TruthIsAll banned from Kos.

[Vidar]

"Kos,
It appears that I have just been banned after two weeks on Daily Kos. I just found out. No warning. Nada. It's obvious that you are referring to me in your midday thread. It cannot just be a coincidence, since yesterday you silently took away my ability to edit or comment. You didn't have the decency to inform me that you would do so, much less tell me why. And you never responded to the three e-mails I sent to the techs telling them I could no longer comment or post.

So, these are these diaries of mine are "crap" to you?"
http://www.dailykos.com/user/TruthIsAll

[wicket]

1. Whoa.

:wow:

[blm]

84. They'd do the same to RFK Jr. Some bloggers are PAID BIG $$$ to distract

from the point that KERRY WON BIG in 2004.

The centrists in the party and the Dem strategist class NEED the perception that a more liberal senator like Kerry LOST because he didn't appeal enough to the center and the south.

That way they get Democrats to shift more right in exchange for a what they claim will be a certain victory based on CENTRIST APPEAL.

BIG MONEY is NOT to be made by pointing to ELECTION FRAUD or the corporate media as the problem for Democratic candidates.

If you were a Dem strategist in DC, could you pay your mortgage with the truth?

[Teaser]

2. Conspiracy Theories mean instabans at Dkos

sorry. you don't like the rules, don't post there.

[MervinFerd]

3. Simple but effective.

[Howardx]

12. gotta make sure kos's ad money keeps rolling in

cant be too controversial. dkos has gone seriously down hill since they saw themselves on cspan.

[Teaser]

13. post about what you know

the no conspiracy rule has been in place since immediately after the 2004 election.
Why? Because the fraudsters annoyed the shit out of everyone else there, got aggressive and offered no satisfactory proof for the allegations.

Hence the instabans, a move which the greater Kos community has applauded.


But don't worry...I hear Art Bell might want you to post something at his site.

[Name removed]

15. Deleted message

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

[Febble]

17. Posts on election fraud are allowed

on Daily Kos.

Check the Election fraud tag:

http://www.dailykos.com/tag/election%20fraud

But Kos appears to be convinced that the case that the exit poll discrepancy was due to massive, multi-million vote fraud was debunked, so anyone making that argument is going to fall foul of the conspiracy theory ban:

Today I did something I've never done before (not even during the Fraudster mess), and wish I'd never had to do.

I made a mass banning of people perpetuating a series of bizarre, off-the-wall, unsupported and frankly embarassing conspiracy theories.

I have a high tolerance level for material I deem appropriate for this site, but one thing I REFUSE to allow is bullshit conspiracy theories. You know the ones -- Bush and Blair conspired to bomb London in order to take the heat off their respective political problems. I can't imagine what fucking world these people live in, but it sure ain't the Reality Based Community.

So I banned these people, and those that have been recommending diaries like it. And I will continue to do so until the purge is complete, and make no mistake -- this is a purge.

This is a reality-based community. Those who wish to live outside it should find a new home. This isn't it.

Update: I've been reinstating some of the banned accounts as they email me. Some people wondered why there wasn't any warning. There have been warnings from others -- repeated pleadings for people to ground themselves in reality.

It's telling that I have NEVER done something like this before. Because this has been an extreme situation. This isn't about disagreeing with what people are saying. If that was the case, everyone would've been banned by now. The myth of the "echo chamber" is just that. A myth.

But as for warnings, well, this has been my warning. I wanted it clear that I was serious, and I think that has come through. I am reinstating those who ask to be reinstated. But the message has been sent.

[Tace]

21. Ahh, Now I Get It -- He Violated A Taboo

The "Reality-based" taboo. There's still people walking around who think the Earth is flat. Astronomical fact was taboo for millenia.

[Tace]

22. By The Way... I Worked For The Exit-Polling Consortium...

during elections. I was a UPI editor who would monitor the collection of exit-poll data at the main database in Michigan on election days. This was in the 1980s. TIA isn't nuts. It's Kos and his ilk who have their heads in the sand.

[Febble]

25. Well, I disagree

I have a kind of affection for TIA, but I think his analyses are wrong.

The math is good - what doesn't make sense are his assumptions. He largely ignores the possibility of "non-sampling error" in the exit polls, despite the fact that biased survey samples are a well known hazard, and that there is good evidence for Democratic over-sampling in particular in both the 2004 exit poll and previous exit polls. He assumes that people report their past votes correctly in the exit polls, even though there is good evidence that they consistently over-report voting for the incumbent. He makes heroic assumptions about the projections of the pre-election polls. And he claims they match the exit polls when in fact there is absolutely no correlation between the two sets of deviations - states in which the vote counts were "redder" than the pre-election polls were not, in general, the same states in which the vote counts were "redder" than the exit polls. This suggest that the same factor was not responsible for both sets of deviations. Far from matching, TIA's own data suggest they did not.

No, TIA, isn't nuts, although I think he knows a lot more about the binomial theorem than he does about surveys. But nor is Kos. And if there is good evidence to suggest that fraud on the scale of millions did not occur (and I think there is) then that has important implications for Democratic party strategy. It means, for example, that although the election might have been rotten (I think it was) and Ohio might have been won had the playing field been level (possibly), Kerry's campaign didn't persuade more voters to vote for him than voted for Bush.

I'd have liked to have known what TIA's response was to the pleasant little exhange of views we were having over there, and I'm sorry he was prevented from responding to my posts on his last DKos thread. I wish him well, hope his health continues to improve, and that he finds another forum where we can continue our conversation.

But I still think he's wrong.

[Tace]

27. Oh, I'm Not Saying He's Not Wrong... In Detail

I don't understand statistical math that well. I just think the fact that exit polls had always been right until 2000... and that they are the accepted way to detect election fraud... point to problems with the vote count in 2000, 2002 and 2004. It's a giant red flag that won't go away. The sky's blue...

[Febble]

30. Well, I'm afraid

not even that is true.

The were fairly close to the count in 2000, but almost as far off in 1992 as in 2004. They were substantially off in 1988 and 1996 as well. In all years they were "off" in the same direction.

And they are not an "accepted way to detect election fraud". This is a pervasive myth.

[southwood]

48. Exit poll

The main point is that the election outcome differed from the Ohio exit poll beyond the stated margin of error. This is from which normal people should be able to draw their conclusions, and is what TIA started from. That is not a conspiracy theory.

And, 2004 was in my recollection quite a bit more off than 1992. Shouldn't the Greatest Exit Polling Firm in the Greatest Nation on Earth have been able to manage a positive learning curve in 12 years, rather than a negative one? This is a second common sense argument that has left me less than convinced of the reliability of the election results.

[Febble]

50. OK, that's reasonable

as far as it goes. The first point I'd make is that there is no question that the polls were "out" beyond their "margin of error". However, the "margin of error" is calculated as the likelihood that you would get a given range of results, simply as a result of chance changes to the sample you drew. And because not only Ohio, but much more dramatically states like New York (and no-one seems particularly interested in why the exit polls were so far out in an all-lever state in which the counted results were close to pre-election expectations) were outside their MoE, we know, for as close to sure as statisticians get, that the discrepancy was not due to chance.

What a lot of people understandably missed (TIA has rather less excuse now, as it has been pointed out to him rather a lot) is that there are many reasons why a survey result should be outside its MoE other than fluke. In short, polls can be biased. This is a well researched phenomenon. Two major sources of bias can be non-reponse bias (this happens if the refusers in your survey have different characteristics from your responders) and selection bias (you inadvertently select respondents preferentially from one group). There is plenty of evidence that this has happened in past exit polls, and plenty of evidence that it happened in 2004. The Edison Mitofsky evaluation reported on methodogical factors that correlated with discrepancy, and I myself also worked, for Mitofsky, on the data, and found large, statistically significant, correlations between methodogical factors and the extent of the discrepancy. I also found absolutely no tendency for the discrepancy to be associated with benefit to Bush, as measured either relative to his voteshare in 2000, or to pre-election polling.

Strangely, this last finding is apparent in TIA's own plot:



As you can see, although TIA posted this plot to support his contention that the pre-election state polls "matched" the state exit polls, it is perfectly apparent from the plot that they do not. They do in one simple sense, which is that the average values for both polls are virtually the same (although how close they are depends on how you estimate the pre-election poll estimates, and TIA's is not only idiosyncratic, but at odds with the estimates made by the pollsters themselves).

But what they don't do is match state by state.

His blue bars represent the extent to which the official count deviated from his pre election poll estimate, and as you can see, there are some very large blue bars, a lot of them at the right hand end of the plot.

The purple bars represent the extent to which the official count deviated from the exit polls, and again there are some very large purble bars. Unfortunately, they are all over the place, and quite a few of them at the left hand end.

In other words, there is no overall tendency for states in which the official count deviated a long way from the pre-election polls (and which therefore, a priori, one might expect to be states in which one might suspect) to be the same states as the ones with large exit poll discrepancies.

TIA has also helpfully included bars that represent the DIFFERENCE between the exit poll deviations and the pre-election poll deviations, and in fact he has ranked the states in his chart by the extent of that difference. The smaller the yellow bars, the better the match. But what his plot demonstrates is that while there are a bunch of states (the ones in the middle) where the two matched quite well, they are not, generally speaking, the ones with the biggest deviations in either poll. The really suspicious looking pre-election poll deviations don't match the exit poll deviations at all. And the really suspicious looking exit poll deviations don't match the pre-election deviations at all.

Although I will grant you, that using TIA's pre elections polls, both deviations for Ohio match quite well.

But even if we grant that Ohio was crooked, and I agree it stinks, TIA's own plot doesn't say much for the relative accuracy of the two polls. Where they don't match (and in many states they don't) one of them must be wrong. Because if fraud was responsible for the deviations in both sets of polls, then large deviations in one set would match large deviations in the other.

So TIA's own data not only do NOT support his contention that they "matched" but they DO support the contention that polls can be outside their MoE for reasons other than fraud. Because if the exit polls meant pro-Bush fraud in New York, the pre-election polls must have had a pro-Bush bias. And if the pre-elections polls in North Dakota meant pro-Bush fraud, then the exit polls must have even more of a pro-Bush bias, than they actually show.

I agree there is reason to mistrust the election results. But there is no reason to trust the polls either. This is what is, essentially, wrong with TIA's analysis.

[druidity33]

67. are you saying...

then that we should never trust the validity of polls? Or do we have to have a doctorate in something to be able to reconstruct what actually is being "proven" ? What hypothesis have you created for the events of the 2004 election (since you say at least that Ohio "stinks")? Do you lend credence to the reluctant responder theory?

How would you fix this broken system?

[Febble]

70. Thanks for the questions

What I am saying is that the "Margin of Error" in surveys is a technical term that refers to the margin of error due solely to the fact that the survey responses are from a sample of the population, not the whole population. But it is not the only source of error in polls, as pollsters are keenly aware. Good polling methodology is designed to reduce "non-sampling" sources of error, and is one of the reasons direct random dialing methods were developed. But you can't use direct random dialing for an exit poll.

So the answer to your question is that all polls need to be interpreted with caution, and the methodology considered carefully when interpreting poll results. The trouble is that there is no way of knowing for sure, in any given poll, how effective the steps you have taken to ensure a non-biased poll are. However, a great deal of experimental research has been done as to what factors tend to produce bias, and large numbers of articles, books and conference papers have been written on the subject.

So the answer to your second question is that to evaluate any given poll independently of the pollster, you probably do need some expertise in public opinion research. Mark Blumenthal's old Mystery Pollster site, and his new one now at:

http://www.pollster.com

offers interesting critiques of all current polls, taking into account the methodology. But that doesn't mean that only professional public opinion researchers are equipped to evaluate polls. It just means that it is worth learning a bit about the kinds of factors that can lead polls to give misleading estimates of whatever it is they are supposed to be estimating.

As far as 2004 goes, my conclusion from the exit poll data was that methodogical factors were highly likely to have produced the discrepancies between poll and count. By "highly likely" I don't mean I had a vague hunch - I actually computed how likely it was in terms of "probability" values; I am then left to interpret whether those factors could have disguised some underlying factor associated with fraud. So I ran lots of checks for that, too. My conclusion is not definitive, and no analysis possible on that data CAN be definitive, but it is extremely well supported by the statistical tests I used. And one implication flowing from that conclusion is that vote-switching is unlikely to have occurred on a massive scale in 2004.

However, that does not rule out vote-tampering in key places (remember that only a handful of precincts are surveyed in each state - only 49 in Ohio for example) and of course there are many ways of systematically disenfranchising one candidates voters that do not involve vote-switching. Voter suppression is one; failing to supply sufficient or well-maintained machines to Democratic areas is another. There is evidence for both of these things in Ohio, as you must be only too aware.

As far as the "reluctant responder" theory goes - I've really covered that above, and elswhwere. I think it is a poor formulation of what is likely to have happened which was, simply, that the pollsters ended up with a sample that was biased in favour of Kerry voters. Evidence suggests that this happened more in precincts in which getting a valid random sample would have been most difficult. This suggests not that Bush voters were more likely to refuse than Kerry voters once selected but that they were less likely to be selected, This could have been because they were "more reluctant" - and therefore more likely to keep out of the interviewer's way (and we know, anecdotally, that some voters did so; it could also have been because Kerry voters were more likely to volunteer (and again we know, anecdotally, that some voters did so). But it could also have been that where refusal rates were high, interviewers possibly without even being aware of it, found themselves tempted to approach more willing looking voters, and to avoid approaching ones that looked more likely to refuse. So in that sense, greater "reluctance" on the part of Bush voters, or greater "willingness" on the part of Kerry voters was probably at the root. That is not the same as saying all Kerry voters were willing and Bush voters were reluctant - refusal rates were high in many precincts where there were plenty of both kinds of voters. It just means that overall, voters who were unwilling to take part in the poll, and who were therefore more likely evade selection, were slightly more likely to be Bush voters than Kerry voters. Interestingly, this has actually been demonstrated experimentally in previous elections. Factors that improve response rates push up Democratic participation more than Republican participation, so that, ironically, you end up with more pro-Democratic bias.

As to your last question, which of course is the important one, and of course I don't claim to know the answers, but my own views are that fixes required include:

• Voter registration requirements need to be overhauled so as not to discriminate against Democrats
  
• Voter suppression incidents need to be prosecuted, and people jailed if necessary
  
• Any technology that requires that people wait in long lines to vote needs to be be junked.
  
• Technology must be secure and reliable
  
• Voter-verified paper records MUST be made, and in a form that can be manually audited and/or recounted
  
• Chain of custody of the ballots must be ensured at all times.
  
• Random manual audits of the paper records must be mandatory.

So I think that HR550 is a good start.

But the reason I think it is so important to consider the evidence (including the exit poll evidence) for election theft in 2004 realistically is that I don't think bad evidence makes a good case. I think the case for radical election reform, including the problem of voter suppression is unassailable and does not depend on demonstrating whether or not Kerry won the popular vote in 2004. I fear that unfounded claims that he did (and I believe those claims are unfounded, because they are largely founded in the exit poll story which I do not consider supports them) simply get in the way. Much easier to dismiss a case if what is claimed as its rationale is fraud on a scale that is extremely unlikely to have happened.

Political scientists demonstrated, unequivocally, that more votes were intended for Gore than Bush in Florida in 2000 - and of course, Gore also won the popular vote. Those same political scientists are not arguing that Kerry won the popular vote in 2004. Claims that he did are therefore not only going to face an uphill struggle for credibility, but risk bringing ridicule on an urgent cause, unless, of course, they are supported by extremely convincing analysis. TIA's analysis is not, for reasons I have given.


Cheers

Elizabeth Liddle

[althecat]

152. Good suggestions all of em....

Voter registration requirements need to be overhauled so as not to discriminate against Democrats

Voter suppression incidents need to be prosecuted, and people jailed if necessary

Any technology that requires that people wait in long lines to vote needs to be be junked.

Technology must be secure and reliable

Voter-verified paper records MUST be made, and in a form that can be manually audited and/or recounted

Chain of custody of the ballots must be ensured at all times.

Random manual audits of the paper records must be mandatory.

[althecat]

151. Shame TIA isn't here to explain himself Febble...

My guess is what he might say looking at the graph is that what it shows is that most of the lines pink and blue are on the up side.... indicating that almost all of the error in both projection polling and exit polling is towards the democrats - i.e. underestimating GOP vote. This in turn seems to me reasonably consistent with the idea of widescale election fraud.

You will also see that all but one of the polarity correlations (HI = Hawaii) are on the upside - i.e. in only 1 instance did exit poll and projection both overestimate GOP Vote.

Secondly he might point out that the difference between projection and exit polls is +/- 2% in around half of all states and with +/- 4% in all but 8 states - which is actually a fairly good correlation.

