Mystery Pollster - Is RFK, Jr. Right About Exit Polls? - Part I

[kpete]

Mystery Pollster
Demystifying the Science and Art of Political Polling - By Mark Blumenthal
June 05, 2006
Is RFK, Jr. Right About Exit Polls? - Part I

Late last week, Rolling Stone published an article by Robert Kennedy, Jr. that asks provocatively, "Was the 2004 Election Stolen?" While it covers many topics involving alleged suppression and fraud in Ohio, the article disappoints in its discussion of the exit poll controversy, because on that aspect of the controversy Kennedy manages to dredge up nearly every long-ago discredited distortion or half-truth on this subject without any acknowledgement of contrary arguments or the weaknesses in his argument. It is as if the exit poll debate of the last eighteen months never happened. With this two-part post, I want to review the article's discussion of the exit poll controversy in-depth, for it provides a good opportunity to learn something about what exit polls can tell us -- and mostly what they cannot -- about whether fraud was committed in the 2004 elections.

But before getting to exit polls I want to make two things clear. First, despite its weaknesses, the Kennedy article raises some important and troubling questions about real problems in Ohio in 2004. As Ohio State University Law Professor Dan Tokaji puts it, the article is "useful in exposing how shoddy election administration practices can result in lost votes, and how some recently enacted laws will make things worse rather than better." The summary of problems deserving attention includes long lines in minority precincts, efforts of the Republican Party to selectively challenge (or "cage") new registrants and the many examples of pure incompetence by local election officials. And then there is partisanship of Republican Secretary of State Ken Blackwell, now his party's nominee for governor. Blackwell will need to answer to Ohio voters for, as Salon.com's Farhad Manjoo writes, having "used his powers for partisan gain," issuing "a series of arbitrary and capricious voting and registration rules that could well have disenfranchised many people in the state" (but interests disclosed: I am a Democratic pollster with clients in Ohio)

more at:
http://www.mysterypollster.com/main/2006/06/is_rfk_jr_right.html

[benddem]

1. RFK synthesized a number of points

made by researchers and did them well. The criticism about exit polls is really suspicious. In Germany exit polls are used to announce the candidates. They have hand counted paper ballots, but the election results are made right after the polls close. It really is an exact science...if someone isn't screwing with the results....or the ballots.

[BeFree]

2. The problem with the data

The data that would clear things up is being kept a secret from anyone but the pollsters.

The pollsters screwed up, is what they say, but whoa! we figured out where we screwed up, and, NO we won't let you look at the data. You just have to trust us.

The present excuse for keeping the data secret is that it might expose a gay black man living in a small town somewhere, and his neighbors might be able to tell who he voted for. I am not kidding.

Ya know, this ain't rocket science - they need to release the data to a Dem analyst.

on edit
Mystery pollster says this: The WPE (within precinct error) data shows the exit-polls being 6.5% off. I believe the Within Precinct polls are the most accurate of polls. It is the WP data that is being kept a secret.

[BeFree]

9. Ohio data was released

If you have any questions about exit polls and whether they should be used to audit elections, check out this link:

http://electionarchive.org/ucvAnalysis/US/exit-polls/Ohio2004-US-future.pdf

Summary: The Ohio data that was released shows that precinct level discrepancies aka WPE, must be due to machine miscounts. Almost all in favor of Bush.
___________________

Since the machines can't be recounted, and since Ohio recounts were obstructed, exit-polls are the only way to examine the vote counts, and this review of the data clearly shows the importance of having access to the data from other states.

[Febble]

21. No, you are not kidding

that is exactly right.

Really, BeFree, you need to find out a bit more about this stuff. Have you any idea what you actually mean when you say "the Within Precinct polls are the most accurate of polls"?

Would you like me to explain?


(edited for typo)

[OnTheOtherHand]

3. well, not exactly

The last German exit poll projections by ZDF/FG Wahlen misstated the margin between the leading parties by 3 points -- between the leading coalitions by 3.8 or so. Not so exact.

Courtesy of a British colleague (no, not a DUer, AFAIK), you can follow the whole thing here: http://service.spiegel.de/cache/international/0,1518,375257,00.html , start at bottom. They did improve the estimates pretty quickly.

Anyway, that's in Germany. I keep asking this question: if we think that U.S. exit polls are accurate, then do we think that John Kerry really won New York by 30 or 31 points?

[billbuckhead]

4. Yes, John Kerry could have won New York by 30 points

Bush doesn't do very in the Northeast.

