Evidence of 2006 Election Fraud

[Time for change]

Though Democrats, the American people, and the world won a great victory last month, we must not let that lull us into a false sense of security about our election system. Democrats took over the U.S. House and Senate because of a popular mandate so large that Republicans couldn’t steal enough seats to maintain control of either chamber. But before anyone conclude that the 2006 mid-term election results provide reason to be unconcerned or less concerned about the threat of election fraud, let’s consider some of the evidence:


Discrepancy between official vote count and exit polls

The official nationwide result for House races on Election Day showed a Democratic margin over their Republican opponents of 7.6%. Comparing that to a nationwide exit poll democratic margin of 11.5%, the difference between the official count and the results predicted by the exit polls was nearly 4%.

That’s just a little bit less than the 5.5% exit poll discrepancy in the 2004 Presidential election, where exit polls predicted a nation-wide Kerry victory of 3%, while George Bush came out 2.5% ahead in the official vote count. The 4% discrepancy in the 2006 mid-terms – if that represents election fraud – was not nearly enough for Republicans to maintain control in either the House or the Senate. But in 2004, it would have been enough to throw the 2004 Presidential election to George Bush – in the nation-wide popular vote, as well as in Ohio, where Bush “won” his electoral vote victory. And, as noted by Jonathon Simon, the 2006 exit poll discrepancy was far beyond the margin of statistical error, with a less than one 10,000 probability of occurring by chance.


Discrepancy between official vote count and pre-election polls

As you can see from this chart, following a brief Republican mini-surge following the arrest of suspected bomb-making terrorists in London, 40 consecutive Congressional generic ballot polls, dating from September 19th through November 3rd, showed Democrats with leads never sinking below 8%, ranging as high as 23%, and averaging above 15% for the latter 20 of those 40 polls. Then, polls ending on November 4th and 5th showed an apparent Republican surge (with three polls showing the Democratic lead down to 4%, 6% and 7%) before the final two polls showed Democratic leads of 20% (CNN) and 13% (FOX News).

The national exit polls are consistent with the pre-election polls in every significant way. They show that voters who decided whom to vote for in the last week (not including the last day) gave Democrats approximately an 8% margin, whereas those who decided before that gave Democrats approximately a 13% margin*. That is consistent with the final week surge by Republicans.

However – and this is very important – those who decided whom to vote for on the last day (10% of all voters) gave Democrats a whopping 25% margin. This strongly suggests that there was a very late Democratic surge that went totally undetected by the pre-election polls. The bottom line is that, any way you look at it, pre-election polls indicated a Democratic victory far greater than the official result of 7.8%.

We see the same story by looking at individual races. Just prior to the election there were 21 House seats where pre-election polls showed a Democratic lead beyond the statistical margin of error. Democrats lost 6 of those seats and appear to have lost a 7th one. And two of those were in Ohio (CDs 1 and 15), which was the site of massive election fraud in 2004.

* Note: See Table labeled “When did you decide who to vote for”. Then add 4% Democratic margin to each figure to account for the fact that the displayed tables represent the exit polls that are “adjusted” to the official vote count. The 4% that is added is the discrepancy between the official vote count (or “adjusted” exit polls) and the unadjusted exit polls.


Solid evidence of electronic vote deletion in Florida

If the 4% exit poll discrepancy is due to electronic vote switching, that does not mean that any significant portion of that vote switching would necessarily have been detected. The vote counts in precincts that use DRE machines are determined by secret computer programs that produce vote counts that are unverifiable.

Nevertheless, as explained by Paul Krugman, the election in Florida Congressional District 13, which was “won” by the Republican candidate by 369 votes, was almost certainly determined by faulty (whether intentional or not) DRE voting machines. In Sarasota County, which used ES & S voting machines, 15% of voters did not register a vote for the House race, compared to 2.2% to 5.3% of voters who did not register a vote for the House race in neighboring counties. That amounted to almost 18,000 ballots that did not register a vote for the House race in Sarasota County. Furthermore, those who failed to cast a vote in the House race were shown by their other votes to strongly favor Democrats.

Why did 15% of the voters voting on ES & S machines in Sarasota County fail to vote for a House candidate? The answer to that question can be ascertained from an interview of voters by the Sarasota Herald-Tribune, which found that one third of voters couldn’t find the House race on their ballot, and 60% said that they did vote for a House candidate, but their vote didn’t show up on their summary page.


My personal experience as a poll watcher on Election Day

As a poll watcher working for Pollworkers for Democracy in Maryland, I encountered a Diebold machine that was missing the tamper proof seal that was supposed to cover access to the voter access cards. Having learned from my poll watcher training that such a finding could very well be an indication of one of several different methods of electronic election fraud, I asked the chief election judges to take the machine out of service. They subsequently called the Montgomery County Board of Elections to ask for advice, and they were told to continue to use the machine. I therefore called my contact person, who sent in lawyers to talk with the Board of Elections, but to no avail. He also informed me that two other identical incidents had been reported to him that day, all three involving machine # 4 at the respective precincts.

At the end of the day machine # 4 showed the highest percent of Republican votes for major statewide offices (Governor, Senate, and Comptroller) of all the machines used in my precinct.


Voter suppression and dirty tricks

In 2004, dirty tricks and voter suppression were major tactics used by Republicans to keep down the Democratic vote. I don’t know what was the full extent of this problem in 2006. But I can tell you that as a poll watcher on Election Day I was informed that there was substantial voter intimidation going on in Prince Georges County, Maryland, causing Pollworkers for Democracy to divert numerous pollwatchers to precincts in that county. I would have been diverted there myself if not for the illegal Diebold machine that I noted in my precinct (discussed above).

Robocalling was a major part of the Republican plan in 2006. CBS News noted that “In at least 53 competitive House races, the National Republican Campaign Committee has launched hundreds of thousands of automated telephone calls, known as "robocalls." In many cases these calls pretend to be taking a poll, when in fact they are being used to slam the Democratic candidate. In many other cases they pretend to be calling on behalf of the Democratic candidate, calling at all hours of the night with the purpose of irritating Democratic voters so badly that they’ll decide not to vote – a practice that is clearly illegal. And that is in all probability why Tammy Duckworth lost her huge lead by Election Day.


Conclusions regarding election fraud in 2006

Most of the issues discussed above do not, by themselves, prove election fraud. Discrepancies between the official vote count and election polls – whether pre-election polls or exit polls – do not by themselves prove election fraud. Election polls can be biased, and it is usually very difficult to prove whether or not that is the case.

However, what we have here is a highly consistent pattern. In 2006, Congressional generic ballot pre-election polls taken by numerous different professional polling organizations consistently showed a much larger Democratic lead than the official results on Election Day. Exit polls on Election Day showed the same thing. And furthermore, the exit polls showed a substantial last day nation-wide surge by Democrats which could not have been identified by any of the pre-election polls. Comparison of Pre-election polls with official election results for individual candidates showed much the same pattern of discrepancies. And the same thing happened with the 2004 Presidential election, in which exit polls predicted a clear popular vote and electoral victory for the Kerry/Edwards ticket.

We also know of numerous instances of specific problems with electronic voting machines which strongly suggest serious inaccurate vote counts at best, and massive election fraud at worst: The negative 16,000 votes for Al Gore in 2000 from a precinct in Volusia County, which caused the 2000 election to be initially called for George W. Bush; the highly unexpected loss of Max Cleland’s Senate seat in 2002, in a state in which all votes were counted by Diebold’s DRE machines; the pervasive switching of votes from John Kerry to George Bush on electronic voting machines from all over the country, but especially in south Florida; and, the “victory” by the Republican House candidate in Florida District 13 described in this post. These and other identified problems in all likelihood represent only the tip of the iceberg of electronic voting inaccuracy or fraud.