[Febble]

153. Well, I've had a bit of a debate with TIA

about this on Mystery Pollster.

Yes, TIA's plot shows most of the bars on the up side. That is because in TIA's projections he mostly forecast that Kerry would get a greater proportion of votes than were actually counted for him, and the same is true of the exit polls.

So then, the interesting question is: did the same cause lead to both these sets of discrepancies (of which the pertinent one is fraud)?

And to do this you have to consider not the polarity per se, but the magnitude of each discrepancy. If the two sets of discrepancy magnitudes are positively correlated, then we can infer that they share a common cause. However, if we plot TIA's discrepancies against his exit poll discrepancies, we find that they are actually slightly negatively correlated - i.e. the slope goes in the opposite direction predicted by the hypothesis that both were caused by fraud.

It's not much of a negative trend, so I wouldn't want to read too much into it, but it is certainly not positive. In other words there is no tendency for TIA's discrepancy to greater where the exit poll discrepancy is greater and vice versa. Sometimes both discrepancies are large. Sometimes both are small, or even pro-Bush. Sometimes one is large and the other is small. Yes, they both tend to be pro-Kerry overall, but no, they are not correlated.

If we plot exit poll discrepancy on the horizontal axis, and TIA's poll discrepancy on the vertical axis, Hawaii, at -3 in the exit poll and -1 in TIA's pre-election poll would be in the bottom left quadrant of the plot. And Alabama, where both are around 4, would be in the top right. But NY, with over 5 in the exit and -3 in TIA's poll would be in the bottom right quadrant, and Tennesee (nearly -2 in the exit, and 6 in TIA's poll) would be in the top left. And the linear best fit line through all the data points, as I said, slopes slightly from top left to bottom right, not bottom left to top right, as would be predicted if a common cause (e.g. fraud) was responsible for both.

That is the sense in which the two do not match, rather importantly for TIA's case.

What I think his response is, is a variant on a very good case that eomer made to me regarding my own similar finding that precinct level discrepancy was not at all positively correlated with "swing" to Bush ("redshift" relative to 2000). He pointed out that if all the variability in the polls was due to polling error, nonetheless the mean value of the shift could be due to fraud. TIA hasn't said this explicitly, but he sort of implied it in a discussion of New York. If both his polls and the exit polls are fairly sloppy, but the average redshift in both was due to fraud, you wouldn't necessarily see a correlation between the two sets of discrepancies. But if this is TIA's case, he then runs into problems with his other claims about the extreme accuracy of the polls. They can't be sloppy enough to account for the lack of correlation AND accurate enough to be gazillion zero probability evidence of fraud.

At least not the way I'm seeing this.

[mom cat]

51. People only "over report" for the incumbent over time, not right after

voting. You state:

"He (TIA) assumes that people report their past votes correctly in the exit polls, even though there is good evidence that they consistently over-report voting for the incumbent."

If people over reported for the incumbent in the exit polls for the incumbent Bush in the exit polls and Kerry won in the exit polls, then Kerry's lead , by your theory, must have even been greater than TIA claims.
You should review your own criticisms before posting against TIA.

[Febble]

52. You misunderstood me -

Sorry I wasn't clear.

I think it is likely that on the whole people report correctly the vote they just cast. But a subample of exit poll respondents were also asked how they voted in the previous presidential election, ie, the election four years earlier. TIA assumes they report that vote correctly. However, Mark Lindeman found in every single exit poll since 1976, the winning candidates winning margin had retrospectively increased since the previous election.


http://inside.bard.edu/~lindeman/too-many.pdf

(You might like to edit your last sentence.)

[mom cat]

63. If you agree that exit polls ate largely correct, then why do you dismiss

the difference between the exit polls in 2004 and the "reported vote"?

[OnTheOtherHand]

65. (1) she doesn't, and (2) she doesn't

(1) Her statement implies that people who participate in the exit polls generally report their own votes accurately. That doesn't entail that the exit polls are accurate, because there is no reason to assume (for instance) that Bush and Kerry voters were equally likely to participate.

(2) Febble has written hundreds of posts, and much else besides, addressing the exit poll discrepancy. How could one possibly say that she "dismisses" it?

[Febble]

66. Well, as OTOH said, far from

"dismissing" it I have devoted a large chunk of the last couple of years investigating it. You might like to run a search on my posts.

But I'll do a very quick summary:

I was so interested in the discrepancy at precinct level (because at precinct level you have the most statistical power) that when the Edison-Mitofsky report came out, I pored over it in minute detail to try to figure out what it was telling us. And I figured out that they probably hadn't measured it in a very fair way - it underestimated the discrepancy in extreme precincts, particularly high Dem precincts. I figured out a better way, and published a paper on the web, and posted some DKos diaries. Warren Mitofsky got to hear about it, realised I probably had hit on an important problem, re-ran some of his analyses, then hired me to run some more, which I did. Then Mark Lindeman, Rick Brady and I worked some more on the problem of how to measure the discrepancy, and I ran them all again. What I was looking for were "correlates" of discrepancy - what factors were associated with bigger differences, particularly differences favoring Bush (what Jonathan Simon, I think it was, termed "redshift"). And I found that redshift was greater under certain conditions - mostly conditions in which it would have been more difficult to get a good unbiased sample. For example where the interviewing rate was low, or where the interviewers were standing a long way from the precinct. And I found that just a few of these factors were enough to account for all the significant redshift. However, there is a fair bit of wiggle room in the data, so I also looked to see whether there was any evidence of the discrepancy being greater in precincts where Bush had done suspiciously well. And I had data on his vote share in 2000 (a year in which the exit poll discrepancy was small). So I tested to see whether the discrepancy was suspiciously large in precincts where he had increased his vote by a large amount, and perhaps smaller where he had done fairly badly. And I found absolutely no tendency for that to be the case, no matter how I sliced and diced it.. The discrepancy was simply not correlated with any apparent benefit to Bush.

So, far from "dismissing" the difference I've probably looked harder at those discrepancies than anyone else on the planet. Sure, I was lucky to have that opportunity, so I'm not bragging about it. But it certainly was precisely because I didn't "dismiss" the difference that I ended up analysing those differences in such enormous detail. And I didn't find evidence that the discrepancies were due to fraud. But I DID find evidence that they were associated with methodological factors in the poll.

So to get back to your question: I think it is likely that people reported their vote correctly (although it is of course possible that not all of them did - there are anecdotal reports of people deliberately telling interviewers the wrong answer). I also don't think that there was a very marked tendency for Bush voters to refuse more than Kerry voters, although there is some evidence that certain groups of Bush voters may have refused more. What the data suggest, strongly, is that bias crept into the poll at the level of respondent selection. Kerry voters were more likely to be selected than Bush voters. This could have been because they looked more enthusiastic, they were more likely to volunteer; or because the Bush voters avoided making eye contact; or the Bush voters were more likely to evade selection in the first place. The data can't tell us which of those it was. But it can tell us that where that where sticking to strict random sampling would have been more difficult, there was greater redshift in the poll.

It certainly doesn't rule out fraud, although it does mean that widespread massive vote-switching is unlikely. And it does suggest that the exit poll data itself is not strong evidence for (indeed it is actual evidence against) widespread vote-switching fraud. But it makes no difference, IMO, to the case for election reform; and it makes no difference to the strong evidence for voter suppression of various forms. And it makes no difference at all to the case that paperless electronic voting is insecure, non-transparent, unreliable and unauditable.

Hope this clears things up. Thanks for asking.

Lizzie

[eomer]

72. A couple of angles for you to consider, and then a toast.

Hi Lizzie,

As usual I have a couple of angles to point out for your consideration. Regarding the correlation you found between the WPD and the distance requirement for exit pollers, it seems to me there are various ways the conclusion of a correlation could be mistaken and/or misleading.

One way it could be mistaken is if there is some other factor that is correlated with WPD and that other factor has a causal or otherwise direct correlation with distance. For example, geography. The distance a poller must stand away is usually set, I think, at the state level. So there is a causal correlation between which state a precinct is in and how far the poller must stand. So now if there is a correlation between which state a precinct is in and the WPD then that will make it look as if there is a correlation between WPD and the distance requirement when really there may not be if you control for the other factor of the state within which the precinct is located.

There could be other, more subtle, relationships between these factors. For example, it is not impossible for there to be a correlation between tendency to commit fraud in a state and tendency to set large distances for exit pollers. So the states with large distance requirements turn out to be the states where fraud is most prevalent. It would appear that the distance requirement caused the redshift when in fact distance, fraud and redshift were all three caused by political tendencies of the state.

Another questions I have about this correlation is what effect on statistical force there is from the fact that the distance is set as a constant at the state level and therefore does not have a freely ranging set of values. I put this as a question to you rather than an argument because I don't have the expertise in statistics I would need to figure it out myself. If the distance requirement were selected randomly for each precinct then it would be clear that the number of precincts would directly imply statistical force. Does the fact that the precincts have a limited number of possible distance requirements and that these distance requirements may, many of them, be all the same value sap away your statistical force? It seems just intuitively to me that it would. Does it and, if so, have you accounted for that? Saying it one more way (more for my benefit than for yours) if we have a sample size of n but three quarters of our data has the same value for our X variable then aren't the standard formulas for significance of a correlation between X and some variable Y out the window?

Finally, a way in which a correlation could be misleading. Even if the distance correlation is a true and valid one when looking across the national sample precincts, it may still not be the full explanation in some subset of the data. Specifically, the discrepancy in Ohio could be explained only partly by the distance correlation and then the rest of it by fraud. Since Ohio had a high distance requirement (100 feet) then fraud in Ohio would ironically reinforce one of your arguments against fraud. Fraud in Ohio would pump up the correlation between distance and WPD and contribute to the conclusion that WPD was not caused by fraud. Regarding your argument that is based on swing/redshift correlation, I'm not sure what fraud in Ohio would do. Ohio had a swing to the blue and a high redshift so you wouldn't expect a strong correlation between swing to the red and redshift in Ohio. Would it be the case that the blue swing would counteract and hide any redswing due to fraud? I'm not really sure -- maybe you have an idea.

As a point of clarification, what I'm disputing is whether the exit poll is evidence against fraud. I'm not as convinced as you are that it is. I think that the system we're examining is too complex, with possibly many hidden relationships and that therefore any conclusions should be taken with much caution. The other question, whether the exit poll is evidence for the existence of fraud is not what I'm addressing here.

BTW, this is my 1000th post. I thought a post to you on this topic was an appropriate way to spend it since a good number of the previous 999 occurred that same way.

So, here is a toast, on the occasion of my 1000th, to a place where people can come who are interested in a debate that is spirited, engaging and energetic but for the most part free from excessive rancor, on a topic that is extremely important to those who participate here and elsewhere and also to those who are totally unaware of it.

Cheers,
eomer

P.S. - I'll be offline shortly and for the rest of the weekend so if I don't reply back, that will be why.

[OnTheOtherHand]

73. happy 1000th, eomer! (with an edit to add)

I've learned a lot from your posts, and I've always appreciated your sense of humor.

Not to preempt Lizzie's response, I note that there is potential confusion about whether the exit polls might be "evidence against fraud" or "evidence against... widespread vote-switching fraud" or "widespread massive vote-switching fraud" or whatever. Some people argue that the exit polls tend to support a swing of some 8 to 10 million net votes from Kerry to Bush. I think both Lizzie and I believe that on the contrary, the exit poll data tend to infirm that scenario. It isn't a straw man, because as far as I can tell, it's what TIA and Freeman (among others) actually think happened.

With respect to fraud in Ohio, I think the exit poll data do not say very much in any direction. I certainly don't think the Ohio exit poll means that Kerry probably won Ohio by 6.5 points, but that's not important in itself! We've pointed out repeatedly that Kerry's apparent exit poll "lead" in Ohio was within the margin of error (although the discrepancy was not), which some folks interpret as an argument that there probably wasn't decisive fraud in Ohio. It isn't. If the exit polls can't support the inference of a 6.5-point Kerry win in Ohio, they aren't likely to support the inference of a 2.1-point Bush win there either.

[eomer]

74. Thanks, OTOH.

No time today to address the substance but I'd like to when I'm back. I think I can convince you on a couple of my ideas, at least partially, and I'm sure you will convince me of some things too. That's the debate I'm toasting on this day, but will have to carry on on another. So, today, it's just: cheers -- and I'll try to pick it back up with you next week.

[Febble]

75. Congratulations on your 1000th post!

And I am honoured to be the recipient of it.

All your points, as usual, are excellent. I'll respond as best I can.

You are of course absolutely correct about correlational analyses. Notoriously, you cannot infer causality from a correlation - all you can infer is that something was causal to both. It is, in fact, exactly the same objection I have to Freeman's inference from the (lack of) correlation between refusal rates and proportion of Bush voters that Bush voters were no more likely to refuse than Kerry voters. While this may be true (it is certainly not a large effect) he omits to consider that a third variable - urban density - may be correlated with both (it was).

So finding a correlation between interviewer distance from precinct and redshift is not sufficient to allow a definitive inference that the distance caused the redshift. A third variable - fraud - could have been responsible for both. For example, the perpetrators of fraud might also have wanted to ensure that interviewers kept from being too close to the the fraudulent precincts. Similarly the correlation between interviewing rate and redshift does not allow the definitive inference that low interviewing rates cause redshift. Large precincts were allocated low interviewing rates. Fraud might have been targeted on large precincts. Therefore a third variable - fraud - could be common to both.

And while it is harder to see how some the other methodological correlates of redshift (interviewer characteristics, for example) the possibility that some of them were collinear with fraud cannot be ruled out.

However, that was the point of the swing-shift correlation. Not only was a substantial proportion of variance in redshift attributable to methodological variables, when those variables were included in the multiple regression model, there was no significant remaining net redshift. That means that if fraud were responsible for the mean redshift, then fraud must have had variance.

I think you probably see where I'm going with this. If fraud had variance, and was reflected in redshift, then redshift should be correlated with swing. And it wasn't.

Your intriguing response to the lack of swing-shift correlation was to suggest that variance in redshift might be attributable to methodological factors but superimposed on largely uniform fraud, and that was why I spent many happy hours trying model a scenario in which I could squeeze fraud into the data without producing a less-than-zero swing shift correlation - but was only able to do it with fairly heroic assumptions about fraud uniformity. But the other reason it doesn't work is that the methodological factors themselves accounted for not only the variance but also the degree to which the mean was different to zero.

But I'm immensely grateful to both you and to TimeForChange for inducing me to check these things out so thoroughly.

And I think you are correct when you say "the system we're examining is too complex, with possibly many hidden relationships and that therefore any conclusions should be taken with much caution." There is plenty of room for fraud in the data. There is very little statistical power at state level, and some of the states of which I am most suspicious (New Mexico, for example) didn't even have a particularly large exit poll discrepancy. What I can conclude, I think, is that it is unlikely that vote-switching on a scale of millions went on, and certainly that the exit poll data is not evidence for it. It remains possible, however, that whoever designed the coup, cleverly targetted fraud in such a way as to defy the kinds of analyses you and I have come up with. But my prior on that is low.

Have a good weekend! And :toast:

Lizzie

[truedelphi]

142. Could you explain how and why after working as you did

You now support TIA's conclusions.

If you don't want to post here, feel free to Private Message me.

[Stevepol]

24. The "conspiracy theories" are quite well supported and have never

been debunked by anybody. I'm sure you've read Steve Freeman's book, perhaps even contributed to it, in an adversarial way. But to my mind it offers the best overall presentation of the facts available.

Here you have machines that are proven to be eminently hackable in completely undetectable ways counting votes in total secrecy without verification, in 40% of the cases come 06 without even the POSSIBILITY of verification, and there are people actually suggesting that theories of widespread fraud are "conspiracy theories"? Come on. Just an awareness of human nature tells you that with in a state where the conditions as described above exist it would be a conspiracy theory to think that fraud DIDN'T happen.

As far as I know the only evidence that has been adduced to indicate that the machines weren't tilted is the fact that some other exit polls have been off and other types of voting machinery had a higher WPD then the machines. Both the other types and the electronic machines of course had a WPD way higher than paper. And there could be many reasons for this fact that a good investigation would likely uncover. Obviously there are other reasons for the fraudulent election than the machines but the machines added a great deal to the mix IMO and I think the facts support that opinion.

Freeman's book, it seems to me, puts to rest that tin foil hat theory that reluctant Republican responders are the reason for the discrepancies between the exit polls and the alleged vote.

It really comes down to who is to be trusted: every computer scientist in the country or the "experts" at Diebold et al.

Avi Rubin recently on Diane Rehm's show revealed that he had been threatened with a law suit by Diebold. They claimed his ideas about their machines were filled with "egregious" errors. Mind you, this ia a computer scientist, one of the most eminent in the country. I was threatened with a lawsuit myself for daring to write a LTTE and making comments about another of the voting machine companies.

