"How to Read Exit Polls: a Primer"

[Mr_Spock]

From DKos - "How to read exit polls: a primer"

http://www.dailykos.com/storyonly/2006/11/4/135126/905

What caught my attention was the statement that exit polls have shown a consistent historical bias for Democrats & that without the massaging based on actual vote data, they are not to be taken as having a normal "margin of error" due to non-sampling error. See excerpt below & full article above. It is stated that exit polls can never have all potential sources of bias taken out and thus must not be used as a check on the integrity of an election.

Discuss...

...

"Exit polls are uncannily accurate"

The precinct level data has shown a consistent Democratic bias over the last 5 presidential elections, and the causes of the bias have been well-researched. In 1992, the discrepancy was almost as large as in 2004. The reputation for "uncanny accuracy" probably, ironically, derives from pollsters' extreme caution about calling states unless they are very sure they are right - and they make sure they are sure by incorporating vote-returns into the estimates in all but the most slam-dunk of races. In the UK, where, for all our faults, we conduct pretty transparent elections, the exit polls are regarded as a bit of a joke (a cruel joke in 1992). Peter Snow, the BBC poll presenter on election night has as his catch phrase "it's all a bit of fun".

"Exit polls are used to monitor election integrity around the world"

Not as far as I know. In Ukraine, there was direct evidence of blatant fraud (acid in ballot boxes; candidate poisoned with disfiguring, potentially lethal poison). Sure, fraud will tend to play havoc with exit polls, but given that exit polls can play havoc with themselves, they can never be a primary instrument for monitoring election integrity. Indeed, here are some cautions:

http://www.cartercenter.org /...

http://www.cartercenter.org /...

http://www.cartercenter.org /...

http://jimmycarter.org /...

Take home message:

The early exit polls will give you a reasonable idea of who is winning on election night, but there is no point in expecting the results to be within any calculated "Margin of Error", as MoE calculations assume random sampling error only, and do not reflect non-sampling error, which polls inevitably contain. Therefore, even if the final results diverge "significantly" from the early poll, it won't necessarily mean there is fraud. Exit polls have too many potential sources of bias for bias ever to be ruled out. If you want to find fraud, don't try and find it in the NEP exit polls. Independent polls designed for the purpose may tell you more, but they are unlikely to be much more immune from bias than the NEP exit polls, and may be more vulnerable.

...

[Glenda]

1. kick

to read later

:kick:

[Mnemosyne]

2. k and 5. n/t

[The Count]

3. Exit polls are always the check and balance for fraud.

They came different in Romania 2004 as well. A revote provide the result the exit polls showed (the other candidate won).
In 2002, networks refused to show the exit polls (there were "discrepancies"
In 2004, the comparison between exit polls/results varied as to the type of machines

[Mr_Spock]

5. The plots look nice

...unfortunately they do not validate the accuracy of the underlying data as described in the article.

In NH, the optical scanning machines appear to be pretty reliable and have a paper trail. The state turned (flipped) completely blue this year for the first time since 1911. It was our best year in nearly 100 years. I can't speak to the other states, but the article points out that exit polling has been proven to be biased toward Democrats - should we take it seriously if even the people who provide it don't trust it's accuracy until they take into account actual voter turnout?

[Sancho]

55. You are not describing this correctly...

What you mean to say is that there is evidence that the pollsters don't do a good enough job of stratified sampling to get a result that reflects the population. Not "Proof of bias". Lawyers prove things, scientists find evidence of various theories and propositions.

I still don't see why people pay someone to do an inaccurate poll and then "adjust" it so that it concludes what you were paid to predict....

This is not a scientific process at this point...it's political, money making, or something else.

If there is gender bias, response bias, etc., good pollsters with lots of experience would deal with it to get accurate samples. No more excuses from pollsters, please.

[malaise]

22. 100% correct

Prof Freeman's paper was all the proof I needed that the 2004 election was stolen. The Diebold exposure showed me how they committed that fraud.

[BeFree]

4. Someone can write all the circles they want

But the truth is that years ago news stations were able to correctly call winners of elections hours before polls even closed. That tells me that the early exit polls are deadly accurate.

But since politics elbowed it's way into the pollsters pockets, the final reports - well after the polls have closed - are not to be trusted.

What we are faced with today is a massive effort to alter votes with the electronic devices and the one sure way to catch that alteration is through exit-polls.

