Landslide Denied! Major Miscount in 2006 Election! BIG STORY!

[emlev]

The Election Defense Alliance has just released its analysis of the 2006 exit polls showing that approximately 3 million votes were denied Democratic candidates in last week's election!

From the link:electiondefensealliance.org/major_miscount_of_vote_in_2006_election|press release]

Major Miscount of Vote in 2006 Election:
Reported Results Skewed Toward Republicans by 4 percent, 3 million votes

Election Defense Alliance Calls for Investigation

BOSTON, MA - November 16, 2006
CONTACT: Jonathan Simon 617.538.6012

Election Defense Alliance, a national election integrity organization, issued an urgent call for further investigation into the 2006 election results and a moratorium on deployment of all electronic election equipment, after analysis of national exit polling data indicated a major undercount of Democratic votes and an overcount of Republican votes in U.S. House and Senate races across the country. “These findings raise urgent questions about the electoral machinery and vote counting systems used in the United States,” according to Sally Castleman, National Chair of EDA. This is a national indictment of the vote counting process in the United States!
{link:electiondefensealliance.org/major_miscount_of_vote_in_2006_election|More...]

Full report can be read here.

This is huge, folks. This means that the Dems finally have the opportunity to challenge the election system without the "sore loser" label. And now it's our job to make sure they do it!

[emlev]

1. EDA leader to be on Hartmann in a few mins--can we get this to Greatest?

My understanding is that they'll be talking about the report at ten minutes past the hour (2:10 EST, 11:10 PST) today, Friday 11/17.

Can you help me get this to the Greatest page before that so DUers can listen?

Thanks!

[texpatriot2004]

2. So, they tried to steal it but didn't have their math right eh? nm

[glitch]

3. That's what happens when you try to create your own reality.

K & R :)

[texpatriot2004]

12. Yeah, especially when you're the only one living in it - as they are

:hi:

[Subdivisions]

92. So true!

[reichstag911]

18. They underestimated their unpopularity!

:applause:

[texpatriot2004]

21. Yeah, they failed to factor that into the equation ;-) nm

[cui bono]

94. Even though Karl had THE math. n/t

[chefgirl]

4. Well, theres a surprise, huh?

No wonder the little idiot looked so 'gut punched' the day after.

They really thought they had perfected their game. I guess they never considered how many people in this country are sick and tired of their bullshit.

We beat the margins, it seems, and fucked up "the math". :rofl:

-chef-

[emlev]

8. moved (posted in wrong spot)

Moved this post to the right place.

[notadmblnd]

15. yep, the huge turnout made it impossible for them to steal

but that's just my opinion.

[RaleighNCDUer]

30. Not impossible. They just underestimated how much they would

have to steal.

My pessimistic self worries that they are not going to make that mistake again, and the next map will look like Reagan/Mondale.

Unless we can get rid of the damn machines before then.

[notadmblnd]

32. the had a mechinisim in place to tweak it just as much as they thought they needed it.

Remember KKKarls satement about his math? but you are right, KKKarl did his "gazintas" wrong and miscalculated turnout. If turnout had been what was expected, the :puke:s would have kept both Houses.

[sinkingfeeling]

5. That would explain why some of the races were so tight!

[emlev]

9. And some of the lost close races, I believe

EDA is now going to be looking at individual races and I think we'll see that some that went GOP should by all rights have gone Dem.

There are still still several Congressional races that are considered unresolved. This information needs to be taken into consideration in the pursuit of justice in these cases at the very least. Here's a summary.

[wakeme2008]

14. Bingo that is the real reason....

and in a number of seats we lost by small margins were in fact wins.

[exlrrp]

128. Don't forget Republican gerrymandering

This was all done against the Republicans re-rigged districts. The count should have been higher just on that basis alone

[AdvancedProgress]

6. Something's gotta give...

Everytime I hear another e-vote story I get a little more jittery about my vote.

:scared:

[emlev]

11. This is about so much more than our individual votes.

This is about our democracy. Check out ways to get active at !

[rumpel]

7. big story, indeed! thanks n/t

[seafan]

10. Link to Hartmann's program here:

http://www.airamerica.com/thomhartmannpage

[emlev]

16. EDA's Sally Castleman on Hartmann right now! eom

[keepthemhonest]

13. music

music to my ears, but now that dems have control of house and senate will they do soething about it?

[emlev]

17. It's up to us to make sure they do! eom

[rman]

19. they're probably going to call us sore winners

[BadgerKid]

31. Nah, just throw their "mandate" back at them. n/t

[seabeyond]

47. or consistant, win or lose n/t

[truedelphi]

20. Greatt o read about this organization n/t

[Einsteinia]

22. Someone (hint!) ought to figure out

how many seats the dems SHOULD HAVE taken!

[rocktivity]

23. Well, James Carville doesn't have anything better to do right now...

rocknation

[StrictlyRockers]

25. Boy would I be pissed at *, that he is so hated that they couldn't steal the election.

I'll bet poppy and Baker took jr to the woodshed, but good!

[nebula]

24. The machines only able to steal 3-4 million votes

That's what Greg Palast said in an article written before the elections.

Edit: link to the article

[bobbie]

26. Sore Winnerman Sore Winnerman!

I can just see the signs.

[grizmaster]

27. This helps put the stolen elections back on the front burner

[Febble]

35. Well I do hope not

If this is what puts it on the front burner, it's going to fizzle out. There is excellent evidence for corruption and miscounts in 2006, but this simply isn't it. The argument has a gaping hole in it.

[spooky3]

59. And that gaping hole would be...?

[Febble]

67. Well it seems to be

entirely predicated on the observation that Bush's "retrospective margin" increased, wherease there is excellent evidence to suggest that the winners retrospective margin always increases. I mean it's possible that this time it was due to fraud, but to base a case entirely on it seems odd.

See the comments to this Daily Kos diary:

http://www.dailykos.com/story/2006/11/17/14169/950

I'm all for getting election integrity front and center, for investigation both corruption and miscounts in 2006, and in getting rid of unreliable, insecure and unauditable voting machines, but selling the case on the basis of an analysis like this just isn't going to do any good at all, and may do harm.

It's a shame, because I have a lot of admiration for both Jonathan and Bruce, but as far as I can see they messed this one up.

[spooky3]

69. that could be an alternative hypothesis

but the authors' hypothesis can't be ruled out based on the data shown here, if I understand both arguments correctly.

Further, if Republicans are less likely to participate in exit polls, that "retrospective margin" bias would have to be huge to offset the "Republicans are less likely to participate" effect.

[OnTheOtherHand]

84. two things

First of all, given the premise of the report, if the best that can be said is that their hypothesis "can't be ruled out," that is a big problem, isn't it?

Second, neither of these effects has to be very "huge," although the retrospective margin bias in some contexts is. (The biggest one I've seen was among respondents in the 1993 General Social Survey, who reported having favored Bush over Dukakis by something like a 41-point margin. Actually, I guess there is one about JFK of a similar magnitude. I wouldn't expect anything like that in a midterm exit poll.)

[spooky3]

142. two things

1) By that reasoning, then both this report and Febble's arguments have "big problems." From an academic standpoint, where one has the luxury of saying "the data are unclear" and should do so if one does not have strong evidence supporting one over other alternative Hs, the answer to your question is clearly "yes". Just as the authors can't persuasively argue that the how-did-you-vote-in-the-last-election question analysis definitively shows fraud (and I am not sure if they have tried), it is possible that there is little or no retrospective bias in this sample (e.g., given how unpopular this president is, and given that the basis of arguing that there is universal bias is a single paper--and it apparently has not appeared in a top refereed journal.) Simply asserting that there is an alternative explanation for the findings does not render the authors wrong, but I certainly agree with you that at least from what little info is in the link, they also can't rule out all alternative Hs. Further a basic tenet in psychometric theory (and exit polling is merely one application using this theory) is that one must have a pure, unbiased, valid measure of a construct (voting behavior) in order to determine whether a new measure is a valid measure of it. Here we have every reason to question whether the actual votes meet those criteria, so it is inappropriate to use the "actual vote counts" to reject the exit polling as not valid, biased, etc.

From a practical standpoint, one's objectives may be different. I'll leave that to others to haggle over.

2) Here the question is simply one of how you define "huge." My point was that the RB would have to be larger than the bias based on proposed undersampling of Republicans that Mitofsky gave as one possible explanation for the variation between exit polls and vote counts. Again, most academics will place little confidence in a single conference paper, as opposed to a body of evidence appearing in top refereed journals.

[OnTheOtherHand]

144. ?

The report says, "...this time there is an objective yardstick in the methodology that establishes the validity of the Exit Poll and exposes the inaccuracy of the election returns." No, there isn't, and I see no problem whatsoever in saying so. (EDIT TO ADD: I agree that we're not in a position to treat either the exit poll or the official count as a standard by which to evaluate the other. You may already have known that, since I've stated it many times, but I don't want to confuse the issue by failing to repeat it.)

It appears to me that the magnitude of the apparent retrospective reporting bias is about the same as the magnitude of the apparent participation bias, but whatever.

[spooky3]

145. when the Lindeman paper is published in a top journal,

I'll be much more interested in it, as I will be with any other papers on the topic.

I'll leave it to others to continue to discuss the OP with you.

[Febble]

148. And will you reserve judgement

on the O'Dell and Simon paper until it has been published in a top journal? Or does it get a pass?

You are of course right to treat all analyses with skepticism. Peer-review is good. But the analysis in the OP needs peer-review too. In lieu of that, both Lindeman's arguments and the O'Dell and Simon arguments are fair game for internet critique. Lindeman's paper - or his data - should, at the very least have been considered by O'Dell and Simon, who do not appear even to have looked at the data Lindeman reports, let alone explained how their argument takes it into account.

[OnTheOtherHand]

151. it would be nice if anyone discussed the OP

I rarely find anyone who is actually interested in assessing the arguments; mostly, people like the way they sound.

So far, there is nothing in this debate sufficiently surprising or controversial to merit publication in a top journal. Sad but true. It's like a scab that I can't stop picking at.

[Febble]

100. No, it can't be ruled out

although given data on retrospective margin inflation from other elections, the fact that the unadjusted crosstabs showed no margin inflation would tend to suggest that Republicans had been underpolled, even if we didn't have other good evidence that Republicans do tend to be underpolled, not least being a poll conducted just before this election in which Republicans indicated that they would be less likely than Democrats to take part in exit polls.

What irritates me about the paper is its conclusiveness. It's a possible indicator, at best, that perhaps the "adjustment" wasn't simply required because of poll bias but might have been because of fraud, and not a good indicator at that. But it's been hyped as something very much more - look at the number of recommends on this thread alone.

And I think that is irresponsible. We had a good case in 2004 for corruption and crappy machines and we have an even better case in 2006. Neither of those cases rested on the exit poll evidence, and neither depended on who won the official vote. The fact that Dems won this one is fantastic news, because it gives credibility to the campaign. But papers like this are in serious danger of jeapardising that credibility. It's simply not a credible argument, whether or not its conclusions are correct.

[eomer]

106. The only way to finally put to bed this continuing debate

about exit polls is to bring transparency to our election system that is sufficient to force one side or the other to admit they were wrong. (Or perhaps for both sides to admit that the truth was somewhere in the middle. You know, wherever it happens to fall out.)

If my credit card company had a shoddy statement system that did not sufficiently justify to me the charges they wanted me to pay for then my position would be that I'm not going to pay until they can deliver enough transparency to convince me. It might turn out, when they delivered a reasonable level of documentation, that the charges were in fact mine. But I still would have been right to make the accusation and put the onus on them.

In our current situation, I think it is similarly reasonable to put the onus on those who are running the elections and say, in effect, until you deliver a system with sufficient transparency to convince a reasonable person then I am going to construe the facts in my favor.

I agree with you that the conclusion in the EDA paper is just one possible explanation for the numbers and that there are other explanations that can be argued (historical pattern of retrospective margins). But from my perspective the EDA position just amounts to construing the facts in our favor and demanding the documentation to prove it isn't so. In essence it says, we're going to believe the exit polls until you deliver enough transparency to convince us we are wrong. So while I lean more toward your side if I look at it from a purely technical POV, I end up siding with Simon and O'Dell on this one because of where I think the onus should be.

[Febble]

111. I agree absolutely

except that I don't share your view that doing an analysis that can support two interpretations and claiming that it supports one is responsible science. Frankly the paper would be more convincing for their own case if they'd actually tackled the counter-case and made the point that nonetheless their own interpretation was possible. As it is, any political scientist is going to take one look and chuck it in the trash. Good cases deserve decent research.

[eomer]

114. Fair enough. But that same criticism should apply equally to both sides.

I don't think that Mitofsky was careful to point out that reluctant responders was just one possible explanation and to then lay out fraud as the other one. Guess I'd have to look back at the paper to be sure but that's how I remember it.

[Febble]

115. No, he wasn't

and it is one of many criticisms I had of that evaluation document (and it was because of those criticisms that I ended up being hired to reanalyse the data from scratch). I still hope my analysis can be published in some form, but it's difficult to know where to go from here now that he's dead. My last email to him is still in my draft folder.

[eomer]

116. And, just to belabor the point,

since it is a point well worth belaboring,

the part of my first reply that you absolutely agree with is that transparency is the way to end this debate, right?

