None Dare Call It Stolen: Harpers Magazine article

[kansasblue]

On the news stands


"And on Election Day, twenty-six state exit polls incorrectly predicted wins for Kerry, a statistical failure so colossal and unprecedented that the odds against its happening, according to a report last May by the National Election Data Archive Project, were 16.5 million to 1. Yet this ever-less beloved president, this president who had united liberals and conservatives and nearly all the world against himself—this president somehow bested his opponent by 3,000,176 votes. How did he do it?"


http://www.harpers.org/ExcerptNoneDare.html

[aquart]

1. Mark Crispin Miller. Always articulate.

[OrwellwasRight]

66. I just bought the DVD of A Patriot Act

and I am going to host a party and have my friends over to watch it.

I'm very excited.

[WannaJumpMyScooter]

2. How did he do it indeed.

Ummmm. Theft?

[understandinglife]

3. "It is a veritable arsenal of “smoking guns”—and yet its findings may be

... less extraordinary than the fact that no one in this country seems to care about them.

That may be changing ......

Matthew Donaldson, an independent visual artist from Auckland. “On this, the 60th anniversary of Hiroshima's bombing, my grief for all of those murdered and yet to be murdered by America's weapons - of mass destruction or otherwise - overwhelms me” said Matt.

It overwhelms me, as well, Mathew.

Peace.

www.missionnotaccomplished.us - How ever long it takes, the day must come when tens of millions of caring individuals peacefully but persistently defy the dictator, deny the corporatists their cash flow, and halt the evil being done in Iraq and in all the other places the Bu$h neoconster regime is destroying civilization and the environment in the name of "America."

[Ellipsis]

6. Took the words right out of my mouth. n/t

[mikita]

4. I can't thank you enough for posting this...

I have been trying to convinvce my best friends that elections are STOLEN, and now I have Harpers (one of the magazine that THEY believe in) for them to read!!! These are the kind of folks, academic types, who need to see things in "believeable" print before they'll even consider it. So this is a BIG one.

One small step for mankind.....


Thanks a bunch!!

[ooglymoogly]

5. kick and nom

proof there is still honest jounalism.

[quisp]

7. Mark Miller is excellent n/t

[Fly by night]

8. Thanks for reposting this mention of MCM's article.

I am glad that Harper's has now posted the excerpt from the article on-line. In speaking with a Harper's editor several weeks ago, he said they would post the entire article on-line one month after it hit the newstands. In the meantime, let's keep buying copies and spreading them around.

[kansasblue]

10. I just saw it at the airport and I was a little shocked.

I had to buy it! Earlier I bought a Vanity Fair mag. They're kicking ass too. Toledo Blade is very good too.

[kansasblue]

9. and he didn't even touch on electronic voting problems

... but it works. Get them curious. Getting hunting for more.

[Fly by night]

13. Actually, he discusses electronic machine "glitches" in 50+ column inches

[mod mom]

11. Please buy copies and leave in public areas to be read. nt

[OnTheOtherHand]

12. oh, dear

Unfortunately, this is not Miller's best sentence:

"And on Election Day, twenty-six state exit polls incorrectly predicted wins for Kerry, a statistical failure so colossal and unprecedented that the odds against its happening, according to a report last May by the National Election Data Archive Project, were 16.5 million to 1."

Several problems.

(1) This seems to say that the polls indicated that Kerry should have won 26 states that he didn't win in the official returns. That is certainly wrong. (I suppose conceivably it might be read to mean that the polls indicated that Kerry would win a total of 26 states, which isn't true either.) Miller probably meant: "there were 26 states in which the estimates produced by the exit poll data overstated the vote for John Kerry by more than one standard error" (E/M report, p. 3).

(2) The state polls did show Kerry ahead in several states that Bush ended up winning in the official returns, but as far as I know, they were all within the margin of error. Being ahead by less than the margin of error doesn't count as a "prediction." (Even being ahead by more than the margin of error doesn't count as a prediction, as far as the people who conduct the exit polls are concerned. From their point of view, if they haven't called a state, there is no prediction.)

(3) This calculation appeared in USCV/NEDA's March paper revised in April, not in the May paper. (It's on p. 30 of http://electionarchive.org/ucvAnalysis/US/Exit_Polls_2004_Edison-Mitofsky.pdf )

(4) The calculation has nothing to do with who was ahead in how many state exit polls -- it is based on Kerry's overall margin in the national exit poll (which combines precincts randomly chosen from the various states).

(5) It is not at all clear that the error in the 2004 exit poll was "unprecedented" -- according to TruthIsAll's post, it isn't even especially large:
http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=334152

Yikes.

[electropop]

14. 17 states (last I heard) had poll results outside the margin of error.

There were several others that got flipped by lesser amounts.

Beyond the obvious problems indicated by the exit polls, the rigging was done by excluding voters. Numerous tactics including last-minute silent cancellation of registrations, false "felons" lists, improperly allocated quantities of voting machines, and thugs at the polling places. None of these things shows up in the exit poll errors, because those excluded neither voted nor were polled. They are nonetheless dirty, immoral, and in most cases, probably illegal.

You can try to ignore the facts, but you can't change them.

[TruthIsAll]

15. Sixteen state exits exceeded the MoE (Z-score >2) for Bush.

Here are the numbers and the probabilities.
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x386293#386381

[OnTheOtherHand]

19. "you can try to ignore the facts"

You seem to be responding to a post other than the one I actually wrote. Where on earth did you get the idea that I was denying vote suppression, or even rebutting the argument that the exit polls prove fraud (although in fact I don't accept that argument)?

Or maybe it was just a generic statement.

[Stevepol]

17. It's not only the flipping of the critical states that is disturbing

but the fact that over the country Bush presumably received 2.5% more votes than Kerry when the national exit poll indicated that Kerry had won by about 3%.