Whoever it was that wrote your Kos piece is obviously working for the crooks who want to steal your vote by denigrating the one tool that can stop the theft cold.

[mom cat]

6. Exit polls do not "favor" Democrats ... they expose the margin of fraud!

[Mr_Spock]

8. The author states that the Democratic bias in exit polls has been verified

What evidence do we have to counter that statement?

[johnaries]

32. What evidence are they using to support the bias? Final Results?

Have they proven the results were NOT fraudulent?

Without proving that, it's only an opinion that cannot be verified.

[Mr_Spock]

36. Isn't proving a negative impossible?

That's not really a valid argument in my opinion.

Negative proof...

This is a fallacy whereby the normal burden of proof is reversed. It is asserted that a hypothesis must be true, solely on the grounds that it has not been proven false.

This is fallacious reasoning because formally, the burden of proof should be on the proposed idea, not the challenger of the idea. This is a crucial point of the Scientific method, that before a claim is thought to be true, it must be proven. All claims must be confirmed by observation. If the claim can not be confirmed this way, the belief must not be asserted. Not-knowing is default.

[followthemoney]

46. You are biased in favor of official vote results.

You have no access to exit poll data, electronic totals, or the software that generated them.

On what basis do you have faith in the political operatives proven to speak with indifference to truth?

Your "normal burden of truth" is based on your perspective of challenged and challenger. You assume that a statement that supports political power claimed by any means they deem necessary, even possibly theft, is valid.
Let the owners of the data challenge the exit poll raw data by full disclosure.

This is circular reasoning at best. You hold that the validity of an election must assumed even when evidence that should be available in an open society is withheld from the public by those that expect the public to have faith in it.

An attempt at objectivity should leave you equally skeptical about either data set. Your statement could be freed of its bias, though relatively boring, if written as follows:

"Both the exit polls and the vote count are subject to error. I cannot assume that either the exit polls or the official vote counts are more accurate."

This makes it plain that nothing can be assumed about either data set and that further investigation is required in order to make a meaningful statement about either of them.

That is to say you know nothing about them despite your status as a scientist. A scientific investigation would be interesting, although more work than a scientific opinion.

Unfortunately, an investigation is not forthcoming because the public business is not held to be any business of the public by the owners of the data.

[Mr_Spock]

47. Putting words in my mouth

I have simply asked for reasoning why we should believe the negative assertion that the previous poster made. The default position is to believe the results of the election. If an alternative theory is to be presented, then simply offer it up with evidence. I have previously stated almost exactly what you have posted there WRT both data sets having the possibility of error. The issue is, one data set deals with "hard" data - that is all of the votes counted, the other deals with a sample of the people who voted that is subject to the foibles of human social interaction phenomenon. All I have brought forth here is evidence that exit polls in themselves cannot be used to prove fraud as they are not unbiased enough (unless, of course, there was a huge & undeniable difference between the two).

[followthemoney]

50. You make the claim that the votes counted are "hard data"

but that claim is what an investigation is required to determine.

In your bias you will not consider that the official vote count is also "subject to the foibles of human social interaction phenomenon."

If you assume the votes counted are "hard data", you can't use that assumption to prove itself or to shield that assumption from scrutiny. That is a tautology.

That is to claim a man is guilty of a crime because he was arrested.Then for what other reason would he have been arrested, you may ask? By that logic he must be guilty. That is to judge before an investigation. To prejudge in the court of public opinion. Your prejudice is in favor of the so-called "hard data."

This type of logic is faulty and all too pervasive.

Furthermore, I do not accept the premise that the owners of that data have a higher interest in it than the public, whose treasure and blood may have been squandered as a consequence of the data's misrepresentation. The obstruction of an investigation and your indifference to its public consequences needs to be explained.

Can you explain why the exit poll discrepancies arouse no suspicion of foul play in you? Why are you so comfortable in your assumption that all is well?

[Mr_Spock]

52. Putting words in my mouth again...

I have not claimed that there cannot be error in the "hard data", all I have done is agree with the DKos article that exit polls are not the be all, end all in proving that there is an issue with the "hard vote", they are a tool, but they cannot be used exclusively to prove voter fraud as they have their own bias that cannot be factored out as the bias is not completely quantifiable.