[OnTheOtherHand]

5. well, no, he doesn't, but

there were four NY statewide polls in the last week, and they showed Kerry +15, +16, +17, and +18 (not in that order). The official returns had Kerry +18.3. So +30 doesn't make much sense.

[papau]

6. but if average final poll was predictive by state, do the Bush results

by state jibe any better?

The red shift appears more or less uniform across the nation I thought - so my conclusion is that all GOP final vote tallies were screwed with to about the same extent - the machines were less important than the dishonesty of the Republican Party.

But in Ohio we can see how it was done. Without random - really random and not selected by the GOP - audits of precincts, plus review of machine placement, registration procedures, etc., we will never get close to a fair election.

The GOP are the classic psych class "transference" example. Because they cheat, they are worried about the other side cheating. Indeed their excesses are just competitive - they will not be out cheated and indeed must "win" the cheating competition by the largest score possible - even if the other side has little cheating going on.

To my mind the hispanic result in the official final exit poll proves that fraud was going on - other more limited area exit polls of heavily hispanic populations show no shift from 2000 - yet our official exit poll tells us Bush was in the mid 40's on election day. Then current polls tell us he has "lost" that hispanic support -

Yeah - right ... :-(

[Awsi Dooger]

7. Here's some PEW data on Hispanic partisanship trends post 9/11

This link was available pre-election, regarding 9/11's impact on party identification toward the GOP and what it likely would mean in the 2004 election. Along with the pasted paragraphs, note the table from this link: Partisan Trends Among Hispanics 1997 to 2003 and particularly the change post 9/11: http://people-press.org/reports/display.php3?PageID=750. Too many DUers want to pretend 9/11 didn't impact the 2002 and 2004 elections, specifically white women and Hispanics who drifted to the GOP due to national security concerns.

"Hispanics and Latinos have also been an important constituency for the Democrats in many parts of the country. But there has been a somewhat larger partisan shift away from the Democratic party among Hispanics than among the public at large.

During the late 1990s, Democrats outnumbered Republicans among Hispanics by a margin of more than two-to-one (41% to 19%). In the aftermath of 9/11, Democrats still lead, but by a smaller margin (36% to 22%).

Republican gains have been greatest among Protestant Hispanics especially those who consider themselves evangelical Christians.

Among Catholic Hispanics, there has been little change in partisan identification.

The Northeast is the only region where the Democratic party has held its own. Hispanics and Latinos living in that region are just as Democratic today as before Sept. 11.

But in several key battleground states in other regions notably Florida Republicans have made gains. With its conservative Cuban-American population, the Sunshine State's Hispanic population is among the more politically diverse in the country, though Democrats outnumbered Republicans by 12 points during the late 1990s. Today, Republicans have a slight advantage over Democrats, 32% to 30%."

[papau]

8. That avoids fact that % of hispanic vote to GOP steady at 33 to 36% for

last 10 years - both pre and post 9/11. Granted Cuban moving to GOP in Florida is real and is as reported by PEW. But Pew does not contradict the election exit poll data of the hispanic exit poll firms. A 36% to 22% current Dem advantage does not translate to a 44% Bush vote.

The jump to 44% that is required to have the exit poll conform to the recorded vote is BULL per the other exit polls taken at the same time as they were in 2000 and 2002 that showed no change in 2004 for the split between GOP and DEM among hispanic voters.

The so called shift to the GOP is a cover for the election theft - or how else does one explain New Mexico, etc and the other hispanic area polls that say the results per the national exit poll (after adjustment to get a best fit with "actual" vote as reported by the various states) for hispanics is bull crap?

[Awsi Dooger]

10. The 44% in 2004 was probably too high

I read reports estimating it was closer to 40%, perhaps a point or so higher. The national exit poll apparently overstated Miami Cubans, and understated many demographics favorable to Kerry, like Hispanic women and Hispanics who live in large cities. The aggregate state exit polls, as opposed to the national exit poll, indicated 40-41.5% may have been more accurate.

Two of the findings from those national and aggregate state exit polls were identical to what PEW forecast -- party identification shift toward the GOP and Hispanic protestants voting in much greater percentage and heavily toward Republicans.

Here's an analysis of the 2004 Hispanic vote: http://www.nyu.edu/gsas/dept/politics/faculty/nagler/abrajano_alvarez_nagler.pdf

[papau]

12. I love testing a false 44% with the data that claimed it was 44%-"only NEP

data" was used because the only polls that I refer to had "data sets too small" for analysis, per the pdf you linked to.

