the data needs to be blurred if it is to be released.

The fact that a blurred set has already been released indicates that there is no problem in principle.

[autorank]

87. Well, I'll save this post for sure. "no problem in principle."

The reason I said they'd publish your letter is because, under your real name and this one, you've been in the Washington Post, maybe not the NYT. So write the Post already. They'll remember you. And, febble, I swear on my favorite cat's life that you know the precinct level data is absolutely essential to answer the final questions answerable by the exit polls. Please, for the sake of my cat and all that is good and proper, don't let me down on this one. Besides, you said there is "no problem in principle."

[Febble]

89. Cute cat

reminds me of a friend of mine. And his life is safe - you can reassure him regarding my views on the precinct level data by referring him to my paper here:

http://www.geocities.com/lizzielid/WPEpaper.pdf

In order to establish the proportion of variance in differential nonresponse that can be accounted for by known independent variables, the use of an unconfounded index of bias as the dependent variable, such as the one described, would seem to be essential. In the case of the 2004 exit polls, it would therefore be of interest to know what proportion of total variance in genuine within-precinct “bias” could be accounted for by the factors postulated in the E-M report, and whether, after thus accounting for known methodological factors, any precincts/states proved to be statistical outliers that might indicate the possible contribution of vote-count corruption to the exit-poll error.

It's a bit dry, sorry about that, but most cats are dry academic types in my experience. Unlike dogs.

But as I've said, auto, I'm not going to write to anyone to request the release of data I do not think should be released, unblurred.

(And when was I in the Washington Post?)

[autorank]

119. Blurred will do with appropriate verification. I'll save this one too.

[Land Shark]

62. the only secrecy even arguably necessary is not connecting voters to votes

anything more than that is not a requirement of democracy and arguably the needs of democracy should prevail. (perhaps somebody's professional "ethics" require a secrecy more than anonymity of data reporters but even if so that "extra" conflicts with and ought to yield to the more fundamental ethics of democracy)

Besides, i can't imagine an exit poll subject demanding anything more (if they request anything at all) than secrecy of their ballot provides. The survey form can be easily designed to make it impossible to connect a particular voter to a particular voting survey form such as (for starters) not collecting their name.

[OnTheOtherHand]

65. they would need to stop collecting demographics

which is not unthinkable (apparently the British exit polls don't), but we can't do it retroactively for the 2004 election. The Census Bureau, Federal Reserve, etc. have extensive policies and procedures regarding the blurring of individual data so that people can't be identified by their demographics. Precincts are small. And the demographics have already been released (as in prior exit polls), so they can't be blurred retroactively.

As I say in my other post, if you can explain to me how releasing the raw data will save democracy -- or even materially help -- then let's talk about it. I am not a privacy purist.

[Name removed]

77. Deleted message

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

[OnTheOtherHand]

92. I think you misunderstand my point

I am not asking you to prove the case before seeing the data. I am asking you to explain how the raw data (as opposed to the data already released for all states, or the blurred Ohio data that ESI analyzed) could conceivably be used to prove the case -- or even to advance the investigation. If no one can actually answer this question, then I see no grounds for releasing the data. Or, to put it differently, the only grounds would be the abstract (and controversial) premise that the data are inherently public.

If someone _can_ answer this question, then let's discuss it. But why are we so hung up on the inherently inconclusive exit poll data, when what we really need is verifiable ballots? I don't get it.

[Melissa G]

129. The reason we are hung up on it is that we believe that it

would prove that the less than savory crowd that are currently resident in the white house (although they spend a ridiculous amount of time on vacation in Crawford) were not in fact EVER elected by the American people. That sticks in the gullet of many of the folks who live in this forum. This CROCK of S### that we have been handed by what passes for the Press in this country that promotes the LIE that B### was actually elected president is a falsehood that needs rectifying!!!
Check my tagline.. I believe we need verifiable ballots as much or more than the next person. BUT this LIE of BUSH winning any Presidential popular vote NEEDS TO BE ERADICATED!!!

[OnTheOtherHand]

155. Melissa G, I think you are missing my question

I understand that you believe that the raw exit poll results would prove that Bush stole the election.

But I do not understand how you would use the raw exit poll results to prove this.

By the way, this isn't the same question as whether Bush stole the election (or, maybe more to the immediate point, stole the popular vote?). Assuming that he did, how could the raw exit poll results be used to prove this?

We know that the exit polls differ from the official returns. We know that some people think the discrepancy is due to fraud, while others think that it was due to bias (and many don't know what they think, or opt for some combination of fraud and bias, etc. etc.). Will access to the raw data resolve that debate, and if so, how? Or, in what other way will access to the raw data be important?

If I am misframing the question(s) because I miss your point, of course please answer the question(s) I should have asked instead.

[Melissa G]

163. Hi OTOH,

You got it mostly right.
I am not a Quant but as you know I talk to Quants and they have told me they believe it would be useful and it make sense to me that it would be useful. I like facts. I like holding them up to the light and seeing what they reflect. Certain people see more in numbers than i do but i see a lot in the concepts that those folks interpret. I don't know if we can accurately anticipate in advance what we can see from looking at the data that we don't yet have access to but I am sure it would be interesting and likely very useful one way or another.
Every bit of data we have looked at so far has been worth glancing at though good folks may differ in interpretation of the various weight of worth.
Is this not what most academics spend their life doing?

The question that has been in my mind since the night of the election
is where did the 9 to 11 million votes Bush could not possibly have gotten come from? (I suspect thin air and programming of some sort or another but not real live voters) I believe the release of the raw blurred data could shed light on this. edit sp

[OnTheOtherHand]

182. OK, speaking as one of that strange tribe...

It is hard for me to think of data that I would not regard as potentially useful. I want it all. And actually, that may be the Achilles' heel of many of the quants who are asking for fuller disclosure. They want it all, but they haven't worked out specifically why they want it all; they more feel in their guts that somehow they could make it prove something.

I'm reasonably certain that some Democrats and liberals work for E/M, so if there were really a conclusive insight into fraud there, I think we would know by now. ("Insight" -- conceivably evidence, but even a strong indication of how or where to look.)

I think the San Diego case illustrates one of the limitations here. They know there is a discrepancy between their survey and the official returns in 11 precincts. In their case, they know which 11 precincts. But knowing the 11 precincts doesn't seem to help very much.

I would like someone to come up with the money it would take to compile "blurred" data for at least Florida and New Mexico as was done for Ohio. It bothers me, however, that there hasn't been much comment on the Ohio analysis. It says to me, as I think I predicted a few months ago, that a handful of people may be convinced and everyone else will go on believing what they previously believed.

[Melissa G]

188. Hi OTOH,

Link on San Diego please...
I have been working harder than normal in my other nonpolitical life and have not followed all that discussion.

It takes a long time to compile a cogent exploration of anything this is likely the case with the Ohio data. I know from following Freeman's work how hard it is to document all of this data. (His book will be out soon though.) Unfortunately, there are not enough folks looking at the issue of Election Frawd and that is only part of the reason some media attention and exploration would be helpful!

[OnTheOtherHand]

191. I know the feeling(s)!

of being behind on discussions, and of how hard it is to compile a cogent exploration of anything.

On San Diego, try
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=389521&mesg_id=389521
""parallel election" WOW ! San Diego"

I think there is another thread on this, but that happens to be the one on the main page right now -- call me lazy ;)

I have problems with the SD study, but it does at least raise the question how we know the vote count is accurate (oh... we don't... oops!).

[Land Shark]

96. the standard of preserving privacy is not whether or not a dedicated

person could conceivably, in a few cases, determine identity by combining demographic data in small precincts....

A person can find anyone's social security number if they really want to. Yet that's private too. THe demographic data is not a bar to disclosure unless someone Wants it to be.

[OnTheOtherHand]

156. the standard for privacy

per AAPOR ethical standards, is:

"Unless the respondent waives confidentiality for specified uses, we shall hold as privileged and confidential all information that might identify a respondent with his or her responses."

Seems reasonable to interpret that fairly strictly when people's votes are at issue.

Of course, by that same token, if someone can come up with a reason why fuller disclosure will make a huge difference, then I might support legal action to compel fuller disclosure. Haven't seen one yet.

[anaxarchos]

116. I am really glad you brought this up again...

It stuck in my head the last time you said it but to have questioned it would have diverted the discussion.

The issue is "the final Bush/Gore proportions diverge from what is possible in exactly the way that one would expect - that people like to report have previously voting for the person they are currently voting for. I think it is known as the "consistency bias"."

I've never heard of that. Could you educate me please...

Also, what is "consistency bias" in this context? I've heard of it in econometrics (where it refers to something completely different) but not in polling. References please?

[Febble]

118. In social psychology

"consistency bias" is sometimes used to refer to a tendency for people to "view the past as more consistent with the present than it was"

http://www.chronos.msu.ru/EREPORTS/aultman_searching.htm (sorry, not a great cite, but it is text book stuff - the basis for the idea that our recall attempts to resolve "cognitive dissonance").

But more specifically, when asked about past vote, two tendencies have been reported: one is that voters tend to recall having voted for the winner, and the other is that they tend to recall having voted for the candidate that is consistent with their current choice.

I'm afraid I only have UK links, but OTOH has US examples.

But the detailed data of Populus polls bore out research at previous general elections, and surveys re-polling the same people during the course of a Parliament, all of which have shown that when asked after a general election how they voted, a lot of voters – possibly as many as one in five – don’t recall correctly: they may lie, or want to be seen to have backed the winner, or are correcting their past vote to match their future intention, or they may simply forget. Voters do not misremember evenly – evidence suggests that they are likely to overestimate the vote of the Labour Party while underestimating the numbers of people who voted Liberal Democrat or for smaller parties.

http://www.populuslimited.com/method_update.html

One of the important controversies about the predictive potential of opinion polls in recent years has been about the extent to which voters’ recall of how they voted at the previous general election could be used as an indicator of possible sample bias and, perhaps, as a variable in weighting. Past research has demonstrated that not only do voters misremember how they voted in the past, but they tend to do so in a way that brings their past behaviour into line with their current preference. Thus although quota polls being conducted in the 1994-96 period found an improbable proportion of the electorate saying they had voted Labour rather than Conservative in 1992, it was argued that this was the product of selective recall rather than evidence of sampling bias.

http://www.crest.ox.ac.uk/reports/report97.pdf

If we apply these two principles (that a proportion of voters will misreport their past vote in line with current vote, and that this will be more marked if current vote is for the last winner) then you would expect that those who reported inconsistent votes (Gore last time, Bush this, or Bush last time and Kerry this) would probably be telling the truth. However of those who reported consistent votes (Gore-Kerry; Bush Bush) a proportion (possibly up to 20%) may be lying about past vote, and that this phenomenon is likely to be greater amongst the Bush-Bush reporters. If 10% of reported Gore-Kerrys were actually Bush-Kerrys, and 20% of reported Bush-Bushes were actually Gore-Bushes - the numbers would be what you'd expect. Ditto if both proportions were smaller.

However, what this would mean is, as TIA points out, that Kerry lost not only around 10% of those who said they'd voted for Gore last time (and who were likely to be truthful, being inconsistent) but also a proportion of those who'd also voted for Gore last time, but reported that they'd voted for Bush in 2000. In other words that Kerry lost more than 10% of the Gore vote. While it seems he got a substantial, if not overwhelming majority amongst new voters (or those that did not vote in 2000).

So one story that would jive with the numbers is that about 10% of Gore voters who were prepared to admit they were Gore voters switched to Bush, and about 10% of voters who were prepared to admit they voted for Bush in 2000, switched to Kerry. People do that stuff. It's why we hold elections. But that wouldn't swing it. So we also have to postulate that an additional percentage of Gore voters swung to Bush and didn't admit having voted for Gore - and that this outweighed the tendency for new voters to vote for Kerry.

Plausible? The misreporting is, as I said, in line with what we know about misreporting. The voting pattern is perhaps more oodd. But that's all. it doesn't prove fraud. I'd put it in Land Shark's suggestive but not persuasive heap. And if I were a cross-examiner, I'd postulate that a non-fraud explanation is that in a polarized election, new (young?) voters were attracted to the Democrat, as predicted, but that older voters voted for security.

But check out OTOH's US links which will be more relevant.

[anaxarchos]

126. Thank you, Febble, for your prompt response....

I know nothing of "social psychology" and I hate to look a gift horse in the mouth (U.S. saying... sorry) but I was hoping to get something a little more usable than what you sent me. All I was able to find from your reference was a single sentence in a review of a book on human memory which reads : "Schacter notes five major kinds of bias: consistency, change, hindsight, egocentric, and stereotypical. With a consistency bias there is a tendency to view the past as more consistent with the present than it was...".

I'm afraid that is even less germaine than my consistency bias from econometrics.

I was specifically looking for the explanation to this: "the final Bush/Gore proportions diverge from what is possible in exactly the way that one would expect - that people like to report have previously voting for the person they are currently voting for. I think it is known as the 'consistency bias'."

No matter...

I think the rest of your post helps. Your next two "tendencies" sound strangely familiar although the wording is a bit different (might be a US/UK thing - "You say tomato and I say...", etc.). So... before we look at TIA, can we confirm that we are talking about the same thing? I think I am much better off talking about two things which I kinda recognize rather than a third thing I've never heard of.

Your first tendency is clearly the over estimation or over reporting of voting which I think is the oldest response error to be documented. What throws me a little about your reference is that in the U.S., while "misremembering", etc. are mentioned, most of the research focuses on the civic pressure to have voted (much like church attendence)... or the "salience of the civic norm of voting" as in this:

Brian D. Silver, Paul R. Abramson and Barbara A. Anderson, "The Presence of Others and Over-reporting of
Voting in American National Elections," Public Opinion Quarterly, 1986.

or this:

Stanley Presser, "Can Context Changes Reduce Vote Over-reporting?," Public Opinion Quarterly, Winter, 1990.

Are we talking about the same thing?

Your second tendency stumps me more. Judging from the time period and the "controversy" in your citation, this seems to refer to "post-election bandwagon effect" but your citation does seem to suggest "consistency" or "selective memory" which is the exact opposite of this response error as I (imperfectly) understand it. One commonly cited paper on this is:

Robert H. Prisuta, A Post-Election Bandwagon Effect? - Comparing National Exit Poll Data with a General Population Survey, SRMS/ASA 1993

available here:

http://www.amstat.org/sections/srms/Proceedings/papers/1993_194.pdf

Are we still talking about the same thing?

[Febble]

133. I'm not a social psychologist either

I'm a cognitive psychologist, so I'm sorry I wasn't able to come up with something a little more definitive in the way of a citation. Here's one from the cognitive literature.

http://psych.wustl.edu/memory/Roddy%20article%20PDF%27s/Roediger%20and%20McDermott%201995.pdf

But also somewhat tangential.

However both are germane in the sense that they deal with the hypothesis that memory is constructed - and that we tend to "remember" things that accord with our current construct of reality even if these are at odds with what actually happened.

But neither would be germane if there was not evidence that it happens with regards to past votes. Which is why I cited the UK pieces.

And my fundamental point is that we cannot assume that all reports of past vote are accurate, particularly when both consistency patterns (same guy both times) and bandwagon patterns (favoring incumbent) would tend to work in the direction of over-reporting of Bush votes in 2000, and would account for the apparent shortage of Gore voter in the survey.

What is rather harder (for me) to believe is that substantial numbers Gore voters would have voted for Bush. But frankly I find it hard to believe that anyone voted for Bush, and clearly they did. Some of them are even friends of mine.

[TruthIsAll]

144. TIA vs OTOH /FEBBLE et al: FACTS, ASSUMPTIONS, ANALYSIS, CONCLUSIONS

A Summary of the Debate: Facts, Assumptions, Analysis and Conclusions

TIA
====
FACTS:
122.26 mm voted in 2004

MAXIMUM NUMBER OF 2000 voters who could have voted in 2004:
Bush: 48.7mm (39.8%)
Kerry: 49.2mm (41.2%
Nader: 3.2mm (2.6%)

MINIMUM NUMBER OF NEW 2004 VOTERS: 21.5mm (17.4%)

National Exit Poll Timeline:
4pm 8349 respondents: Kerry 51%
Kerry % of NEW voters 59%

7:33pm 11027 respondents: Kerry 51%
Kerry % of NEW voters 59%

12:22am 13047 respondents: Kerry 51%
Kerry % of NEW voters 57%

1:25pm 13660 respondents: Kerry 48%
Kerry % of NEW voters 54%


ASSUMPTIONS:
All Bush 2000 voters (100%) came to vote in 2004.
Only 86-100% of Gore Voters came out
Kerry % of New voters (above)

SENSITIVITY ANALYSIS:
KERRY WINS ALL 120 POPULAR VOTE SCENARIOS

CONCLUSION:
IT IS EXTREMELY UNLIKELY THAT BUSH WON.
IT IS VERY LIKELY THAT KERRY WON BY 4-6 MILLION VOTES, DEPENDING ON HIS PERCENT OF NEW VOTERS


OTOH
====
FACTS
As given above

ASSUMPTIONS
A 2002 VOTER SURVEY INDICATES THAT GORE VOTERS FORGOT WHO THEY VOTED FOR. SOME MAY HAVE WILLFULLY LIED TO IDENTIFY WITH BUSH. THIS TRAIT WAS NOT THE CASE FOR BUSH VOTERS.

QUANTITATIVE ANALYSIS
None

CONCLUSION
IT IS LIKELY THAT BUSH ADDED 13 MILLION VOTES TO HIS 2000 TOTAL VOTER TURNOUT (49 TO 62 MM) AND WON THE ELECTION BY THE RECORDED 3 MILLION VOTE MARGIN: 62-59.

[autorank]

151. Rich in facts, elegant analysis.

This represent the culmination of many threads and strands of analysis. The facts are simple and straight forward. I hope we get a debate, once and for all, on the facts. Let's see what happens. Substance over style, reason over rhetoric; that would be nice. This should wrap up shortly.

[OnTheOtherHand]

184. you think?

I started to laugh when I reached, "ASSUMPTIONS
A 2002 VOTER SURVEY INDICATES THAT GORE VOTERS FORGOT WHO THEY VOTED FOR."

I've heard of assuming away inconvenient facts, but it doesn't come much more straightforward than that.

There is no prospect of closure on the debate unless TIA actually thinks about my evidence before he attempts to paraphrase my "assumptions." He might as well say, "I'm not listening! I'm not listening!"

And it is disappointing that you would cheer this as "elegant analysis."

[TruthIsAll]

187. Are you still doodling? So far, you have offered nothing in

the way of quantitative analysis.
Just talk.
Your evidence?
Talk is cheap.
You are very good at it.

anax, melissa, autorank and others are waiting for the outcome of your doodling.

Lets see some analysis.
You told anax you would doodle something - whatever that means.

How about firing up Excel, downloading some data, state your assumptions and run it through some scenarios?

I do it every day.

You've been here since April and have still not done it.
You talk a lot though. Very good at that.

Why don't you get your hands dirty and do some real work for a change?

So you are LOL.
If I were you, I would be COL.

[Febble]

195. My computer is groaning

with OTOH's Excel spreadsheets and more.

Disagree with him if you like, but don't accuse him of not working hard.

I don't think he sleeps.

[TruthIsAll]

202. Who is he crunching numbers for, you or DU?

So far.

Nada.

Febble, we may disagree on some things, but I have great respect for your efforts to communicate in a very clear, direct manner. And I know you will be there to counter in an engaging, well-thought out manner, even if we disagree.

BTW, am I wrong in assuming that if you do NOT reply to my posts, you agree with the basic content? I am surprised that you have not commented on several of my recent analyses. In particular, the post in which I applied the exit poll optimization model to state exit poll response - a sort of corollary to the original precinct-based optimization analysis which you have commented on. The results confirmed that alpha's are higher in Bush strongholds, be they individual precincts or states. Perhaps you could take a look.

In any case, the bottom line is that I show all my work in full view of DU. I don't pass the buck. Anyone can review the work. DUers who follows my posts knows that they will not get any BS or disruptive jargon from me. I happen to think numbers tell the story a lot better than words ever will.

Ok, maybe I do a little trash-talking at times, but only out of frustration with particular naysayers. That's because I recognize their MO and have no further patience for their relentless nit-picking and hypothecating with nothing of substance (I mean analysis) to back it up.

[Febble]

205. Thanks, TIA for the gracious comment

I'm juggling too many things at the moment (including two small boys in need of amusement) so no, please don't assume that silence means assent! And I've also been on holiday (southern France, since you ask, it was lovely).

But as it's hard to get much done with constant interruptions, I thought I'd drop in to see you guys again, having missed you all.

OTOH presented developments from my work at the ASA meeting, which is why my computer is full of his spreadsheets. I just wanted to scotch the notion that he was lazing around doing nothing. Alpha needs still needs a tweak or two.

I also respect the fact that you do your work in full view. It's a high risk strategy, of course, so I admire your courage. However, I don't think it's fair to assume that because others don't post tables that the tables don't exist (for a start I still haven't figured out how to format tables on DU - how do you do that?) Also tables aren't necessarily the form of communication of choice, especially to "non-quants" (although I've now figured out how to paste your output into Excel, by going into Word first and using convert-text-to-table.)

And although I realise you have categorized me and others as "naysayers" and therefore lose patience with us, I also don't think it is fair to assume that those of us who remain unconvinced that Kerry won the majority of the votes are saying "nay" to the case that the election was unfair, corrupt, and unauditable and that therefore the voting system requires radical reform.

Some of us think that the two issues need to be emphatically decoupled. The case for investigation into fraud is unassailable, whether or not it resulted in the wrong man being elected, as is the case for an auditable voting system. However, the case that Kerry won the popular vote IS assailable. As I've said elsewhere on this thread, people who are unconvinced that Kerry won the popular vote include people like Walter Mebane who has nonetheless written damningly on voter suppression in Ohio, and concluded that Gore would have won Florida in 2000 by over 30,000 if the voting system had been just. So the "naysayers" are not necessarily statistical ignoramuses or Republican shills.

If you post a link to the optimization thread you refer to, I'll take a look - though I don't promise a comment!

[Dynasty_At_Passes]

216. "and concluded that Gore would have won the vote in 2000"

Why in the world did Walter Mebane conclude this?

How could he automatically guess and assume that, if the optical scanners were used in 2000 Gore would have won the election?

I just can't see it anymore that this man was really doing his homework. A statistician analyzes all the factors, and one clear factor that was horridly overlooked and dismissed was the fact that the optical scan machine of at least Volusia County was rigged by a blamed memory card installed into the central tabulation.

Now, when I look at the results in Ohio and notice that the seals were broken on the memory cards stored in Lucas County I can not imagine how anyone, let alone a scientist, can conclude the optical scanners produced a fair count. Insecure memory cards or scanning devices, which have been tampered with, never produce accurate or fair results. If this happened all over Ohio, the popular vote as well is back at being suspect. Why has he never even DISSECTED this?

[OnTheOtherHand]

196. just a crude (but pertinent) factoid

In the 2000-04 NES panel study, among the folks who reported a 2000 vote in both the 2000 postelection and the 2004 postelection survey:

Among the 308 who had voted for Gore as of 2000, 29 of them said in 2004 that they had voted for someone else in 2000 (23 Bush, 3 Nader, 2 other, 1 DK -- you can skip that one).

Among the 309 who had voted for Bush as of 2000, 6 of them said in 2004 that they had voted for someone else in 2000 (5 Gore, 1 Nader).

It will take a bit more work to see how (reported) non-voters fit in.

[TruthIsAll]

224. Wow. 600. Big sample ( 4% MoE) NEP: 13,047 (1%); State exits:73,000 (.4%)

Sorry.

No sale.

[Febble]

225. plenty of statistical power there

9.4% of Gore voters misrecalled their vote, 1.9% of Bush voters.

Fisher's Exact Test gives a probability of 1 in 27,740 that this difference could have occurred by chance.

Looks like Gore voters really did have worse memories than Bush voters. However it is possible that they regained their powers of recall by 2004. Or not.

Or maybe it wasn't a random sample.

[OnTheOtherHand]

230. Febble, don't let him off too easy

As I have understood it, TIA's contention is that the actual proportion must be 0.

[autorank]

219. I'm so upset that you're "disappointed" in me. It is elegant and

regrettably you offer no analysis, just nipping at the ankles, a point here or there. Put you thoughts together, as Melissa G suggested (scolded?) and do a thread. Let's see what you can do.

It's one thing to play an ersatz Socrates, it's another to write The Republic.

[TruthIsAll]

215. You can postulate anything you like...

"So one story that would jive with the numbers is that about 10% of Gore voters who were prepared to admit they were Gore voters switched to Bush, and about 10% of voters who were prepared to admit they voted for Bush in 2000, switched to Kerry. People do that stuff. It's why we hold elections. But that wouldn't swing it. So we also have to postulate that an additional percentage of Gore voters swung to Bush and didn't admit having voted for Gore - and that this outweighed the tendency for new voters to vote for Kerry."