This to me indicates widespread fraud, i.e., the machines are defaulted to pad the Repub numbers.

To me, looking at the statistical analyses along with the by now thousands of other bits and pieces of evidence and common sense deductions, there's no question but that the 04 election was a fraud, and I don't think there's a court in the country where, if the facts were presented honestly, the voting machine companies wouldn't lose.

Kerry won the election hands down.

[OnTheOtherHand]

21. ok,, we could talk about that for a long while

I personally don't think the exit polls prove a darn thing about fraud (one way or another), and as far as I can tell, most polling folks agree. But what they add up to, in combination with all the other facts -- well, I'm at least open to argument on that. I just want to weed out the factually inaccurate arguments first.

Whether or not the 2004 election was hacked, the election system is hackable, and that stinks.

[Peace Patriot]

28. OnTheOtherHand, please see my reply, post #27, below. I posted

it in the wrong place. It is a reply to this, your post #21 ("Whether or not the 2004 election was hacked, the electon system is hackable, and that stinks.")

[OnTheOtherHand]

43. ok, that sequence makes more sense, thanks. n/t

[electropop]

37. What about Ukraine?

Were the exit polls useless there as well? They overthrew a government based on them.

[OnTheOtherHand]

42. Ukraine is interesting --

On Monday, at the Joint Statistical Meetings in Minneapolis, I heard a presentation about the Ukraine exit polls by an expert in polling in the former Soviet republics. She explained that exit polls were conducted by a consortium, an anti-consortium, and an anti-anti-consortium -- and, just looking at the consortium, there were four separate firms, each of which did its own sample.

Basically, she argued that the exit poll results from the first run-off were flawed (a bad precinct sample -- imposing a regional imbalance, a big problem since the election was very regionally polarized -- and faulty assumptions about when people would vote) and simply didn't provide enough information to call the election either way. She didn't say this in her presentation, but I think she basically thought that if the U.S. government had been on the other side, the first run-off would have stood -- right or wrong.

In my gut, I think Yushchenko won -- certainly he deserved to win, after surviving a poisoning attempt. And obviously the exit poll results were politically significant, whether or not they were valid.

[electropop]

44. Interesting details

The concept of consortia and anti-consortia (et al) seems like a darned good idea. Each would have to find ways to acheive credibility - "celebrity" (prominent pols) board members, or some kind of certifying commission, or mutual cross-monitoring. Given the disastrous state of our election system, we have to start somewhere.

[OnTheOtherHand]

46. healthy competition

I vaguely recall that some observers predicted disaster when the competing U.S. exit polls were winnowed down to one as a money-saving device -- and arguably they have been proven right.

There is some talk of fielding an independent exit poll (or some other independent way of verifying the vote) in future elections. I personally am not so interested in tinkering with exit polls per se, but I can see the appeal.

[Time for change]

18. I agree that Miller did mis-state some of this -- however

I would characterize the discrepancy between the national exit poll (and many state exit polls as well) as large and unprecedented. Bush's official margin of victory was 2.5%, and he lost the national exit poll by 3.0%. A difference of 5.5% seems large to me. Also, it is "unprecedented" in that no previous E-M poll has shown that much of a discrepancy, though it was almost as large in 1992 (5.0% I believe).

Also, the discrepancy in Ohio definitely did make the difference, and it was WAY outside the margin of error. Officially Bush won Ohio by 2.3%, and the exit poll had Kerry winning by 4.2% -- a difference of 6.5%.

[OnTheOtherHand]

20. hi, Tfc

I agree that the discrepancy was large (and won't quibble about "colossal," in this world of jumbo shrimp).

I sort of agree about "unprecedented," but (1) I think it is hyperbolic in light of the 1992 result, and (2) it seems worth noting that one of the 1988 exit polls showed Dukakis beating Bush. (I will let folks assess TIA's numbers, in the link I gave, for themselves. There are various plausible ways to measure exit poll error.) I think Miller is playing on a lot of confusion and exaggeration about how accurate exit polls have been in the past.

It's true that Ohio was well outside the margin of error ("way" is debatable -- the t-score, per the E-M report, is -2.2). As you know, lots of states, including both safe-Bush and safe-Kerry states, had larger errors. So, everyone agrees that the gap between the exit polls and the official results has to be explained (and the explanation should accommodate the fact that the biggest WPEs were in Vermont and Delaware).

It was an awful, awful sentence, but that doesn't discredit the entire article.

[Fly by night]

22. "...lots of states, including both safe-Bush and safe-Kerry states, ..."

...had larger errors"

And they all favored Bush. The few states (four, I believe) where Kerry did better in the "reported" vote than the exit polls had relatively small (within the MOE) deviations, as I remember.

TIA, can you or TFC elaborate on this -- it is pretty ancient news to me now.

[OnTheOtherHand]

24. yes, I know the large errors favored Bush (except

...in South Dakota, the 6.8-point Best Geo WPE favoring Bush is actually outside the margin of error, but the composite estimator incorporating pre-election polls pulls it back within; North Dakota is kind of similar).

It stands to reason that if the poll was biased in favor of Bush, it should be biased in favor of Bush in many different places. And -- whether the results reflect poll bias, election fraud, or both -- they are certainly ancient news.

[Time for change]

36. State exit polls

I'm out of town on vacation now, and I don't have my files with me. Here's a link to a thread where I discussed this some time ago: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=371726.

See section on swing states. Depending on what specific method you use, states that were outside of the margin of error amounted to 13-17 states, and they all favored Bush (though OTOH points out that one state could be considered outside the margin of error in favor of Kerry if one specific method was used).

The method that I referenced (from the Mitofski report) in the above thread had 13 states outside of the margin of error, all favoring Bush. 5 of 11 swing states were outside of the margin of error, and 8 of the other 39 states.

