Florida E-Vote Study Debunked

[Kelvin Mace]

http://www.wired.com/news/evote/0,2645,65896,00.html

A study by Berkeley grad students and a professor showing anomalies with electronic-voting machines in Florida has been debunked by numerous academics who say the students used a faulty equation to reach their results and should never have released the study before getting it peer-reviewed.

The study, released three weeks ago by seven graduate students from the University of California, Berkeley's Quantitative Methods Research Team and sociology professor Michael Hout, presented analysis showing a discrepancy in the number of votes Bush received in counties that used touch-screen voting machines versus counties that used other types of voting equipment.

But Bruce McCullough, a decisions science professor at Drexel University in Philadelphia, and Binghamton University economics professor Florenz Plassmann released an analysis (.pdf) of the Berkeley report criticizing the results.

According to the Berkeley study, the number of votes granted to Bush in touch-screen counties far exceeded expectation, given a number of variables -- including the number of votes those counties gave Bush in 2000 -- while counties using other types of voting equipment gave Bush a predictable number of votes.

The analysis was not peer-reviewed, although Hout and the students said that seven professors examined their numbers. They would not speculate about what occurred with the voting machines, but voting activists on internet forums seized the study as proof of faulty voting machines or election fraud. Drexel University's McCullough, however, found fault with the study.

Kim Zetter is a good reporter and friendly to anti-BBV issues. I fear now that she will be attacked as (pick one):

A Freeper
Agent of Karl Rove
Diebold Shill
Sleeper agent planted in her mother's womb by The Evil Cabal(tm)

David Allen
www.thoughtcrimes.org

[nashville_brook]

1. so, which is it?

this doesn't specify what was wrong with the Berekley study -- just that it was not peer-reveiewed. if they weren't presenting as an "academic" work, it doesn't *have* to be peer-reviewed. i figured it was more a think-tank exercise, and given the timliness of the material, it seems the responsible thing to do to put it out there and get useful crit -- not a technical foul.

beyond being a prof of "decisions science" at Drexel Univ and prof of economics at Binghamton Univ (??) these folks don't appear to have more legitimate/impressive credentials than the berksters.

just my immediate reaction.

[reality_bites]

2. Here ya go....

Berkely study:

A George Mason University scientist completely debunks Berkely here:

http://elections.gmu.edu/Berkeley.html

Patrick Ruffini makes short work debunking Berkely:

http://www.patrickruffini.com/archives/2004/11/the_last_word_o.php

A non-fraud explanation for the Berkely result is pointed out by a University of Arizona professor here:

http://www.kieranhealy.org/files/misc/election.pdf

Another author adds another variable to the Berkely study - the Jewish vote - which entirely nullifys the Berkely anomoly. See Here:

http://newmarksdoor.typepad.com/mainblog/2004/11/more_on_the_stu.html

This site also makes the case that there are non-fraud related reasons for the results in the Berkely study:

http://rightonthemark.blogspot.com/2004/11/did-democrats-cheat-in-2000.html

[nashville_brook]

13. can you sum up their basic argument?

are these de-berkers all ringing the same bell?

[daleo]

25. Two of these seem to be right wing bloggers

So, I won't comment on them. The more substantive criticism, by the university researchers given in the links, is that the "voter machine" effect is mostly the result of two big counties, and if you take them out of the analysis the effect goes away (as far as I can tell). In other words, they claim this was a spurious relationship, and "something else" was behind the increase in the Bush vote in these counties, although they don't say what that was. Without really delving into the data, it is hard to say if this is true or not.

The other criticism is that the empirical results don't have an adequate theoretical basis - the claim is, it is mostly after the fact data dredging. I would agree that the paper was exploratory, and the model was a bit ad-hoc, but that is hardly unusual in the social sciences in my experience.

I agree that the paper isn't be enough to overturn an election on its own. But, I think it should be enough to initiate a recount, along with the reported voting irregularities. A recount of paper ballots would settle the academic and political issue. The fact that this is not possible, due to their being no paper ballots to recount, is the real scandal.