You can postulate anything you like. But this implies that you also assume that the recorded vote count was perfect. Only respondents memory is at fault. It implies that there was no fraud.
But we have much evidence which proves there was fraud.

So we are back to square one.Your hypothesis of faulty-recall is quite faulty - because it relies on a faulty (shall we say naive) assumption that Bushco did not cheat in the first place.

What is encouraging is this:
Your case has been reduced to absurd hypotheses.
Gore voters forget.
Gore voters lie.
Gore voters identify with Bush.

Bush voters don't.

This is all quite pathetic.

[OnTheOtherHand]

19. let me rephrase it

I don't see how anyone can participate usefully in serious debate while ignoring basic facts. That is, in fact, the general principle that I intended to enunciate -- I didn't intend to hand-tailor it to one person alone.

I have a real problem here, if I attempt to explain TIA's arguments to a colleague. The colleague might say, "What does TIA say about false recall?" And I will say... what? that he calls me a crypto-Republican?

Please understand that this is a real problem whether I mention it or not. Anyone can overlook a pertinent argument, or disagree about its importance, but it ought to be engaged.

EDIT: That said, I am happy to move on.

[Name removed]

15. Deleted message

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

[OnTheOtherHand]

21. I do not understand your definition of "facts"

In another thread, I wrote:

"Again, if we assume that the raw data were wrong, there is no particular reason to assume that the Bush/Gore results should be accurate. However, in the 2002 National Election Study (which interviewed both voters and non-voters), 51.6% of respondents reported having voted for George Bush in 2000, and 44.3% reported voting for Al Gore, among those who reported a 2000 presidential vote. I therefore invite you to explain why you think that the 2004 exit poll result proves fraud but the 2002 NES result proves nothing in particular."

And your response was:

"Forget polling. Deal with the facts. Answer ONE simple question. Would you calculate the MAXIMUM number of Bush 2000 voters who could have voted in 2004?"

Now, anyone can see that I posted a fact, not a hypothetical. And anyone can see that you ignored it -- and actually seem to have asked me to address a hypothetical question, instead.

If you cannot understand why the fact is pertinent (and why it does not entail that "Only Gore voters forget," or "Only Gore voters lie," or "Only Bush voters are shy"), then -- well, then at this point I don't think it's my fault.

You are giving me nothing to tell my colleagues. Not only are you strangely claiming that almost everyone already agrees with you, but you aren't giving me any way to convince them.

[TruthIsAll]

26. Ok, give us the facts about this poll.

Who sponsored the poll?
How many were polled?
Who was polled?
What were the poll internals?
What was the purpose of the poll?

Do you recall the Gallup/CBS polls which had Bush ahead of Kerry by 12 points? Do you recall the party_ID weightings in these polls?


You
"You are giving me nothing to tell my colleagues. Not only are you strangely claiming that almost everyone already agrees with you, but you aren't giving me any way to convince them."

Me
Oh, so you are on a mission to bring the TRUTH to your colleagues. And you want to convince them they are wrong?
And that Bushco stole it?

So altruistic of you.
I'm flattered that you would even try by having them look at my posts.

Well, why don't ask them to prove that Bush won?
Why don't you ask them why the raw EM data is STILL not public?

You say that I am "strangely claiming that almost everyone already agrees with me". Well, at least 90% of DUers think the election was stolen. Of course they're biased. They're democrats or independents.

Just who did your colleagues vote for, anyway?
Who did YOU vote for?

Probably, John Tenure and Bill Job and Tom Security.

[foo_bar]

31. how many DUers think Kerry won the popular vote?

Looks to be in the 14% range:

What about the other three and a half million votes that Bush won by? You know, it's -- you cannot solve this problem by just blaming the cheaters.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=132&topic_id=2008168

Re: Well, at least 90% of DUers think the election was stolen.

[TruthIsAll]

57. mea culpa... it's only 89%

foo.. you are right.
I lied.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=4062106

[mgr]

99. Foo's point, as I see it, is

If members of DU are intelligent, then they should by your criteria, overwhelmingly agree that the popular vote was stolen. They do not. These are the progressive democrats, and it appears you have not made the case to the majority of them. Say good night Dick.

Mike

[TruthIsAll]

124. You never even read my poll. 89% is a majority to anyone but you.

Oh, I know mgr, its not scientific.

But neither are you.

[mgr]

245. LOL--Bogus poll

The responses don't make you right. God, are you the partisan neophyte, or are you pulling another Zell Miller, Bishop Ussher?

Seems there's a missing element that the DNC addressed around this time. A poll is only as good as the questions asked. In this case, that is not science, it is fairness.

Choices are

1. Exit polls right & other documentation and show vote was stolen.

2. Fraud but insufficient to steal election

3. No fraud.

So, if you agree with the DNC, you either choose to not vote (no choice for this), or if you think something hinkey went on, you go with #1, although the exit polls are immaterial to your position.

You could have split it out and see if there is a difference, but you chose not to.

As to being scientific, what fruitful avenues of further investigation into fraud has your exit poll analysis and determination that fraud was on the east coast brought. You have made that point over and over the last few months, where has it gotten you, which states require further investigation?

My position that it is widespread voter suppression, should have folks looking into (among other states, I still need to refresh my memory regarding Missouri, Iowa, and Wisconsin) Colorado (Eastern Counties--communities of Aspen, Denver and Colorado Springs), and Arizona (Phoenix); as well as the connection between Jewish voting population and long lines in Palm Beach, Volusia, and Miami Dade County (possibly Pinellas). I have pointed out that suppression of the Jewish vote possibly lost Florida.

They are specific hypotheses, that can be looked into, where are yours? I will take the ad hominen attack or silence that follows to show that there are none.

Mike

[OnTheOtherHand]

35. are you aware that you are desperately changing the subject? n/t

[TruthIsAll]

95. Are you aware that you have failed to respond to # 26 and #81?

Still waiting for my questions to be answered regarding that 2002 poll.

[OnTheOtherHand]

98. I will tackle the 2002 poll in a moment

but I certainly don't accept the premise that you are entitled to ignore the 2002 poll until you run out of questions to ask.

[TruthIsAll]

81. Once again, regarding your 2002 poll, a few more questions..

"Again, if we assume that the raw data were wrong, there is no particular reason to assume that the Bush/Gore results should be accurate. However, in the 2002 National Election Study (which interviewed both voters and non-voters), 51.6% of respondents reported having voted for George Bush in 2000, and 44.3% reported voting for Al Gore, among those who reported a 2000 presidential vote. I therefore invite you to explain why you think that the 2004 exit poll result proves fraud but the 2002 NES result proves nothing in particular."

What is the point?
What information are you gleaning from the poll?

Unlike the 2004 exit poll, the 2002 poll did not have to match the vote.
It was just a poll.
Very good.
The results did not match the 2000 vote, did they?
Why am I not surprised?
So we can be 100% sure that the poll was pristine, right?
The 51.6% of respondents who said they voted for Bush was really 51.6%.

But how do we know that 43% of those polled in the 2004 National were Bush 2000 voters? You agreed that the weightings were ADJUSTED to match the vote.
So again, I ask you:
Was the 43% an actual sample?
Or was it a weighting applied to match the vote, as you have stated?

And if it was a weighting, not a polling result, are we not comparing apples and oranges?

In any case, do the 2002 results
1-change the fact that Gore got 540,000 more votes?
2-change the fact that at most 48.7mm Bush 2000 voters could have voted in 2004?
3-change the fact that 122.26mm voted in 2004?

What does it have to do with the questions I raised about the 2004 National Exit Poll?

About your 2002 poll:
Was there a timeline of votes earlier in the day which indicated a trend?
Like the first 8349, 1107, 13047 in the National Exit Poll which all showed Kerry winning?

Who commissioned the 2002 poll?
For what purpose?
How was the sampling done?
Where was the sampling done?
What organization was commissioned to do the actual sampling?
How many were sampled?
What was the sampling MoE?
What was the party_ID weighting of the respondents?
How was the poll weighted demographically?
What is the NES track record?

WHAT WERE THE POLLING INTERNALS?
ARE THEY PUBLIC?

Had enough?
I may have some more for you after you answer these.

[OnTheOtherHand]

102. "What is the point?"

The NES, the GSS, and other surveys in the United States and Britain (and probably elsewhere) routinely show a propensity for people to recall voting for the incumbent when they didn't. (Of course we can't know who voted for whom, but in panel studies we can actually see people changing their votes retrospectively, not all in the same direction, but collectively in favor of the incumbent.)

So, the 2004 exit poll's unweighted result that almost as many respondents reported voting for Gore as for Bush in 2000 actually suggests bias in the exit poll. The weighted result is more consistent with other surveys.

The exit poll could be unbiased, if turnout was substantially higher among Gore 2000 voters than Bush 2000 voters. But nothing requires us to believe that it was. Given the extreme partisan polarization of views about Bush, I imagine that turnout among Bush2K and Gore2K voters was fairly similar.

(Now I am bracing for a snark about my imagination, and how you stick to facts -- but no, you don't, because nobody has all the relevant facts. You certainly cannot say, accurately, that you have _proven_ that turnout was substantially higher among Gore 2000 voters.)

The NES and GSS are noteworthy in this respect because they are widely regarded as the best political surveys in the U.S., due to all the effort that goes into maximizing the response rate.

If you know as little about the NES as your questions imply, one might wonder how you have any business expressing opinions about survey research. Or you may know the answers to these questions, in which case one might wonder about your motives in asking them. You may wish to refer to http://www.umich.edu/~nes/ Similarly, for more information on the GSS you might consult http://webapp.icpsr.umich.edu/GSS/

I don't need to change any of the facts you cite. I only have to attempt, over and over, to explain why they don't prove your case (even in a very weak sense of "prove"). Your case depends on assuming, implausibly, that marginal results from people who _say_ that they voted for Bush/Gore/nobody in 2000 can be mechanically applied to the numbers of people who _actually_ voted for Bush/Gore/nobody in 2000. That exercise may yield interesting results, but it cannot yield evidence of fraud.

[TruthIsAll]

117. I'm afraid we are at a dead-end.

These are very simple observations.

Let B= maximum number of Bush 2000 voters who voted in 2004.
Let G= maximum number of Gore 2000 voters who voted in 2004
Let N= maximum number of Nader 2000 voters who voted in 2004.
Let TV = Total Votes in 2004

We know B,G, N and TV.
B= 48.7 mm
G= 49.2 mm
N= 2.6 mm
TV = 122.26

Therefore we can compute the Minimum number of New 2004 Voters:
NV = TV - B - G - N.

We know Kerry got .54* NV, based on the final exit poll.
We know Bush got .45* NV, based on the final exit poll.

We know Kerry got .71* N, based on the final exit poll.
We know Kerry got .21* N, based on the final exit poll.

Assume
X1 = percentage of Gore voters for Kerry
X2 = percentage of Bush 2000 voters for Bush

Total Kerry vote is
KV = X1*G + X2*B + .54*NV + .71*N

Total Bush vote is
BV =(1-X1)*G + (1-X2)*B +.54*NV + .21*N

Next step:
Calculate the percentages X1, X2 for which KV = BV.

Anyone care to try?

[TruthIsAll]

125. Two typos fixed

Let B= maximum number of Bush 2000 voters who voted in 2004.
Let G= maximum number of Gore 2000 voters who voted in 2004
Let N= maximum number of Nader 2000 voters who voted in 2004.
Let TV = Total Votes in 2004

We know B,G, N and TV.
B= 48.7 mm
G= 49.2 mm
N= 2.6 mm
TV = 122.26

Therefore we can compute the Minimum number of New 2004 Voters:
NV = TV - B - G - N.

We know Kerry got .54* NV, based on the final exit poll.
We know Bush got .45* NV, based on the final exit poll.

We know Kerry got .71* N, based on the final exit poll.
We know Bush got .21* N, based on the final exit poll. <fixed to Bush

Assume
X1 = percentage of Gore voters for Kerry
X2 = percentage of Bush 2000 voters for Bush

Total Kerry vote is
KV = X1*G + X2*B + .54*NV + .71*N

Total Bush vote is
BV =(1-X1)*G + (1-X2)*B +.45*NV + .21*N < fixed to .45

Next step:
Calculate the percentages X1, X2 for which KV = BV.

Anyone care to try?

[foo_bar]

132. I'm trying to picture the assumptions

It's x1=83.27%, x2=13.04%. Anything higher or lower makes KV <> BV (including negative values, x1/x2 is linear not calculus). Your answers on the other thread neglected the primary constraint:

If Kerry gets 2% of Bush 2000 voters,
he needs 94.20% of Gore voters to tie

KV 60916997 <> BV 60917403 (Nader's a constant for lack of Nader variable, NadV=425600 (0.08*n2000)+(0.01*nv2004))

If Kerry gets 8% of Bush 2000 voters,
he needs 88.26% of Gore
voters to tie

KV 60917107 <> BV 60917293 (almost to the tipping point)

If Kerry gets 13.04% of "Bush 2000 voters", 60917200=60917200. Assuming you don't buy into all that sociology crap.

[TruthIsAll]

147. Foo-bar.. you are out of your league.

You should know better than to venture into the world of
mathematical analysis.

Stick to what you are good at.
But, what are you good at?

**********************************************************

There is no unique solution. 

But... we can still come up with a mathematical relationship
to calculate the breakeven Kerry vote. 

Given a Kerry percentage of Bush 2000 voters, we can
determine the percentage of Gore votes he needed to tie Bush.
					
TV	122.26	Total 2004 votes	
	MaxVotes	Pct2004	
G	49.2	40.24%	Gore
B	48.7	39.83%	Bush
N	2.6	2.13%	Nader
Total	100.5	82.20%	
			
NV	21.76	17.80%	New voters
Kerry			
KNV	11.750 54%	New votes
KN	1.846	71%	Nader votes
KGV	X1*49.2	X1	Gore votes
KBV	X2*48.7	X2	Bush votes
Total	13.5964		+ X1*49.2+ X2*48.7
			
Bush			
BNV	9.792	45%	New votes
BN	0.546	21%	Nader votes
BGV	(1-X1)*49.2	1-X1	Gore votes
BBV	(1-X2)*48.7	1-X2	Bush votes
Total	10.338		+(1- X1)*49.2+ (1-X2) *48.7
			
Total Kerry                         = Total Bush			
			
x1*g + x2*b +.54*nv +.71n =(1-x1)*g+ (1-x2)*b +.45*nv
+.21*n			
			
2x1*g  - g  +2x2*b -b +.09*nv + .50*n = 0			
			
g * (2x1-1) + b *(2x2-1) +.09*nv +.50*n =0			
			
49.2 * (2x1-1) + 48.7 *(2x2-1) +.09*21.76 +.50*2.6 =0			
					
98.4*x1 - 49.2+  97.4 * x2  - 48.7 + 1.9584 + 1.30 = 0					
					
98.4*x1 + 97.4*x2 = 49.2 +48.7 -1.9584 - 1.30					
					
98.4*x1 + 97.4*x2 = 94.64					
					
x1= (94.64 -97.4x2)/98.4					
					
x1=.9618 -.9898*x2					
					
x1	x2				
K%G	K%B				
95.19%	1.00%				
94.20%	2.00%				
93.21%	3.00%				
92.22%	4.00%				
91.23%	5.00%				
90.24%	6.00%				
89.25%	7.00%				
88.26%	8.00%				
87.27%	9.00%				
86.28%	10.00%				
85.29%	11.00%				
84.30%	12.00%				
					
Using the Final Exit Poll, modified for the new
weightings.					
If Kerry gets 2% of Bush 2000 voters, he needs 94.20% of Gore
voters to tie					
2000	Pct	K	B	N	Votes
NV	17.80%	54.00%	45.00%	1.00%	21.76
G	40.24%	94.20%	5.80%	0.00%	49.2
B	39.83%	2.00%	98.00%	0.00%	48.7
N	2.13%	71.00%	21.00%	8.00%	2.6
Total	100%	49.83%	49.83%	0.35%	122.26
					
If Kerry gets 6% of Bush 2000 voters, he needs 90.24% of Gore
voters to tie					
2000	Pct	K	B	N	Votes
NV	17.80%	54.00%	45.00%	1.00%	21.76
G	40.24%	90.24%	9.76%	0.00%	49.2
B	39.83%	6.00%	94.00%	0.00%	48.7
N	2.13%	71.00%	21.00%	8.00%	2.6
Total	100%	49.83%	49.83%	0.35%	122.26
					
If Kerry gets 8% of Bush 2000 voters, he needs 88.26% of Gore
voters to tie					
2000	Pct	K	B	N	Votes
NV	17.80%	54.00%	45.00%	1.00%	21.76
G	40.24%	88.26%	11.74%	0.00%	49.2
B	39.83%	8.00%	92.00%	0.00%	48.7
N	2.13%	71.00%	21.00%	8.00%	2.6
Total	100%	49.83%	49.83%	0.35%	122.26

[foo_bar]

148. reposting the wrong answer doesn't make it right

per x1=.9618 -.9898*x2, KV has a positive slope as x2 increases (x1 decreases), BV has a negative slope as x2 increases (NaderV is a flat line with zero slope, since it's constant regardless of x1,x2)

Thus you have two intersecting lines that converge at x1=83.27%, x2=13.04% and never again the twain shall meet:

[TruthIsAll]

149. Once again, an exercise in futility on your part.

You did not look at the three detailed breakeven calcs.

The breakeven is 49.83%.
This has been derived by formula.
The formula contains two independent variables, X1 and X2

We input a value for X2 to derive the value for X1 which satisfies the breakeven.

And yes, we ignore Nader.

Once again, your convoluted logic and lack of understanding has been exposed for all DUers.

You would not be able to design an analytical model if your life depended on it.

Stop flagellating.

[foo_bar]

153. it doesn't satisfy the breakeven unless x2=13.04%

We input a value for X2 to derive the value for X1 which satisfies the breakeven

Your matrix restates x1=.9618 -.9898*x2 as a tautology without testing its effect on BV=KV. The effect is KV+=18.4 (BV-=18.4) for every +1% change in x2. You might want to click the "Assume Linear Model" checkbox in Excel Solver's options, it's only y=mx+b vs. y=-mx+b (i.e., algebra).

[OnTheOtherHand]

159. looks like confusion about rounding error

Foo is talking about margins on the order of 400 votes out of 120 million. I think he took your "x1=.9618 -.9898*x2" (from the other thread) as exact rather than rounded.

So, to summarize, your equation shows that under these assumptions, the break-even point is approximately x1 + x2 = 96.3% across the range of interest (where x2 is somewhere around 0.1). x1 is the percentage of Gore 2000 voters that Kerry holds, and x2 is the percentage of Bush 2000 voters that Kerry takes. If x1 = 90%, and x2 = 6.3% or better, then Kerry wins. If x1 = 90%, and x2 = 6.2% or less, than Bush wins. (The popular vote, that is.)

Would you say you have demonstrated that this is impossible, because x1 + x2 must equal at least 96.3%?

[foo_bar]

200. the formula is in the post you replied to

Two Quibbles: the aforementioned rounding error (either my assumption of it or its provision as such), and the N for Nader factor. Nader can be whisked away with more precise constants, or variables where the .71/.21/.08 go (say KV: X3*N, BV: X4*N, NV: (1-(X3+X4)*N)).

Which is why these "what if?" scenarios are invariably chained to unrealistic assumptions: if you model the assumptions found in real life, you need the National Weather Service to solve for X5 through X10. That's where you find the real subcurrents, like partly racist in Florida and a 90% chance of corruption in Ohio.

Look small, national polls are a kaleidoscope. Build a better mousetrap.

[Febble]

203. Oh, I SO agree

re realistic assumptions. Occam's razor is fine for mousetraps, but this thing needs a shredder. And once it's been shredded there's a lot less power. There might have been a lot of voters in the sample, but it covered the nation and there's no reason to think that a Bush voter in Florida would behave like a Bush voter in Ohio. Or, for that matter, that a hacker in Florida would behave like a hacker in Ohio.

[Land Shark]

218. Occam's razor is for charlatans though

if the pop culture version of it (the simpler explanation is the better one) is used. That will too often lead one astray unless Einstein's pop culture quote "things can not be made simpler than they are" is kept in mind at all times.

But what occam's razor most directly refers to is General Theories of the universe and (given systems of unbelievable complexity) the theories that contain only the minimum number of necessary ASSUMPTIONS are best.

See http://pespmc1.vub.ac.be/OCCAMRAZ.html

Of course, trying to model something with the minimum assumptions is exceedingly COMPLICATED, often. Especially if you're thinking about the universe.

[mgr]

246. Minor off center point

The conclusion as to the application and utility of Ockham's razor is a conclusion based upon the site's point of view. Parsimony has a significant role in hypothesis formation and testing, and applies across the board in the sciences. The razor is a standard to compare comp(delete l)eting arguments that may equally fulfill an observation, and its use is only predicated on predicting the future (e.g. the hypothesis does not need to be precisely right). We apply the concept all the time in working out evolutionary patterns using claudistics, and I would not equate systematics with general systems theory.

The other thing to pay attention to is the mention of the relationship between the concept and epistemology, which links back to the Late Medieval Catholicism, the original context of Ockham's concept. I would be a little careful attributing an aspect of Augustiniam philosophy to charletanism, the jesuits among others, may take exception.

Mike

Edited for typo.

[TruthIsAll]

189. I searched NES for "FRAUD" and " EXIT POLLS" (no luck).

How come?

I am not going to waste any more time for you.
Why don't you provide DU with some specific information?

Why don't you respond to the anaxarchos request?
Make your case already.

I just posted NUMERICAL ANALYSIS in a new thread in order to advance the debate with a novel examination of the data. It's called sensitivity analysis.

You can look at multiple scenarios. See what fits. What makes sense.

Where's yours? Even one scenario?.

You have pointed out two errors in my prior analysis. I appreciated that and made the fixes. You obviously have an interest in my work.
Now we would like to see your work.

Let us have the opportunity of critiquing YOUR work for once.

Do you have anything of substance to add to the debate other than palaver? Or must you continue to just target my threads for debunking?
Which you never really do.

You should take a new pro-active role in making your case. You have not debunked the case for fraud. You have only made it stronger.
And it will get stronger as time goes on.

My work is out there. All the posts, lots of data, assumptions, analysis - and the only logical conclusions one can infer from it all.

Where is your work, OTOH? You keep everyone twisting in the wind waiting for your analysis.

But it never comes.

[OnTheOtherHand]

190. charitably, we are talking past each other

Are you asserting that false recall in the NES is obviously irrelevant to exit polls because you don't find the phrase "exit polls" on the NES website? Oh-kay.

I've made my case, and you haven't laid a glove on it. Conceivably you would find it easier to understand it if I presented it in the form of 120 scenarios, and I will try to do something like that, but I don't construe it as an obligation.

Your assumptions are not strengthened (or weakened) by characterizing them as "NUMERICAL ANALYSIS."

Suppose we stipulate that Bush had to win 17% of Gore 2000 voters. Do you have any evidence that he didn't?

I keep looking for the "clincher" -- the impossibility proof -- and it always seems to come back to assumptions. John Kerry reportedly said something like, "I can't believe I'm losing to that idiot." Yeah, well.

You have an unusual position in this argument, because while many people believed that Kerry was doing better than the pre-election polls seemed to indicate, you were one of the very few to believe that he had a 98%(?) chance of victory. Certainly you consistently start from different assumptions than most political observers. And you like to present NUMERICAL ANALYSES that you believe buttress your assumptions.

[TruthIsAll]

197. Well, I guess we will never see your doodles after all.

So that just about ends it. You have lost the debate because you never made the case to begin with.

As far as your snide comments on my pre-election probabilities, I based by numbers on detailed analysis using publicly available data. I had a very clear rationale for making those projections - and did a thorough quantitavive analysis showing all the work.

That is something you have never done. The vast majority of interested DUers will surely agree with me. You have heard from a few of our best to that effect already.

If you look at my projections and probabilities on the election model site, you will see that they are grounded in logic based on factual data and very reasonable assumptions as the Kerry's expected share of the undecided vote.

If you bother to check, you will see that 11 out of the final 18 pre-election polls had Kerry ahead. These are a mix of registered voters and likely voter polls. They were NOT cherry picked. I used registered where available because I felt they were a better indicator due to the massive democratic registration efforts.

Even assuming Kerry was tied with Bush at 47% with 6% undecided on Nov. 1, it was clear that as challenger he would win, since he could reasonably expect 60-75% of the undecided based on all historic precedent. Do the simple math 47+4=51%.

