ASSUME WE FLIP A COIN 50 TIMES...

[TruthIsAll]

Assume we flip a coin 50 times
What is the probability it will turn up heads at least 41 times?

n= number of coin flips:50

Probability of the coin turning up Heads:50%

Let's calculate the probability that the number of heads in the 50 flips is at least 41.

Solution:
Using the Excel function, we calculate the probability:
Probability = 1-BINOMDIST(41,50,50%,TRUE)

Prob (number of heads is at least 41) = 0.000058%

or 1 out of 1,718,869

Fact:
The actual vote percentage for Bush exceeded his exit poll percentage in 41 out of 50 states.

Question:
What is the probability that this would occur due to chance alone?

[Mojambo]

1. They told me there wouldn't be any math. n/t

[jdog]

2. Thank you! for simplifying this for me.

On this issue, on Meet The Press it seems I recall Tim Russert saying that Karl Rove "knew" the exit polls were wrong and immediately started an email campaign "correcting" the impression going around in the press that Kerry was winning. Have you heard anythingn about that, and do you know what the emails said?

[bigwillq]

3. That's why exit polls should not be released

Never. Ever. Not even one.


And candidates should not be allowed on TV until all the polls close, especially on the West Coast.

[BlueDog2u]

14. I heard that also

Its the first thing I remember hearing out of the Bush camp during what appeared to be a mandate if not a landslide against them.

[Degeezer]

4. I suppose it depends on the poll takers

If they stopped and asked people randomly, one might expect results that are different that if they stopped and asked people because of some factor or other.

Ask all the cute girls and all the cute guys, and you may get a skewed answer.

As every 10th person, cute or ugly, old or young, male or female, black, white, asian, hispanic, etc and the results may be more accurate.

Also, ask 10% of people in precincts known to be Democrat and 5% in precincts known to be Republican, the results may be skewed.

There are many many more variables to consider, but you should be able to see what I mean.

[jdog]

5. But for it to break so consistently for Bush

all the time, wouldn't they all have had to be making the same errors????

[Yupster]

10. What if everywhere in the nation

A third of the grizzled old angry white men that are asked tell the pollster to fuck off? What if overwhelmingly those angry old men are Rush Limbaugh listeners?

It's just not the same as flipping a coin. You have to have random to begin with.

Also, how do exit polls capture early voters. In Texas a third of all votes are cast before election day.

[electropop2]

22. angry old white men are not a new phenomenon

This is the kind of hokum peddled by the repukes. It's just a desperate attempt to explain why a completely reliable and proven polling method would suddenly disagree with official results. You are assuming there is a new class of voters who are suddenly ashamed of the way the voted, immediately after doing it. While we may think they should be ashamed, Bushies are quite shameless. Besides, the pollsters will keep asking angry white men how they voted, until they have their quota of angry white men. As for the early votes, earlier posts here reveal that the early votes match the exit polls with excellent precision.

[TruthIsAll]

11. Exit polls have always been accurate to within 1%. Until Bush.

They are used throughout the world to monitor elections as protection against fraud. Very successfully, I might add.

They have always been right on the money in the U.S.
Until Bush started running for president in 2000.

Are our exit pollsters lacking in expertise of the foreign pollsters? They were accurate up until the 2000 selection.
When Bush started running for president.

The one who started it all, Mitofsky, has always been very accurate.
Until Bush started running for president.

Explain that.

[Degeezer]

13. I have seen reports

That say that the exit polls are not that far off. That they have a margin of error like pre-election polls do and that the final results are within the margins of error.

Also...

I have seen reports that poll takers had trouble getting close ebough to the polling places.

But the most striking thing I have heard is that Bush voters simply didn't want to talk about it. That the country has been so polarized by the war in Iraq that they had developed the habit of just not discussing politics, while Kerry supporters actually wanted to talk about it because it was easy to criticize Bush's handling of the war and the lack of WMDs in Iraq.

At the same time Kerry lacked the ability to convince most people that it would be a good idea to change presidents at this time (the war and all). The result was that if a pollster approached someone who had voted for Bush, the voter kept walking, but if a pollster approached someone who had voted for Kerry, he stopped and answered the questions.

I don't know about that.

But one thing I DO know is that the Kerry people are a pretty worldly bunch. They have been around the block a time or two and if there was a chance that they thought this election had been stolen, they would be ALL OVER IT like stink on a skunk.

They remain silent on the subject of fraud.