The "guilty of a crime" argument is a straw man argument and completely irrelevant.

Do not call me indifferent - I am anything but indifferent - putting words in my mouth again. I think the data should be examined by independent statisticians to verify if the techniques used to factor out sample error are valid.

The exit poll discrepancies DO arouse suspicion in me - once again you are putting words in my mouth - I am interested in how we can get a definitive determination if the bias in in the exit polls or if there is an error in the vote tally. Stating definitively that voter fraud occurred without an independent investigation is not valid IMHO. Why are you so interested in castigating someone who simply brings forth evidence that exit polls in themselves are not sufficient to prove voter fraud? Why would you assume that I do not want to find, or believe in voter fraud? That is disappointing to me - my last word was "discuss...", not "castigate...".

[BeFree]

53. Well, here is one idea about using data

From one statistician on DU:
"None of which is to argue that no fraud happened in 2004 (or indeed in 2006) - I strongly suspect fraud in both elections. However, it is a caution against gung ho use of statistics, and also against giving to much credibility to any one analyst, regardless of how starry their credentials.

There may be good cases to make in 2006 - voter suppression in Virginia for a start, and undervotes in Florida. Let's not dilute good cases with crappy inferences from dodgy data."

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=457943&mesg_id=457977

The data infering corruption of vote counts by use of computers is clear.

But your OP's position - that of using data from one analyst - has now been refuted by an *expert*. Don't you think?

[Mr_Spock]

54. Not at all - I have read quite a few posts from that same O/P here on DU

I did the same research and will not cherry pick a particular statement to make my point - the bottom line is that the exit poll data is not in itself proof of voter fraud, though with other supporting evidence, it surely is excellent evidence of it. What I am saying is - in conjunction with other evidence, exit poll data is an excellent tool for determining the validity of election data - oh, and to get an early idea of who won! We can narrow down where to look further by seeing the difference between "official" voter tallies & exit polls, then further investigation can commence. Unfortunately, the exit poll companies do not seem to want to stand by their data when it comes to calling for investigations - they are "suspicious" of their own data!

[BeFree]

37. Good point!

The OP's theory is based upon the results being correct. No study was undertaken to check that assumption, as far as I can tell. It is just blindly accepted. And that blind acceptance flies in the face of all the evidence of machine failures, misprogramming and malfeasance.

Therefore the theory is corrupted from the beginning, eh?

[mom cat]

45. Precicely! The "evidence" is the tallys that do not include the possibility

of fraud.

[Mr_Spock]

7. They can call races that are not close, but...

can they call close races now any worse than previously? I would think we would have even better models today than in the past, but even the exit pollsters have so little faith in their data that they feel compelled to verify their data against actual turnout patterns before they "officially" publish their "final" poll.

I am not convinced by your statements that the person who wrote the article is not a reliable source. The conclusions of the article strike me as making sense.

[BeFree]

11. The article is bs

The article's premise is that the exit-polls can't stand on their own data. That's bs. The data gathering is time tested and respected around the world. In 2004 it cost $10 million to produce and anyone who thinks the data was not worth every penny thinks ABC and the rest were fools for spending that kind of money.

The other part of that trash was the idea that the final results have to be blended into the final exit-polls to be accurate. Well, why spend all that money and all that effort to just come up with the same numbers that come out by 11pm? If the early numbers are not accurate then why go through the exercise at all? It makes no sense to argue, as that foolish bushco ass-kisser who wrote your article has, that the exit-polls can't stand on their own two feet.

We are up against a massive undertaking to steal our votes and your writers position seems to be one of support for the theft.

[Mr_Spock]

13. The person who wrote the article is a bushco ass kisser?

I was wondering if this charge would be leveled - do you have any more information on the author?

I assume that the Democrats have some extremely good statisticians on their payroll, I can't imagine that they would not be aware of statistically significant deviations in exit polls. I can't imagine that they are that accurate - I have yet to see any data that convinces me that exit polls are reliable enough to state with certainty where voter fraud has occurred.

[truedelphi]

9. You can believe TIA and Autorank

or you can believe Febble but you cannot believe them both

[autorank]

43. It's Occam's Razor...but not concealed.

We had a great victory and I'm in a celebratory, yet relaxed state.