PEW did not forecast anything that I can see other than party self identification had a change that left 78% of hispanics not identifying as GOP, which was better for the GOP than before.

Sorry, NEP exit poll data certainly did not show that, after adjustment to fit the "real vote", that it had any use in any discussion of voting trends, when it can not explain the massive hispanic error in its after adjustment results.

A massive hispanic error that was looked at by the exit poll folks - and no way to reconcile the small independent hispanic polls and exit polls to the NEP data was found.

But I do find amusing current media stories about a shift away from Bush by Hispanics, based on media comparisons of NEP data to current polling. Nobody wants to note that the hispanic polling has been constant and the NEP is the "outlier" - although "outlier" implies more stat value than I think it deserves.

[Awsi Dooger]

14. PEW forecast at least a 7 point swing in three of the four regions

7, 7, and 10. All but the northeast.

You can dismiss the 36-22 as 78% not identifying with the GOP. Or you can consider the previous 41-19 and look at it as an 8 point net swing post 9/11. That 8 points would fall in line with the rise from 33 or 36% to 40%. The PEW sample was 3850 nationwide.

You didn't address the Hispanic Protestant issue, mentioned by PEW and NEP. I tend to pay attention when a specific prediction is made and evidence indicates it held up. Protestant Hispanics jumped from 25 to 32% of the Hispanic vote and went 56% for Bush.

We really don't know how Hispanics voted in 2002. I remember 10 FOX election day polls in the battleground states and some of them a joke in reference to Hispanic totals, like South Dakota which has maybe 1% Hispanics and Minnesota with less than 3%. The indications were Hispanics voted the typical 33% for GOP senators but much higher, closer to 46%, for GOP governors. So your assertion Hispanics didn't move toward Republicans immediately post 9/11, nothing higher than 33 to 36%, is not something I fully accept, given the flimsy data from 2002.

I show one 2002 House estimate in my notes indicating Democrats won the Hispanic vote 62-38. In fact, now I found that source, Emerging Democratic Majority: http://www.emergingdemocraticmajorityweblog.com/donkeyrising/archives/001227.php

They also seeem to be accepting the 40% number from 2004, based on aggregate state polls.

Regarding the current shift of Hispanics away from the GOP, they been turning on the TV, right?

[OnTheOtherHand]

11. what we see is that

the states that were furthest out from pre-election polls aren't states with large red shifts. (No, the red shift isn't uniform at all. Ironically, at precinct level, that is one of the few findings of Ron Baiman's that I agree with. But it's pretty obvious at state level; TIA used to post a pretty picture of this, which he used to argue propositions like "THE BEAST WAS IN THE EAST!" And indeed most of the largest red shifts were in the Eastern time zone.)

It's hard to answer unambiguously your question about (as I understood it) whether the official returns are closer to the pre-election polls or to the exit poll projections, because some folks have argued that the pre-election polls were biased against Kerry. But using the pre-election polls as even a crude baseline, I don't see anything in the official returns that is nearly as hard to believe as the exit poll result that Kerry won New York by 30 (actually 31.3 by the Best Geo estimate), or that Kerry won New Hampshire by 15.

Honestly, even Kerry winning Ohio by 6.5 is a big stretch. Even the folks who think they've made up Kerry's 118,000 vote deficit in Ohio don't seem to think they have turned around and found him another 350,000-plus votes to win by. And if we are basing those numbers on exit polls, they should apply just among people who would have participated in the exit polls, i.e., presumably people who at least thought they voted.

So, frankly, I don't know anyone who acts as if s/he believes the exit polls were right when it comes to analyzing specific states.

As we've discussed, I just think the Hispanic issue is a mess, and even if we conclude that the final NEP Hispanic estimate is way off, I don't see how we can conclude that all the NEP estimates would have been basically right if they hadn't been weighted to the official returns. But I haven't even looked at Awsi's stuff downthread yet.

[Stevepol]

18. I don't understand the figures here.

I couldn't find any results of the real elections to compare with the exit polls. Is your point that the various exit polling orgs come up with different results that vary by the percentage you give?

On the other hand, a bunch of statements indicate that much trust is put in the exit polls in Germany. To wit:

"11:27 p.m.) According to forecasts by broadcasters ARD:

CDU - 35.2 %

SPD - 34.2 %

But the results may be narrower because of the so-called "overhang mandate," a technicality in the German parliamentary process, which means late elections in Dresden on October 2 could be decisive (if the CDU doesn't win the district outright)."