[Febble]

26. Yes, they've been

"debunked" - by me, for a start. You might not agree with the debunking, but they've been debunked.

I don't think the machines are secure, and I agree they could be hacked. And that even when they aren't they are unreliable and not fit to serve a democracy. But it is not true that WPD was lower where paper was used. It was one of the things I checked out (although it was also fairly clear from the E-M report). When precincts serving the same kinds of communities that used paper ballots were compared with the paper ballot precincts, there was no statistically significant difference. When precincts serving large urban communities were compared (and no paper was used) WPD was higher where punchcards and levers were used, and lower where DREs and opt scans were used.

Steve makes a good case for a rotten election system, especially the case that hispanic and black communities are systematically disenfranchised. This is a disgrace. But when it comes to the exit poll analyses, almost every one of his analyses is fundamentally flawed. In fact, to be honest, I can't think of one that isn't.

edited for dodgy grammar

[Katzenjammer]

53. I don't think "debunked" means what you think it means

A claim is "debunked" only when the counterclaim is accepted as definitive. Yours isn't.

[OnTheOtherHand]

54. then is it ever possible to debunk anything?

Are we spiraling into total subjectivity -- "You're only as wrong as you feel"?

[Febble]

55. "is accepted"

by whom? How is "definitive" defined?

But in any case, I am not claiming that anyone has accepted my "counterclaim" as "definitive" - I wouldn't claim it was definitive myself. But there are certainly serious flaws in Freeman's arguments. I can tell you what they are if you are interested, and you can evaluate them for yourself.

[bleever]

28. Gotta love any post with the word "adduced" in it.

And it's a great paragraph. Nicely put.

[OnTheOtherHand]

36. no, this argument doesn't work for me at all

If you look at the returns from Ohio, the DRE results are basically no different than all the other results. So it seems basically meaningless, or circular, to say that DREs "added a great deal to the mix" there.

This has nothing to do with the computer science critique of DREs. There is little distance between me and Avi Rubin on DREs, except of course that he is an expert in that field and I am basically saying "ditto."

Perhaps surprisingly, there is not much distance between us on the 2004 election, either. Here's what Avi Rubin says on p. 258 of his new book:

The many activists who definitively claimed, with no evidence, that the election had been stolen did a great disservice to everyone who had thoughtfully and seriously criticized the e-voting technology.

The way I interpret that, it doesn't mean that anyone who thinks, or argues, that the election was stolen is thereby separated from the ranks of thoughtful and serious critics. It is unsupported definitive claims that arouse Rubin's ire, because they undermine other work. (Rubin's own stated view is that Bush probably won, but extremely clever fraud is also possible; "(n)either fraud nor legitimacy in the voting can be verified." He doesn't get into vote suppression at all.)

You can take Freeman's book seriously, but as far as I can tell, there aren't many survey professionals who do, at least with respect to the exit polls. Febble and I have spent a lot of time trying to explain why. We're in the minority not because we don't think the exit polls indicate fraud -- ours is the mainstream opinion -- but because we try to explain why.

I sort of agree with you about "conspiracy theories." I don't care whether a theory is a "conspiracy theory," I just want to know how likely it is to be true.

I think it's amazing that TIA can insist that the pre-election polls "matched" the exit polls even as he posts a graph that shows that they didn't. I wouldn't have considered that a banning offense; I just don't understand the thought process.

[Einsteinia]

126. Kos also admits he likes Chuck Hagel & CIA,

& and he applied to work for the CIA, but says he declined because he didn't want to move to Washington D.C.

I've heard so many people complain that he continues to not believe the election integrity issues are real NOW, despite the fact that they've been proven so many times over.

And for me, having recently read, "Confessions of an Econonic HitMan," which is the true story of a guy who declines the CIA but who ends up working for the civilian apparatus to it, it makes me wonder who's really on our side. Probably overly suspicious about someone who simply disagrees. I dunno.

[Febble]

127. Actually,

I see very little evidence that Kos "continues not to believe the election integrity issues are real NOW", and election integrity issues are regularly aired in Kos diaries, as a search on the "election fraud" tag:

http://www.dailykos.com/tag/election%20fraud

will show.

What seems apparent is that he has zero tolerance for election fraud diaries that he considers make inflated or unsubstantiated claims, and that he himself does not believe that massive electronic vote switching occurred in 2004.

Which is a perfectly legitimate view that I happen to share, although I do not share Kos's zero tolerance of even inflated claims. But then I'm an academic and he's concerned with Democratic strategy.

I suggest Booman Tribune for Kos fraud refugees:

http://www.boomantribune.com/

edited to complete unfinished sentence

[Einsteinia]

130. It was only about six weeks ago that Kos

It was only about six weeks ago that Kos was continuing to say he did not buy it. He's NEVER once admitted to the factual accuracy of ANY of the findings.

At best, I think he holds his nose and tolerates posts that would make him look really bad should he ban them, i.e. the Princeton report, the Brennan Center.

He is NOT a believer, and I find that at this juncture quite damning.

But I would love to be proved wrong.

[Febble]

131. Do you have a link?

I am sure Kos does not think that millions of Kerry votes were stolen in 2004, but he has certainly tolerated diaries (even consented to the frontpaging of that major Ohio diary by Georgia10 on Ohio) that make a decent argument. Georgia10 is even a regular frontpager now, and I have no reason to think she has reneged on her magnum opus.

But I'd like to know what you mean by "a believer" - a believer in what? That the election system needs reform? I wouldn't know, and I can believe it's probably not high on his agenda, but do you think he thinks everything is hunky-dory?

Or a believer that Kerry won the EV? In which case his view is shared with a fair number of people who nonetheless passionately urge election reform (me for instance - I think the jury is out on Ohio, but my prior is that probably Bush would have won it anyway).

Or a believer that Kerry won the popular vote? In which case do you damn all those who share his disbelief?

I'm not trying to be argumentative here, I'd just like to know what you think it is that Kos doesn't believe and doesn't tolerate. He certainly outlaws what he sees as bad arguments that 2004 was stolen. But I see no evidence that he outlaws good arguments, even though he may not be convinced. And there are plenty of election fraud diaries on DKos.

edited to add:

from On the Left Tip:

What do Markos and Jerome think of the subject of election fraud? If Diebold and others have the voting machines, how do we deal with this?

• Markos: The tone of the entire voting machine issue is so negative. You get people who just say "well, my vote won't count anyways" and then they don't vote. There are so many other pre-voting issues - the Secretaries of State who decide how many machines go to what district - the canvassing boards who decide who will be include on the felon list. The issue of electronic voting is less important that addressing the systemic and structural issues.

[althecat]

148. So Markos Gets CNN Advertising & Turns Into Stalin....

What a fucking moron..

Its shit like this that is why your democracy is so totally fucked up.

[truedelphi]

78. Conspiracy? Conspiracy?

Conspiracy is just another word for NOT THE OFFICIAL STORY

In 1933, in Germany, you would have been branded a conspiracy theorist if you expressed feeling uneasy about Adoph Hitler's fast rise to power.

In 1936 in Germany you would have been branded a conspiracy theorist had you stated openly that bad things were going to result from the Jewish populace being forced to wear the Star of David.

In 1937 after the invasion of Czechoslovakia, you would have been branded a conspiracy theorist for saying that this was the beginning of Germany's decline.

In 1939 in Germany you would have been branded a conspiracy theorist if you had stated that it was absurd to think that the nation of Poland had invaded Germany.

In 1941 when Goering announced that the Allies bombs and missiles would never afffect German soil, and you denounced this sttement, saying simply, What goes around comes around, comes around, you would have been called a conspiracy theorist.

In late 1941, when the last of your Jewish neighbors are rounded up, you say, "I bet we never see them again. I doubt that where they are going has anything to do with an establishment of community for the Jewish populace" you will be branded a conspiracy theorist. (Remember the reason that the German police originally started to round up their Jewish friends is that after Kristelnacht, the German people were told that measures designed for the protection and safety of Jewish people were going to be implemmented.)

In late 1943 if you are a German in Germany, and you mutter to yourself that there is no way that Germany can win, yes, you guessed it, you would be branded a conspiracy theorist.

Yeserday here in America, the President of the United States announced that "It is not right to think."

many of us who do think will be branded conspiracy theorists by the right wing talk show hosts,
demonic by the right wing "Christian" crowd, and threats to the national security by those in Homeland Security.

But hey, there are always those who think outside the box.

[dchill]

154. In THIS case ...

Conspiracy Theory = Unprosecuted Proven Case. (CT=UPC)

[Pirate Smile]

4. Ahhh. That is who the PSA Kos mentioned was regarding.

"A PSA -- if you get yourself banned on Democratic Undergound, don't think you can come here and carry on whatever crap that got you banned at DU in the first place."

[Canuckistanian]

19. Markos wrote that?

That's an ignorant statement. People get banned here for all sorts of reasons, mostly because of personal conflicts and in flame wars.

And personally, I think we should have "temporary bannings", wherein a poster who breaks the rules gets a 2 week "timeout", instead of an immediate, non-rescindable tombstoning.

[Tace]

5. I Don't Understand Why TruthIsAll Is Poison

I didn't understand why he was poison here, and I don't understand why he is poison there.

[Vidar]

6. He's a hero in my book.

I was temporarily absent from DU when he was banned here--unsure of details.

[RadiDem]

7. Whoa! Its not like he was talking about 'Bush's Tsunami Machine'! That sux

[Monkeyman]

8. I got banned for not posting a link

My fault and rules are rules

[Vidar]

9. Sounds like a good place to avoid.

[Febble]

10. Tell TIA

to try here:

http://www.boomantribune.com/

It's where the DKos election fraud refugees hang out.

[Vidar]

11. Thanks.

[Land Shark]

14. I don't go to Kos much, having a hard time understanding stories I hear nt

[mod mom]

16. I agree-the set up sucks. DU is a million times easier to navigate.

[blm]

87. Understand this - Dem CONSULTANTS and advisors have mortgages.

And DC ain't a cheap place to eat.

There is NO MONEY in it for them if the Dem lawmakers and candidates wake up to the realization the elections are being stolen through mechanisms they have no say in.

THey make ALOT of money telling pols they need to focus group, or finesse issues towards the center, or tweak their language to appeal to the south, or anything BUT that the machines are likely rigged and the election fraud is going on 24/7 throughout the 4yr. cycle.

[truedelphi]

132. I'd never thought of that but having met my

Fair share of consultants - you may well have hit the nail on the head.

[truedelphi]

136. From the lips of the damned (Oops I mean banned)

OTOH
Political scientists at least intend to be a reality-based community. The political science literature pretty much takes for granted that Bush won the popular vote in 2004. It's fine for people to assume that that is just a collective mental flaw or character defect, but it isn't especially helpful.

TIA
I guess no one has written a plausible explanation of how Bush won, since it would have to refute all the evidence (statistical, EIRS, Conyers, RFK Jr., Fitrakis, etc) of massive fraud, all favoring Bush. Now do I expect to see one, since we haven't yet seen: "Intelligent Design: How Bush Found 14 Million New Votes in 2004"
...
...
...
By the way, I'm still waiting for one of your poli-scientists to write the book: " How Bush won by 3 million Votes". You promised that it would be out by now.
-------------------------------------

[OnTheOtherHand]

140. thanks, I will not read TIA at that certain external site

But sorry, I don't know what TIA is talking about here. There is absolutely no difficulty whatsoever in finding books, by political scientists and others, that offer explanations -- right, wrong, or mixed -- of how Bush won in 2004. I suppose TIA is joking, but I don't understand the joke.

As for the evidence, as TIA's friend autorank once wrote here on DU, "If all the evidence of election fraud, the exit polls are the unifying force, the only national evidence...." Indeed. So, if the exit polls don't hold up as evidence that Kerry won the popular vote, what does? (As eomer pointed out at #113, whether Kerry won or should have won Ohio is a different and more important question.)

[salinen]

18. You might consider that a trophy

[bleever]

20. If the Kos defines "reality-based community" as acceptance of a

actual Bush victory in 2004, that's his own reality, and he's welcome to it.

Data-phobic and afraid to be excluded from the commentariat. That's one way to go.

[glitch]

29. "excluded from the commentariat" nails it. Desire for acceptance

kills innovative thought, IMO. Kos was tacky about the banning too. Tsk.

[Febble]

46. Whatever else Kos might be

"data phobic" seems scarcely appropriate. There is good data to support the case that Kerry won more votes than Bush. Do you think, for example, that almost the entire political science profession is "data phobic"? The academic consensus appears to be that Bush won. The academic consensus may be wrong. But academic social scientists are not data phobic. The data supports their case.

Ironically, even TIA's data supports their case.

But I was sorry to see him banned from DKos.

[Febble]

47. Freudian slip

and too late to edit:

I meant of course that "There is good data to support the case that Bush won more votes than Kerry", although I understand that is a minority view on this thread. It is, however, a majority view among many data-philes, including several political scientists with quantitative skills way beyond mine.

[bleever]

59. Kos says "fraudsters" have "crap" theories outside of any reality.

Yet even you acknowledge that there is in fact room for debate and detailed analysis. I give you credit for a much more sophisticated view of the matter than Kos's "off with their heads" approach.

[Febble]

61. Well, thanks

I certainly acknowledge there is room for debate and detailed analysis. My computer is groaning with detailed analysis.

And I like debate, so I hate bannings.

Cheers,

Lizzie

[meganmonkey]

23. I'm not surprised

As people really grasp the election fraud problem, they often become more critical of the Dems for letting it go unchallenged. Kos is just one of many sites which rely on loyal Dem support. I think 9/11 and I/P are very similar issues in that way.

Fortunately there are still some forums without that fear, where folks like TIA can post and discuss their work.

:hi:

[Tace]

31. TIA Undermines Kos And Skinner

It occurs to me that this is probably at the root of his banning from these sites. Think about it, Kos and DU make their bread and butter from Democratic activists who don't want to believe that the whole exercise in electoral Democracy could be futile. I don't want to believe that either, but I'm beginning to understand their aversion to the factual truths he presents -- it undercuts what Kos and Skinner see as the purpose of these sites. I'm starting to get it. : )

(No offense, Skinner. I'm a long-time DUer, and I'll follow the rules. I love this place, and try not to rock the boat. I got a warning for taking part in a weather-control thread a couple of years ago.)

[GuvWurld]

32. Why Old Election Numbers No Longer Matter

http://guvwurld.blogspot.com/2005/08/why-old-election-numbers-no-longer.html

Wednesday, August 17, 2005
Why Old Election Numbers No Longer Matter

This apocryphal parable contains only true facts. I know because I made them up myself.

The date was Sept 33, 1965. Samdy Kouflax ate eight live pigs, a new record among those in his Pagan cult. This sign of virility and luck soon paid dividends as Kouflax threw one of golf's most memorable games. Going all nine chuckers, Kouflax completed a no-hitter to lead his Chicago Whitehawks to the Super Bowl with a victory over the Boston Celtics. Kouflax's solo grand-slam leading off the first inning was all the offense needed that season. Many also remember this game for the unusual brawl that broke out between two Celtic mid-fielders. The ensuing power play enabled the 'Hawks to expand their zone defense with 20 men on the ice. All in all, it was a fabulous year for Samdy and his wife Whora.

Can you imagine yourself getting into a debate about who won this game or what was the final score? Any such discussion would validate the legitimacy of the idea that there was an actual game with a winner, a loser, and a knowable point tally. If you were asked who won, or what was the final score, the only reasonable answer is that it cannot be known from such nonsensical reporting. There is nothing in this story to suggest that a real game of any sort was actually played.

U.S. federal elections are held under conditions that are equally farcical. At this point, any discussion of the numbers, or the outcome, is only serving to reinforce that an actual election took place. It is imperative that we make it conventional wisdom that no actual election took place, that instead we witnessed and partook in simulated competition. The core argument that summarizes this frame is: there is no basis for confidence in the results reported from U.S. federal elections.

MORE...

This essay is also included in my book, We Do Not Consent (free .pdf download)

[Febble]

33. I've always liked

that frame. It's bullet-proof.

[Cookie wookie]

64. Hear Hear!

No electronic voting system (that I know of) is a transparent process. Transparency is an essential element of democratic voting systems. Therefore, no electronic voting system (that I know of) is a democratic voting system.

Add to that those evoting systems that lack any means for independent audits and we have a system that is not transparent, verifiable, and therefore we (the people) can not determine if the results are accurate.

80% of the US voting on electronic systems in November.

Democratic elections in the US
RIP

[Straight Shooter]

34. Ironic that someone who goes byTruthIsAll can't get the truth out.

This isn't the first time he's been blockaded.