Based on the AVERAGE of very public national polls, I had Kerry winning the popular vote with 51.63% of the two-party vote. The probability based on 18 independent pre-election polls (which is the equivalent of an 18,000 sample) was close to 100%.

In addition, the 5000 trial Monte Carlo simulation of the state exit polls showed that Kerry would win over 330 Electoral votes with 51.80% of the two-party vote with near 100% probability of getting over 270 electoral votes.

Turned out, I was almost exactly right.

And you have the absolute gall to suggest that Bush may have gotten 17% of Gore voters.

You have succeeded in proving to all readers that you are totally disingenuous. You are rich in palaver, short on analysis.

Very short.

[OnTheOtherHand]

231. darling!

Following one of your links on pre-election polls, I find Bush ahead in 10 "final trial heats" and Kerry ahead in 3, with one tied.

http://www.pollingreport.com/2004.htm

One can also follow the electoral-vote.com link, download the database of state polls, and do some tinkering to again show Bush ahead in the popular vote (weighted state results).

You are entitled, I guess, to alter the assumptions of (e.g.) the final Pew Research Group poll and conclude that it "actually" shows Kerry up by one (registered voters), even though the people who conducted the poll concluded that it showed Bush up by three (likely voters). That said, your pre-election "Recent National Poll Trend" graphic probably violates AAPOR professional standards. It's OK to jigger people's numbers based on your own assumptions, but it at least should be crystal clear how you are jiggering, and it isn't. Pew reports, "Slight Bush Margin in Final Days of Campaign" (http://people-press.org/reports/display.php3?ReportID=232). How hard would a reader of that site have to work in order to figure out that you were betting _against_ Pew's likely voter model while claiming their data as support for your own assumptions?

And it's not just Pew -- this is inded the basic move by which you change a Bush lead into a Kerry lead. Some folks agreed with you at the time, others (obviously including the designers of the likely voter models) disagreed, and some didn't know what to think. I would say that we still have no way to know for sure, and that is why we need election reform. But you seem to think that you do know for sure, and I regard that as surreal.

Yeah, I suggested that Bush may have gotten 17% of Gore voters, but I am backing off that now. I think it is more likely that Kerry got 6% (not 10%) of Bush2K voters and Bush got 12-13%, or so, of Gore2K voters (some of whom retrospectively defected to Bush in 2000 also). That fits the panel data better -- but maybe too well.

I am inclined to say that the one thing that is clear to people who actually look at the innards of the survey data is how unclear this entire issue is. That just might be why you rarely look at the innards of survey data. Your way is faster, and apparently confers a greater sense of certainty, to those who can believe it.

[TruthIsAll]

235. Do your homework, darlin', I did no jiggering. I used registered voter

polls when they were available. In lieu of that, I used likely voters. I used 18 of the most popular polls.

You infer that I was cherry-picking. I will not infer, but I will state very clearly that you out of control.

If you preferred the likely voter model, be my guest. But you were wrong. In this election, the registered voter polls made more sense because of the large number of new voters who were not "likely" voters having never voted.

As usual, your propensity to naysay anything I do gets in the way of logical thought. If you check MP (I don't have the link) he agreed with my assertion that Kerry led in a majority of the final national polls.

I said it was 11 of 18. I was correct. I believe he also had it just about the same.

Once again, my model predicted that Kerry would win 51.6-51.8% of the two party vote. As it turned out, I was wrong. He did even better than that. Eat your heart out.

What did your model project?

[OnTheOtherHand]

249. one issue is controversial; the other is straightforward

Whether it is more appropriate to use RV or LV was and is debated.

But if you want to alter a poll's assumptions, you owe it to the polling firm and to your readers to make clear that you have done that.

And, incidentally, when you then ring Monte Carlo changes on your assumptions and declare a 99%+ probability that Kerry will win -- well, you are entitled to your belief, but it is only as good as your assumptions. Personally, I think that is a bad use of P values.

I have no idea what MP quotation you might be referring to, but folks can read his November 3 column and judge for themselves:

http://www.mysterypollster.com/main/2004/11/first_impressio.html

If you mean that MP thought before the election that the polls understated Kerry's chances, I agree with that. To quote from the above:

"The average of the final polls had Bush ahead by just under two points (48.7% to 47.1%).... Obviously, these topics deserve more careful consideration once we have final results, but it is obvious that (a) the incumbent rule did not apply and (b) that the consensus of the national polls was reasonably close to the final result. While the turnout was heavy, it did not conceal any hidden Kerry vote, as I speculated it might. If anything, the polls slightly underestimated Bush's national margin. Why? One big clue should have been the failure of the incumbent rule in 2002...."

It really seems to me that you are so certain that Kerry "did even better" than 51.6-51.8% of the two-party vote because you think your pre-election model proves it.

[TruthIsAll]

24. Are these the same intelligent people..

You:
If you mean that I am assuming that 43% of 2004 voters _actually_ voted for Bush in 2000, and only 37% _actually_ voted for Gore in 2000 -- no, I neither assume nor believe that. I don't know anyone who does. (And so you might stop to wonder: how can anyone see that figure and not think the election was stolen? But you should ask the question seriously, not rhetorically, because many intelligent people who have seen the figure do not think the election was stolen.)

Me:

Are these the same "intelligent" people who believe that

- Bush legitimately won the 2000 election in Florida?
- the Repubs legitimately won the senate in 2002?
- Bush won Ohio without the help of Ken Blackwell?
- Bush won Florida without the help of Jeb Bush?
- The reason why Diebold/Es&S didn't include a voter-verified ballot was to save cost?
- All those new, young female voters wanted Bush?
- The reason why the early exit poll results were never mentioned in the media was because they had no meaning?
- The reason that the Final Exit Poll matched to the vote count was because the vote count is accurate?
- Mitofsky is a great exit pollster with 25 years of experience who only hired inexperienced polltakers when a Bush was running?
- The exit polls were uniformly biased throughout the states to favor Kerry?
- that it's just a coincidence that the exit polls were accurate, (according to Mitofsky's own data) in those districts where paper ballots were used?
- that Republicans who controlled the design, manufacturing and programming of the voting machines fought against a paper trail and with machines that could never pass certification?
- who give you talking points?

You mean those people?

Oh.

[OnTheOtherHand]

27. I will try to respond to part of this post

Gee, which part should I choose?

Maybe the first bullet point.

Political scientists wrote quite a few articles arguing umpteen reasons why they were convinced Gore won Florida in 2000. One of them has the delightfully succinct title, "The Wrong Man is President!"

Several of these analysts have been at the forefront of attempts to analyze data from 2004. Perhaps I will leave it to you as an exercise where the author of "The Wrong Man is President!" fits in.

The idea that you are entitled to take every colleague of mine who doesn't agree with your analysis, shove a bunch of polemical claptrap in their mouths, and then dismiss them as Republican stooges...

...umm, is certainly a barrier to effective communication. I think it also discourages lucid thought.

[Name removed]

30. Deleted message

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

[Name removed]

34. Deleted message

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

[TruthIsAll]

41. Ok, I don't attribute all of these to you; only to those who you appear

to be in total sync with.

We have heard variations on these themes.
All are excuses designed to obscure the facts.

Most, if not all, are accepted by the whore media and RW Bush apologists without question.

The obscurantists (including many DINOs) have done their job well.

[Name removed]

93. Deleted message

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

[Melissa G]

168. Huh? OTOH ,what are you saying about bullet one?

There is pretty definitive recount analysis that shows Gore won Florida in 2000. Are you saying otherwise?

[Febble]

177. I think he's saying

that a political scientist, Professor Walter Mebane, wrote a paper called "The Wrong Man is President" about Florida 2000:

http://macht.arts.cornell.edu/wrm1/overvotes.pdf

The Wrong Man is President! Overvotes in the 2000 Presidential Election in Florida

Abstract

Using ballot-level data from the NORC Florida ballots project and ballot-image files, I argue that overvoted ballots in the 2000 presidential election in Florida included more than 50,000 votes that were intended to go to either Bush or Gore but instead were discarded. The primary reason for this was defective election administration in the state, especially the failure to use systems that warn the voter when there are too many marks on the ballot and allow the voter to make corrections. If the best type of vote tabulation system used in the state in 2000—precinct-tabulated optical scan ballots—had been used everywhere in Florida, Gore would have won by more than 30,000 votes. The experience in Florida points to the need to gather ballot-level data to evaluate the success of election reform efforts now underway in much of the United States.

His work on machine allocation in Ohio in 2004 is reported here:

http://www.democrats.org/a/2005/08/ohio_election_r.php

He was also in the study team that dealt with Quantitative Analysis of Precinct Level Data in the DNC study: Democracy At Risk: The 2004 Election in Ohio

http://a9.g.akamai.net/7/9/8082/v001/www.democrats.org/pdfs/ohvrireport/section02.pdf

The point being that political scientists who remain unconvinced that the election was stolen include those who are on record as concluding that the electoral system unfairly deprived Gore of the presidency, and Ohio of Democratic votes in 2004.

[Dynasty_At_Passes]

178. This one has raised a red flag for me...

Why does Walter Melbane insist that the "voting would have been perfectly accurate" if done on optical scan machines?

Does he know the vendor is ES&S and that a proven demonstration has been done, where a memory card was switched and the votes were switched?

Has he seen the Hursti hack on Black Box Voting?

Does he realize these irregularities were proven on memory cards in both 2000, and 2004?

If and true of the above, how can he possibly indicate 2004 was fair or that those machines would have made 2000 vote fair?

[Febble]

179. You need to read a little more carefully

Mebane wrote - on April 6th 2004, i.e. before the 2004 election - not that the "voting would have been perfectly accurate" (where did you find that? if it's a paraphrase you shouldn't use quotes) but that:

"If the best type of vote tabulation system used in the state in 2000—precinct-tabulated optical scan ballots—had been used everywhere in Florida, Gore would have won by more than 30,000 votes."

In other words, his study found that in 2000 the most accurate (note, not "perfectly accurate") machines were optical scan ballots, and that if all machines had been as accurate as the most accurate, Gore would have won by over 30,000 votes. In other words's he'd have been elected president. Pretty damning of the Florida 2000 election I would have thought. And based on direct evidence that in 2000, optical scan systems had fewest problems with over-votes. (And as I have pointed out elsewhere, whether those optical scanners compared favorably or otherwise with the new e-touch machines used in 2004 is not easy to determine. My findings were that they did worse; Hout et al found they did better. Others have concluded that there was no significant effect of machine. I found no effect of vendor.)

Then read Mebane's work on Franklin county. My point is that a man who has done sterling work on exposing the inequities of the US voting system still believes that Ohio probably wasn't stolen. Maybe you can convince him otherwise. But it might be worth reading his work first.

[Dynasty_At_Passes]

181. I have read his work. But here's where Melbane's point falls apart.

He claims that clearly the optical-scan machines were accurate in 2000, where direct evidence has been revealed that at least one memory card was switched by Diebold/ES&S which lost around 16,000 votes. And that memory card, as BlackBoxVoting has shown contains secret instructions.

Edit: Theresa Lapore, the democrat Board of Election worker held up this memory card even and claimed the company said it had "failed".

The fact they fail to account for this alarming discrepancy casts doubt on the whole report, not that Melbane is insincere in his findings. But he seems to take a rather laid back approach to verifying the count for these machines, or really exploring the larger implication of fraud and election tampering. In Washington, seals were broken. The case for this is very strong in Ohio based on Board of Elections workers leaving the equipment in partisan hands unsupervised, and the evidence of ballots altered in Lucas County through various "non-detectable" means. Right there is a red-flag on the whole report. They didn't lock down the op-scan machines in Ohio or the memory cards, nor did they do so in Florida. So there is no possible way to even guess on what their accuracy or discrepancy based calculations are.

[Dynasty_At_Passes]

214. It could very well be Mebane is simply behind the times, too..

But this information has just recently come out, and black box voting altered a memory card which deleted votes the same way the card did in Volusia County, during 2000 recount.

Essentially the same thing happened with the optical scan precinct tabulator, that happened with the machine tabulator.

[Melissa G]

186. Thanks for the links and explanation Febble! n/t

[mgr]

28. What is it you believe?, and what is it you know?

To be intelligent is to know, not believe. Belief is an act of faith, not necessarily grounded in empirical fact. Interesting choice of words from one who deals exclusively in facts. Only belief can make that graphic imply that the election was stolen.

Mike

[TruthIsAll]

29. Surely, not a belief in rBr or that 13mm new Evangelicals voted for Bush.

I'll leave those beliefs to you.

I believe that Land Shark gets it just right.
And you never will.

That's what I believe.

[mgr]

44. Bishop Ussher--It is obvious you do not know with any certainty

what you purport to affirm. I take your response to reflect that. This was an opportunity to affirm what you know, and you took it to construct a strawman and make an ad hominem attack, as well as some odd claim of support of your position from an authority. Don't bother to respond, it will only make it worse.

Landshark sees our purposes as one in the same in the grand scheme of things, but blurs over the practical implications of what the differences imply as to further research and hypothesis testing; and moving the cause forward. Read what you like into what he said, but no where does he make the claim that he thinks my position is that of a republican apparatchik. Only you would twist it that way.

Let me show you how its done, and what it means:

I know that the exit polls performed by NEP do not give any clear indication of widespread fraud such that Kerry was deprived of the popular vote. I suspect that DREs did not play a significant role in switching votes, but know they played a role in producing long lines on election day. I know by accounts of long lines I heard over the radio, and confirmed in the EIRS database; and, from accounts of similar methods of depriving people of their vote in Florida in 2000, that the same was done in 2004; and that the final electoral college result was probably wrong becasue of this.

Nowhere do I claim to know with absolute certainty, but neither do I believe, which is the absolute nadir of knowledge with any certitude. One knows their morals, one does not believe their ethics. This post-modern sophistry is why we have ascientific nabobs like yourself clamoring for equal time for intelligent design, creationism, the anthropocentric mumbojumbo, alongside the most elegantly stated law of evolution; who have no clue what phyrronic scepticism is, and why it is so important to the development of the technology we enjoy today; and have no idea who David Hume is, and why he might be so important to this discussion.

Mike

[Febble]

32. Well, speaking for myself

and I'm fairly intelligent, and I'm not convinced by this "clincher" story, so, do I believe that:

- Bush legitimately won the 2000 election in Florida?
No

- the Repubs legitimately won the senate in 2002?
DUNNO, sorry, wasn't following that story

- Bush won Ohio without the help of Ken Blackwell?
Possibly not

- Bush won Florida without the help of Jeb Bush?
Probably, but he might have helped anyway

- The reason why Diebold/Es&S didn't include a voter-verified ballot was to save cost?
I doubt it

- All those new, young female voters wanted Bush?
Disagree with premise

- The reason why the early exit poll results were never mentioned in the media was because they had no meaning?
Possibly, but they were mentioned in the media I read

- The reason that the Final Exit Poll matched to the vote count was because the vote count is accurate?
No - it matched because it was matched

- Mitofsky is a great exit pollster with 25 years of experience who only hired inexperienced polltakers when a Bush was running?
I think he hired inexperienced interviewers this year, and probably in the past.

- The exit polls were uniformly biased throughout the states to favor Kerry?
Probably

- that it's just a coincidence that the exit polls were accurate, (according to Mitofsky's own data) in those districts where paper ballots were used?
False premise; data does not support this inference

- that Republicans who controlled the design, manufacturing and programming of the voting machines fought against a paper trail and with machines that could never pass certification?
Probably

- who give you talking points?
Cheap shot

It is possible to think that there is good evidence that the election was won illegally, and that there is some vulnerable evidence. IMO presenting vulnerable evidence is a bad tactic. Think Dan Rather.

[TruthIsAll]

36. Boy, this post must have really hit a nerve.

Good to see all you guys again.
Let's see:

What are the ODDS that Kerry would win ALL 120 scenarios in which we have given Bush a HEAD START by assuming that ALL of his 2000 voters (still alive) turned out to vote in 2004 while AT THE SAME TIME assuming a range of scenarios in which ONLY 88-100% of Gore voters CAME OUT TO VOTE?

AND TO TOP IT OFF:
LET'S ALSO ASSUME THE FULL RANGE OF NATIONAL EXIT POLL KERRY/BUSH NEW VOTER SPLITS: FROM 59-39% TO 57-41% TO 54-45% - ALL FOR KERRY?

A FULL 4-D SENSITIVITY ANALYSIS.
Kerry Vote = f(G, B = 100%, KP)
where
G = Gore Turnout,
B = Bush turnout = 100%,
KP = Kerry percentage of New voters

KERRY WON THEM ALL.
ALL THE COMBINATIONS.
ALL THE PERMUTATIONS.
ALL 120 OUT OF 120.

WHO WUDDA THUNK IT?

GO AHEAD.
TEAR APART THE MATH.


Remember MC Hammer?
YOU CAN'T TOUCH THIS.

YOU CAN'T SPIN THIS.

[Febble]

37. And nice to see you again too

I must have been missing you.

And I won't tear apart your math, for reasons given above. I'm not disputing your math (although I'd like it if you didn't tear apart my friend).

But proving that something is not statistically possible is not the same as proving causality, as you will know. There is more than one way of interpreting the result.

And I simply do not share your assumption that people correctly reported who they voted for in 2000. They tend not to.

I actually agree with Land Shark and GuvWorld that at this point, the numbers cease to matter. Where I think we all agree is that a country with an electoral system that leaves its citizens in doubt as to whether the man who is president is the people's choice is not a true democracy. You do not doubt that Bush lost. I doubt whether he won. The fact that there is room for doubt is the issue. And even as a Brit that matters to me. It matters to the world.

(I'll crawl back into my hidey hole again now.)

[Name removed]

38. Deleted message

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

[Febble]

40. Land Shark wrote:

Given that, based simply on nondisclosure of data and analysis we can confidently conclude that there is NO BASIS FOR CONFIDENCE IN ELECTION RESULTS

What
on
Earth
More
Do
We
Really
Need
to
call
for
election system
CHANGES?

and it was that I was referring to. What I meant was that the case does not depend on the numbers. It is self evident.

(You might like to stop putting words into OTOH's mouth as well.)

Actually, numbers DO matter to me, as you know. I like numbers. Which means that I care fairly passionately about how they are interpreted. It's my job.

[TruthIsAll]

43. LS on exit poll internal "inconsistencies" and "impossibilties"

While "inconsistent", there's nothing wrong with accepting final national exit poll numbers as true in order to show that various internal inconsistencies or impossibilities are therefore necessary to support them when accepted as true, and thereafter using those same inconsistencies or impossibilities as evidence that the final national exit poll numbers showing bush victory (which otoh expressly agrees in the above post had to be changed to show a bush victory) are not correct because they can not be a valid reweighting process leading to MORE accuracy when subject to such unlikely internal tensions and inconsistencies.

[mgr]

46. Except you failed to provide the necessary proof

of the inconsistencies or impossibilities. Your Hobson's choice was anything but a lady and a tiger, but a tactical evasion from the arguments developed in the previous thread. A Hobson's choice must show inconsistency within each statements connected by the or, that means f and not f follow for two trees. The statement, backed by reseach, that some forget who they voted for, makes this impossible for one of the trees; and the prima facie outcome when reweighting the data that poll non respondants consisted of a population distinct from the those polled, cannot be show(n) impossible within the poll's own methodology. You need evidence that widespread fixing of the popular results occurred to have the exit poll's rBr hypothesis falsied; but you also need to make the case to show that the exit poll shows that the popular vote was fixed. I call this teleology, or circular reasoning. Logically, your exit polls argument demonstrates nothing, unless it's along the lines of Augustine's ontological argument.

Mike

Edited for typo.

[mgr]

100. Correction

the term should tautology, not teleology. Age and the workplace are catching up with me.

Mike

[Febble]

48. No, I agree

that as an exercise in logic you can hold one thing to be true, and test whether other things become difficult to explain. We can, for example, assume the sun goes round the earth - it's just that planetary motion becomes very difficult to account for, and the wanderings of the planets are more neatly explained if we assume that the earth goes round the sun.

However, in the case of the exit poll responders who report having voted for Bush or Gore in 2000, I believe the inconsistencies you point to are explicable in terms of false recall - in a pattern well documented in previous elections. In fact, on my calculations, only a small number of apparent "Bush 2000" voters would have had to have been actual "Gore 2000" voters to account for the discrepancy. And we know that people who voted for the loser tend to report having voted for the winner. So at least one alternative, fairly parsimonious, explanation to the fraud explanation is available.

The inconsistency is not, therefore, a "clincher".

[OnTheOtherHand]

42. asked and answered

I (among others) already tore apart the math -- or, rather, I smashed the assumptions on which the math was based. You insist on treating the question "Who did you vote for in 2000?" (or whatever the actual wording was) as if it had a one-to-one correspondence with whom people had actually voted for in 2000.

We have ample evidence that this assumption is wrong.

For whatever reason, you change the subject every time I cite the evidence.

(EDIT: Oh, but I think I see why you are confused. I will try to show you the numbers in another way. --I will be going to my daughter's soccer practice soon. Given that false vote recall typically benefits the incumbent and demonstrably did so in 2002, we can expect some people who _didn't_ vote in 2000 to claim that they voted for Bush. And we can expect some people who voted for Gore to claim that they voted for Bush. Not only does that throw off the percentages in each category, but it will throw off the marginal percentages within the categories. The poll can be expected to exaggerate Kerry's support among new voters and Gore voters, because some of them will not only vote for Bush in 2004 but "retroactively defect" to Bush in 2000. So rejecting the first set of percentages and accepting the second does lead to strange results, but that doesn't represent an internal contradiction.)

[TruthIsAll]

45. You have not disputed a damn thing.Your inconsistencies are glaring.

All you have done is offer faith-based arguments that we should all just accept your hypothesis that Gore voters lied or forgot who they voted for. And Bush voters did not. Or if they did, at a much lower rate.

Your inconsistencies are glaring.

1) First you claim that the final poll was weighted to match the vote.I take that to mean the 43/37 split.

2) Then you claim that the 43/37 split was due to Gore voter Alzheimer's and/or democratic propensity to lie. That's a low blow. I'm a Democrat.

So which one is it?

If it's (1) why match the exit poll to the vote at all? I can't believe I'm asking a political scientist this simple question.

The demographic weighting was done earlier at the 13047 mark.
But the demographics said Kerry won. So it's Ok with you that weights and percentages are changed to match the vote.

Because Bush won.
After all, the vote count tells us so.
And Bush is an honorable man.

Then why bother to investigate the election at all?
Why bother to exit poll at all?


If it's (2) why was it that only Gore voters lied or forgot who they voted for? Are not Bush voters subject to the same dementia?
Do you have ANY factual medical, and psychiatric data to back up your claims?

Or must we accept all of this hypothetical nonsense as fact?

[mgr]

53. What hokum.

OTOH has done the following:

1.) Cited Mitofkski's methodology is to reweigh the categories based upon the outcome. Find a quote to dispute that.

2.) OTOH and others have cited several studies that the only explanation of the phenomena is misrepresentation of what they did in the past. Find a study that disputes them.

Good luck.

Mike

[Land Shark]

47. OTOH has indirectly shown us the Abyss of elections

Given the radical uncertainty to which all election conclusions are subject, given conditions of data secrecy and analysis secrecy, the comment by otoh above (critiquing TruthIsAll for assuming that what someone *reports* as their vote in 2000 is necessarily the same or a one to one correspondence to who they *actually* voted for) is very interesting.

Before I say why, consider also that legally speaking nothing is 'admissible' in terms of evidence extrinsic to the ballots themselves to impeach (or otherwise change) a ballot's votes; i.e. the ballot stands alone and even if one could (but this would be illegal, due to ballot secrecy) find one's own ballot and argue forcefully for a correction to the votes on that ballot, that evidence would not be heard by a court and allowed to affect the ballot's votes. (Perhaps a hypothetical voter, based perhaps on a lifetime history of voting a certain way cobmined with repeated announcement to friends and family of both one's intent to vote a certain way and one having actually voted that way might make such a challenge with abundant evidence, but it would still not be heard)

So, given the lack of any basis for confidence in election results/conclusions we are left very uncertain. Given the legal inability to change the ballot even with abundant evidence we are rendered helpless, but given OTOH's statement above challenging the very memory of voters as being unreliable as guide to how they actually voted in 2000 even under exit polls circumstances where there is no reason to believe there will be consequences to honest responses, we are now left with the following:

No amount of EVIDENCE, nor any amount of a voter's MEMORY, will be allowed to impeach RESULTS for which there is no basis for confidence in the first place.

Election NIHILISM is now status quo? I struggle to find the right word for this state of affairs where voters don't matter.

WE seem to be staring down an electoral abyss of no escape, if we accept principles like otoh suggests (this is more due to the underlying lack of basis for confidence, but otoh's position is another necessary component).

As such, democracy has vertigo.

