Votergate 2004; We Don't Need Paper to Prove Fraud,

[kster]

*An Oldie"



11/9/2004)
- By Sheldon Drobny, Op-Ed News

But We Do Need Money and Leadership, NOW.

Since last Tuesday there has been a justifiable uproar about the major differences between the exit polls in Ohio and Florida and the actual results. Democrats and Republicans, who both saw the same exit polls that showed an electoral landslide in favor of Kerry, have confirmed this. Investigative reporter Bob Parry confirmed from his sources that the Bush campaign was convinced they were going to lose. George H. W. Bush also confirmed this in an interview with The Today Show. So why have the exit polls been so wrong in the last two elections? It is clear that there must have been manipulation in the voting machines. While there's been a lot of talk of problems with not having paper trails, computer fraud is uncovered most of the time without paper trails.

As a former C.P.A and auditor, I have used statistical sampling throughout my career with great confidence. With electronic record keeping, it's easy to create a program to falsify the books. But there are ways to uncover that. Auditors have developed statistical ways to cut right through corruption in companies. You don't even need a paper trail. These statistical approaches can be used with almost 100% accuracy to uncover fraud.

With the votergate 2004 it's a numbers game just like it is with corporate accounting, even easier. All you're talking about is one number-- total votes for each candidate.

There's a huge difference between polling what WILL happen and polling something that has already happened. The reliability of polling something that has already happened is highly reliable vs. predictive polls, like Gallup or Zogby, which is very risky. The reliability can be, not plus or minus 4 percent as we see with predictive poplls, but rather a much more reliable plus or minus one half or one tenth of one percent with exit polls, because those are based on asking people who already voted. I would even say that if the exit polling were done in the key precincts of Florida and Ohio, which it was, then these results should be practically “bullet proof.”

http://www.zogby.com/Soundbites/ReadClips.dbm?ID=10385

WHAT IS GOING ON HERE? NOW THIS? AND STILL THE HONEST DEMS AND REPUBS, REMAIN SILENT ABOUT HOW THEIR RACES ARE BEING COUNTED AND TABULATED???? If you where in the race would'nt you be, the least bit concerned as to how the race was being tallied? COME-ON.

...........92% of Americans Oppose Secret Vote Counting Machines

Zogby official release

http://www.zogby.com/news/ReadNews.dbm?ID=1163

[OnTheOtherHand]

1. fact check

"So why have the exit polls been so wrong in the last two elections?"

The exit poll discrepancies were larger in 1992 than in 2000. (The blown calls in Florida in 2000 incorporated vote totals, not just interview results -- and they were influenced by absentee voting, which obviously are a problem for exit polls.)

"The reliability can be, not plus or minus 4 percent as we see with predictive poplls (sic), but rather a much more reliable plus or minus one half or one tenth of one percent with exit polls, because those are based on asking people who already voted."

Jeebus. Where to start? The most important thing is that Drobny doesn't point to any instance where the national exit polls actually were 'reliable' within plus or minus one tenth of one percent. It sounds good, but at best it is a circular argument.

Second, as Drobny presumably knows, sampling theory dictates that to get a survey result that is 'reliable' to within plus or minus one tenth of a percent, one would need to interview about one million people. Sampling theory doesn't know or care whether the question is predictive or retrospective.

And, by the way, "reliability" is the wrong concept here. In principle, a research method that consistently gives the wrong answer is highly "reliable," but not very "valid." (EDIT TO ADD: I'm not saying that exit polls are such a method. The national exit polls do seem generally to have overstated the Democratic vote share, but the magnitude of overstatement has not been consistent -- and apparently the ABC poll back in 1984 was pretty much spot on. And, no, I'm not assuming that the official vote counts have been accurate.)

"I would even say that if the exit polling were done in the key precincts of Florida and Ohio, which it was, then these results should be practically 'bullet proof.'"

The exit polling was done in a (complex) random sample of 49 precincts in Ohio and 55 or so in Florida. It is either wrong or meaningless to say that these were "the key precincts." It is likewise either wrong or meaningless to say that the state-level results should be "practically 'bullet proof,'" and astonishingly wrong to imply that they would be expected to be accurate within a fraction of a percentage point even if we could assume that the survey was unbiased.

[kster]

2. The vote counting is done in secret

and you try to explain why the exit polls where off, Do you have a crystal ball?

[OnTheOtherHand]

4. facts are facts, and the OP is not factual

If your point is that the vote counting is done in secret, you can make that point without having to state non-facts about the exit polls. Right?

[kster]

7. Can you get an accurate exit poll if

the votes are being counted in secret?