Not discussed in the thread was your question of in how many states did the exit polls favor Kerry (relative to the official count). I don't remember exactly, but I believe it was about 8, and the margin was very small in each case, and none of them were swing states.

This may be "ancient history" in some respects, but I nevertheless believe that it should form an important part of any overall assessment that looks at the possibility of fraud in the 2004 election.

[Time for change]

38. The bottom line is this

No matter how you figure it, the probability that either the state or national exit polls could have favored Bush by as much as they did, solely on the basis of chance, is astronomically small. About one in a million is the most conservative estimate I've seen. NOBODY, not Mitofski or anybody else would argue that this happened by chance. Mitofski said as much in his report. And the discrepancies were especially notable in the swing states (not only the swing states, but much moreso as a group than the other states as a group).

So the remaining question is whether or not bias could have accounted for the discrepancies. This is not something that can be figured out by looking at margins of error, etc.

[Stevepol]

23. Mitofsky himself admitted the discrepancy between the exit polls

and the alleged results in 04 were statistically impossible.

Are you disputing the statistics or the exit polls? If the exit polls, why not investigate to see where the problem lies?

Do you also dispute the exit polling in GA 02 (as well as all the pre-election polls in GA 02)?

Unfortunately, the only way to assess the accurcy of electronic voting machines is exit polls. The machine's owners and programmers refuse wherever they can easily do so to provide a paper ballot as a way to test the accuracy of the machines, the electioins officials stonewall audits and recounts, the vendors build a myriad of back doors and secret passages so the machines and tabulators can the more easily be hacked and patched, they lie about the accuracy of their machines, they use uncertified software, they bribe and line the pockets of the secretaries of state, they act for all the world like criminals.

At what point, are you willing to say if it walks like a duck and quacks like a duck, it's a duck?

[OnTheOtherHand]

25. point by point

Of course Mitofsky admits that the discrepancy is beyond the nominal margin of error. Miller could have found any number of ways to make that point accurately, and I wish he had.

I've been investigating the exit polls since last November, and also trying to pitch in on investigations of the official returns. Is there some particular form of investigation that you are assuming I might be against? I don't understand your question at all.

I don't know enough to have an opinion on GA-02 (and I don't really understand the point of that question either).

Exit polls really stink as a way of assessing the accuracy of electronic voting machines, which basically means that EVMs stink all the more. I won't stake out a position on whether the vendors are criminals -- it really doesn't matter. Any system that counts on a few companies' honesty is inherently unreliable. (It isn't even a matter of the companies, per se. A computer security guy said to me the other night that even if 99+% of Diebold's employees, including the CEO, were honest, a few people in the right places could rig everything.)

[Fly by night]

26. "I won't stake out a position on whether the vendors are criminals ..."

Since you've been around DU less than three months, welcome to DU. I haven't corresponded with you much, so I don't know if you've ever seen the report, "20 Amazing Things About Voting in the US" before. It deals almost exclusively with the EVM companies. Here are points 12-14:


"12. Diebold employed 5 convicted felons as senior managers and developers to help write the central compiler computer code that counted 50% of the votes in 30 states.
http://www.wired.com/news/evote/0,2645,61640,00.html
http://portland.indymedia.org/en/2004/10/301469.shtml

"13. Jeff Dean, Diebold's Senior Vice-President and senior programmer on Diebold's central compiler code, was convicted of 23 counts of felony theft in the first degree.
http://www.chuckherrin.com/HackthevoteFAQ.htm#how
http://www.blackboxvoting.org/bbv_chapter-8.pdf

"14. Diebold Senior Vice-President Jeff Dean was convicted of planting back doors in his software and using a "high degree of sophistication" to evade detection over a period of 2 years.
http://www.chuckherrin.com/HackthevoteFAQ.htm#how
http://www.blackboxvoting.org/bbv_chapter-8.pdf"

It is not a theoretical question as to whether the EVM companies have employed convicted computer fraud felons to develop their software. Most of us have been aware of these facts for a while now.

That's why it's always surprised me that Republicans fight so hard (at least here in Tennessee) to prevent ex-felons from obtaining the right to vote again while they are seemingly unconcerned when the companies (owned by prominent Republicans) hire computer felons to write the software programs that count all our votes. Maybe there's a difference between plain ole' felons and "their" felons.

If you haven't seen this report and want all "20 Amazing Facts", PM me and I'll send it to you.

[OnTheOtherHand]

29. good point -- the convicted felons are criminals by definition!

I found a different version of the 20 Amazing Facts, which may actually be more accurate on these points:

http://nightweed.com/usavotefacts.html

Its point 13 reads: " 13. Jeff Dean was Senior Vice-President of Global Election Systems when it was bought by Diebold. Even though he had been convicted of 23 counts of felony theft in the first degree, Jeff Dean was retained as a consultant by Diebold and was largely responsible for programming the optical scanning software now used in most of the United States." That seems to be what Bev Harris's book says, too -- that Dean was retained as a consultant. It makes my point rather nicely: I don't really care whether Wally O'Dell hired Jeff Dean to program back doors into the election equipment or not, and I certainly don't care what Dean's title was. If anyone is in a position to hack an election, it is a Bad Thing.

(I bet neither one of us stays up late at night wondering why Republicans are so dang inconsistent....)

[Stevepol]

52. The point of the GA 02 remark is that the exit polls and the pre-election

polls as well were even more out of whack than they were for 04. Roy Barnes had anywhere from a 9% to an 11% lead in pre-election polling up to 4 days before the election, then lost by 5% points, as much as a 16 point swing. He was never behind in any poll before the election. Max Cleland had about a 5% point lead in the pre-election polling (I don't think he trailed in any pre-election poll either but I'm not sure) and lost by about 7% or 8% points, a 12-13% pt swing, give or take a few points. The odds of these two things happening in the same election? I've read 960,000 to 1 but I think that's very conservative.