[electropop2]

23. The "corrected" exit polls match the tabulations...

...because the networks "corrected" the polls by mixing in tabulations from the boards of elections. That's like bragging about how well you "predicted" the outcome of the football game, just by looking at the score near the end of the 4th quarter. The original uncorrected data match very accurately to the final pre-election polls, and match even more accurately the early voting results.

And you are right about the worldly Kerry people. They have 17,000 lawyers working on it right now. As an experienced prosecutor, Kerry knows to remain silent until he has assembled the case and brought the indictment. Looks like your guy is gonna go down with a bang.

[BlueDog2u]

15. I deleted this post.

[MrUnderhill]

20. One problem with that theory...

exit polls would be much easier to "cook" than election returns.

[electropop2]

25. Why would repuke and dem pollsters both cook the books?

Wouldn't they cook differently at least? No, both were trying to be as accurate as possible, because their clients needed real information, not BS. The networks for example are still trying to recover from 2000's embarassing fiascos. They would not pay somebdy to disgrace them once again. Besides, the exit poll data was not for direct public consumption, so it was pretty useless for influencing voters.

[MrUnderhill]

35. Repuke & Dem? It was a media-hired firm.

The exit polling that was released online wasn't from the parties. The party polling that was discussed didn't match it... but how should we take THAT... both parties would have incentive to report favorable numbers.

[Nederland]

58. NOT TRUE

Exit polls were off in 1988 and 1992 by larger margins that in 2004.

[reality_bites]

92. This is not true.

http://www.washingtonmonthly.com/archives/individual/20...

We don't stand a chance if we can't get the facts right.

[TruthIsAll]

97. I see exit poll numbers: but no source.

And Dukakis never led Bush in EXIT polls.

Gore led Bush. But it was stolen. Remember FL? Over 175,000 votes spoiled.

Exit Polls have always been accurate. They use them in Germany and they have been correct to within 1/4%.

They are correct here also. But I want you to source those exit polls.

I have a lot of respect for Texeira and Drum. But the train has already left them. Instead of questioning the fraud, they question the exit polls. Short memories. They forget FL 2000 and GA 2002 And the thousands of lies out of BushCo.

Texeira disregards his own writings about the incumbent rule, Bush approval below 50%, and that most (at least 2/3) of undecided split to the challenger. He's taking all his polling experience and throwing it right out the window. He does not even entertain the possibility of fraud.

What ever happened to "The Emerging Democratic Majority"?

[lizzieforkerry]

18. I am pretty sure the poll takers were professionals

They poll a certain number and then only poll what they are missing after that.

[electropop2]

21. The poll takers were seasoned professionals

such as John Zogby, who had accurately predicted every outcome up until 2000. They used reliable, scientific sampling, and interviewed many more voters than the typical "horserace" prosepective polls. Exit polls are inherently very reliable, because you are not asking "likely" voters who they might vote for; you are asking known voters who they already voted for.

[reality_bites]

94. Again, not true

Exit polls are not necessarily accurate. I don't know who started this rumor (falsehood), but it has damaged the cause in that it has sent hundreds of DUers down the wrong argument path. It is a real waste of time and effort.

http://www.washingtonmonthly.com/archives/individual/20...

[goclark]

121. Got it!

And even though the exit polls could have been fixed, Kerry still came out ahead!

That is a real reason why the exit polls didn't lie this time.

I can't believe that many Kerry voter would have been proud to say they voted for Bush when they didn't.

My other point is, let's flip the coin. If the Exit polls had indicated that Bush won and Kerry was declared the winner, case closed. The Exit Polls would have been blessed by Rove and used as proof positive.

So, let's act like BushCo for a change, declare them PROOF!

[tridim]

6. Once in Vegas I saw a 25 spin streak of "red" on a roulette wheel

Obviously a roulette wheel is completely random, unlike the election, but the odds of 25 reds in a row is huge. I don't really have a point, but it was funny to watch all the players place large bets on black over and over because they just knew that a black was due. Everyone was pissed.. Of course on each spin there is exactly a 50% chance of landing on black regardless of the past red streak.

You've been doing wonderful work TIA. What gets me is that the repukes are all discounting statistics as junk science, when obviously it's not. The methods are sound.

[Rumba]

7. Not 50%

Minor correction, tangential to the point of the thread and your post... but the green 0 and 00 on the roulette wheel reduce the odds of either black or red below 50%. Hence the 1-1 payoff for betting black or red favors the house.

[tridim]

8. You're right, I forgot about the green.

It's still fairly close to 50%.