[autorank]

50. "Election NIHILISM" Great,. They "Made It Happen On Purpose"

Think about it. This is the corollary to they're all crooks; it's the same everywhere (see my daily news thread of today); and it leads to "what's the point," "nothing ever changes." Stasis--that's the goal; we lay down and take a beating while they keep stealing elections. Pretty soon, we'll have the machines do the voting entirely. It will be sold to the public as an efficiency measure -- no more missing work, no long lines, no paid board of elections--just machines voting, programmed by Diebold, ESS, and the other radical right ownership groups. Wonderful.

This is the core of the fight -- getting people to act while there is still enough faith in democracy as an alternative.

[Febble]

51. That is the abyss indeed

but you don't have to go there, surely? What you need is an auditable electoral system. Without one, clearly, there can be no redress (exit polls are a very poor way to audit an election). But auditable elections are easy to devise. We have them in the UK.

(For the first time, this year, there was some cheating, but that was due to a change in the regulations for postal voting. It is being rapidly fixed.)

But you need some kind of hard record (paper seems obvious), scrutinised counting, and some kind of mandatory recount system that is legally enforceable. I still fail to understand why Kenneth Blackwell is walking around free when the Ohio recount was simply a sham. I dispute the fraud interpretation of the exit poll discrepancy on the principle that the sample was unlikely to be random. But by the same principle I dispute that the Ohio recount was a recount precisely because the sample of hand counted precincts was not random.

If a true recount followed a challenge or a close call, you'd have taken a big step back from the abyss.

[foo_bar]

55. it's rationalism, not nihilism

The evidentiary standards in civil court (9 out of 12 jurors, unreasonable doubts implicitly permitted) have more to do with the perception of truth than the empirical establishment of it. And that's fine, Lord Blackstone would want it that way, otherwise you get the infinite Talmudic debates like you see here.

Election NIHILISM is now status quo? I struggle to find the right word for this state of affairs where voters don't matter.

I won't speak for OTOH, but nihilism to me means believing that humans act deterministically, and their intentions can be divined more accurately from an exit poll sample of <0.05% than the voters' stated intentions (as established in the NH recount, for instance, even though the NH exit poll deviated twice as much as Ohio's, the focal point of the hijinks investigation). That doesn't prove Bush tried to steal NH, VT, and MA; if anything, it implies an inverse correlation between early exit poll "blue shift" and hotspots of fraud/suppression (FL, OH, and NM were all within the margin of error, yet they're the least believable results; VT and MA don't let republicans near the levers of power).

but given OTOH's statement above challenging the very memory of voters as being unreliable as guide to how they actually voted in 2000 even under exit polls circumstances where there is no reason to believe there will be consequences to honest responses, we are now left with the following:

No amount of EVIDENCE, nor any amount of a voter's MEMORY, will be allowed to impeach RESULTS for which there is no basis for confidence in the first place.

Behavioral science doesn't exist just to give people headaches, it's something of an emerging science (albeit a primitive one). But the notion that people's recollection fades over time is a theory like evolution is a theory. (hands mic to OTOH)

[TruthIsAll]

58. FL and OH wthin the MoE? Are you ready to back that up?

You
FL, OH, and NM were all within the margin of error, yet they're the least believable results.

Show us:
1) the exit poll sample size
2) the calculation

While you are looking...
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=318693&mesg_id=318693

[foo_bar]

61. there's no sources for your MoE calculations

It's 1/sqrt(n), right? Here's what an actual pollster had to say:

Ohio, n = 2020. Sqrt (.5 X .5) / Sqrt 2020 X 2.6 X 1.6 = +/- 4.6%.

Pennsylvania, n = 2107. Sqrt (.5 X .5) / Sqrt 2107 X 2.6 X 1.6 = +/- 4.5%.

Florida, n = 2862. Sqrt (.5 X .5) / Sqrt 2862 X 2.6 X 1.6 = +/- 3.8%.

http://www.mydd.com/story/2004/11/30/171641/64

[TruthIsAll]

63. You quoting MP? He adds a 60% cluster effect.

MP just makes the MoE as big as necessary so that he can say that it was not exceeded by Bush. His analytical shortcomings were exposed a long time ago.

Everyone knows exit polls are more accurate than pre-election telephone polls.And pre-election polls always use that same formula: MoE= 1/sqrt(n).

You give FL a +/-3.8% MoE for almost 3000 respondents?
You never stop smoking, do you?

Argue with Mitofsky:
Notice the MoE for 11,500 respondents.
Then estimate the MoE for 13047 respondents.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=357345

Foo, you are a great source of misinformation.

[OnTheOtherHand]

66. are you staking your reputation on the position

-- gee, so many choices here --

that Mitofsky's table indicates a zero cluster effect?

(Oh, please. I could link to that and never have to rebut you again.)

TIA, exit polls are more accurate in the sense that major sources of error are removed (people, at least in theory, know whether they voted). But it doesn't mean they have smaller margins of error. That you could apparently believe that indicates that you don't understand the meaning of "margin of error."

You have some nerve.

[Name removed]

68. Deleted message

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

[foo_bar]

80. it's a mystery jpeg, with whole numbers where the percentages should be

It appears to be coming from mysterypollster.com, the person you said was "exposed" a post ago.

[TruthIsAll]

82. Right. From MP. He got it from E-M.

Thanks for the heads up.

[Febble]

88. Lets get this in context:

Here's the link to the document containing the table:

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

It's from the Methods Statement for the E-M National Exit Poll. A similar table for the State Exit Poll can be found here:

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

Both are preceded by this explanatory passage:

All samples are approximations. A measure of the approximation is called the sampling error. Sampling error is affected by the design of the sample, the characteristic being measured and the number of people who have the characteristic. If a characteristic is found in roughly the same proportions in all precincts the sampling error will be lower. If the characteristic is concentrated in a few precincts the sampling error will be larger. Gender would be a good example of a characteristic with a lower sampling error. Characteristics for minority racial groups will have larger sampling errors.

The table below lists typical sampling errors for given size subgroups for a 95% confidence interval. The values in the table should be added and subtracted from the characteristic’s percentage in order to construct an interval. 95% of the intervals created this way will contain the value that would be obtained if all voters were interviewed using the same procedures. Other non-sampling factors, including nonresponse, are likely to increase the total error.

{My bold}. Note that the table gives typical figures for an unspecified "characteristic" and that figures are rounded to the nearest whole number.

The context in which your image was taken was a piece by MP here:

http://www.mysterypollster.com/main/2004/12/what_is_the_sam.html

which contains links to further sources that discuss the design effect.

Having said all that, the MoE is largely irrelevant to the present discussion. Nobody disputes that the discrepancy between the poll and the count was massively significant, and that something other than sampling error, clustered or not, must account for it. One question is - what? Another question is - was Ohio special?

The E-M report

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

(p 28) tells us that the problem did not lie in precinct selection p . The discrepancy was at precinct level. So to check out Ohio, the sensible question is not so much: was Ohio's estimate outside the MoE, but was the WPE in Ohio an outlier?

Well, it was higher (-10.9) than the average of -6.5 for all 1460 precincts in the state samples (pp31-33), but Connecticut, Delaware and Vermont all had mean WPEs greater than -15, and New Hampshire, which, at least according to two apparently disinterested sources

http://www.invisibleida.com/

http://www.votenader.org/media_press/index.php?cid=413

had a "clean" recount, also had a higher mean WPE of -13.6. Alabama, Missouri, New York and North Carolina were also all more "red-shifted", according to the mean WPE figures, than Ohio. So Ohio doesn't particularly stand out; moreover, the mean WPE in Ohio was less than in one state (NH) where an apparently properly conducted recount confirmed the count rather than the poll.

Look, there is plenty of evidence for fraudulent activity of all kinds regarding the election. I take Land Shark's point that all a court needs is a preponderance of evidence. But it is my considered view that the exit polls do not constitute evidence that meets anything like the 51% confidence mark, never mind a 95% confidence mark. Whereas the evidence that Kenneth Blackwell obstructed a fair recount in Ohio is damn near 100%, as is the evidence that more Democratic votes than Republican votes were lost in Ohio due to voter suppression, whether that suppression arose from systemic negligence or malice. It wouldn't have taken much for this to have swung Ohio, and thus the presidency, for Bush. And it probably wouldn't have made much impact on the exit poll.

I don't see the point in making a vulnerable case when you have a watertight one. It simply saves your opponents the trouble of making their own straw men.

[TruthIsAll]

84. Getting a little testy, aren't we? You guys get all worked up over MoE.

Cuts to the core of the case, I guess.

You have to keep those MoEs as wide as the holland tunnel so nothing exceeds it. I see you are back to the old talking points: MoE. I thought we MOE'D that one down a long time ago. Still trying to wear me down, eh?

But Mitofsky say the MoE is 1% for 10,000 respondents (rounded).
I calculate it as 0.84% for 13,047. That's 1% rounded to include your "design effect". Just like Mitofsky.

Jeez, how many times must I keep repeating that?

But I digress.

Exit polls are MORE accurate AND also have a SMALLER margin of error than traditional polls because of two basic reasons:

1) More voters are sampled in exit polls than in standard surveys. So N is greater. And therefore 1/sqrt(N) is smaller.

2) People who have just voted know who they voted for.
No undecided or change of mind. Makes up for a lot of the "cluster".
Or as you would call it, the (intelligent) design effect.

Of course if they voted for the Democrat, there is a good chance they will forget or lie about it four years later, if asked.

[OnTheOtherHand]

97. actually, no part of my argument hinges on the MoE

Everyone here agrees that the discrepancy between the exit polls and the official returns can't be explained by random chance alone. This business about "hav(ing) to keep those MoEs as wide as the holland tunnel" is quite obviously wrong.

But analytically it is essential to distinguish between MoE and accuracy. Sampling error is just one form of error, and you like to talk as if it is the only kind.

And analytically it is useful, at least for the sake of credibility, to make a good-faith effort to calculate the MoEs correctly. Disregarding the design effect fails that test.

Your post seems to assert that E/M's table implies a design effect multiplier of 1.0, but it clearly does not. Look at it harder.

Since the design effect multiplier is incorporated directly into the MoE, it is confusing to say that the factors you cite "make up for" clustering. The MoE refers only to sampling error, which is why it incorporates a design effect correction.

If we wish to assess the plausible magnitude of non-sampling error, a good approach might be to take a dispassionate look at recorded errors in other exit polls. Merely asserting that exit polls are more accurate than telephone polls doesn't help much (especially if you never actually assess how accurate the telephone polls are).

[foo_bar]

79. you stopped reading after "Mystery Pollster"

You quoting MP? MP just makes the MoE as big as necessary

Conclusion. All of the state estimates above are well within their error calculations below.

Nick Panagakis

http://www.mydd.com/story/2004/11/30/171641/64

His analytical shortcomings were exposed a long time ago.

No comment.

[Land Shark]

59. Foo bar's 9 of 12 jurors points to why one can win a fraud case in court

while scientists may sit around forever with "coulda been" this and "coulda been" that. A preponderance of the evidence (say 51%) or clear and convincing evidence (80%+?) is enough to get a judgment and thereby, ultimately, the gun of the sheriff on your side to seize property or otherwise enforce the judgment.

A lot of people's notions of "proof" is amorphous and may follow a principle of excluding every other reasonable possibility. Scientific proof is not amorphous but is still a very high standard.

For those enamored of proof before action (as opposed to my "no basis for confidence" approach) I would ask the following question: since coulda beens will be insufficient to defeat legal proof unless they rise to the level of more likely than not, and if you accept my assumption (based on arguments and facts elsewhere) that a preponderance of the evidence shows a corruption of democracy and that thereby (but given ONLY full funding and a proper plaintiff, etc.) we already HAVE the power of the sheriff.... WHAT ARE PEOPLE WAITING FOR before they defend democracy?

OK, OK, probably some of you will say you are defending democracy in your own way by being scientifically careful from your point of view, but consider this:

People rationally over protect against the possibility of large losses that are relatively unlikely, such as tending to overbundle children against cold, over protect against stranger kidnappings when known family members are the real danger, defense forces are relatively speaking on a hair trigger in terms of investigating threats (though not necessarily for launching missiles). Therefore, even if it were UNLIKELY but possible that a threat existed to democracy, if we value democracy there would be an investigation of that potential threat, and a robust one.

Yet there's no investigation, of course, and even among those here committed to democracy there is some reticence to act upon the real implications of what we know.

Someone needs to get up in the middle of the night and investigate the suspicious noise we heard in a robust and honest way, even if it were more likely than not that it was "just the cat". Even if it coulda been a dozen or more innocent explanations.

Thus, the secrecy of the data and analysis and the stonewalling of Mitofsky and company in the face of credible possibilities or probabilities is quite arguably a faithlessness toward democracy that is itself cause for serious concern. That Mitofsky may claim that he's convinced himself it's all ok is meaningless, because the right to be "secure about one's democracy" is not a personal right to people like Mitofsky, it is a public right. Damn him for not caring about the public, and keeping the public's business secret.

[OnTheOtherHand]

64. it is a pleasure to converse with you

I only hope that you will try to interpret the arguments I (and others here) am actually attempting to make, some of which seem distinguishable from the arguments you seem to address.

I am not calling for scientific proof of anything. (I do call for proof as a condition of claiming to have proved something, but that doesn't seem so unreasonable to me.) I am not a practitioner of tobacco science.

I am not suggesting that exit polls are radically unreliable and meaningless. We know that exit polls 'usually' do 'pretty well.' A view that has somehow taken hold in certain quarters, despite abundant knowledge to the contrary, is that exit polls are almost always supremely accurate. Some seem to think that nominal margin-of-error calculations are binding upon the universe (what was the nominal margin of error of the 1936 Literary Digest poll, I wonder?). And some may believe, or argue as if they believe, that known sources and patterns of error in polling data can or should be discounted, at least if they undermine The Cause. I have no patience for that.

Very few -- I am tempted to say vanishingly few -- people with knowledge of survey methodology believe that the exit polls establish a preponderance of evidence that Bush won the popular vote. Some of us are on DU and voice this skepticism. (We are of some range of opinion on whether there is an overall preponderance of evidence.)

That does not mean that we argue that everyone should just ignore the exit polls, or that everyone should just ignore other evidence of fraud, or that everyone should ignore the glaring holes in the system because we cannot be certain that they were exploited in any particular year. On the contrary, we investigate fraud, and we call for election reform. I must confess that the repeated claims that I am out to obstruct the investigation really tick me off. One of us was at JSM last week talking about the possibility of stolen elections and calling for reform, and I'm pretty sure it wasn't TIA.

If you or anyone else can explain to me how having Mitofsky release more data would materially advance the investigation of fraud in 2004, I would be pleased to know. If I'm not mistaken, in San Diego some folks recently conducted what we might think of as an "open-source exit poll," and it seems to me that the only advantage that accrues from that complete access is that they know the exact precincts that were surveyed. By analogy, conceivably the raw exit poll data would give us some ideas about which precincts to obtain a full recount in. But I think we already have plenty of such ideas. I am perplexed and troubled by the apparently widespread conviction that Mitofsky believes he has in his possession information that could save democracy, and/or that full release of the data would provide some huge advance in our ability to assess fraud. If the latter, at least, is true, someone should explain it. I don't know whether anyone on DU (including me) has taken the time to do a close assessment of the Election Science Institute's analysis of "blurred" Ohio exit poll data. I would love a thoughtful assessment of whether it is worthwhile to obtain similar data from other states, or -- well, I appreciate thoughtful assessments of almost anything.

[Land Shark]

67. i apologize in advance for my latest post on "bridge out ahead" but...

it stands for the proposition that the requirement of scientific truth doesn't necessarily apply.... at least in debates where we as a society are considering corrective actions. OK, but you say above that you don't require that.....

Now we can't directly violate scientific principles in our actions but on the other hand we don't have to prove to a scientific certainty the cause for our actions. IF the action is disclosure of data or even a new election under proper conditions, that IMHO need not be proven with anything approaching scientific standards. But again, you've clarified you don't advocate that position.

Let me say this however, by way of explaining why you are being treated the way you are (from your point of view) and why TIA (assuming he is stretching the case as you say) may be closer to "correct" from a certain point of view:

This is not just science, but politics and public policy. As I see nearly every day in law, an adversarial approach is taken by all parties whereby the truth if often found somewhere in the vast middle, but is not unknown even in the positions of the most extreme from time to time. Bottom line however is that a bit of exaggeration or a lot of it is PRESUMED in public actors, so if they start out with a modest claim that is rigorously true, that claim will also be scaled back in order to grant "proper" credit to the claims of other stakeholders (because we have no mechanism for ultimate truth determination).

so, if someone "representing" the election fraud-proof contingency doesn't push all available inferences in the favor of that side, it is dangerous to that point of view, because the public will cut it back to arrive at it's "truth"

OF COURSE there's danger too of not being treated seriously at all if one is "discovered" doing this spinning and stretching, yet the system almost requires it.

Problem is OTOH and everyone else, an honest academician can be DESTROYED on the stand for his very honesty, and politics is for these purposes too much like the courtroom.

[TruthIsAll]

69. I have never stretched the case. Just used publicly available data.

Land Shark, just take a look at the Mitofsky MoE table cited in this thread.

In the table (as well as in notes accompanying the National Exit Poll demographics) Mitofsky gives a 1% MoE for an exit poll with greater than 10000 respondents.



OTOH and foo-bar have argued incessantly (and in this thread) that the MoE is not 1%, although the standard MoE formula and Mitofsky's table state that it is.

Their arrogance in denying the obvious is purely to debunk my analysis of both state and exit polls. The state exit polls show that Bush exceeded the MoE in 16 states (a 1 in 19 trillion probability) .

The National exit poll shows that Bush easily exceeded the 1% MoE by going from 48% in the poll to 51% in the vote. The odds are of this deviation are astronomical as well.

DUers would be most interested to hear your comments on this specific matter.

[autorank]

72. Correct...this is the only "paper trail." You work with what you have.

We could have more if EM would release the data rather than building "back fires" to avoid it. Rep. Conyers specifically asked EM for the data and they said no, too busy to show up in person to do so by the way...the arrogance.

If it's good enough for the Ukraine, exit polling evidence is good enough here.

[Land Shark]

75. well the graph above certainly gives an MoE of 1% though i'm

unsure "DUers" (all of them) are entirely interested in hearing my comments on the matter since my last post, together with 5-6 posts under it, disappeared without a trace. It is referred to in post 67 of the "Return of the Clincher" thread.

[autorank]

70. Politics is just fine. Bush is going down, down, down....of course

one can always doubt the polls. When he gets to s certain point, people will wake up and realize, "hey, this guy is not at all trustworthy" and the "election fraud" argument will have it's fertile ground.

Look at this from Annenberg, prior to the election:

ONLY 62% OF REGISTERED VOTERS EXPECT THEIR VOTES TO BE ACCURATELY COUNTED, ANNENBERG DATA SHOWS.
http://www.annenbergpublicpolicycenter.org/naes/2004_03_votes-counting_11-01_pr.pdf

11/1/04

Republicans were “very confident” at a rate of 77% while Democrats and Independents shared that view at rates of 48% and 56% respectively.

The total for the poll was 62% of respondents were “very confident” that their votes would be accurately counted.

Well 42% not "very confident" pre election 2004 is weak ground if the legitimacy of this government is questioned.

At this point, we've had all the questions about 2004 raised and disseminated haphazardly. What do you think the confidence rate is now, even 50%? I don't know but a drop of 12% from pre election opinion is not at all unreasonable. What do you think the confidence level will be when CM (corporate media) actually covers election fraud?

This isn't an academic debate or an effort to help OTOH persuade his friends, it's about the fury of the people when they realize that they were had. The exit polls will be part of the rationale that is put forth to question legitimacy, the "hack-ability" of the system another part, and various acts of fraud uncovered in Ohio and elsewhere will arise also. In addition, we can expect the public to pay attention to new and current whistle blowers.

The only consistent paper trail is provided by the exit polls. However, the other consistent features of 2004 are voter suppression, lousy voting machines, insecure networks, and "vanishing ballots" that were never real in the first place. These elements all go together and are mutually reinforcing and supportive. For example, in Ohio, Kerry beat Bush as per the exit polls AND there is tremendous evidence of voter suppression, problems with machines, bogus recounting procedures, and bizarre happenings (the fictitious national security alert for Warren County). One more time for everybody: these elements are related and mutually supportive.

We were had, no two ways about it. That's the conclusion people are about to reach, in my opinion, and when they do it will be a great day for America. That conclusion is, as they say inside the beltway, "actionable."

[Febble]

90. interesting

and depressing.

But it seems sensible to probe your own case for flaws in advance, even if you still present the case.

Isn't this how you prepare a case?

In this instance, the first thing a cross-examiner is going to do, I would have thought is the same thing a peer-reviewer would do - raise the issue of false recall, especially if, as in this case, he can cite prior evidence that people do, in fact, tend to falsely recall having voted for the incumbent. This argument needs to be addressed. Simply to accuse OTOH of being a Republican, or worse, as has been done here (not by you), for raising it, seems rather ostrich-like, to me. It is your cross-examiners who will raise it.

Ditto with New Hampshire. NH was recounted, to the apparent satisfaction of those requesting the recount. The vote count had a large "red shift" relative to the poll. Any cross-examiner/peer-reviewer is going to raise the argument that the NH recount results appear to confirm that the exit polls can differ widely from the count, even when the count is correct.

It is not freeping to point this out. It merely points to the importance of gathering evidence in NH if you want to make the case that the exit poll discrepancy "could not" have been due to polling error. NH is a big obstacle to shift.

However, I think there are other grave problems with the exit poll evidence as evidence for fraud. The strongest is the finding, cited by Mitofsky in the press as well as in the E-M report, that bias was greatest where interviewing rate was lowest. This strongly implies that as the opportunity for non-random sampling increased, bias also increased, and leads to the interpretation that non-random sampling contributed to bias. If so, the MoE becomes useless, as it is predicated on a random sample.

Again, raising this does not make me, or anyone else, a freeper. It makes me what I am, a concerned non-American who cares about the way the president of the world's most powerful nation is elected, and keen to see a watertight case made for electoral reform. Which means your electoral system must be fair, secure and auditable. At present it is none of these things. My concern is that the exit poll evidence merely muddies the water, and is too easy to destroy.

Because in the end it doesn't matter whether Bush actually stole the election or not. The fact that he could have done, and that no-one could stop him, or know for sure that he'd done so, is what needs to be fixed. The essence of a democracy that the minority consent to be governed by those elected by the majority. If there is no confidence in the election (your point) the consent of the people to be governed is compromised. Whether or not the minority is, in fact, a minority.

[Land Shark]

91. Febble, one thing you can NOT do, given a bias toward careful

approach that you and others insist upon here, is say that the NH recount "confirms" an accurate election.

Why? Because there are numerous types of error that are not recount-detectable. One easy example is spoilage, another is voter suppression, a third is ballot loss or destruction.... there are more.

My belief is that there would be far fewer charges of freeping if the following two conditions were altered:

1. Those who think the situation here is urgent with regard to something as important as democracy, will be hard pressed to understand the good faith of someone who wants to slow down and raise every minute objection. How understanding would Romans be of Nero fiddling while Rome is burning? "Fiddling" = sophistry and academic debates over non-major points.
2. I've been on many legal teams all building toward accomplishment of an objective, up to and including million dollar fraud causes and a death penalty defense. YES, you do anticipate weaknesses and prepare for them, but you do so privately, and always with the express purpose of making the case stronger, and you never outright tell your lead counsel (much less publicly) that he has no credibility (or somesuch).

If these conditions of open debate among people that are supposed to be on the same team, I expect suspicions and charges regarding motive to come up.

Let's say for the sake of argument that your counterarguments have apparent validity, and that it "coulda been" low interview rates that contributed to all (or part) of the exit poll discrepancy. You know, even in the death penalty case, we had what we thought were great arguments at every turn, yet our client was convicted, but not sentenced to death. (This was a victory under the circumstances, I was third chair on this team). The existence of counterarguments SIMPLY DOES NOT MEAN there is insufficient evidence even to execute someone!

Good faith team playing from my point of view would seem to require an overall reflection and affirmation of a conclusion on your part such as the one that follows here (but only if you agree of course)

1. While substantial evidence and argument suggests otherwise, the preponderance (51%) of the publicly available evidence favors inaccurate presidential election returns in 2004 such that Bush did not win by the 3 million vote margin indicated.

The above is not attempting to fairly summarize your views, or mine. But I think regular "reminders" of what the strongest possible defensible statements are in support of inaccurate or fraudulent election results would be helpful in showing continuing good faith (I think most people other than trained lawyers and academics do this naturally as they tend to be agreeable and diplomatic). However, the habit here seems to emphasize disagreement.