[Febble]

117. Yes.

unless you are arguing that exit poll data should be released in such a way that respondent confidentiality would be compromised. I don't think that. I also think secrecy of the vote is as important as transparency of the count. I think both are achievable, but getting both right isn't necessarily easy.

Confidentiality of exit poll reponses and votes is important. But so is transparency. And transparency in vote counting is absolutely key.

[eomer]

118. Transparency of the count

is what I meant.

Although you do need transparency of a certain sort in the casting of the votes too in order to achieve transparency of the count. But since the system that you use to cast your own vote where you live and to then see it counted is what I have in mind (or something equivalent), I'm sure you were including that part just like I am.

[eomer]

119. BTW,

the credit card example I used actually happened to us.

Both of my daughters had a debit card that was sponsored and promoted by their university and handled by a major bank. On a regular basis debit charges would appear on their accounts out of nowhere with no indication of where they had used the card and with dates on which they knew thay had not made any debit purchases (like dates they were home for a holiday). We called the bank and said we were not going to pay the charges and not going to pay the overdraft fees. The bank was actually very responsive at that point and did a good job of researching and telling us what had happened. It turned out that there was a food vendor who ran many of the food plaza operations on campus and who was very careless about processing the charges. They would put them through weeks or even months after the fact and use the date they processed instead of the date the charge was incurred. Sort of the Ken Blackwell of credit card processing. We finally got (after many phone conversations) what we thought was a fair resolution on the charges that had been incurred up to that point but then closed the accounts because every indication was that the poorly documented charges and late processing were going to continue.

We don't have the option of closing our accounts in the case of elections (unless we all move to Canada I guess). Maybe we should all stop paying our taxes until they fix this problem?

_{Edit: dumb grammatical errors, doh!}

[Febble]

127. Canada is perfectly capable

of messing up bank charges too. But they do have a better election system!

[Febble]

123. Yes

you need to be able to check your own vote. Sounds like we agree.

[philb]

169. If there argument is valid and the hypotheses true, then the conclusion is true.

Is someone questioning the validity of the argument or whether their hypotheses are true?
They do have support for their position, and the fact that its clearly easy to manipulate touch screens and compilers that can't be audited, that there was documented to be widespread manipulation in 2004, and the majority of manipulation/"glitches" in 2004 was against Dems, and the history of exit poll accuracy > all support the argument and the truth of the conclusion. I think they have a pretty good case

2004
Florida www.flcv.com/fraudpat.html & www.flcv.com/fla04EAS.html
Ohio www.flcv.com/ohiosum and Richard Phillips documentation
many states throughout the country www.flcv.com/summary.html


similarly in 2000

[Febble]

172. Well one of your givens

isn't a given, namely "history of exit poll accuracy". The history of "exit poll accuracy" is that they aren't, and that in the US the direction of the discrepancy is "redshift" (count redder than poll". Moreover, the history of exit poll accuracy is that in the "past vote" question, the previous winners margin is consistently inflated. With these two givens, their argument based on the exit poll data falls apart. The remaining arguments (that there was documented "widespread" - how widespread?- manipulation in 2004) and that the majority of reported manipulation/"glitches" in 2004 was against the Dems remains an indictment of the electoral process. However the finding that in 2004 there was absolutely no correlation between exit poll discrepancy and change in Bush's vote share lends itself to the interpretation that the scale of the voting problems nationwide was not reflected in the nationwide scale of the discrepancy.

This is why I consider the argument made in this paper to be a poor/misleading one. I am all for collecting real evidence, as RHP and others have done, and is being done in Florida right now. I think scaling up the magnitude of this evidence by reference to an exit poll discrepancy is simply not justified by the exit poll data.

[philb]

178. I think you have to be careful to look at whether the shift is explained by known manipulations

Being from Florida and a part of election protection monitoring for 30 years, past legislative lobbyist for Common Cause,
and active in CC and more recently EIRS, I am aware that in Florida its clear that we haven't had a fair election in the last decade, and that the bias in manipulation has been large and in the direction of many thousands of legal Dem votes not being counted each election in recent years. So since this is well documented by major media recount in 2000 and by the EIRS and Common Cause major monitoring systems in 2004-augmented by reports to country SOEs, VotersUnite, etc.
there was a clear and documented explanation of why the exit polls found more Dem votes than the "official vote totals"
Florida may be different than some areas because elections have been mostly controlled by one party for the last decade, but its clear that the system has had a major partisan bias that has made it hard for Dem votes to be counted in many areas and for dems to have a fair playing field.

see the Gore thread and
www.flcv.com/fraudpat.html
www.flcv.com/fla04EAS.html
for compilation of hundreds of thousands of votes disapperared or swung by manipulations
widespread illegal dirty tricks in many counties to prevent minorities votes from counting in conjunction with new rules
promulgated by authorities that voting in the wrong precinct means your vote doesn't count
widespread illegal purges of minority voters from rolls so they aren't allowed to vote
biased motor voter registration system that doesn't register thousands of the "wrong type of voters" in spite of legal intent, and other registration irregularities
widespread polling place irregularities and intimidation, shortage of machines and excess "broken" machines in minority precincts resulting in very long lines and resulting ironically in very low official turnout in precincts that had long lines with waits of up to 3 hours
widespread absentee ballot manipulations and irregularities affecting over 100,000 voters
widespread electronic machine glitches
and touch screen switching in many counties;
partisan election officials in control of easily manipulated compilers with no transparency

I think its clear that in Florida any blue bias is easily explained by the massive manipulation of elections over the last decade.

The same was true in Ohio in 2004, as documented by the large EIRS and Common Cause and other monitoring systems.
There was again a huge amount of manipulation in many counties through out the state, of the same nature as that described above for Florida, as described by compilations of the Common Cause reports, Dr. Richard Phillips and others data and analyses, and the EIRS data- which fully documents major manipulations of all the types listed above
www.flcv.com/ohiosum.html

The same was true in many states, as documented by the Common Cause compilation and the EIRS compilation,
and the majority of the manipulations in 2004 were clearly away from Dems, which would explain why a lot more Dems intended to vote and attempted to vote than the "official records" show.
www.flcv.com/summary.html

While its true that manipulation can go both ways, and there is a tendancy for partisan control of elections to influence
the election in the direction of the controlling party, its clear from the election protection monitoring of the last decade that the majority of the manipulation shift has been to the detriment of "official Dem votes"

This seems to me to be the best supported explanation for the Exit poll results since 2000, and from what I can see fully supports the accuracy of the exit polls in all cases that I've had a chance to look at.

[Febble]

180. Well, I have every sympathy

with anyone from Florida, and I would be utterly astonished if Florida elections were clean.

The only sense in which I diverge from you is in thinking the exit poll discrepancy in Florida has much to do with corruption. The discrepancy wasn't large in Florida in either 2000 or 2004. I suspect that the most corrupt precincts in Florida are the strongly Democratic ones, and if corruption is concentrated in that demographic, it isn't going to impact on the exit polls much.

In other words, I think the exit polls tell you nothing about fraud one way or the other in Florida. I do think, however, that they actually contra-indicate widespread nationwide vote-switching fraud in 2004. I think they say nothing (yet) about 2006.

In addition, the exit poll discrepancy doesn't actually need further explanation - the differential participation rate hypothesis is well-supported.

Which is not to be contentious. I'd just much rather see a decent and watertight case made on the basis of the kind of evidence you mention, than a case made on the basis of flimsy exit poll evidence which can just as easily be interpreted as consistent with a biased poll. I'd like to see exit polls relegated to a footnote, if that.

And of course, I'd like to see mandatory manual random audits of paper ballots in all jurisdictions, with automatic manual recount triggered by any departure from the specified accuracy rates! Oh, and an end to robocalls, push polls, registration purges, undervotes, overvotes, and voter suppression.

Cheers

Lizzie

[philb]

182. I agree with some of what you said, I don't suggest that all machines in all counties

of all states had systematic manipulation of touch screens or compilers.

But it is well documented that there were systematic "default to Bush or "xxx" " on many machines in many states and
"default to blank" likewise on many machines in several states, and other "glitches" affecting many machines, and unpublicized defaults that cost voters in many states their vote in Presidential race when voting by party slate, etc. and manipulations of compilers in various manners such as use of wrong compilers for some votes, especially in multiple precinct polling ares with more than one ballot type, etc., manipulation of punch cards in some areas, misallignment of punch card equipment resulting in shifting of votes, compilers with wireless patches under control of manufactuer reps before and during elections, etc. These manipulations appear to have affected large numbers of votes, but in specific areas not on a generic across the board manner. Some of the default mechanisms only affect those who have problems voting and don't check the review screen or have the ability to know what to do about problems. But its clear that there have been widespread manipulations that have affected large numbers in some areas and surely swung some major elections, for certain in the case of the 2000 Pres. election and for other races throughout the country, likely including the 2004 Pres. election. Its very hard to measure the total impact of such systematic problems, but my detailed analysis of one county in Florida shows that the effects can be significant, even given the limited amount of information due to inability to fully monitor all votes.

Likewise the manipulations of the registration system and election system against minorities, both by race and against students was widespread and systematic, though mostly in 10 or some major counties of mostly 20 or so states.
And there have been widespread and systematic manipulations of absentee voting, again not affecting all states or all counties in any uniform manner, but again adding to the bias.

www.flcv.com/studentv.html

Simiarly for the many other types of manipulations documented by the large EIRS, Common Cause, state SOE, and other monitoring systems in recent years.

[Febble]

184. I agree with all that

and I applaud the hard work that has gone into compiling these kinds of documentary evidence.

[eomer]

186. "The longevity of an error...

is no excuse for its perpetuity."

I'm sure you agree it is good to question even well-supported hypotheses. That's the essence of the scientific method and the only way, in many cases, to advance the state of the art.

Best I recall where we left off on this question, I agreed that your argument was strong but vowed to continue questioning it, an endeavor that you welcomed.

[Febble]

187. Indeed.

And I will continue to welcome your questions!

[Sancho]

122. The only scientifically credibility problem are pollsters and defenders!

No thinking persons, much less professional or responsible pollsters, would continue to deny exit poll evidence YET NEVER change the methods, sampling, release detailed data, etc., etc., and then continually claim, "we have no basis for a conclusion"!

You have no basis for a conclusion that you don't want to find! With 3 tries since 2000 after the original controversy: E-M strikes out as credible! It simply doesn't take rocket science to make an up-front decision that NEXT time we'll put more interviewers in Florida and Ohio, we'll ask questions that identify the generalizibility of the sample, we'll take a few of the interviewing locations and mark down all the demographics of the sample, and on and on.

I think that IF there were a paper trail and lie detector tests of 100% leaving the polling station, some people would claim that there is still not evidence of how they voted.

It's a DUMB assertion that the pollsters don't want to find out why the polls differ from the actual vote, of course they want to know (unless they really already know). They just don't want to tell what they know.

Why not post the questions and sampling method that YOU think would prove or disprove accurately evidence of voting integrity or voting fraud: and we will get 10,000 DUer's to call for E-M to give it a try next time!

How many "that won't work" arguments are going to be posted without a "this is what would work"? Why not serious demand for ALL the raw data? I'd bet that I'm one of many who could look at the "flawed" design of the pollsters and fix it for the next round.

This is like the cigarette makers yelling that smoking doesn't cause cancer for 50 years because there is "no scientific proof" while the proof is in their own data that they don't release and the studies that they don't want to do...the "proof" is in the continuing unchanging methodology of the pollsters!

[Febble]

124. Are you serious?

You don't think the voting machine vendors have a credibility problem? Or some of the BoEs?

Your post makes no sense. Of course the pollsters try to get their methodology right, which is why they did a detailed in house investigation in 2004, and, indeed, why Warren Mitofsky hired me to do a bit more in 2005. And why they changed some aspects of their protocol. And they have plenty of "basis for a conclusion" - you just don't happen to like the conclusion.

If you think you can run an exit poll with no bias, then I suggest you hire yourself out to E-M. But if your prescription is to "ask questions that identify the generalizibility of the sample, we'll take a few of the interviewing locations and mark down all the demographics of the sample, and on and on." then I suggest you make your proposal a bit more specific before you try and sell it. Of course the exit polls are evidence of how people voted. But the sampling will be prone to non-sampling as well as sampling error. Responsible pollsters acknowledge this.

And who is supposed to be making the "assertion that the pollsters don't want to find out why the polls differ from the actual vote"? No-one I have seen on this thread, or anywhere for that matter, because, of course, it would be DUMB.

No exit poll will "prove or disprove accurately evidence of voting integrity or voting fraud". It's not something polls could ever be designed to do, and this is why, of course, they aren't. It's why you need proper manual recounts and audits.

It's nothing like "cigarette makers yelling that smoking doesn't cause cancer". It's like cigarette makers identifying the ingredients in their cigarettes that are most likely to cause cancer, trying to cut down on them, but arguing that there is no way that their product can eliminate the risk of cancer, whether caused by their ingredients or by other causes.

And the methodogy is subject to constant re-evalutation.

Was there anything else?

[OnTheOtherHand]

129. block that metaphor

The pollsters are not the "cigarette makers." That seems to be the missing insight around here. (EDIT TO CLARIFY: What I mean is, many DU posters don't seem to realize that neither are pollsters the source of the problem, nor can they solve it. Of course, many DU posters do realize this, and so are off doing other things.)