So why didn’t the Republicans perpetuate enough vote fraud in 2006 to allow them to maintain control of Congress? Many Democrats, including a good percentage of DUers, thought that they would do just that. I believe that the obvious answer to that question is that they were unable to. Through a combination of vigilance by Democrats and voter rights organizations, the knowledge that they were being carefully watched, and a fear of getting caught, Republicans in 2006 were perhaps able to steal only about 4% of the vote. And that wasn’t enough this time. But it was enough to elect a Congress considerably more conservative than it could have been, drive Democrats to the right for fear of losing future elections, and pose the threat that most close elections in the future will go to Republicans. Or worse yet, if they are able to better perfect their methods they might be able to steal even more elections than they are currently able to steal.

The bottom line is not so much the evidence that points towards election fraud as it is the simple fact that our vote counts cannot be verified when votes are counted by electronic DRE machines. That is totally unacceptable in a Democracy, and is the main reason why Jimmy Carter has said that our voting system fails to meet international standards for free and fair elections. Until Congress enacts laws that make vote counting in our country transparent to all, it is difficult to see how we can be called a Democracy.

[Peace is Possible]

1. Thanks for posting

wish I could recommend and get it on the greatest...

[nadinbrzezinski]

2. I simply do not trust the system

period... I vote, but don't expect my vote to be counted any longer... so this is not surprising, actually expected

[leftofthedial]

3. we must fix the election system

both the voting/counting process and the financing system

[Larry Allen]

29. Very correct, LOTD.

And the remedies are not rocket science. 1) mandate paper ballots hand counted at the precinct level for federal elections. 2) make contributions to political parties and campaigns by for profit corporations illegal. It is up to us to turn these provisions into main pillars of a progressive manifesto.

[Time for change]

68. Yes, the financing system too

When our elected representatives care more about pleasing the CEO of a large company they they do about hundreds of their other constituents, it seems to me that we are bordering on tyranny.

And our corporate news media too. Something has to be done to restore balance in the news media, and we can start by enacting and enforcing strict barriers against monopolization.

[The Doctor.]

4. There's so much to do.

Plenty of weeding ahead.

Thanks for posting TFC.

[Gwerlain]

5. This is a permalink for me.

Nice piece of research.

[rumpel]

6. Wow, the # 4 observation...

do you know whether the lawyers continued looking further into this?
Do these machines "go home" with the election officials or poll workers prior to election day or are they distributed in the morning?

[Time for change]

40. No, I haven't received an update on that

And I don't know if the machines went home with election officials prior to election day, but I can say from my 16 hours of contact with the election officials at the precinct where I worked, that I doubt very much that they were involved in any fraud. Therefore, if fraud was involved it was perpetrated without the knowledge of those election judges, I believe.

Thanks for reminding me -- I will e-mail "Pollworkers for Democracy" and see if they have any more information on this.

[Zan_of_Texas]

46. Time for Change -- that is an important observation on #4.

Can you get any info on the two other places where machine #4 had a problem?

I think the "tamper proof" seals are barking!

Here are two other looks at those seals.



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

first, yours.

Montgomery County MD -- Diebold


"My personal experience as a poll watcher on Election Day



As a poll watcher working for Pollworkers for Democracy in Maryland, I encountered a Diebold machine that was missing the tamper proof seal that was supposed to cover access to the voter access cards. Having learned from my poll watcher training that such a finding could very well be an indication of one of several different methods of electronic election fraud, I asked the chief election judges to take the machine out of service. They subsequently called the Montgomery County Board of Elections to ask for advice, and they were told to continue to use the machine. I therefore called my contact person, who sent in lawyers to talk with the Board of Elections, but to no avail. He also informed me that two other identical incidents had been reported to him that day, all three involving machine # 4 at the respective precincts.

At the end of the day machine # 4 showed the highest percent of Republican votes for major statewide offices (Governor, Senate, and Comptroller) of all the machines used in my precinct. "


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Cuyahoga County OH AccuVote

"Even leaving aside the possibility that voters will deliberately break seals, broken seals are an unfortunately common occurrence. The most comprehensive study of AccuVote DRE election processes in practice examined the May 2006 primary election in Cuyahoga County, Ohio, which used AccuVote-TSx machines.

The study found that more than 15% of polling places reported at least one problem with seals <9>. The available evidence is that machines and memory cards are not handled with anything approaching the necessary level of care."



-- Security Analysis of the Diebold AccuVote-TS Voting Machine, by Ed Felten et al.,

http://itpolicy.princeton.edu/voting


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


MARYLAND, AccuVote


"I haven't written at all about the Accuvote machines. I guess I've made my opinions about that known in the past, and my new book deals primarily with them. Nothing happened today to change my opinion about the security of these systems, but I did have some eye opening experiences about the weaknesses of some of the physical security measures that are touted as providing the missing security. For example, I carefully studied the tamper tape that is used to guard the memory cards. In light of Hursti's report, the security of the memory cards is critical. Well, I am 100% convinced that if the tamper tape had been peeled off and put back on, nobody except a very well trained professional would notice it. The tamper tape has a tiny version of the word "void" appear inside it after it has been removed and replaced, but it is very subtle. In fact, a couple of times, due to issues we had with the machines, the chief judge removed the tamper tape and then put it back. One time, it was to reboot a machine that was hanging when a voter was trying to vote. I looked at the tamper tape that was replaced and couldn't tell the difference, and then it occurred to me that instead of rebooting, someone could mess with the memory card and replace the tape, and we wouldn't have noticed. I asked if I could play with the tamper tape a bit, and they let me handle it. I believe I can now, with great effort and concentration, tell the difference between one that has been peeled off and one that has not. But, I did not see the judges using that kind of care every time they opened and closed them. As far as I'm concerned, the tamper tape does very little in the way of actual security, and that will be the case as long as it is used by lay poll workers, as opposed to CIA agents."



--- My day at the polls - Maryland primary '06, by Avi Rubin

Tuesday, September 12, 2006

http://avi-rubin.blogspot.com/2006/09/my-day-at-polls-m...

[Time for change]

60. I have no more information on it at this time

Though I have written to Stan Boyd at Pollworkers for Democracy, in an attempt to get follow-up.

I believe that this is a very interesting and important issue. We were told in our training of an example similar to the one I talked about in the OP, where the results of the tampered machine (which they wouldn't take out of service) was vastly different (and much more Republican) than the other machines.

I asked Stan why they needed to remove the tamper tape. Why couldn't they just program the machines to cheat, and then they wouldn't have to worry about removing seals? I suppose it has to do with the fact that it would be easier to get caught that way -- i.e., if they programmed the machines to cheat, that may show up when they're tested. But it's not at all clear to me how much control we have over that process. Anyhow, Stan's answer to me was that, in his opinion, they want to try out a variety of methods for vote switching or vote deletion, and some of those methods involve removal of the tamper tape, whereas others probably don't. I think that we have a great deal to learn about this.

But as I said in the OP, the bottom line is that we shouldn't be using any unverifiable machines, period.

[Zan_of_Texas]

67. "We shouldn't be using any unverifiable machines, period."

Oh, I agree. But, part of how we get there, I believe, is somehow to catch some of these goons in the act, tampering with the machines. Or, at least get enough forensic or statistical evidence (as you pointed to with your machine #4), piled high, to make even the most Doubting Thomas admit that fraud is evident.

It's a difficult task -- the apologist election officials keep saying "Well, no one has ever proven fraud with DREs." Well, that's because they're built to shield a goddamn perfect crime -- any "error" can be written off as a "glitch" - a random error of the machine. And, the rigger is long gone, remains anonymous, and can just try something else next time.

[Time for change]

83. Yes, I agree with all of that

Hopefully with the Dems in power in Congress we'll get some action on this now. You can't obtain the evidence you need to catch these bastards if the government isn't interested.

[livvy]

7. Nicely done! Thanks. Was there any further inquiry into machine #4...

that you know of? That must have been a very frustrating experience.