So if I were the mediator or facilitator here (I think some interesting things could come out of this debate here if it were moderated) I would advise you febble, otoh, foobar and mgr to

(1) state the pro-fraud case in the strongest defensible terms you can justify, identifying the standard of proof you are applying, and
(2) expressly deal with the issue of urgency; i.e. just how much time do we have to perfect arguments prior to further action and what burden of proof needs to be met prior to action.

On this last issue #2, I previously raised the metaphor of seeing a sign on a Mountain Road that says "bridge out ahead" and raised the issue of what amount of information, proof, and certainty is necessary before taking corrective action in the face of possibilities or counterarguments that the sign could be old, out of place, a joke, the road could have an easy detour available, etc.

There are (arguably ambiguous) signs on the election road that say "fraudulent elections ahead". What is the standard of proof necessary for someone who believes in democracy to take action?

I am saying that under conditions of urgency (a topic not directly discussed here but which I think has validity) some technical scientific objections really have no validity in terms of stopping the efforts toward corrective action. There are ALWAYS decent arguments on the other side, if one can hear them.

But what is the WEIGHT of the evidence favoring???

Under conditions of data and analytical secrecy that violate the rules of free inquiry and thus of science itself, how can you or anyone else insist that all the requirements of solid science be met IN THE FIRST PLACE when that task is inherently made impossible?

As a legal matter, the spoliatory inference (in most jurisdictions) is a jury instruction that tells the jury they are entitled to infer that missing data would be unfavorable to the parties that have the data but fail to produce it for public court. The public interest in election integrity trumps everything else, if there is even a true conflict.

[Febble]

101. OK

see my PM, But, for the record, what I said was:

NH was recounted, to the apparent satisfaction of those requesting the recount. The vote count had a large "red shift" relative to the poll. Any cross-examiner/peer-reviewer is going to raise the argument that the NH recount results appear to confirm that the exit polls can differ widely from the count, even when the count is correct.

It is not freeping to point this out. It merely points to the importance of gathering evidence in NH if you want to make the case that the exit poll discrepancy "could not" have been due to polling error. NH is a big obstacle to shift.

[Melissa G]

170. 11 precinct recount does nothing to account for Tabulator Fraud

This is not hard evidence of a correct count if there is an 'election machine management' error as Bill Bored might say. More legs for his view of the elephant of election fraud. A county wide recount would have been more useful but money and time were big issues as I recall.

[OnTheOtherHand]

103. I appreciate your thoughtfulness, and let me be candid

If I thought that I could advance the cause of election reform by lying through my teeth about my assessment of the evidence, I would at least seriously consider it. I don't. (I'm not saying that anyone else here is lying through their teeth -- actually, I almost wish they were. But I would be lying if I claimed to give any credence whatsoever to certain of the arguments.)

And of course, like presumably everyone else here, I think that people should be entitled to some standard of proof that an election _wasn't_ stolen, not that anyone should have to muster even a 51% probability that it was.

As a lawyer, I assume, you want to make the strongest possible argument for your side -- but that means the most persuasive, not the most extreme. If you assert, as obvious truth, claims that many people believe to be obviously wrong, this will not strengthen the case.

There are lots of interesting ruminations on the relationship between academic uncertainty and political discourse (for instance, Steve Schneider has a great website on politics and climate science).

[Land Shark]

105. law sometimes engages in "process review" to determine what

level of deference a conclusion deserves. If the process is excellent then higher deference is appropriate. The law already tries extremely hard to make election results unchallengeable, at least after 10 day statutes of limitations on election contests (in many states) expire (they start at certification of results).

However, the deference one may wish to give election results for the sake of "stability" reasons isn't justified by the transparency and security of the present process.

[OnTheOtherHand]

107. we may agree

although it might depend on the sort of "deference" being considered. I wouldn't support, for instance, setting up a Kerry shadow government. But I certainly don't argue or believe that it is A Bad Thing to question the 2004 election results because, after all, we must have Stability. (Lest anyone be confused, I don't argue that it is a bad thing to question the election results, period.) And indeed, if we want our election results to merit deference and to confer stability, the process needs to be more transparent and secure.

[Land Shark]

108. How an honest advocate and an academician might differ in answers

Question: Isn't it true that you haven't reviewed dataset X?

Honest Academician Answer: Yes.

Honest Advocate Answer: Though not important to my analysis, it's true I haven't reviewed it.

The advocate, while still being honest, is alert to ohw inferences may be used against her in the future and attempts to protect against them.

In cross examination one is not allowed full sentence answers though, so it is up to the other lawyer, on re-direct examination, to allow the opportunity to explain if the cross-examiner has insisted on yes or no answers.

So, OTOH, I'm not talking about taking extreme positions, but I am talking about covering all bases and taking all available territory, and claiming that territory as one's own at each reasonable opportunity (unfortunately some lawyers interpret this to mean "crowing" about one's claimed victories which is not very effective however)

Applied to the debates here, I don't think everyone's put the best construction on what TIA has said, thus they appear in the adversarial mode (taking all available territory and claiming success) rather than in the friendly mode of mutually strengthening positions. Then, of course, one's own words are then also not interpreted in the most friendly way. And no, i don't want to get into "who started it".

Flipping between team mode and adversarial mode typically leads to questions about loyalty in all communities of discussion. Sometimes this is merited and sometimes not. Motive is hard to divine, yet important. Trust, when deserved, allows for efficiency and effectiveness in work by elimination of duplication.

Simply put, if one is in attack mode, you leave yourself wide open to either "outta control" inferences or else "freeper/disrupter" inferences. Occupational hazard.

[OnTheOtherHand]

109. yes, these are inherent problems

I meant it when I said elsewhere that I would, and will, try to be TIA's ally.

But there is a danger in putting the best construction on arguments that are incorrect. And there is a danger in disregarding arguments that impugn the integrity of many good people who aren't here to defend themselves.

Or, OTOH, maybe there isn't.

[Land Shark]

123. There's a diplomatic or constructive way to say almost anything

so i don't agree with your "danger in putting the best construction on arguments that aren't correct". One is not proceeding blind when being constructive, and the failure to be that way leads to immediate negative consequences.

YOu could, for example, say "the issue of MoE needs substantially more exploration because ___________ will lead to potential vulnerabilities in presenting the best possible case." You've probably said similar things at some point, but it gets wrapped in the adversarial tone of battle around here and thereby lost.

Given TIA's out there (at some substantial personal cost) fighting for legitimate democracy, I question the tone. You can say almost anything, but laying intellectual landmines that are laborious to address or remove or tossing eggs is out of order.

Though i know you feel you've suffered abuse as well, if the train is heading at full speed in the wrong direction, land mines and brickbats are not the best strategy for a change of course, IMHO. It may however be somewhat more justifiable to clear the way of perceived debris....

[OnTheOtherHand]

142. I'm not keeping score

I am willing to try anything. I have tried being really polite. I have tried being really rude. I have tried being exquisitely detailed. I have tried being concise. I will keep trying.

[foo_bar]

114. ever read Hermann Hesse's "The Glass Bead Game" (1943)?

When he begins his studies at the university, the doctor, lawyer, or engineer is forced into an extremely rigid curriculum which ends with a series of examinations. If he passes them, he receives his license and can thereafter pursue his profession in seeming freedom. But in doing so he becomes the slave of base powers; he is dependent on success, on money, on his ambition, his hunger for fame, on whether or not people like him. He must submit to elections, must earn money, must take part in the ruthless competition of castes, families, political parties, newspapers. In return he has the freedom to become successful and well-to-do, and to be hated by the unsuccessful, or vice versa.

For the elite pupil and later member of the Order, everything is the other way around. He does not 'choose' any profession. He does not imagine that he is a better judge of his own talents than are his teachers. He accepts the place and the function within the hierarchy that his superiors choose for him - if, that is, the matter is not reversed and the qualities, gifts, and faults of the pupil compel the teachers to send him to one place or another. In the midst of this seeming unfreedom every electus enjoys the greatest imaginable freedom after his early courses.

The natural teacher is employed as a teacher, the natural educator as an educator, the natural translator as a translator; each, as if of his own accord, finds his way to the place in which he can serve, and in serving be free. Moreover, for the rest of his life he is saved from that 'freedom' of career which means such terrible slavery. He knows nothing of the struggle for money, fame, rank; he recognizes no parties, no dichotomy between the individual and the office, between what is private and what is public; he feels no dependence upon success. Now do you see, my son, that when we speak of the free professions, the word 'free' is meant rather humorously.

http://www.amazon.com/exec/obidos/tg/detail/-/0312278497/ref=x/002-4442829-0682457?v=glance&s=books

[autorank]

73. Refusal of open release questions credibility; selective release implies

collusion. Sorry, those are the facts. If you have nothing to hide, talk, open up, release the data. The election process belongs to the public, not you or Mitofsky or the venal network consortium. There are real questions that abound. Release the data and allow precincts to be identified. Let's see what happens. It is so painful to hear you ask us to prove the data would be useful when the data is not available. This is just awful reasoning, stunningly so.

Releasing things to ESI and febble, selectively, is an outrage. Who are they? Why should they be the arbiters of the most important decision of the new century? A broad release diminishes questions, selective releases, well... Your logic is painfully inadequate. It's reminiscent of the selective release of the Dead Sea Scrolls -- we know what that's about, conformity with the prevailing views of the holders of the data, in both cases.

Release the data is the appropriate position. You can't lose, if you really want to settle the issue. If it proves fraud, you win because that's new data. If it doesn't or is inclusive, you win again. But to say the following is making about you: "I am perplexed and troubled by the apparently widespread conviction that Mitofsky believes he has in his possession information that could save democracy, and/or that full release of the data would provide some huge advance in our ability to assess fraud" . The American public has a right to the data, regardless of you being "troubled."

Release the data to everybody, period.

[Name removed]

54. Deleted message

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

[OnTheOtherHand]

60. good golly

When you offer your opinion of how you think poll respondents' propensity to misreport past votes will affect other analytical conclusions, then we will have something to talk about, perhaps.

[TruthIsAll]

71. When you can explain to me why democrats would differ

from republicans in that regard, perhaps then we can discuss your hypothesis.

Until you provide the factual rationale for your inference that lying and forgetting is primarily a Gore voter trait, indeed, we have nothing to talk about.

Until you can provide a scientific basis for your great unifying theory that Repubs are like elephants and never forget, but that democrats are just asses who suffer from alzheimer's, you should not even call your theory a hypothesis. I would just include them in your faith-based intelligent design curriculum, along with that pathetic reluctant Bush voter theory.

[Febble]

120. See my post 118 above

There is certainly a scientific basis for the theory.

It has been shown that

a) people like to report consistency (so voters who report Bush-Bush or Gore-Kerry are more likely to be lying to maintain consistency than voters who report Bush-Kerry or Gore-Bush) and

b) people like to report having voted for the winner.

Lying is not the right word. "Misremembering" is usually used, and it is a well-researched human trait. We construct our memories, and reject bits that don't fit. It's actually how we stave off the fragmentation of memory that afflicts those with Alzheimer's and maintain our life narrative in some state of cohesion.

And to the extent that misremembering voting for the winner is the stronger trait, it would give rise to more people "misremembering" voting for the Bush than for Gore. I'm sure it's not a Democratic trait. I bet no-one remembers voting for Dole (does anyone even remember Dole?)

As I say in the post above, the more striking thing is that if misremembering accounts for the numbers (and they are well within the reported "up to one in five") then it does mean appear to mean that Kerry lost more than 10% of the Gore vote (although the MoE on that sample will be quite high - it was a sub-sample), while gaining the majority of those who did not vote in 2000.

But a swing to the left in young voters and a swing to the right in older voters is certainly not unprecedented.

Suggestive, I would say, therefore, only. By no means a clincher.

[TruthIsAll]

128. I would think that Democrats would be more likely to remember that...

Bush stole it and that it was Gore who won the popular vote. I would think that would be engraved in their minds permanently. They would never forget the 36-day Florida fiasco, the butterfly ballot, the overpunched and underpunched Gore votes, Jeb Bush, the thousands of disenfranchised blacks, the 96,000 Nader votes, Volusia, the Miami-Dade GOP riot stopping the recount, Katherine Harris, the the 65-4 SCOTUS vote for Bush...

Never forget it. Never get over it.

Not only that, but a signifcant fair-minded republicans, true conservatives, would feel the same way. They would be vastly more motivated to kick the thieving bastards out. Especially after 9/11, no WMD, lost jobs, American prestige in the toilet, Abu Ghraib, Plame, AWOL, the Bin Laden/Bush/Saudi connection,

But perhaps I'm just being old-fashioned in believing that when people get kicked in the gut, they are very unlikely to forget who did the kicking.

[Febble]

141. Now that's what I call a valid argument

and is the reason I myself was predisposed to disbelieve Bush's victory.

But, as I've said before, it's a behavioural argument, not a statistical one. And for many people, politics doesn't matter that much. Weird. But there it is. Some people don't even bother to vote.

[Melissa G]

76. Hi TIA and OTOH, Let's get back to civility.. You both have some

interesting things to say. I think you may not be re-reading your own posts from someone else's POV. Going through this thread having read probably all of OTOH's's posts and most of TIA's since Nov. I can extrapolate what you both intend but you both leave things out that you think are implicit. I believe this is one of the things OTOH is IMO attacking TIA for but OTOH does it also. If we are asking one another to clarify perhaps we could STOP here and ask in a more gentle manner. I know I will probably be slammed again for bringing this up BUT not attacking folks is a Good DU Rule.
Peace,
Melissa

[autorank]

78. There is a difference between style and substance.

So what if the debate is vigorous. Some like it some don't. That is a stylistic question. The question that's more important is what are the substantive claims of each poster. When you look at it that way, you've got:

TIA: serious questions about the legitimacy of the election(Land Shark's point on this thread, "legitimacy); serious questions about the connection between funny numbers from exit polls and outrages on the ground; serious questions about defenses of the exit poll discrepancies and inconsistencies including the demolition of "reluctant Bush responder;" a demand for release of all the data (save that which identifies the individual) including Precinct level data (which is essential to validate the fraud hypothesis); over 100 individual threads where the work is shown in detail and anyone who wants can question the work; years on DU; an unabashed Democrat.

OTOH: questions, questions, questions.

When you go to Paris or New York City, you can get hung up on the superficial style of the denizens and miss the true substance of the people's character (in both cases wonderful) and the great assets the cities have to offer.

I'm here for substance. The substantive contributions, arguments; the substantive demand for election reform and investigating election fraud is 100% in TIA's corner. That's why I pay attention what he produces and find it compelling.

It all ties together-- problems with voting machines -- take a look at how odd the results of elections are--both issues interact and support each other; problems with hacking -- take a look at how odd the results of the elections are -- both issues interact and support each other; problems with obvious voter suppression -- both issues interact and support each other.

IT ALL FITS TOGETHER AS A STRONG CASE TO DOUBT THE LEGITIMACY OF BOTH PRESIDENTIAL ELECTIONS (AS YOU SO CORRECTLY POINT OUT IN A LATER POST DOWN-THREAD). YOU'RE *NOT* BEING SLAMMED, JUST REDIRECTED TO...

SUBSTANCE.

[OnTheOtherHand]

94. but your paraphrase does not match the subject headers

What TIA and you offer at the top of this thread is not "serious questions" or a "strong case to doubt" -- it is absolute certainty, coupled with what looks to me like a preemptive smear of anyone who dares to challenge the case.

Which is why I will ask, yet again, because I don't think anyone has answered me yet: am I supposed to take the arguments seriously, or are they just intended to rally the troops?

Because the dual claims that (1) we _know_ that Kerry won the election, and (2) anyone who disagrees is (whatever) are, in my opinion, incorrect, incendiary, and counterproductive.

If you and TIA actually stuck to "serious questions" and a "strong case to doubt," you would never get guff from me. Lots of people around here offer serious questions, and I have serious questions myself, and we actually try to make progress on them together.

[TruthIsAll]

104. OTOH, let's review where we are right now

I have presented a sensitivity analysis of 120 scenarios of Bush/Gore voter turnout and Kerry share of new voters. You appear to object very strongly to this study on its face.

OTOH, do you know the purpose of sensitivity analysis? Have you ever used this methodology? Just because I am using this very standard approach to illuminate the implausibility of a Bush win, you accuse me of rallying the troops and avoiding the facts.

On the contrary, as you purport to be a quantitative analyst, you should commend me for the effort. There is no justification in your personal attack. Why does it bother you so?

To your credit, you have attempted to reconstruct the numbers. But I must give you an I for Incomplete, professor. You neglected to check your work and fell 4mm votes short (118mm), coming up with a virtual tie vote. I then showed you that the shortfall had to be due to New voters and since Kerry won a clear majority of them, he also wins the scenario - as I said he did.

You have not refuted that.

Well, let's take a look at the model.
Let's look at my assumptions.
Then we will look at your facts.
And we look at your assumptions.

I gave Bush (and you) the benefit of the doubt in the assumptions.

-I gave you a 100% turnout of Bush 2000 voters.
You have not refuted that.

-I gave you the final exit poll 37% Kerry share of 2004 voters.
You have not refuted that.

-I gave you the final exit poll percentages which had Bush a winner at 13660 respondents because the numbers were matched to the vote. All prior exit poll time lines up to 13047 at 12:22am showed that Kerry was the winner.

You have not refuted that.

In spite of all these assumptions in favor of Bush, Kerry still won all the scenarios.

You have not refuted that.

So now you resort to shrill, desperate personal attacks.

You said I had "a lot of nerve" because I agreed with the Mitofsky MoE table, which shows a 1% MoE for exit polls over 10,000 respondents.

You have not refuted that.

And what is the reader left with after all of this? Your "false recall" argument. It is a very weak, unsubstantiated faith-based case when juxtaposed next to the numerical voter turnout sensitivity analysis, which is based on facts, not "faith".

You are asking us to assume "on faith" the hypothetical that Gore voters forgot that they voted for him in 2000 when asked the question. But at the same time, we should also accept "on faith" that Bush voters did in fact recall who they voted for. Do you realize how utterly ridiculous that conjecture is? I did not base the sensitivity analysis on faith.

I understand your position. When you are devoid of any plausible explanation to refute the sensitivity analysis of cold, hard numbers, you must resort to faith-based arguments.

You are left with a very, very shaky hypothesis based on your last gasp 2002 poll argument. I have asked you for details on that poll, as if it is even germaine to my analysis. It's a Hail Mary pass on your part. Still waiting.

As Land Shark has stated, this argument brings us to the abyss.
As as he has also pointed out, there is absolutely nothing wrong in the way I have presented the analysis - to show the extreme implausibility of Bush win since Kerry won all 120 scenarios.
You have not refuted that.

Know this: I will keep posting the analyses. And they will continue to withstand your vain attempts to refute them.

Eventually, you will just give up trying.

[OnTheOtherHand]

106. a few corrections

since I can't go through every argument every time.

Yes, I am familiar with and have conducted sensitivity analysis. But of course the validity of the analysis can't be measured by the sheer number of scenarios tested. You could have run one scenario, or five million, and the force of the argument would be determined by the limiting case.

I did not fall 4 million votes short, but you come to this conclusion by apparently failing to read my post #9 beyond the line labeled "Totals." Not that I am wedded to the assumptions in that post.

As I explain elsewhere (at least twice), you cannot extract the marginals for reported 2000 vote and apply them to your estimates of actual 2000 vote. That is your faulty assumption.

I will let others tally the personal attacks as they will, but they have no bearing on the arguments regardless.

Since you have ducked the question twice: the E/M table reflects a design effect square root (multiplier) of approximately 1.5, at least if we can trust the "15" at lower left. Be advised, however, that these are described as "typical sampling errors" because the design effect varies from question to question. This is something you could actually read about.

You say I am assuming some desperate hypothetical. But false recall is demonstrated in many surveys, and I and others have provided the figures and the links, and you have ignored them. The "faith-based" and "Hail Mary" rhetoric is, well, rhetoric.

[TruthIsAll]

110. As I expected, your counter is very weak..

You:
Yes, I am familiar with and have conducted sensitivity analysis. But of course the validity of the analysis can't be measured by the sheer number of scenarios tested. You could have run one scenario, or five million, and the force of the argument would be determined by the limiting case.

Me:
The number of scenarios reflect incremental changes to the assumptions. It does indeed increase the force of the argument.
You can't argue that any plausible scenarios were ignored.

You
I did not fall 4 million votes short, but you come to this conclusion by apparently failing to read my post #9 beyond the line labeled "Totals." Not that I am wedded to the assumptions in that post.

Me
But you do not dispute the fact that, contrary to your premature result, Kerry won the scenario. As he won all 120 scenarios.

You
As I explain elsewhere (at least twice), you cannot extract the marginals for reported 2000 vote and apply them to your estimates of actual 2000 vote. That is your faulty assumption.

Me
Sorry, you have never refuted the fact that Bush could not have received 43% of the 2004 vote. This is your Achilles heel and you know it.

You
I will let others tally the personal attacks as they will, but they have no bearing on the arguments regardless.

Me
Nevertheless, I have noticed a distinct change in your attitude.
Maybe its due to frustration on your part.
You never expected I would be so relentless, did you?

You
Since you have ducked the question twice: the E/M table reflects a design effect square root (multiplier) of approximately 1.5, at least if we can trust the "15" at lower left. Be advised, however, that these are described as "typical sampling errors" because the design effect varies from question to question. This is something you could actually read about.

Me
Then why is the MoE = 1.0% for 10,000 respondents?

You
You say I am assuming some desperate hypothetical. But false recall is demonstrated in many surveys, and I and others have provided the figures and the links, and you have ignored them. The "faith-based" and "Hail Mary" rhetoric is, well, rhetoric.

Me
I have yet to receive your response regarding that 2002 survey.
I asked you to document the purpose, advocates, executors, internals and results of the poll. You have shown us nothing. It was just a poll. The numbers were not in agreement with the actual 2000 vote.
I fail to see relevance nor any facts to substantiate your "false recall" hypothesis.

Is that the best you can do?

[OnTheOtherHand]

111. of course I believe that a plausible scenario was ignored

namely that Kerry won less than 90% of actual Gore 2000 voters who voted in 2004.

I have never even _denied_ the fact that Bush could not have received 43% of the 2004 vote. That isn't the argument. I tried to understand _your_ argument, now it is your turn.

I have never doubted your relentlessness. I wish it were more discriminating.

The NEP table does not assert that the MoE is "1.0%", as anyone can readily verify. There is obviously some freehand rounding in the NEP table, since it is mathematically impossible for the cut-point between 2 and 1 to be at 8000 voters in both the third and fourth rows. However, the 15 at lower left seems to imply a design effect square root between 1.45 and 1.55. Conveniently, a DESR of 1.45 would yield an MoE of precisely 1.45 for n = 10000, and a DESR of 1.55 would yield an MoE of 1.55. The mean value, 1.5, could be rounded either down to 1 (not 1.0) or up to 2. If you can find a DESR estimate that fits the table values better, I would be interested to know.

I sent you a link from which you can learn all about the National Election Study, if you care to. Yes, the 2002 NES was just a poll (and the other polls in which false recall has been documented were just polls). The exit poll was just a poll. Why would you assume that false recall in the 2002 NES would have no bearing on false recall in the exit poll?

[TruthIsAll]

113. SURELY, YOU JEST. KERRY LESS THAN 90% OF GORE VOTERS?

What are you smoking?

I would like you to rationalize over 10% of Gore voters
willing to forgive Bush for stealing the 2000 election by
voting for him in 2004.

Do you really believe that over 5 million Gore voters would
vote for Bush? If anything, there is substantial anecdotal
evidence that Bush lost more voters to Kerry than vice-versa.

Get real.

As far as the MoE is concerned, you are really stretching it.
The Moe for 13047 voters is 0.87%. Go ahead. Add your 50%
cluster and you are at 1.3%. That's if you buy 50%.

I will read the link.

[OnTheOtherHand]

115. I am not rationalizing anything

I am _reasoning_ about the implications of the fact that, in surveys, some folks who actually voted for Gore in 2000 (or who did not vote) presumably report having voted for Bush in 2000. That is not something I desperately made up; that is something supported by the 2002 and 2004 NES, the 2002 GSS, and by analogy in a host of other studies from the U.S. and Britain. (And no doubt some Bush 2000 voters now "recall" having voted for Gore, although all survey results known to me indicate that the tide runs in the other direction. Incidentally, Febble has pointed out that the tide of false recall does not always favor the incumbent; she knows of at least one counterexample in Britain.)

The fact that you and I can hardly imagine having voted for Bush in 2000, much less falsely remembering having voted for him after the fact, does not constitute a refutation of poll results evincing false recall.

I think that my quantitative evidence can at least hold its own against your "substantial anecdotal evidence." Of course, if you can come up with some reason why (or at least some evidence that) exit polls are uniquely immune from false recall, then we have something further to discuss.