The exit polls were removed almost immediately from the internet, so I can't say what they were, but they must have been close to the pre-election polls because they were way off according to the note on the IT and the remarks of the polling rep who answered questions about the exit polls in GA. He said that his exit polls were all very accurate in all the other states where they were used, well within the MOE. Only in GA were they wildly off.

Of course, this is the first year that Diebold touchscreens were used exclusively for the vote and that means no audit, no recount. Just trust me. If a computer did it, it must be right.

The program that was used in the GA touchscreens was likely devised by Jeffrey Dean, an ex-felon working as a consultant for Diebold, and there are a hundred other facts that indicate something rotten in the state of GA.

My question again: At what point do you say that if it walks like a duck and quacks like a duck, it's a duck?

[OnTheOtherHand]

55. I would rather leave ducks out of it

I was trying to figure out why the ducks were bugging me, and then I remembered: this spring one of my students gave a presentation arguing that liberals, as a group, secretly advocate "secular humanism," and the whole presentation was organized around the duck metaphor. The ACLU opposes prayer in schools -- there go the liberals, quacking again! Politics and religion are confusing enough without injecting ornithology. Same here. I have to try to translate your question into questions that I can actually answer.

Am I convinced that Barnes and Cleland actually received more votes than Perdue and Chambliss? No. (Am I convinced that they didn't? Not hardly. As I said, I have no opinion yet.)

-- Calculations from pre-election polls alone are inherently unconvincing. The business about 960,000-to-1 odds is just wrong -- there is no way of calculating the odds of an election result from a pre-election poll. Or rather, one can make assumptions that let one calculate a P value, but there is no way to prove the assumptions. A 16-point swing is very weird, but we can't meaningfully say that it is one-in-a-million weird.

-- As far as I can tell from Lexis-Nexis, contemporaneous observers in Georgia didn't assume that the gov and Senate elections had been stolen -- so it would be pretty nervy of me to assume that they were wrong. But they may have been.

-- Obviously I can't speak to the exit polls without knowing how they turned out. I will try to obtain them via ICPSR this afternoon.

Do I trust the Georgia 2002 touchscreens, or any other touchscreens? Nope, not at all. Do I trust Diebold? no. Do I trust computers? no.

Could one believe that Georgia 2002 may have been rigged, but Ohio 2004 wasn't? Yes. As you know, Ohio counties used various kinds of vote mechanisms. Bush doesn't seem to have done better on DREs than elsewhere. And in the national exits, there is no clear relationship between machine type and error rate. We can probably say, at least, that hacking the touch screens would not have been enough to steal the 2004 popular vote -- whereas I assume it would have been enough to steal the 2002 Georgia elections.

Am I saying that touch screens weren't hacked in 2004? No. I don't know whether they were. I have seen no compelling evidence that they were. (Many folks cite visual indications of Kerry votes flipping to Bush -- if that was a hack, it was a really stupid one. A good touch screen hack would be invisible to the voter.)

Am I saying that Bush received more votes than Kerry in 2004? I think he did, yes, but that isn't an entrenched opinion that I strive to defend. I do think I owe it to the DU community to point out, at times, when particular arguments have unacknowledged weaknesses.

[OnTheOtherHand]

57. Georgia 2002 exit polls (raw results)

Senate: Cleland 200, Chambliss 202, other 2
Governor: Barnes 210, Perdue 191, other 1

So for gov, that's an apparent margin of 4.7 points for Barnes, whereas Perdue won in the final returns by something like 5.2 points (I'm using a CNN screen, probably not the absolute final).

For what it's worth, the nominal 95% margin of error for one candidate around 50%, with 400 respondents, would be +/- 5 points (or about +/- 10 points on the difference). So statistically speaking, this discrepancy is marginal. (I have no way of judging, at least immediately, whether sensible geographical weightings would make the results more or less anomalous.) And at this point, I think it would be crazy to assume that exit polls are usually unbiased, or that the 2002 exit poll in Georgia was. The exit results at least support the inference that these two elections could have been stolen, although I don't think they provide much independent reason to believe that they were.

[Stevepol]

65. Where in the hell did you get these exit poll results?

As far as I know they were pulled almost immediately off the IT at the time of the election and haven't been available since. I searched the IT for them and came up empty.

I assumed that they showed about what the pre-election polls showed: that Barnes was winning, as he won in every pre-election poll before the election, the last one 4 days before.

I also might ask why you would suggest that exit polls can't be trusted and are useless as a way of testing the accuracy of the election results as reported in cyberspace. Again as far as I know, the exit polls are the only possible check we have at present on the fraud that is being committed in every election. Would you suggest that the American voter just roll over and play dead?

Or maybe you have some constructive suggestion, some way of rooting out the fraud. If so, what is it?

[Time for change]

60. Electronic vote switching

I believe that you don't take the evidence of electronic vote switching seriously enough when you say "Many folks cite visual indications of Kerry votes flipping to Bush -- if that was a hack, it was a really stupid one. A good touch screen hack would be invisible to the voter".

I study I did of the National Electronic Incidence Reporting system (EIRS) showed that out of 94 reported incdidents that favored one or the other candidate, 87 of them favored Bush: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=371211

Are you aware that in Clint Curtis' testimony before Conyers' committee, he noted that he was told to develop a vote switching program that would be undetectable. He said that he could develop the program, but he couldn't make it undetectable. So, I don't doubt that an occasional screw up where the vote switch became detectable to the voter didn't necessarily mean that the hack was stupid.

One could just as easily (I think) dismiss the Nixon Watergate tapes by saying that they were probably bogus because Nixon wouldn't have been so stupid as to allow such tapes to see the light of day. Or, one could dismiss the medical evidence in the case of the Kennedy assassination by saying that the assassins wouldn't have been so stupid as to have him shot from the front because that would have made their plot medically detectable (which indeed it was, as evidenced by the testimony of several doctors who treated Kennedy at the hospital).

