The unexplained exit poll discrepancy - Debunking the debunking?

[54anickel]

A new member, JMDEM posted it here (and a couple of other places):

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=102&topic_id=979770#981854

http://www.buzzflash.com/alerts/04/11/The_unexplained_exit_poll_discrepancy_v00k.pdf

edited for subject line.

[alittlelark]

1. If it has not been posted - it needs to be kicked

[54anickel]

2. Thanks for the kick, was just going to do that!

[alittlelark]

4. I PMed it to one of the principles.

[54anickel]

6. Thanks! Meanwhile it seems JMDEM can now start threads, so it's

popping up all over. I'll try and gather the "spread". ;-)

[NuttyFluffers]

3. thanks, good read. n/t

[regnaD kciN]

25. The conclusion of that article is just right...

...and should be adopted by those of us pushing for further examination.

My purpose in this paper, however, has not been to allege election theft, let alone explain it. Rather, I have tried toi demonstrate that exit poll data is fundamentally sound, that the deviations between exit poll predictions and vote tallies in the three critical battleground states could not have occurred strictly by chance or random error, and that no solid explanations have yet been provided to explain the discrepancy. In short, I have tried to justify the discrepancy as a legitimate issue that warrants public attention.

In other words, he's not a tinfoil-hatter, nor a conspiracy theorist. Nor are we. None of us is pushing a theory (claiming, for example, that the Warren County headquarters locked-off to reporters was really where the hired G.O.P. hackers were changing the vote results elsewhere in Ohio). Therefore, since we are not advancing a particular theory, we're under no burden to produce "proof" of any such theory. All we know is what can be shown through statistical analysis -- that the difference between exit polls, usually the models of accuracy, and the final results were way outside the realms of probability, and that we want to find a credible explanation for what happened. We don't know the answers yet -- but we are trying to find them.

[jsamuel]

26. Well stated... agree

[RuleofLaw]

40. Or like Rumsfeld would say:

Absencee of evidence of voter manipulation is not evidence of absence of voter manipulation.

[JaneEyrez]

5. Wow!

Very scholarly and methodical presentation of the argument, pro and con. Would be hard for anybody on either side of the fence to dispute the validity of this information. :bounce:

[dbonds]

7. Not hard

Not hard at all when they just make stuff up.

[54anickel]

8. ..

[alittlelark]

9. kick up

[Helga Scow Stern]

10. Good to finally have an academic telling the truth.

He will be heavily criticized, of course, especially by the "experts" who have already pontificated that Bush won it fair and square.

[Hobbes199]

11. Haven't even finished reading it...

And I love every word. Clear and concise. Sources and methods documented. The media bunking this crap should be ashamed.

[Bozos for Bush]

12. Kick hard for this incredible and factual analysis that shows we

WERE RIPPED OFF!

[ewulf]

13. Getting the raw data:

Professor Freeman writes that he "attempted to get the raw exit poll data from AP, Edison Media Research, Mitofsky International, and the New York Times, but as yet received no response." If he had the raw data, he might be able to make a much stronger case.

exit-poll.net gives the following list of National Election Pool subscribers:

Newsweek


The New York Times


Los Angeles Times


USA Today


The Wall Street Journal
The Washington Post

Chicago Tribune
National Public Radio

Atlanta Journal Constitution
Baltimore Sun

Boston Globe
Cleveland Plain Dealer

Dallas Morning News
KCRG-TV Cedar Rapids

Milwaukee Journal-Sentinel
Newsday

New York Post
NHK - Japan National TV

NY1-TV New York 1 News
Raleigh Observer

RAI - Italian National TV
Sacramento Bee

St. Louis Dispatch
St. Paul Pioneer Press

State News - Columbia SC
WCBS-TV New York

WCVB-TV Boston
WHDH -TV Boston

WNBC-TV New York
WNYW-TV New York

Detroit Free Press

and there are probably many more we don't know about. They were all privy to a lot more data than we, or Freeman, have. Will none of them give it over?

Are their no civically minded polling firm employees or responsible journalists out there who might give us, and the academic community, more information?

Is their something DUer's can do about this?

[Old and In the Way]

17. I wonder why the DNC did not subscribe to that service?

It amazes me that they wouldn't have wanted access to the raw data so they could understand the results directly.

[DELUSIONAL]

18. We need to ask a lot of questions about what the

DNC leadership was doing.

Like why did they completely ignore this BBV issues? Why didn't they hire their own exit pollers to target precincts with BBV and other e-voting machines as well as a random selection of precincts with paper ballots?

If -- IF these guys are pulling back and going quiet because it is part of a strategy -- then pardon us ignorant grass roots fools.

[ewulf]

20. what the hell could that strategy be?

Work really hard, spend half a billion dollars, get your guy Ohio, Pennsylvania, and maybe Florida, then let it be stolen because, in the end, Bush created a huge mess in this country, and he should be the one saddled with the political baggage of cleaning it up?