When it comes to the MoE, please bear in mind, again: I am not trying to argue that the exit poll error was within the range of sampling error. It is worth getting the MoE calculations right because any calculation worth doing (yadda yadda yadda). It also may shed some light into why no one called Ohio and Florida for Kerry based on the exits, although that could be explained in lots of ways.

While we are at it, let us consider what we are arguing about. I get the sense that, in your heart, you think I am trying to convince everyone that Bush won the 2004 popular vote. I am not. I plunged into this thread based on your invitation (dare?) to refute your arguments one by one, but even if I had time to undertake that challenge, I would not wish to dispel real, legitimate questions and doubts about the 2004 outcome. I sincerely deny that your 120 scenarios are any sort of "clincher." I think we need to look for clinchers somewhere beside the exit polls -- and, for that matter, somewhere beside the past. And I think we need to recognize that a lot of people think that Bush won the 2004 election, and probably a lot of them will continue to think that, and in order to get election reform we will need support from some of them, too. (And, yes, I am one of those people, but it's just a current personal opinion, not a cause I intend to prosyletize for.)

[autorank]

121. Reality time...

"And I think we need to recognize that a lot of people think that Bush won the 2004 election, and probably a lot of them will continue to think that, and in order to get election reform we will need support from some of them, too."

We do not need to agree with those who think Bush won, nor they with us, in order to agree on the need for election reform. That makes no sense at all. Everyone who wants honest elections, and I'm sure that includes the overwhelming majority of Republicans and Democrats, will agree on reasonable procedures. The problem is that the HAVA and other methods of "reform" are really just vendor jamborees where decisions are made based on schmoozing rather than real issues. We could go back to paper ballots tomorrow and provide machines for the handicapped from the current stockpile and every objection to the election, from the standpoint of machines would be answered. Then the observation of the count, tally and reporting could be monitored as well, something everyone except the thieves would agree on, and we'd be done with it.

The theft of the election 2004 is a separate question. Separate arguments.

I've noticed that you're starting to deny the entire validity of polling in your discussion of 2002. You are, at times (not always) reminding me of Elizabeth Loftus PhD. She made a name for her self as a "researcher" and defense witness asserting that eye witness testimony is rarely valid (radical epistemology). In the case of 2002, who cares, it's after the fact. In the case of the exits, we care a lot because it is contemporaneous data collection. "Lets see, I just voted for Kerry and I'm going to talk to this pollster, so, yeah, I'll lie"...please.

If you think that polling evidence is as tainted as it is, why are you even involved in studying it and why are you taking so much time with people who do, according to your view that reporting is largely meaningless?

On edit: And the reality is that as * approaches the mid thirties, people will cry out and ask how did this guy get selected/elected twice. We know he stole Florida (that evidence will come forward)! Damn it, we now suspect he stole 2004 (that evidence is there). That's the political reality. The argument that prevails will follow the political realities. Formerly, only a minority thought it was a stolen election. Now we're in a state of flux (I assume since I have not seen any polls, which would be meaningless by your criteria anyway). Soon, the * team and the apologists for Election 2004 will be in full retreat because the majority will be willing to listen to and accept the arguments we've been making. The arguments are valid and persuasive regardless of the popular mood and trends but they will be come delightfully explicated as the crescendo grows for relief from the misery brought to us by lousy elections and judicial decisions, i.e., * and company.

[OnTheOtherHand]

143. excellent, we have a huge and crucial agreement

"The theft of the election 2004 is a separate question. Separate arguments."

Yes.

"I've noticed that you're starting to deny the entire validity of polling in your discussion of 2002."

No.

Look, false recall is an observed empirical phenomenon -- we can't (or shouldn't) just discount it. I don't see how recognizing sources of error in polls is tantamount to arguing that polls are meaningless. In fact, it isn't. (But there may be a confusion here: I am not suggesting that Kerry voters, or anyone else, consciously lied in the 2004 exits.)

If you are counting on convincing the country that Bush stole the 2004 election, I hope we can find better evidence. Otherwise, with respect to the election reform agenda, I prefer a Plan B.

[Melissa G]

131. I have a problem with you OTOH, I do not believe you are reasoning

.. I believe you are just doing crit...You may not have a problem with this but I do..
This is why..
TIA is doing something. Quite a lot of Something. Febble did something also. They produced work. They put their work out there and took and take a lot of heat for it. I have a lot of respect for both of them because of it. I agree with TIA (though not so much with Febble, much as I might possibly like her) but I respect them both for their efforts and the C### they have to take because of it.
Now we come to you, OTOH... You have a skill set and you had promise. I had my popcorn ready. :popcorn: I was informed that you were likely an earnest fellow so in the face of a fair amount of heat I have stood up for you to say your piece in peace. Mostly, so far, you have delivered nit picky goobledy gook. I have waited for a decent effort on your part to put up a coherent counter argument or alternate scenario. IMO it has yet to be presented. This IMHO BS argument about any significant number of DEM's forgetting the election was stolen from them in 2000 and misremembering their vote is IMHO LUDICROUS! Febble is in the UK she could maybe get away with saying this but you, as an American, must have been on Mars at the time to try to get away with that statement. I own a business...Everyone was glued to their tv's at that time... Restaurants that were usually full had only three tables being occupied. Commerce shut down. It is is ridiculous to suggest some huge memory loss on something that impacted folks so strongly.. It was economically as impactful as 9/11 to many small business owners.
OK, for some to my mind, utterly incomprehensible reason, you believe the Shrub won the election. (Damn few of us here do). Make your case.
Explain to us how you believe this sorry excuse for a resident who had driven the country into financial disaster and lied us into a war the majority of the country did not want in the first place kept all his votes and found 9-11 million more?
'Splain it to me please... (Edit clarity)

[Febble]

134. Well thanks, Melissa

but I need to correct you on the subject of HARD WORK. Just because you weren't there with the popcorn to watch doesn't mean it wasn't done.

Quite apart from a HUGE amount of work for USCV, OTOH last week delivered the substance of my paper at a meeting of the ASA in Minneapolis, having developed it very much further (in collaboration with myself and Rick Brady) than I had done.

And, contrary to the way my work has been interpreted on DU, what I started to do, and what OTOH, Brady and myself have continued to do, is to attempt to develop a much more sensitive measure of "bias" (ie. discrepancy between poll and count) than the WPE, so that more valid analyses of the factors that contribute to the bias can be done. It's not as easy as it looks. It's not as easy as I thought it was. In fact it's been HARD WORK. But it's important, because blunt an instrument as the exit polls are, anything that can be done to sharpen them is likely to shed light on what went wrong. People scoffed at my work when I described it as "noise reduction". But noise reduction is what is required here.

But here is the problem I have:

If we (generic we) start from the premise that Bush cannot have won the election, ANY argument to the contrary will fail. OTOH (and I) raise the issue of false recall - you laugh it out of court, despite good evidence that it happens. We raise the issue of sampling bias in the poll - it is laughed out of court ("do you believe such a wonderful pollster as Mitofsky could have run such a bad poll?). We raise the issue that the pattern of bias in the poll does not support the hypothesis of highest fraud in Bush strongholds - or widespread vote-switching - and are accused of - I dunno, statistical incompetence, using bad data, you name it.

Fair enough, if you believe Bush lost, you are free to believe it. I wish I still believed it (though I still think Kerry may have won Ohio had the playing field been level). But the fact is, it is possible to wish Bush had lost, to believe that the election was corrupt, to believe that electoral reform is vital for the future of American democracy (and for the world) - and to believe that he probably won. Furthermore, it is possible to believe that the best hope of electoral reform is to look this possibility (probability, in my view) squarely in the face, and to construct an argument for electoral reform that is consistent with the statistical evidence as it is viewed by a great many statistically literate people who also support the case for electoral reform.

Because OTOH, and I, and others, do not believe that TIA's evidence is anything more than suggestive (if that) and certainly not "the clincher" that he claims, does not mean that we do not earnestly want, and work for, electoral reform. It means that we want good, not flawed, statistical arguments to make the case. TIA's interpretation of the data he cites (in itself) is perfectly plausible. So is OTOH's. Statistics will not tell you which is correct. Suggesting they can is a misuse of statistics, and sets up a straw man which, IMO, runs the risk of damaging the case it attempts to support.

And no, I cannot explain how Bush found those votes, if he found them. But then I cannot explain how Britain voted for the disaster that was Thatcherism for all those years. But it did. And the only way Labour could finally regain power was to become, in some senses, almost as Thatcherite as Thatcher. Almost. And not in all senses, thank God.

[Dynasty_At_Passes]

135. Who are these people in this paragraph?

Why has Rick Brady, the United Kingdom nor anyone else ever discussed this telling document of collected evidence?

http://www.answers.com/topic/2004-u-s-presidential-election-controversy-voting-machines

What exactly is the reasoning for hiding this information, I found it through a web search. This was nowhere else to be found, and not one single polling firm, not even Edison Mitofsky has brought up this information. Why not, is the answer I want.

[Febble]

136. Ha!

They even cite my work on machines in Florida.

It's an interesting compilation. What is your point?

[Dynasty_At_Passes]

137. My point is it proves an awful lot, and nobody is talking about it....

Including the mentioned Rick Brady or pundits on TV like Tim Russert.

Why the silence? An astounding amount of evidence, and real reports, backed documents is shown that the elections are fradulent in at least 2004, 2002, and 2000. Yet there is not a single report about it anywhere. Why?

[Febble]

138. It's a Wikipedia

compilation. Lots of people are talking about it - which is why it exists. It's how Wikipedia exists. Some of the reports it links to have been reported quite widely, and also widely critiqued. Some have not stood up to critique (including my Florida stuff, probably).

Not sure what Rick Brady has to do with your point though.

[Dynasty_At_Passes]

139. I don't think its from Wikpedia.

All I know is it was compiled by a forensics group from answers.com and has never been refuted.

What I'm trying to get at, is if the stolen elections evidently conducted by both major parties are such a huge national issue why in the world is no one reporting about this?

Like the mountains of evidence here, yet no one is talking about it on TV in statistical terms.

In order to take elections seriously, one must be able to see the whole picture. What you have here is outright proof of people being bribed and coerced into doing very unethical, and other illegal things.

Legally, these machines should not even be in operation. Why? Well if you read the proof there, they are violating the voter protection clause and also were NEVER legally state certified.

Apparently they were all certified by Ciber systems and some other conducted ISO groups, who have never been questioned by federal authorities. They were endorsed by a rank and file committee called NASED.

Under that legal requirement, the equipment here was not secure nor even suitable for electing a city council person. Yet they are being dropped into cities left and right anyway, so how in the world could this get out of control so quickly?

What kind of conspiracy is this?

[Febble]

140. The piece you linked to

gives the source as Wikipedia and a link:

http://en.wikipedia.org/wiki/2004+U.S.+presidential+election+controversy%2C+voting+machines

I don't know whether your link is updated as the source file is updated. If not, the two will get out of synch. But the material seems currently identical in both places.

[autorank]

157. So what! Comment on the posters substantive issue. This is damning

evidence. What do you have to say?

[Dynasty_At_Passes]

162. From what I can gather they did comment....

And said by and large the machines are a problem but didn't go into specifics.

What I'm citing here has serious national implications, and should be less about political gamesmanship and more about security.

So evidently it seems to me, people choose to make this about the democrats having elections stolen from them but fail to account for republicans who have elections stolen from them as well.

McCain? All the primaries, or honest republicans losing their seats. Same for dishonest democrats taking seats in cities like San Francisco, does Dianne Feinstein ring a bell?

Stolen elections are not about scorecards, but are about security. Why have you NOT made this the real puncher?

[autorank]

167. I agree. This is a bipartisan issue. Specifically...

The machines, the software, the vendors, the networking, the lack of security, the lack of real certification, the personnel, the arrogance of bureaucracy, the docile public, the cynical, sleeping CM (corporate media)...all of these make a mockery of democracy. I've never said corruption is exclusively the province of the Republicans, although it certainly seems that they're the big winners: Congress, the presidency (twice).

There should be nothing controversial about Voting Rights. The reason the act passed 40 years ago was simple: no one could mount an argument against a person's right to vote. This time around, the voting rights movement has to establish fraud in order to make the connection between the right to vote and the problems with the current system.

Interestingly, there has been a voting rights issue with election procedures for decades called "spoiled ballots" which routinely make up 1-2% of the presidential ballot. These types of disqualifications take place 7 times as often in minority precincts, typically black precincts. We all tolerated this as though it were a fact of life but, in fact, it was blatant racism operating to steal the votes of a specific segment of the population.

Part of the answer to your question about the excellent summary in Answers.Com is "go along to get along." If any established part leader stands up and says, "It's a sham," they will threaten to topple the whole system because "it's not legitimate" follows closely after the first statement.

I say, let's put it all on the table, go back to paper as the only solution, and make appropriate accommodations for handicapped voters with our 'voting machine surplus.' Simple isn't it. Paper, trail of evidence, observed counting, recounts are possible, precincts announce the results so no temptation to mess with network transmissions.

That would be easy and it would solve the problem.

Why don't you think a major political figure from either party has said a word?

[Dynasty_At_Passes]

169. Complicity? Money? Racketeering? All of these charges are equally valid..

Does it not seem odd to anyone that the DLC never talks about the machines? It is a fully forbidden topic, as in, it may be how the DLC type democrats are getting elected into office through under the table means?

Why can't all citizens follow this initiative?

http://rawstory.com/news/2005/Citizens_Request_Recount_IN_SAN_DIEGO_MAYO_0818.html

This proves without doubt, and with backing of the authorities that illegal machine tampering was putting GOP candidates and dishonest politicians into office by a 4% swing!


Someone is making big bucks off this scam, and the audit proves Donna Frye won the election. Make a huge issue of this everywhere, and Diebold/Sequoia/ES&S will retreat and be forced to abandon the game.

But too many leaders seem to be unwilling to stand up to these coy crooks, as if they are completely unworthy of opposing the establishment. Please get over it, they say. Well as long as we get crooks into office I say never get over it, and lets sink these losers with a full assault.

Bob Taft convicted of bribery pay to play, and Thomas Noe covering up the election. Ring any bells?

[autorank]

192. We agree totally...a "bright light" will scare the crooks off even as

we get checks and balances in place. San Diego is inspiring and so is Arcata, CA's resolution. The money involved must be huge given the stakes, literally control of trillions of dollars, a kick-ass military, and a court system that can enable great progress toward repealing the New Deal (the Holy Grail of the extremists). It is appalling that there is NOT ONE major Democratic leader shouting from the roof tops on this. It's just pathetic. I've been complaining about this, locally to my party folks, and nationally for some time.

[Febble]

172. the poster asked

why this compilation of evidence had not been reported on.

My answer was that a lot of it had, and that the report itself was a Wikipedia piece. Nothing wrong with that, but it's not itself a published report. Which might explain why it hasn't been reported on as a piece, although it contains material that has been reported and critiqued fairly widely, on the internet. In fact the piece cites some of the media reviews that the material has received.

As to the substantive material in the report, well, as I said, some of it's stuff I did myself (on machines in Florida), so I could comment on that. The issue of whether touchscreen or optical machines benefited Bush or Kerry in Florida is difficult to resolve as touchscreens were not used in small counties, and optical scans were not used in very large counties, so there was a major demographic confound. I looked at mid-size counties only, in order to compare counties where both types of machines had an equal chance of being selected, which also eliminated most panhandle counties, where it had been suggested a "Dixiecrat" effect might be a confound. Mitteldorf and I concluded that Bush did significantly better that expected in counties that used optical scanning. However there were significant demographic differences between the two groups of counties even so (percentage of party registrants; percentage of ethnic groups), so the differences cannot be clearly attributed to the machines. Moreover, Hout's Berkeley team found that when the whole state was considered, and a large regression model was used, it was the E-touch counties that looked suspicious. But it turned out they had omitted a crucial term (interaction between machine type and county size).

So regarding machines in Florida, I think machine type doesn't tell us much, except that statistics is an enfuriating field, and doesn't give the unambiguous answers people would like. The evidence certainly doesn't rule out fraud - as people here have pointed out, any machine can be used to implement fraud, and in any case, if it's the central tabulators that were hacked, then the voting machine type won't be the critical factor.

So there's my comment on the one bit of analysis cited that I know something about - because I did it.

[Dynasty_At_Passes]

173. But this doesn't answer the underlying question....

My point is, why isn't this study about the Florida counties being widely discussed not to mention dissected?

It seeks to make the issue more complicated by factoring in demographics and all these other variables, and from where I'm sitting, this is not a complicated issue in any respect.

http://rawstory.com/news/2005/Citizens_Request_Recount_IN_SAN_DIEGO_MAYO_0818.html

Reading this information and on how they weighted the results, has shown this issue is very cut and dry and displays a clear factor of certain machine vendors exhibiting unexplained 4% swings.

It isn't complex, it isn't a huge enigma, its a very simple problem largely and it has gone under recognized and not reported. The problem isn't with the counties, all the different demographic adjustments or anything else and clearly that is not the purpose here.

The problem is with the machines plain and matter of factly, and the machines by vendor, not by type are the culprit. A Diebold manufactured tabulator or ES&S tabulator are almost guaranteed to show a disproportionate swing or be involved in some fraudulent activity. Another machine/tabulator such as the new model by Hart Intervic is about 90% less likely to show gross manipulation or errors. That's the clear problem, and there is not any study to address it.

I ask about the study by registered machine vendor and lobbyist, not by model, type, or demographic. What has become of this, given the reported bribery charges and allegations?

[Febble]

175. Sorry, which study are you referring to?

You linked to a compilation at Answers.com which included links to work done by myself and Josh Mitteldorf, and by a team at Berkeley.

I did analyse the data by vendor (Sequoia, ES&S, Diebold, etc), and found no significant effects. Machine type was what I found significant - but because machine type was "confounded" with demographics, it is not possible to say whether any difference was due to machine type or demographic differences. It was suggestive, nonetheless.

If you can link directly to the study you are referring to, I may be able to answer more fully.

[Dynasty_At_Passes]

176. I've been referring to this parallel study that has been ongoing.

http://www.wired.com/news/evote/0,2645,65665-2,00.html?tw=wn_story_page_next1

http://ucdata.berkeley.edu/new_web/VOTE2004/election04_Sum.pdf

http://www.scoop.co.nz/mason/stories/HL0310/S00211.htm#66

They discovered that in the 15 counties using touch-screen voting systems, the number of votes granted to Bush far exceeded the number of votes Bush should have received -- given all of the other variables -- while the number of votes that Bush received in counties using other types of voting equipment lined up perfectly with what the variables would have predicted for those counties. The total number of excessive votes ranged between 130,000 and 260,000, depending on what kind of problem caused the excess votes. The counties most affected by the anomaly were heavily Democratic. "
"Sociology professor Michael Hout, who chairs the university's graduate Sociology and Demography group, said the chance for such a discrepancy to occur was less than 1 in 1,000. 'No matter how many factors and variables we took into consideration, the significant correlation in the votes for President Bush and electronic voting cannot be explained...'"

When you say "significant effects" what differences are you citing? Doing a study by machine vendor, would mean studying all of them in counties which were democratic or republican, I.E. the largest counties and analyzing the swing percentage on a statewide basis. I have seen no such study investigated.

[Febble]

194. I wondered if you meant that study.

It was interesting. I'm not convinced however, for the same reason that I am not, ultimately convinced by my own. IIRC there was a critical variable they failed to include in their model (county size x voting machine type). There have been a number of critiques of the study.

The study I am "citing" regarding machine vendor was my own. I investigated machine vendor across all counties, and had data on party registration for both 00 and 04. I did not find anything significant, so have not reported anything. The only thing I found were machine effects, which, as I have said, do not necessarily convince me. I found greater Bush advantage for Optical Scanning. Hout et al found the same for mid-size counties but an advantage for e-touch in large counties (Broward I think was the greatest). I excluded these for reasons I have given, so the studies are not actually at odds. I think both are flawed.

Regarding Mebane's piece "The Wrong Man is President": you have NOT read it correctly (you have even mis-spelt his name). He does not say Optical scanners were "accurate" in 2000. Look at the table on page 29. It was a detailed and meticulous piece of work investigating just one mechanism by which votes for Gore were lost in 2000. He concluded that precinct-tabulated optical scanned ballots recorded fewer over-votes than any other type of ballot, and that if all the ballots in Florida had been cast and counted this way, Gore would have won Florida. There may well have been other problems with optical scanners in Florida in 2000 (though you do not make it clear about what incidents refer to which election) but that does not invalidate Mebane's finding, which is very important. Whatever is wrong with electronic voting, one heck of a lot was wrong with levers and punchcards in 2000 in Florida, and probably still is where they are still used.

[Dynasty_At_Passes]

213. Regarding the Mebane piece.

"He concluded that precinct-tabulated optical scanned ballots produced fewer over-votes than any other type of ballot,"

But the problem is he stated this would have ENSURED GORE WON THE ELECTION BY 30,000 VOTES when the very opposite is true.

This is the red flag I'm talking about.

"A "faulty memory card" was cited as the cause by the manufacturer. Experts and Diebold's own technical staff dismissed this as implausible for substansively the same reasons cited by the author:

"A memory card is like a floppy disk. If you have worked with computers for any length of time you will know that a disk can go bad. When it does, which of the following is most likely? In an Excel spreadsheet that you saved on a 'bad disk,' might it read a column of numbers correct the first time: '1005, 2109, 3000, 450...' but the second time, replace the numbers like this: '1005, 2109, -16022, 450...' Or is it more likely that the 'bad disk' will ... fail to read the file at all, crash your computer, give you an error message, or make weird humming and whirring noises."'

Diebold internal emails:

Ken Clark (Diebold ES R&D Manager) – January 18, 2001 1:41 PM
"My understanding is that the card was not corrupt after (or before) upload.
They fixed the problem by clearing the precinct and re-uploading the same card.
So neither of these explainations washes. That's not to say I have any idea what
actually happened, its just not either of those ... The problem is its going to be
very hard to collect enough data to really know what happened. The card isn't
corrupt so we can't post-mortem it (its not mort)."



John McLaurin - Diebold ES - 18 Jan 2001 15:44:50
"...the negative numbers ... occurred when Lana attempted to reupload a card or
duplicate card. Sophia and Tab may be able to shed some light here, keeping in
mind that the boogie man may me reading our mail. Do we know how this could occur?"



Tab Iredale - Diebold ES - 18 Jan 2001 13:31
The problem precinct had two memory cards uploaded ... on the same port approx.
1 hour apart. As far as I know there should only have been one memory card uploaded.
I asked you to check this out when the problem first occurred but have not heard back
as to whether this is true. Given that we transfer data in ascii form not binary and
given the way the data was 'invalid' the error could not have occurred during transmission.
Therefore the error could only occur in one of four ways:
...
<4> There is always the possiblity that the 'second memory card' or 'second upload'
came from an un-authorised source.



John McLaurin – Diebold ES - Thu, 18 Jan 2001 16:56:06
I will be visiting with Lana on Monday and will ascertain the particulars related
to the second memory card. One concern I’ve had all along is 'if' we are getting
the full story from Lana.
("Lana" is Lana Hires, the Volusia election employee described by Blackboxvoting as "particularly unhappy about seeing the Black Box Voting investigators in the office" in Nov 2004, described by BBV as having initially rejected their request to visit the warehouse containing election voting data, and "ordered them out")

BBV summary of Diebold memo's:

"What we know from the memos can be summarised as follows...

Two memory cards were uploaded from Volusia Couny's precinct 216, the second one was loaded sometime close to 2am in the morning. It automatically replaced the first card's results and reduced Gore's total by 16,022 votes and added several thousand votes to Bush plus a variety of minor candidates;
Both memory cards loaded into the system clean and without errors, indicating (contrary to the official line) that they were not faulty;
After the error was noticed the original card was reloaded and the mistake was rectified;
The error was introduced in such a way that the total number of votes remained unchanged (again something that could not happen by chance);
According to the technical boffins, the chance of the memory card being corrupted and still passing the checksum error test are less than 60,000 to 1;
The technical managers at Diebold Election Systems considered it a reasonable possibility that the second card was part of deliberate conspiracy to rig the election results." "

http://www.answers.com/topic/2004-u-s-presidential-election-controversy-voting-machines

Now, Mebane says that the optical-scan machine would have given Gore the election based on its ability to filter out overvotes. This is false, because it was this very optical-scan machine which virtually costed Gore the election. 16,000 votes in the deciding chapter, were lost by a "blamed memory card" Theresa LaPore says referring to Diebold. However the memory card was switched, and was proven to not be faulty. So someone basically deleted 16,000 votes at least, which erases Gore's margin, and it was the optical scanner who was the culprit in 2000.