[GAspnes]

9. You might enjoy participating in this discussion too

over at Crooked Timber.

http://www.crookedtimber.org/archives/002890.html

Further Analysis of Electronic Voting Patterns
Posted by Kieran

Mike Hout and some colleagues at Berkeley have a working paper called “The Effect of Electronic Voting Machines on Change in Support for Bush in the 2004 Florida Elections”. A summary is also available as well as the data itself. Hout is a well-respected sociologist, so if he thinks the data for Florida show some anomalies he’s worth listing to. Hout et al try to estimate whether the presence of touch-screen electronic voting made a difference to the number of votes cast for Bush, controlling for various demographic characteristics of the counties as well as the proportion of votes cast for the Republican Presidential candidate in 1996 and 2000. Here’s the punchline:

As baseline support for Bush increases in Florida counties, the change in percent voting for Bush from 2000 to 2004 increases, but at a decreasing rate. Electronic voting has a main, positive effect on the dependent variable. Furthermore, there is an interaction effect between baseline support for Bush and electronic voting, and between baseline support for Bush squared and electronic voting. Support for Dole in 1996, county size, median income, and Hispanic population had no significant effect net of the other effects. Essentially, net of other effects, electronic voting had the greatest positive effect on changin percent voting for Bush from 2000 to 2004 in democratic counties. … Summing these effects for the fifteen counties with electronic voting yields the total estimated excess votes in favor of Bush associated with Electronic Voting; this figure is 130,733.

Hmm. I’m going to go mess around with the data for a while and see what we can see.

<more>

[Carolab]

66. Old news

from last week--a press conference was held in California and this was all discussed then

[stellanoir]

12. Just an aside. . .

in as much as a value TIA's conclusive data tremendously. . .

FYI. . .try flipping a New Hampshire quarter. It's lopsided. It has the now crestfallen Man of the Mountain on it. (in full accordance with the curse of the state quarters tin foil theory)

You can rig a coin toss with this quarter if you have a very measured throw.

You can rig an election with partisan control of voting machines, their tabulations, and their certification.

The seemingly slumbering populous will never notice.
The corporately controlled media will be complicit.

I know I'm preaching to the choir but this is getting exceedingly frustrating.

[TruthIsAll]

16. Remember Brando in Streetcar? Stella, Stella...

Thanks
tia

[stellanoir]

49. Oh sheesh

I've never heard that line before. Still it's pretty ironic since one of my distant relatives wrote that play. Aaaand no, it's not my given name.

Too silly really.

best

[TruthIsAll]

79. Stella, are you related to Tennessee Williams?

tia

[Nederland]

17. Ho Hum

Same old same old.

Assume that the exit polls weren't crap and procede.

Ever hear of the phrase "garbage in, garbage out" TIA?

[TruthIsAll]

24. Why is it only when Bush runs, Exit Polling is "Garbage In, Garbage Out"

What about ALL the other elections where Exit Polling is used throughout the world to discourage election fraud?
Are they all just GIGO?

What about elections in the U.S. prior to 2000?
Was Exit Polling GIGO before Bush? Or only since 2000?

Nederland, you are in never-never land.

[Nederland]

42. Response

First, in the past the exit polls were done by a different company. Second, and more importantly, you are completely ignoring the facts. It is a FACT that the company that conducted the exit poll has said that the early results they distributed to the networks were WRONG. This is not disputable, it is a FACT. The question is, why do you continue to behave as if the early exit polls are gospel? It makes absolutely no sense whatsoever.

[TruthIsAll]

43. Nederland, you are WRONG again. The company just changed it's name.

It was always headed up by Mitofsky.

Remember VNS? You know, the consortium which refused to exit poll in 2002? That was Mitofsky. He's been doing it for years. A monopoly.

Go Google.

[Nederland]

55. Response

You are correct about Mitofsky. He was involved in VNS.

However, you completey ignored the other point I made in my post. You know, the one I prefaced with the phrase "more importantly"? Funny how you completely ignored that point and concentrated on a minor piece that is ultimately irrelevant.

Here is the transcript of an interview with Mitofsky:

http://www.pbs.org/newshour/bb/politics/july-dec04/exit...

Read it and answer a very simply question: do you trust him?

[MrUnderhill]

19. That assumes the chaces are 50-50 for each state.

And that there are no systematic problems inherent in the poll methodology (different voting population for different times of day perhaps?)

It also doesn't address whether the "problem" identified is with the electoral returns or with the poll.