[Time for change]

65. Thanks -- No, I have no more information on that than what is in the OP

I address that question here:
http://www.democraticunderground.com/discuss/duboard.ph...

Actually, I found it more interesting than frustrating. It was the first time I'd ever done that, and it was kind of exciting, though I was thinking that I sure wish that I had access to a lot of data that would allow me to try to figure out what was going on.

And, of course, I was very much looking forward to getting home and seeing some election results

[corkhead]

8. I took the fairness of the election process for granted until 2000

I think it is the most important issue congress needs to take up in January.

Excellent article.

[ktlyon]

48. I think voter fraud has been around since the beginning

in one way or another.
I voted for years in Houston Tx on punch cards (pre 1997) and wondered until 2000 why it was so hard to push the chads out, now I know. I don't think my votes were counted back then.

[indepat]

9. Many, many 1 in 10,000 probabilities are needed to accept plausibility of 9-11 and pre-emptive war

claims.

[Sancho]

10. No more excuses! Let's hold election supervisors and pollsters accountable.

I also saw vote flipping in 2004 and that motivated me to consider election issues for the first time. The DRE wouldn't let my wife vote for a democrat no matter what she did...the review screen reversed the vote.

We also had unsecured machines in Pinellas, Florida in 2006 (which I complained about before the election). I think ELECTION OFFICIALS with ANY evidence of possible tampering should be tossed out. If there is evidence of suppression, let's through the bums out!

Most importantly, after all the problems in 2000, 2004, and 2006 - WHY are pollsters unable to produce large samples and interviews from targeted problem districts in Ohio, Florida, etc??!? Instead of newspapers calling after the problems appear, pollsters need to have a revote project, lots of interviews, and questions about voting problems on the day of the election? They don't want to document the fraud is the only reason I can think of at this point. NEXT TIME, let's all call and email the media tell them we expect to see pollsters at the disputed precincts in large numbers asking questions about voting experience! No more secrect and manipulated data!

Thanks for the great post.

[Time for change]

76. Thank you -- Fortunately, there are many organizations working on

various aspects of this. The following are just a few of them:

Election Defense Alliance
http://www.electiondefensealliance.org /

Pollworkers for Democracy
http://www.pollworkersfordemocracy.org /

Vote Trust USA
http://www.votetrustusa.org /

Voters Unite!
http://www.votersunite.org /

Is your first name Ian, by any chance?

[yowzayowzayowza]

11. Clumsy Poll Analysis is Evidence of Squat.

The exit poll analysis ignores well known polling bias effects. And, we're supposed to believe a
Sept 19th generic poll proves something about election day? These analysis are statistically
insignificant due to poor assumptions.

TIAs pre-election 2004 polling tallies were fun & informative. But, from the millions of phantom
votes in his 2004 exit poll analysis to these pant loads, enuf already. I miss the pre-Monte
Carlo TIA who reserved his bombast fer trolls rather than folkz who agree with 'em 98% of the time.

Otherwise, TfC, appreciate your compilation.

[Time for change]

12. As I said

The discrepancies between the polls and the official vote counts, by themselves, are not proof of anything, for the very reason that you said -- the potential for bias.

However, there is a big difference between saying that something does not constitute proof and saying that it is evidence of squat. And I have to strongly disagree with you on that point. If polls were evidence of squat there would not be so many millions of dollars spent on them.

Also, it is not the discrepancies between the polls and the official vote counts alone that should lead us to believe that there is a serious problem, but the combination of so many things taken together: Hundreds of voters testifying to electronic vote switching right before their eyes; testimony to Congress about a computer programmer (Clint Curtis) being told to write vote switching programs; the bizarre Busby/Billbray election, where election officials took home the voting machines. And certainly, you don't think that the analysis in Sarasota County regarding this year's Florida 13 election is due to bias?

And most important, we have secret vote counting and unverifiable vote count all over the country. Certainly, all the evidence, taken together, strongly indicates that we need to be very vigilant about this. And I do believe that it was this vigilance, with such actions as stationing pollwatchers all over the country in unprecedented numbers, that prevented more cheating from occurring in this election.

[yowzayowzayowza]

13. Apparently, you "strongly disagree" with something...

I did not say:

If polls were evidence of squat there would not be so many millions of dollars spent on them.

I spoke of the clumsy poll analysis in yer OP; not *ALL* poll analysis, let alone all polls.

A faulty analysis due to poor assumptions is worthless even if the conclusion is correct. Propagating those analysis as evidence of election fraud is irresponsible, esp in light of the volumes of concrete evidence.

[Time for change]

16. Can you be more specific about that?

What "clumsy analysis" are you referring to?

[yowzayowzayowza]

19. The same two analysis in my first reply.

Your:

Discrepancy between official vote count and exit polls

...and...

Discrepancy between official vote count and pre-election polls

[Time for change]

22. Yes, but what is your reason for referring to them as "clumsy"?

[yowzayowzayowza]

24. Throughly discussed:

http://www.democraticunderground.com/discuss/duboard.ph...

...and...

http://www.democraticunderground.com/discuss/duboard.ph...

esp:

http://www.democraticunderground.com/discuss/duboard.ph...

[Time for change]

32. I thought you were referring to my analysis, and now I'm not sure

The examples you give regard TIA's analysis. I don't have time to defend someone else's analysis.

But if you were referring to my analysis, I don't understand what about it you consider "clumsy".

I recognize the potential for bias, and there is no way with the data I have, to measure how much of it there might have been. I also recognize that that the discrepancies between vote counts and polls is not proof of bias, by themselves. But I certainly do believe that they constitute evidence, especially when combined with other evidence.

[yowzayowzayowza]

111. In short, my esteem for your...

...contributions and the levity of your direct evidence intersected your first two poll analysis citations with a (A coupla sets of Al last nite made it awebedder.)

Your more targeted pre-election poll analysis mite prove useful if the alleged fraud number exceeds undecideds. Evidence ... of a sort.

On edit: Sorry for the tardy reply.

PixelatedGruntXlation.Disable;

[OnTheOtherHand]

15. we'll have to come back to the generic polls

It's actually unfair to pick on the "Sept 19th" date. But I agree with your bottom line: the generics aren't evidence of fraud. The prior lit indicates that the "polls of polls" tend to overstate the popular vote margin. Gallup has a pretty good track record over time, and its poll was actually close to the official return. (Of course that can be explained away, if one has a mind to.) And in the aggregate, the polls in individual races were close to the official returns. There were some deviations beyond margin of error that favored the Republicans, others that favored the Democrats.

Some of this needs more sorting out, but not from me right now. I'm happy to agree with Tfc about a continued need to be concerned about election fraud (and system failure). From a forensic standpoint, I think it makes sense to look case by case. From a policy standpoint, I don't think we even need the forensic analysis.

[Time for change]

25. Thank you OTOH -- One point of contention

You say "And in the aggregate, the polls in individual races were close to the official returns. There were some deviations beyond margin of error that favored the Republicans, others that favored the Democrats."

In my OP I provide a link to a pre-election analysis I did, where I categorize 21 races where Democrats had the potential of gaining a seat and where they had leads in pre-election polls beyond the margin of error in both the last poll and an average of polls taken in the past 10 days. Out of those 21, they lost at least 6 and probably 7. I think that's a lot, considering that their lead in several of those races was way beyond the margin of error. Those races were IL6, NE3, NY25, NY29, OH1, OH15, and PA6.

I could not do anything similar for races where Republicans had the potential of picking up a seat and also had leads beyond the margin of error, because there were none. However, looking at ALL races, there was only 1, and possibly 2, where a Democrat won while being behind beyond the margin of error (NH1, and possibly PA4).

[OnTheOtherHand]

51. those could both be true

I will have to look at the races more closely. But tampering focused in some competitive districts certainly seems plausible to me, quite apart from the overall results. (Not that the polls would be slam-dunk evidence, but reason to look some more.)