This is a very serious distinction which harms Mebane's credibility immediately, because he gusses on the probability of Gore's win without citing the facts. If the optical scanners were really better and a lesser liability than the other system, it would not have costed these politicians the race. The fact is it is just as insecure, because the memory cards can be preprogrammed to alter and delete votes. That is what Diebold did here and Mebane does not even scan over it, hence why his argument about the Al Gore race falls apart.

[Febble]

222. Mebane did a study of the over-vote problem

in Florida, in 2000, and it was a problem. Relative to the system that had the least over-vote problem (precinct-tabulated optical scanners), the other systems (county-tabulated optical scanners, levers, punchcards) lost Gore 30,0000 votes. The butterfly ballot cost another few thousand didn't it? And your Volusia incident would also have done if it hadn't been rectified.

But Mebane wasn't addressing these last two. He was saying that levers, punchcards and county-tabulated optical scanners alone cost Gore 30,000 votes in over-votes.

Don't knock Mebane for addressing a different problem to the one you want him to address. Of course it doesn't harm his credibility. Over-votes are a serious problem. He tells us what it cost Gore. He's not saying "therefore you should use insecure precinct-tabulated optical scanners". I expect he'd recommend secure precinct-tabulated optical scanners.

OK, I'm done. Trash Mebane if you feel you need to. I fail to understand this compulsion to trash good people on the same side. But there it is.

Mebane concluded Gore would have been president if equipment less good at preventing over-votes than precinct-tabululated optical scanners had not been used. If precinct-tabulated optical scanners are also less secure than the other systems used in 2000 (probably) then that needs fixing too.

[Dynasty_At_Passes]

223. I understand what he was trying to say.

But it just came off as completely hack-jobbed in the context he was saying it.

If he had said that had overvotes not happened for Al Gore and a secure based precinct counter had been used than Al Gore would have won, I would have understood.

But he specifically stated optical scan machines being used in three of those counties would have eliminated Gore's problem, and given him a victory.

And he did so without realizing that those optical scan machines had their memory card switched, and ended up losing votes of at least 16,000 that threw the election. Did those problems ever get rectified? No, not in the recount. That's why the court decision came down the way it did. No one wanted to count the real votes. But that is not the issue here.

The issue is a serious oversight relating to security, and I guess I've addressed that. I just wish that if someone is going to do a study on one vague area of e-voting, they explain specifically what the study is out to address rather than lump it together and call it a comprehensive study.

I want to see scientists take responsibility for that, and if a comprehensive study is going to be done get a team of at least 100 people and staff and then explain all the basis of the study, the logistics, and compare machines by vendor. Its that kind of study which will really be able to tell and indicate what level of fraud and disenfranchisement is happening.

And it takes place with citizens such as in San Diego, who decided they should run another election alongside the "machine" election. This is what uncovered the discrepancies, not a statistical guess.

[Febble]

229. You say:

"I just wish that if someone is going to do a study on one vague area of e-voting, they explain specifically what the study is out to address rather than lump it together and call it a comprehensive study."

Who called it a comprehensive study? And what is vague about the title:

"The Wrong Man is President! Overvotes in the 2000 Presidential Election
in Florida".

Headline grabber (nice!) followed by a very specific indication as to what the paper is about.

And if you think most science ventures can "get team of at least 100 people" to address a question, you are certainly an optimist. This was a good piece of work. Don't knock if for not answering the question you wanted it to answer. It tells you the question it addressed, and tells you the answer. Nothing vague about it at all.

The reason for the precinct-tabulated optical scanner comparison was presumably to give a practical baseline. So the conclusion was not "if there had been no overvotes, Gore would have won" but "if the overvotes had been no worse than in precinct-tabulated optical scanners Gore would have won". In other words, even within the realms of available technology, over-votes could have been reduced to a level at which Gore would have won.

Nowhere in the paper does Mebane argue for insecure systems. That is a separate issue, addressed by others. And as far as I know, no-one has ever claimed that this study was a comprehensive study of electoral problems in Florida in 2000. Try this paper for references to more studies of Florida 2000 (including one by Mebane and others on the butterfly ballot).

http://www.capc.umd.edu/rpts/MD_EVote_HerrnsonNiemi.pdf

[Dynasty_At_Passes]

241. Ok, and I understand that. But the way the title came across....

Was very vague, saying this was an "overvote study related to the election" (2000)

When Mebane was writing that piece I thought he was going to discuss what the affects were in Ohio, or what was actually happening in Ohio. I thought the whole point for him being on the team was to address that issue, not study the differences between overvotes for a totally different election.

I guess what I'm trying to say, is it was simply worded poorly to be able to address what the report was about. If he wanted to do a seperate study about overvotes, I would have preferred it to be a cliffnotes version or something like that. It didn't make any sense to me to put out misleading information about Al Gore in the middle of the Ohio election investigation, it just didn't really make sense.


And I think it is possible for 100 people to do a comprehensive study, but they must first have the will and the money to do that study, as at this point it is completely neccesary.

We need to know exactly how many votes are getting lost, stolen, or likewise and discover a permanent solution to this problem. This is only getting done by citizens standing up and rolling out parallel elections like in california, its not getting done by throwing numbers back and forth using statistical guesswork I'm sorry to report.

[Febble]

244. Please look at the date on the piece I linked to

and you claim to have read. It was a study of over-votes in FLORIDA in 2000 . It was published on the 6th April 2004.

Mebane could scarcely be expected to address the issues of the NOVEMBER 2004 election in OHIO, when the 2004 election had not yet occurred

People do tend to be a little out of date when writing about things that have not yet happened. Give the guy a break.

Critique his analyses in the DNC report on Ohio 2004 by all means, but don't expect him to cover Ohio in a different paper written seven months before the event, about a different election in a different state.

[Dynasty_At_Passes]

247. Well where I read about it was in the DNC VRI report.

He didn't go into lengthy detail about the election 2000 but he compared touch-screen and other voting systems to opti-scan, and this seemed to be the general conclusion in the machine index.

Opti-scan machines were far better off because of their ability to filter out overvotes is what jumped out at me. It was un-neccesary I feel in this research report, and he would have been better off to simply have a cliffnotes section and link to his "earlier" report where this whole thing gets discussed.

To me it detracted from the overall paper. That just didn't seem to correlate with the problem.

[TruthIsAll]

204. Florida 2004 - Touchscreens and Optiscans

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x358495

[Febble]

208. So you sided with me, rather than Hout et al?

I found that OpScans benefited Bush. But it WAS apples and oranges. There were no touch screens in the smallest (pop. wise) rural counties, and no opscans in the largest urban counties. This was what undermines Hout's study, IMO - they found that it was touchscreens that benefited Bush. I tried to control for that by considering mid size counties only, where both machines were equally likely to be chosen. But there were still big demographic variables.

The thing is, TIA, that it is one thing to say that the probability of something happening by chance is gazillion to one (though it wasn't that significant once you leave off the rural counties). It is quite another to say that the factor you are considering is the critical one. It's what collinearity is all about. If machine choice was related to some other factor, you can't tell whether it was the other factor or machine that was the critical one. Sometimes you can control for other factors in some way, but in this case the collinearity was high - presumably because touch screens had been introduced to replace the old levers and punchcards that had disenfranchised so many African American voters in 2000.

So that's really where we tend to differ, you and I - you like to keep things simple - and it's often a virtue. But there's a reason for the development of multivariate statistical methods - most things in real life have multivariate causes. It makes analysis complex. And appealing to complexity can look like nitpicking. But I would argue that it isn't - it's trying to get things right. Mebane and Sekhon did a study of Florida where they modelled many demographic variables, and didn't find a machine effect. Which doesn't mean there wasn't fraud - it just means if there was fraud it wasn't particularly associated with one type of technology rather than the other.

[OnTheOtherHand]

145. let me make another point

Most of the substance of this article is about the voting machines, yes?

I'm not in a position to say how sound all the analysis is -- there are several people here who know the issues inside out -- but I think we all agree that the electronic voting machines stink.

In pluralistic polities, concentrated interests tend to beat diffuse interests. The voting machine manufacturers know what they want, and the public doesn't -- or, if it does, isn't organized to act upon it.

Some people think the manufacturers intend to throw every election to the Republicans; some people think they just want to sell lots of machines. Some people think the machines were used to steal the 2004 election, others don't, and others are unsure. And many people don't think about the machines because they are busy doing something else.

[autorank]

158. 80% of the votes will be tabulated by Reich Wing Republican Firms

Dominionists etc. Are you comfortable with that.

It's all very clear if you just look and pay attention. The game is rigged. You'd do better to bet on a three card Monty game.

Read about the absence of certifications for security for example.

The public doesn't have to get organized' to have its constitutional right to a free and fair election where all votes are counted and tabulated fairly.

Wake up, it's already too late for many efforts that should have been recognized and made, i.e., global warming (how about that ice formation the size of France that broke off of the Polar cap; or 5 f'ing years without stem cell research and people in pain and dying as a result! This is why we care and take the evidence seriously. People are dying because of crappy, fixed elections. Enough with the quibbling. It's tedious and very costly.

Are you working on Melissa G's project?

[Melissa G]

164. Okay Febble, Thanks for your articulate response..

Is this possibly a case of blind folks describing an elephant?

Here is my thinking. Your group is looking at the tail and we are looking at the trunk or vice versa if you will, but I am slightly colored by my perspective (grin) We will all in this group admit the enormous elephant of an unfair unverifiable election fraught with problems... The framing of the question then diverges.

In my group here in this forum and other places as well there was the question of were the Blivet votes came from when Dems did an unmatched effort of new voter registrations and motivated base turnout. This was unequaled on the repub side. We frame our analysis from this question. Your group doubts the validity of our approach.

Your approach, and I am being purposefully very vague here, centers around technicalities about polling noise and polling process. There are other threads where i have gone into detail on why i have my doubts about this approach but i am being general here and am just responding to concept. (No personal disrespect to you or OTOH but note my reference to the anatomical reference of what part of the elephant I ascribe this question to. grinning here)

Since it does not really answer the part of the elephant I am looking at, I think your question is more useless than you do. This in no way means I am right merely that I am more interested in my question than yours, which I generally understand as questions that speak to the technicalities of the polling process. Your group's work might give us valid information about process but it does not IMHO really do much to explain what is interesting to me about the enormous unexplainable vote shift. I find the heavily re weighted republican pre election poll defense unmoving and very weak as a supportive argument of the alleged shrub victory which is a rationale which has been posited as reason for the blivet win elsewhere..
Neither question is complete. They are tiny parts of a big elephant.

Bill bored has another leg of the elephant, Tabulators and election management systems. He thinks we election poll watchers are noisy too...
Best,
Melissa
edit clarity

[Melissa G]

174. Hi Febble, I kinda doubt we will have much agreement here but I'll

respond since you were so complete and polite enough to post.

I am clear that noise reduction is important to you. I hear that OTOH and Rick Brady are your friends and that they find it interesting to pursue this as well. I am always happy to encourage folks to do what interests them. My problem here is that I don't think your efforts at noise reduction can address the screaming discrepancy of the enormously loud Shrub vote shift. It may well reduce it somewhat but it does not silence it's screaming racket.
I do not necessarily agree with you about the actual statement of premise of our argument. you write..
'If we (generic we) start from the premise that Bush cannot have won the election, ANY argument to the contrary will fail.'

While we now pretty standardly posit fraud around here, we actually started looking precinct by precinct in many, many spreadsheets about where the hard to imagine B### vote discrepancies were coming from first. Archival searches should bring this up.. I actually believe it is might be possible to show Blivet possibly won the election I just have not seen anyone come up with what i think is a reasonable explanation. We obviously disagree here. Though we agree that the election is unverifiable and that is a HUGE problem regardless.
On the evidence presented, unless i missed something I will continue to scoff at the recall argument. I see no evidence other than assertion for it and as i remember the studies i saw were not apples to apples... telephone polls much after the fact do not in my mind equate to exit polls immediately after voting..we will likely just disagree here also.
you write..
'We raise the issue that the pattern of bias in the poll does not support the hypothesis of highest fraud in Bush strongholds - or widespread vote-switching - and are accused of - I dunno, statistical incompetence, using bad data, you name it.'

I don't think i accused you of these things. Here I thought you were a clever academic on the Bush stronghold point. I think you got an academic two pointer by exploiting that weakness in the paper. My shots at you on this issue were about it's irrelevancy to the discussion. Your bringing this up was an academic win for you but in my view a red herring in the general discussion which I frame differently than you do.
Again, in my view, what is important and interesting to you was misleading to the case, which is more or less what i hear you saying about my/our position. Standing in each other's shoes i can see how this makes sense to us individually.
Best,
Melissa

[Febble]

180. Well, we probably won't, Melissa

but I would nonetheless argue that my concerns are not irrelevant, are not red herrings (except in the sense that I think the exit polls are a probably red herring) nor of merely academic "interest". My own academic interests are quite other.

And if your estimate of the magnitude of the alleged vote-shift is based on the exit poll evidence, then I think it is shaky. The strongest evidence that the exit poll discrepancy was not due to fraud is given in the ESI study of Ohio - at least for Ohio. Discrepancies were not greater where Bush's increase in vote share was greater.

It is of course perfectly possible that Ohio was won by voter suppression rather than vote fraud. Or that vote fraud swung it, but escaped the exit polls (also perfectly possible). But that undermines the exit poll story in itself, as Ohio has been the poster child for the exit poll evidence. In other words, if the exit polls were not more off in Ohio where Bush's increase was greatest, it looks as though some other explanation is required for the exit poll discrepancy in Ohio. NH is also a stumbling block (though not an insuperable one, as Land Shark has pointed out - the discrepancy could have arisen through vote destruction - but that is not the mechanism generally alleged for the exit-poll red shift).

You may consider these nit-picking details. But the devil is often in the details. An apparently simple case can fall apart when looked at closely. And I don't want this case to fall apart. It was a corrupt election.

[Melissa G]

185. Yes Febble, it would be a boring world if we all agreed!

I am not at all attacking you and saying you are irrelevant or other than interesting and i am deeply grateful for your kind posts. Although I disagree with you, I have really enjoyed the exchange and your view of the elephant of election fraud. It was fun and educational to look at the issue through your eyes and see how the work I have been involved in is perceived.
We are both bright people but seem to lack compelling evidence to convince the other. In my mind, that gives weight to Bill Bored's perspective of the elephant, (grin) though I am still a great fan of the work of TIA and Freeman.
Thanks for chatting. I would never have learned all the lovely things you have taught me by considering your perspective. I'm still unconvinced of your position (as are you of mine) but I feel clearer about why and I have new insights about how and where to look at the fraud issue. Triangulation is a lovely tool and disagreement is a fine engine for motivation to explore issues one might not have looked at without the difference of opinion.
Best Regards,
Melissa

edit clarity

[anaxarchos]

122. Finally, after all this sound and fury, we get to it....

Good.

You were wrong when you said: "I don't need to change any of the facts you cite. I only have to attempt, over and over, to explain why they don't prove your case (even in a very weak sense of "prove")."

You also have to present an alternative thesis which explains the data as well or better. You just did that (at least the "present" part). In fairness, there weren't many options to chose from. As you pointed out, there are only those who voted for Bush, voted for Kerry, voted for someone else or those who didn't vote in 2000. It's kinda like a water balloon, isn't it? You push in here and it pops out there. So, if you reject fraud and you reject Mitofsky's mix of 2000 Gore and Kerry voters (I won't quibble with your "notoriously unreliable", etc. - I might quibble with Febble, though, because that almost rhymes) and instead you believe that "Given the extreme partisan polarization of views about Bush, I imagine that turnout among Bush2K and Gore2K voters was fairly similar", then all you are left with is a red-shift larger than that implied by the NEP survey questions in one or more of the groups listed above (I think Mitofsky has used the "all groups" option already in his final weightings).

You choose Bush2K voters. OK.

Now all we need is some empirical evidence.... Make the case.

[anaxarchos]

130. Ooops... I meant

you chose Gore 2000 voters, of course.

So since I can't edit the above, let me restate the question:

If you stipulate that Gore2K and Bush2K voters were roughly equal in turnout and that those that voted for others in 2000 and those that didn't vote broke roughly in the way indicated by the NEP... then:

Exactly how much red shift among Gore 2000 voters was needed (using any assumptions you like) in order to give Bush his plurality?

AND

Where is the empirical evidence (pre-election/post-election polls, etc.) that supports such a previously undetected red-shift among those who voted for Gore in 2000?

Simple question....

[OnTheOtherHand]

146. I will try to doodle up some numbers

of how the splits may have gone.

I don't know whether any pre-election polls (from 2004) ask about 2000 vote as well as vote intention. I believe there is a 2000-2002-2004 NES panel study which may shed some light.

But let's not get too subtle. The pre-election polls typically showed a very close race with, on average, Bush slightly ahead (see http://pollingreport.com/2004.htm). If we assume, as we seem to with decent warrant, that new voters (i.e., non-voters in 2000) split for Kerry, then as with your water balloon, Bush's lead or even competitiveness had to come from somewhere else. Wouldn't the obvious back-of-the-envelope inference be that more Gore 2000 than Bush 2000 voters were either defecting or opting out?

It is of course possible to subject the pre-election polls to a style of meta-analysis that indicates that properly understood, they clearly showed that Kerry had the advantage. I was open to those arguments before the election, and I still am. But I daresay that they are speculative, not definitive.

I suspect that for forensic analysis of the 2004 election, attention is best focused on particular states -- but if you have a killer argument that generates new avenues for testing the fraud hypothesis, please do share. (It could even be a half-baked argument in progress. We are all friends here, on good days at least.)

[anaxarchos]

150. I think that it is worthwhile to doodle up some numbers...

I've been doodling them up since the day after the election. Sometimes, I remind myself of the scene in the old movie "War Games", where the computer is trying to find a winning scenario to nuclear war. "Let me try the New Voters break evenly/Limited War option...", etc.

Let me put the math problem out front (in the spirit of friendliness). Of the four groups (Bush2K, Gore2K, Other, New), only Bush2K matters (because they break 90+% to Bush). You could have New break to Bush but to make up for Other and the 3M Bush plurality, you need to break significantly to Bush. That leaves "trace", to use your forensic analogy. 'Snot "plausible". You could leave Gore voters at home (snot plausible either for the reasons you stated). You could make Gore2K "red-shift" but it is a BIG shift and you have the trace problem again. You could do all of the above but then you end up defending many "slightly less plausible scenarios" at once... "I tell ya it was the perfect storm, dammit!"

The easiest way is to have more Bush2k voters. The problem is 122 million votes. You need more Bush2k voters than actually exist. A minor inconsistency. (BTW, I'm not saying Mitofsky did this... I think the exit polls "don't matter" except as a detector of "trace").

The irony is that it is NOT a killer argument. It is too obvious, much like a bar bet to win a beer.

I take all of your posts VERY seriously, OTOH. But I think TIA is 100% right this time. This IS "the clincher". Game, set, and match. But hey, its a friendly game as you say and he beat me too (he has this "sleeper" serve which kinda creeps up on you and then he "trash-talks" you while you are trying to figure it out). And... let's go through the scenarios anyway.

MUCH more importantly than all this, I think the fuckers stole the election. I'm still not exactly sure how. I'm not sure what constitutes "proof" or to whom one offers such "proof". I dunno what to say to your friend. But I am certain of this.

[OnTheOtherHand]

154. very interesting

I take your posts very seriously, too (especially the funny stuff -- I've found myself humming, "quibble, Febble, fibble, quebble, let's call the whole thing off!").

From first principles, I don't deeply see what your problem could be. It always drove me nuts that Bush was leading in some polls even though I was pretty sure he was gonna lose among new/returning voters, but it never seemed to me that this could be mathematically impossible.

But I do know that sometimes the more one doodles, the more doo-doo one discovers. So I accept your friendly warning about the sleeper serve.

Right This Moment I am doodling with Febble on something quite different. Do you have some doodles that I could look at? You could either post them here or e-mail them -- I will PM you with an address in case you don't have it.

Or in the alternative, I think Febble and I have also done some preliminary doodling on _this_ question, so maybe we can compare scrawls.

[TruthIsAll]

160. Googling? How 'bout crunching? Bush needed 16.2%-17.7% of Gore voters!

to get his 3 million margin, depending on the exit poll time
line.. 

We assume equal 100% turnout of both Gore and Bush voters.

USING THE FINAL 1:25PM EXIT POLL (13660 RESPONDENTS):
ONE OUT OF 6 (16.2%) GORE VOTERS HAD TO SWITCH TO BUSH,GIVEN
THAT ONE OUT OF 11 (9.0%) BUSH VOTERS SWITCHED TO KERRY.
THIS ASSUMES KERRY WON NEW VOTERS BY 54/45%.

AT THE 12:22AM TIMELINE (13047 RESPONDENTS): 
KERRY WON NEW VOTERS BY 57/41%. BUSH NEEDED 17.7% OF GORE
VOTERS TO GET HIS 3 MILLION VOTE "MANDATE".

Note: I adjusted Nader/Other percentages slightly (21/71/8) to
10/64/26) to back into the actual final numbers as close as
possible. In fact, it is more accurate, since the third party
vote totaled near 1%.


Final
1:25pm 13660 respondents
Voted	Pct	Bush	Kerry	Other	Votes
No	17.31%	45.0%	54.0%	1%	21.16
Gore	40.25%	16.2%	83.8%	0%	49.21
Bush	39.82%	91.0%	9.0%	0%	48.68
Nader	2.62%	10.0%	64.0%	26%	3.20
	100%	50.81%	48.34%	0.85%	122.26
	122.26	62.12	59.10	1.04	

***************************************************

Pristine:
12:22AM 13047 respondents
Voted	Mix	Bush	Kerry	Other	Votes
No	17.31%	41.0%	57.0%	2%	21.16
Gore	40.25%	17.7%	82.3%	0%	49.21
Bush	39.82%	91.0%	9.0%	0%	48.68
Nader	2.62%	10.0%	64.0%	26%	3.20
	100%	50.72%	48.25%	1.03%	122.26
	122.26	62.01	58.99	1.26	

[anaxarchos]

161. Assume 90 or 95% (you can't really do100%) and it gets worse....

What are the chances of one out of 5 or 6 Kerry voters going to Bush with no indication of that in the polls before or since? How often has this happened in U.S. history? How often has it happened at the same time that New or Other voters broke towards the challenger?

Nope...

Had to be more Bush voters. So what if they weren't alive? Ever read the old Russian novel Dead Souls (Gogul)? If it's more serfs we need, it's more serfs we get. Nobody said nothin' about LIVING...

[TruthIsAll]

171. Here is additional analysis which illustrates the point.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x367600

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=367160

[OnTheOtherHand]

183. I think that with 95% turnout, 16% Gore 2000->Bush 04 is close

I want to dood-- I mean, DO SOME SENSITIVITY ANALYSIS (grin) to see how things bounce around. (I assume that you meant "one out of 5 or 6 Gore voters," yes?)

I think that my own sleeper serve may have gotten by you. The national pre-election polls generally showed a close race with, on average, Bush slightly ahead. It was widely anticipated that Kerry would win among new voters (although we have to be careful here -- I think the main expectation was that he would win among first-time voters). And various surveys exemplifying false vote recall have been out there forever. (I think last time we went down this road, I used the online GSS interface to run crosstabs going back 30 years -- poor Mike Dukakis, by 1993, lost retrospectively by 41 points, but that was an extreme case. Some might think that Clinton would be struggling retrospectively against Bush in 1994, but nope, he was up 12 points, having won by less than 6. Other curious minds can roll their own at http://webapp.icpsr.umich.edu/GSS/ .)

So I think we might have reasoned all along that Bush at least had a good shot at taking more Gore voters than Kerry took Bush'00 voters -- we could have made essentially the same argument, contemporaneously, from the pre-election polls that TIA is making from the exit poll weightings, except that he takes it as a reductio ad absurdum. And we should have anticipated that the retrospective crosstabs would confuse the hell out of us about the issue. We even might have expected that the exit polls would have a Democratic tilt, as they have in past years, although we wouldn't have guessed how large. I know some think it is contrived to say both that "exit polls favor Democrats" and "surveys retrospectively favor incumbents," and to apply both those generalizations (only generalizations, not laws) to the 2004 exit poll, but I don't see how one can not.

Now, I honestly don't know the answer to your questions. I don't know whether anyone ever bothered to work through the implications of false vote recall for retrospective crosstabs. But I imagine that this has happened rather often in U.S. history. Just as, rather often in U.S. and British history, the unweighted exit polls have diverged significantly from the official results -- something most of us hadn't noticed, so we didn't spontaneously laugh when some folks told us that they were uncannily accurate prior to 2004.