People can do whatever hey want to prop up the Bush "victory" in 2004. Heck, delphi (if I may
be familiar;), they'll probably go back and argue that Bush won the popular vote in 2000. That's
not as wacky as it sounds because it would solve some of the major problems with the 2004 National
Exit Poll. This is getting too deep. Gore won, Kerry won, we won last Tuesday ... and in Virginia
we won big no matter what the count. George Allen is seeking opportunities elsewhere and sanity
has once again returned to the Commonwealth. CHEERS!!!

[truedelphi]

51. Duh! I keep forgetting that we shoild be able to relax for a

moment.

Still and all, when you ask someone who worked for Mitofsky about the Nov 2 2004 exit polls and say how could it be that the whole republic was led to believe that Kerry would win,what was that about? only to have that person say "Well even as we were posting those exit polls we were having problems with them..."

I mean just simple journalistic ethics would demand that in that case (if indeed you are having problems with your own numbers) that you would not run those numbers OR you would run them with a disclaimer

[daleo]

10. I don't get why exit polls are supposed to slew Democratic

Do Republican voters skulk away shamefully, unwilling to admit their sin to exit pollsters? I doubt it. In my experience, the Conservative mindset is not one that hides its light under a bushel.

[Mr_Spock]

12. It's apparenlty a proven statistical anomaly

I have no way of disproving the authors statement.

[daleo]

14. Well, I think it requires a theoretical justification (and empirical data)

Probability theory says that statistical anomalies should break one way as often as the other. For a consistent bias to apply, there has to be an underlying reason.

In the past, polls and electoral votes have sometimes varied for financial or technological reasons (e.g. Dewey polled higher in 1948 before the election because Dewey voters were richer and more likely to have telephones). I just don't see why people being polled on their way from voting would be more reticent to talk to a pollster because they voted Republican.

Nothing personal, your post generated discussion, which is always valuable.

[Mr_Spock]

15. The article states that the last 5 presidential elections had Dem bias

...and that the Democratic bias in the exit polls in 1992 (Clinton won) was as bad as the bias in 2004 (Kerry lost).

[daleo]

18. Maybe election rigging isn't that new

Maybe Clinton actually won by a larger margin than we think.

I wonder how things were before 1992. Sometimes analysts pick their cutoff dates very carefully when they are making an argument.

[autorank]

44. Here's some data...

From TruthIsAll's OP on Gretest: http://tinyurl.com/yfnozv
The image at the bottom says it all. The Generic national poll was heavily in favor of Democrats. The generic preference gets translated into a lesser share due to local politics or huge margins in some areas, cities for example, which up the percentage. However, this is no reason to think that the poll is pro Democrat. Given the "spoiled" ballot subtraction which has gone on forever and the mroe recent voter suppression tactics, both of which SUBTRACT from the Democrats, its hard to conceive this poll as favoring us.
------------------------------------------------------------------

THE DEMOCRATS DID MUCH BETTER THAN THE FINAL EXIT POLL
INDICATES!

They always do.

Remember Kerry in 2004?

He won the 13047 respondent National Exit Poll timeline
at 12:22am by 51-47%.

But the Final Exit Poll (13660 respondents at 2pm) said Bush
won by 51-48%.

Fast forward to 2006.

The 7pm Montana exit poll said Tester won by 53-46.
The Final Exit Poll: 50-47.5
The recorded vote: 49-48

The 7pm Virginia exit poll said Webb won by 53-46.
The Final Exit Poll: 50.1-49.9
The recorded vote: 50-49.

What do 2004 and 2006 also have in common?
The Final Exit Poll was matched to the recorded vote.
It always is. That's SOP.
The Democratic vote was 3% too low.

Bottom line:

If the recorded vote was bogus and the election was rigged
through uncounted ballots and switched votes, you would
never know it from the Final Exit Poll.

But if you view the earlier exit poll timeline, you would
be alerted to fraud. And if you analyze the demographics,
you would confirm the theft.

Let's start our analysis with the 116 GENERIC PRE-ELECTION
POLL TREND LINE. The Democratic vote share has been a
steadily increasing trend line.

On Nov. 7, the Dems held a 14.6% lead over the GOP.

Here's graphic proof:

[Mr_Spock]

49. I don't think using a "generic" poll is useful

How would one correlate generic poll data to actual results - I just can't even conceive of it. I'll grant you that those two states exit polls were off, though other less controversial state results had error in the 2-3% range as well. It is a given that a large deviation in exit polls may indicate voter fraud. The default position is that the exit polls were in error and I am not sure how we can do anything more than point out the deviations & press for an explanation.