[OnTheOtherHand]

69. OK, I was wondering why these hadn't kicked out

IL 6, Roskam vs. Duckworth, the Zogby poll was the last one out and had Duckworth up 14, but two other polls at almost the same time had Duckworth +1. I thought Duckworth had the inside track, but I didn't especially believe the Zogby result.

NE 3? All I see is Penn Schoen and Berland, Dem +6, within the margin of error (small sample).

NY 25, RT Strategies had a late IVR poll with Maffei (D) +9, but two late phone polls had Maffei +4 and Walsh +2.

NY 29, Kuhl-Massa, didn't show up because all the polls were IVR (RT Strategies). The RT Strategies results often didn't jibe with other results in the same districts; pollster.com didn't use them in the averages. (I'm not expressing an opinion beyond that.)

OH 1, Grove Insight (D) had Cranley (D) +9, within the MoE because it was a small sample (and this time the RT Strategies poll came in Cranley +2).

OH 15 (Pryce-Kilroy, extremely close), the one poll (Kilroy +12) was early and IVR.

PA 6 (Gerlach-Murphy), it looks like the last 4 polls all put Murphy ahead, but none of them beyond the margin of error.

Not that I'm vouching for the counts in any of these races, but I was trying to understand the contrast between your analysis and Mark Blumenthal's at http://www.pollster.com/mystery_pollster/house_district... .

The survey evidence doesn't have much bite. EDIT TO ADD: whoops, distracted. I mean, if I tried to get a colleague interested on the basis of these data, I would evoke an eye-roll. There would have to be something else -- as there is, in spades, in FL 13.

[Time for change]

73. Margin of error

We may be calculating it differently.

Take NE3, for example. I get a standard error for the Dem vote of (0.54 x 0.46)/square root of 404. That comes to 0.0124. Multiply by 2 to get the 95% C I, and you get 0.025, or 2.5%. The poll difference was 6%. So how do you get that that was within the MOE.

[OnTheOtherHand]

87. extend that square root

The standard error is sqrt(pq/n), not pq/sqrt(n).

TIA's shortcut is perfectly handy (when used judiciously!) -- to get a margin of error, use 1/sqrt(n). Either way, around 5%, or 10% on the margin.

Given what we can see of house effects (or whatever they are) in races with multiple polls, I would be leery about relying too much on any one survey result regardless of MoE.

Of course if all the close races seem to break the same way, that would still raise eyebrows. I have to admit that my eyebrows aren't up -- could be because I came in by a different analytical route. I do know that I haven't lost my capacity for quantitative alarm, because Sarasota set me off right quick.

[Time for change]

89. Thanks OTOH -- I will go back and have a re-look at these races

later when I get some time.

[BeFree]

20. You can fool some of the people

some of the time, as the saying goes, but you can't fool all of the people.

Some of us are not fooled by the faulty vote counts produced by Diebold and ES&S. So the onus is on those who believe the vote counts to show us the evidence that the vote counts were correct. Of course, no one can prove that Diebold and ES&S counted the votes as cast; all they can do is try to make fun of the analysis of the faulty counts. One would think they'd have better things to do than that, eh?

And using the canard that there is or was a polling bias has been so thoroughly discounted that I find it amazing anyone even brings up the subject, but I guess if one has nothing else then that is what is to be expected.

[yowzayowzayowza]

23. You do realize that ...

believing there were "faulty vote counts produced by Diebold and ES&S" does not preclude error in TIAs analysis?

[BeFree]

26. Of course

But why not spend time going after the real culprits? Ya know, the ones that you know are vote stealing bastards?

And if you don't like TIA's analysis, do your own.

This constant harping, nit-picking, and mud-slinging against TIA is ridiculous, assinine, and fruitless.

Haven't you people something better to do?

[yowzayowzayowza]

30. You do realize that both...

itz important that the evidence supporting our claims (esp our first two points) is rock solid, not washed-out statistics?

...and that...

one can review TIAs analysis AND go after the real culprits at the same time?

[BeFree]

31. All one needs to know

Is that the vote counts by Diebold and ES&S are not rock solid. That the people can not trust in any way the results that come from those companies.

Tell you what, when you get a court case going, and you need TIA to take the stand, then we can insist that TIA's statistics be unimpeachable. But get that court case first. That is what you want, right?

[ktlyon]

50. so let's impound some machines and run some tests and look at

their internal workings and logic. Lets get some real evidence by your standards. I personally believe the republicans have been suppressing the vote for a long time and before that it was Democrats that were stuffing ballot boxes and letting dead people vote. Before that we had Jim Crow. Voter fraud is nothing new but this wrinkle is. Let's crack the machines now.

[Time for change]

66. The people should have total access to the machines

But for some strange reason, the voting machine companies are able to get away with calling their machines "proprietary" and limiting public access to them.

Beats me how this could happen in a "democracy".

[ktlyon]

107. F... them I want to know

we can run test votes can't we to see if they work right even without seeing the program, but I want to see the code.

[Time for change]

108. Yes, we can, but I'm pretty sure they have the capability of writing programs

that will allow their machines to pass the test votes, and yet on Election day will steal votes.

[Febble]

27. Ditto

It seems to me there are three issues:

1. Can we trust the count?
   
2. Do we have reason to believe the count is actually wrong?
   
3. On what scale is the count wrong?

The answer to the first is a clear no, because without any transparency there is no basis for trust.

The answer to the second is also a pretty clear no, because of the evidence you cite, and more

The answer to the third is the controversial one, and it is the one where the polling evidence (pre-election generic; exit) tends to be called into play, as you do. And, as you know, I don't think it supports the inference you put on it. There is very strong evidence (some of it from actual experimental data) that exit polls tend to have a Democratic bias, and that they did so in 2004; there is also no evidence that the exit poll discrepancy at precinct level was associated with change in Bush's vote share. And that all adds up, IMO, to a strong case for substantial Democratic bias in the poll in 2004. Ergo, I do not think much, if anything, can be inferred from the a slighlty smaller exit poll discrepancy in 2006. To be epidemiological about this: exit polls may be sensitive to fraud, but they have very little specificity. The False Positive rate is high. And I don't think the generic poll resultd, for reasons discussed elsewhere, support the case that in this instance the exit poll discrepancy might have been due to fraud.

But thanks for posting, anyway! I agree with your conclusions.

Cheers

Lizzie

[BeFree]

33. I love it

"...actual experimental data..." That's classic!

So, you are asking us to believe that there is nothing we can believe, yet you want us to believe your opinion that we can't believe what the numbers tell us because the numbers - even after all these years - you believe, are biased?

Thank gawd some of us still believe that we CAN add 1+1 and come up with 2.

[Febble]

43. You know what I am asking

I am asking you to bear in mind that polls can be biased, and that there is actual experimental evidence that exit polls tend to be biased in the Democratic direction.

I am not asking you to believe that they can't tell you anything (they can), nor am I asking you to believe that there wasn't fraud.

I am simply asking people to base the case for fraud on better evidence than polling data.

But you know this.

[Time for change]

34. It is a matter of degree

If you are saying that the discrepancies between the polls and the official vote count do not constitute proof of fraud, I agree with that.

But if you are saying that they constitute NO evidence, I strongly disagree with that.

More likely, we both believe that the truth is somewhere in between, though I believe it is more in the direction towards fraud than you do -- partly or largely because there is so many different kinds of evidence that point in the same direction.

With regard to the Democratic bias in polls, that has generally been shown to be small, and as Steven Freeman points out, some or all of that "bias" may be more apparent than real, being due largely to disproportionate undercounts in Democratic areas. And of course, any analysis that purports to identify a Democratic bias in polls also should account for the possibility that the apparent "poll bias" may be due to election fraud.

[Febble]

36. Not so much degree

as probability. That's why I used the sensitivity/specificity analogy. If there was large-scale fraud, it would be unlikely that the exit polls would match the count. So in that sense, the exit polls are sensitive to fraud, and raise the probability that fraud occurred.