[distantearlywarning]

34. I agree with your analysis.

IMO, the most useful paper of the four cited is by the University of Arizona student. Two of them are totally worthless due to the clear biases that come through in their writing.

[papau]

28. Silly - making an equation to fit the reported results is easy-proves NADA

The debunkers have proven nothing other than the fact that an equation that "fits" 2000 to 2004 must have both plus votes for Kerry as well as Bush in order to "fit" on average.

They have certainly not shown that the math shows no fraud.

[jwcomer]

41. Their math is flawed

Papau,

I got into this when the paper first came out and offered specific reason's that the math and models used by Berkley were flawed. I got accused by several DU'ers or being a freeper. Go figure. The only counterarguments ever offered to my assessment suffered ad hominen fallacies - not unlike what I'm reading in this thread.

Summary of flaws:
1) the model used has a dual which leads to the opposite conclusion.
2) the t-tests assume Gaussian error distribution which is impossible given a domain constrained dependent variate. Ergo conclusions from those test may not mean much.
3) the robust fit should have picked up on the 2 (3 in my opinion) outliers and thrown them out, but this didn't happen. Why not?
4) their interpretation of the t-test is erroneous.
5) the overall chi-squared showed a weak fit for their model. Certainly not the sort of fit with which 99.99% conclusions could be reached.
6) Some of their independent variates aren't independent and there are non trivial constraints governing the relationships between some of the independent variates and the dependent variate.

There are more, but that is just off the top of my head without looking back at the paper.

regards,
Walton

[papau]

49. I agree JW - but the debunkers have not shown an indication of no fraud

They only show that the study did not give a 99.99% proof of anything.

I agree with your reasons 1,3,4,5,and 6 is a guess but likely. As to 2, it sort of goes with 4. But the there is something there that smells, at least to me. And I doubt that anything less than a precinct level analysis -with type of BBV noted, and also with type of precinct sumary operation, and location of summary location noted -will bring it out.

I grant I am biased from the 1960 gross vote fraud I saw the GOP do in downstate Illinois in 1960 (and when tossed ballots floated to the surface in the Desplaines River in the spring floods of 62, I noted that only media was the Waukegan News Sun which noted that fact - but then only because the fellow editing the publication of the Libertyville police blotter did not know better to toss it).

I simply come down wanting a complete stat anaylsis.

But your points are well taken.

[johnfunk]

40. It's spelled "Berkeley," not "Berkely," and the GMU "debunk" has big holes

I don't think you've actually read it -- it's a joke from a right-wing "university."

I have yet to read the rest, but if they're as bad as the GMU "study" (cue laugh track), then thank you for the free laff links!

[depakid]

3. I would hope that what people would want to see

is an honest and forthright analysis among experts- with full access to the actual data and to the proprietary code used to compile and tabulate results.

I think that's all most of us have wanted since the whole BBV issue first broke- way back in 1984 with David Burnham's NY Times article:

http://www.newsgarden.org/columns/burnham1.shtml

Ronnie Drugger's New Yorker article also turned a few heads (including mine) when it was published back in 1988:

http://www.newsgarden.org/columns/dugger.shtml

Hell, you can even go back further and see disturbing things. Check out this 1975 GAO report!

http://www.vote.caltech.edu/Reports/NBS_SP_500-30.pdf

Bottom line is that unless voting procedures are open, transparent and free from coercion, the US isn't any better than some petty, corrupt third world country- and there's not really any arguing with that....

[intheflow]

5. You are so right on.

We'll never know anything for certain until we see the proprietary code. :shrug:

[depakid]

11. Before HAVA was even introduced, the MIT/Caltech Report

Recommended in no uncertain terms to:

"Make source code for all vote recording and vote counting processes open source and source code for the user interface proprietary."

http://www.vote.caltech.edu/Reports/july01/July01_VTP_%20Voting_Report_Entire.pdf

Like that was going to happen with a Congress full of corrupt Republicans (and some corrupt and/or naive Dems).