[OnTheOtherHand]

9. you might, or you might not -- but would you KNOW if it were accurate? n/t

[kster]

11. Thats where we are right now, isn't it.?.............nt

[OnTheOtherHand]

14. pretty much, which leads back to my post #4 n/t

[kster]

15. And my post #2....nt

[truedelphi]

12. Okay fact check, explain this

Presumably the same structured exit polls were "right" in Illinois
perhaps the only state that had worked very hard on having cheating
put out of the equation. Putting the cheat out of elections occurred
there not because of fer of Republicans stealing the vote because
of the experiences with the Daley machine (Democratic by the way)
in Chicago.
(It took many years of dealing with the Daley machine of Chicago to
help activists and visionaries come up with the type of voting system
that worked in Illinois, mroe or less across the state, for t
he Nov 2004 Presidential election)
unfortunately due to the misnamed "Help America Vote Act" (HAVA),
the election techniques and machinery used In Illinois in this Last
Presidential election (Honest Abe himself would have been proud)
will probably have to be scrapped in order to produce the kind of
machinery called for by HAVA

[Febble]

19. Well, Kerry's winning margin in Illinois

was 4.8 points less than the final exit poll estimate (made on the basis of exit poll responses only).

But I'm not quite sure what your question is.

[kster]

22. There has to be an OFFICIAL HAND COUNT

By the people, for the people, FOR PEACE OF MIND, No matter how the ballots get counted in the first place, there has to be a second and automatic, hand count, by THE PEOPLE and FOR THE PEOPLE of these Optiscanned ballots, at the precinct level.

There is absolutely, no reason with the use of OP scan ballots, that the people, should be DENIED this second hand count of the ballots that were op scanned, and if WE are DENIED, we need to come out swinging, Because it is indeed a totally fake democracy.

[OnTheOtherHand]

39. sorry I didn't see this before

As Febble says, the exit poll discrepancy in Illinois (based on the "Call 3 Best Geo estimate," the projection made from interviews alone) was 4.8 points, which is very close to the national average. The sample size is small enough in IL that one can't really tell whether the discrepancy there is "real" or just a matter of chance, but it is consistent with the national trend.

The basic problem with treating the exit polls as a fraud index is that there is no rhyme or reason. As I've mentioned here, one of the biggest discrepancies -- third in the country after Vermont(!?) and Delaware -- was in New York, where the initial exit poll projection had Kerry winning by 31.3 points. He officially won by about 17, which is consistent with pre-election polls. It's hard to figure that Kerry had a last-moment double-digit surge, and then Bush cancelled it out by hacking a lot of lever machines. It's more parsimonious to think that the exit poll was wrong. Looking at all the states together, I come to the same conclusion: the exit polls don't seem to work as an "index" of fraud that varies from state to state. (You can check out the graph on page 7 of http://inside.bard.edu/~lindeman/surprise01.pdf -- someday I should do a more user-friendly version of this, but....)

That doesn't mean that I'm vouching for the accuracy of the official count. It just means that I can't pretend to believe that the exit polls are bias-free, much less accurate within a fraction of a point. (Full disclosure: I do think that Bush got more votes in 2004, but even if that's true, it doesn't mean that the vote count is reliable, or the election system is fair. It doesn't even necessarily mean that Bush won, since Ohio is a subject unto itself.)

A lot of us here in NY are pretty happy with our lever machines, so I know what you mean about HAVA.

[Febble]

3. Two questions are raised here:

1. Is the vote counting accurate?
   
2. Are exit polls accurate?

This author of the article claims the answer to the second question is yes, and he is wrong. The fact that he is wrong about the second question does not mean that the answer to the first question is yes.

A lot of people got the answer to the second question wrong immediately after the election, including this guy. No point in reposting his mistake.

[starmaker]

5. are exit polls accurate?

We don't know because an accurate vote count is questionable
we should assume fraud in elections. Without a truely verified
ballot count we don't know if exit polls are accurate or manipulated.

[Febble]

6. There are lots of ways

of estimating the accuracy of exit polls, even without assuming the count is accurate. Exit polls simply do not have the accuracy claimed for them in the OP. Here are a few reasons why not:

1. They are not a simple random sample of voters. The voters are sampled from a sample of precincts.
   
2. Although a random sample of voters is attempted at precinct level, those who refuse to be interviewed are not a random subsample. This can result in "non-response bias".
   
3. Voters may be selected but "missed" if the interviewer is busy. If busy times coincide with clusters of voters for one particular candidate (and there is no reason why this would not be the case, seeing as voters do not arrive randomly at the precinct), then this will skew the poll.
   
4. When the "interviewing interval" is large, or when the interviewer has to stand a substantial distance from the polling place, unwilling voters may take the opportunity to evade selection.