[Starlight]

16. I agree. Without some sort of rationale for their position, how can anyone accept it as valid?

[PATRICK]

17. This is more smoke

I would like to see data how and why and even when dem biases are put into polls when not only pressure has been put upon those companies to correct for the GOP result and the image of GOP resurgence but that the companies have been correcting for this for several campaign cycles. If this is a roundabout way to show that the Dem surge is catching the GOP propaganda flatfooted everytime despite vote suppression, etc, let them say it. And what is this excuse supposed to be for? It looks like it was prepared to explain away Dem defeats and avoid scrutiny of the "useless" polls that are practically the treasure of MSM media and secret tools of both parties internally.

If such things are problems we need new exit pollsters separate from media and political pressure and less, of course, of the obvious fraud problems that cause alarming discrepancies their salesmen quail to see.

Even this critique seems mystical mumbo jumbo carnival sales talk more like those justifying- then disclaiming- online popular polling. Maybe it is too important a tool to be left the whims of the current marketplace pollsters. If you are set up for expectations in preordained sampling that is building in error as opposed to simply recording actual turnout polling. At least, they'd rather fall on that sword than anger politicians and corporations with signs of criminal fraud.

[Mr_Spock]

21. Good questions

If the bias is known, why can't they adjust for it? The author was clearly trying to educate people before the election, though I am not prepared to ascribe any motivations to this person just yet.

I am not ready to use phrases like mystical mumbo jumbo carnival sales talk - this seems to be written in the same serious tone that our dear poll watchers use. Since these organizations must necessarily be non-partisan, it is unlikely that politicos from either party will be getting the inside scoop on their sampling techniques & raw data anytime soon. If we knew how the sausage was made, we'd likely not buy it!!

[bling bling]

20. Well this is what I heard.

Supposedly the majority of the poll takers (people asking the questions) in 2004 were female college grad students. The networks blamed this for the skewed results and in this years midterms they "corrected" that issue. Not sure how.

That's what I heard on CNN or somewhere the night of the election.

[ProgressiveEconomist]

27. mhatrw has a DU discussion on what I think you're talking about

See http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=364&topic_id=2654934&mesg_id=2654934 and the US News and World Report link therein.

Before he died, Mitofsky, who devised and implemented the first exit poll in 1968 and was THE guru on exit polling thereafter, apparently had a theory why Rs might be less willing to be interviewed than Ds.

Exit polling companies have to deploy personnel to all the thousands of precincts they pre-select for their samples. Generally, they try to spend as little as possible on staffing. They generally only hire just one person per precinct, so there are times when no one is "minding the store". Whenever more than 10 or so people exit while the interviewer is busy with one voter at a popular voting time, more people slip through the survey net.

Mitofsky apparently measured differential (between Rs and Ds) nonresponse by interviewer characteristics and found female graduate students did the worst with Republicans. This may reflect Repubublican voters' knowledge of the gender gap in party affiliation, their tendencies toward misogyny, tendencies for contempt of young people, and tendencies toward the self-important feeling that "my time is too valuable". Thus Republicans may tend to especially shun potentially stressful and hostile interviews with young female graduate students.

[bling bling]

38. Yes, I think that's what the media was referring to.

They had a whole new "system" this year for exit poll data where all the data was kept super-classified until the polls closed. Come to think of it, I don't remember seeing the exit poll data this year to see how closely it matched with the actual results. Not that I trust either one completely anymore.

[daleo]

30. And conservatives won't talk to attractive young women?

I don't know, it still doesn't compute for me.

[bling bling]

40. Yeah. I don't know if I buy it either.

It's one of many possible confounding variables. I'm not yet ready to dismiss my suspicions about the integrity of the election process in the last two presidential elections. If the exit polls have been so unreliable in other elections, as some have stated, why did the media just now change their research methods?

[ProgressiveEconomist]

23. Republicans may be less likely to agree to be interviewed than

Democrats ("differential nonresponse bias"), and failure of exiting voters to complete interviews ("nonresponse bias") has grown dramatically over time.

Thus proportions of Republican voters in exit polls may be lower than proportions of Republican voters in final precinct vote tallies.