[nashville_brook]

16. hmmm.... maybe take this thread

as sooooo five minutes ago.

:+

[Ima]

22. professor

http://216.239.63.104/search?q=cache:s7CPU7iPn5EJ:www.lebow.drexel.edu/admin/index.cfm/fuseaction/lebow_admin.faculty_detail/facultyid/59/+Bruce+McCullough+drexel+OR+philadelphia&hl=en&lr=lang_en



Ph.D. in Economics, University of Texas


Senior Economist at the Federal Communications Commission

[Ima]

24. plassmann wrong

Lott has posted some comments by Florenz Plassmann on the controversy about “More Guns, Less Crime”. Plassmann writes:

In their response to our paper, Ayres and Donohue have pointed out that our extended data set contains errors. Correction of these errors leads to estimates that differ from those that we published in our paper in Tables 3–8. As far as I can tell, the public discussion of the three papers in the Stanford Law Review has focused exclusively on this fact, and has used it as a reason to dismiss our entire paper as irrelevant.
Plassmann does not mention that correcting the errors causes the results in tables 3–8 to mostly go away. The discussion of the papers has focussed on the coding errors because Lott’s behaviour in refusing to admit to the errors for four months and dropping the clustering correction has made it into a big issue. An issue that is not just a normal academic dispute about who is right, but one that calls into question Lott’s integrity. I have discussed the other half of the Lott/Plassmann/Whitley (LPW) paper

http://216.239.63.104/search?q=cache:3xGJc96970MJ:cgi.cse.unsw.edu.au/~lambert/cgi-bin/blog/guns/Lott/more_guns_less_crime/index.plain+%22Florenz+Plassmann+%22+whois%3F&hl=en&lr=lang_en

[distantearlywarning]

35. WOW! This is not just wrong...

This web page implies that this Plassmann person was a co-author on a paper that was accused of data manipulation in a peer review. That's a big deal! If this is true (and it looks like it probably is) that would be enough to completely and permanently discredit this author in my eyes.

This website also suggests that his co-author Lott has a history of "spinning" his writing in such a way that it covers up his academic dishonesty and supports his claims. While it doesn't necessarily suggest that Plassmann does that, I also find that suspicious in light of how the Plassmann "Berkeley Debunking" is written (suspiciously hostile and overly emotional language usage).

I also find the subject matter of this disputed Lott-Plassmann VERY interesting... Here's my tinfoil hat conspiracy theory of the month: This professor has enough of a desire to promote a right wing agenda that he is possibly willing to engage in dishonest data manipulation in his own scientific paper - can we trust what he says about the Berkeley paper (the results of which would certainly be very threatening to a hardcore right-winger)?

I would repeat my sentiments of last night about this Plassmann paper (which were based on the tone of the writing and the alternative explanations they gave) - something seems a little fishy to me here, and I would be cautious about completely accepting their criticism of the Berkeley study. I think these people have an agenda to promote.

[Merlin]

46. Problems with "access to proprietary source code"

First, debugging code is difficult, let alone looking for holes and rigs in someone elses elaborate code. Especially true where updates are frequent.

Second, so long as the machine mfg. can "update" code at will, we have no way of knowing what code is in use during various parts of the election day.

There are more, but these two are killers.

[intheflow]

4. Charles Stewart?

(snip)

"Charles Stewart, an MIT political science professor, called the study 'the type of exercise that you do in a graduate data-analysis class' rather than as an academic paper."

"'If I were to get this article as (an academic) reviewer, I would turn it around and say they were fishing to find a result,' Stewart said. 'I know of no theory or no prior set of intuitions that would have led me to run the analysis they ran.'"

(snip)

Is this the same MIT Dean Charles Stewart who was quoted in The Oakland Tribune as saying, "There is an interesting pattern here that I hope someone looks into," after duplicating the Berkeley findings using the raw polling data?