Some of these effects can be observed in the data even without assuming the count is accurate. Indeed, accuracy can, and is, estimated without recourse to the count at all, by examining the variance in the precinct level responses. It also can be, and is, estimated by comparing precinct level responses with pre-election expectations. These estimates are the kinds of estimates used by the pollsters use to compute their margins of error. It was on the basis of these estimates that E-M warned its subscribers before any results had been reported that there was a pro-Kerry bias in the 2004 exit poll.

[starmaker]

10. Polls have a margin of error

It's the vote count that we question.
If pollsters are basing their #'s on history.
would then manipulated totals skew the expectations of
new election exit polls?

To me the accuracy of the exit poll exists only with an accurate vote count.

And estimating the accuracy of an unknown accuracy ?

[OnTheOtherHand]

16. to answer your first question, probably not

As far as we know, the exit poll estimates aren't torqued to compensate for past 'inaccuracies.'

But I basically agree with your main point: we can't evaluate how accurate the exit polls are without knowing how accurate the official counts are. There actually are good reasons to think that the exit polls were inaccurate, but that doesn't settle the question about the official counts. Both could be wrong. Well, both are wrong. The questions are how wrong they were, and why.

[kster]

17. Absolutely, And the main thing is

We need Leadership "Then" and Now. But Silence, From both parties?

Suspicious? I'd say so.

Neither party Cares to know how their very own votes are being counted and or tabulated, and or discuss anything about these secret machine counts or totals in public, Then and still now,

Suspicious? I'd say so.

The way I see it, the politicians hope that the exit poll discrepancy will, make people forget about the fact that the votes are being counted by "SECRET VOTE COUNTING MACHINES".

[Febble]

18. Well my point is

that there are baselines other than the count for the current election with which to estimate the accuracy of the poll, as well as years of investigation into factors that give rise to inaccuracy. Therefore, claims, as in the OP, that raw exit poll data should be "bullet-proof" are false, as are claims, made frequently elsewhere, that the only error in exit polls is what is called "sampling error".

To give one example: a few years ago, pollsters experimented with giving free folders to interviewees in an attempt to improve response rate. They randomly assigned precincts to experimental and control groups - they found significant differences in response rate between the two groups, but also significant differences in the degree to which the poll differed from the count. Ironically, the precincts with the higher response rates had the greater discrepancies.

Now, the point of doing a true experiment is that you control the variables. There could have been fraud in the election in which the experiment was done. But there is no reason to suppose there was more fraud in the experimental group than in the control group - that's why you randomly allocate your manipulated variable in an experiment. So you can infer, pretty strongly, from the result, that the thing you manipulated (the giveaway folder) resulted in bias.

Well, that tells the pollsters something about how not to try to eliminate bias. But it also confirms for sure that bias can be introduced by methodological factors. And the bias was in the direction of "redshift".

So yes, we know that exit polls can be biased; that the bias tends to be in the direction of "redshift"; and that the pollsters knew that they had a greater than usual "redshift" problem in 2004 before they had any results at all. That doesn't rule out fraud; it doesn't rule out fraud as the cause of some of the redshift in 2004 (although there are other reasons for thinking it was not a substantial factor). It simply says that the OP is factually wrong. Exit polls are not "bullet-proof" - this is known. It is not a guess.


Edit to clarify:

We might be at cross purposes here. It is difficult to estimate the accuracy of any given poll without reference to the count (although comparison with past counts, or pre-election polls can indicate potential problems) - what you can do is make estimates of the degree of precision you have. The OP indicates that exit polls have much greater precision than in fact they have. We know they are not that precise (even when they do not display apparent overall "bias") for the reasons I gave, i.e. because variance in discrepancy correlates with methodological factors.

[kster]

23. Wow, so if the vote count is done in secret

can you please explain why the exit polls has a chance of be off? I know you could explain it, I'VE read your work.

Your country fixed the problem, they banned the secret vote counting mahines, SMART MOVE.

Maybe you can help us do that, here in the great US of A.

[Febble]

25. I don't understand your question

Exit polls may or may not be "off" the true count. The official count may or may not be "off" the true count.

Having an accurate official count will not make the exit polls any more accurate - it will simply mean that we can be more confident about how inaccurate they were.

And I think a good way to get accurate, trustworthy official counts in America is to make sure that there is a paper record of every vote, and that all elections are randomly hand-audited. It's why I support HR550, and why I support those who are working on developing effective manual audit protocols.

Once that happens, maybe hand-counts of entire elections could be the next step. I don't know. Certainly if HR550 succeeds, and states across the nation discover that machine counts are unreliable, then the case for HCPB would become politically much easier to make.