Maybe we should have skipped the money and hard work in the first place?

[DELUSIONAL]

21. My question is WHY was BBV apparently ignored by the DNC?

They were warned by the 2002 election loss in Georgia -- that BBV could have a negative impact on democratic candidates.

Sure they had poll watchers -- but was there any organized strategy?

Perhaps there was -- but we just don't know about it yet.

Also the lesson many of us take away from this election -- It does NOT matter how hard everyone works and how large the vote turn out it -- what matters is who counts the vote and WHO controls access to the computers. In other words the computer tech guys warned that computers could be hacked.

Congressman Wexler in Florida had a lawsuit to demand a paper trail -- but the GOP played out the clock.

The GOP fought at every turn for a paper trail.

No matter what we did -- it seems that the fix was in LONG before this election.

If there is nothing that the dems can do about the lack of a paper trail -- then the dems might as well give up. Democracy is DEAD in America.

[Helga Scow Stern]

24. I agree. Is it the money the machines bring as someone suggested,

for instance in NC, where Democrat Dan Bartlett is trying to get the state to spend 80 million on them.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x42458

I have written my senators Feinstein and Boxer. Feinstein says she voted for HAVA, which clearly is not adequate protection. I also wrote Bartlett, and said are you crazy?

We have to demand integrity from our party and from all elected officials and candidates, and it starts with the vote.

[ewulf]

19. Candidates can get the data a week after

according to exit-poll.net. I wonder, however, if it would include all the raw numbers that were given out on election night, or would just have the tabulated-vote adjusted numbers that you can find on cnn.com, etc. (but maybe with greater detail)?

Maybe we could ask Nader, Cobb, or Badnarik for help with this also?

If we could get their raw, precinct level data, that would be awesome: we could compare their numbers with the actual numbers tabulated for those precincts, and we could essentially prove fraud (if the repugs or whomever weren't smart enough to only target precints that weren't being polled: but this would still help us show discrepencies in the untargeted precincts).

[jsamuel]

14. GREAT JOB! Emailed that out to my family and friends!

Message:

I read the attached paper and it is right. Using the exit poll data released so far, the way this election turned out in Ohio, Florida, and Pennsylvania had a probability of one in 250,000,000.



This means that one of two things happened that spoiled this result.



The election results are wrong.
The exit polls are wrong.


It points out that exit polls have been used to give reason to investigate voter fraud in many countries. It shows that there is currently no explanation for why the exit polls were wrong, only theories that have no evidence. The theories of voter fraud also lack evidence, however many irregularities have become apparent.



The main point is that we have no reason to think that the exit polls were wrong and that Kerry should have won Ohio, New Mexico, and Nevada with Florida and Iowa being very close. It is up to us to investigate which one is wrong, the exit polls or the election’s result, neither having been debunked thus far.

[melwoods]

15. kick

:)

[Old and In the Way]

16. Very interesting.

I recall on election evening that actor-Democrat Ron (?) Silver on MSNBC who tried to tell the audience, in most assertive terms, that one could not believe the exit polling data because, of course, more Bush voters were "afraid" to let people know they voted for the President.

I remember saying to myself, what pure and utter nonsense. What makes this person, an actor for crissakes, an authority on what the voter was thinking? And why wouldn't his "logic" be just as applicable to people who would say Bush, but actually vote Kerry?
Obviously, they (the MSM) had to address the exit v. actual discrepancies....so they chose this guy, probably based on his acting skills, to make the case that exit polls could not be believed.

He was, and is, full of shit.

[high density]

22. Ron Silver was probably given the GOP talking points before showing up

I'm really curious as to what would be going on right now if the exit polls had shown Bush winning but Kerry won the tallied vote.

[jsamuel]

23. Why the heck was this B movie actor on the floor durring the RNC??????????

I mean, is that the best they could do... Something is strange about that...

[ewulf]

27. The Freeman article really compares nicely

with the garbled crap coming out of MIT-CalTech:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=43258&mesg_id=43258

[thedutch]

28. agreed

This paper is the anti-anti-missile-missile-missile of bad math. The people of this glorious blue county are going to wake up monday with a thousand of these under their windshield wipers.

and a cookie to anyone who can tell me what cartoon is referenced above.

[prairierose]

29. JMDEM, thanks for posting this.

I am mathematically challenged but managed to pass all my ed stats classes :shrug:

This paper is very well written and I thought really easy to understand. He lays out the case & backs it up with data as well as explaining how data is aquired & used.

I would thnk that MSM might just pay some attention this.

[Robert Oak]

30. Shouldn't the Kerry campaign have all of the raw exit poll numbers?

I believe the campaigns were getting updated, county per county exit poll numbers, the question is did they get the raw results
that are now in a "lock box"?