Voting could be highly reliable, but isn't; polls will probably never be highly reliable (at least not demonstrably highly reliable). We need to fix the voting.

[Febble]

130. Granted

When cigarette makers begin to have the integrity of most pollsters, cancer rates might start to drop.

We need to fix the voting.

Edit in response to your edit:

Yes, good point. The most perfectly conducted poll cannot ensure that bias is eliminated if there is an underlying differential between the willingness of Republicans and Democrats to participate in the poll. Conversely, even a badly conducted poll might not exhibit bias, if there was no underlying differential willingness. If Sancho has not understood this basic point, he is certainly not going to be in a good position to advice Edison-Mitofsky on the conduct of exit polls.

[Sancho]

140. So the expert answer is that it is an unsolvable problem?!?!

Here's the bottom line. Hypothetically, IF someone were going to design an exit poll that would provide credible evidence of reported results that are not the actual voting intent of the voters, how would they do it, i.e...what can you tell us about the undervote in 2006 in Florida?

We all understand there is sampling bias and variability and error...but a some point, the evidence becomes convincing to "experts". We only need one subset of data as an example, not the whole country...we'll come back to the 2006 example you propose in a second...

How would you advise E-M to fix things? I've hunted down non-ignorable non-response in many situations, and it can certainly be a challenge. Here's how that applies to the exit polls:

If there was an interviewer gender bias, then E-M surely hired some male interviewers this time. If there were too many reluctant republican responders, then E-M probably put up a sign, "Republican voters exit poll here, drawing for a new automobile." Did E-M do whatever was needed so that E-M can produce results that we can all believe in?

It's not good enough to always find that the post hoc evidence is weak and so the only answer is to have verifiable elections. There are plenty of election officials who are using that logic to report that we don't need to fix anything because there is no evidence of problems! On the other hand, I simply don't see enough effort or response on the part of the exit pollsters to fix a continuing methodological issue.

Post hoc analysis that identifies outliers is nice the first time. Holding the data until 5PM is not a fix. Specifically...your example:

In Florida District 13 the exit interview would surely ask, "Did you vote in the US Senate/House races?" A number of interview forms should have already asked that question! What pollster would not anticipate an undervote issue after the 2004 election in Florida and all the numerous antecdotal reports? If ONE interviewer with only 100 interviews out of 250 voters in ONE precinct out of 150 precincts in Sarasota had reported that 2% reported skipping the race and 15% in that precinct were undervoted, don't you think that would be useful evidence - even in ONE sample of interviews? Would that be used in court? Absolutely! What would you then suspect in the other 149 precincts that reported 15% undervote. Did E-M not anticipate issues in Western Florida? Do they have data from ONE subset of datal in Western Florida that would be useful on Monday to the election officials? If so, let us see it. If not, then E-M doesn't really want to do accurate polls or they would have anticipated the issue (I'm giving them credit for being competent). Saying that, "we only poll one person here and another one there in Florida" is evidence that E-M is avoiding the issue! Did E-M asks if anyone had a problem with missing or difficult touchscreens?

My 2004 exit poll data show 2898 ID's in Florida...so, what light can E-M shed on the contested undervote in Sarasota in 2006? Don't they have a hundred interviews from Sarasota for 2006? Fifty inerviews per county would make about 200 in district 13.

We don't need to know which precinct. You can argue relunctant responder all you want, but at some point the sample within that precinct (the obvious unit of analysis) becomes overwhelmingly convincing. Is there NO example in all the 2006 or 2004 and 2000 data that can't be explained in all likelyhood except by fraud or manipulation of the vote....hmmm.....I'll bet there is one! What I want is evidence that either a precinct is a bunch of very, very weird voters in that district OR that someone is hacking the vote in those machines. Can you get that information?

Specifically, I would have a blast working for E-M! So please let us know what E-M is proposing that will fix the sampling problem or why they have failed to do so to date, if only in the most likely and contested districts! Let us know if we can see the precinct level exit poll data or how you explain the undervote!

[OnTheOtherHand]

143. sigh

In Florida District 13 the exit interview would surely ask, "Did you vote in the US Senate/House races?"

"Senate/House races"?

No, actually, there is no reason to assume that the Florida exit poll interview asked about House vote, and good reason to imagine that it didn't bother, since results from 2 or 3 precincts per congressional district would not be especially interpretable. I assume that you've looked at the final state questionnaires from 2004 at http://www.exit-poll.net/election-night/Final_State_Questionnaires.pdf , yes? And you understand the difference between the state and national questionnaires, of course?

What pollster would not anticipate an undervote issue after the 2004 election in Florida and all the numerous antecdotal reports?

Huh? If you can point me to the DU post that said "I bet there will be a double-digit undervote rate in Sarasota County, but only in the House race -- Edison/Mitofsky really ought to be looking for that!", I might give this further thought. Meanwhile, I file it under Perfect Hindsight.

What I want is evidence that either a precinct is a bunch of very, very weird voters in that district OR that someone is hacking the vote in those machines.

Have you examined the election returns? What, specifically, do you find inconclusive about the election returns, and the other evidence, that 100 (hypothetical) exit poll interviews in one or two precincts would resolve for you?

Specifically, I would have a blast working for E-M!

Great, send them your resume. Febble has a full-time job on the other side of the pond.

[Sancho]

146. Exactly...you said it.

It's the best way to detect error... ask the questions that demonstrate if there is error. I personally posted multiple notes about the 2004 Castor/Martinez race being hacked with flips/undervotes on the west coast of Florida, and E-M didn't need to anticipate a hack in the Jenkins race. It was pretty much the talk of the town for the last year. In the paper, courts, and on TV for months before the election.

The fact that the Primary in Sarasota was also widely published as a mess would also be a clue. In fact, I think that a ballot referendum in Sarasota to get rid of the electronic machines after a court case in the last year to put in on the ballot (it passed) on the same election as the disputed race may have been a bit of a pollster's target for "what in the hell is going on down there?" Why are the voters passing a referendum to get rid of electronic machines over the protests of election officials in response to their experience in 2000, 2004, etc...?

Also video's of unsecured machines, Jeb and Laura and all the others flying into town to campaign, and pre-election complaints by half a dozen watchdog groups would be another hint...hmmmm....I can't think of any reason to ask anyone in an exit poll around here IF you voted or If you had any problem with the machine?!?! Why would a pollster do that?

Would there be a national interest IF a single house or senate race were hackable on E&S machines? This was the Katharine Harris home ground! Why would a pollster have an interest in that? Oh, I forgot 2000, 2004, etc...you've got to be kidding. Let's see, we're rock throwing distance from Tom Feeney and Clint Curtis....hmmmm, why would I have asked IF you voted and what races did you skip?

I find LOTS of suspected inconclusive results (like why would a district vote for a democratic tax referendum and 5 out of 6 democrats down to the dog catcher by wide margins and they barely elect a single republican to congress by a narrow margin with lots of complaints of disappearing votes (undervote) on the screen for that one election). That precinct (and sometimes district data) is not available from exit polls as far as I can tell.

E-M likely doesn't pay enough for me to move out of hurricane alley, but hey, I'm always up for consulting!

[Febble]

149. Well, if you are interested

in selling your consulting services to E-M I suggest you read their own home page:

http://www.exit-poll.net/

and perhaps my own primer here:

http://www.dailykos.com/storyonly/2006/11/4/135126/905

as you seem to be very confused about their purpose, scope and methodology.

[Sancho]

156. I understand...

I'm not sure I'd fit the E-M philosophy. I'm a little more interested in getting to the bottom of things and I don't agree with a "can't be done" attitude, if that's the way I read it. Some things that you have linked (like the WPE article) have some interesting merit. Also, I have worked on plenty of projects where the client contracted for a certain thing, and didn't want to answer to what I thought was an interesting question. That's why I'm aware of pressure to avoid finding something that may not be what the client is happy to find! E-M is in the same boat.

What the world wants is an external way to detect fraud, and as long as pollsters set themselves up to call races on TV, they will be accountible to the public when they claim that their methods aren't accurate and say, "oops". If they can't improve the product or explain the discrepency, they will take the heat! I think the public would accept a heroic effort by E-M, like anticipating a close election and setting up a report problems website and sending in extra pollsters to get as much information as possible on the spot! Then they could at least report what they did find, even with limitations.

I don't think the pollsters have earned the respect of most of the public or the statisticians yet (at least this statistician). At best there are some wait and see groups along with many who don't think the effort is good enough (even though I will give you due credit for being willing to debate).

I am beginning to understand the methology within the general context of social science methods that I use, I just don't agree with some of the theories from the reported results that I've seen. I'm not convinced the explanations for the non-response or gender interactions account for some of the larger discrepancies of exit polls from reported votes. There's as much evidence of other theories at this point. I certainly don't agree that better sampling can't resolve some of the issues.

I think that pollsters, like lots of us in various "scientific" pursuits, tend to do what we're used to doing, and sometimes we are challenged!

[Febble]

158. Well, we are moving incrementally closer

but it still seems to me that you take a very odd view of the purpose of the exit polls. They are commissioned by the news networks to provide information regarding two very specific purposes: to enable them to "call" each state before all the results are in (and to do this accurately they use vote-returns for all but very close slam dunk* races) and to provide information on "who voted for whom and why" - for which they design a long questionnaire, and crosstabulate data against the voteshares of each candidate. They are not even attempting to monitor the election, which would involve a quite different design. They anticipate bias in their poll, although they try to minimimise it, and compensate for it in various ways, including the use of pre-election polls. They also use post-stratification re-weighting, based on the vote returns. Clearly if the vote returns have greater error than the polls, this won't work, and people might, and do, want to reverse engineer the exercise to find out whether the it was the vote-returns were in error. But as the entire design of the polls is geared to predicting, and accounting for, the official count, reverse-engineering it to do something else isn't exactly straightforward.

Nonetheless, statistical analyses of the discrepancies, particularly at precinct level, would seem worth doing, and it was what I was hired to do. And try as I could, I could find no evidence for a major role for fraud (see here for details), some possible evidence that older technologies in urban areas were associated with greater discrepancies (and one could postulate that under or over votes on these technologies might have played a role), and extremely strong evidence that methodological factors likely to have made pollster avoidance easier were also associated with greater discrepancies.

But expecting E-M to actually take the iniative in setting up election monitoring sites is somewhat absurd. I'm all for such initiatives, but are you expecting this to be done pro bono? In any case, their poll, huge though it is, is simply spread far too thin (as it has to be to be done at all).

*whoopsies. Started to write one sentence and ended with another. That's the second time I've done that today.

[Sancho]

163. Ok...I've seen the red shift discussion and it makes sense....

Here's my take on it.

Everytime I do large samples, I'm going to follow up and target all those points that are outside some prescribed interval. In testing we're looking for cheaters or some Einstein with test anxiety. Regardless, we have to chase down the outlier.

To me, the shifts need to be explained.

Next, I don't think it would cost that much. Pro bono is an appropriate descriptor! If I had a few targeted races and a small crew of hot shot interviewers ready, I'd bet that SOMEONE would pay LOTS for evidence of what happened in that debated district or precinct! Got to be as cheap as $500 an hour lawyers! Even if the emergency crew showed up at noon after an early reported problem and handed out flyers for voters to record their exit interview online and to interview all they could, by 7PM they could get a hugh sample from a chosen precinct or two in a disputed district. Added to the evening report, that would make the rating go up for sure! Can't you see the Ohio or Florida races from 2004 with an extra in-depth result from 4 or 5 disputed precincts where 50 or 100 people were interviewed in those precincts. Generalizibility doesn't matter at that point, but what is happening in that precinct or district is critical. E-M doesn't need to take sides, just target the areas that need variability explained.

If E-M did it and put it on TV for ABC or CNN: that would be consistent with calling the election. They already have the system set up to collect the data and compare the results. Who else?

The "who voted and why data" at the precinct level is critical to explaining the outliers as very, very weird voters vs. election manipulation. That's something I would be interested in....heck, I think I'd put a couple of DRE's in the parking lot in a tent and ask voters to fill out some demographics and repeat their voting just like they did in the election! What a comparison for a hack! Can't you see a truck of pollsters showing up unannounced at precinct #9 with DRE's in the back. Watch the election officials and Deibold techie's sweat! That would be a much fun as anything I can imagine. What would you say then, that the undervote was due to not knowing how to vote? In Florida, I'd bet 75% would participate! What the hell, you're standing in line anyway.

[Febble]

164. Well, why don't you suggest it?

Anyway, thanks for bothering to read the links.

In fact, the shifts were well explained by methodological variables. About half the variance was attributable to sampling error, and a third of the remainder, and all the net shift, to methodological factors likely to have made pollster avoidance easier. A small amount of additional variance was accounted for by older technology in large urban areas; this may or may not have reflected higher residual vote rates in these precincts.

Cheers

Lizzie

[Sancho]

167. Now we're getting somewhere...and a little technical and more specific....

Twenty years ago, I used linear regression, parallel regression lines or related GLM techniques (discriminant analysis, etc.) for everything. The more research that I dug in and studied, the more I found hidden covariates and non-linear components (HLM)....etc...but it is a logical start and not a bad effort with the data available. I usually do that first.