It hardly matters that later in the article it states, "Stewart said valid questions about the election results in Florida and elsewhere remain unanswered." The damage was done in the title.

[Old Mouse]

6. I think the word "debunked" is being thrown around too casually

when we are actually dealing with different opinions of data. One expert's criticism of another's conclusion does not equal proving a hoax to be false.

With the large volume of stories and analysis and opinions flying around, we should not be using a word so freely that has the extra connotation of gullibility on one side and superiority on the other.

[Randi_Listener]

7. Who the fuck "bunked" it in the first place?

[nashville_brook]

14. yeah, we aren't talking about UFOs here either

i think we can leave behind the TFH imagery. we have affidavits and stats and history and movitive4 and opportunity. balls in their court.

[distantearlywarning]

8. I don't have enough knowledge to understand the stats in this one.

But their tone throughout makes me uncomfortable. In my field, even a "debunking paper" usually isn't written in the format of a personal attack on a scientist or a particular political orientation, and this kind of does. Examples of writing in this paper supporting my claim here: 1) the quote at the beginning seems a little harsh given that this is supposedly a scientific peer-to-peer criticism; 2) implying that the Berkeley study is somehow suspect because they called a press conference to release the results and putting the word RESULTS in quotations; 3) the study is completely without merit and the results are meaningless? (no authentic researcher in my field would ever ever make such a bold and arrogant claim about any colleague's study, even if they found totally conflicting results; 4) "apparently never bothered to check this aspect of their model" - again, overly excessive hostility that seems unwarranted in a "neutral" scientific paper; 5)"empirically vacuous"; 6) ALL the claims HCMB made are without ANY foundation??? Again, awfully arrogant and sweeping for a scientific criticism.

Overall, their mode of writing seems unprofessional to me. This is not the way any of my professors would present a professional criticism of another's work. That would be particularly true if the work was known to be that of a newer graduate student - most Ph.D.s I know are sensitive to the fact that students make mistakes and need informed guidance rather than condescension.

Additionally, the "alternative hypotheses" they present in their discussion portion kind of sound like Republican talking points to me, and don't seem to be all that viable from a theoretical standpoint (i.e., Bush got more of the Jewish vote this year? That's not what I've seen anywhere else... It also seems unlikely to me that all the evangelical Christians live in the most urban and Democratic counties in Florida.)

Something about the way this is written makes me think that this paper may not be totally on the up and up. I'm not sure that I would trust this particular "debunking". But again, I don't have enough stats to comment on this paper scientifically.

Can someone who is well versed in stats please tell me whether or not their actual empirical criticism is true/warranted/well done? I would be curious to know...

[indigo]

18. the tone is scathing because

You are absolutely correct that academics are normally far more reserved ("professional") in a critical follow-up. The reason is because if you anger your peers you may suffer a backlash the next time you submit a paper for publication. Peer review for publication is usually undertaken by anonymous peer review (only the journal editor knows who is reading your paper). After a paper is presented/published, commentary is public. So, if you criticize a peer's work, be diplomatic because they'll have the opportunity to get back at you without your knowing it.

There is one exception -- if something slips through that makes everyone (all the peers within a specialty/subfield "look bad"). A public spanking is necessary to restore the public trust. Graduate students are also easy to attack because they essentially have no stature.

I did not read HMCB, but, if their model is presented as described in McCullough and Plassman I am not surprised by the latter's tone. The criticisms read as a compendium of very basic errors that a beginning student of multiple variable regression is warned of in basic texts. If anything, they have shown restraint.

[papau]

29. "compendium of very basic errors"-true - but only because the grad

students pushed the wording in saying what they had.

Debunked does not mean not useful, and certainly does not mean the opposite has been proven.

[distantearlywarning]

36. Indigo - Read the link ref. in post #24 and tell me what you think....n/t

[George_S]

23. I don't understand it, but ....

Maybe this will help.

http://elections.gmu.edu/Berkeley.html

[SomthingsGotaGive]

32. Hello DEW

I agree and enjoyed reading you analysis.