I hope someone contacts the kerry attorneys and gets those numbers
to some statistical experts.

[ewulf]

34. I don't think they do have the NEP numbers,

Certainly, the DNC did have a substantial ground effort, but as a small part of that effort, I saw it mostly focused on GOTV, on counting reliable Democratic voters rather than actually surveying everyone on their votes in a systematic way. Maybe they did that in some precincts, but I doubt it: the focus was on winning, and (perhaps unwisely?) not on setting up for a challenge.

As I stated previously, I don't think the NEP data is released to candidates untill a week out (so they might have it now) and I'm not sure how raw or detailed it is.

[Dem2theMax]

31. How big is that pdf? I cannot get it to open and it's pissing me off!

aaaarrrrgggggg. Stupid computer is acting up. Who knows computers and lives in San Diego area? Double aaaarrrrgggggg.

[jsamuel]

32. Here is another way

Right click on the link, go to Save As, then save it. Navigate to that spot and open it. Sometimes browsers have problems with Adobe plugins.

[lamp_shade]

33. kick..

[Sideways]

35. KICK

[althecat]

36. Saw this thread getting long and didn't want to bury this...

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x44423

[Mr Creosote]

37. Debunking the debunking of the debunking

"We have every reason to believe that exit polls are accurate survey instruments".
If anyone wants to allege the election was stolen on the basis of this assertion, then they're going to get shot down quickly. So exit polls are OK in Mexico, Germany and Georgia. Like the US, the UK is a mature industrialised democracy. In 1987 the exit polls predicted a hung parliament. In fact it was a fairly easy Tory win. In 1992 the polls predicted a narrow Labour win, in fact it was a narrow Tory win. Very similar error - under-estimation of the right-wing vote.

What I'd like to know is this:
1. Are the same differentials seen in non-battleground states?
2. Is there any relationship between sample size and differential?
3. Was the same sampling methodology used everywhere?
4. Did they poll for the other races going on at the same time? Does the same "error" show up?

That last one is the key for me: if the exit polls were "right" for other elections, and only "wrong" for the Presidential election, then that suggests to me that the problem does lie with the counting of votes, rather than the polling.

[Old and In the Way]

41. good point.

I'd like to see the exit polling data broken down by counties.....is there are correlation between accuracy when the exit polls are controlled for paper vs. paperless voting systems?

[ewulf]

42. The exit poll shift and sampling size

do seem to be correlated: the bigger the sampling size, the bigger the shift, on average. This is not what you would expect (nor that the shift would almost always be toward Bush).

[Sideways]

38. Kick

[MelissaB]

39. Kick!

for the morning crowd.

[sarahlee]

43. Minor Update to the Freeman Paper

Dr. Steve Freeman sent me a minor update this morning. He is working on another that will take a bit longer and says those changes will be more significant.

Latest version is: The unexplained exit poll discrepancy v00l.pdf
(replaces v00h and v00k)

Please keep checking for updates at:
http://www.dakotatechnics.com/downloads/

I will continue to upload updates as soon as he sends them to me and my hosting service puts up with the demands.

This is all over this board and the net now. Please help me make sure others referring to this document by Dr. Freeman know are aware of this update.

Dr Freeman told me on the phone this am that he is getting a lot of email from this (including some not so nice ones), so please do not email him unless you have something substantial to contribute to his analysis.

I originally sent the paper to BuzzFlash and have notified them of the update.

Note the server at this site is a bit slow, so if you get an error, just try back in a bit. Getting a lot of traffic.

I also find some of the work Mike Lewis is sending me re Ohio is very interesting. It is also downloadable from the above page.

Thank You!

[mak3cats]

45. More significant changes?

Do you know if these changes will strengthen or weaken (or neither) the "debunking of the debunking"?

[sarahlee]

46. Update

What he told me he is writing is more "documentation" (my translation), for the work he has done - work a document of this type would normally have, but which generally takes a good bit of time to actually write.

He was trying to get his work out there to get feedback from others in the field. The document clearly states it is a draft and that he seeks input.

This has been occurring.

Obviously, I am simply a "distributor/messenger" and not trained in math or statistics so the terminology he was using was not familiar to me. However, he gave no indication that his conclusions had changed.

I believe he is just adding more of the details of his work, which other statisticians require before judging the veracity of his conclusions.

[ewulf]

44. 3rd part votes are interestingly

consistent in their swing between the exit polls and tallied votes; in FL, OH, and PA, their was always an increase in the number of votes that went to neither Kerry or Bush: .3, .5 and .1 respectively, according the Prof. Freeman's data. If one considers a statistical model of vote error, this would make sense; as the number of votes for the 3rd party candidates is small compared with those for the main candidates, if one were to randomly take votes from any candidate, and reassign them at random, 3rd parties would benefit. One could say that the state with more third part votes has higher entropy.