There is a similar kind of potential bias in pre-election telephone surveys, but it runs the other way: Since pollsters must exclude cell-phone exchanges from their "random digit dialing", "cell-phone-only" households, which tend to be younger and lean more Democratic than landline households, are excluded from pre-election polls. Virtually all pre-election telephone surveys except those done by Gallup try to correct for this bias by giving Democratic-party-affiliated responses higher weight than Republican-party-affiliated responses.

But such reweighting would be unreliable in an exit poll to CHECK election results; party weights used for exit polls generally come from precinct tallies. Such weights may help pollsters explain who voted Republican and who voted Democratic, but can't be used in auditing elections.

According to http://www.mysterypollster.com/main/2004/12/what_about_thos.html , the "nonresponse rate" in 1992 national exit polls was 40 percentage points. Eight years later, in Dubya's initial "victory", it was 23% higher, at 49 percentage points. Higher nonresponse bias over time may also mean higher differential nonresponse bias over time.

See also Febble's DU posts today on the "nonresponse" and "differential nonresponse" issues, at http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=364&topic_id=2654934&mesg_id=2654934 .

Also, increasing percentages of voters are using absentee ballots or voting before Election Day in the 15 states that allow early voting. Differential (between Rs and Ds) trends in these modes of casting ballots could be compounding any differential Election Day nonresponse bias against inclusion of Republican voters in exit polls.

[Mr_Spock]

25. Thanks for contributing to the discussion

I can see how attempting to re-weigh exit polls would be fraught with potential problems and why the bias continues to show up in the exit polls. Once actual precinct voting percentages are weighed in to correct the poll, the ability to prove fraud is eliminated, but the data is unreliable without this correction - a catch 22.

I hadn't even thought about the increasing use of absentee ballots - no wonder polling is half science & half art - it seems obvious to me why they do not share raw data with the public. As I said in another post - if we saw how the sausage was made, we'd likely never buy it!

[BeFree]

26. Mystery pollster is biased

But looking at the OP we see that the *bias* supposedly existed in 1992 so all your new-age theories are just that - Theories.

It's crazy to think that these days, with all the computing power available to the pollsters, that they can't figure out the MOE. Just plain crazy.

They know how to get the right numbers, and they do. They don't need to blend the official results with the $10 million early estimates, to get correct and accurate reads. In fact the official results are most likely incorrect.

It is crazy to think that somehow the pollsters have it all wrong and armchair quarterbacks like Mystery Pollster are anywhere near correct.

It is fact that the theft of elections is such an organized crime that the criminals influence would lead to articles and innuendo like this OP's.

[Mr_Spock]

28. "the official results are most likely incorrect"

I've never understood this logic. Don't you have to first assume that there is fraud occurring, then somehow attempt to prove that a sampling of data that is fraught with potential errors is more accurate than the actual vote count? I don't trust the crappy ass machines that have no paper trail because they are just plain shitty designs that require "calibration" and they crash and are hackable to boot! :eyes: But knowing how smart Dems are, wouldn't you assume that Dems are more or just as likely to know how to hack the vote?

[BeFree]

29. These days, for sure

But exit-polls - the early numbers - have historically been dead on. And with the added computer power the pollsters now have, they are most likely even more correct than days gone by.

And Dems don't have control over the machines. The Publicans own the companies.

All this is very important right now because it is being estimated that there was a theft of 5% in a number of races last Tuesday.

[ProgressiveEconomist]

31. Election theft well may have been responsible for Dubya's "victories"

in 2000 and 2004. But IMO few people are going to be convinced by crude analyses that don't give proper credence to alternative, less conspiratorial explanations first. IOW, those who allege hacking need to step up their game, and get some real experts to consider election fraud along with other explanations. Even Stephen Freeman's efforts have been questioned fairly effectively IMO, for their step away from his area of expertise: See http://www.mysterypollster.com/main/2004/11/the_freeman_pap.html

I guess this is what you meant when you said "Mystery Pollster (Mark Blumenthal) is biased."

[BeFree]

34. Well, the question here is

How much credence can we give to exit-polls? I think anybody reading this exchange will see that the OP is a crude theory of why the exit-polls and the offical results are often contradictory.

The real world explanation is that the introduction of the computerized voting systems has created the fault. Instead what we are presented with is second-hand, probably biased information coming up with a laundry list of other explanations for the differences.