This is the type of posts we need in these types of thread.
I'm so sick of hearing the "rovian plant" argument.

Has anyone else heard of this school before?

I'm not sure how this story will be received in the Heartland but here in Canada a Berkley study carries a lot more weight than a report from a school most, if not all Canadians, have never heard of.

[brindis_desala]

9. head in the sand rationalization

Democrats do themselves a disservice by refusing to face reality: Paperless voting machines are a mockery to democracy. Exit polls are trumpeted in Ukraine yet here in the U.S. we are denied the basic data even though exit polls were never questioned before the FOX shenanigans in 2000 and Jimmy Carter announced before the election that Florida would not pass the smell test. The truth DUers, is that your party needs to prove that it is not being led by effete elitist frauds who would rather trust Jeb Bush than a Democratic president. Sorry to be so blunt but until we clear the elitists and deadwood and get transparency back in the process this party will wane into irrelevancy.

[nashville_brook]

15. haven't seen many DU'ers or dems argue for more opacity

so, it's a point well-taken.

[Emillereid]

10. Have the Berkeley people responded to any of the alternate

analyses or made any attempt to defend their work?

[leftchick]

31. another question.... If the Berkeley people have...

will the MSM print it?

[distantearlywarning]

12. One more comment on the "Jewish Vote"

One of the other "debunkings" of this paper I just read in this thread referenced the Jewish vote again.

I really can't see why there would be any theoretical reason to assume that Floridian Jews voted for Bush. Across the nation, polls showed that over 75% of Jews voted for Kerry despite Bush's proclaimed policy on Israel.

Furthermore, just as an anecdote, I live in a predominately Jewish neighborhood in Pennsylvania. The Kerry signs in my urban neighborhood outnumbered Bush 10-1, and I believe my district went over 85% for Kerry. Why would we assume that Jewish districts in urban neighborhoods in Florida would be any different? Are Florida Jews different from Pennsylvania Jews? Are they different from Jews in the rest of the nation? I haven't seen one thing so far that suggests that Jewish people anywhere voted in large numbers for Bush, despite the claims of these "debunking" articles.

Can someone on DU claim differently and provide evidence? If not, I am going to assume that this "Jewish Vote" hypothesis is just one more "explanation" presented to gullible MSM viewers to distract them from actual problems.

Also, all of the "debunkings" of this particular study seem to be a lot more hostile and agitated than the stuff I saw about the Freedman study and the Dopp/Liddle report. It kind of makes me wonder if the Berkeley study WAS actually on to something (flawed model or not), and they are now getting scared and desperate.

[6th Borough]

56. Most Jews in Florida come frome the tri-state area.

(FYI, for us current and former residents of said geographic region, "tri-state" refers to Connecticut, NJ, and NY).

Much more from Jersey than Connecticut, and more again from New York (especially the city, Nassau County, and Westchester county). Must ex-Jersyites are former New Yorkers as well.

Not to mention retiree FDR "Greatest Generation" Democrats. If you mention Bush's name in a Century Village, you would be lucky to escape with all of your limbs intact. Though the death blows would rain down...slowly...

By nature, Jews vote overwhelmingly Dem; NY'rs vote overwhelmingly Dem, and elderly New Deal Dems vote overwhelmingly straight ticket Dem.

In other words, the Jewish vote in Florida is as liberal as nearly anywhere.

[alfredo]

17. I like

Sleeper agent planted in her mother's womb by The Evil Cabal(tm)

It rolls off the tongue nicely.

We got to be better than the Republicans. We have to stick to the truth, no matter how much we want it to be otherwise. It does us no good to put our hopes onto something without merit.

Let's let them figure it all out. If Berkeley is correct, fine, if they're not, fine. Let's act on the facts, and it appears there is a dispute over the facts.

I can wait.

[aikido15]

20. My favorite is...

Agent of Karl Rove...so covert sounding.