It sounds like they put a great deal of credence in the polls. I think that's what most of the writers on the subject indicate. I've never seen the figures you've referred to that indicate that the polls are way off from the results in Germany. Is that what you intended to say? Or just that the polls have varying results?

Where is any indication of exit poll inaccuracy?

Here's a typical entry:

"7:50 p.m.) The Social Democrats, according to projections from ARD and ZDF, have defended their position as the strongest party in East German states. ZDF reports that they've earned 29.5 of easterners' votes, ARD reports 30.7 percent. There is a neck-and-neck race in the east between the CDU and the Left Party. ARD reported that the CDU led the Left Party by o.3 percent, with a total of 25.3 percent, while ZDF reports that both parties received 25.9 percent. The Green Party and the FDP received, according to ARD, 5.4 percent and 8 percent, respectively; ZDF reports 4.4 percent and 8.6 percent, for the Greens and the FDP, respectively."

Could you make your point a little clearer?

[OnTheOtherHand]

19. sure -- go down to the bottom

+++ First ZDF Exit Poll Released +++

(6 p.m.) Public broadcaster ZDF and Forschungsgruppe Wahlen have released their first exit poll:

CDU - 37%
SPD - 33%
FDP - 10.5%
Greens - 8 %
Left Party/PDS - 8%

That's a 4-point margin. The final official margin was about a point. A 3-point apparent error in margin is larger than the official margin in the 2004 U.S. election. It's a substantial error (and indeed an unusually large error for Germany).

What happens in Germany is the same thing as happens in the United States and other countries (not all other countries) -- the interview results are progressively supplemented with official returns. (I think the Germans get turnout figures first, but I don't know -- I have never even been to Germany.)

Or, to put it polemically, you are here watching the process of the German exit polls being twisted and distorted, in real time! But that's just how the projection models work.

[Febble]

20. Ours are supplemented too

but because we are in a single time zone, it is clear that what is changing is the projections, not the poll results.

The projections change as the votes come in. Because they come in haphazardly we get comically large changes in the projections at first, and the BBC presenter (Peter Snow) gets very excited as he demonstrates that "if this result in Putney were to be repeated across the country, then we'd see..." (pause for jazzy CGI graphic) "Tony Blair with 30 seats, and the Liberal Democrats with 170 - but it's all a bit of fun.

As the night wears on, the swingometer settles down, and by morning it just twitches a bit as the results limp in from Wales and Orkney.

Check the link and watch the video!

[papau]

13. An excellent answer to Mysterypollster is the Ernest Partridge post below

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=103&topic_id=215665&mesg_id=215665

| Ernest Partridge |

"Simplify! Simplify! Simplify!" -- H. D. Thoreau, Walden

Complication of the election integrity issue works to the advantage of the status quo; which is to say, the increasing use of paperless, unauditable Direct Recording Electronic (DRE) voting machines. More complications abound as critics of the status quo attempt to prove that past, and presumably future, elections were and will be fraudulent.

In fact, the controversy can be reduced to two simple questions http://www.crisispapers.org/essays6p/gulliberal.htm :

1. Can defenders of the status quo prove that the 2004 (and also the 2000 and 2002) elections were fair and accurate?

2. Can defenders of the status quo refute the critics?

The answer to the first question is simple and straightforward: they cannot, because the DREs (and also the central compiling computers) were designed to exclude proof. The software is secret, and thus closed to inspection and validation, and there is no independent record of the votes against which the totals can be verified. (Running the same computations again is not a "recount"). Moreover, computing experts have found, and demonstrated, numerous "holes" in the machines through which voting totals can be finagled, and reports of still more flaws continue to come in.

The response of the private election industry and the Republicans to demands of proof are (1) "trust us," (2) ad hominem attacks on the critics. ("Sore losers," "conspiracy theorists," "get over it!"). And finally (with the collaboration of the mainstream media) (3) no response. There are no substantive proofs of validity because, once again, the machines are designed to exclude them.

Regarding the second question, every now and then an attempt is made to refute the critics. The most recent of note was published last Friday in Salon.com, and was written by Farhad Manjoo http://www.salon.com/news/feature/2006/06/03/kennedy , who has made something of a career out of debunking the critics. Whenever an important critique of the electoral status quo is published, by John Conyers' committee, by Mark Crispin Miller, or most recently by Robert F. Kennedy, Jr. http://www.rollingstone.com/news/story/10432334/was_the_2004_election_stolen/1 , we can generally count on a rebuttal by Manjoo. Last week, he did not disappoint us.