However, because polls can be biased, they don't have good specificity. And, worse, because the evidence suggests (and I believe it does suggest) that the polls in 2004 had a large magnitude bias, then we know that large scale bias not only possible but likely. In which case, I think it is not so much evidence "to a degree" but simply consistent with large scale fraud. But not, in itself, indicative of it. However, the data in 2004 if anything contra-indicate fraud on a scale of millions

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

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

which means that my prior for fraud on the scale of millions in 2006 is low.

But my prior for miscounted votes is high. And we know there was rankly fraudulent voter suppression, which wouldn't affect the exit poll data (but might affect the generic poll), so actually my prior for fraudulent vote theft of some magnitude is also high.

[Time for change]

39. I don't believe that 2004 is a good year for comparison purposes

(Yes, I know, I referred to it in my OP, but I didn't emphasize it).

I agree with you that voter suppression (including voter registration fraud especially) was probably a much more important issue in that election than electronic vote switching. And I agree that the evidence is good that there was some bias in those exit polls.

However, 2004 demonstrated the largest discrepancy ever, in a presidential election, between the exit polls and the official vote count, and for that reason it isn't a particularly good example to bring up in an attempt to discredit this year's discrepancies.

[Febble]

41. OK

if you are not arguing from precedent, then I'm with you, I suppose. Except that I don't see that 2004 was particularly exceptional - sure it was larger than usual, but 2006 seems to have been about the same as 1992. And the discrepancies in 1988 and 1996 were also substantial. Michael Butterworth investigated correlates of the 1996 discrepancy and got very similar findings to those I found in the 2004 data. Ironically, in 2000, the exit polls were unusually close to the count. And they weren't bad in NM in 2004, which is a state I think it is highly likely that Kerry won. So there is precedent for polls being biased, and precedent for the Wrong Result even where the exit polls are fairly close to the official result.

[Time for change]

52. I look at it like this Febble

The discrepancy (i.e., between the polls and the official count) = election fraud + bias + random error. We know for a fact that random error is going to be quite small in a very large poll, so for the sake of this discussion, let's just say that the discrepancy = election fraud + bias.

Therefore, a very large discrepancy is virtual proof that we have EITHER election fraud or bias or both. The only question is which one (or both), and how much.

Therefore, I believe that it is accurate to say that a discrepancy beyond the MOE provides at least some evidence of both election fraud and bias. To say that it provides NO evidence of one or the other means that you've pre-determined that the whole discrepancy is accounted for by the other one. I just don't believe that we can do that. You may believe that one is more likely than the other, but to say that a discrepancy provides NO evidence of one or the other, I think, is not warranted by what we know -- especially in 2006, where substantial investigations have yet to be done.

On edit: It is also reasonable to assume bias is likely to be differently determined in pre-election than in exit polls, because the reasons for the bias in the two are different. Therefore, if the pre-election polls match the exit polls, which they do pretty much in this case, then that provides good additional reason to believe that it is more likely that the polls are reasonably accurate.

[Febble]

58. Well, that's coherent!

I still wouldn't put it at the top of your summary of evidence, though. The one I'm interested in is your experience as a poll-worker. Even if the machines weren't tampered with, the fact that the seals were broken means that surely using them was illegal (not that I am a lawyer). Did you do a chi square on the proportion of Republican votes on the unsealed machines? Presumably the assignment of voter to machine should have been random.

Yes, I take your point re the analysis of the exit poll discrepancy, although the work I did on the data demonstrated that about 50% of the variance was sampling variance. But that, as you imply, would have no net bias (or not after we'd figured out how to get rid of the inbuilt bias in extreme precincts). But the reason I put a high probability on the discrepancy being accountable by bias is not simply that methodological factors were a strong predictor, but because of the complete absence of any correlation between redshift and swing to Bush. If fraud variables were being masked as methodological variables, then it follows that the fraud had variance. And yet that variance was not shared with any benefit to Bush, even after baselining the discrepancy to state mean levels.

That certainly does not mean that there was no fraud in 2004, nor even that it didn't contribute something to the discrepancy (it may have done). But in terms of effect size, it was completely swamped by other factors, which tells me that the probability that it was on a scale of millions is low. And if 2004 had large bias and fraud below the scale of millions, then I find it hard to follow the inference that 2006 had small bias and fraud on scale of millions. You may be right. But I simply don't think it amounts to more than a footnote that the exit polls are consistent with fraud on some scale. And I certainly don't think the generic poll discrepancy is much of a corroboration. I was following them, and the debate about what they meant, pretty closely, and the pundits I read who attempted to translate them into House seats got the prediction about right. And the close races where there were actually polls of specific races again, seemed to get them about right.

So if I were your editor (ha!) I'd put the stuff at the top of your OP in small print in the footnote to an appendix, and get on with taking the people who refused to take the machines with tampered seals to court. And start finding out who tampered with them.

[Time for change]

63. Yes, I did a chi square on the machine differences

p value was not statistically significant at 0.05, but it was kind of borderline, around 0.1 -- I don't remember exactly. The numbers were small. I was thinking at the time that it might make it less likely that they get caught if they cheat a small amount on a large number of machines than a great amount on a small number. At this time, there is no proof either way, as to whether or not the removal of the tamper proof seal (in this instance, or in general) is likely indicative of fraud. But I do consider it to be just one more suspicious incident, and since I was personally involved with it and told to watch out for it ahead of time, it seems important to me.

With regard to whether or not individual races were skewed to the Republicans (compared with polls), please see this:
http://www.democraticunderground.com/discuss/duboard.ph...

And with regard to whether the generic polls were on target in this election, I don't believe that is the case at all, unless you want to pick out three outlier polls among several dozen, as I discuss here:
http://www.democraticunderground.com/discuss/duboard.ph...

[Febble]

64. Well the fact that

the seal was broken yet the machine was used is de facto corrupt, isn't it? I mean, even if there was no intentional fraud, the whole point of seals is that they are what little assurance you have that the machine is OK. If you can allow an election on a machine with a broken seal, then you might as well not have seals.

We seal our ballot boxes in the presence of witnesses. If the seal was broken on arrival at the count there would be hell to pay. I think it would invalidate the election.

Did you check out the results on all the machines with broken seals?

I'll check out the polls you mention, but I've gotta run right now (to bed, in fact....)

[Time for change]

77. Yes, I believe you're correct about the importance of the seals

But what I don't understand is why they even need to break the seal in order to fix the election. There just seem to be so many ways to cheat. Most importantly, in my mind, is that they can write a program to count the votes the way they want them counted. Since they so frequently use the excuse "This machine is 'proprietary', so the public doesn't have the right to check it out", and our government lets them get away with that, I don't understand why they need to break a seal. Like Stan Boyd (my contact person at pollworkers for democracy) said to me, "It's probably just one of many methods they have to steal the vote, and they want to try as many as they can to see which work the best".

I don't know what the results were on the other machines with broken seals -- Stan is checking it out.

[LiberalEsto]

14. Thank you for your service at the polls

I salute you and everyone else who put in long hours on Election Day, trying to safeguard democracy.

My husband was also a poll watcher in Montgomery County, MD. We live in Derwood.

We were happy to hear that NIST recently issued a study condemning these electronic "cheating machines."

[Time for change]

80. Yes, I was very happy to see that too

though I don't know how much clout they have.

Did you husband have a good experience on Election Day as a poll watcher?

[LiberalEsto]

96. This year, yes

but that's because the senior election judge at our polling place finally retired. Last year he and Dave got into a fuss over paper ballots, which are supposed to be available as an alternative.

[porphyrian]

17. I'll kick that. - n/t

[Subdivisions]

18. 4% discrepancy. Just enough to win an election and not raise

suspicions to the point of investigations. However, they tried it one too many times.

Anyone found guilty of tampering with an election should get life in prison with no parole.