By no means do I think that this analysis rules out massive fraud. The 43/37 retrospective Bush2K/Gore2K split in the weighted 2004 exit poll seems plausible to me, and I don't think anyone could convince me that clearly it ought to be 40/40, but conceivably I might be convinced that it ought to be somewhere between. (I know that we are discussing Gore2K and Bush2K holds/steals, but again, my point is that false recall is gonna skew those figures.) I certainly don't see how anything in this line of analysis can be the "clincher."

[anaxarchos]

193. You are WAY over-thinking this, PoliSciGuy,...

... and drifting toward comfort, I think. Let's stick to the de facto truce we just established above and see what we get. But, in order to do that, we also have to stick to some basic rules... which we just make up, of course.

Proposal:

1) The game is a game of "Best Fit" but the boundaries are subjective. Someone (maybe Descartes?) once said that "mathematics has a soul but it has no heart" (i.e. "conscience"), meaning any value is acceptable in calculation but not necessarily in life. We use a standard of "plausible", stretch our self-discipline to the max, and see where we end up.

2) The game gets played right here and anyone can play. Despite the "crazy" tone we sometimes adopt, there are close to 65000 experts here on what is "plausible" in politics. A few are truly nuts so we identify them early and make them the board of directors.

3) We keep it really simple, just as you said.

4) No changing assumptions in mid-stream without calling a time-out and saying... "This assumption can no longer be sustained because...". I am speaking specifically about: "Given the extreme partisan polarization of views about Bush, I imagine that turnout among Bush2K and Gore2K voters was fairly similar"(YOU) and "The 43/37 retrospective Bush2K/Gore2K split in the weighted 2004 exit poll seems plausible to me, and I don't think anyone could convince me that clearly it ought to be 40/40, but conceivably I might be convinced that it ought to be somewhere between." (also YOU). No shaving points allowed. This will test whether we can even get started. I'll start. I expect no more than a one to two point differential between returning Gore and Bush voters (I could argue it the other way). Some demographics favor the Republicans (re-registration) but others favor the Dems (Age). ...Let's see if we can even start this game.

5) No controversial applications of controversial interpretations of controversial theories (as in: "I don't know whether anyone ever bothered to work through the implications of false vote recall for retrospective cross tabs"), until simple sources of trace data are exhausted. We are dealing in gross magnitudes here and we have a thousand sources of info. For example, we are starting with the question of whether there was an erosion of the base for Democratic voters in 2004. On this, we have multiple survey questions in the 2004 exit poll (and from previous years) alone, on subjects such as "Were you contacted by the XXX campaign", where NO implication of response error applies.

6) You and I have a standing side agreement that I won't quibble with you about "post-election bandwagon effect"... I'll quibble with Febble (or was it queble with Febble?). If you really, REALLY need it, we will cut a new side deal.

7) I further agree not to comment on your serve until we get to the "double-dare-you" stage.

8) This game is only indirectly about Exit Polls and Polling in general. We are actually trying to reconstruct the election of GW Bush, in a "plausible" way.

9) This game is deadly serious for me, but turning it into a game avoids what is "conceivable" for some, what is "proof", what is "acceptable", and all the rest... the only standards for this game are "reason" and "virtue".

10) First one who says "the real vote count has to be right because it is the real vote count", loses.

Wadda you say?

[OnTheOtherHand]

198. slight puzzlement

You invite me to this task by asserting that I am way over-thinking, and then propose ten rules of which one is to "keep it really simple"? Hmm. Probably I am in one of my hyper-literal moods.

I'm not quite sure where the "simple" starting place is here, but I can try to "play" it your way.

On point 4, I wonder whether you are seeing more wobble in my statements than I do. I would be comfortable assuming no difference in turnout between returning Bush2K and Gore2K voters, unless otherwise convinced. That being the case, I expect the retrospective question to yield more (reported) Bush2K voters, and a six-point gap seems fine. (I would have to download all the other exit polls to refine that, which I will not do any time soon.)

On point 5, I can't concede that rule if I don't know what it means. If you mean that I shouldn't spin the crazy idea that Gore2K voters would become Bush2K voters, I can't accept, because in the NES 2000-2004 panel study, some demonstrably did -- confirming my expectation from other panel and cross-section studies.

On point 6, ditto. TIA asked the reasonable question (only I think he thought it was rhetorical), whazzup with that 43/37 split? I think false recall is the obvious answer.

I think your point 9 is well put.

I'm fine with point 10. I thought the debate was whether the official vote count plausibly _could_ be right, not whether it _had to_ be right. And obviously people's criteria of plausibility vary, as per your point 9.

Nu?

[anaxarchos]

199. I've always thought that negotiating the shape of the table was legit...

As far as 10 points go, they are entirely arbitrary. There are only a few points in my post that really matter. The rest are to "keep it light", etc.

On wobble, if that is all you imply, I stand corrected.

On playing by "my rules", I'm making it up as I go along... same as you. I'm cool about any changes you may propose. Same goes for how we start... although we probably start with basic assumptions (i.e. 122 million votes, how many returning, how many new, etc.).

On 5 and 6, I'm also cool with setting aside the retrospective question entirely for now (Shouldn't be that tough if the object is the election and not the "validity of the exit poll", etc.). What concerns me are the obvious circular arguments: We can't trust the retrospective question because of response error ("notoriously unreliable") but that response error suddenly becomes "notoriously reliable" in order to prove a red-shift that is indicated by no other data... We'll use "reason" and "virtue" and self-restraint.

'Sok?

("I think we may already have started", he thinks to himself as he hits the ball in a long arching lob back over the net...)

[Febble]

201. Looking forward to the match

But why don't you start a new thread? It's hard to follow this one, now it's got all squashed against one wall (more like squash in fact). How about a proper court and a new title? I'll quibble as and when.

[OnTheOtherHand]

207. I'm happy to negotiate

and restraint is good. I don't think it's dirty to say that some sort of retrospective reporting error could be expected, but the ground should swallow me if I even hint that it _proves_ (or offers strong independent support for) a red shift in the absence of other data.

Maybe I could make one more meta point about this "game," one I have made many times: I'm just as happy to "lose" as to "win," as long as we both feel at the end that the rules we made/played by together were fair. And with any luck, we might find some things to check out that hadn't occurred to us yet.

I agree with Febble that another thread makes sense. Either you can start, or I will keep working on my scenario of "what I think might've happened" and then figure out how to write it up.

[Melissa G]

212. Now THIS is what i bought popcorn for!!!

:popcorn: save me a ringside seat! Dueling Quants! I love it!

[anaxarchos]

217. OK, let's go...

As I understand the agreement: of the 10 rules, #4,#5,#6 go away as being paranoid fixations of mine (guilty as charged). Instead they are subsumed by rules #1 and #9 (which is now #6). I suggest we also get rid of rule #7 in order to cross OTOH's simplicity threshold (well known that 6 is simple but 10 ain't).

If you guys want to start a thread, it works for me. I think it should include the posts going back to OTOH's "of course I believe that a plausible scenario was ignored" and all those since so people know what we are talking about.

If I understand the form; the objective is to let George Bush win the White House "plausibly". One person puts out a scenario and puts forward an initial defense. Everyone then viciously attacks the plausibility of that scenario (but reasonably, virtuously, and with decorum). The defenders defend. We see what we learn and move on to the next.

I suggest one test scenario first to nail down the game. Then, I suggest we do the fraud scenario together (to the extent we can agree) because we all seem to agree that it is a "fit" even if some don't believe it is the "most plausible fit". Then, we try alternate scenarios that will fit equal or better. OK?

I suggest you set the initial assumptions, OTOH, and once we have agreed, I will drive the first test scenario. That will be "The Brilliant Karl Rove's super Brilliant strategy for Brilliantly winning in 2004". Mr. Rove has graciously agreed to let me drive this scenario for him as he has a previous engagement at his lawyers.

One agreement with TIA - if dead people vote, it counts as fraud.

One disagreement with everyone - notice rule #2. EVERYONE PLAYS.

Does that work for y'all?

[OnTheOtherHand]

232. sorry, delayed by children and antelopes

Although I decided to leave the antelope to its own devices, the children required a firmer hand.

I'm not sure I understand the relationship between the "initial assumptions" and the "first test scenario." But I will try to rough out something based on TIA's calculations and seemingly pertinent meanderings, and post it separately. Actually, I think I can do that fairly efficiently right now.

[anaxarchos]

233. OK, ...

If you don't think we ought to try it out first, then let's go straight to your red shift case. We can refine initial assumptions in that context just as easily. We'll do fraud later.

You SHOULD start a new thread though. This monster has become impossible.

[OnTheOtherHand]

234. I'm about to post it

A quick(?) problem summary and the redacted rules. You can amend to taste. I've finished the summary, just need to copy-paste-revise the rules.

[TruthIsAll]

206. OTOH AND ANAX: BEFORE YOU START THE MATCH, READ THIS..

Before proceeding, let's establish some BASIC FACTS RIGHT NOW.
DO NOT start the match without stipulating to it.

The maximum number of Gore and Bush 2000 voters still alive in 2004 are FACTS.

USE THEM.
THESE ARE FACTS.
DO NOT IGNORE THE FACTS.

USE THESE FACTS AS A BASIS FOR THE DISCUSSION AND SENSITIVITY ANALYSIS OR DON'T EVEN BOTHER TO PLAY THE GAME.

By that I mean, we must start with the MAXIMUM weightings:
Bush 39.82% of 122.26 million 2004 voters
Gore 40.25% of 122.26 million 2004 voters

THE FINAL EXIT POLL'S 43% for Bush is ABSOLUTELY IMPOSSIBLE.
The 37% for Gore is POSSIBLE, though highly unlikely.

Start from there.
Work with the REAL numbers.
WORK WITH FACTS.
YOU MUST NOT IGNORE THEM.
YOU DO SO AT YOUR PERIL.

OTOH, don't try to split hairs.
UNBLOCK YOUR MIND.
ACCEPT YOUR SAVIOUR.
ACCEPT THE FACTS OF LIFE.
AND DEATH.

THIS WILL BE A FAIR MATCH IF:
YOU ACCEPT THE REALITY OF LIFE AND DEATH.

ONLY BUSH VOTERS STILL ALIVE COULD VOTE IN 2004.
ONLY GORE VOTERS STILL ALIVE COULD VOTE IN 2004.

THE MAXIMUM NUMBER OF BUSH 2000 VOTERS STILL ALIVE WAS 48.7
HE GOT 50.456 MILLION VOTES IN 2000.

THE MAXIMUM NUMBER OF GORE 2000 VOTERS STILL ALIVE WAS 49.2
HE GOT 50.999 MILLION VOTES IN 2000.

WE ASSUME AS GIVEN THE 0.87% ANNUAL U.S. DEATH RATE.

I HAVE ALREADY GIVEN YOU BOTH A HEAD START WITH ANALYSIS OF WHAT IT WOULD TAKE FOR BUSH TO WIN BY 3 MILLION VOTES.

YOU CAN ARGUE ABOUT THE COMBINATIONS AND PERMUTATIONS OF ASSUMED PARTISANSHIP SWITCHES.

BUT YOU HAD BETTER STIPULATE THIS BEFORE PROCEEDING:
THE APPROXIMATE 1.7MM BUSH 2000 VOTERS WHO DIED BEFORE ELECTION DAY 2004 COULD NOT HAVE VOTED.

I CANNOT BELIEVE WHAT I JUST SAID:
PEOPLE WHO DIE, STAY DEAD.
UNLESS YOU BELIEVE IN INTELLIGENT DESIGN.
AND THE IMMORTALITY OF BUSH VOTERS.

AGREED?

YOU MAY WANT TO BEGIN BY ASSUMING 100% TURNOUT.
FOR BOTH BUSH AND GORE VOTERS.
THAT'S WHAT I DID IN MY POST.

I HOPE YOU WILL REFER TO MY ANALYSIS.
BUT EVEN IF YOU DON'T, IT'S THERE.
IT'S NOT GOING ANYWHERE.

AND I JUST MAY CRUNCH SOME NEW NUMBERS.
JUST IN CASE.


OK?

WHO SERVES FIRST?

[Febble]

209. I'm sure OTOH won't invoke zombie voters

I've suggested a death rate of 8.25/1000 per annum, which I found on some website or other. That suit you?

[TruthIsAll]

211. No, it doesn't suit me. My site said 0.87% No nitpicking, please.

Not that it would make a difference, Febble.

As long as you stipulate to the fact that BOTH Gore and Bush voters who died are no longer walking among us.

So what's the point?
Do you think making the ball slightly smaller will help your side?

[autorank]

220. What happened? OTOH call the game on account of "humidity"?

[Febble]

221. of course not

It's just that I found that figure here

http://www.cia.gov/cia/publications/factbook/geos/us.html

and used it. I couldn't remember what you'd used. Thanks for providing the figure. As you say, it shouldn't make a difference.

PLEASE don't fly off the handle, TIA! I once knew a cat like you, much beloved, but boy did you have to be careful not to stroke his fur the wrong way.

I'm trying to make sure the playing field is level, not slant it in one direction. That's why I posted a figure for you to check. I'm sure everyone will be happy to use yours.

And as I said, no zombies sounds fair. I hate zombies too.

[Febble]

228. Apologies

Just noticed you gave a figure in that post. I DO wish you wouldn't use ALL CAPS! It makes your posts so hard to read. My trifocals aren't up to it.

0.87% is fine, I'm sure. Who knows where the CIA got their 0.83% figure from. I just used theirs as I had lost track of your previous posts.

[TruthIsAll]

226. Breakeven analysis: point of discussion

We can derive a mathematical relationship to calculate what it
would take for Bush and Kerry to tie. Given Kerry's percentage
of Bush 2000 voters (X1), we can determine the percentage of
Gore voters (X2)he needed to tie Bush. Or vice versa.

We hold the Nader and New voter percentages constant.

x1	x2				
K%G	K%B				
94.94%	1.00%				
93.95%	2.00%				
92.96%	3.00%				
91.97%	4.00%				
90.98%	5.00%				
89.99%	6.00%				
89.00%	7.00%				
88.01%	8.00%				
87.02%	9.00%				
86.03%	10.00%				
85.04%	11.00%				
84.05%	12.00%				


Assume Final National Exit Poll percentages, with maximum
weightings based on actual 2000 votes:

Assume Kerry gets:
1) 2% of Bush's 2000 vote, he needs 93.95% of the Gore vote.
					
2000	Pct	K	B	N	Votes
NV	17.31%	54.00%	45.00%	1.00%	21.16
G	40.24%	93.95%	6.05%	0.00%	49.2
B	39.83%	2.00%	98.00%	0.00%	48.7
N	2.62%	71.00%	21.00%	8.00%	3.2
Total	100%	49.81%	49.81%	0.38%	122.26



2) 6% of Bush's 2000 vote, he needs 89.99% of the Gore vote.

2000	Pct	K	B	N	Votes
NV	17.31%	54.00%	45.00%	1.00%	21.16
G	40.24%	89.99%	10.01%	0.00%	49.2
B	39.83%	6.00%	94.00%	0.00%	48.7
N	2.62%	71.00%	21.00%	8.00%	3.2
Total	100%	49.81%	49.81%	0.38%	122.26




3) 8% of Bush's 2000 vote, he needs 88.01% of the Gore vote.

2000	Pct	K	B	N	Votes
NV	17.31%	54.00%	45.00%	1.00%	21.16
G	40.24%	88.01%	11.99%	0.00%	49.2
B	39.83%	8.00%	92.00%	0.00%	48.7
N	2.62%	71.00%	21.00%	8.00%	3.2
Total	100%	49.81%	49.81%	0.38%	122.26


4) 10% of Bush's 2000 vote, he needs 86.03% of the Gore vote
to tie					
2000	Pct	K	B	N	Votes
NV	17.31%	54.00%	45.00%	1.00%	21.16
G	40.24%	86.03%	13.97%	0.00%	49.2
B	39.83%	10.00%	90.00%	0.00%	48.7
N	2.62%	71.00%	21.00%	8.00%	3.2
Total	100%	49.81%	49.81%	0.38%	122.26


5) 12% of Bush's 2000 vote, he needs 84.05% of the Gore vote
to tie					
2000	Pct	K	B	N	Votes
NV	17.31%	54.00%	45.00%	1.00%	21.16
G	40.24%	84.05%	15.95%	0.00%	49.2
B	39.83%	12.00%	88.00%	0.00%	48.7
N	2.62%	71.00%	21.00%	8.00%	3.2
Total	100%	49.81%	49.81%	0.38%	122.26


************************************************************

Derivation of the formula:

	MaxVotes	Pct2004	
G	49.2	40.24%	Gore
B	48.7	39.83%	Bush
N	3.2	2.62%	Nader
Total	101.1	82.69%	
			
NV	21.16	17.31%	New voters
Kerry			
KNV	11.4264	54%	New votes
KN	2.272	71%	Nader votes
KGV	X1*49.2	X1	Gore votes
KBV	X2*48.7	X2	Bush votes
Total	13.6984		+ X1*49.2+ X2*48.7
			
Bush					
BNV	9.522	45%	New votes		
BN	0.672	21%	Nader votes		
BGV	(1-X1)*49.2	1-X1	Gore votes		
BBV	(1-X2)*48.7	1-X2	Bush votes		
Total	10.194		+(1- X1)*49.2+ (1-X2) *48.7		
					
Total Kerry                         = Total Bush					
					
x1*g + x2*b +.54*nv +.71n = (1-x1)*g+ (1-x2)*b + .45*nv +
.21*n					
					
2x1*g  - g  +2x2*b -b +.09*nv + .50*n = 0					
					
g * (2x1-1) + b *(2x2-1) +.09*nv +.50*n =0					
					
49.2 * (2x1-1) + 48.7 *(2x2-1) +.09*21.16 +.50*3.2 =0					
					
98.4*x1 - 49.2+  97.4 * x2  - 48.7 + 1.9044 + 1.60 = 0					
					
98.4*x1 + 97.4*x2 = 49.2 +48.7 -1.9044 - 1.60					
					
98.4*x1 + 97.4*x2 = 94.3956					
					
x1= (94.3956 -97.4x2)/98.4					
					
x1=.9593 -.9898*x2					
					

[TruthIsAll]

227. Typo fix: Kerry% of Gore voters =X1; K% of Bush 2000 voters =X2

Typo fix: Kerry% of Gore voters =X1; K% of Bush 2000 voters =X2

[TruthIsAll]

248. Are we done yet?

eKv Early Kerry Voter.
lBv Late Bush Voter.
oFv Overpolled Female Voter.
dEp Democratic Exit Pollsters.
iEp Inexperienced Exit Pollsters.
yEp Young Exit Pollsters.
hMe High Margin of Error.
kCe Kerry Cluster Effect.
kDe Kerry Design Effect
bBe Bush Bandwagon effect.
mFt Massive Fundie Turnout.
rBr Reluctant Bush Responder.
eBr Exuberant Kerry Responder.
lGv Lying Gore Voter.

fGr Faulty Gore Recall.

What's next?

[anaxarchos]

152. Hiya MG....

How can you be a rap artist if you don't like to "battle"?

;-)

[Melissa G]

165. Rap artist? My middle intial is C but the resemblance ends there I think.

[anaxarchos]

166. Didn't mean you "YOU"...

... meant generic you (as in the battlers). Didn't mean an actual "rap artist" - meant the art of rapping (i.e. "talking")...

My jokes are getting as old as I am...

[Melissa G]

210. I get your jokes anax.. we are likely close to the same age....

That rap artist shot was just a hedge in case you were younger.:toast:

[Melissa G]

74. Hi Febble, In all the really plausible recounts done by the newspaper

consortium that recounted Florida, Shrub lost! This story was buried by the press after 9/11 but i am happy to send you links.
Melissa

editing to include what I am responding to..your partial post below
" Bush won Florida without the help of Jeb Bush? (TIA)
Probably, but he might have helped anyway (febble)

[Febble]

83. I am convinced Gore won Florida in 2000

as I said.

I assumed this question this referred to 2004.

[autorank]

86. And isn't that the start of the illegitimacy! What a fiasco Florida was.

It's bothered me ever since the ugly events during which thugs stole the election. At least Gore fought back. He missed the boat, unfortunately, by not challenging on race and the felon purge (would you trust your private data with ChoicePoint).

The most tragic scene in modern American politics was Al Gore, as Vice President, presiding over the certification of the vote, having to gavel down the challengers, and watching 100 US Senators sit there and do nothing (these are the same folks who helicoptered out of the DC area right after 911 while the rest of us waited for the other shoe to drop).

The theft of the presidency 2000 and Florida 2000 are the beginning points of the new Voting Rights movement.

Here's an article of interest, my favorite long running outrage, "spoiled ballots" ...

http://www.washingtonpost.com/ac2/wp-dyn?pagename=article&node=&contentId=A18220-2001Nov12

[Melissa G]

127. I think Bush lost Florida in 2004 as well. I'll try to dig up supportive

documentation which I do not have at my fingertips before I try to convince you though...

[Melissa G]

238. Here is a Zogby article on how he believes Kerry won Florida

http://marc.perkel.com/archives/000642.html
and there is of course the Freeman paper about this..

[TruthIsAll]

112. THE CALCULATION FOR THE 37% GORE/ 54% KERRY NEW VOTER SCENARIO - AND MORE.

He wins by 1.03 mm votes

....	Weight	Kerry	Bush	Other
New	0.2056	0.54	0.45	0.01
Gore	0.37	0.91	0.09	0
Bush	0.3982	0.09	0.91	0
Nader	0.0262	0.71	0.21	0.08

	1.00	50.22%	49.37%	0.42%
	122.26	61.39	60.36	0.51

Because Kerry, Bush and Nader weights are given, the remainder
(for the 122.26 total) must be New voters.

But New voters comprised only 17-18% of the 2004 electorate. 
Thus, Gore voters comprised more than 37%. 

And Bush did not get 39.82% (his maximum). 
Some Bush voters stayed home.

It was probably very close to: 
Kerry 39.5%, Bush 39.0%

....	WEIGHT	k	b	n
n	0.1888	0.54	0.45	0.01
g	0.395	0.91	0.09	0
b	0.39	0.09	0.91	0
n	0.0262	0.71	0.21	0.08
	1	51.51%	48.09%	0.40%
	122.26	62.98	58.80	0.49


BUT WHY SHOULD WE BELIEVE THE FINAL EXIT POLL PERCENTAGE?
KERRY won 57%, NOT 54% of NEW voters,
ACCORDING TO THE PRISTINE 13047

That means he won by 6 million votes (52%-47%)
It was a Kerry landslide.


....	WEIGHT	k	b	n
n	0.1888	0.57	0.41	0.02
g	0.395	0.91	0.09	0
b	0.39	0.09	0.91	0
n	0.0262	0.71	0.21	0.08
	1	52.08%	47.34%	0.59%
	122.26	63.67	57.87	0.72

[Chi]

236. WOW...Looks like the band's back together for this one 8)

Chi grabs his slippers and kicks his feet up.....
:popcorn:

[Melissa G]

237. Hi Chi! Come play on the other thread...

I saved some popcorn for you! :popcorn:
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x390193

[Chi]

239. Heya Melissa 8)

(waves spastic-ally)
Actually, I went to post something funny on two occasions, but I'm really intrigued by the game they're inventing/playing.
So I didn't want to distract.
Figured I'll pretend to be watching golf or tennis, since they're all playing so nice.
I'm warming up my golf-clap as we speak.

Thanks for the popcorn offer, if I run out before I :puke:, I'll take you up on it 8)

[anaxarchos]

240. Since when do you distract, Chi my man.... n/t

[Chi]

242. Awwww....Thanx.

Give me an A
Give me an N
Give me an A
Give me an X....
LOL
Heya Anax, how the hell are ya 8)

Been needing to ask since I read it....who the hell is the "Rocks eat Popcorn" guy?
Hehehe, my brain's been trying to make sense of that since I read it.
(Chi grabs two fist-fulls of hair and screams "MAKE IT STOP!")

[anaxarchos]

243. Rick Brady, Fundie Election Blogger...

His website is "Stones Cry Out" (What? I was close...).

He came to visit TIA once. Might have been the shortest visit in history (at least under that name).

A different one wrote to me once to ask since I used Abe Lincoln as an avatar, wouldn't I "reconsider" the Republican position?

I wrote back and said: "It's not Abe Lincoln... It's John Brown".

Never heard from him again....

[tommcintyre]

250. kick - In tribute to his tremendous contributions - WHY??? :( n/t

[freedomfries]

251. KICK

[kster]

252. kick.nt

[OnTheOtherHand]

253. but it's as wrong as it ever was...

OK, if someone gets to choose some of his assumptions, he can vary others -- can come up with dozens, hundreds, millions, billions of "alternative scenarios" -- and always get the same result.

From shell game to Excel game.

Or maybe it's more like green exit polls and ham. "Would you eat them with a tart? Would you eat them in a chart? Do you like them good and plenty? Maybe times one hundred twenty?"

Grrrrrr.