[burythehatchet]

35. Kos is a whore

just with a different clientele

[fooj]

37. There's so much info in those KOS threads...

TIA rocks! He's got it going on. I'm hoping he has communicated his findings to RFK Jr. and John Conyers. Glad to see he is feeling better. :)

[Awsi Dooger]

38. TIA needs a bit of finesse

That's not his style but I knew he wouldn't last long at Kos when I saw his first diary a couple of weeks ago. It was full blast, as always. He should have commented in some of the diaries under the election fraud tag and eased into the website with an occasional short diary. I was going to respond to some of his stuff at Kos but I handicapped the situation properly and realized it was going to be a quick hook.

[Blue Shark]

40. So much the sadder...

...for a reality-based community.

[OnTheOtherHand]

49. where to start?

Political scientists at least intend to be a reality-based community. The political science literature pretty much takes for granted that Bush won the popular vote in 2004. It's fine for people to assume that that is just a collective mental flaw or character defect, but it isn't especially helpful.

If TIA intends to contribute to discourse within a reality-based community, he needs to say something interesting. He doesn't. Assuming that exit polls are inherently accurate within sampling error is not interesting. Saying that the exit polls "match" the pre-election polls, while posting data and charts that demonstrate that the exit poll discrepancies and deviations from pre-election polls are essentially uncorrelated, may evoke voyeuristic schadenfreude, but is not otherwise interesting.

Basically, TIA has no clue how to even start to convince someone who (1) has some familiarity with survey methods and election returns and (2) doesn't already agree with him. I guess it's pretty fun if one does agree with him, but I don't see what good it does.

(All this quite apart from whether he should have been banned from DKos. I would have been perfectly happy to continue to point out, to anyone who cares, why his arguments leak like sieves. The whole "Saint TIA, Prophet Without Honor" business doesn't benefit anyone.)

[southwood]

68. Think!

The fun actually starts when people refuse to 'pretty much take things for granted'. Isn't that what being a scientist, an academic, an intellectual or a critical mind is all about?

[OnTheOtherHand]

69. gee, thanks for sharing that thought

Here's some of my work on the subject: http://inside.bard.edu/~lindeman

Do you really suppose that the reason the political science literature pretty much takes for granted that Bush won the popular vote is that political scientists never considered any alternatives?

Well, to repeat myself, it's fine for people to assume that that is just a collective mental flaw or character defect, but it isn't especially helpful.

(edit to add) It occurred to me that since you probably don't know me, you might read something into my post that isn't there. My point isn't that the mainstream opinion of political scientists must be right. It's just that I don't see how it is helpful to assume it is wrong, to ignore it, and to make no attempt to formulate arguments strong enough to convince at least some political scientists. But the point isn't limited to political scientists. It applies to many other groups of people.

[truedelphi]

79. See my post inthis topic labelled "Conspiracy? Conspiracy?"

they who control the discussion, control the discussion - be it academia, mainstream media,
the clout of a totalitarian political party.

[OnTheOtherHand]

85. I don't think there is one "discussion"

and in many cases there is no one in particular who can control the discussion. Certainly some people have gone to some trouble to discourage Febble and me from posting what we believe here on this forum, or failing that, to encourage other people not to read it. From my point of view, of course, they are attempting to control the discussion. They apparently think that we are attempting to control the discussion. I have my opinion about what a neutral observer would conclude, but I suppose it really doesn't matter.

Academic discussions aren't very tightly controlled. If someone says something controversial and interesting, certainly those who disapprove can try to shut it down, but they have no assurance of succeeding. Personally, I think the reason that critiques of DREs have more traction among computer scientists, than arguments that Kerry won the popular vote have among (inter alia) political scientists, is that the DRE critiques are better. Certainly political scientists had no trouble saying that Gore should have won in Florida.

But backing up to Kos banning TIA -- well, Kos can do what he wants. Every board has the right to decide who it will ban. I think it's ridiculous for TIA to characterize it as "pure fascism." I've tried to carry on civil conversations with TIA, and I don't think he is any good at it. It's astonishing that Febble gives him the time of day. He should be grateful, and embarrassed.

[Febble]

71. I couldn't agree more.

The first thing any scientist does with a hypothesis is to test it.

And having tested it, and found it supported, the next thing a scientist does is to try to disprove it. Possibly by turning to different data; possibly by running a different test on the same data.

Where I think misunderstanding has occurred is that people assume that those who investigated causes of error in the exit poll (including me) failed to consider the hypothesis that the discrepancy was due to fraud. It is certainly true that pollsters were likely to regard methodological factors as highly likely - after all, they know, from past analyses and from the nature of polling that a number methodological factors sometimes lead to bias, and indeed they knew in 2004, before a single result had come in, that some of their poll results were out of line with their pre-election estimates. So they suspected bias from the get-go. But that does not mean the alternative was not considered or tested.

And I can state, unequivocally, that I both considered it and tested it. One finding was presented in public by Mitofsky, and appears, with commentary, by my colleague Mark Lindeman, here:

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

A critical mind, as you say, is about refusing to take things for granted. In fact, it was because I refused to take the explanation that the discrepancy was due to bias in the poll for granted that I found myself in a position to test both that hypothesis and the hypothesis that it may have been, at least partially, due to fraud. The fact that the data suported the first, and not the second is not due to any failure to take the second seriously. It's due to the data itself.

Elizabeth Liddle

[truedelphi]

80. Your statements are concise and well phrased

I find that when I discover something "new" about a topic, it has to do with
the curious side of my brain refusing to take some overriding premise as the central truth.

Sometimes this central truth is just a bunch of hooey! At least in the terms of investigating
the principle ingredient in the pesticide RoundUp (contains formaladehyde - how do I know? Well my body has a certain reaction to formaldehyde - and my body always responds that way to RoundUp. Then my scientific source Bob Simon came into my life. He revealed as a forensic expert that he saw papers in court that monsanto released proving Monsanto lied about the formula of ROundUp in order to get EPA licensing.
In looking into the vaccine issue, especially with regards to Sudden Infant Death Syndrome. (I am not finished working on this one).
In termsof the voting issue, I find it very "wobbly" as for instance, I just do not have the statistical background that would let me determine whether your data is or is not superior say to TruthIsAll's work.

[Febble]

82. You are certainly right

that central assumptions are often wrong. In fact, the thing that I "discovered", and which led to my contract to reanalyse the data, was fairly central - the metric that had been used to measure the discrepancy between poll and count at precinct level had a serious problem. It was called the "Within Precinct Discrepancy"(WPE), and was simply the arithmetical difference in percentage points between the margin between the leading candidates in the poll and the margin between the leading candidates in the count. The trouble with that is that where one candidate's voteshare is very much greater than the other, it underestimates the discrepancy. As more precincts had very high proportions of Kerry voters than Bush voters, it meant in particular that the discrepancy in Kerry precincts tended to be underestimated.

So I suggested an alternative.

The WPE wasn't exactly a bunch of hooey, but it did cast doubt on the inferences made from the data.

It doesn't make any difference to most of TIA's analyses though. The main differences between mine and TIA's is that TIA makes certain assumptions that I don't think are valid. Well, which are widely agreed within the polling profession to be invalid. It isn't really a question of statistics, it's more a question of survey methodology. In essence, TIA assumes that in a survey, the respondents are a perfect random sample - the kind of sample you'd get if you tossed a coin, or drew numbers out of a hat, or used a random number generator. Public opinion researchers don't make that assumption, because they know that their samples are not random, although they do their best to make them as representative as possible. But people are not numbers in a hat. They are more like rats in hat, and the smart ones may avoid being picked. And even when picked, they may escape before you've seen what number they have.

TIA's pre-election poll estimates also make assumptions about the way the "undecideds" would vote that are not the same as the assumptions made by the pollsters themselves. There is legitimate debate about the assumptions that are most likely to produce an accurate estimate (including whether "registered voter" or "likely voter" models should be used). TIA's assumptions may be correct. But again, they are mostly at odds with the assumptions of the pollsters themselves.

In short, if you share TIA's assumptions, you may come to his conclusions. There's nothing wrong with his math. But if you don't, you won't, and they are not assumptions shared by the polling profession. And to end on a statistical point - strictly, you should also put a probability value on your assumptions being correct. And if you are out on a limb, you should discount your own probabilities by the probability that you yourself are wrong.

We should all do that, in fact. Including Monsanto.

[Febble]

83. A response from TIA

TIA:
Febble, you continue to trash my "assumptions", but agree that the math is correct. Well, would you also agree that if my assumptions are essentially correct, then my conclusions must also be correct? Let's take a close look.
______________________________________________________________________

Example #1: The "Clincher" sensitivity analysis. Assumptions: 100% Bush 2000 voter turnout as opposed to 71-100% Gore voter turnout; Kerry's 51-59% share of those who did not vote in 2000.

<snip>

The assumption gives Bush the head-start advantage that 1) 100% of Bush 2000 voters turned out in 2004 and 2) Gore voter turnout ranges from 71-100%.

The Clincher is a 300-scenario sensitivity analysis (based on the NEP) of 2000 voter turnout vs. Kerry's assumed share of those who did not vote in 2000 (the NEP had it at 54-59%) depending on the timeline. You can't argue with the assumptions, because they show the utter implausibility of a Bush win by giving him the turnout advantage.

Kerry wins ALL plausible scenarios. In fact, assuming EQUAL 100% Bush/Gore voter turnout and a Kerry 57% share, the analysis shows that Kerry won by 6.73 million votes, a 52.28% share. With a 95% turnout, Kerry wins by 5.05 million, a 51.60% share.

Who woulda thunk it? Febble, have YOU ever done an analysis remotely like this?
Yes, is "Clinch".

___________________________________________________________________

Example #2: The Margin of Error assumption

I do NOT assume a PERFECT random sample and never said it. I DO assume that the MoE provided by pollsters is generally correct; otherwise the would not supply one with their surveys. Why do ALL posters use the simple formula:
MoE = 1/sqrt(n) if they do not believe it to be valid? Why would professional pollsters even refer to the MoE if they did not believe that it's a valid measure of 95% accuracy? Doesn't their expertise in sampling design enable them to obtain close approximation to a pure random sample?

Mitofsky said that the National Exit Poll was a "randomly-selected sample of voters as they exited the voting booth". He also said the NEP MoE was 1.0%. According to your logic, Mitofsky was wrong and the pollsters should all just quit the business, because the MoE is not a valid measure and polls are not true random samples. Really?

My analysis proves that the average of pre-election and the average of the exit polls, state and national, match to within 0.5%. In other words, the difference in the group average is well-within the expected MoE. For example, the typical 1000 sample national poll has a 3% MoE. The combined MoE of the 18-poll (total 18000 sample-size) group is 0.60%. No wonder the 18-poll group average is within 0.5% of the 13047 sample NEP.
________________________________________________________________________

Example #3: The average of 18 national pre-election polls was within 0.5% of the 12:22am NEP.

For you to claim the following does not constitute a match is pure hand-waving:

NATIONAL PRE-ELECTION 18-POLL WEIGHTED AVERAGE
POLL KERRY BUSH OTHER
AVG 47.80 47.14 1.0

Sensitivity analysis of 18 pre-election polls to UVA:
UVA = percent of undecided voters allocated to Kerry.
UVA KERRY BUSH
50% 49.83 49.17
55% 50.03 48.97
60% 50.24 48.76
67% 50.52 48.48
75% 50.85 48.16

NATIONAL EXIT POLL DEMOGRAPHICS
(12:22am, 13047 respondents)

CATEG KERRY BUSH
Gender 50.78 48.22
PartyID 50.69 47.50
Voted2k 51.41 47.62
_________________________________________________________

Example #4: The average of pre-election state polls was within 0.5% of the average 12:22am exit poll.


For the states, a typical pre-election 600 sample-size yields a 4% MoE. The corresponding exit poll MoE is 2-4%, depending on the number of respondents. The average difference between pre-election polls (adjusted for that undecided assumption) and the exit polls was also under 0.5%

For you to claim the following does not constitute a match is pure hand-waving:

1. There was a 0.37% deviation (assuming 60% undecided to Kerry) between
the average state pre-election poll and the average exit poll.
2. There was a 0.05% deviation (assuming 55% undecided to Kerry) between
the average state pre-election poll and the average exit poll.
3. In 39 states, one individual led BOTH the pre-election and exit poll.
4. In 15 of 17 battleground states, one individual led in BOTH polls.
5. In 15 states, pre-election vote shares differed from the corresponding
exit poll vote shares by less than 1%; in 29 states, by less than 2%; in 32
states, by less than 3%; in 42 states, by less than 4%.
______________________________________________________________________

Example #5: The assumption that undecideds break for the challenger

As far as undecideds are concerned, virtually ALL professional posters agree that the late undecided vote goes to the challenger by 2-1 or 3-1, especially when the incumbent is unpopular (Bush's rating was 48.5% on election day). So your statement about this assumption is totally erroneous. In fact, NOT allocating undecideds using this logic would be foolhardy.

For you to claim that undecided voters do not break to the challenger is pure hand-waving.

For example, if Kerry and Bush were tied at 47%, with 6% undecided (ignoring third parties) then it is perfectly reasonable to suggest that Kerry is actually leading the poll by 51-49%, assuming he will win 67% of the undecided.

I suggest you read up on what Harris, Zogby and others have to say about the undecided vote.

Here is what Harris had to say on election day about the late undecideds breaking for Kerry:
http://www.harrisinteractive.com/harris_poll/index...

________________________________________________________________

The Harris Poll® #87, November 2, 2004

Final Pre-election Harris Polls: Still Too Close to Call but Kerry Makes Modest Gains

The final Harris Polls show Senator John Kerry making modest gains at the very end of the campaign in an election that is still too close to call using telephone methods of polling.. At the same time, the final Harris Internet-based poll suggests that Kerry will win the White House today in a narrow victory.

Harris Interactive’s final online survey of 5,508 likely voters shows a three-point lead for Senator Kerry. The final Harris Interactive telephone survey of 1,509 likely voters shows a one-point lead for President Bush. Both surveys are based on interviews conducted between October 29, 2004 and November 1, 2004. The telephone survey is consistent with most of the other telephone polls, which show the race virtually tied.

Both surveys suggest that Kerry has been making some gains over the course of the past few days (see Harris Polls #83

http://www.harrisinteractive.com/harris_poll/index... and #78 http://www.harrisinteractive.com/harris_poll/index....

If this trend is real, then Kerry may actually do better than these numbers suggest. In the past, presidential challengers tend to do better against an incumbent President among the undecided voters during the last three days of the elections, and that appears to be the case here. The reason: undecided voters are more often voters who dislike the President but do not know the challenger well enough to make a decision. When they decide, they frequently split 2:1 to 4:1 for the challenger.

....

Three Key States (Florida, Pennsylvania, and Ohio)

Another piece of evidence pointing to a likely Kerry victory is that online Harris Polls in these large, key states, which may well determine the Electoral College result, all show modest Kerry leads. However, all these leads are within the possible sampling error for these surveys. Assuming the forecast is correct, Kerry is likely to win all three large states, and almost certainly the White House along with it. The sample sizes were well over 1,000 likely voters in Florida (1,433), Pennsylvania (1,204), and Ohio (1,218).

My brief response (but I won't continue hereafter with this inter-forum discussion):

1. TIA's analysis assumes that people did not over-report having voted for Bush in 2000, when in every exit poll since the seventies, the "past vote" question shows that people over-report having voted for the winner. There is no reason to believe 2004 was different; ergo, at least some of the "Bush2000" voters were likely, in actuality, to have been Gore voters. This fundamentally alters the math.

2. Mitofsky himself said, in an address to NYAAPOR, that he supported giving MoE figures because he disagreed with the view that it would mislead "poor mindless souls" into the mistaken view "that sampling error was the only error in the survey". Not only that, but if TIA looks at his own pre-election poll data he will observe that the between-poll variance is greater than could arise from sampling error alone. Several polls are well outside the MoE of other polls. This must mean that at least some of the polls have non-sampling error. This is not particularly surprising. Even if the selection of respondents is truly random (which is unlikely) unless the response rate is 100%, the resulting sample will not be random at all. It will be determined by whether or not each selected respondent chooses to respond. The exit poll response mean response rate was 53%. Polls are, in fact, not true random samples. Really.

3. I do not dispute that the mean value of TIA's pre-election polls matches the exit poll, given his UVA assumptions, although the MoE will be wide, given the between-poll variance. However, even taking into account the between-poll variance, I agree that the pre-election polls, pooled, were closer to the exit polls than to the counted result. What I dispute is that this is a valid procedure. The between-poll variance tells us (see 2 above) that some of the polls were biased. We do not know which, and there is no reason to suppose that the mean bias was zero. The official result was within the MoE of several of the polls. These could have been the biased ones. On the other hand, they could have been the correct ones. But math will not tell us which.