Manjoo's latest was a pathetically weak piece of work which, due to its flaws, only serves to strengthen the case of its target, the RFK article. Or so I shall argue in the remainder of this essay.

At the outset, I should note that with all due respect to Robert Kennedy Jr., I must hope that he is wrong and the Manjoo is right. If so, then the Democrats have an excellent chance of regaining control of at least one house of Congress in the November election, and with it oversight of the Bush Administration. But if Kennedy is right about the ability of the Republicans to "fix" elections, then it may be impossible to budge the GOP from power, whatever might be the will of the voters. False optimism is the enemy of reform.

To begin, let's address a few minor points, which can be dispatched quite briefly.

Is Kennedy just rehashing old complaints?

Manjoo writes: "If you've spent time on Democratic Underground or have read Mark Crispin Miller's "Fooled Again," you're already familiar with everything Kennedy has to say." Because Miller is about to publish a rebuttal of this claim, I would prefer to let him reply in his own behalf. However, having read the books by Conyers, Fitrakis and Wasserman, and Mark Crispin Miller, I am willing to stipulate that most of what Kennedy presents is "old stuff", however this time with the added advantage of scrupulous documentation.

But so what? Those "old stories" are no less substantial for being "old." On the contrary, after a year and a half of examination and criticism, they still stand up. For this reason, the "old" possesses an advantage over the "new."

Were the Ohio (and other) anomalies nothing more than expected "screw-ups' and coincidences found in all elections?

If so, then these anomalies would be expected to work, approximately evenly, to the advantage and disadvantage of both sides. They did not. Almost all of the alleged "screw-ups" and "coincidences" worked to the advantage of Bush. Typical of defenders of the Ohio outcome, Manjoo also points out that individual anomalies were not sufficient to alter the outcome of the Ohio election. But he fails to address obvious rejoinder: the cumulative effect of several anomalies (by no means all of them) were quite enough.

Manjoo has nothing whatever to say about paperless Direct Electronic Recording (DRE) machines.

He presumably says nothing because he can say nothing that can advance his case for the validity of the 2004 election. So there is not a word in his article about secret source codes, lack of independent paper record, impossibility of auditing, or the GOP partisanship of the manufacturers and code-writers. Even if DREs in Ohio in 2004 (and elsewhere, and in 2002 and 2000) were 100% honest and accurate, there is no reason whatever to know this and an abundance of evidence (statistical, circumstantial and anecdotal) indicating that they were "fixed." As I noted at the outset, "Trust us," and ad hominem attacks on the critics are not evidence. And the silence of the media (not to mention the Democrats) about this compelling issue is deafening.

Now to some more substantial issues:

Is Kennedy guilty of telling half-truths and omitting embarrassing data?

Is Manjoo? Manjoo complains that Kennedy commits "numerous errors of interpretation and ... deliberate omission of key bits of data." But "the whole story" cannot be told in the allowed space. Even so, with his 206 endnotes, RFK makes a valiant attempt. More telling are Manjoo's omissions. With Manjoo's complaint of "deliberate omissions" in mind, I re-read Kennedy's essay. There I found at least twenty key elements of Kennedy's case for fraud that were totally ignored by Manjoo. Among them:
Half of the six million American voters abroad either did not receive their ballots, or received them too late. (Polls of these voters indicated that they were overwhelmingly pro-Kerry).

In New Mexico, decided by 5988 votes, malfunctioning machines failed to register the presidential vote on 20,000 ballots. (Kennedy fails to mention that these were in predominantly Democratic districts).

"A precinct in an evangelical church in Miami County recorded an impossibly high turnout of ninety-eight percent, while a polling place in inner-city Cleveland recorded an equally impossible turnout of only seven percent.

In Warren County, media monitoring of the official vote count was prevented by a totally bogus "terrorist warning."

In one precinct, exit polls indicated that "Kerry should have received sixty-seven percent of the vote... Yet the certified tally gave him only thirty-eight percent." The statistical odds? Almost one in three billion.

"A New York Times analysis before the election found that new registrations in traditional Democratic strongholds were up 250 percent, compared to only twenty-five percent in Republican-leaning counties."

"In heavily democratic Youngstown ... nearly 100 voters reported entering ‘Kerry' on the touch screen and watching ‘Bush' light up... Similar ‘vote hopping' from Kerry to Bush was reported by voters and election officials in other states."