Now, the "variance explained" would not totally satisfy me, but I realize we have different backgrounds. That's really why I'd have to see precinct data from the election officials (including races NOT discussed in the exit polls), the precinct poll data, and any demographic data on previous voting or registered voter stats that were available. I'd look for what I call "misfitting" precincts and their type of association with any "facet" of the data (polled or not). There have to be comparisons from the "non-important and not likely hacked" precinct races to the "possibly hacked races in the exit polls" at the precinct level. There have to be demographic profiles that are "predictible" vs. "not predictible" and how good are the profiles at predicting elections where we have demographics but no chance of hacking? Are there systematic but predictible changes in variability by precinct profile?
Finally, I'd use some pretty serious analysis (IRT) of the survey items and other predictors to detect lying, anxiety, bias, and any pattern of person misfit that was useful. I'd also estimte all the interviewers for "rater effect" techniques (Facets for example) to see exactly how much gender or age effect was evident. To me, each precinct would be like an item or rating of the candidate and the known information about the precinct would be used to decide if the precinct fit or misfit in a logical conjoint measure that led a precise or fuzzy decision about a particular race or candidate. Let's say that all the patterns were precise EXCEPT a particular race where certain precincts or races were misfitting and the type of election machine was the variable implicated as unpredicible (I made that up, but you get the idea). Hmmmm...what would be my pollster's plan for that precinct or distict next election? Why was that race misfitting?

I can't really say where I would end up, but my goal would be to identify outlier precincts by any factor or combination of factors that had a logical possible explanation: interviewer error or fraud or whatever. Once I started getting those patterns, I'd really look for combinations, ask certain questions in the poll, and hire interviewers that would catch the outliers in the next election to see what were flukes and what were manipulations. Two elections would likely do it, but who knows. By the second or third go around, I wouldn't expect to see an error that couldn't be spotted so quickly in the first 5 hours of incoming data so that I could send in the team of hot shots! No stone unturned.

I think the perspective difference is that pollsters tend to be more interested in "what can we get from what happened" and I'm most interested in "what can we do to make the system serve all purposes from the smallest unit of data". Perhaps that's not fair (it's not a criticism), but a point of view, since to me, the instrument and voter intent are the focus...not the sampling and "how big is your standard error" and "null hypothesis" focus. We've all fgured out that either fraud or badly designed voting machines are possibly causing the elections to declare the wrong candidate winner. I think it's time for pollsters and everyone else involved to detect the problems and fix them. I haven't seen this level of analysis reported, but I'm still looking at your links. The election officials will not fix things until it's force on them. Thanks, you are at least informative and civil.

[Febble]

176. I am completely confused

as to whether you are confused about what data was actually collected, or whether you are suggesting some mega-study at some future unspecified data, funded by some unspecified mega research fund.

Most of what you propose (where I understand it) involves data that was not collected ("lying, anxiety, bias"). What was possible was, essentially what I did. Yes, I use HLM, but the state-level power is very weak.

Essentially, all that was possible was a reverse-engineering job, to find out what caused a discrepancy between poll and count. It was thus limited to largely to data that was acquired in pursuit of a quite different question.

I think we'd better leave it there. PM me if you have any questions.

[Sancho]

154. This is where we disagree...

"It's not something polls could ever be designed to do, and this is why, of course, they aren't. It's why you need proper manual recounts and audits."

To be silly, if I had 109 people vote on a computer in a precinct, and 100 completed a 10 page interview and polygraph and waterboarding that said something different than the reported computer, that would be a poll that could convince me of fraud! In a single precinct of 109 voters, there would be convincing evidence at some point for whatever stratification of 100 out of the 109 that you happened to get that there was a problem with the computer, not the sample! It's like the guy who voted for himself and got a 0 vote reported for his name - there's got to be a problem outside of the sample. A report from the computer of 50 missing registered votes from a poll of 100 out of 109 is pretty convincing that something is amiss. Non-response can't be it....

If the power to detect the variability is not there because of the measure, sample size, sample method, etc...then there is no conclusion. BUT, even a small targeted and valid measure will detect convincing evidence of likely fraud in specific samples were the effects are large and therefore, detectable!

There has to be a poll that would sometimes be able to detect fraud, mistakes, or whatever issues was the aim of the measure. It's a matter of the practical time and money and design for a given purpose. As we've said many times, potential fraud is lost in the variablilty of the data, small representation at precinct level, etc. Pollsters explanations are non-response, interviewer gender, etc. for the noise.

So why not target 10 or 20 areas (like ones with a history of problems and easy access to lots of subjects) with specific forms of the exit poll and larger samples and follow up calls, etc. DON'T announce to the TV or politicians the details of the targets...and send in the troops for "pollsters shock and awe" on Nov. 7th? There only needs to be ONE precinct with evidence of clear fraud to bring down the house of cards. The reason for targeting instead of random selections is to get the detectable effect in that sample, NOT to necessarily generalize to everywhere else. That's why the polls accuracy will always be undetectable to hacksters who know the approximate exit poll sample size and methodology and plan manipulations within the SEM. There are some variables like the size of people showing up to vote and things, but there's a pretty good chance that manipulation would always be explainable as error UNLESS there was an unpredictible focus on a specfic subset of races or precincts!

After all the crazy elections, why is there NO reported attempt to put the issue to rest by selecting a subset to get an absolutely valid and reliable match to the final result within selected precincts, races, or districts? If there were such polling attempts, what were the results? I would think such a result might trigger the manual recounts and audits!

That's where we untimately disagree:
"It's not something polls could ever be designed to do, and this is why, of course, they aren't. It's why you need proper manual recounts and audits."

[Febble]

155. Well, I remain dubious

but if I really wanted to design a poll to detect fraud, then I suppose I'd do it along the lines you suggest. Indeed, it's the kind of advice I've been giving to people who have been thinking about this kind of poll. But a random precinct audit of the total precinct vote would still be better. Even a poll that aimed at interviewing 100% of the sample would be unlikely to net 100% response rate, so you'd end up with a volunteer sample, not a random sample. And the evidence is that attempts to boost response rates in exit polls actually increase the participation bias.

But sure, polls could be designed for the purpose. It's just not a purpose I think they are ever likely to serve well, and one for which there are far better (and far cheaper) alternatives. Do read my DKos piece.

[papau]

135. Hi Feeble - I see your doing your best to educate re memory problems in

any poll for what the respondent did in the last election.

I am curious on why the Dem vote is always remembered as larger than recorded - at least of late.

I have a built in distrust of the social sciences and their behavior studies, but I am curious as to why it is always the Dem vote that is remembered in excess of the recorded vote.

What correlations have been shown as to the bias, and do the Dems, or the elections, in 2000, 2002. 2004, 2006, have those characteristics?

[Febble]

137. Well, here is Mark Lindeman's paper

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

His finding is that it is the winner's margin that is overstated, whether the winner was Dem or Rep. Clinton and Carter both had increased retrospective margins, but, weirdly, so did Nixon, even though by then he wasn't even president.

It's curious, but it seems to be an extraordinarily pervasive effect, and therefore one that needs to be taken into account when evaluating its implications. But the authors of this paper don't seem to have considered it at all.

Oh, I'm FEbble, BTW, not feeble (it's from my initials: FEBL).

Cheers

Lizzie

[RaleighNCDUer]

28. And here I offer my personal apology to all those who said it

would happen this way.

I maintained that the computer fraud would give the repukes the election, but we should vote anyway. There were many here who insisted that a high enough turnout would overwhelm the fraud. It appears we were both right to a degree - the fraud was there, but they just couldn't steal enough to prevent the Dem victory.

I've never been so glad to be wrong.

[Raster]

36. To believably throw an election you've got to be within the accepted

margin of error. That way you can write off the anomalies. This time the point spreads were too wide. And bushco*'s dirty tricks couldn't be too obvious this year. There wasn't a decent, clear-thinking person in the United States that didn't already suspect cheating long before the first ballot was ever cast. And to boot, everyone already knew the Dems were coming in with subpoena power. This is why Allen was not screaming for a recount, or we didn't see a drunken Peter King slobbering on about counting the votes.

[roamer65]

45. I would add...GOTV in 2008!! We hafta make it unpredictable.

We'll hafta bugger their "system" good.:evilgrin:

[krispos42]

91. They tilted the margin of error 100% to the GOP side

It was enough in some races, not enough in others. It seems to me we should have at least one more Senate seat (go the ash-heap of history, Holy Joe) and a dozen or more House seats.

I thought we should have won more Senate seats anyway, because those are statewide and cannot be jerrymandered.

[seabeyond]

48. i didnt. after watching virg and mo, iknew once again it was happening

you could see the way the votes would shift, when they were doing it

[Febble]

29. Well, it's not really a big story

Essentially it's the same story as the too-few-Gore voters story in 2004.

Here's the response I made on Daily Kos:

http://www.dailykos.com/comments/2006/11/17/14169/950/9#9

for anyone who can't second guess what I would be likely to say.

But I do agree with this comment of yours: "This is huge, folks. This means that the Dems finally have the opportunity to challenge the election system without the "sore loser" label. And now it's our job to make sure they do it!"

I just wouldn't use this paper as part of your ammunition. There is far better data out there (undervotes; push polls; robocalls) and this stuff is probably wrong. Even if it's actually right, it's far too easy to refute. I just did.

[Pokey Anderson]

75. Febble, your refutations have caused huge commotions in the past.

Rightly or wrongly.

I hope people will read Bruce and Jonathan's piece, and send it around to competent pollsters and statisticians they know. See what they think.

Science relies on many minds and many proofs.

Bruce and Jonathan are two very smart fellows. I hope others inspect their data closely, and find it sound.

If they don't, I hope they can supply better methodology to use with the data at hand.

[Febble]

108. So do I

When it was presented on Daily Kos, I actually recommended the diary. I think it should be discussed. But I also wish it had been presented as a work in progress, not hyped like a major conclusion. It is a seriously flawed paper.

http://www.dailykos.com/story/2006/11/17/14169/950

Yes, Bruce and Jonathan are two very smart fellows, and I have reason to be rather fond of them both. But they have made a huge assumption that is not supported by exit poll data, which is that retrospective margins are accurate. Mark Lindeman looked at every presidential exit poll (and is currently looking at every mid-term exit poll) and found that in every single one, the previous winner's margin was inflated, regardless of whether that winner was running, losing, or even had resigned (Nixon).

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

They do not reject this data - they do not even appear to have considered looking for it, and that is an extremely serious omission. It also completely undermines their case.

It's an understandable error, because neither of them are experts in social science or statistical analysis (although Bruce is extremely numerate). But I don't, frankly, think that is an excuse. If you are not an expert, you consult experts, and they don't seem to have done so.

[Name removed]

90. Deleted message

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

[Febble]

109. Not true

Please read my other posts.

on edit: it is "not true" that I posted "a buncha shit" and also "not true" that I "pretended it was a refutation".

What I posted on DKos was the point that this analysis is based on an unsupported assumption that is actually not supported by previous exit poll data. This is not "a buncha shit" - it is simply true. It is also not a "refutation" nor does it pretend to be one. Their conclusions might well be true. But they are not the only inferences that can be made from their data, which they imply, and alternative inferences are better supported, given the evidence for the untenability of their assumption.

[Febble]

133. Apologies and clarification

on the "pretended refutation" - you are in fact, literally correct. I did "pretend" to "refute" the argument. What I did was to demonstrate what the counter-argument would be - it is dead easy.

It remains possible that the reason the retrospective margin increased is because there was fraud. But the fact that the retrospective margin of the winner always seems to increase means that there is absolutely no way of inferring fraud from such an increase, and that is what this paper does.

So the argument is easy to "refute" in the sense that it doesn't make the case it claims to make. But that does not mean that fraud was could not have been a factor. In that sense, I did not "pretend" to "refute" the conclusions of the paper.

And of course my post was not a "buncha shit".

[OnTheOtherHand]

136. call me cynical, but...

I think the apology should be coming from other quarters. (EDIT: risk of apparent call-out)

[Febble]

138. Well, he'd misunderstood my post

and I hadn't really unpacked the notion of "refute". It was no excuse for the language he used though.

[Catrina]

141. I disagree with your advice frankly

and will continue to use my own judgement as to what I decide is credible or not. Everything needs to be assessed.

'This stuff is probably wrong. Even if it's right, it's far too easy to refute. I just did.'

No, what you did was to raise a question about it. And in fact contradicted yourself by saying first 'even if it's right' and 'this stuff is PROBABLY wrong' ~ certainly not absolute statements. But then your absolute statement ''it's far too easy to refute. I just did'. Not meaning to be rude, but you admit yourself that you did not, what you did was TRY. Had you said 'I tried to refute it and may have convinced some people of the unreliability of this 'stuff', your position would be more acceptable.

Even as someone who knows little about this subject, although I am learning fast, as are many Americans now, I can see huge holes in the 'Republicans are shy' argument. Where's the proof first of all? And without proof, what I have to go on is just as believable, our own experience with Republicans, family members, friends, adversaries.

Republicans, especially righwingers, are not only not shy, their operatives seek out opportunities to skew the results of polls their way. Logically it makes ZERO SENSE to say that the same people who will miss NO opportunity that can be used to their advantage, suddenly become shy when a golden opportunity to affect the exit poll results is presented to them. Why this argument is even entertained, is laughable actually. And even if it were true, it was easily corrected, wasn't it? So, why was it not?