4. TIA is of course correct that in both the exit polls and the pre-election polls (on his assumptions, which I dispute, see above) the deviation of the count from the poll was in the "red" direction. However, at state level, the extent to which each pre-election poll deviates from the count is completely unmatched by the extent to which each exit poll deviates from the count. There is simply no correlation between the two sets of deviations. I invite TIA to consider my post #50.

5. It is probably true that in most elections undecideds break for the challenger. I hoped it would be the case in 2004. But it is certainly not a hard and fast rule. Nick Panagakis analysed 155 elections, and found that in 82% the undecideds broke for the challenger, in 6% they broke equally, and in 12% they broke for the incumbent. It was cause for optimisim back in October 2004, but it still left a worrying 1 in 8 chance that they would break for Bush.

The pre-election polls, particularly in the last week, certainly had me hoping. I was following Sam Wang's site with bitten nails. And if the degree, at state level, to which the count was "redder" than the final polls had matched the degree to which count was "redder" than the exit polls, there would be grounds for suspicion. In any case, I agree there were grounds for suspicion. It's why I spent so much time trying to figure out what might have happened. But TIA's insistence that the only error in polls is sampling error, when his own data, not to mention a vast literature on the subject, show that this is not the case, and his insistence on ignoring strong evidence that a significant proportion of voters for the loser in one election tend to report having voted for the winner by the time the next election comes around (an effect found in EVERY SINGLE EXIT POLL SINCE 1976) casts his inference into doubt. In contrast, the "undecideds break for the challenger" rule admits of relatively frequent exceptions (1 in 8 according to Nick Panagakis).

I rest my case.

Seeya, big guy. Keep well.

Lizzie

[OnTheOtherHand]

86. the man does know a thing or two about "pure hand-waving"

I wonder whether he has a list of those "virtually ALL professional pollsters" who agree that undecideds break 2:1 or more to the challenger. I see something like 15 national polls in the last week, so, being a reasonable person, I would settle for any twelve.

(He had better not put Harris on the list, at least based on this evidence. Harris says that undecideds "frequently" break 2:1 or more to the challenger, but I see no sign that Harris actually allocates undecideds that way.)

[truedelphi]

98. A comment on one of TIA's assumptions

Febble says that people who vote for a loser in one election will vote for a winner in a second election

Well, possibly. But here are several caveats to that below:

One never before had the "winner" also be someone who was considered a winner by theft (or as some have put it by "Judicial Coup")

But many people who voted for Gore did not, especially after a powerful experience of attending the Michael Moore film (Seen by the millions in Ohio just like everywhere else) consider Gore to be the loser. They considered him to be someone who had the election stolen from him.

Some people may of course end up not liking Gore as he was the victim in the theft; others like me would still vote Democratic, considering that the what Republicans did in Florida 2000 was something so heinous that they had not only vicitimized one man but had vicitimized a nation

Other things to consider: This "winner" was caught in countless lies. He also had brutally savaged a nation in a war that he was losing on election day. A war that Vietnam era folks were rapidly concluding was as much a mistake as the previous war. A budget that was drowning in red ink. He was not likeable - could not speak well, acted funny. His school policies were not popular especially among young families. (The Nickelodean Children's poll showed Kerry over Bush 59% to 40%.)

Also Gore and Kerry followed in the tradition of Bill CLinton. Had Clinton been able to run in 2000 several polls had indicated that Americans would proceed to elect him. Only two candidates in history might have had the "Clinton" brush off effect to rattle the voting experience. Those two being Al Gore and John Kerry (At age 88, My father,a lifelong Republican, voted for Gore in 2000, resulting in MY Mom not speaking to him for two days. He could not even bring himself to tell me that he voted for Gore for three full months. I asked "Why." He said, "Clinton showed me that the Democrats could be smart in ways that matter just as Republicans *used* to be." I should add that when my dad died July 2002 he was as mentally sharp as ever. The Republicans did not lose him because his mind had gone.

In these remarks I am considering the one area of assumption only. I have not yet digested the other areas.

[Febble]

99. Oh dear, I seem to be having difficulty

in making myself clear on this point.

What I said was that a small but significant proportion people who vote for the loser in one election will,when asked about it four years later say they voted for the winner. In other words, in the 2004 exit poll, some of the people who said they had voted for Bush in 2000, probably actually voted for Gore. Mark Lindeman has demonstrated this with reference to at least two sets of data.

The relevance of this is simply to TIA's argument about the proportion of defectors. He assumes that only those who actually reported having voted for Gore in 2000 did so. Lindeman has argued, persuasively, in my view, that this is unlikely. And I quite agree that a huge swathe of people rightly did not consider Gore the loser in 2000. I expect those people strongly and correctly recalled voting for him, and not having their choice, which they shared with a majority of their fellow voters, translated into a President Gore (but I am hopeful we may yet see President Gore....)

And, like you, I find it incomprehensible that Bush won. But "argument from incredulity" is not sufficient - and I don't find the evidence credible that a majority voted for Kerry (although my mind is not fixed regarding Ohio, which would have given Kerry the presidency, of course).

And you have a point (and I believe TIA has made the same point) that the "rule" about a proportion of people forgetting having voted for the loser may not apply in the case of a candidate who actually won. Against this is the evidence that people even misrecalled having voted for Nixon - suggesting that for those poor benighted souls who aren't particularly interested in politics (and yes, there are quite a few) name familiarity may be the crucial factor. And one of Lindeman's data sources was actual longitudinal data on a panel of people whose recall of their vote in 2000 was asked for later - and, indeed, significantly more people misrecalled having voted for Gore (thought they'd voted for Bush) than the reverse (although there were some of those too)

The point, therefore, is, like most things about the exit poll, that the data points are not as solid as they look. A biased sample is possible; the responses people give are not necessarily reliable, especially about an even that took place four years previously. And so, I would argue, TIA's defection rates are not the "clinching" argument for the accuracy of the exit poll data that he claims. That was my point.

[truedelphi]

103. Ah, my brain fell for the same old 25% more chicken

end of the argument I mentioned in our PM's... And how I disparaged people who
fall for that kinda thing

Sorry about that Febble. (But what can you expect from a woman who spent last night
inventing a web site entitled "CAtFarts - The air freshener for felines)

[Febble]

104. I have to have

that URL!

[Pachamama]

39. Why???

:shrug: Why would they prevent TIA from posting? TIA has so much amazing information and perspective that we should all be able to access and to think about and contemplate.

So I ask again: Why?

[Vidar]

41. He's dismissed as a conspiracy theorist. Never mind six years.

of one proven conspiracy after another intersperced with random evil & depredatations. Kos is an ass.

[Pachamama]

42. Thanks for the explanation....

Guess I'm a Conspiracy theorist too because I have been suggesting things to my family and friends for years, only to be later having them come true.

I'm proud to be a conspiracy theorist...I think a better term is a "seeker and seer of truth"....

Thank you to all my fellow Seekers and Seers of Truth like TIA! :grouphug:

[Vidar]

57. Entirely welcome. I agree.

[kster]

43. I'm not surprised

when someone starts talking common sense, that no one can deny, the only thing left to do is ban them.

TIA, ROCK THE BOAT, GET BANNED, MAKE WAVES, MAKE NOISE, BE A THORN IN THEIR SIDES, :yourock:

THEY CANNOT BAN THE TRUTH!!!

Future America (our kids) depend on people like you, to keep spreading the truth. :patriot:

[autorank]

44. Think I have a career at KOS...autodiaretics

This is a fun thread:

http://tinyurl.com/mrntm

[Kurovski]

45. I wonder how long JFK Jr. would last as a diarist at DK? (nt)

TIA deserves better. At the very least he's owed an explanation.

[New Earth]

58. haha yeah

let's see them ban JFK, Jr. :eyes:

[bleever]

60. You meant RFK Jr., right?

It's a very easy mistake to make, like calling Fitzgerald "Fitzpatrick". I've done it myself.

[Kurovski]

62. Yes, I do.

And thank you. Would I get banned at DK for that faux pas?:-)

[New Earth]

76. lol oops

i meant RFK Jr. too :blush:

[PATRICK]

56. The continuing power

of the Bush Lost belief, spreading nationwide with very little evidence savvy going for it, is probably based on the following real premises:

Successful, shameless and utterly self-evident vote suppression that even many GOP people believe happens but is OK because it helps them win.

And unsophisticated but absolutely correct apprehension that invisible ballots and e-counting is untrustworthy.

In both these beliefs there is an extreme amount of common sense and practical evidence despite the utter
failure of the establishment of anything in this nation to come to grips with. Any local experience of suspicion and visible failings spreads by word of mouth at least, especially without any paper record or recount and a great deal of suppression and rationalization why one CAN'T do it. The courts, hands tied by the self-protective legislation of incumbents, resorts to cost benefits and its own set of faiths to avoid confrontation or contests with the "law" and the "results" and the "good of the community".

The point about all of the above is that none of them contain evidence that materially change the vote tally. Suppressed votes are simply non-votes unless a successful grievance for negated elections results can be won in court. E-votes are simply transformed codes and even if they vary in re-running the machines it is a self contained unverifiable system based on the trust that the position of the finger corresponds exactly to every ballot counted.

In the arguments somehow the major common sense points are lost along with the common evidence wider than ballot tampering itself that the GOP intended to do everything possible to fix the election, the campaign, the spin and the interpretation of any and all controversies related thereunto. This is not protecting one's legitimate election from false practices and mis-perceptions but creating those things wholesale to pro-actively set up a myth of "victory", sacrificing verifiability in advance for image.

I have long maintained that so-called conspiracy theories are legitimate when the darkness descends on all evidence because of malice and aforethought. Not enough is done however to avoid the Catch-22 of being swallowed up in the same secrecy trap by not keeping the focus on the need to end the secrecy(get proper, verified elections). Catch-22 or the circular logic that FIRST you must prove an unverifiable election that purposefully excludes voters in order to show the need for reform is followed swiftly by the OUTRAGE(the last step of the con) when the defeat of common sense questioning by victims is followed by "reforms" presented by the winning side to make matters even worse. The the INSULT, to prove that stupidity is fixed in stone is to turn around and manipulatively overthrow the foreign Ukrainian elections using many of the same arguments denied to suppressed voters in our own country.

Insults and outrages and abuses don't prove that Kerry won. To get annoyed and help the Bush team with their unscrupulous myths is very strange- to compare stiff and cautious Dem campaign errors with relentless MSM Bush pimping and tolerance for smears and suppressing unfavorable news items for Herr Bush. It all has the appearance of victims, some laughably ignorant, others frustrated and over-stating, fighting each other after the dominance sniffing has shown them the GOP is out of bounds as a target.

So forgive me if I simply join the mob and posit from prejudice at having faith in my fellow citizens and an instinct for having been screwed over if I ask simple questions like:

When polls appear in error and pollsters are not by definition election verifiers, is there not a consensus that future samplings and methods be re-weighted to project in the future closer adhesion to the results? Do parties commission both optimistic(publicly useful) biases in "private polling" and the real deal polling that betrays itself on grim faces during election day? or do they just log onto Gallup and the others and do averaging? Is this just about Ohio(Florida and Georgia have long been surrendered as lost to GOP tactics) in 2004? Several other states have their own question marks that cannot be resolved, believe what you will.

The main issue was revealed in 2000 that Dems win despite the hook and crook and removal of the fraud would simply boost their chances, besides serving democracy. Instead the cons spun, performed the outrage and insult and wrote even worse "reforms" to improve their undemocratic chances and then, not so surprisingly, did nothing to serve the vast majority or the nation at all. They even found a 'war" substitute to further sideline both their simple duty and populist chances to win hearts and minds with the power they chose to steal.

The elections as such are viscerally untrustworthy and weighted for the GOP, and not as once we naively believed because of poor Dem message and money disadvantage alone. Regardless of the spats between people arguing over election results and reality, the legitimacy of the government and democracy are draining out of the system at the grass root level and will continue to do so until the revealed abuses are obliterated and hopefully brought to justice.

The obfuscation of the figures and the plot graphs mean nothing at this point. We argued that the instinctive reaction to Ohio's failed ballot initiatives could be very wrong, but the perception, minus any ability of polling or recount to settle, has grown simply because the appearance of cheating has helped the public awareness of the Great Fraud simply to see confirmation of a prejudice ironically born out by evidence and distrust despite massive mis-dis-un-information and top-down surrender.

So one would be left, were not the public distrust becoming an emotional complex that no longer needs much evidence in thinking that brainless doltish American voters swayed by the machine and turned off by inept Dem leaders elected George Bush(the dark view of humankind) or that the majority not only resisted the crap but voted almost always against it but were defrauded at every turn. Naturally I want to believe TIA. Most people, regardless of the ephemeral complexities now would. In fact, the darkness of human history would be enlightened had we the ability to go back and see if people basically did not choose monsters and fraud is the consistent thread of tyranny when all stupid popular controlled forums have exhausted their limits. If we believe Bush "won" we believe monsters and the worst represent humanity and there is little hope. If we say Bush didn't- and all that was needed was fair evidence- then we deny our worst side. I believe in the challenge of hope and if fair proof and practices are denied why should I believe in THEM- the killers?

Given firm loss of popular belief, eventually tyranny may be overthrown with extreme prejudice. But that this time, perhaps, a better understanding would really instigate reforms with enforcement to establish democracy as something more than a childish myth for the privileged to manipulate.

[truedelphi]

77. Sorry this happened to you TruthIsAll

But Kos is a gatekeeper.

If you try to work out things with the gatekeepers, you do not get to far

Could you move all of your fine work that is now sitting over there at Kos
over to this discussion forum?

It is excellent work and there is no guarantee that it will be kept up on
Kos for long.

Carol

[jeffuppy]

81. It's a shame about Kos

Egos run amok are an ugly thing. The proper response to speech that you disgree with - or that you know for a fact to be wrong - is speech of your own in disagreement or correction. Banning and purging and deleting are authoritarian tactics that only serve to bury the truth in the end. They erode trust. Why even waste your time in that self absorbed echo chamber at Kos to begin with?

[Vidar]

107. Agreed. Welcome to DU.

[Name removed]

88. Deleted message

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

[OnTheOtherHand]

89. well, I have yet to find anyone here

who can actually defend TIA's arguments in detail. I do draw inferences from that fact, but they could be mistaken.

[OnTheOtherHand]

91. by the way

lest the last sound too dark -- what I infer is that TIA's arguments aren't very good, but people really want to believe them anyway. That doesn't bother me so much. When people start making stuff up about other people who disagree (academics, bloggers, politicians, whoever), that does bother me.

[truedelphi]

92. I don't think it is ust that people want to

Blindingly believe something, but it is a difficult thing to fathom.
How would the average person know which methodology is better when it comes to analyzing the
exit polls, how would the average person know how to determine whether or
not the statistical analysis was done correctly?

I think it is good that there are two disparate sets of analyses - Febbles' and TIA's.

I want to spend time this week thinking about them both. I find it hard if not possible
to set aside my belief that fraud was committed in Ohio - Andy Stephenson got to spend
just a short bit of time there, but he had observed personally a technician screwing around with the machine in the precinct where he went in (I think during the time of the recount) Why would they need to have the technician screw around - if they had not done something originally that needed to be coverered up? (If memory serves he was alerted to the occurrence of the technician being there by someone who had been a poll watcher and observer for the Nov 2nd election day event.
In other words, it was not just his perception, but the pereption of a second person who was more familiar with what should and should not be happening inside that precinct.)

[OnTheOtherHand]

93. I'm talking about the people who already Know

Honest confusion is totally appropriate, and I think everyone should be honestly confused about some aspect of the 2004 election. And we don't have to know all the answers in order to know that some things have to change.

You might be genuinely shocked at some of the things that people have written about Febble. I was.

It's totally possible that fraud was committed in Ohio (I know Febble agrees). As I wrote somewhere else on this thread, there aren't enough exit poll data from Ohio to say much one way or another. There are reasons why we aren't convinced that enough vote-switching fraud went down to change the outcome, but in some ways that is beside the point. The machines are way too vulnerable -- not knowing that they were hacked (even if we're that lucky) isn't nearly good enough.

(Of course, what happened in Ohio isn't just a matter of outright fraud. For instance, we may not be sure why the machines were brutally misallocated in Franklin County, but we know that lots of people didn't get to vote because of it.)

What really troubles me is the short distance that some folks travel from "Kerry won the election" to "Febble, Kos, and the Democratic National Committee are covering up the truth." If people understand the reasons for being skeptical that Kerry won by millions of votes, they may still believe that he did, but they are less likely to flame out on actual or potential allies who can't go there with them.