Arguing in favor of the polling methodology being in error are two facts:

1) This happens regularly
2) The exit polling showing an incredibly high gender bias in the sample group cannot reflect actual voting.

[TruthIsAll]

28. Show us historical evidence of exit poll "error".

Just because you say so does not make it true.

Source it. Prove it. Otherwise you are just blowing smoke.

And don't use 2000, 2002, 2004 as your source. They were Bushian elections.

[MrUnderhill]

34. 2000 & 2002 would have been my "sources"

SO I guess there isn't much to talk about.

Except that the way the exit polling was done was changed for the 2000 election, right? Didn't each network used to do it themselves?

[reality_bites]

95. Here

http://www.washingtonmonthly.com/archives/individual/20...

[John Q. Citizen]

30. There are two options; either the polls were flawed or the tally

is flawed. The statistical analysis by UCBerkley gives more weight to the theory that the tally was incorrect.

Plus deep down in everybody's heart we know there is no way.

Suppressing the vote isn't even in dispute, it's wide spread in particularly Democratic strongholds, long lines, hundreds of thousands of "lost" absentee ballots, the ridicules problems of overseas voters, These people have demonstrated and said that they needed to suppress key precincts. This is a huge spike in crime across the US, more than 100,000 election crimes reported from across the country in one day. How many crimes weren't reported? People who waited in line for two hours then had to split?

If the Republicans were willing to aide and abet this felony crime of suppression, what makes you think they wouldn't tweak with the vote tallies, especially when we have seen widespead tally problems in many areas.

Exit polls are not regularly wrong and there is no evidence of a gender bias in the polling that I have seen. If you have evidence to support the gender accusation, please show it MrUnderHill

[MrUnderhill]

37. Berkley's work is not sufficient. Their gap was like 1%

The polls were much farther off than that.

[nmoliver]

26. Another way to look at the exit poll anomalies

I'd throw light on the answer this way.

On the morning of Sept. 11, there were four hijacked planes flying around in the sky within two hours of each other. All left from East Coast airports.

What is the probability that these four hijackers were not associated with each other at all, and the simultaneous appearance of 4 hijacked planes in the skies was a coincidence?

However low the probabilities, maybe we should - prima facie - reject the possibility that the hijackers were coordinated with each other. After all, that would make it a conspiracy, and we good Americans don't believe in conspiracy theories.

Has anybody really proved that these hijackers were all in cahoots under a central command? Yet we all knew right away. Nobody has ever doubted it. Why? Because to figure otherwise would have been TOO MUCH OF A COINCIDENCE.

Or, maybe only Arabs conspire?

-- Nina

[On the Road]

27. Of Course It's Not Due to Chance

but if that's all the information you have, there's no reason to conclude it was due to fraud.

[TruthIsAll]

29. It's NOT all the info. What about the deviations being BEYOND the MOE

1. Look at the number of states which shifted to Bush.
2. Look at the deviations, based on (1). Compare them to the MOE.
3. Compute probabilities for each state based on (2).
4. Calculate the TOTAL probability, based on (3)
5. Do you smell a rat yet? If not, keep your head in the sand.
Otherwise,
6. Count ALL the votes. At least those which did not disappear in Cyberspace.
7. Investigate ALL instances of voter intimidation, voting methods and spoiled votes.
8. Find whistleblowers.
9. Pound the sleeping media. Except for Olbermann.
10. Express outrage.
11. Study Election 2000. Was Gore cheated?
12. Study Election 2002. Was Cleland cheated?
13. Study Election 2004. Did Bush have a 48% approval rating?

[Shalom]

31. Good Corrections, But Makes the Case Even Stronger

Your points are correct - the best example is the study by Steven Freeman, who only looked at Florida, Ohio, and Pennsylvania, and found that the likelihood of Bush winning those states with the MOE distribution was 1/250,000,000. It would have been 1/8 with the "coin toss" analogy.

As such, the assumption of the coin toss makes it an extremely conservative calculation. Even under these circumstances, the likelihood of 41/50 "heads" is so small, it hardly matters. (neither does the fact that the actual spread is 42/51 states)

P.S.: For those who think the binomial distribution is rocket science, it isn't.

Take a simple example: what is the probability of throwing 3 heads in four tries ?

Let H= head, and T = tail.

Now if we looked at 3 heads, followed by 1 tail, we can express it as:
HHHT.