Yet historically the exit-polls were quite accurate before the new-fangled machines were put in place. Why don't your armchair quarterbacks offer up a missive on that point?

Are they biased in favor of the new-fangled machines? Sure looks that way... otherwise we would see them undertake a proper examination of that well known problem. Instead, all we get are theories steering us away from the machines.

[ProgressiveEconomist]

39. I'd like to see the next Congress investigate exit polling and

voting-machine irregularities, in preparation for mandatory national election standards in the next replacement for the Help America Vote Act.

More useful than a debate based on insufficient data would be questions the exit polling companies can be forced to answer with subpoenas.

One quesiton would be, "What are the discrepancies between unadjusted exit-poll crosstabulations and precinct Election-Day tallies, by voting method and machine type, if any?"

Let's look to the future. "Everything changed" on Tuesday, and now we have a chance to demand some real answers from those who have the proprietary data.

[Mr_Spock]

41. I'd like to see something demonstrating the accuracy of the exit polls

You have stated the opinion that the article (and I'm assuming it's references by default) is a "crude" theory of why the results of exit polls can be contradictory compared to official results, yet I don't really see much in the way of a definitive argument against the facts as stated in the article. You can't just state that the new machines are causing the discrepancies, yet assume that human sampling techniques are flawless and never change over time. And I am having trouble believing that the older exit poll data was accurate to within +/-2% (what is the MoE anyway?) consistently over history. Why is it assumed that human behavior cannot have changed over time - it seems possible that social/political behavior can change over time and this would effect the results. I'm not sure why people who would argue one side of this issue are labeled with dismissive labels such as "armchair quarterbacks" etc. in this thread - what does that accomplish? It is not knowable if there is a problem with the official results until someone undertakes a scientific study specifically designed to examine this issue. We can conjecture based on the numbers released by the polling companies, but we cannot assume anything to be definitively true since we are not privy to the sampling and statistical techniques or raw data in these exit polls. I know cheating occurs and that diligence is extremely important, I'm just not sure that exit polls are the silver bullet to election fraud - at least not until they figure out how to eliminate polling bias so the skew from raw data to final data can be eliminated.

[Awsi Dooger]

42. It has nothing to do with computer power

Come, on BeFree. If the response rate is uneven and doesn't accurately reflect the voting population the computers can't magically know it and adjust for it.

I'm still trying to figure out how the Republicans were so incompetent they stole 5 or 6 percent in Virginia and Montana, yet let Missouri alone and didn't rob a ripple. That's what the fraud crew here is apparently asserting. In that brilliant fraud watch thread last week they listed Virginia and Tennessee as the danger spots, then apparently couldn't figure out if the third state was Montana or Missouri. They were alternately identified within a couple of pargraphs in the same thread.

The Missouri exit poll was dead on, McCaskill by 2. That would have flipped senate control but maybe the GOP forgot about that state, or was too gentlemanly to steal from a woman.

[progressoid]

19. Well that article is just full of sense and logic - we have no use for that here.

K&R!

[followthemoney]

24. Reality has a liberal bias. End of discussion.

[johnaries]

33. The Election Science Institute weighs in:

The panel concluded that current media exit polls, by themselves, could not be used to monitor elections because of inherent polling imprecision. Nonetheless, for future elections, exit polls do have important uses in election monitoring. Based on this conclusion, the Election Science Institute offers recommendations on their appropriate role.
http://www.electionscience.org/Members/stevenhertzberg/report.2005-05-10.0988159730/view?searchterm=None

(emphasis added)

[BeFree]

35. Just remember

ESI is a group that found nothing wrong with the New Mexico election in 2004.

ESI's credibility is questionable when it comes to facts about election fraud.

[Ediacara]

48. If there's ALWAYS a bias toward Democrats, why don't they recalibrate?

Honestly, this unexplained bias is the silliest thing I've heard. It's exactly the same as a scientists using a thermometer that measured temperatures 3 degrees too cold and consistently forgetting about it, recording the incorrect temperature, and then fudging the numbers when the day is done, instead of either recalibrating the thermometer before you start or throwing it away and getting a new thermometer.

I don't know why it should or shouldn't have shown up initially, but what I want to know is why it continues to show up time and again despite the pollsters knowing it should show up. Are they just incompetent, or is it the only excuse they could come up with?