"An electric machine at a fundamentalist church in the town of Gahanna recorded a total of 4,258 votes for Bush and 260 votes for Kerry. In that precinct, however, there were only 800 registered voters."
And twelve more. None of them mentioned by Manjoo. And once again, in almost all cases of voting "anomalies" throughout the country, the "errors" favored Bush.

If one were to concede most of Majoo's criticisms (the exit poll issue excepted) which, of course, I do not, even so the remaining unanswered elements of Kennedy's essay would add up to a compelling case for fraud.

On the misallocation of voting machines: Manjoo gives away his argument.

Manjoo appears unaware of the fact that through his attempt to explain away the misallocation of voting machines, he has supplied strong evidence of significant voting fraud.

To begin, here are some quotations by Manjoo which set up the trap into which he falls. (The emphases are mine. EP):


Kennedy says that "more than 174,000 voters" in Ohio did not cast a ballot due to long lines at the polls. He considers the GOP directly responsible for this failure. "The long lines were not only foreseeable -- they were actually created by GOP efforts," he says...

Kennedy's argument that Republicans deliberately engineered the long lines, on pretty shaky ground. To be sure, there is ample evidence that election officials throughout the state failed to respond to the surge in voter registration seen in the 2004 race. But it is far more accurate to see their actions as part of a larger picture of incompetence in the midst of massive changes in election procedures - especially changes in voting technology - than as part of a GOP plot...

Franklin County's allocation of voting machines can be seen as biased if you look at the number of black voters who were registered by Election Day, but decisions about how to allocate voting machines are made months before then. That's why Mebane also notes that "if the allocation of voting machines is compared to information about the size of the active electorate that was available to Franklin County election officials at the end of April, 2004, then the allocation of machines is not biased against voters who were active at that time in precincts having high proportions of African Americans."

The difference reflects the reality that in the last few months of election season, registration surged in Ohio. That Franklin County's voting-machine allocation was considered unbiased in the spring and biased in the fall arises from the fact that the county failed to respond to these electoral changes.

Note now, as Manjoo concedes, that there were "shortages" of voting machines in democratic precincts and "longages" of machines in republican districts. And why? Because in April the election officials did not anticipate the "registration surges." But Manjoo fails to take note of the obvious implication that the misallocation shows that the "surges" were primarily among the Democrats. Kennedy is explicit: "A New York Times analysis before the election found that new registrations in traditional Democratic strongholds were up 250 percent, compared to only twenty-five percent in Republican-leaning counties."

So now the trap is sprung: Where, Mr. Manjoo, did the Democratic vote "surge" resulting from the Democratic registration "surge" go? Is it just possible that those votes were either "lost" or, through some hidden hocus-pocus within the Diebold "black boxes" switched from Kerry to Bush? Clearly, they are not apparent in the final vote totals.

Robert Kennedy has a ready answer. I am curious as to how Manjoo and like-minded apologists would respond.

Of course, none of this "misallocation theory" accounts for the following, as described by Kennedy (and ignored by Manjoo):


At liberal Kenyon College, where students had registered in record numbers, local election officials provided only two voting machines to handle the anticipated surge of up to 1,300 voters. Meanwhile, fundamentalist students at nearby Mount Vernon Nazarene University had one machine for 100 voters and faced no lines at all

Surely the election officials knew in April that Kenyon College was strongly liberal, and Nazerene College was conservative. Why the misallocation?

Explanation please, Mr. Manjoo?

Manjoo's dismissal of statistical evidence is absurd.

Because a paraphrase of Manjoo's treatment of statistical proof may appear too outlandish to be believable, a direct quotation is in order.


As for Freeman's 660,000 to 1 statistic (of the improbability of random error), it is irrelevant... The statistic measures the probability that the errors in Pennsylvania, Ohio and Florida occurred due to chance or random error, and according to Freeman, that probability is very low. But nobody argues the errors happened by chance. Everyone in the exit poll debate agrees that there was a systematic cause for the errors in the poll. Freeman, Kennedy, et al., claim that the systematic cause was fraud, while Mitofsky and many in the polling community claim the cause was a problem with the poll. So Freeman's argument that it would take preposterous odds to produce a random sampling error is a straw-man assertion. (My emphases, EP)

Of course "nobody argues the errors happened by chance"! Freeman's whole point is that chance error is in effect impossible. But that doesn't make the statistic "irrelevant" or the argument "a straw man." On the contrary, the statistic, and the entailed "impossibility," is central to Freeman's argument.