The probability of HHHT is:
(1/2)*(1/2)*(1/2)*(1/2) = 1/16

However, there are 4 ways of getting 3 heads:
HHHT
HHTH
HTHH
THHH

The probability P of 3 heads in four tries is simply:
P = (Number of ways it can happen) * (probability of one way)

In this case,
P = 4 * (1/16) = 1/4

When we have larger numbers, the calculations get more complicated, but the principle is the same.

[On the Road]

32. What I'm Saying Is: It's a Straw Man Argument

Nobody thinks the deviation from the exit polls is due to chance. Nobody.

The point is that in addition to fraud there are a number of other factors such as sampling, weighting, and response variables that can account for the difference. Mitofsky, who was responsible for the study, has a detailed, plausible explanation. I'm not saying it's correct. I'm just saying these are not the terms of the debate.

To make progress, facts must be submitted to a candid world. And publicizing this "what are the chances" argument is one of the worst things that can happen to the fraud investigation.

[reality_bites]

96. ditto

[On the Road]

33. Dupe

[MrUnderhill]

36. You need to reevaluate "MOE"

That margin of (sampling) error. Meaning if you took the same group mixing and asked the same question... you would get the same answer within the MOE percentage.

It does NOT account for other survey problems.


What if, for instance, Democrats vote earlier (on average) than Republicans? It could throw your numbers off much more than the sampling error.

If the exit polling was actually 100% accurate, then we would ahve to assume that women vote much earlier (on average) than men do. The poll could have ZERO sampling error and still be way off if it didn't take actual demographics in to account.

And, as I said, the books could have been tainted. They may have simply done a poor job of selecting precincts, or party activists may have gotten early word on where those polls were being conducted (it would have been brilliant).

[slackmaster]

38. Your analysis is based on a flawed assumption

That variations in the exit poll numbers are random, i.e. not biased in any direction one way or the other. The coin-flip is a totally inappropriate analogy here.

TruthIsAll, you are approaching this analysis from a completely bass-ackwards direction. If you want to provide a credible analysis to show the actual election results were rigged, you have to start with a null hypotheses that the discrepancies between the reported vote counts and the pre- and post-election straw polls are the result of a flaw in the polling process, not in the election.

It's really that simple - Either the election results are wrong or the exit polls were done improperly. If you want to prove the election results are wrong you must start with the presumption that they are correct.

[IAMREALITY]

39. The coin flip

Is a perfect visual for sake of simplicity to DU'rs.

TIA, thank you as always for your gracious and hard working efforts to help keep the issues on DU in perspective.

[slackmaster]

40. My remarks were addressed to TruthIsAll

Do you two know each other?

[IAMREALITY]

41. Nope

But since this is an open forum we don't need to. TIA is a true patriot.

[Name removed]

100. Deleted message

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

[Name removed]

102. Deleted message

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

[foo_bar]

104. "disciplined"

thanks in advance.

Ad hominem attacks are second-to-last.

[foo_bar]

107. "patriotism is the last refuge of a scoundrel." -Samuel Johnson

[TruthIsAll]

44. No, we don't. I know no one on DU.

And so what if we did?

[slackmaster]

45. Nice dodge but still no response to my critique of your methodology

Got answers?

Why are you so willing to take someone else's word that their exit polls were conducted properly? That their sampling methods were sound? That their calculations of uncertainty are accurate?

TIA, you're approaching this issue with the zeal of a "creation scientist" poring through rock slices looking for "Po rings" or scouring Miocene sediments looking for signs of modern horses - You know the answer you want to reach, and you're looking for anything that might suggest your foregone conclusion is correct. That's not how science works.

You're passionate, you need to be dispassionate.

You're emotionally involved, you need to be detached.

[IAMREALITY]

51. Interesting

I'd give anything to see you come up with a better study. I'd actually give anything to see you contribute something to this forum. TIA has put forth monumental efforts to further the cause. If you have any studies that you have done that are worthy I'd love to see them.

And no, I'm not answering for TIA. He is more than capable of answering for himself. But since this is a public post rather than a private email any member here has a right to offer an opinion on any post made.

[slackmaster]

54. Burden of proof

I'm playing Devil's Advocate here - A flaw in the exit polls requires less accommodation than a gigantic conspiracy to influence the results in favor of the GOP. Occam's Razor and all that.

Null hypothesis has to be the election was relatively OK and the exit polls messed up. The real truth may be a combination of the two, but if your real goal is to impugn the official election results you must start with the presumption that they are OK or close to OK, then rule out all other explanations (besides massive fraud) for the apparent failure of the polls to predict the result.