With no disrespect intended Febble, mainly due to the fact that it was so easily refuted, your argument is weak at best.

All material needs to be evaluated and tested, including this. I do believe now that we are getting closer to the truth as more and more Americans are becoming aware that their votes may be stolen.

When the American people get involved, they won't want to hear nuanced arguments such as yours. They will demand facts, machines confiscated and taken apart, source codes revealed etc. etc. This has already happened and unlike previous elections, I believe may be successful this time. It's hard when the people are involved and not just a few activists. The activists are not alone in this anymore. They have a growing, shocked and angry populace behind their demands now.

The only reason this subject (not even open for discussion on so-called Progressive boards to their shame) has not caused riots in this country is because it has been suppressed. The media now has little choice but to cover it. Americans on the whole are very impatient. Once alerted to a problem they want answers right away. They can be maddeningly oblivious for a while, but when they wake up, watch out.

I believe that election fraud will be revealed very shortly in this country, and the response to it will not be pretty. Anyone who attempted to cover it up will need to run for cover. Despite appearances, we do take our Democracy very, very seriously here.

[Febble]

147. Thanks for a civil response

Let me take it point by point. I should have been more careful with the word "refute".

Where the analysis is faulty is that it fails to note that its "objective yardstick" is not only not objective, but has a history of being biased - the retrospective margin of the winner of the previous election tends to be inflated in exit polls. This means that there is an alternative (and indeed likely) interpretation of their data - that after adjustment, the polls reflected a more likely retrospective margin. The authors do not consider this. At the very minimum they should have flagged it as a possible alternative interpretation. The data may be consistent with fraud (2006 may have been a rare year in which the winner's retrospective margin was not inflated, but seeing as Nixon's was, this seems to be an outside chance), but it is also perfectly consistent with what we know about retrospective margins. Any decent analysis has to consider - even if to reject - alternative interpretations of the data. This does not.

The sense in which I "refute" the argument is that I simply cite evidence to show that there is a strong alternative hypothesis. Given that their argument does not even attempt to consider the alternative, it is "refuted" as an argument that the data strongly suggest fraud. It is not "refuted" as an argument that there may have been fraud, but in that case it is a weak argument. A heck of a lot weaker than any of mine.

As for the "holes" in the participation bias account of the exit poll discrepancy in 2004 - there is lots of evidence for it. I explain some of it here:

http://www.dailykos.com/storyonly/2006/11/4/135126/905

also in my response here:

http://www.dailykos.com/comments/2006/11/17/14169/950/28#28, although it contains an error that was picked up by the following poster, for which I am grateful.

But the rest of your post makes sense. I am extremely keen to see verifiable, secure and transparent elections in your country. But I see absolutely no point in making the argument from exit poll data that does not support the case it is used to make. That is why this I find this kind of analysis, published without the kind of peer review that would immediately have picked up the confound, a bit irresponsible. The case for election reform is to good to dilute with bad arguments.

[Catrina]

165. Thanks for your response ~ I have followed your

link and read your analysis ~ it's impressive, obviously you have done a great deal of work on this.

The average person will not be focused on this kind of data however. The reason it has perhaps, influenced a number of people is only because the debate has remained between 'experts' on either side of the issue many of whom have focused in on a particular area, as you have. While their work deserves credit, it is not where the debate will focus once out in the general public. If anything, the work of one expert cancels the work of another.

But even if I focus on your work, this is what I see, as a totally non-expert on the subject. In relation to the 'shy Republican' argument, I see no definitive proof that this is the case, despite the excellence of your presentation. Why? There are too many 'ifs'. Here's an example:

Unfortunately, only non-response bias by visible characteristics can be observed.

You state that pollsters 'judge' by the appearance of those who refuse to respond by their age, ethnicity etc. I can't think of a less reliable method of judging someone's party affiliation than that, to be honest. This sounds like the 'law of averages', which is fine, but not really reliable in this case, depending on the area of the country eg, the times.

Here is how the average person is judging whether or not the likelihood of the past three elections 2000, 2002, 2004 and now 2006 were subjected to fraud.

Watch the face of someone who is just learning about this when they are told that private companies now count the votes in this country. That 80% of Americans vote on machines that are owned by Republicans! Even Republicans exhibit shock when given this information.

When 'glitches' such as voters being unable to get their choice of candidate to register correctly without trying three times, always favor Republicans, word gets around and serious questions arise.

When there are not enough machines provided in districts that are predominently poor causing voters to have to stand in lines for hours, there is a sense of outrage and no one misses the implications of this.

You no doubt know of all the other 'problems', ie, undervotes, missing votes, deliberate threats to minorities, misinformation given regarding where to vote, phone-jamming, mal-functioning machines, refusal by the machine owners to allow inspections of the machines etc. etc.

But when told that the machines on which they vote have been allowed to have proprietary software, that there can be no recounts in contested elections, it's hard to describe the outrage I've seen.

Add to that the exit polls switching as we watched in the 2004 election, and it's just one more piece of information to consider. The focus on this one of many problems, is like not seeing the forest for the trees to most ordinary Americans at this point. Even if they cancel out this one problem, they are left with all the others which are actually, facts! I personally know people who were unable to get their candidate's name to come up (naturally a Democrat) on their screen until after three tries.

Speaking of that ~ what happened to the other three tries? Did those votes register or not? Did their final vote register? Should a voter in a country like this even have to ask the question, and worse, should it be impossible for the voter to get an answer because of a 'secert' code?

While the experts have argued about the exit polls, slowly over the past two years, average Americans are learning this facts and I can tell you, forget about Internet Forums, in Real Life there is outrage, not only that these problems exist, but that they were suppressed by the media and by those who should have known better.

My own family are mostly Republicans, as are many of my friends. But they are Americans first and this has shocked them as much as it as it shocks me. That it is six years later and that this election has been as much of a mess as it is, is even more shocking.

We cannot be as benign as you can about this with all due respect. It is more than possible that this shameful era of American history with its disasrous results for so many people could have been avoided had the American people not been kept in the dark about his issue.

Not to diminish the work you have done, but it has done nothing to help restore Democracy to this country. Only the outrage of the American people will do that. They made a good beginning on Nov. 7th, now hopefully, we will get to the bottom of this and many other things.

When demands are made to confiscate machines this time (as they have been) it will be a lot more difficult to refuse. Personally, what has shocked me the most has been the suppression of discussion of this most important issue on many progressive boards. It has caused many of us to wonder why, and no, the CT argument doesn't explain it. They have lost a lot of trust as a result. The goal this year was to restore the Democratic Party to power and many chose to remain silent in order to accomplish that goal. Now that this has been accomplished, and with the election itself still a mess, none of the problems resolved, I doubt any kind of suppression will be acceptable from now on.

Anyhow, thank you for work ~ I'm sure it will be considered along with all the other work done by those who care about this country. They will be receiving far more support from now on ~

[Febble]

166. Thanks for getting back to me

and for following the links!

One little thing to clear up:

You say:

You state that pollsters 'judge' by the appearance of those who refuse to respond by their age, ethnicity etc. I can't think of a less reliable method of judging someone's party affiliation than that, to be honest. This sounds like the 'law of averages', which is fine, but not really reliable in this case, depending on the area of the country eg, the times.

No this is not what it is done. Sorry if I wasn't clear. What happens is that for each respondent, one of the questions is age, sex and ethnicity. And for those who don't respond, interviewers also note (by eye) age, sex and ethnicity. Then, when the pollsters look at the responses, if they find that, say the ratio of women to men in the responders is 55%:45% and in the non-responders it is 48%:52%, they know that they have an unrepresentatively large proportion of women in their sample (in other words, that men were less likely to respond - were more likely to refuse - then women). Ditto with age and race. So what they then can do is upweight the responses of the men in the sample. Each man, for example is treated as, say 1.2 voters, to make up the difference. Or each women is treated as .8 of a voter.

And from these data we know that there is really "non-response bias" - the "non-responders" have different characteristics as a group (maybe contain more men) than the responders. However, as party ID, or who you voted for is an invisible characteristic, you can't weight for these characteristics. It may help a bit, however - if men are more likely to vote Republican (yes) and less willing to participate in the poll, upweighting the men will tend to reduce the pro-Democratic bias. But if Republican men are more likely to refuse than Democratic men, it won't help much.

Is this clearer?

Anyway, much of what else you write I agree with. I just don't think the problem was on the scale indicated by the 2004 exit polls (and I think that is the simple answer to the question as to how the Dems were able to win in 2004!). It bothers me that the worst arguments (the exit poll arguments) have often been the most prominent. I remain convinced that the biggest loss to Kerry was various forms of voter suppression, and residual votes on punchcards. But whether or not votes were stolen by DREs, they are dreadful and outrage is entirely appropriate. They need to be abolished. I hope they will be. I am tremendously encouraged by the Dem win. I'd just hate the campaign to be sidetracked (yet again) by the exit poll red herring!

Cheers

Lizzie

[Catrina]

168. Yes, that is more clear and obviously I misunderstood

what you wrote. However, now I have more questions. The argument that Republicans were undercounted because they may not have responded has been used to explain the discrepency in the 2004 exit polls. But your response to my post seems to say that pollsters compensated for this assumption. Iow, if a larger number of white men refused to respond than did respond, the pollster then counted those who did as 1.2 men.

1) Are you saying that in 2004, this methodollogy was used but since the results (mostly in swing states as I recall) did not correspond to the adjusted poll result, the pollsters' methodology was wrong?

2) If so, then why not simply use their original results?

3) Did they do this, and if so, did the polling then more closely correspond with the actual results?

4) Is this why they did not want to release their data? Because they readjusted it to match the results and were afraid of a) Being accused of using faulty premises (1.2 men) and/or 2)worse, fixing the data to fit the results?

As far as exit poll arguments being the most prominent case for fraud, I don't know. Maybe in online discussions. In real life, it is the machines that most people are concerned about in my experience and I have not heard anyone mention exit polls. By a large majority, people do believe that in the probability is that several elections have been stolen through use of voting machines, and as a result, want them removed from the process.

Btw, I still do not buy that Republicans are less likely to respond to pollsters. I still see absolutely nothing to prove that. My instincts are that this crosses party lines.

The media has attempted to divide the country into Right and Left, true, but in real life, even families are made up of people who vote for both and even third parties ~ I can say that there is nothing about the people I know that would make them more or less responsive to a pollster based on party affiliation. It would be based more on personality, some more patient than others with being asked questions by strangers.

Americans are Americans, like everyone else. They are not defined by their party affiliation, except for a minority who are deeply involved in the political process. Before Bush, we tended to be interested in the election up to the point the results were in. After that, even if our candidate did not win, we accepted it and went on with our lives. This changed because of the current administration and has swung now to a majority of the people wanting a change in govenrment, and a uniting of many different people to accomplish that goal.

[OnTheOtherHand]

170. a point on weighting

A while ago I assembled an example to show, in somewhat gruesome detail, why weighting to the official returns is expected to yield more accurate results -- even if the official returns are somewhat erroneous. (Of course, if there is actually massive hacking, this won't be true.)

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=2490768&mesg_id=2499695

That's why they weight to the official returns. You were also asking about age/race/sex weighting, and other things, but I won't try to answer for Febble -- just thought you might find the example interesting.

[Catrina]

173. Thank you ~

Of course, if there is actually massive hacking, this won't be true.)

Yes, which is why many believe that the machines may be only one of the many ways elections are stolen. Iow, they may be programmed to alter a small % of the vote on the assumption that the country is pretty evenly divided. As insurance, other means are employed. If this is the thinking, it would explain why it didn't work this time. The country is no longer evenly divided. It doesn't mean though, that there was no attempt to change the vote, just that it wasn't enough to overcome the anti-Bush vote.

Watching Karl Rove's interview before the election, he seemed to believe that the polls showing Dems leading were wrong. He claimed to be looking at different polls that the public had no access to, with more favorable results for Republicans. If they believed that, that would explain why they did not adjust their calculations.

[OnTheOtherHand]

175. well, OK

There are reasons that Febble and I (and, for what it's worth, almost all political scientists with an opinion, as far as I can tell) think that widespread vote switching on machines (on the order of millions of votes) didn't occur in 2004. I know less about 2006 so far, but provisionally I have the same opinion: I haven't seen good evidence that millions of votes were stolen nationwide. That isn't to rule out anything in particular races.

When I talk about "vote stealing," generally I'm talking about things that could show up in exit polls -- circumstances where people were under the impression of having voted. Robo-calls (for instance) don't count as "vote stealing" in this sense, but they certainly count as an attempt to steal an election.

Personally, I have a hard time believing that Karl Rove would fake himself out by looking at the wrong polls. I think Karl Rove is overrated, mostly. But I don't know.

It's good to 'talk' with you.

[Febble]

174. Well, here are some possible answers:

1) Are you saying that in 2004, this methodollogy was used but since the results (mostly in swing states as I recall) did not correspond to the adjusted poll result, the pollsters' methodology was wrong?

Not exactly. As I tried to explain, the weighting for non-response bias by visible characteristics may reduce non-response bias by invisible characteristics (e.g. candidate choice) but only if the invisible characteristics do not "interact" with the visible ones. For example, if men are less likely to respond than women (yes), and men are more likely to vote Republican than women (yes) then if over-sampling women is your only source of pro-Democratic bias, upweighting the men in your sample will eliminate it. However, if your male responders are more Democratic than your male non-responders, then weighting up the men in the sample won't compensate for the overall shortage of Republicans.