[Norquist Nemesis]

21. Just about the same amount as in 2004

Wasn't it about a 4% discrepancy in the 2004 election as well? (serious question)

[Time for change]

75. It was 5.5% in 2004

http://freepress.org/images/departments/PopularVotePape...

[Perky]

28. Oh for Crying out loud....I have to take you to task on the first two items.

Generic ballots and national exit polls are nothing close to what actually happens on election day. It is a scientifically selected sample of 1,000 to 1,300 or so voters across the country.

The FACT of the matter is that as the seat becomes safer the number of votes cast actually goes down and Conservative are generally more likely to vote than liberals. Also in certain areas of the country Blue is very very blue and Red is extremely Red. The national generic ballot question does not wight for any of these factors. You would have to have a sample size of more than 100,000 voters or about 230 per district.

On the first two questions, I would suggest you wade through a book on polling or statistics. It is not evidence of fraud....not hardly

[BeFree]

35. Hey!

lookie here folks we have an expert on polls and polling and voters!

Forget all you ever heard about polls and about all that money spent on polls, and what the *other* poll experts have said about polls, because we now know that none of the pollls are capable of telling us anything - they have no value!

////

[PATRICK]

38. Polling aside

and I was very intrigued why (TIA) early and few polls were used to indicate the strength of my candidate in NY who went on to lose despite being labeled a sure thing. I think that was to optimistic for many of the reasons posted above. However I was there for it all on the phone lines and I could feel the time when we made progress by talking as real people to receptive voters one on one, then when phone fatigue and the dirty TV ads started hitting with lies so huge people thought there must be something to them. THEN at the last minute the robocalling finally hit right in the middle of wearying and sometimes repetitive efforts by the campaign, unions and Moveon. Before that I had to argue with otherwise sensible Democrats that not voting because the campaign had turned nasty is exactly what the GOP wanted.

I don't necessarily need the thermometer of polling up my re..un to know I am sick. It cost. It was unable to be countered quickly enough. It hit at our necessary strength. With a recording no less. The only way to remove that threat short of strong legal measures is to make sure that everyone knows that your candidate neither does nor condones recorded messages at all. But to do that you have to pick some other medium and they are all open to various poisons. But ALL that aside, it was Cheney's sculptured gerrymandering that set the stage. You have to be some kind of complete moron to debate about polls, machines, dirty campaigning etc. in separate pieces as if one negates interest in all the others. That is how Rove succeeds with one crime enabling another by people's dumb reaction that deals with none of the competing victimizations.

Fraud=GOP. With a 60,000 advantage in registrations they almost lost. Their "representative" has little legitimacy and was not meant to have to suffer the will of his own party members. Everything is suspect, nothing is verifiable, no benefit need be given because no proof will ever be forthcoming. Maybe they WILL let go a little bit and choose candidates who please people and serve their interests a little more? Well rah rah for the GOP. They are a political mafia before they even get started on policy and ideology horrors.

[OnTheOtherHand]

54. heh

Seems to me the last time our paths crossed, I was pointing out to you how you had cherry-picked a poli sci article on generic polls when you mysteriously vanished.

[BeFree]

74. Moi? Cherry pick?

No say si, senor.

Then we have a link from:
http://people-press.org/commentary/display.php3?Analysi...

Which pretty much says the same thing that moi said before when claimed to have been picking cherries. Really, if you want to accuse me of something make it low hanging fruit, at least, please!

What it all boils down to is simple math. Or, if you desire, Karl Rove's math. Whatever.

I desire the simple math. The math that says: add a little vote theft by machine programming, too touch screens that won't allow a voter to pick their choice, and machines owned operated and controllled by friends of the republican party, with evidence of funny numbers in the exit polls, and you come up with the clear conclusion that millions of votes were stolen. Add it all up and you too will conclude that it really kinda equals Karl Rove's math. EH?

Yep, that miraculous republican turnout in 2002 and 2004 is nothing more than manipulation of the vote counts. Manipulation by the same folks who stole the election from Al Gore in 2000. What? You think because they got away with it in 2000 they wouldn't do it again?

That is whacked to think they wouldn't steal an election again. Just whacked. Not only would they do it, they had the means, the motive and the diebold-tunity to do it in 2002 and 2004. In 2006, we stopped them. Mostly. But they'll be back in 2008. Where are you gonna be? I hope not still cherry picking the polls as you are so prone to doing.

[OnTheOtherHand]

88. no, it doesn't

It says, "The lines plotting the actual vote against the final poll-based forecast vote by Gallup and the Pew Research Center track almost perfectly over time." This year Gallup had Democrats +7; Pew had Democrats +4. Could you please pay attention?

[BeFree]

91. Easy fella

http://people-press.org/commentary/display.php3?Analysi...

What I see at the link is this:

The final forecast of the generic House vote and the actual vote totals have paralleled each other very closely for nearly a half-century in U.S. elections. The average prediction error in off-year elections since 1954 has been 1.1%. The lines plotting the actual vote against the final poll-based forecast vote by Gallup and the Pew Research Center track almost perfectly over time.

Which follows what we were talking about - how accurate are generic polls? The average error has been only 1.1%. That's pretty accurate, eh?

Of course, when one uses generic polls, one lumps them all together and averages the results, ya know, like TIA did. That's how you get the 1.1% error rate.

Except for the last few elections using e-voting. Something seems to have skewed the historical accuracy since e-voting. Do you have a clue?

[OnTheOtherHand]

93. the question is, do you?

Read the quotation again, and look at the graphic. If you are not aware that you are misrepresenting Pew, it is not too late to figure it out.

"Of course, when one uses generic polls, one lumps them all together and averages the results, ya know, like TIA did."

Well, that's what Bafumi et al. did, and that's how they arrived at the conclusion that averages of generic polls tend to overstate the winning margin. What Pew did -- as you accurately quoted, but apparently could not bring yourself to read -- was to cite "the final poll-based forecast vote by Gallup and the Pew Research Center." Is that an average of all the generic polls? No, it is not.

"That's how you get the 1.1% error rate."

No, it obviously isn't. To get the 1.1% error rate, Pew used "the final poll-based forecast vote by Gallup and the Pew Research Center." More specifically, they used the Gallup result from 1954 through 1990, and the Pew result for 1994 and 1998. They told us all what data they were using. Why would you conclude that "Of course" they used some other data instead?

Look, if you want to sit around and be suspicious of e-voting, I'm happy to do that with you. But you don't have to go and butcher polling arguments.

[Raster]

37. Better than you have waded through the books on polling and statistics and are

currently wading through the data (as it is released). If you wish to choose to believe there has been no fraud, fine, your choice. You are now FIRMLY in the minority.

[Perky]

49. I did not say there was no fraud...sheesh

I said that the first two points are not evidence of fraud.

[OnTheOtherHand]

55. hmmm...

For pre-election polls, I would rather have non-generic polls with samples in each district -- which, of course, we actually have (for competitive districts). But the samples tend to be small (not as small as "230 per district") and the results sometimes vary from poll to poll in poorly understood ways.

The national exit polls are much larger, of course. Sample size does matter, but it doesn't fix (or even improve) biased polls.

[AzDar]

42. Kick n/t

[bleever]

44. What a well-written overview of this topic.

Recommended with enthusiasm!

[mopinko]

45. tammy was robbed

the robocalls helped, but i think that votes were stolen outright. she was running in a very, very, very dirty district.

[Mrspeeker]

47. The Democratic Voting system lives on!

LOL, well almost every post I've made on these forums is about a failed democracy, a failed voting system, a failed way of life... why has it failed? Because the people do not elect their government, so the Government that is spose to work for the good of the people works for someone and something else. How can we preach all around the world about freedom and democracy, when we no not what that is?

note* I had to write this fast for fear of outsourcing my writing to Halliburton and its subsidiary's All my writing is subject to review by the Department of Homeland Security, the CIA, The FBI, The ATF, the Illuminate, the PNAC, and the world banks. Thanks to large corporations and the GREED of a few, I feel so Free in this total democracy!