[IAMREALITY]

57. I'll Repeat

Please, I implore you, submit a study for us DU'rs to see. You seem to want to protray yourself as so knowledgeable, so I'm sure you could come up with something that we all would respect and honor as much as TIA's......

I look forward to seeing your work (as opposed to your rhetoric)

[slackmaster]

60. Gee, IAMREALITY

How could someone of your stature possibly need any help?

[Name removed]

61. Deleted message

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

[slackmaster]

62. It's more entertaining to watch TIA shout "WRONG" to every criticism

And to watch you back TIA up by bashing anyone who criticizes the alleged numerical analysis.

I posed the question first: Validate that the exit polls are good. Responding to that challenge by requesting that I do an independent analysis is not satisfactory. That would be like a medical researcher responding to rejection of a journal article submission by challenging the editors of the New England Journal of Medicine to do a better study.

I volunteered to take on the role of academic review here. So far I've seen nothing but schoolyard taunts and couched accusations that I'm not a patriotic American in response.

[Name removed]

63. Deleted message

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

[slackmaster]

65. You're saying that calling me "Sean" isn't taunting?

No one is taunting you...

That's pretty funny.



My name isn't Sean BTW.

[Name removed]

67. Deleted message

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

[slackmaster]

87. How about if I call you "TruthIsAll's Lapdog" or just 'Dog for short?

If the shoe fits...

[foo_bar]

108. that's unnecessary

the truth will out; even creationists can't prolong the inevitable.

[slackmaster]

86. Calling me "Sean" isn't taunting?

Yeah, suuuuuuuuuuuuure it isn't.

[Name removed]

73. Deleted message

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

[Heewack]

77. I don;t think he is trying to get your goat here.

He is simply pointing out that you have to use the vote total as the null, and assume it is right to go about proving your contention. A poll by nature is a guess at what the outcome will be. The outcome is the outcome, so you must start from that basis point to prove or disprove the hypothesis. You can't start with the notion that the poll was correct and the outcome was wrong because the poll is inherently less accurate than a person by person tally.

Don't start calling me Sean, I just don't think your methodology as put forth is able to back up your hypothesis.

I admire your steadfastness to the cause, but let's make sure we are doing things right.

[Name removed]

81. Deleted message

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

[Heewack]

82. Not at all.

I'm just saying that the flipping the coin analogy is not valid in this case. We all want ALL of the votes counted, but the "probability" in this election was 50/50. That's what the pre-election polls told us.

[TruthIsAll]

88. Sorry. Undecided votes swing to the challenger. Kerry 51-48.

tia

[Heewack]

91. Again that is based on a probability

It is not a certain. The only certain is the vote tally itself. You can't use the guess to set the rule for the reality of the outcome. It's basic statistics.

I don't know why you see this as something that must be refuted on a persoanl level. The premise is flawed, we all make mistakes.

[TruthIsAll]

93. It's based on historical statistics. Where are your facts?

.

[foo_bar]

105. the fact is, one firm's polling statistics aren't facts

To wit:

The margin of error is a simple transformation of the number of respondents into an ambiguous term that is neither a "margin" nor the whole of "error".

The margin of error is being confused with the confidence interval of reported percentages.

The 99 percent confidence interval radius is smaller than the margin of error for any percentage besides 50 percent.
It is much smaller and more asymmetric for very high and very low percentages.

It is not a "margin" at all; the probability of the true percentage being outside the margin of error is low but nonzero.
There is no agreed-upon confidence level. Most pollsters use 99 percent, but many use 95 percent or 90 percent; this makes their polls look more accurate.

When the purpose of polls is to compare percentages, the use of the margin of error is tempting but inappropriate.

Perhaps most importantly, there are many different sources of error in polling, and variance due to sample size is not likely to be the only contribution. Other possible contributions to error include:

Sampling bias, when the sample is not a representative sample from the population of interest.

The phrasing of the question may not be appropriate for the conclusions of the poll.

Response error (Sudman & Bradburn, 1982)

Deliberate distortion (fear of consequences, social desirability, response acquiescence).

Misconstrual (not understanding the question).

Lack of knowledge (guessing to try to be helpful).

source

[TruthIsAll]

111. You have accomplished nothing but copy and paste.

Now, get to work and address a SPECIFIC issue in any of the analysis I have done. Go ahead. If you dare. And if you don't respond to this post, I'll know you just gave up.

So far, you have proven that you can copy and paste from Wikpedia and create Powerpoint slides.