In addition, non-response bias is not the only form of bias, and probably not the major one in 2004 - selection bias appears to have been a bigger factor - Democrats were more likely to be selected (or Bush voters were more likely to avoid selection). If this is the mode by which bias is introduced, the missing voters won't be recorded at all, and, ironically, reported completion rates may be high. in 2004, some reported completion rates were higher than they should have been, given the size of the sample and the interviewing rate. This suggests that some voters were avoiding the selection process altogether.

2) If so, then why not simply use their original results?

Well, as I said, the pollsters actually expect bias in their poll, although they can't quantify it in advance. And they assume the count is correct (they are not, as I said, attempting to provide an independent check on the count). So when their poll sample is found to systematically diverge from the count, they assume that their sample was biased. This happens in survey research all the time. You attempt to get a random sample. You find that the incidence of something in your sample (income; age; gender; race) is not equal to the incidence you know exists in the population you are sampling from (from census data, for instance). So you reweight your sample so that it more closely matches the known characteristics of the population to which you want to generalize the findings from your sample. In the exit polls, the pollsters use the vote count - because, in principle, it should be the most accurate source of data against which to match their sample. Of course the concern is that it isn't.

3) Did they do this, and if so, did the polling then more closely correspond with the actual results?

I assume so. What we don't know is exactly what the weights for age, race and sex of non-respondents were, as we only have the final weights. These would also have included weightings for other things, such as the geographic strata from which the precinct sample was drawn. But we do know that some groups were significantly undersampled (I can't remember which).

4) Is this why they did not want to release their data? Because they readjusted it to match the results and were afraid of a) Being accused of using faulty premises (1.2 men) and/or 2)worse, fixing the data to fit the results?

No, not at all. They weren't hiding their methodology - it was all laid out in their FAQ on their website, and it's they way they've done it for years. There are three things to bear in mind regarding data release: one is that in any survey, the data actually belongs to the clients (in this case, the NEP) and so release is up to the clients (they paid for it); secondly, one of the important things that Warren Mitofsky insisted on was that all raw data (including weights) should be placed in a public archive, and it is. The 2004 data was actually available for free download for over a year. Thirdly, there is an important issue regarding participant confidentiality. Because detailed responses are included in the archived datasets, including highly personal information regarding not only age, race and sex, but also marital status, sexual orientiation, religious beliefs, political ID, and of course, vote, it is essential that the data cannot be matched to real people. For this reason, no precinct identifiers are given, nor are vote totals for the precincts (otherwise precincts could be identified from the vote totals). In fact "blurred" precinct vote-shares were prepared and release for Ohio, and were analysed by the Election Science Institute. I had understood that more might have been commissioned, but I do not know what became of this project. As you probably know, Warren Mitofsky died suddenly a few months ago.

As far as exit poll arguments being the most prominent case for fraud, I don't know. Maybe in online discussions. In real life, it is the machines that most people are concerned about in my experience and I have not heard anyone mention exit polls. By a large majority, people do believe that in the probability is that several elections have been stolen through use of voting machines, and as a result, want them removed from the process.

I couldn't agree more. And I am delighted that you came to that conclusion independently of the exit poll evidence, which merely suggests the scale of the problem, and, IMO, suggests it spuriously.

Btw, I still do not buy that Republicans are less likely to respond to pollsters. I still see absolutely nothing to prove that. My instincts are that this crosses party lines.

It is a difficult thing to study directly, as by definition we do not know the views of non-responders. However it can be strongly inferred from a number of studies, and was indicated directly by a pre-election poll last month that found that Republicans were significantly less likely to express willingness to take part in an exit poll. Arguments from incredulity occasionally have force, but this one is contradicted by substantial evidence.

The media has attempted to divide the country into Right and Left, true, but in real life, even families are made up of people who vote for both and even third parties ~ I can say that there is nothing about the people I know that would make them more or less responsive to a pollster based on party affiliation. It would be based more on personality, some more patient than others with being asked questions by strangers.

Well that is an excellent point, but it is not really relevant to the question of differential participation. No-one is saying that people systematically check their own party ID before deciding whether to participate. All that is hypothesised is that people who are unwilling to respond to exit polls (and that was about 35% of the electorate in 2004) are slightly more likely to vote Republican than Democratic. But your point is one that is often missed on political forums - nobody on DU can understand, for example how anyone could have forgotten voting for Gore. But people do forget who they voted for (hard evidence from longitudinal panel studies), and I expect those people are the people who didn't care that much one way or the other.

Americans are Americans, like everyone else. They are not defined by their party affiliation, except for a minority who are deeply involved in the political process. Before Bush, we tended to be interested in the election up to the point the results were in. After that, even if our candidate did not win, we accepted it and went on with our lives. This changed because of the current administration and has swung now to a majority of the people wanting a change in govenrment, and a uniting of many different people to accomplish that goal.

I agree. It happened in the UK in 1997. It's gone a bit sour since, but finally, our rightwing party has moved to the left, which is interesting. I hope the Labour party will shunt itself back in the same direction too!

Thanks again for responding.

Lizzie

[Catrina]

183. Febble, first, I want to thank you for your

patience in responding to what must be questions asked and responded to do thousands of times, by you and others. I did not pay much attention to the exit-poll debate, largely because, to put it mildly, numbers are not my forte.

I was not aware that Warren Mitofsky had died ~ that is sad news especially for his family and friends.

So when their poll sample is found to systematically diverge from the count, they assume that their sample was biased.

This statement struck me because one of the things I did wonder about at the time, was Mitofsky's seeming willingness to take the blame for the exit polls apparently being so wrong in 2004. Having read your answers to my questions which explain how pollsters reach their conclusions, it seems he was being honest and was assuming the vote count was accurate. I also see though, why this caused so much controversy and how difficult it would have been for him to explain in the climate leading up to the 2004 election. Granted because of there being no way to recount contested elections (the real travesty imo) he really had no option but to accept the results and conclude that his own findings were faulty.

What a position to be in. I am one of those who was watching the returns that night and was certain at approx. 6.00 that Kerry had won. Many of us went off to celebrate. But then, the numbers began to change, as did the exit poll numbers on the screen. Adding to the suspicions this naturally aroused, the media refused to answer questions regarding the exit poll numbers. I do object therefore to the term 'conspiracy theorist' since people watched with their own eyes, what seemed like a complete reversal of the results and never got a satisfactory explanation.

Two years later, you have explained this to me ~ (ie, that the vote count itself is used). Ordinarily this would put to rest any suspicions that there was an attempt to 'adjust the exit poll numbers to fit the results' once it was established that nothing was done differently than before.

But can you see why, even when this is understood, in the atmosphere of suspicion that already existed, many people saw it as an attempt to justify results that were contrary to most indicators before the election, and especially since the Bush administration had lied so consistently to the American people?

Iow, this is too fine a point to overcome the suspicions (which I believe are justified) and the anger that another election may have been stolen. Sadly for Mitofsky, some people suspected him of facilitating the theft of another election. Logic seemed to say that he should have questioned the vote count himself, even if only to protect his own reputation. But I do see how, considering his line of work, he could not do that honestly without proof, regardless of what he may have thought privately. I wonder what he did think?

I imagine the quandary for him and other pollsters in the aftermath would be if the vote count was wrong (stolen election) then his methodology was good. But if not, then how would they compensate for possible fraud in the future? And did they allow for that possibility when preparing to poll the next election?

I do agree regarding the exit polls, while still inexplicable to many (ironically because of respect for Mitofsky's work) we do not need them to prove fraud.

I do believe that there was fraud ~ (even today I see the news of the Florida election controversy that may result in a demand for a new election) I think when it is proven, Mitofsky's work will be validated.

But your patience in responding to my questions has clarified so much for me, particularly in relation to Mitofsky. I have more questions, (such as why adjust the numbers with the actual vote count?) but with the foundation you have given me, I can do some research with a better understanding of what I'm looking for. I'm sure we'll talk again, I do comment at DK also but missed your posts there on this topic.

[Febble]

185. Well, just one other point

worth making, is that the discrepancy had been almost as large in 1992, but no-one much noticed at the time, because Clinton was way ahead anyway. So having to do fairly large adjustments to the vote count was not unprecedented.

And the other thing to say, re Mitofsky, which you can read in my sig, is that he hired me to do further analyses on the data (can't remember if I said that before), knowing that I'd been critical of his first report, and knowing that I'd be looking for fraud. I never met him, but got to know him a bit by email. I still have an unsent email to him in my inbox, asking about the report I did for him.

He really wanted to know what went on in that poll, and I had a carte blanche to test any hypothesis - including fraud - that I needed to. And I did.

Nice talking to you!

Lizzie

[Catrina]

188. I wonder if he was ever satisfied ~ sad

that we'll probably never know ~

I did notice in another post that you had done some analysis for him ~ it sounds as though you never did learn what he thought about it. Disappointing for you also, I'm sure.

Thanks again for your time ~ I hope one day we get to look inside those machines. Maybe this year if all goes well ~

trina

[nam78_two]

33. K and R!

Go for it Dem leadership :bounce: :bounce: :bounce:

This is HUGH11111seriesley

[mom cat]

34. I wonder how many of the Dem "losers" were liberal/progressive.

[SaveAmerica]

37. Hmmm, skewed 4%, Kissell was up 4%+ before the election then

on election day there are only 450 votes between them. The poll I'm talking about was on electoral-vote.com (NC-08). Still waiting for the provisional ballots to be counted. I can't wait to see what a recount would reveal.

[happydreams]

38. I sent this baby to my local newspaper.

[DoYouEverWonder]

39. So now that the Dems are back in charge

Can we just throw these machines away and go back to pen and paper?

[No DUplicitous DUpe]

52. And besides being able to truly recount, There is no need for long lines..

..another way the machines suppress the vote, if you have enough pens available to fill out the ballots with!
edited to add:
(I am featured in this story)
El Jebel gives electronic voting thumbs down -Aspen Times

Delays drove voters away

By Scott Condon
November 10, 2006

Eagle County's top election official said she wants to avoid future delays like those that plagued polling places such as El Jebel in Tuesday's election.

Eagle County Clerk Teak Simonton (R) said she wants to purchase additional electronic voting machines, which were put into service this year. Extra machines are needed in the busier precincts "and El Jebel certainly qualifies," said Simonton.

Voters waited up to an hour to cast their ballots at El Jebel, according to Ken Ransford and Jon Barnes, poll watchers for the Democratic party. Ransford said that created problems for some voters. One man had to leave to pick up his kids after school, according to Ransford. He was uncertain if that voter ever made it back to the polls.

Another man tried to vote in the morning but left to go to work because the line was so long, Ransford said. That man returned at 7:02 p.m. but was turned away by election judges. State law mandates that people must be waiting in line as of 7 p.m.

Ransford, an attorney, felt they should have made an exception since the man tried to vote in the morning but couldn't remain because of the long line.

The clerk's office set up five electronic voting machines in El Jebel. Three to five stations were also used over the course of the afternoon to allow people to use the old system of marking a paper ballot, according to witness accounts.

Barnes said more paper ballot stations were needed and could have easily been set up once it was clear people were waiting so long. Barnes estimated he saw 20 people waiting in line to vote at 3:30 p.m. People were taking eight to 15 minutes to complete their ballots, he said.

Barnes felt the delay was "artificial" and could have been avoided with more paper ballot stations.

"I didn't see the logic of making everybody stand there 30 to 40 minutes to vote," said Barnes. "It was a disincentive to vote."

He said he urged the election judges to establish more paper ballot voting stations since they had plenty of space. Judges said they were instructed to emphasize the electronic voting machines and that the paper ballot stations were only secondary.

Ransford renewed the request to set up more paper ballot stations when he came to poll-watch at 5:30 p.m. Eventually, there were 13 paper ballot voting stations in use as well as the five electronic machines, he said. By about 6:30 p.m. the lines started disappearing.

He credited the election judges for responding to pleas for more paper ballot stations, but also questioned why adjustments weren't made earlier in the day. Like Barnes, Ransford suspects the lengthy delay discouraged some people from voting.

"Why did it take so long to react?" Ransford asked. "The bigger issue is, we ought to be making it easier to vote, not harder."

Simonton agreed that "we need to re-evaluate" what is a reasonable amount of time for a voter to wait. But she was adamant that the proper solution is buying additional electronic machines rather than mixing electronic and paper voting. The duel system created headaches for the people counting the votes, she said.

Other polling places in the middle and upper valley didn't experience the severity of delays reported in El Jebel, and some voters complimented the electronic voting system.

"It's great. It's cool," said Missouri Heights rancher and former U.S. Congressman Mike Strang after he voted Tuesday on the only machine in use at the Carbondale Town Hall precinct. He said he found it easier to use than a paper ballot, particularly in that it allowed him to review his votes electronically before pushing the "cast vote" button, and then to see them on paper to make sure it was all correct.

Others, however, were not so positive.

"I won't go through those self-checkout lines and I won't . I want to mark a piece of paper," said Carbondale resident Brad Hendricks.