[brettdale]

53. keep kicked

kick

[Kelly Rupert]

56. Knock it off with the generic ballots.

Generic ballots do not bear any resemblance to the final result, nor have they ever. Taking a district-by-district approach, the polls were pretty good--and since there were 435 races, finding an outlier or two doesn't show anything meaningful. You can make a case without spinning numbers to show things they don't.

[Time for change]

57. Where do you get that from?

Here's an analyses by Pew Research that shows not only that the generic ballot has a resemblence to the final result, but it has generally been within 1.1%:
http://people-press.org/commentary/display.php3?Analysi...

Also, I don't believe it's fair to say that the district by district polling has been good this year. In 7 of 21 races where Dems had a chance to take a GOP seat, and where they had clear leads in pre-election polls beyond the margin of error, they lost, as I note here:
http://www.democraticunderground.com/discuss/duboard.ph...

That's a lot of losses for having a lead beyond the margin of statistical error.

[OnTheOtherHand]

59. personally, I think Pew was probably not far off

but since it had Democrats +4, that doesn't especially help the cause. Bafumi et al. pointed out on Pollster.com that generic "polls of polls" are predictive, but also that they tend to overstate the winning margin (follow their link for more details). (I wouldn't have said "do not bear any resemblance.") I don't remember if it shows up in that article, but the prediction error (root mean square) of the polls of polls is so large that one just ends up shrugging.

Still haven't had time to look at your list of races.

[OnTheOtherHand]

61. oh, by the way

I already have someone lighting me up all over the place for citing Pew, so I had better clarify my meaning. Your link shows historicals for Gallup and then Pew; both Gallup and Pew projected single-digit Democratic margins. Since the margin of error on a margin is pretty much double the margin of error on a vote share, there is lots of slop in the estimate.

As Mark Blumenthal argued in a different pollster.com article, the final Gallup numbers appear to have been more accurate than the poll-of-polls numbers.

I'm off to daughters' concert; more later.

[Time for change]

62. If I understand correctly what you're saying

Your argument hinges on giving substantially more weight to three polls ending on Novebmer 4 and 5 than to 40 consecutive polls prior to those 3 polls and two polls that kind of followed those three polls, the vast majority of which indicated substantially greater Democratic margins.

I believe that there are two potential rationales for doing that -- one, that Pew and Gallup are substantially more accurate than the other generic polls; and two, that it has to do with the timing. I don't believe that we have significant (or any) evidence that Pew and Gallup are better than the rest. I'm more inclined to think that it has to do with the timing. But then we have the two latest polls shooting right back up there, and we also have substantial evidence that a Democratic surge did in fact occur on the last day, because the 10% of voters who say they decided on the last day demonstrated a Democratic margin of 25% -- and that's without adding the 4% that we need to add because the 25% figure is "adjusted" down.

In summary, I see little reason to judge the true prediction based on three selected polls that are surrounded by a vast amount of other polls with radically different results.

[OnTheOtherHand]

70. actually, there are two lines of argument

One has to do with the historic accuracy of final generic Gallup polls; the other, with the historic margin overstatement of generic "polls of polls." Both point broadly to the same conclusion: single-digit point estimate with plenty of room for error. (As for final Pew polls, there isn't nearly as much history to fall back on. Two nice things about the final Pew poll are that it has a large sample and a four-day field period that began before the weekend.)

It's worth bearing in mind that there are good theoretical reasons to distrust generic poll results. Even the official count isn't especially useful, given how many uncontested races there are -- some of which people can vote in, some of which they can't.

I agree that the exit poll result about last-day decisions is a striking one, although I am less confident than you are about how to interpret it. It would be interesting to have someone extend the Bafumi et al. analysis, or something like it, with exit poll results to see how late-breaking voters seem to affect the vote.

You can read the generic polls however you like; I'm trying to report how political scientists read them.

[truedelphi]

110. Ya know, on the 4th thru the 7th of Nov

I was forced to watch a lot of right wing talking heads (due to visiting
an elderly relative)

And they seemed tobe glibly quoting from polls I had never even heard of.

I did not hear Gallup mentioned very often.

[NotGivingUp]

71. great post! k & r !

[In Truth We Trust]

72. K&R Hand Counted Paper Ballots NOW!

[BeFree]

78. I forgot

To thank you, Time for change, for this great thread. The way you put it together is one of the best compilations I've seen as concerns this matter. Way to go!

The only real flaw I see in your OP and in the underlying replies is the flaw of believing the exit-polls show a bias. I know the arguments for the bias have been somewhat convincing, but there still is no sound basis for that bias, since the bias theory has never been scientifically repeated by anyone other then Ms. Febble.

[Time for change]

79. Thank you Be Free

I didn't say that there was bias in the 2006 election polls. I just said that I can't prove that it's not there.

I'm an epidemiologist, and in every research study that we do we need to consider the possible role of bias and compare that with the likelihood of other explanations. It's very difficult to prove that it's not there, so what you're left with is considering all of the evidence taken together.

Part of my reason for believing that election fraud is a more likely explanation than bias is that I have no confidence whatsoever in the integrity of the Republican Party, and I consider most of the leaders of the Republican Party to have no morals whatsoever, and I believe that they would do anything to stay in power. I'm sure that most DUers who believe as we do feel the same way about that. We have motive, means and opportunity, nobody can argue about that. That puts us more than half way there to showing the likelihood of election fraud, I believe. That's what makes any additional evidence all the more persuasive in my view.

[BeFree]

81. Can't argue with that

I must have been mistaken, then. I thought I read you agreeing with Ms. Febble that you saw a bias in the exit-polls. I stand corrected.

It's funny how a little thing like that can be blown into such huge importance. But that's what it has come down to, these elections of ours. This bit of evidence against that.

What a shame. A shame that we can't trust how are votes are counted. A shame that we have to gather all these bits of evidence just to show the world that it isn't what it seems - this president is really not our president!

But we are doing it, because we have too. We have too because without showing the world that jackal is not really our's the world will turn against us and our American experiment will wither away.

Now I'm really just a keybord hacker, hacking away at the wall protecting the jackal. So when I see the likes of you - the educated and comfortable joining we hackers at that wall, and see yall blast mighty fine, irrepairable holes in that god damned wall, I just get to feeling better and better about our chances. Know what ah mean, Vern?

[Time for change]

84. Yes, it is a terrible shame

Our current administration is a disgrace to our country, and a blight on the whole world.

And we need to be in this together, because it is oppressing all of us. And we've got lots of good people ready to go at it now too. I'm so glad to hear about all the investigations that various Dems are talking about now. Conyers, Leahy, Boxer, Feingold, Kerry. This could be very interesting.

I did mis-speak slightly in my last post to you. I do agree with Febble that there was probably some bias in the 2004 exit polls. I believe the evidence is fairly persuasive on that count. But I also believe that there was a great deal of election fraud, and that John Kerry won the election, just like Al Gore did in 2000.

But one thing that you and I and Febble and almost all DUers agree upon is that we need to be very vigilant, and our government needs to be much more vigilant than it has been, in protecting our elections.

[BeFree]

98. Probably some bias?

What Febble has done has come up with a theory, at the request of Mistofski, to theorize why his exit polls had so much discrepancy.

The data that Febble was allowed to see has not been disclosed to other researchers so that they could replicate and test her theory. It remains an untested theory: an assumption.

But you are convinced? Do tell us convinced you to adopt that theory. And I do expect a thorough explanation.

[Febble]

100. Not true

I didn't "come up with a theory" - I tested hypotheses arising from a theory. I found correlations in the data that supported the theory.

It does not remain an "untested" theory, although I agree that all you have for 2004 is my report. But the report is precisely of tests of the theory. The theory is not an assumption. What may be an assumption is that I am an honest and competent scientist. And actually, I am.