Come on, wise guy, let's see what you've got.

[foo_bar]

112. project much?

What has the guru of one spreadsheet command accomplished beyond cut 'n'paste and Microsoft presentations?

On edit: for specific, unrebutted assertions as to the "garbage in garbage out" nature of your estimative polls of polls, read the odd-numbered posts on this thread. Or click the "my posts" button, and attempt to respond without ad hominem rule violations.

[slackmaster]

90. Well, you have about 600 more posts than I do

I guess that settles it.

Uncle.

[slackmaster]

85. Now WTF would I want with your old goat, TIA?

I already have three cats, two turtles, and a desert tortoise to care for.

[TruthIsAll]

46. Wrong! My premise is that the Exit Polls were correct. For a few reasons.

1. They have always been correct. Except when Bush is involved.
2. Bush stole the 2000 election. If he did it then, he would of course do it again.
3. The exit polls agreed with the pre-election polls from Zogby and Harris (who were both spot-on in 1996 and 2000) and others.
4. The Repubs fought tooth and nail against a paper trail. Use common sense. WHY?

The burden of proof is not on those who question the votes.
It's on those who question the exit polls.

I say let's count ALL the votes. At least the ones not stolen in Cyberspace by hackers. Let's include the spoiled punched cards. And I want to see ALL the Optiscam ballots.

I'm getting tired repeating the same mantra. It is always thus with you guys who turn up to defend the indefensible.

[slackmaster]

48. Exit poll results are like mutual stock funds

1. They have always been correct....

...3. The exit polls agreed with the pre-election polls from Zogby and Harris (who were both spot-on in 1996 and 2000) and others....


Have you ever heard the expression "Past performance is not a guarantee of future results? Same principle applies here.

The burden of proof is not on those who question the votes.
It's on those who question the exit polls.

It's really very sad that you are so wrapped up in your foregone conclusion that you are unable to see the folly of that statement.

By all standard scientific methodology you should be looking for the flaw in the 2004 exit polls. Are you afraid to look at them? If you find no problem with them then you can start saying it looks like the election was rigged.

I say let's count ALL the votes. At least the ones not stolen in Cyberspace by hackers. Let's include the spoiled punched cards. And I want to see ALL the Optiscam ballots.

I fully agree with those statements.

I'm getting tired repeating the same mantra. It is always thus with you guys who turn up to defend the indefensible.

You're the one who is clinging like a Gila monster to flawed methodology. Your defensiveness on this point is telling.

[TruthIsAll]

50. Flawed methodology? I refer you to Prof. Freeman and Dick Morris.

Try as you might, the TRUTH will one day hit you upside the head.

Even you will have to accept it.

By the way, where is YOUR analysis?

[slackmaster]

52. That's not an adequate answer

You are apparently not willing to do the work required to construct a credible attack on the election results. You talk the talk but aren't walking the walk. I'm throwing hardball questions and you're just ducking.

By the way, where is YOUR analysis?

Changing the subject won't fix the problem. You proffered an analysis and I'm criticizing it as would any peer review committee member for an academic journal. I'm trying to help you do a better job. If you don't want to listen to critics like me you will fail. I guarantee it.

[IAMREALITY]

53. Sean Hannity

Would act the same way.

"You are apparently not willing to do the work required to construct a credible attack on the election results. You talk the talk but aren't walking the walk. I'm throwing hardball questions and you're just ducking"

That reminds me so much of him it makes me want to vomit.

[slackmaster]

56. Feeling nauseous is an emotional reaction

You're too involved in this to look at it objectively IMO.

Shoot the messenger, ignore the message, rinse, repeat.

[IAMREALITY]

59. Sean, Please, I Urge you

Please submit your own study that will garnish the attention that TIA is used to.

[MrUnderhill]

69. Dick Morris?????!!?

I hesitate to ask... but I have to know.

What does HE have to say?

[reality_bites]

98. Exit Polls are not reliable, historically or otherwise.

It is up to you to show that exit polls are accurate. But you can't, because they aren't.

http://www.washingtonmonthly.com/archives/individual/20...

You are wasting your time. Please try another approach.

[bleever]

47. This example is indeed relevant. All of the other factors have been

controlled for in a number of ways. The historical accuracy of exit polling, and the very close conformity of the exit polling to the final pre-election polling (when considered as an aggregate of the most historically reliable sources)are critical elements here.