An informal poll of voters leaving the polling places up and down the valley showed that most who picked paper ballots over the electronic machines did so mainly because using paper seemed to go faster.

Poll watchers reported no major delays in Basalt. Two Basalt High School juniors, Jannae Swanson and Megan Southward, were available to help voters confused by the electronic machines.

Donna Linnecke, the chief election judge at Basalt, added paper ballot stations as needed, noted poll-watcher Gerry Terwilliger. "I have nothing but praise," he said.

Linnecke noted that "our preference is for to go electronic," just as it was in El Jebel.

Pitkin County Clerk Janice Vos Caudill said election day went pretty smoothly, especially considering the turnout was about 52 percent.

On election day, election judges at Precinct 1 reported that voters seemed happy with the machines.

"It's working out really well," said Cindy Christensen, a precinct election judge, noting that even voters who were skeptical initially ended up voting electronically after learning that the machines all had paper backups to be used in recounts.

[0rganism]

64. I love this rancher guy: "It's great. It's cool"

"It's great. It's cool," said Missouri Heights rancher and former U.S. Congressman Mike Strang after he voted Tuesday on the only machine in use at the Carbondale Town Hall precinct. He said he found it easier to use than a paper ballot, particularly in that it allowed him to review his votes electronically before pushing the "cast vote" button, and then to see them on paper to make sure it was all correct.
---
Yes, former Republican Congressman and wealthy rancher Mike Strang got to vote in what was probably a damn-near deserted precinct at the time on one of these computerized election stealers, and he's happy because it allows him to review his votes?

This machinery probably cost five thousand dollars, and he's overjoyed that he doesn't have to look over both sides of a piece of paper.

What a maroon.

[EFerrari]

40. Wow. K&R

:wow:

[texpatriot2004]

41. Are "undercount" and "undervotes" connected? I would think so.

Anyone care to guess?

[Febble]

46. There is strong evidence for undervotes

but the "undercount" inferred in this paper, is, unfortunately, inferred from an analysis based on a faulty assumption.

So, if they are connected, it is pure coincidence. I agree with EDA that an investigation should be launched into the corruption in 2006. But to use this argument to do so is to risk jeapardizing the case in my view. I'd start with undervotes in Florida, push polls in Maryland and Tennessee and robocalls everywhere. Then investigate every reported anomaly in every race.

But this exit poll story, like the 2004 exit poll story, is a red herring.

[roamer65]

42. We now are getting an idea of their workings.

Hmmm...interesting. I trust the exit polls MUCH more than the machines, especially since I found out Germany uses exit polls to announce election results.

[In Truth We Trust]

43. K&R Hand Counted Paper Ballots NOW!

[Mnemosyne]

44. I want a recount of 2000, 2002, 2004 and this year. Now. n/t

[pola]

49. CALL YOUR REP - VOTE FOR HR 6200, paper ballot bill !

Kucinich's paper ballot, hand counting BILL - CALL YOUR REPS today to vote for this bill !!!

[pola]

50. THROW THE MACHINES INTO BOSTON HARBOR !

[Straight Shooter]

55. Nah. Dump them on the Imperial Hog Farm in Crawford, Texas.

:D

[nznow]

58. Carville wanted Dean to put our money where the machines are.

I've said it several times because I think its important to note. They wanted Dean to waste money in the touch screens areas that could be flipped. The only way around this was the fifty state strategy. We were supposed to fail in 06 so Hillary could win in 08 and so Bush could finish the agenda..It's no wonder the 50 state strategy is under attack. It makes the outcome harder to control.

[NMDemDist2]

82. welcome to DU nznow!

:hi:

or are you a retread? either way HI!

[diva77]

51. I hope they have a chance to look into the initiatives in CA that were

close -- especially Prop. 87 and 89 that lost.

[mzmolly]

53. I read an article by Greg Palast recently called "How They Stole The Mid-Term

Election."

Theft #3: Votes Spoiled Rotten

The nasty little secret of US elections is that three million ballots are cast in national elections but not counted — 3,600,380 not counted in 2004 according to US Election Commission stats. These are votes lost because a punch card didn’t punch (its chad got “hung”), a stray mark voided a paper ballot and other machinery glitches.

Officials call it “spoilage.” I call it, “inaugurating Republicans.” Why? According to statisticians working with the US Civil Rights Commission, the chance your vote will “spoil” this way is 900% higher for Black folk and 500% higher for Hispanics than for white voters. When we do the arithmetic, we find that well over half of all votes spoiled or “blank” are cast by voters of color. On balance, this spoilage game produces a million-vote edge for the GOP.

That’s where the Black Boxes come into play. Forget about Karl Rove messing with the software to change your vote. Rather, the big losses occur when computers crash, fail to start or simply don’t respond to your touch. They are the new spoilage machines of choice with, statistically, the same racial bias as the old vote-snatching lever machines. (Funny, but paper ballots with in-precinct scanners don’t go rotten on Black voters. Maybe that’s why Republican Secretaries of State have installed so few of them.)

More here: http://www.gregpalast.com/how-they-stole-the-mid-term-election

Thankfully, Mr. Rove's "math" was a bit off.

[Kurovski]

54. They won't ever stop on their own. They will never stop, ever.

Let's get rid of the machines.

I thought it would be MORE than 3 million.

[Zambero]

56. So this was Karl Rove's Ace-in-the-hole!

Surprise, that's 4-for-4 highly questionable elections beginning with Bush stealing it in 2000. Gerrymandering, RW media bias, skewed election results, voter intimidation, massive fundraising superiority, relentless negative campaigning, divide & conquer manipulation of social "wedge" issues -- all part of a master plan for the GOP to retain power forever? Yes, if it all goes like clockwork. When the clock went haywire and "too many" voters start identifying the bullshit factor as it affected them directly, things started to change. What we're seeing with the Dems regaining power is a rare opportunity to right some wrongs with the electoral process, and hopefully preventing them from ever happening again.

[dchill]

74. Ace-in-the-hole...

turned out to be a deuce of clubs.

[go west young man]

57. We need a brave dem house member who won by a landslide to

call for a recount anyway and show that the vote tallys don't match. Who says the winner can't ask for a recount? If we expose fraud this way it shows guts and integrity.

[donkeyotay]

60. This is consistent with the post-election spin of the media

Dems won on a "razor thin margin," better not push too liberal agenda, candidates were actually conservatives, Pelosi's the lame duck, etc...

We have a system of “faith-based” voting where we are simply asked to trust the integrity of the count produced by the machines that tally our votes, with little if any effective checks and balances. In the context of yet another election replete with reported problems with vote tallying,18 the continuing mismatch between the preferences expressed by voters as captured in national exit polls, and the official vote tally as reported to the public is extremely disturbing.

If you can't outright steal the election because suspicion is high, the least you can do is claim it's so close that the spin is that we practically re-elected Poppy so that nothing gets done.

Remember that Gore won. By 2002 it was a shocker that the pubics won, and by 2004 people were going to make sure the Democrats won. Yet, still it was sooo close and oooh the repos won again. Now in 2006 we're still hearing the same thing: the country is evenly split even though the Democrats have been registering more people and have been more and more motivated...yet can never pull ahead. Interesting.

[David Zephyr]

61. A group of us were chatting about how the exit polls (again) favored Dems

This is really getting old.

Here's hoping that our great Democrats in both houses will now, once and for all, take control of our election system. I have confidence they will.

[jgraz]

72. Just show them this line from the study

"In order to match the results of the official tally, the 2006 exit poll adjustment was so extensive that it finally depicted an electorate that voted for Bush over Kerry by a 6% margin in 2004: very clearly an undersampling of Democrats and an oversampling of Republicans"

[Febble]

105. Well, it's one of the many lines

that are wrong. The first part is correct; the bit after the colon doesn't follow. The "retrospective margin" of the previous winner is routinely inflated, regardless of the size of the "adjustment" required. Even Nixon's was, and he'd resigned by the time of the next exit poll. So if the "unadjusted" poll showed zero inflation, it would actually, if anything, be a reason to suspect undersampling of Republicans, not an undersampling of Democrats. The data simply do not support the inference made by these authors.

[The Count]

62. So...Not dean's fault then? Okey d-okey. They didn't miss a trick!

voter supression, exiot polls, machines hacked - Rove's math at work.

[Senator]

63. It was "only 3 million" because of Rove's Targetting

There was no need for a "national popular vote" smokescreen, like in 2004.

And remember, Rove was prepared to "study" into submission anyone who challenged "THE Math."

From this Newsweek article:

Rove placed so much faith in his figures that, after the elections, he planned to convene a panel of Republican political scientists—to study just how wrong the polls were.

We need to "convene a panel" for impeachment.

Nothing else will stop these traitors.

--

[stepnw1f]

65. My Favorite Line:

"....a national election integrity organization, issued an urgent call for further investigation into the 2006 election results and a moratorium on deployment of all electronic election equipment, after analysis of national exit polling data indicated a major undercount of Democratic votes and an overcount of Republican votes in U.S. House and Senate races across the country."

[Raksha]

66. Remember what we were all telling each other before the election?

"We need to really get out the vote this time. We have to win by margins too big for them to steal." Now I truly appreciate the wisdom of that approach. So it turned out to be literally true: we needed to win by margins too big to steal--AND WE DID!

[BootinUp]

68. exit polls are not a suitable means

to audit an election, sorry.

If you want an audit, then push for those reforms.

[eridani]

77. And just because you found a corpse in your back yard--

--does not by itself prove that there was any foul play involved. It is damned well grounds for an investigation, though. The best secondary data is still secondary data, which is all that polling can be. Proof of wrongdoing? Absolutely not. PROBABLE CAUSE?? Fuck yes!

[BootinUp]

79. 4% is not a figurative corpse

its within normal polling limitations.

[eridani]

96. If the variation is always in the same direction it is

A 4% variation with points above and below the mean is within normal limitations. That the 4% is always at the expense of Dems is a smoking gun.

[BootinUp]

102. So you are arguing against the various

causes of errors in polling? There is always some amount of error in a sampling that could go either way. Exit polls have never struck me as being highly representative.

Read this report, you find there are quite a number of adjustments and assumptions that are made to the data. In the end what you have is a fair approximation of the actual vote nothing more.

And I don't know what you mean by always at the expense of Dems. Sounds like a conspiracy that involves people with no reason to skew the results.

[eridani]

110. No, just saying that errors in polling that always go against the Dems--

--are inherently suspicious. Particularly if there are also reams of reports of machine malfunctions that almost always favor Repubs.

[Febble]

113. Yes, this argument

is legitimate. It is why it was really worth checking out the 2004 exit poll evidence pretty thoroughly. But it is not the argument made in this paper.

[BootinUp]

120. We need to fix or eliminate the machines

not claim we were wronged based on flimsy evidence. Thats my feeling on it FWIW. I'll see you around. :)

[Febble]

126. Well said.

There is plenty of non-flimsy evidence.

[Febble]

103. It is a smoking gun

that non-random factors are at work. It is not a smoking gun that that factor was fraud. The evidence is much stronger in favor the role of selection bias in the poll, than for fraud, and this particular analysis is flawed at a fairly basic level.

[eridani]

107. It is impossible to know what random factors are at work--

--unless you can examine voting machines and their software. The basic flaw is tha poll data isn't the real thing, but only secondary data.

[Febble]

112. Yes,

That is a good argument for examining voting machines and their software - because it is impossible to know what is going on otherwise. It's irrefutable, unlike this poll argument.

[Cookie wookie]

132. Thanks Febble for your contributions here.

I can't add to these discussions, since I don't have the knowledge and background needed to assess the arguments, but I do read them. Thanks for continuing to share your expertise with those of us who value science.

[Febble]

134. Well, I think reality-based argument is

important.

Thanks.

[Peace Patriot]

81. Exit Polls are used worldwide to verify elections and check for fraud.

They are especially useful in conditions in which the "official count" is not truthworthy. Like here.

And this is one of the ways that Edison-Mitofsky--the consortium pollster for all the war profiteering corporate news monopolies in 2004--let us down. The conditions of the vote count in 2004 were egregiously non-transparent--conditions created in a very short time, 2002-2004, by the cancerous spread of untested, insecure and extremely insider hackable electronic voting machines, run on TRADE SECRET, PROPRIETARY programming code, owned and controlled by Bushite corporations (fast-tracked around the country by a $3.9 billion electronic voting boondoggle from the Anthrax Congress). Edison-Mitofsky should have objected to these non-transparent conditions and used their exit poll capability to FIND THE FRAUD. Instead, they acted as apologists for the election thieves.

[BootinUp]

83. correction... they are misused then.

If participation in exit polls is voluntary and there is a high percentage of people who decline to participate the sample is placed in doubt to the same degree as telephone polls.

[Febble]

101. No they are not

Please give an example.

[Einsteinia]

152. UNTIL these pollsters reveal their raw data, then

their feeble attempts at excusing their anomalies are going to be indefensible.


Show us the raw data! It's the scientific method that all the facts be transparent.

Until then the "reluctant responder" and "oversampling of dems" explanation sound like propaganda.

[Febble]

153. Sigh

the raw data was released in January 2005, together with a 77 page report of an exhaustive analysis. More has since been done (by me, as it happens) and that has been reported too. The findings support the hypothesis that differential participation rates were a major contributor to the discrepancy. They do not support the hypothesis that fraud was. You may not like that conclusion, but it was, nonetheless, supported by the data.