[WillYourVoteBCounted]

105. many had to hold their noses

I was just talking with an old friend who lives in a different state
than me.

We were reminiscing about our decision to vote for Kerry in Nov 2004.
We discussed how repugnant the idea was, but that we would hold our noses
and vote for him.

IF we as Die Hard Democrats had to hold our noses to vote for Kerry,
what do you think might have happened to non internet educated
and less involved but registered democrats?

Not saying the election was not rigged, but saying that not all dems
are as politically educated.

And, I myself didn't have faith in the results of the New Hampshire
Democratic Primary in 04.

ITs not that the Ohio Election wasn't rigged, its that we can't prove it.

But anyway, I am focusing on now - because we can never prove it either way -

We can't prove that the election was rigged,
We can't prove that the election was NOT rigged.

[OnTheOtherHand]

106. Right, and we can't prove that the next one won't be. n/t

[Febble]

100. Well, I certainly haven't set aside

my belief that fraud was committed in Ohio. I don't think we have evidence that it wasn't, and I don't think we have any reason to think it wasn't, given what we know.

But I don't think the exit poll data are evidence that it was.

[Febble]

90. Paid?

Would you like to be more specific? Or, alternatively, edit.

[kster]

97. Oh come on Lighten up......nt

[kster]

94. The never ending debate, just where the politicians

want us, stuck at the exit poll controversy, The voters weren't allowed to see the votes being counted in 2000,2002,2004 and still now in 2006.

An 8 yr old would know what would happen if the votes were BEING counted in secret, its a NO BRAINER.

Where is their evidence that the exit polls were accurate, they don't have any, where is our evidence that the exit polls were not accurate, we don't have any, WHY? Because we are not allowed to see how the votes were or are being counted. The debate is never ending and thats where the politicians want us. They want us to Forget the secret vote counting machines and keep debating the exit polls, its cover for them.

Politicians, where do I go to see the votes being counted, thats what I would like to know, if there is not a place to go to view the votes being counted, HOW DO I KNOW THE VOTES ARE EVEN BEING COUNTED?

[OnTheOtherHand]

95. dunno about the politicians, but you have a point

There are lots of folks (Republicans and Democrats) who will have a hard time believing that Kerry won by millions of votes, but will have an easy time believing that the machines aren't trustworthy.

This is probably why Avi Rubin sounds so annoyed at folks who get hung up on the first argument. Here's that quotation from Brave New Ballot again:

The many activists who definitively claimed, with no evidence, that the election had been stolen did a great disservice to everyone who had thoughtfully and seriously criticized the e-voting technology.

I think people should wrestle with what Rubin is saying here, even if it came out a bit tart. (See my other comments at #36.)

[kster]

96. Neither side of the exit debate has evidence to back up their claim, why?

because the Politicians "Dem's and Repubs" want it that way or better yet, made it that way.

I don't agree that the debate was a disservice, because the debate opened up a whole lot of eyes.

Now its time to DEMAND to see the votes being counted, How long can the Politicians D&R remain silent about the fact that the votes are being counted in SECRET?

[Febble]

101. I think you'll find, kster

that I have never ever disagreed with you on the need for transparency in American elections. I have been advocating the need for transparency for a couple of years now, and indeed, it was the lack of transparency that led to my involvement in this debate in the first place.

I also agree that the debate has been a useful one, because as you say, it opened up a whole lot of eyes. After the election, I didn't know about TIA's work, but I did read Steve Freeman's paper on the Unexplained Exit Poll Discrepancy, and it intrigued me. The discrepancy intrigued a lot of people, and history may record that it was what kick-started the process of Election Reform, although I would also give credit to those who were pointing out the vulnerabilities of electronic voting well before the election. And to those political scientists who did the meticulous work that demonstrated that had the election system been fair, the SCOTUS would never even have been involved in Florida 2000 - and Gore would simply have been declared the winner in the usual way.

And I also think it was worth investigating the exit poll evidence. However, based on what I know know (and what is also in the public domain) I don't think the exit polls demonstrate that Kerry won, and I do think the argument that they do has the potential to damage the case for Election Reform, simply because I don't think the argument holds water, and bad arguments don't help a good case.

As for your "lighten up" comment - perhaps. But I have had my integrity called into question too often on this board to take kindly to smears. I suspect you would feel the same in my position. I welcome debate, even disagreement. But I don't welcome innuendos about my motives.

And FWIW - what opened my own eyes, was not electronic voting, or the exit poll story, but the footage of those voters in Columbus standing in the rain for several hours to vote. It broke my heart.

[Blue Shark]

108. Febble I have asked this before...

...why not turn your brilliant analytical skills to things other than the exit polls.

...You know, the number of new registrants to the democratic party, the margin of Kerry voters in first-time voters as opposed to Rove's secret army of which no proof ever existed, the disenchantment with Bush even among Repbulicans, the numeric advantages in nearly every arena that never translated into reported votes for Kerry, the 110% turnouts in Bush Country, and the 23% turnouts in Kerry country all used to argue that GWB got 3.2 million more votes than Kerry?...and on and on ad nauseum? How about the farcical Ohio (non)Recount?

...This week's undetectable Princeton Diebold hack which put Benedict Arnold into power by a 3-2 vote when the actual vote was 4-1 for George Washington shows absolutely what happened. Votes were electronically flipped. TIAs most important analysis had nothing to do with exit polls which is still the venue where you argue exclusively. His most important analysis is those 20 things OTHER THAN EXIT POLLS that you had to believe to believe GWB won.

...How about addressing those for a change?

[Febble]

109. Fair questions

(well, I don't claim any brilliance, just training in data analysis) - I did in fact do some work on machine allocation in Franklin County, Ohio, and on undervotes in New Mexico, in case you are interested - but they don't seem to be so controversial.

And I am also happy to do any number crunching if people want to send me stuff, although I am a bit busy with some other stuff right now. But the reason I have posted on exit polls is twofold; one is that having found myself in a unique position with regard to the data, I think people are entitled to know what I found; the second is that I do NOT think that the exit polls reveal (or even hide) evidence of widespread fraud, and I think that arguments based on them are not only mostly wrong (though that wouldn't matter in itself) but potentially damaging, for the simple reason that I think that bad arguments for a good case can bring the case into disrepute. Actually, I will go one further than that - I think the exit poll evidence, if anything, suggests that vote-switching in 2004 was not on a scale of millions, and that although Kerry may well have won Ohio on a level playing field, I think the evidence is scant that he was the intended choice of a majority of voters.

It's interesting that you think that TIA's most important analysis was not the exit polls but the electronic vote-flipping. I agree. The trouble is that even there, I am not convinced by his probability analysis - you can't apply the binomial theorem to a sample that isn't random, and there is no guarantee that the sample is random. In fact there is no guarantee it was a sample at all - it could have been the entire "population" of incidents. I agree that the vast majority of those 100 odd reported vote flip incidents were Kerry-to-Bush. But to know whether that was a real majority you would have to know whether a) any of those incidents related to the same machine or precinct and b) whether the EIRS reporting number was known equally in Democratic and Republican precincts. But it is certainly enough to make one worry. Against it, is my own exit poll finding that the exit poll discrepancy in urban areas was greater not in precincts with DREs but in precincts with levers or punchcards. And when I looked at rural and suburban areas there was no significant difference in exit poll discrepancy between any of the voting technologies, including paper.

As for the rest of TIA's list, I don't have a link handy. What would be helpful,if you are really interested, is if you would choose the ones that you found most convincing. and I will tell you whether or not I share your conviction.

And I certainly share your conviction that electronic paperless voting is absurdly insecure, and that, even if 2004 wasn't massively hacked, there is far too great a chance that the next election may be. And even if it isn't - how would we know? A democracy that doesn't know whether it is a democracy or not is not a democracy.

Cheers

Lizzie

on edit To be honest, my heart's not really in fighting with TIA. The guy did a great job of energising the Election Reform activities on this forum and elsewhere. Sure, I disagree with him that the exit poll evidence shows that Kerry won the popular vote, and sure I take issue with some of his probability calculations. And sure he and I had our share of fights. But I think it's time for me to let him get his blood pressure down! We both agree that the both the 2000 and the 2004 were a disgrace to democracy. We only differ in our estimates as to the magnitude of the corruption. And he's treated me with respect. I return that respect, and wish him a serene recovery.

[kster]

111. I'd like to see your evidence, to all these claims...nt

[Febble]

112. Well I didn't claim much

in that post.

What I did claim is that I consider that the exit poll data if anything contraindicates widespread, massive, vote-switching fraud.

The evidence is in the data analysis here:

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

And you might also like to check my journal entries here:

http://journals.democraticunderground.com/Febble/8 (from this thread)

And here:

http://journals.democraticunderground.com/Febble/3

And take a look at this plot:



The plot shows, on the vertical axis, how well Bush did relative to 2000 in each precinct. On the horizontal axis is how far off the exit poll was in that precinct.

As you can see, there are more precincts above the zero line than below, because more precincts "swung" towards Bush than away. And you can also see that there are more precincts to the right of zero than to the left. That is because more precincts had "redshift" (count was "redder" than the poll) than "blue shift" (count was "bluer" - more in Kerry's favour- than the poll). The units on the horizontal axis are "z scores". That tells you much further "off" the exit poll was than you'd expect by chance. Any z score higher than 2 or lower than -2 has a 5% probability of having occurred simply by chance, assuming a random sample of voters. And as you can see, a large number of precincts have scores outside these values (more than 5% of them) and more of these are on the "redshift" side than the "blueshift" side. This means that the overall "redshift" could not have occurred by chance.

However, as you can also see, there is absolutely no tendency for Bush to do generally better, relative to 2000, in precincts where the exit poll had greater "redshift" or worse where the exit poll had "blueshift". In other words, "redshift" in the poll does not seem to have been associated with benefit to Bush - and nor was "blueshift" associated with benefit to Kerry. In other words, if fraud was causing the "redshift" to Bush (or even if pro-Kerry fraud was causing "blueshift"), it isn't reflected in his performance relative to 2000. From this, therefore, it seems more likely that what was causing some precincts to have redshift and some to have blueshift was non-random sampling, and that more of that non-randomness was favouring Kerry than Bush. Other evidence from within the data suggests the same thing - redshift was strongly correlated with methodological factors that would have made strict random sampling more difficult, and thus allowed any underlying tendency for Bush voters to be less keen to participate than Kerry voters to be translated into a biased sample.

In addition, there was NO tendency for redshift to be greater where DREs were used, or optical scanners, then where paper was used, when the type of community served was taken into account. In fact there WAS a tendency for redshift to be less in precincts using DREs and optical scanners than in precincts using punchcards or levers.

Added to that, the greatest redshift was in the "bluest" states - like New York, for example. New York votes entirely on levers, and the counted result was close to pre-election polls. Nonetheless, the "redshift" in New York was one of the largest in the country - and considerably larger than Ohio. This evidence alone suggests that a subtantial proportion of redshift was due to something other than fraud. Coupled with the other evidence I've cited, it doesn't look to me that there is any actual evidence that any of it was due to fraud, although some of it might have been. However, given that massive widespread fraud WOULD have shown up as redshift, it is hard to postulate any massive fraud mechanism that would not also have shown up as a positive relationship between Bush's performance and the degree of redshift in the poll. Eomer, however continues to try to think of one! I applaud his efforts.

But I'll add, as I always add: none of this means there was NO FRAUD IN 2004; none of it means that 2004 WAS A FAIR DEMOCRATIC ELECTION, and none of it means that there aren't serious reasons to worry that future election hacks may take place. It is one of the reasons I think it is so urgent that paperless voting machines are made illegal. There HAS to be some independent way of checking that the electronic vote count matches what was cast. That Princeton demonstration is chilling.

Let me know if you have any questions

Lizzie

edited for typos

[eomer]

113. Well, I do have a few ways that massive fraud would not show up

as swing/redshift correlation, but I don't think any one of them really works as an explanation by itself.

Scenario #1 - Avoidance of exit poll precincts
One way is if they avoided exit poll precincts when committing fraud.

The likelihood of this first scenario is doubtful because of the degree of control over the fraud mechanism that would be required. They would have to either somehow obtain knowledge of the secret exit poll precinct list in advance or else find out at the last moment and then control the fraud to avoid them. For old technology fraud like punchcard stuffing or opscan ballot tampering this may have been easy enough but for DRE or tabulator fraud some kind of control mechanism would be needed. Maybe they could pull off such a feat in one or even a few states but it is hard to believe they could pull it off across almost all of them.

Another point about this scenario is that even if it is true it still would not be the explanation of the exit poll discrepancy. Since this scenario is based on fraud having avoided the exit poll precincts then fraud cannot be the explanation of redshift in those precincts.

Scenario #2 - Uniformity over time
Another way is if fraud occurred in 2000 in pretty much the same amount and pattern (distribution over precincts) as it did in 2004. Swing would then have no measure of fraud in it while redshift would and there would be no reason to expect that swing and redshift would correlate.

This second scenario seems to be ruled out by the fact that redshift in 2000 and 2004 had different patterns. If it were caused by fraud that was much the same in both years then it (redshift) should also be the same in both years -- but it wasn't, at least at the state level.

Scenario #3 - Uniformity over geography
If fraud were uniform across all the exit poll precincts -- say 3% everywhere -- then it would not show up as swing/redshift correlation.

This scenario seems unlikely for the same reason that the first one does -- the degree of control that would be required to accomplish it.

Scenario #4 - Smorgasbord
Another scenario is that of a combination of various factors -- some of the above plus some unknown ones. It could be the case that some of the exit poll precincts were avoided, that some of the precincts had fraud in 2000 that was the same as in 2004, and that some of the precincts had a fairly uniform percentage of fraud. For any given precinct, if it was covered by just one of those three factors then it would slip by the correlation test. You would need enough, but not all, of them to slip by so that you would not conclude there was a significant correlation.

Scenario #5 - Morass
This scenario builds on the previous one by adding hidden relationships among the various factors. For example, if fraud was intentionally committed mainly when swing wasn't going the way they wanted, if fraud took the form of intentionally duplicating or approximating the vote count from 2000, if fraud was intentionally uniform in exit poll precincts, and so on.

The last two scenarios are difficult to assess without doing a bunch of modeling and, even if you did, I'm going to guess that assessing the morass hypothesis could be challenging. I wouldn't be inclined to spend a lot of time on it because personally I think this whole line of investigation is not the most important point anyway. To me the question of whether there was massive fraud of millions of votes that gave Bush a popular vote mandate is much less important than the question of whether fraud in the tens of thousands of votes were one component (along with vote suppression and other tricks) of stealing the electoral college result. I understand why you address this question repeatedly -- it is because the argument is made repeatedly and you only respond to the argument when it is made. What I want to do every time the massive fraud argument is made and then disputed by you is to argue that the more important point is whether the electoral college result was stolen. The massive fraud argument is a bit like two (American) football fans arguing over which team had more total yards gained running and passing. That may be interesting but is much less important than the question of which team put more points on the scoreboard. The popular vote is like total yards gained. The electoral college is points on the scoreboard.

Summary
So here's how I sum up the question of whether there was massive fraud:

1. I haven't seen evidence that makes a clear and compelling case that there was.
   
2. I haven't been convinced, on the other hand, that it is clearly and indisputably ruled out.
   
3. If there was massive fraud, the exit poll discrepancy appears to be orthogonal to it rather than an index of it and massive fraud is therefore not a sufficient explanation by itself of the exit poll discrepancy.
   
4. Conversely, but for the same reason, the exit poll discrepancy by itself is not a sufficient argument to prove massive fraud.
   
5. Whether there was or wasn't massive fraud isn't, to me, the most important point.
   
6. What is the most important point is whether the electoral college result was stolen. I know it was in 2000 in Florida and I think it probably was in 2004 in Ohio.
   
   

[Febble]

114. Bravo!

I think I agree with every word of this, give or take a hunch or two.

I think Scenario 1 is also contra-indicated by the good match between the results in exit poll precincts in each state and the state results as a whole. If fraud had been targetted at non NEP precincts, they would tend to be "redder" than the NEP precincts, and they weren't. The mean difference was insignificantly different from zero, and the variance was small.

Re scenario 2: although I didn't have redshift data at precinct level for 2000, I did covary for state mean WPE, and it didn't make any difference. In any case state mean WPEs were small in 2000.

Scenario 4 is my favorite, I think. OTOH modelled 5 a bit (and so did I) and you need a heck of a lot of assumptions re control, as you say. Scenario 4 fits the bill best, IMO, partly because I find opportunistic piecemeal fraud inherently more plausible than a co-ordinated effort to fool the data analysts. And I agree that the EV result is up for grabs. I'm possibly more skeptical than you are that Ohio didn't go the way its voters intended, just because so far I haven't been convinced by the numbers (the slam dunks don't look like slam dunks to me), but it wouldn't surprise me if eventually the numbers added up (although some are incalculable). I'm sure Ohio wasn't clean. And I find it hard to believe that Florida could possibly have been clean in the circs. (I think that NM quite probably should have gone to Kerry, but that wouldn't have been enough for the EV).