And it is not unreplicated; other researchers in other US exit polls have found the same thing. Even Steve Freeman found evidence for "reluctant Republican Responders" in his exit poll.

Pro-Democratic bias in exit polls is well-supported by evidence.

[OnTheOtherHand]

101. that's hardly all he has for 2004

The dirty little secret of the "exit poll debate" is that it doesn't take access to the data, or even to your report, to see some of the gaping holes in the arguments that the 2004 (or 2006) exit polls point to massive fraud. One doesn't even need to read the E/M evaluation report, although it helps.

For instance, in order to believe that the exit polls were highly accurate prior to 2000 or so, one also has to believe that the documentary "The War Room" was faked to depict exit poll error in 1992. It's frankly pretty silly to think that Mitofsky had to hire you to whip up an explanation for a problem that he had been tracking for at least a decade-plus.

Also, screen shots and TIA-approved poll numbers would suffice to demonstrate that in 2004, the states with the biggest exit poll "red shifts" weren't the states where Bush did best compared with pre-election expectations. And screen shots support E/M's precinct-level report that jurisdictions with e-voting didn't have larger red shifts in 2004. And so on.

I can understand why people who haven't taken much time to look at the evidence think that you are saying something unusual. But I have a harder time understanding why a few devote so much energy to 'shooting the messenger' when they could look at so much of the evidence themselves, or else stick to subjects that they are more interested in.

[Time for change]

109. 3 things

3 issues mainly caused me to believe that there was probably some bias in the 2004 exit polls:

1. The New Hampshire recount: NH was the one state where there was a massive exit poll discrepancy and a recount (which was regarded as fair) was performed. The recount indicated no problems with the official vote count. Therefore, it is fair to suspect a serious problem with the recount in NH.

2. Walter Mebane's study, which was part of Howard Dean's DNC report: That study showed strong correlations between the vote for John Kerry and the vote for the Democratic candidates for Ohio Governor in 2002 and Senate in 2004. If there had been systematic widespread massisve vote switching throughout the state of Ohio (where Bush 'won' the election), those correlations would not have been so strong. However, it is certainly possible that selected areas of widespread vote switching occurred, which would not have shown up in Mebane's analysis.

3. The ESI study: This study showed that precincts where Kerry did considerably better than Al Gore in 2000 did not have larger exit poll discrepancies than other precincts, on average. If exit poll discrepancies were due to vote switching, one would expect to see large differences between the Kerry and the Gore vote in those precincts where the biggest exit poll discrepancies occurred.

Also keep in mind that in Ohio, which is where Kerry won the election, and where there was a very large exit poll discrepancy, there was very little electronic voting in 2004.

Therefore, I do not believe that the 2004 exit poll discrepancies are the key to showing that the 2004 election was stolen.

But I do feel that there was a good amount of evidence showing that the election was stolen, specifically in Ohio, as I show here:
http://www.democraticunderground.com/discuss/duboard.ph...

and here
http://www.democraticunderground.com/discuss/duboard.ph...

[Febble]

86. Actually, BeFree

that isn't true. Many other people have demonstrated pro-Democratic bias in exit polls, including, as I said, people who have conducted actual experimental studies, in which two methodological conditions were randomly assigned to precincts, and in which one condition produced significantly more pro-Democratic bias than another. In a true experiment, with random assignment of variables, you can actually attribute causality - and they were thus able to conclude that bias resulted from methodological factors.

This is why the pollsters are so aware that pro-Democratic bias is a hazard in their polls.

It is true that I found evidence of this in the data; E-M also found it, and reported it, well before I came on the scene; Michael Butterworth also found it, independently, in a previous exit poll; it was found in UK exit and pre-election polls, where it is called the "Shy Tory" phenomenon.

So it is a well replicated finding. But for some reason it seems to bother you, despite the fact that I have said to you, on countless occasions, that it does not mean that fraud can't, or didn't occur. All it means is that polls are a poor basis on which to argue that fraud did or could occur.

If someone says 2+2=5, therefore the sky is falling, it means nothing. It does not mean that the sky isn't falling. But it does mean that any mathematician is going to say: "but 2+2 doesn't equal 5, so we can't conclude the sky is falling". And people may then ignore the other good evidence that it is. Or that something slightly less catastrophic, but nonetheless extremly serious, may be.

[Name removed]

92. Deleted message

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

[Febble]

94. No, the odds

of bias producing a 5 point discrepancy are not astronomical. They can't actually be computed at all.

But yes, let's stick to getting rid of machines that can't count and can't be checked. I'm really sick of trying to explain for the n millionth time why you can't use the binomial theorem to calculate the odds of bias in a poll.

And one good place to start would be to find out why machines with broken seals were allowed to be used on election day. The answer to that question might spell the end of the machines.

[Febble]

95. Failed to notice

on an initial read, that you are alleging that I was asked to cover for Mitofsky.

I don't cover for anyone. Please retract that assertion, for which you have absolutely no grounds.

[Name removed]

97. Deleted message

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

[Febble]

99. No, he didn't hire me to "cover"

He hired me to try "discover" the reasons why his poll differed from the count. "Discover" is the opposite of "cover". And yes, I "discovered" a lot of reasons, some of which weren't actually bias in the poll, but may have been miscounted votes, particularly in urban precincts using older technology. But I found strong evidence for bias in the poll. As others have done in previous exit polls.

And I am perfectly happy to respond to criticism. What is simply not allowed, on this forum, is for you to allege that I am a shill. As it says in the rules, you may not call a jerk a jerk, unless I am tombstoned. Then you can call me what you want.

[Name removed]

102. Deleted message

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

[OnTheOtherHand]

103. going after the mods now?

No use blaming it on Febble -- she's not a moderator.

[Febble]

104. For the last time:

I did not cover for Mitofsky. Cover means concealing. I did not help Mitofsky conceal anything. He hired me to try to discover the causes of the exit poll discrepancy - whatever that cause was, whether miscounted votes or bias. In other words to un-cover the causes of the discrepancy.

To allege that I accepted a contract to "cover" for Mitofsky is the same as alleging that I am a liar. I am not a liar, and I was not paid to lie. I would never accept a contract to lie. If I did, that would make me a shill. I am not a shill. To allege that I am a liar and a shill is to break DU civility rules, which forbid calling another poster a liar.

And yes, I was already a member of this board when I was contracted by Mitofsky. If you consider that that accepting that contract is a banning offence, then take it up with the mods. But while I remain a member of this board I am entitled not to be called a liar on this board.


edited to insert missing 'this' and 'that'

[BeFree]

105. No

I was not attacking you. I was attacking Mistofski. He couldn't figure out what his own problem was so he went for outside council. That, IMO, is his problem, his mistake, and shows he needed cover.

Why else would he go outside his organization?

You are the one making all this personal when it is a simple matter of E-M's mistakes. If it hadn't been you, he would have found someone else, eh?

[Febble]

106. No, you are alleging

that I was hired as Mitofsky's shill - that I was hired to cover for him. I wasn't, and I didn't. I am not making this personal, you are. You are also, of course, wrong about Mitofsky - he had a pretty shrewd idea of what the problem was, as he had already analysed his data in detail. However, I was critical of that analysis, and, far from attempting to "cover" himself from my criticism, he actually hired me to re-analyse the data using the approach I had suggested. I went into the job determined to find evidence of fraud if it was there.

And the reason, I assume, that he went "outside his organisation" was that he thought I had a good point - that the data should be re-analysed in the light of the criticism I had made.

He required no "cover" at all, and I did not provide that "cover". What he required was information - data analysis. That was what I was hired to do.

Now please stop alleging that I was hired to "cover" for Mitofsky. I absolutely deny the charge. I ask you to retract it. You are entitled to disbelieve me. But you are not entitled to call me a liar on this board.

[bleever]

82. Kick for a lucid post about a dense subject.

[truedelphi]

85. Keep up the great coverage... n/t

[lonestarnot]

90. We definitely have work to do.