See here for other details:

http://www.dailykos.com/storyonly/2006/11/4/135126/905

on edit:

and here:

http://www.dailykos.com/comments/2006/11/17/14169/950/28#c28

[Einsteinia]

157. nyet! redux: Pt. 1 & 2

Respectfully:

1. As to your explanation on the DailyKos as to the true function of exit polls:
This obvious obfuscation of the facts on what exit polls may be used for is dishonest. To say, exit polls are to answer all questions, except who'dja vote for is blatantly false. The truth is that is THE question on everyone's minds and why the polls are conducted. The other questions, the secondary questions, are to factor in "weighting" variables. It's true that exit polls are not 100% accurate, but most are accurate at LEAST within +/- 5%.

We've toppled many governments on the basis of exit polls and they are based on a statistical science that shows how people voted that directly translates into a superb index of the legitimacy of an election, e.g. the Ukraine (which was instigated by our CIA).

2. The RAW data of Edison/Mitofsky, though I believe you were one of the privileged few who actually saw it?, was NOT released sans the "reluctant responder" poppycock. Again, this explanation sounds much more like propaganda than science. And here's the essential clue, statistics are defined in advance about parameters inclusive of weighting BEFORE the event and set with a +/- margin of error. Anything factored in AFTER the event to "correct" findings is dubious at best.

[Febble]

159. I take exception to that, Einsteinia

and it isn't like you. I am not dishonest, and I am not obfuscating. I am simply telling you the purpose for which the polls are designed. You may wish they were designed for a different purpose (in which case they would have a different design) and you may be interested in a different story to the one they are designed to tell. In any case, of course they are designed to answer the question "who'dja vote for" - what they are not designed to answer is the question as to whether the voting system correctly recorded those votes.

And to assert that "most are accurate to at LEAST +/- 5%" is just that - an assertion. It is true that the MoE of most polls is smaller than that. But the MoE is calculated on the assumption of sampling error alone. Polls are also subject to non-sampling error, which includes bias, and is well researched. But it is not possible to calculate in advance.

And you have not "toppled many governments on the basis of exit polls". Please name one. Ukraine was not "toppled" on the basis of the exit polls. There were, I believe, two, which didn't even agree with each other. There were many many reasons for the CIA to investigate Ukraine. You may recall that one candidate was poisoned by a disfiguring and potentially lethal poison.

I don't know what you mean by "sans the 'reluctant responder' poppycock". It was released sans precinct identifiers, but that has nothing to do with "reluctant responders" poppycock or not. It may sound like propaganda to you, but I am a scientist and I use the scientific method, and I practice sound science. And as for "statistics are defined in advance about parameters inclusive of weighting BEFORE the event and set with a +/- margin of error. Anything factored in AFTER the event to "correct" findings is dubious at best", well, I'm trying to make sense of what you are saying, but frankly, it's gobbledygook. Please clarify.

[OnTheOtherHand]

160. "respectfully"?

What a dreadful post. If you are going to characterize another DUer's arguments as "obfuscation," "dishonest," or "blatantly false," you could at least pay us all the courtesy of citing specific arguments, so we can judge what you are taking exception to. As far as I can tell, either you are refuting arguments that Febble never made, or you are just out at sea. Some of each, I think.

The truth is that is THE question on everyone's minds and why the polls are conducted. The other questions, the secondary questions, are to factor in "weighting" variables.

Say whaa? "to factor in 'weighting' variables"? auf Englisch, bitte?

Misinformed or abusive: please choose no more than one.

[Einsteinia]

161. Febble & OTOH --

My points are these:

1) I am highly suspicious of the campaign to discredit exit polls as a means to measure the legitimacy of an election when it becomes poltiically inconvenient to the ruling class

2) I am disappointed that you two are apologists for what appears to be hijinks as usual. I'm certain your intentions are honorable and fact-based as you say. But I also cannot figure why Kenneth Blackwell, who otherwise seems a reputable man, has gone out of his way to keep voters out of the polling booth. Why? I honestly cannot imagine.

3) Again, post-mortem "corrections" must be looked at with great suspicion. Science should not be matching results to fit the desired hypothesis.

4) If you have the Edison/Mitofsky RAW uncorrected data, please share it in the spirit of genuine scientific plausability. Then if I'm wrong, I will most certainly apologize profusely.

[Febble]

162. OK, point by point

1. One of the things I really objected to about the O'Dell and Simon paper was its reference to a "campaign to discredit exit polls". I know personally of no such campaign, and if such a thing exists, it has got absolutely nothing to do with me.

2. Please support your allegations that OTOH and myself are "apologists" for "hijinks". OTOH, I happen to know, has been involved in investigating undervotes in Sarasota, and I am certainly an "apologist" for nothing. In any case, you say, next sentence that you are certain that our intentions are "honorable and fact-based". So what the hell are you alleging? That we are comparable to Kenneth Blackwell? WTF?

3. The pollsters do not "match{ing} results to fit a desired hypothesis". You seem confused as to what constitutes a hypothesis. They reweight the cross-tabulations on the assumption the vote count is correct and the bias is in their poll. They also re-estimate their projections of the winner in each state on the basis of the vote returns. The purpose of the exercise is to correctly predict the winner, not to act as an independent check on the count. They assume (with some justification) that their cross-tabulations will be more accurate if they correct for assumed bias.

However, when it comes to actually investigating the cause of the discrepancy, they assume nothing. They (and I) tested both hypotheses arising from the theory that their was bias in the poll, and hypotheses arising from the theory that there was fraud in the count. The data supported the former, but not the latter.

4. Of course I have the E-M data, but the data in the public archive is also raw - the weights are supplied in a separate column. However, what is not publicly available are precinct identifiers, nor precinct vote counts, as these would potentially allow the some voters to be identified, and highly confidential data to become known. Nonetheless "blurred" voteshare values were given for the 49 Ohio precincts. But clearly I am not at liberty to "share" the data I was contracted to analyse. It was not "released" to me, I was simply hired to analyse it. It remains bound by the confidentially constraints under which it was gathered.

[Jim Sagle]

89. Oh yes they are! And everybody goddamn well knows it.

[Febble]

131. Well, I don't

and I know a fair bit about polls. What is the basis for your assertion?

[Febble]

104. Absolutely

Papers like this do nothing to advance the cause of election integrity, and, IMO, simply get in the way, or worse. They are hostages to fortune because so easy to refute.


Audits are the way forward. Also prosecutions for corruption and voter suppression.

[holboz]

70. We're on our 4th set of recounts here in Benton County, AR

Loads of problems with the machines, cases of deleted votes. etc. Races have switched back and forth and back and forth again. Candidates and voters are steamed and certainly not confident in our system.

[Gelliebeans]

71. K&R

Big news

[dchill]

73. K&R

Knew it!

[grmamo]

76. Election Defense Alliance has another good read, hope this isn't a repeat

Reported Results Skewed Toward Republicans by 4 percent, 3 million votes
Election Defense Alliance Calls for Investigation
BOSTON, MA - November 16, 2006
CONTACT: Jonathan Simon 617.538.6012
http://www.electiondefensealliance.org/major_miscount_of_vote_in_2006_election

<skip>

As in 2004, the exit polling data and the reported election results don’t add up. “But this time there is an objective yardstick in the methodology which establishes the validity of the Exit Poll and challenges the accuracy of the election returns,” said Jonathan Simon, co-founder of Election Defense Alliance. The Exit Poll findings are detailed in a paper published today on the EDA website.

The 2006 Edison-Mitofsky Exit Poll was commissioned by a consortium of major news organizations. Its conclusions were based on the responses of a very large sample, of more than 10,000 voters nationwide*, and posted at 7:07 p.m. Election Night, on the CNN website. That Exit Poll showed Democratic House candidates had out-polled Republicans by 55.0 percent to 43.5 percent – an 11.5 percent margin – in the total vote for the U.S. House, sometimes referred to as the “generic” vote.

By contrast, the election results showed Democratic House candidates won 52.7 percent of the vote to 45.1 percent for Republican candidates, producing a 7.6 percent margin in the total vote for the U.S. House — 3.9 percent less than the Edison-Mitofsky poll. This discrepancy, far beyond the poll’s +/- 1 percent margin of error, has less than a one in 10,000 likelihood of occurring by chance.

By Wednesday afternoon the Edison-Mitofsky poll had been adjusted, by a process known as “forcing,” to match the reported vote totals for the election. This forcing process is done to supply data for future demographic analysis, the main purpose of the Exit Poll. It involved re-weighting every response so that the sum of those responses matched the reported election results. The final result, posted at 1:00 p.m. November 8, showed the adjusted Democratic vote at 52.6 percent and the Republican vote at 45.0 percent, a 7.6 percent margin exactly mirroring the reported vote totals.


more at link

[bleever]

78. Transparent elections NOW.

I am overjoyed, and cite the words of the Moody Blues, in Land of Make Believe:


We're living in a land of make-believe
And trying not to let it show
Maybe in that land of make-believe
Heartaches can turn into joy

We're breathing in the smoke of high and low
We're taking up a lot of room
Somewhere in the dark and silent night
Our prayer will be heard, make it soon

So fly little bird
Up into the clear blue sky
And carry the Word
Love's the only reason why, why

Open all the shutters on your windows
Unlock all the locks upon your doors
Brush away the cobwebs from your daydreams
No secrets come between us anymore

[rdmtimp]

86. One of my favorite Moodies songs as well...

... and highly appropriate.:hi: :hippie: :toast:

[bleever]

87. Welcome to DU, rdmtimp.

Happy you're here.

:toast:

[A Simple Game]

179. Another Moody Blues fan, excellent! n/t

[nonconformist]

80. Ooops! Looks like they "misunderestimated"! nt

[followthemoney]

85. The purpose of comparing official vote totals to exit polls is to ...

cast doubt and suspicion on elections merely to make requests for honest elections sound reasonable. This occurs because the official vote totals are not verifiable.

Would you attempt to justify a request to your bank that it supply an accounting of balances and a reporting of transactions?

Would you reason with your bank or would you tell them they are fired?

The whole concept of this nation as an open democratic society is a JOKE if we have to present an excuse to the owners before asking for transparency in public affairs.

There should be no hesitation in saying to our so-called elected representatives, "I don't trust you. You have not earned my trust." No Excuses. No apologies for insults possibly taken.

Tell your elected representatives to fix the system.

The Office of Technology Assessment was an independent agency disbanded by the Republican Congress in 1995 after their takeover. This allowed them to spend money on junk.

The Democratic Congress needs to reinstitute the OTA before any further attempt is made to fix or buy voting machines.

[Land Shark]

88. that's right: There should be no hesitation in saying to our so-called elected representatives,

"There should be no hesitation in saying to our so-called elected representatives, "I don't trust you. You have not earned my trust." No Excuses."

You got it followthemoney. The election was suspicious with its secret vote counting before it even got started. The slightest additional facts and we can and should move from the opening position of "no basis for confidence" right up to "probable cause for major investigation."

The secrecy itself is major evidence or probable cause related to wrongdoing since elections should be public.

This report should launch a major discussion and investigation. Proof at this point is neither required nor expected but it is beginning to materialize.

[Febble]

125. Well I could think of better

reports to "launch a major discussion and investigation" such as one that wasn't so easily demonstrated to be founded on an unsupportable assumption.

As usual, I am simply arguing that poor arguments risk damaging your case, rather than helping it. The idea that any argument for the right side is a good one is not one I have ever found persuasive. The case for investigation, and for reform is a slam dunk. Arguing for it on the basis of a flawed inference from a flawed analysis seems, well, asking for it to be sidelined.

Try the Florida undercounts instead.

[southerncrone]

93. Must have let Chimpy do the math for 'um!

:silly:

[rndmprsn]

95. K+R

[barbtries]

97. it doesn't surprise me

no wonder fuckass gw was surprised at the results: he thought the fix was in and the repubs would get away with it the way they did in 2000 and 2004.

imagine what a landslide it really was. imagine how bad and unamerican the powermongers who would take and retain power in this country the one supposedly governed of, by and for the people, if it had not been such a fucking landslide.

death to electronic cheating machines. let the motherfuckers have them in prison.

[nolabels]

98. Is that 31% approval rate for * starting to make sense yet?

They got their thumpin and ass handed back to em at the same time. Going from a minority party to a minor party is starting to sound more credible all the time.

[OneBlueSky]

99. that should probably read "BIGGER landslide denied" . . . because . . .

what I saw was a pretty damn close approximation of a landslide . . . yes it was . . .

[Pooka Fey]

121. Kick!!!! & R'd

[IWantAChange]

139. Irrefutable or flawed - reports like this can be a lightning rod to help cause change..

and for that reason I say thank you to the Election Defense Alliance.

[Febble]

150. I hope you are right.

My fear is that anyone opposed to change uses it as a conveniently ready made straw man.

[bleever]

171. Have no fear.

We've got torches for straw men. The six-year march towards fascism by this administration as lasted longer than World War II, and so have we, and we're ready now for it to be over.

[AtLiberty]

177. Why isn't this thread on the home page?

?!?!?!?

[rocktivity]

181. Maybe James Carville's complaints are justified after all.

Misdirected, of course, but possibly justified.

:headbang:
rocknation