Is Manjoo really so foolish as to believe this nonsense? I doubt it, for he is obviously a bright fellow. But he apparently hopes that his readers are fools. Well, not all of us are.

So we are left with this: Yes, random error is impossible, therefore, yes, there was "systematic cause" for errors. Lacking plausible explanation of error in exit poll design and execution, the compelling conclusion is fraud.

But is there a plausible explanation in design and execution of the exit polls? Manjoo says yes, and so to this consideration we now turn.

The desperate attempt to explain away the exit polls.

Logicians and Philosophers of Science describe "ad hoc hypotheses" as assertions that explain (better, "explain-away") phenomena, although these assertions are without any independent evidential warrant. Scholarly Choctaw aside, the concept can be clearly illustrated by examples.

Q: "Why haven't any of Saddam's weapons of mass destruction been found?" Ans.: "They were all shipped out and hidden in Syria." (That's the ad hoc hypothesis). Q: "Any evidence of this?" Ans: "Unfortunately, no." ("But just you wait!").

Another example: When, as a child, I asked how, if God created the world in 4004BC, there are dinosaur bones in the ground. I was told that "it is possible that Satan put them there to lead us astray." Any independent evidence of this? Of course not! (This is where "faith" comes to the rescue).

"Ad hoc-ery" is commonly revealed by such phrases as "it is possible that..." and "could have..." and "there is reason to believe..."

Consider now the theory that the exit poll "error" predicting the Kerry victory was the result of "the over-sampled Kerry voters."


"According to interviewers assigned to talk to voters as they left the polls appear to be slightly more inclined to seek out Kerry voters than Bush voters. Kerry voters were thus overrepresented in the poll by a small margin.

Kennedy is unimpressed with Mitofsky's explanation:


"Now, thanks to careful examination of Mitofsky's own data by Freeman and a team of eight researchers, we can say conclusively that the theory is dead wrong. In fact it was Democrats, not Republicans, who were more disinclined to answer pollsters' questions on Election Day. In Bush strongholds, Freeman and the other researchers found that fifty-six percent of voters completed the exit survey - compared to only fifty-three percent in Kerry strongholds. 'The data presented to support the claim not only fails to substantiate it,' observes Freeman, 'but actually contradicts it.'"

Now things begin to get dicey for the Manjoo/Mitofsky faction. (My emphases):


The numbers Kennedy cites fit the theory that Kerry voters were more likely to respond to pollsters than Bush voters. For instance, in the Bush strongholds - where the average completion rate was 56 percent - it's possible that only 53 percent of those who voted for Bush were willing to be polled, while people who voted for Kerry participated at a higher 59 percent rate. Meanwhile, in the Kerry strongholds, where Mitofsky found a 53 percent average completion rate, it's possible that Bush voters participated 50 percent of the time, while Kerry voters were willing to be interviewed 56 percent of the time. In this scenario, the averages work out to the same ones Kennedy cited: a 56 percent average response rate in Bush strongholds, and a 53 percent average response rate in Kerry strongholds. But in both Bush strongholds and Kerry strongholds, Kerry voters would have been responding at a higher rate, skewing the poll toward Kerry.

Yes, "it is possible that..." Independent evidence? None!


What's more, these numbers are not set in stone. That's because, as Mitofsky has pointed out, it's not possible to measure the actual completion rate by Kerry voters and by Bush voters. (When someone refuses to talk to a pollster, it's not possible to say whether he was a Bush voter or Kerry voter.) Mitofsky says that a hypothetical completion rate of 50 percent for Bush voters and 56 percent for Kerry voters would have led to the error we saw in the poll.

Independent evidence? None!

Next, from these unsupported ad hoc hypotheticals, Manjoo draws a substantive conclusion:


"In other words, Kerry voters were very slightly more likely to talk to pollsters than were Bush voters."

Obviously a non sequitur.


Ultimately, nothing in Kennedy's article, and nothing that the research he cites, refutes Mitofsky's theory that there was a true difference in the willingness of Kerry voters to participate in the poll compared to that of Bush voters.

But why should Kennedy be required to "refute Mitofsky's theory," when Mitofsky offers nothing to substantiate his "theory?" Manjoo concludes his "explanation" of the exit poll "error" with still more empty, hypothetical hand-waving:


Mitofsky noted a broad array of methodological errors that could have contributed to this difference in participation rate by Kerry and Bush voters. Such a difference would not have been a surprise; Democrats have historically been overrepresented in exit polls. There is no reason to think that the error in 2004 was anything substantively different.