The value (as I see it) in forming some decent probabilistic estimate as to the magnitude of likely theft in 2004 is threefold (I think)

1. While we know the stable door is unlocked, it is worth knowing whether in fact the horse bolted - I think it didn't (or not the whole herd), but I'm not sure the stable door is going to be locked in time to prevent a future escape bid. However, if I'm right, it does mean there is a sporting chance of spotting it when it happens (i.e. it is unlikely to be invisible to data analysis).
   
2. Again, if I'm right, it suggests that efforts to bring the culprits of 2004 to justice might be best focussed on voter suppression efforts, and specifically anomalous cases/counties, rather than on a particular manufacturer, or on trying to find a few key hackers.
   
3. I think it is worth knowing which arguments are watertight and which aren't. Even if I'm wrong about the exit poll data, what I am certainly right about is that the data aren't unambiguous evidence of massive theft. Asserting that they are is, IMO, a potentially counter-productive tactic, and I would really like to see what I see as the twin evils of 2004 - insecure and non-transparent voting systems and voter suppression - decoupled from the case that Kerry won even the EV, let alone the popular vote. Simply because it is handing the opponents of election reform a ready made straw man.
   

But this is a fantastic post. Can you add it to your journal so it can be easily found and linked to?

[OnTheOtherHand]

116. thanks for that carefully argued post

I really agree with just about everything you said. Even what I disagree with, I don't really disagree with. You think Ohio 2004 was "probably" stolen, I don't see it yet, but we agree about so much with respect to Ohio that it hardly seems like a disagreement. Bottom line: we don't know.

I will add a perhaps tangential thought: In principle, the question whether Bush stole millions and millions of votes (among people who thought they had voted) is more consequential than "which team had more total yards gained." If Bush stole at least half a million votes in New York alone, as a resident of New York I would actually like to know that. (I hope even some non-residents would share my interest.) So I do think that the massive fraud argument is important in that respect. And, well, you probably know what my next two or three points in that vein would be.

That said, I absolutely agree: whether there was or wasn't massive fraud isn't the most important point. The legitimacy or illegitimacy of Bush's election doesn't hinge on the massive fraud debate. The need for a slew of election reforms doesn't hinge on the massive fraud debate.

[eomer]

128. You're right -- if millions of votes were stolen then I also would like to

know it.

Both massive fraud and focused fraud that stole the Presidency are important.

My concern is that the massive fraud argument, which I personally am not convinced by at least at present, could end up acting like a strawman. I know it's not really a strawman (if that double-negative way of putting it makes any sense) because it has real proponents who argue it persistently and energetically. But the effect can be the same as a strawman because it often comes across as the argument that the election was stolen and therefore as if refuting it means that you've refuted that the election was stolen. I think this concern is one that you hold too, more or less, but you can correct me if I'm wrong.

So it doesn't bother me if people argue over whether there was massive fraud as long as they are at the same time aware of the separate question of whether the election was stolen by fraud that was not massive.

[OnTheOtherHand]

129. yes, that is a big part of my concern

and I would again add the next tier of questions: whether future elections will be stolen, and whether we can prevent it.

For many people, the arguments that massive fraud occurred in 2004 wonderfully concentrate the mind upon the urgent need for change. But this simply will not work for everyone. I think we need to accept that fact, accept disagreement about 2004, and find arguments that don't depend upon enforcing agreement about 2004. We don't necessarily have to stop arguing about 2004; we don't necessarily have to continue arguing about 2004.

Sorry if I'm belaboring the obvious, but some folks seem to regard disagreement about 2004 as a luxury we can't afford. It is what it is.

[kster]

115. No way to know for sure, because America counts their vote in secret

if you said this from the get go, you wouldn't of had to go thru all this work.

Some good work you put together, but why? You know the votes in America are being counted in secret.

[Febble]

117. You are absolutely right!

Yes, I would quite like that year of my life back!

On the other hand, I learned a lot, both from the data and from DU. Thanks to all.

Although, actually, I did say it from the get-go. What I also said from the get-go was that the fact that people had to queue for hours in the rain to vote was a disgrace to any democracy, let alone one as rich and influential as America.

Getting the votes counted properly and transparently is one half of task ahead. Making sure everyone entitled to a vote gets to cast it is the other half.

And neither of those two things have anything to do with exit polls!

[kster]

119. Give us some suggestions, how did your Country ban e-voting? nt

[Febble]

120. Well, we haven't got it

much yet, but I think it's coming. I think we need to learn from you. I think the Scots are keen to bring in optical scanning.

I'm sure it will come actually. Sorry to be depressing. But at least we know how NOT to do it.

One thing that is good though, is that we have something called the Electoral Reform Society. It has been going for decades, advocating for election reform, and also actually running elections for bodies like unions. They are an excellent resource - but they do use optical scanners for the elections they run.

The thing is that handcounts aren't enough to ensure transparent elections - the biggest worry I have in the UK is postal-voting-on-demand. We had our first real attempt at election theft in the last election - theft of postal ballots. I hope they tighten that loophole.

But our biggest strength I think is the way our elections are structured. Even if optical scanners come in, I can't imagine we will lose the oversight we have at present. The counts are too ceremonial for that - and I can't imagine they will cease to be public, televised events! And that is a huge bulwark against fraud.

But I'm keeping a wary lookout, I assure you.

Cheers

Lizzie

[kster]

123. Your Media will be stolen, Your Government will silenced

What else do they need to know about e-voting.....

Tell them to stick to what this article says, Keep the vote count out in the open for all to see. Anything else is just plain crazy.

http://www.theregister.co.uk/2005/09/06/govt_voting/

[Blue Shark]

118. Thanks Liz...

...You are certainly well informed regarding exit polls and I love eomer's take on how red shifts wouldn't show if the same fraud had been practiced in the pastin the same way.

There are some issues that are exclusive of the exit polls in Ohio and other swing states such as:

This is what occured to believe GWB got 3.2 million more votes than Kerry

--A big turnout and a highly energized and motivated electorate favored the GOP instead of the Democrats for the first time in history.(despite over 70 newspaper endorsement reversals from 2000)

--Even though first-time voters, lapsed voters (those who didn’t vote in 2000), and undecideds went for John Kerry by big margins, and Bush lost people who voted for him in the cliffhanger 2000 election, Bush still received a 3.4 million vote surplus nationally.

--The fact that Bush far exceeded the 85% of registered Florida Republicans’ votes that he got in 2000, receiving in 2004 more than 100% of the registered Republican votes in 24 out of 67 Florida counties, more than 200% of registered Republicans in 10 counties, over 300% of registered Republicans in 4 counties, more than 400% of Registered Republicans in 4 counties, and over 700% in one county. This could only be explained by a massive crossover vote in these specific counties by registered Democrats and/or Independents. Bush's share of crossover votes by registerd Democrats in Florida, however, did not actually increase over 2000 and he lost ground among registered Independents, dropping 15 points. Floridians were just so enthused about Bush and Cheney that they somehow managed to overrule basic math.

--The fact that Bush got more votes than registered voters, and the fact that by stark contrast participation rates in many Democratic strongholds in Ohio and Florida fell to as low as less than 8%, do not indicate a rigged election.

--The fact that Bush “won” Ohio by 51-48%, but this was not matched by the court-supervised hand count of the 147,400 absentee and provisional ballots in which Kerry received 54.46% of the vote doesn’t cast any suspicion upon the official tally.

--CNN reported at 9 p.m. EST on election evening that Kerry was leading by 3 points in the national exit polls based on well over 13,000 respondents. Several hours later at 1:36 a.m. CNN reported that the exit polls, now based on a few hundred more - 13,531 respondents - were showing Bush leading by 2 points, a 5-point swing. In other words, a swing of 5 percentage points from a tiny increase in the number of respondents somehow occurred despite it being mathematically impossible. (This is TIA's strong trump card)

--Exit polls in the November 2004 Ukrainian presidential elections, paid for in part by the Bush administration, were right, but exit polls in the U.S., where exit polling was invented, were very wrong. (this directly counter-argues your position that historically the exits are NOT looked to to decide elections)

--In every single instance where exit polls were wrong the discrepancy favored Bush, even though statistical probability tells us that any survey errors should show up in both directions. Half a century of polling and centuries of mathematics must be wrong.

You will no doubt notice that this is all from Dr. Dennis Loo's commentary posted at http://www.projectcensored.org/newsflash/voter_fraud.html and I would sent you there for the footnotes to support these issues.

Everthing, EVERYTHING, about 2004 sez NO WAY King George II got 3.2 million more votes than Kerry and much, if not most of it, has nothing to do with exit polling.

Ohio was just the flash point, but NM, Florida, Penn. all the way around the country reported results that just don't match the empirical, mathmatical, observational, and statistical data.

In the "Big Picture" we all know (s)Election 2004 doesn't pass the smell test, and it is vital that we get it turned around before the next (s)election, and the next and the next...cuz God knows it has been going on a while.

[Febble]

121. Thanks for the post,

Coupla three points:

--CNN reported at 9 p.m. EST on election evening that Kerry was leading by 3 points in the national exit polls based on well over 13,000 respondents. Several hours later at 1:36 a.m. CNN reported that the exit polls, now based on a few hundred more - 13,531 respondents - were showing Bush leading by 2 points, a 5-point swing. In other words, a swing of 5 percentage points from a tiny increase in the number of respondents somehow occurred despite it being mathematically impossible. (This is TIA's strong trump card)

This is based on a misunderstanding of the timeline. Clearly the difference was not made by a tiny increase in respondents. The thing that made the difference was that the later estimates were made after the results had started to come in. And in fact, this is how the exit poll projections were designed to be done - and the way they are done every year. The details are actually posted on the E-M site, and were before the election. The exit poll responses are one data source for making the projections (the things that allow the networks to "call" a state) but for all but the most slam-dunk states, the projections aren't made until a substantial number of vote-returns are in. And when they are, the crosstabulations are weighted to reflect what is assumed to be bias in the poll. Now the fact that the weighting was so substantial in 2004 certainly indicated that something was wrong somewhere - and it could have been fraud - but it doesn't reflect any kind of mathematical impossibility, and it doesn't (and this is an important point) indicate a cover-up. It was widely misunderstood, but the information was in the public domain. I read it myself on election day. And I should also add that the pollsters themselves knew that there was something wrong with their samples before a single result was in, because they weren't matching their pre-election polling.

--Exit polls in the November 2004 Ukrainian presidential elections, paid for in part by the Bush administration, were right, but exit polls in the U.S., where exit polling was invented, were very wrong. (this directly counter-argues your position that historically the exits are NOT looked to to decide elections)

You might like to check out this post and the links:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=364&topic_id=1328708&mesg_id=1336021

--In every single instance where exit polls were wrong the discrepancy favored Bush, even though statistical probability tells us that any survey errors should show up in both directions. Half a century of polling and centuries of mathematics must be wrong.

This one makes me a bit cross because Dennis Loo really should know better. For a start it isn't true - there were states in which the discrepancy favored Kerry. But that is a minor point - it is certainly true that there was apparent net bias in polls that cannot have been due to sampling error, because it significantly favoured Bush, and which I think is what Loo means by "statistical probability". Sampling error is the error you get simply because you are polling a sample not the entire population - and samples will vary from the population simply by chance. But that variance, as Loo says, will cancel out. What Loo misses is that "error" in surveys is not solely "sampling error". There are many sources of "non-sampling error" and although some of them may cancel out, some of them may be of a type that may not. These sources of "non-sampling error" include: non-response bias, coverage error and selection bias. And they are a hazard in any poll. Loo's point about "half a century of polling" is, not to put to fine a point on it, simply ignorant.

But I agree with you that election 2004 doesn't pass the smell test, which is, after all, why I'm here. I could smell it across the Atlantic (I'm a Brit). What I dispute is the magnitude of some of the estimates of the crime, when they are based on dodgy statistics (do Americans say "dodgy"?) and naivete about the sources of error in surveys. Getting an unbiased random sample in a survey is notoriously difficult, and exit polls, contrary to widespread belief, are not especially accurate. Ironically, I think the reason most people think they are so accurate historically in America is that they have always been weighted by the incoming returns.

Some of Dennis Loo's other points I won't dispute, and they are similar to Mark Crispin Miller's. Essentially they boil down to: "Bush couldn't have won because he was a bozo". Which is a perfectly valid point and one of the reasons I was suspicious. But when it comes down to actual data, and inference from data, most of Loo's points seem to be flawed.

But I don't want to get in a fight over this. A lot of people were rightly indignant at the stink (including myself) and a lot of people waded in without specialist knowledge In a sense they did a good job in keeping the questions going, and I think some inferences were reasonable at the time, given the state of knowledge and the availability of data. But I think we have more data now, and we actually have answers to some of those questions. That means we can concentrate on the ones we don't have answers to, for instance, why weren't there enough voting machines in Franklin County? How could America allow the introduction of voting technology that is so insecure, opaque and unauditable that no-one can be sure that the person with the most votes was the person people voted for? And how the heck was it possible for the Secretary of State for Elections in Ohio to simultaneously be Chair of the committee to re-elect the president?

Cheers

Lizzie

[Blue Shark]

125. Thanks again Lizzie!

[unc70]

146. We found various "red shift" effects in NC in 2004

A couple of us looked at what happened in various parts of NC in the 2004 election. We found a variety of really puzzling situations, most of which went unreported, even by us, because there were so many more-obvious issues. Because a couple of these other issues such as lost ballots on DRE systems hurt both parties, NC law was changed to require paper ballot or paper trail, random audits, access to source code, etc. The people leading that fight are heros.

I did not want to muddy the waters with issues that might appear too partisan and thus dismissed by opponents or possibly used against our collective efforts to change the NC election laws. For background, in 2004 NC had a full slate of state-wide elections including US Senate, which provide more ways to compare the data. And over the 100 counties NC used just about every technology and vendor in 2004 including hand-counted paper, opscan, punch card, DRE, ...

Example results we found were:

1. Consistently more votes by precinct for Bush and Burr(R) than for other prominent R candidates. People often dismiss this with the Reagan Dem argument, but other findings seem to rule that out. NC Republicans had just had another very contested primary that has tended to force voters to register in their "natural" party. The demographics of some of these precincts would make that highly unlikely, particularly with Edwards on the ballot.

2. Almost all of the shift towards Bush and Burr happened on election day, not in early voting. In many precincts on election day, Bush would have needed the equivalent of all the Rep/Ind/Other voters plus some of the Dems. The early voting generally appeared consistent with voter registrations and with pre-election polls and election-day exit polling.

3. The red shift even happened in precincts that are 70-90% African American!

4. The red shift was primarily on optical scanned ballots. The DRE systems had lots of other problems.

5. A state-wide recount of one race provides additional comparison data.

6. It took about a year before one state-wide race was finally decided after recounts, courts, and legislative involvement.


I am nearly certain that something strange happened in our vote results, but I don't know what or how it happened. (I do have some suspicions.) Since I don't have access to the individual ballots, just the precinct totals for candidates and for voter party registrations, and since I don't have access to raw polling data, I really can't solve this puzzle at the moment.

While I have other concerns with 2004 in NC (security, vendors, dirty tricks), so much has changed in 2006 that I decided to just file those concerns until after this election. Overall, I now believe that NC should be in relatively good shape.

BTW I am a software, security, and data analysis professional. Lots of training and experience dealing with messy, real-world data.


As for Ohio, I have followed the various reports rather closely and think that their results are explained by "all of the above". The punch card issues with ballot-position rotation/multiple precints in same building certainly caused big problems. Strange results in the "lockdown" counties. Various voter-suppression actions. Blackwell's near-total control of county election boards, sometimes loaded with faux democratics. General incompetency.

[Febble]

147. Interesting

Thanks for the response.

Messy, real-world data is indeed messy, but can be useful! And precinct totals will tell you more than polling numbers. There are very few precincts in the poll per state (tens), and the precinct level samples are often small (average 80). Do you have precinct totals for past elections? That would give you a benchmark for your redshift.

[OnTheOtherHand]

149. good stuff

I think it's tricky to assess early versus ED voting because early voting is expanding so fast, we don't have clear historical expectations. There's no doubt in the aggregate that the presidential and Senate races had a shift that the down-ticket races mostly didn't, and certainly not to the same extent. I've offered some reasons here why I don't find the shift highly suspicious in itself*, but that leaves plenty of open questions. I think I also looked at presidential returns at the county level, 2000 vs. 2004, by machine type and nothing jumped out that way, so I moved on.