[kster]

27. You don't understand, OK.....

keep doing what you do, Long answers so no one understangs, You are doing good. I'm impressed.

[Febble]

28. Well, you need to clarify your question:

"Wow, so if the vote count is done in secret can you please explain why the exit polls has a chance of be off?"

I don't understand what that question means.

OK, let me try to explain what I am saying again, and perhaps you can rephrase your question.

Exit polls are surveys. A survey can be "off" if the sample is biased. For example, if you did a survey about what people thought about gun control, and your sample of respondents included 85 men and 15 women, you'd have a biased sample. So you couldn't trust the answer. Fortunately you'd know it was biased, because we know that the population consists of roughly 50% men and 50% women.

Well, exit poll samples can be biased too. You might get too many women (women might be more prepared to be interviewed). You might get too few older people. You might be short of a particular category of people - say middle-aged male entrepreneurs, who thought they felt too busy or too important to hang around answering a long questionnaire. You have a rough idea of the bias, because interviewers note the age (roughly), race and sex of each person selected for interview who does not take part. But of course what they can't note is who that person voted for.

We know that the samples are biased, because we know that adjustments are made for the age, race and sex of non-respondents. What we don't know is whether they are biased by vote.

So to try to get back what I think might be behind your question - an exit poll may be biased, whether or not the count is correct. A correct count will not eliminate discrepancies between poll and count due to bias. In the UK we pretty well trust our counts to be correct. But we still have biased polls.

edited to fix typo

[kster]

30. Do you think votes should be counted IN SECRET? ...NT

[Febble]

32. No. n/t.

[kster]

35. Thank you......nt

[kster]

8. You are correct...........nt

[kster]

13. They do go hand in hand, no?........nt

[Febble]

20. In this sense:

fraud will tend to result in a discrepancy between poll and count.

What is wrong with the OP is the implication that the only factor that could produce a substantial discrepancy is fraud. This is not the case. Exit polls simply do not have the accuracy claimed in the OP. Methodological factors can introduce bias. It is not necessary to know that the count is accurate in order to establish this; it has been demonstrated experimentally, and by correlational analyses that show that methodological factors correlate with degree of discrepancy.

So the OP is factually wrong. That does not, as I said, mean that fraud did not occur in 2004, or in 2000. The exit polls came out closer to the count in 2000 than in 2004, but many forms of corruption would not show up in the exit polls anyway, including voter suppression.

[kster]

21. Why did your country ban "big brother style e-voting?

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

Was it because

1) The exit polls would be constantly off, because, trying to get an exit poll, correct, using secret vote counting machines, would be impossible, because you absolutely, need an accurate vote count, first and foremost to predict the out come of a race using exit polls.

2) They didn't think the vote count, from a secret vote counting machine would be accurate? or

3) They just didn't trust the machines? (PERIOD)

[Febble]

24. Number three, I expect

it certainly wouldn't be because it would upset the exit polls, which are often "off" anyway. In the UK the projections are regarded as "just a bit of fun".

Trust is the important issue, as I'm sure you agree.

[kster]

26. You know the vote count is being done in secret

in America, but yet you defend why our exit polls might have been off, I would not do that to the people in your Country, You are a sharp person, I know that, and I also know that, You know better. Thats what I don't get about you. You definitely know better.

[Febble]

29. Because I have analysed the data

and I know that the exit poll discrepancies correlated strongly with methodological factors. And they didn't correlate with benefit to Bush.

That DOES NOT MEAN THE COUNT WAS ACCURATE.

[kster]

31. Do you think that the vote COUNTING SHOULD BE DONE IN SECRET?

[Febble]

33. No. As I've said. On countless occasions.

[kster]

34. Thank you, So then you should be fighting with ME or US

As I Would be with you, exit polls is an attraction that gets everyone to forget about the FACT the the votes are being counted IN SECRET! I don't know where America screwed up, but thats what we did. Now we need all the help that we can get.

[Febble]

36. I agree completely with this.

Exit polls are a huge distraction from the real problems facing American Democracy:

• voter suppression
  
• corruption
  
• lax chain of custody of ballots
  
• unauditable, insecure, secret technology
  
• Few states with mandatory or effective audit/recount protocols.

This bill won't solve all these problems but it will go a substantial way towards solving some of them, especially if we can work towards improvements when it gets to committee stage. I've got a few suggestions, although what I can do is limited, given the fact that I am not an American citizen. But I'm with you all the way. And if this bill succeeds, you are a long way further forward with your own goal - you will have paper ballots to count in every state/

Cheers.

Lizzie

[kster]

37. See post 26......nt

[Febble]

38. Huh?