And, then one would need to explain why 87 of 94 reported incidents favored Bush. I'm not saying that there couldn't be some non-fraud reason for that, but I sure as hell can't think of a plausible one.

And lastly, I think that the fact that the software that counts our votes remains a "trade secret" and the fact that the owners of the machines are all die-hard Republicans (not to mention many of them are convicted felons), and I believe that the weight of evidence strongly favors the conclusion that these vote switches did indeed represent fraud.

[OnTheOtherHand]

61. yes, we shouldn't rule out detectable hacks a priori...

If our starting point is that a good hack would be undetectable, well, I guess we're done. ;) That wasn't the argument I was trying to make, but you are right to have some fun with it!

When I wrote that, for what it's worth, my starting point was my conversation with Bruce O'Dell about the BBV report on specific vulnerabilities. As I understand it, much of the analysis was based on a source code review that was possible because BBV had access to an, umm, Accubasic compiler and could thereby reliably disassemble the compiled code on one of the memory cards and see what it was programmed to do, and what "back doors" existed. The analysis concluded that the cards could be preset to a large number of votes for a particular candidate -- so large that when, say, 50 more votes came in, the total would roll over to 0. That particular flaw was detectable by analysis of the source code (or possibly it was established by other means -- I may have misunderstood Bruce, and I read the original report very quickly and some time ago), but it wouldn't be detectable to a voter.

This seems to jibe with Clint Curtis's statement as paraphrased/partly quoted at http://residentbush.com/Aftermath-2004_BradBlog.html : "Curtis explained to the Congressman that it would be 'virtually impossible to hide such code written to change the voting results if anyone is able to review the uncompiled source code.'" (By the way, if I remember rightly, Curtis wrote a mockup that would allow the vote to be hacked via 'hidden buttons' right on the touch screen -- or if that wasn't Curtis, someone else did it. Of course it would be very hard to do that in a way undetectable via source code review -- but it underscores in how many ways this technology is potentially vulnerable.)

Now, it would be stupid of me to argue that the 87/94 _must not evince_ actual fraud because the actual fraud 'should have been' undetectable by voters. For all I know, the DREs were hacked by different people in different ways in different places. And I have no particular explanation for the 87/94, and no particular desire to explain it away. But I don't quite grasp the hypothesis. I understand and accept your argument that these visible vote shifts would be radically underreported, but at the end of the day, I can't see a centralized effort to steal lots of votes visibly in lots of different places, if a centralized effort to steal lots of votes invisibly was technically feasible. It just seems too risky. So for now I have to stand by my original statement -- not that it wasn't a hack, but that if it was a hack, it was a really stupid one. (Or maybe a clever one designed to undermine confidence in the systems?)

[Time for change]

62. I think that one of the main problems is

that fraud in this election was committed in many different ways.

Why do that if "a centralized effort to steal lots of votes invisibly was technically feasible" (which I believe was the case and was done), as you say?

I think that it's obvious that nobody - on our side - can answer that with a great deal of specificity. However, the obvious general answer is that they believed that there was a limit to how many votes could be stolen in any one manner without making it obvious -- so they went for many different avenues. That's not necessarily the most obvious choice, but I believe that that's where the evidence points, especially given the lack of any apparent and plausible way of explaining the 87-7 split in vote switching incidents (which occurred all over the country but were heavily concentrated in the Democratic strongholds of southeastern Florida).

Also to cite some evidence of the serious under-reporting to EIRS: 8 cases (all in favor of Bush) were reported in Mahoning County, Ohio. But an investigation by the Washington Post discovered 25 machines in Mahoning County that were doing that kind of thing all election day long.

[Time for change]

39. Ohio

I agree with most of what you say here, though I wasn't familiar with that Dukakis exit poll you refer to.

I think Ohio needs to be emphasized here, since it made the difference in the election.

When I said it was WAY outside the margin of error, I was referring to the discrepancy between the raw numbers regarding Bush % of the exit poll vs. Kerry % of the exit polls -- not the WPEs, as discussed in the Mitofski report. Looking at those raw numbers, we have a discrepancy of 6.5%, and a p value of .001. That's a probability of one in a thousand -- which seems quite impressive to me, especially in view of the fact that similar numbers for arguably the next two most crucial states, FL and PA, were similar.

One could argue I think, whether these raw numbers are more or less important than the calculations based on WPEs presented in Mitofski's report. I think that one thing that favors the raw numbers is that, as far as I could tell, I couldn't decipher an explanation as to how Mitofski came up with the best geo-estimates, etc.

On the other hand, you could say that the raw numbers weren't weighted, but I'm not sure how important that is.

[OnTheOtherHand]

45. OK, I think we mostly agree

(On the Dukakis exit poll -- maybe TIA could source it faster than I can! I think it was the ABC poll that initially gave Dukakis a lead, but I'm not sure.)

The Ohio error was substantively large no matter what p value we end up using. And it's true that the raw result gave Kerry an even bigger lead than the "best geo" estimate (or the composite estimate, which was narrower).

From Mitofsky's POV, the poll hasn't made a prediction until he does -- and he never called Ohio for Kerry. Apparently he is staring at the estimators, not the raw results, on election night, so in that sense the estimators are more important. (Although the estimators are outside the margin of error for the official result of Bush winning by over 2%, they are within the margin of error for Bush winning at all. And if I were them, I would probably hold out for a t score of at least 3, not 2 -- buttressed by some official returns -- before making a call.)

On best geo vs. raw: I would think that geographic weights would be pretty darn important in a race that is sharply polarized by region. But I agree that the best geo estimates are totally opaque. I won't fault you for looking at raw numbers that we can at least decipher. I I think it's hard to calculate a standard error for a cluster sample once we discount the weightings. I could try a bootstrap resample of the Ohio results on a precinct level -- care to make a friendly wager about how that would turn out? (grin)

But let's keep it in perspective. I agree with your "bottom line" post -- the exit polls were way, way off, and no one thinks it was by random chance alone.