[alfredo]

38. No, my choice is better.

Mine has more words, so it must be superior to your choice. Take that!

:spank:

[lovuian]

19. Let me see Berkley I know who you are but Drexel is ???

In other words Berkeley has a Huge reputation and Drexel is where on the food chain of colleges???

[indigo]

21. well-known for engineering (n/t)

[Ima]

55. Ranks 50th in nation

Not exactly at the top I'd say.

[Ima]

26. Stewart

The UC Berkeley report has not been peer-reviewed, but a reputable MIT political scientist succeeded in replicating the analysis Thursday at the request of the Herald and The Associated Press. He said an investigation is warranted.

"There is an interesting pattern here that I hope someone looks into," said MIT Arts and Social Sciences Dean Charles Stewart III, a researcher in the MIT-Caltech Voting Technology Project.


http://216.239.63.104/search?q=cache:S-Z2H3Rf7yEJ:www.truthout.org/docs_04/112104E.shtml+%22Charles+Stewart%22%2Bmit&hl=en

[indigo]

27. hmm ...

Interesting. One of the first academic "studies" posted here at DU after the election was a study from MIT-Caltech Voting Technology Project. People here were not happy with it. A link to the updated version of the "study" is:

http://www.vote.caltech.edu/Reports/VotingMachines3.pdf

Summary of its position is on the second page. Basically,it comes to the exact opposite conclusion! No interesting correlations.

Not published, but it is linked from:

http://www.vote.caltech.edu/Election2004.html as a VTP report, and right below it is a link to a discussion w/ Stewart on NPR to "discuss these issues." So, I'd guess he's backing the position in the VTP report.

[papau]

30. hang their hat on NY - AN OUTLIER - where the type of fraud may have

been in the reporting - and not in the recording - of votes

But their points are well taken - a very detail precinct level analysis is need broken by machine type used.

[distantearlywarning]

33. I personally thought the Cal-Tech study had a major methodological flaw.

They assigned Op-Scan machines to the "paper" category in their analysis. This is a serious confound because although Op-Scan machines have a paper trail, the tabulated (computerized) results can be hacked by an outside source just like E-Touch machines. If your end goal is to determine if election fraud happened through the manipulation of electronic voting machines or tabulators (essentially all the machines that can be hacked electronically), then Op-Scan machines need to be placed in the non-paper category.

Otherwise, if you find no difference between categories on a regional basis, as they did, it could be due to the fact that half the hackable machines are in your "control group" and half are in the group you want to investigate. It's really bad methodology - if any of my undergraduate Research Methods students did that in a class project I would probably have to fail them.

Nobody has "debunked" that article based on this serious and easily correctable problem yet (except for me), but I guess it's easier to pick on graduate students than MIT professors.

I personally would like to see many more academic studies out there - both investigating the issue and critical of one another (in a "bi-partisan" way). I believe in the power of statistics and critical peer-review to get at the truth of a matter.

And I do agree with the poster who criticized the use of the word "de-bunking". The use of that word should be discouraged in scientific endeavors - it suggests that someone now knows all there is to know about a problem, therefore discouraging further thought, discussion, or study. And that's the last thing we need. (That's why I've been putting the word "debunking" in quotes during my posts on these issues.)

[papau]

50. I agree :-)

:-)

[JohnOneillsMemory]

37. Can't wait to hear Zogby at Conyers' voting fraud forum today. 12/8/04 n/t

[lakeguy]

39. hmmm..this guy is upset because it wasn't peer reveiwed...

but that could take up to a year. in addition, his correction wasn't "peer-reveiwed" so why should we take it seriously?

[Redleg]

42. Good point.

Some people here at DU seem to take glee in seeing the voter fraud accusations "de-bunked." I wonder why...

[distantearlywarning]

43. None of these studies has been peer-reviewed, either for or against.

Does that mean that all of these authors are completely suspect and irresponsible? This is a timely issue, and the data should be put out in the public sphere ASAP with the understanding that it *hasn't* been peer-reviewed. Authors shouldn't have to be afraid that their integrity will be attacked if they widely distribute their data in an effort to serve the public interest.