"It's possible that..." "A hypothetical completion rate..." "Would have led..." "Could have contributed...." "There is no reason to think...." These are the plaintive cries of despair of the evidence starved. They are howling indicators of shameless ad hoc-ery aforethought.

So it comes to this circular result:

(1) Why the exit polling error?
(2) Because of the oversampling of the Kerry voters.
(3) And why should we believe that the Kerry voters were oversampled?
(4) Because it explains the exit poll error.

Pathetic!

How does one escape the circular argument. By supplying independent evidence of "the over-sampled Kerry voter." And as we have seen, there is none.

So if this ad hoc "explanation" of the polling error fails, and if no other explanations are brought forward and "random error" is judged impossible, what other explanation are we left with?

What else? The election was fraudulent.

Do articles like Kennedy's, and rebuttals like Manjoo's, illegitimately "frame" the controversy in favor of the status quo and against the critics?

Have the critics been "suckered" into "playing the game" according to their opponents ground rules? Unfortunately, it appears that they have. Why must it be the task of the critics, private citizens all, to "prove" that the past three elections were fraudulent? Why have these critics conceded this burden of proof? Do not the citizens have a right to secure and accurate elections? Shouldn't the burden of proof be on the government to provide verifiable procedures? Should it not suffice that the critics demonstrate that the procedures fall short? Even if Farhad Manjoo and others succeed in showing that Robert Kennedy and other critics fail to make a convincing case for fraud (and I submit that Manjoo has done no such thing), shouldn't it be enough that the critics have raised reasonable and unanswerable doubts, and the election officials and the defenders of the status quo can not supply the citizens convincing evidence and proof that the elections are honest and accurate? This much at least, the defenders have accomplished. Nothing else should be required. So why does the controversy continue?

Robert Kennedy Jr.'s argument that the 2004 election was stolen emerges essentially undiminished, and arguably strengthened by the weakness of Manjoo's "rebuttal." That's the logic of it.

But the practical effects are another matter. Will this controversy finally break into open public debate? And will it do so in time for the public will to overcome the formidable barrier of "black box" voting machines with their hidden secret codes and unlocked "back doors" open to real time manipulation and fraud?

Emerging from the Constitutional Convention, Benjamin Franklin was asked: "What do we have, Dr. Franklin?" He replied, "A republic if you can keep it."

Today it is uncertain whether we still have a republic, much less whether we can keep it.

Today, as in 1787, the answer to that question is up to We the People of the United States

-- EP

[charliecat]

15. Mystery Pollster is Not an Exit Poll Expert

Mystery Pollster Blumenthal does not understand the mathematics of exit poll analysis and has blindly followed the disproven analysis of Liddle and Lindeman.

This recently released short compendium by the National Election Data Archive rebuts the Mystery Pollster's allegeded "experts" Liddle and Lindeman:

http://electionarchive.org/ucvAnalysis/US/IncorrectElectionDataAnalysis-06.pdf

The Mystery Pollster is not a good source for correct information on 2004 exit poll analysis.

Why doesn't Mitofsky release the exit polling data that would support his case? Exit polling factors would not violate voter privacy, so there is no reason to withhold the data that Mitofsky claims proves his case.

[BeFree]

16. It's a mystery....

...why he listens to Liddle/Lindeman and doesn't check their work.

Really, his observations aren't worth squat because he bases his observations on evidence not worth a squat.

He'd probably tear his hair out reading the link you posted, or anything else that disproves the L/L theories. Oh well, just another in a long line of bush lickers.

[Febble]

17. A few points

Mystery Pollster is a professional pollster. He can therefore be fairly described as an expert on polls.

The paper you link to has three substantive mis-attributions to Liddle and Lindeman and ESI on the first page alone (this can can easily be confirmed by following the references provided), so cannot be considered an authoritative source of information.

E-M has released detailed data in the manner in which it is always released it to two public archives. Further disclosure of data in a form that would compromise the confidentiality of the respondents would violate the ethical guidelines of his own profesional organisation. An alternative is "blurred" data as was commissioned by ESI for Ohio, and I understand that further datasets may be in preparation. However preparing such datasets requires a considerable amount of data collection in itself in order to preserve the statistical properties of the data. Note also that the more "factors" that are released, the less effective the "blurring" will be at preventing identification of the precinct, and thus of the respondents. This is one of the key principles of the "blurring" technique.

Elizabeth Liddle