[Fly by night]

47. As I remember, the mistaken Dukakis call was for one state - California.

But I may be wrong -- that is just from memory. TIA/others, facts please.

[OnTheOtherHand]

48. the 50.3/49.7 wasn't a call...

It is widely (including by TIA) reported to have been the raw result of the national exit poll in 1988 -- although I'm not sure which one of the exit polls (apparently there were 4).

[kiwi_expat]

53. WPE is not weighted either

"One could argue I think, whether these raw numbers are more or less important than the calculations based on WPEs...
On the other hand, you could say that the raw numbers weren't weighted..." -TfC

From a PM from Febble (I'm sure she wouldn't object to my posting this)-

"From: Febble

"WPE is not weighted.

"It is simply the difference between the margin between the responses and the margin between the votes.

"So it is a direct (but not good, see my paper!) measure of error at the level of sampling or vote counting.

"WPE = (Kerry votes-Bush votes)/all votes

minus

(Kerry tallies-Bush tallies)/all tallies"

[OnTheOtherHand]

54. yeah, I assumed Tfc meant the Best Geo (which isn't WPE)...

In case anyone is playing along at home, the best geo and composite estimators appear on pp. 21-22 of the E/M report. (The composites are most likely to be used in election-day calls -- but these are the Call 3 results only; the results then get updated as official returns, right or wrong, come in.)

The actual errors in this table are calculated using basically the same formula as WPE -- but the overall error in a state wouldn't likely equal the average WPE in the state. In fact, we can see on p. 33 that the average WPE in Ohio was even larger than the estimator errors. (The WPE for Ohio was -10.6 to -11.2, depending on the exact calculation, while the error in the best geo estimator for Ohio was -8.6.)

Tfc may have meant that the best geo estimator is a weighting of precinct estimates, and so we can think of the error in the best geo estimator as a weighting of WPEs. Anyway, his basic point seems reasonable to me, but it's good to pin down the terms so we know what we are talking about.

[Time for change]

56. Thanks OTOH, yes, that's basically what I meant to say

I figured that since since the best geo and composite estimators seem to be stated in terms of WPE that they were based on WPEs, and so my statement was based on that assumption.

I have two problems with those calculations: One, it was very unclear to me from reading Mitofski's report, how they were calculated. Maybe some statisticians could tell (or maybe not), but I sure couldn't.

And secondly, since as you say they are a weighting of precincts, they don't give proportional weight to the larger precincts, even though the larger precincts contribute proportionally more to final vote count. The overall national exit poll discrepancy was -5.5%, whereas the overall WPE was -6.5%. Since calculation of WPE gives equal weight to each precinct, I would assume from this that the red shift was greater in the smaller precincts in general.

Is that a correct assumption?

[OnTheOtherHand]

58. estimator calculations

I don't claim to know much about how the estimators are calculated (I don't think they really want us to know, exactly), but I have a few thoughts.

First of all, I think it's confusing to think of an estimator as "based on WPE." The estimators are calculated before WPE is available (at least the Call 3 versions should be). However, the errors for the estimators are in the same units as WPE (difference in net margin).

The estimators certainly should give greater weight to larger precincts -- they shouldn't just average together all the percentages -- but I'm not sure I'm understanding the point you were making with that particular sentence.

I don't think the -5.5% and -6.5% are directly comparable in the way you suggest, because the WPE is based on all(?) the precincts in the state exit polls, whereas the -5.5% is just from the national subsample, and because not all error is due to within-precinct error (the precinct-sampling error was minimal, but not zero). But I haven't stared at those numbers closely trying to sort out how I think they ought to wash out.

Since the interviewing rates are probably based to some exten ton size of precinct, I think it is a good bet that red shift was greater in _larger_ precincts in general (see p. 36), and the greater WPEs in urban vs. rural areas may also support that hunch. A confounding factor here is where there were multiple precincts at a single polling place, so maybe my hunch is actually that red shift was greater in bigger polling places, which might not hold up at the precinct level. The whole multiple-precinct business is sort of nightmarish.

[Time for change]

59. Ok, thanks OTOH, I guess that one bottom line is that the method of

calculation for the best geo estimates and composite estimates is not really clear. So I think that that is one reason in favor of giving more credence to the raw numbers when calculating probabilities of a given discrepancy between exit polls and official results on the basis of random error (i.e., chance) -- though calculations based on the raw numbers weren't that far off from calculations based on the weighted numbers.

I forgot that the national sample was based on a subset of the total, so my theorizing about what the differences might mean would not be valid.

[Nothing Without Hope]

33. Yes, this is a problem, one the Rethugs would not hesitate to use to

try to undermine the credibility of the article. That's one of their favorite tricks - pick a detail to attack and imply that everything else is also wrong.

[Blue Shark]

16. Gore Vidal...

...Mark Miller, Jim Lampley, Bob Koehler all Dare call it stolen...and they are not alone.

...In 2000 we got over it and moved on and got screwed
...In 2002 we got over it and moved on and got screwed
...In 2004 we got over it and moved on and got screwed

...MY ASS IS TOO SORE TO MOVE ON AGAIN!

[Peace Patriot]

27. OnTheOtherHand, you treat the hackability of the electronic voting

machines as an afterthought. Yeah, that stinks, too, you say.

But, 1) you leave out an important couple of facts about these hackable machines--that 80% of the "official result" was tabulated by two companies, Diebold and ES&S, that both of these companies have Bushite or far rightwing connections, and that they tabulated the vote using SECRET, PRORPIETARY SOFTWARE; and 2) you don't put these things in the right order, thus...