[Merlin]

44. Here's my FL analysis. Finds Op Scan problem in '04, but also in '02.

Just straight-forward numbers here. No "t-tests", "chi-squared", "multi-linear regressions" and other statistics buzzwords. The numbers in basic math form alone tell the story.

I did this right after the election, but then the issue disappeared from public focus.

Analytical Method:

-- I took the total votes for each county and broke them into EXPECTED votes for Kerry vs. EXPECTED votes for Bush, based on party registration. (On a binary basis--e.g. disregarding 3d party candidates.)

-- Then subtracted EXPECTED from (binary) ACTUAL votes for each, and came up with a + or - VARIANCE.

-- Divided VARIANCE by EXPECTED to come up with a PERFORMANCE %.

-- Then did the same thing for 2000 results of Gore vs Bush.

-- Then subracted the 2004 variance from the 2000 variance to show significant differences between the two elections.

Results are shown separately for touch screen vs. op scan.


Touch Screen Machines
---------------------
Results show that for '04 we did just 3.5% worse than expected. Not a serious difference by itself. However, in '00 (before the machines were in use) we did 14.3% better than expected. Now the difference becomes 14.3%+3.5%, or 17.8%, and that gets to be problematic.

Optical Scan Machines
---------------------
Here the results show a serious negative variance of 18.4% for JK, and a plus variance of 18.4% for Bush, for a total of 36.8%.

But, ironically, while that is a remarkable fact, nonetheless the PERFORMANCE % in 2000 in these counties was far worse! Kerry actually outperformed Gore--as to Expected vs Actual, by a whopping 228.8%. That's one reason why debunking comes easily. However, the serious negative variance in these counties still needs to be examined carefully.

Essentially this analysis shows that counties with Op Scan machines had huge negative differences in '04 between Expected Democratic votes and Actual Democratic votes, but that those differences were far worse in 2000.

If somebody would like the .xls file to do further analysis (like taking it back to 1992 and 1996--an important contribution), let me know and I'll give you a link for it.

[papau]

52. Piece of data re Florida optical scan 2000 results. The error correction

switch - allowing over and under and double votes to see the error and correct - was turned on in White GOP areas, but was turned off in minority, heavily Dem areas.

The result is a better than 8% throw away in Dem areas, compared to a little over 1% in GOP areas.

And this is using the same equipement.

The only hope of finding the various GOP frauds is a Stat analysis - and it need more detail than can be done in a week or two.

[Igel]

53. This assumes randomness in machine distribution. n/t

[dbonds]

45. They used Multilinear regression

This is not a bleeding edge technique. All it depends on is finding the factors for all the variables in an equation. Pretty straight forward stuff.

[Merlin]

47. But correlation analysis is always suspect. And it buries the problem.

By hypothesizing based on correlation, we're always subject to the accurate accusation that "correlation is not causation."

Then the news story shifts from the dramatically unexpected pattern of results to the possibility of error in the hypothetical explanation for it.

We wind up losing when we engage in this kind of correlation conjecture.

The only persuasive case is one based on fact.

[distantearlywarning]

48. I agree with your assessment. The problem is...

you can't do a truly experimental study on this kind of data. The only "smoking guns" are the correlational statistical analyses showing highly improbable outcomes.

What we really need is a whistleblower or 50. Preferably with impeccable credentials and credibility. Or maybe a spy at Diebold...any programmers out there want to undertake a risky mission for your country by applying for work at a voting machine company? :-)

[papau]

51. That is what happen with smoking deaths from 60 to 1995

But the Stat "proof" held up!

[Igel]

54. The stat "proofs" got experimental confirmation.

In other words, causation was independently proven.

Moreover, hundreds of studies examined many possible sets of assumptions, and looked for underlying connections. They found some, but the correlations held. In other words, the stats did exactly what they were expected to do.