* IIRC, since the NC "exit poll" also included a telephone poll of some early voters, and had the gov race as well as the pres and Senate races, I found some evidence in the exit poll that there actually was a shift in the pres and Senate race as opposed to the gov. The data are very noisy, to be sure -- the telephone poll was n=380 or so.

I would have some follow-up questions, but there's no way I will be looking at NC 2004 in the next five days -- and after that, who knows what we will be thinking about? But I do want to come back to it.

[BeFree]

155. Have you ever seen this thread about NC 2004?

It goes along with what you wrote, fits real well, actually.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=170&topic_id=2562&mesg_id=2562

Following is the bottom line from the extensive post:

So what the heck is going on here??? Kerry was behind by 6 points in the absentee/early voting. The result is consistent with the pre-election polls and most importantly with the exit polls of November 2nd. THE EXIT POLLS TELL US THAT PEOPLE VOTED IDENTICALLY TO THE OTHER THIRD OF THE ELECTORATE. By all standards of reason, the other two-thirds of the vote should be very close to the same result. But look at what happens -- a sudden and unexplained plummet in the very same electorate of NINE POINTS at the election day polls, more than doubling Kerry's overall margin of defeat. A 15 point edge for Bush in North Carolina on election day??? Come on -- I'm not that gullible. I honestly don't know how to account for that outside of computer programming -- and if it's there, there's a damn good case with the nationwide inconsistencies between exit polls and results on election day to say that it follows everywhere electronic tabulation goes. My gut tells me that this is why there is a reluctance in Florida and Ohio to push the absentee counting and that the ballots and counts had best be watched very damn closely. They present a paper trail challenge that if understood will provide a key benchmark for election day fraud. I also want to point out that the differential was not there prior to election day -- meaning there either had to be a *date specific* alteration in the software, a hack, or a specific activation just prior to the election. And lastly, it is not only the Presidential election day vote that is spurious -- the close Senate races also bear close scrutiny.

[OnTheOtherHand]

156. I addressed this in my post n/t

[unc70]

157. Saw all of those threads at the time

I have been lurking here since very near the beginning, so I have seen nearly everything relating to voting systems, particularly wrt NC.

What I found most curious about the NC red shift was that it happened in some really strange places, given their demographic profiles. Fundamental problems with the results passing the "common sense" test (e.g. red shift in 80% AA precinct.) Even Orange County looks somewhat "off" at the precinct level.

But as I said, I decided not to spend anymore time on analysis, but rather do what I could to change our laws by supporting those who were leading that successful effort.

[OnTheOtherHand]

122. Loo's piece is pretty much a muddle

I won't go over the ground Febble has already covered. Really, the one about "centuries of mathematics" is sort of painful. As for some of the other points:

Loo doesn't even try actually to document that high turnout never benefits Republicans. He just asserts it.

There is no reason whatsoever that Bush couldn't win while losing among first-time voters, etc. He did have to get more Gore votes than Kerry got Bush votes, which may seem unlikely to us, but it isn't an impossible thing before breakfast.

There is no reason Bush couldn't get more votes than there are registered Republicans in any number of Florida counties.

Even if there were two precincts in Perry County with impossible turnout (check the footnote -- not much relation to the claim in the text, but that is probably just an honest editorial goof), that comes nowhere near documenting massive election fraud. Nor were most turnouts in Cleveland unusually small.

There is no reason to expect the absentee/provisional ballot proportions to match the overall proportions.

If I may say so, the reason political scientists aren't convinced by these arguments is that they aren't convincing. In contrast, Febble's work on machine misallocation in Franklin County actually was convincing. There's no way to reconstruct who was convinced by Febble rather than by others, but the work holds up.

[truedelphi]

135. Regarding Florida - you need to talk to

Activists who dealt with widespread voter disenfranchisement. Always Democratic, never Republican.

The continued use of voter purging from registration lists.

Ever ask Bev Harris why she and Andy Stephenson attended a Registrar of Voters convention (early Dec 2004) in order to subpoena a Florida elections official?

[truedelphi]

134. Great list of things that the analytical crowd

Needs to examine.

I would add, who is, I mean, *is* Mitofsky?

And just why *is* the Mainstream Media in league with him, and vice a versa?

[Febble]

137. Well, some answers to your first question

are here:

http://www.aapor.org/warren.htm

http://www.pollster.com/exit_polls/more_recollections_of_warren_m.php

http://www.pollster.com/warren_mitofsky_an_appreciatio.php

http://www.washingtonpost.com/wp-dyn/content/article/2006/09/05/AR2006090501477.html

Essentially, he invented the exit poll system, and he died suddenly three weeks ago.

The answers to your other question are probably in some of these links as well. The networks were "in league" with him in the sense you are in league with your doctor or lawyer. The networks were his clients. They commissioned the polls.

You might like to read his address to NYAAPOR on the occasion of a lifetime achievement award:

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

As for the list of things, there is plenty of evidence other than the exit polls that the 2004 was unjust. I worked on some of it myself. But that doesn't transform the exit poll data into good evidence that the election was stolen. The exit poll data remain, if anything, counter-evidence to the case that the theft was on a scale of millions. Apart from that, it doesn't tell us much at all. It certainly doesn't tell us anything about voter suppression.

[truedelphi]

138. The Mainstream Media does not get into bed

With ANYONE who will not tell them what suits their purposes

After the 1996 Telecommunications Act, we ended up with only four or five major news media providers. One of these, General Electric, is a major War/defense conglomerate. Long before this TC Act occurred, there was a huge move to the right in the media. Some researchers document that immediately after the War ended in 1945, the major TV broadcasters were brpought in line with the CIA.

(You do not have to know that you are in line with the CIA to be considered a valuable asset by them. However throughout the nineteen sixties and nineteen seventies the FBI had yearly meetings with the head people at America's most successful ad agencies. Is it any wonder that the 1980 ad Campaign for Time magazine featured a photo montage of Reagan looking royal, and Carter being attacked by a rabbit?)

ETHOS is a whole lot in my book. I don't mean for this exchange to be an attack on your work, Febble. But I would find it difficult to believe in any organization's findings if they were sponsored by Mainstream Media.

I am not a statitician. All I have to go on is my experience with the pesticide industry in this country, fifteen long hard years. When Monsanto tells me that University such and so had released findings on its pesticides' safety,a nd I find that that University received X amount of monies for its research laboroatories, I have to weigh that heavily. Ethos, ethos, ethos. Follow the money and if the money taints you, then the research - even if there be volumes of it, even if there be dozens of awards attached, - is only tainted research.

I like science, I like researchers. But truth has become "Truthiness" as Stephen Colbert puts it. (Or as Rumsfeld must believe, "Say it often enough, and it becomes true.")

It is a shame that Mitofsky did not feel it necessary to address this issue.

How hard would it have been to get a team of researchers that were independent and let them - untainted by the networks' money or Mitofsky's influence - look into this?

[Febble]

139. Well, frankly

that is pure assertion.

And I'm not sure I even understand your post.

Are you saying the exit polls would have been biased because they were commissioned by the networks? Well, they probably were biased - with a pro-Democratic bias - which is a perennial hazard with polls. But because they were commissioned by the networks? What is your rationale? How would having a pro-Democratic bias in the poll suit the purpose of the networks? Frankly I think it was a damn nuisance.

And what is the "issue" that "Mitofsky did not feel it necessary to address"?

And as for "How hard would it have been to get a team of researchers that were independent and let them - untainted by the networks' money or Mitofsky's influence - look into this?"

What are you actually suggesting? ESI looked at the Ohio data, but clearly they couldn't be independent "of Mitofsky's influence" because Edison-Mitofsky collected the data.

Can you articulate what it is that you think should be looked into?

[truedelphi]

141. Yep, I am saying that I fear or believe or both

network polling is done in a biased way. The Media is Right wing and any polling that comes up with anything wil be altered to fit in with the mainstream media wants.

The White House, The Right Wing, The Media, all wanted George W to be President. They got him.

The actual polling might be done in a perfectly honest, perfectly scientific way. The numbers that you yourself run may be handled in a perfectly honest, perfectly scientific way.

But if Mitofsky was being paid by the Media, directly or indirectly, then I imagine that there was a big old switcheroo between the numbers that you crunched and the numbers as they were collected.

The Powers That Be are always going to swing things to supporting that their premise is scientifically supported. Whether it be pesticides, whether it be a political election.

Polling techniques themselves might actually allow for a Democratic bias. But the person(s) in charge of handling the paperwork after it gets collected is going to be able to alter the findings.
If that person is ultimately paid by the Media, the alteration will occur along a Right wing breakdown.

All in all it is much easier to deal with the science of the pesticide industry. after they tell me that I need such and such a University's studty to confirm that RoundUp is safe, and after I find out that that same Univ. is basically on Monsanto's payroll, I also find a scientist or two who has been let go (and then blacklisted by industry) when they discovered that much of the research that they colleted never made it into the survey. So I have confirmation of that pit of the stomach, something is not right feeling.

Don't have that yet with the polling situation.

[Febble]

143. I'm sorry, I am still

really unclear as to what you are alleging might have happened.

What do you think might have been altered, and in what direction?

[truedelphi]

144. Okay for starters

People who watched Tv on election night in 2004 saw that the
exit polls were giving the "win" to Kerry.

Then almost all the networks had the Bush family come on and
immediately after that, there was change in the nature of reporting.

All of a sudden, everything was okay for Bush. At least one DU member
posted right after the elections that they believed that the Bush family interview was
a signal for the dirty work to be done - that around the country, at many key
Republican controlled voting Counties - the personnel that were in on the fix
and the "shave" were now
to go in on the ES & S and the Diebold and the other hackable equipment and switch
to Bush. (I even believe that this may have happened where I live in Marin County
- I am not able to reconcile why Marin with all its liberals went 72% for Kerry and
SOnoma,one County north, with all its good ol boys went 82% for Kerry. In this
case, what occurred was a "shave' - it did not alter that Kerry still carried the
electoral votes for California)

Little of the above is pertinent to "exit polls" but then:

Now after the Bush win is announced, There is a "correction" of the exit poll numbers
as they were and the way they need to be to support the outcome of the election.

I am no longer remembering who it was that went in and made this "Correction." If it
was Edison, who released it to the Media and they made the change. if it was
Mitofsky and he made the correction figures abailable to the Media and they made
the cahnge. or if the Media just went in and they made the change.

But the fact that this "correction" was made so swiftly stuck in my craw and then
the idea that the Media just accepted this "correction" with out much (if any!)
discussion only made me feel worse. If a mistake has been made, wouldn't it take time
to consider all the factors and then come up with the "correction"

The only other analogy I can think of is when
the government did not know anything at all about the plot on Sept 11th, but then
24 hours later they knew exactly who the hijackers were. And again, the Media did
not find anything peculiar about this turn of events.

[Febble]

145. OK, I wondered if that was what you meant

I've actually posted about this on this thread I think, and certainly elsewhere.

I'll try and find some old threads, but you might like to start by looking at the E-M website, particularly the FAQ.

As briefly as I can:

The pollsters offer two services:

1. Information that allows networks to "call" each state once it is clear that its Electoral College votes have been won by one or other candidate.
   
   
2. Data on who voted for whom and why.

Both these are so that the networks can discuss who's winning, and why they might have done so.

The first, the "projections" are made by a combination of exit poll responses and actual vote returns. Unless a state is incredibly obvious, it won't be "called" until a substantial number of vote returns are in. As the results come in, the exit poll projections are gradually reweighted in line with the actual vote. They are designed to project the vote, not act as an independent check on it.

The second is based on a subsample of exit poll questionnaires. These initially showed Kerry ahead. However, because the pollsters assume that if there is a discrepancy between poll and count, it is the poll that is biased, in order to increase their perceived accuracy, these too are "weighted" in line with the vote returns.

So there is not a "real" exit poll, that was surreptiously "corrected" and then "released" to "the Media". The networks are involved at all stages; their own statisticians weigh up the projections, and decide when to call each state, depending on how confident the pollsters are that their current projections are going to agree with the count.

Certainly in 2004 a substantial amount of "reweighting" occurred, although it was not very much greater than the amount required in 1992. However in 1992 Clinton was ahead anyway, and the reweighting just reflected a reduction in his margin.

So the question as to why there was such a big mismatch between the exit poll responses and the count is a legitimate one - the big question was - was it due to fraud in the count, rather than bias in the poll?

But even before close-of-polls, the pollsters knew that they had an unusually biased sample because it wasn't agreeing with their pre-election estimates. So they were pretty confident that it was their poll. And their subsequent investigation (and my analyses) put a lot of detail on just where the bias had come from.

But it wasn't unprecedented, the reweighting is what is done every election, the discrepancy itself had been similar in 1992, and no-one tried to hide anything as far as I can see. The information was all on the FAQ on their website before hand, and the early crosstabs were on the CNN website.

edited for stupid typo

[OnTheOtherHand]

102. I don't think Rubin was discussing a debate

If you want to say that the exit poll debate was productive once, and now is "just where the politicians want us," OK. I think the debate used to be more productive myself. But Rubin didn't say that it was unproductive to debate the exit polls. He said (roughly) that it was a disservice to insist that Kerry had won, because it gave opponents an opening to dismiss all DRE critics as factually challenged sore losers.

Plenty of activists managed to 'rage against the machines' without falling into that particular trap. You've pretty much kept your own focus on the machines. We have disagreements, but I don't think that is one of them.

[autorank]

110. Here's a hot KOS thread...blast from the past. My commentary on this.

autorank Donating Member (1000+ posts) Thu Aug-25-05 10:17 PM
Original message
American Legion and DailyKOS Agree – “`Like Minded

WORST OF THE WORST

[truedelphi]

133. In a totalitarian system, Conspiracy theory is the only viable option

Conspiracy is just another word for NOT THE OFFICIAL STORY

In any totalitarian regime, those who hold onto conspiracy theories have little other option.

Can I believe Mitofsky? Well, looking SOLELY to the ethos of the situation, he would *have to be* in cahoots with the mainstream media - they certainly did not immediately after the election and the dirty deed was done, start advocating the independent position of people like Zogby. If the media is relying on him for information, by the very nature and definition of the situation, Mitofsky is part and parcel of the whole can of stink that is the current mainstream media event here in the USA.

(And I have yet to hear the MSM refer to Kennedy's piece in the Rolling Stone.)

In another year and country, in another land where the mainstream media was not free, there I detect the conspiracy theories of those who said "But what can I do? I make my fist into a stone and put it into my pocket."

Let's examine the conspiracy theory fanatics of that recent (relative to history) time and place:
In 1933, in Germany, you would have been branded a conspiracy theorist if you expressed feeling uneasy about Adoph Hitler's fast rise to power.

In 1936 in Germany you would have been branded a conspiracy theorist had you stated openly that bad things were going to result from the Jewish populace being forced to wear the Star of David.

In 1937 after the invasion of Czechoslovakia, you would have been branded a conspiracy theorist for saying that this was the beginning of Germany's decline.

In 1939 in Germany you would have been branded a conspiracy theorist if you had stated that it was absurd to think that the nation of Poland had invaded Germany.

In 1941 when Goering announced that the Allies bombs and missiles would never afffect German soil, and you denounced this statement, saying simply, What goes around comes around, comes around, you would have been called a conspiracy theorist.

In late 1941, when the last of your Jewish neighbors are rounded up, you say, "I bet we never see them again. I doubt that where they are going has anything to do with an establishment of community for the Jewish populace" you will be branded a conspiracy theorist. (Remember the reason that the German police originally started to round up their Jewish friends is that after Kristelnacht, the German people were told that these measures were designed for the protection and safety of Jewish people.)

In late 1943 if you are a German in Germany, and someone hears you muttering to yourself that there is no way that Germany can win, yes, you guessed it, you would be branded a conspiracy theorist.

Monday Sept 18th 2006 here in America, the President of the United States announced that "It is unacceptable to think." Only Keith Olbermann took on this President's dismal performance with the manner of stridance that the situation demands.

Many of us who do think will be branded conspiracy theorists by the right wing talk show hosts, demonic by the right wing "Christian" crowd, and threats to the national security by those in Homeland Security.

But hey, there are always those who think outside the box.

[In Truth We Trust]

124. Ya can't keep a good man down! Go Truth! Hand Counted paper ballots NOW!

[WeHoldTheseTruths]

150. I posted this at DailyKos days ago and haven't been banned

yet.

Only FRAUD can stop the Democrats from winning the House AND Senate
http://www.dailykos.com/story/2006/10/31/64630/768

Beyond that I don't know what's happening about this. (I hadn't seen any TruthisAll posts at Kos and didn't know he had posted there.)