Major Bush donors, campaign chairs, and billionaire funders of righwing causes, gained control of the 80% of the vote tabulation in 2004, using secret, proprietary programming code for the tabulation--code so secret that not even elected Secretaries of State were permitted to review it. The independent national and state exit polls showed a Kerry win all day long on election day--a major "red flag" for an election in which Bushites controlled the tabulation of the votes.

Do you see what I'm saying? You say, oh, that stinks, too. But you don't get its import. The election was non-transparent, and unverifiable--and controlled by Bushite companies, and...

AND...

Kerry won the exit polls.

---------

Another point that I think is critical is that these secretly tabulated results were then fed through an AP computer to all the major TV networks, where the exit polls were showing a Kerry win. The TV networks then FALSIFIED the exit polls data--they "adjusted" the exit polls to FIT the results coming from the Diebold and ES&S tabulators (Bush wins.)

They thus denied the American people major evidence of election fraud--and they did this knowing that, 1) this was a brand new election system being tested out nationawide for the first time--and one that had already raised controversy; and 2) who the owners of it were.

A non-transparent, unverifiable election system controlled by Bushites. Falsified evidence. And...

AND

Kerry won the exit polls.

----------

We owe the fact that we know that Kerry won the exit polls to alert alternative journalists who grabbed screen shots before the networks buried the information. And the networks have yet to disclose the raw exit poll data--not to a senior Congressman who requested it for an investigation, not to reputable researchers--9 Ph.D.'s!--who requested it.

You can't just talk about facts and numbers, OnTheOtherHand. You have to grabble with what these facts and numbers mean, politically, and as to people's perceptions of things. I myself was given the FALSE PERCEPTION on election night that Bush had won the exit polls. I soon knew he hadn't--but I was given FALSE INFORMATION, FALSE NUMBERS, POLLUTED DATA, in order to MANIPULATE my perception of the election. That tells me as much as the numbers themselves tell me--that it was in someone's interest to LIE to me, to manipulate me, and to make me THINK that Bush had won the election with no evidence to the contrary.

I don't cotton to that. I really don't. It makes me suspicious. And now--these days--I don't really have to know that Kerry won the exit polls to know that the Bush Cartel rigged this election. They've rigged everything else. They rigged a war. They rigged Halliburton contracts. They rigged Bush up with a machine to feed him talking points during the debates--when not even pens were allowed. They rigged his appearances with people who signed loyalty oaths. They've rigged the CIA so that it no longer has covert eyes and ears on WMDs around the world. They rigged the outing of a CIA agent, and endangerment of her life and the lives of all her covert contacts, by laundering it through reporters. They rigged the Niger documents, trying to prove Iraq had nukes. They've rigged everything they've touched. I no longer believe ANYTHING these people say--about Iran or anything else. I think they are major criminals and traitors. And I would damn well believe they'd rigged the exit polls, too--if that outcome had said to me that a majority of Americans had voted for these thieves and murderers.

But it didn't. No, it didn't. It said Kerry won the only independent, verifiable part of the election. The exit polls.

Context, meaning, connecting the dots. WHY did Tom Delay block a "paper trail" requirement for these electronic machines, owned and controlled by major Bush supporters? Why not, instead, insist upon a "paper trail"--if it is your intention to do an honest vote count? Why permit "trade secret," proprietary programming code--such a smelly little item--if you don't intend to use it to steal the election? Why give this appearance of fraud, if you don't intend to benefit from it? Why not just hold an honest, transparent, verifiable election?

[OnTheOtherHand]

30. look, I'm not gonna defend the Bush cartel...

I am focusing on the exit polls in this thread because the thread began by quoting two sentences from MCM's article, and then a bunch of people said nice things, and I thought someone really ought to point out that one of the two sentences was riddled with inaccuracies.

I have to disagree with your characterization of what the media outlets did with the exit polls. First, CNN posted raw results as the polls closed. Then, E/M reweighted the results to official returns as those returns arrived -- exactly as they do every time, exactly as they say they are going to do every time. I can't speak to why you had a false perception of this process (for one thing, I don't know what network(s) you were watching, or whatever), but there was nothing bizarre about the process.

I don't know why you believe that "We owe the fact that we know that Kerry won the exit polls to alert alternative journalists who grabbed screen shots before the networks buried the information." Thousands upon thousands, for all I know millions, of people knew on election day that Kerry was ahead in the exits. The tabulated surveys are now all available for download, as they have been in past years.

Even if the exit poll surveys were matched to original precincts, that wouldn't prove anything about who actually won the election. For that we need recounting efforts, wherever they are even possible.

Otherwise, I agree with a lot of what you say, and other parts I doubt but think are plausible. I do think that the idea of a conspiracy to suppress the exit poll results is unproven, very probably wrong, and more confusing than helpful.

[autorank]

32. Oh yes you are, that's your job. How's Ken Mehlman?

:hi:

[paineinthearse]

40. Mehlman's latest lies

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=1999619&mesg_id=1999619

[OnTheOtherHand]

41. huh? n/t

[foo_bar]

49. go directly to #10

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=132&topic_id=43868

[OnTheOtherHand]

50. LOL^2 -- thanks, that cleared it up nicely! n/t

[Time for change]

63. The re-weighting of exit poll results at about 1:30 a.m. Tuesday morning

I don't doubt that you are correct when you say: "E/M reweighted the results to official returns as those returns arrived -- exactly as they do every time, exactly as they say they are going to do every time."

But consider this: I (and I would assume most people) consider the presentation of the exit poll results to be news, since they were presented to us by CNN (and perhaps other news channels as well, I don't know).

As in the presentation of any news, since I believe that our mainstream news should be considered a public affair (which is why, after all, we used to have many federal regulations to govern it), in order to present the news in a responsible manner, there should be a significant attempt to present it in a responsible manner. This is similar to the issue of presenting the case for war in Iraq. If a news source presents the administrations official line on it without noting to its viewers that the official line is a lie, then I believe that is irresponsible.

That is not to say that the sudden adjustment of the exit polls at about 1:30 a.m. was a "lie". But what do you think that most viewers made of that change? My guess is that an extremely small percent of viewers have your knowledge of this kind of thing, which would enable them to realize that this is standard practive.

Rather, it looks to most people as if suddenly the exit polls changed. Most people would simply believe that these are the final exit polls. And why shouldn't they believe that? They were not labeled to say "These results are not really exit polls, but rather they are adjusted to reflect the official results". No -- they are simply presented as final exit polls.

Again, I'm not saying that the purpose was to deceive (though I'm not saying it wasn't either, and I don't have enough time to go into that right now). But I really do think that it was not responsible to present it in the way that it was presented.

And remember, the owners of the exit polls are the same MSM who decieve us on a myriad of other issues on a regular basis.

[OnTheOtherHand]

64. a matter of perspective

I see your point of view, and I'm not trying to refute it. I will try to convey what I take to be the likely point of view of E/M and their subscribers.

From your point of view (as I understand it), what we might call the raw results of the exit polls are a matter of public importance, because they shed light into the possibility of massive electoral fraud. And so, for the media to obscure, or to conceal, the discrepancy between the raw results and the official returns was at best irresponsible.

From their point of view, the failure of the raw results to come much closer to the official returns is a matter of some embarrassment, and certainly not "news we can use." I really doubt that Warren Mitofsky would agree with your characterization that "These results are not really exit polls, but rather they are adjusted to reflect the official results." The exit polls are designed to be adjusted in that manner, not to be used as an independent check on the official returns. The adjustment serves two purposes: it should allow the analysts to measure and compensate for actual bias in the survey and thus make quicker and better "calls" of various states, and it should allow for more accurate cross-tabs after the fact. Those both seem to me like legitimate purposes -- but they do assume the accuracy of the official returns.

I think the more stories one has heard about (apparently) egregious error in political surveys, the more natural it seems that these folks would -- rightly or wrongly -- assume that the discrepancy was poll error rather than evidence of fraud. And I don't see why poll error "should" be a big story (if it is poll error!).

We probably agree about a lot of other things the media has done wrong (partly because Democratic leaders haven't held their feet to the fire) -- on Iraq, on HAVA and voting methods, and migosh where does one begin, never mind end?

[Time for change]

67. That sounds about right

You understand my perspective correctly.

As far as Mitofsky's perspective is concerned, I'm not even going to attempt to guess what that might be, because that would have to include his motives. I don't know him, and I don't know his motives, and I don't want to guess as to what they might me at this time.

But I believe that it's fair to say that Mitofsky really isn't the main issue, since it is the MSM who owns the exit polls, not him. And my opinion of them is at most times such that if I expressed them in this forum my message could be deleted by the mods.

But I would like to add a critique of the perspective that you attribute to Mitofsky. I understand your explanation of why the exit polls are adjusted. But if the unadjusted polls are not considered to be news, then why are they displayed by CNN for everyone to see? In other words it just doesn't seem consistent to me. They display partial results all day long, for everyone to see. And then, at 1:30 a.m. they are suddently changed, giving the impression to most people (I strongly believe) one of two things: To the vast majority of the American people it seems as if the "final" exit polls acutally predicted a Bush victory; whereas, to those with a more suspicious and inquiring nature (such as many of us on DU) it looks like a deliberate attempt to cover up what the exit polls showed. In other words, what I am saying is, if they displayed partial raw results all day long, why couldn't they display the final raw results? -- even if they had to accompany them with a disclaimer.

Then, to your last point, regarding the errors in political surveys: The question here is whether there is more evidence of error in political surveys or fraud in this particular election. One could write a book on that question alone, so I'm not going to even attempt to do that here. I will simply say here that I think that there is a tremendous amount of evidence for fraud, especially in Ohio -- so I think that any attempt to resolve the discrepancy between the exit polls and the official count without seriously addressing all those issues is seriously flawed.

And lastly I want to say that I think that it is a good thing that you started this discussion -- because there are a lot of misconceptions about this issue, and I believe that they deserve public discussion.

[bananas]

31. It's the cover article

The cover is on their webpage now.
http://www.harpers.org/MostRecentCover.html

[Nothing Without Hope]

34. Audio file link for Miller radio interview on the article is on this page:

http://wnyc.org/shows/lopate/episodes/08092005

[MelissaB]

35. Thanks!

[tandem5]

51. thank you nt

[kster]

68. kick.nt

[garybeck]

69. this is a good article but it doesn't mention electronic voting.

this is a big problem. the few articles that do make it seem to avoid the biggest issue.

[OnTheOtherHand]

70. well, it _mentions_ it a couple of times

e.g. it refers to the machines in Mahoning County (Youngstown) and Lucas (Toledo). On the tabulation side, it discusses the role of Triad in the recount in Hocking County and elsewhere.

The article certainly doesn't offer a strategic analysis of what to do next. In his defense, clearly Miller had limited space.

[garybeck]

72. yeah, but this is my concern about the greater issue

in many cases, there has been too much focus on long lines in Ohio, Blackwell, etc. These are all important things, but there has been a lack of attention on electronic voting, Diebold, etc. In my mind if I could change just one thing for the next election, it would be the electronic voting. The other stuff, at least you can see it. When there are long lines people take photos and film and it gets some coverage. But with the electronic voting machines, it's the silent and invisible thief. It doesn't get coverage, and no one talks about it. and the bottom line is, it has more of an effect because it's not just one state. I wish someone like Miller who at least has a toe in the door at the regular media would write an article that focuses on electronic voting, Diebold, etc.

[AtLiberty]

71. The above link was just an excerpt -- here's the full article...

http://www.harpers.org/ExcerptNoneDare.html

I started a thread with the full article almost a month ago, btw...

[Name removed]

73. Deleted message

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