One could hypothesize that the exit polling was systematically skewed in favor of one candidate, and if anyone has any evidence to back that hypothesis up, we'd all like to see it. But this election played out like a rigged game of three card monty, and it would be more constructive for critics of TIA's work to support opposing hypotheses with some solid evidence, rather than nitpick at his methodology. And if it isn't just "nitpicking", then it shouldn't be hard to provide something solid as an alternative explanation.

[slackmaster]

64. It's not nit-picking, it's playing Devil's Advocate!

Assuming that the 2004 exit polls were done properly because some earlier polls accurately predicted the outcome of different elections does not cut the mustard. How do we know the pollsters in question didn't change their methodology? Did the demographics change? Did consolidation of voting precincts cause unexpected changes in voting behavior (i.e. higher percentage of people voting after work as opposed to early in the day)?

You have to dissect the 2004 polls to figure out what, if anything went wrong with them. If you can't find a smoking gun there, it becomes much easier to make a case that something went wrong with the election.

If you can't address this kind of criticism from a friendly person like me you have no hope of making a credible case in court of public opinion.

[Name removed]

68. Deleted message

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

[Heewack]

70. You were onto something...

with the fact that voting results can't be given a probability because it is not made up of chance. There is no other analysis that could be done because the premise is flawed. Each person went into the booth knowing whom they were going to vote for. That is not chance. If each person had flipped a coin to decide whom they were voting for then the analogy would be valid.

[TruthIsAll]

75. BRILLIANT, JUST BRILLIANT. NOT.

YOUR SECOND POST WILL BE FOREVER ENSHRINED IN THE DU PANTHEON.

HOW TO LOSE CREDIBILITY INSTANTLY BY VIRTUE OF AN INANE SECOND POST.

JEEZ, SEEMS WE STRUCK A VEIN HERE.

SLACKMASTER'S TROOPS ARE COMING TO RESCUE THEIR CAPTAIN.

[TruthIsAll]

71. HOW DO WE KNOW THAT YOU ARE NOT SEAN HANNITY?

.

[ahyums]

74. the worrying thing is that Hannity actually predicted the results

pretty much exactly two days before the election. It's another one of those things that make me suspicious.

[slackmaster]

84. I DON'T KNOW

HOW DO WE KNOW YOU DON'T BEAT YOUR KIDS?

[TruthIsAll]

72. NO, SEAN, EXIT POLLS WERE CORRECT IN 2000 AND 2002

IN 2000, MORE VOTERS WENT TO THE POLLS IN FLORIDA TO VOTE FOR GORE THAN FOR BUSH.

ONLY THEY HAD NO WAY OF KNOWING THEIR VOTES WOULD BE DOUBLE OR TRIPLE-PUNCHED, OVER 110,000 OF THEM.

HOW WOULD VOTERS IN PALM BEACH KNOW WOULD BE BUTTERFLIED INTO VOTING FOR BUCHANAN?

THE EARLY EXIT POLLS SHOWING GORE WON FLORIDA WERE CORRECT. THOSE WHO SPOILED THE VOTES FOR BUSH KNOW THAT.


IN 2002, MORE VOTERS WENT TO THE POLLS IN GEORGIA TO REELECT MAX CLELAND, A TRUE AMERICAN HERO WHOSE CHICKEN-HAWK OPPONENT SAXBY CHAMBLISS (WITH BUSH'S BLESSING, I'M SURE) HAD HIM MORPHED INTO SADDAM HUSSEIN.

WHO WAS TO KNOW THE VNS EXIT POLLS WOULD BE SHUT DOWN THAT DAY? WHY WOULD THEY DO THAT? TO HELP THE REPUBLICANS STEAL THE ELECTION? NO, THAT CAN'T BE. WHY WOULD THEY DO THAT? REPUBLICANS ARE VERY TRUSTWORTHY. JUST LOOK AT THEIR TRACK RECORD.

THEY ALWAYS CLAIM COMPUTER GLICHES, SEAN, GLICHES. ALWAYS GLICHES.
LIKE IN VOLUSIA WHEN DIEBOLD SWITCHED 16,000 VOTES MEANT FOR GORE TO THE SOCIALIST CANDIDATE. DAN RATHER CALLED THAT A COMPUTER GLICH.

HOW ABOUT THIS TIME, WHEN THE EXIT-POLL GOD MITOFSKY CLAIMED THE EXIT POLL MAINFRAMES WENT DOWN AT 9PM. COMPUTER GLICHES. ALWAYS AT THE RIGHT TIME.

I SAY, GIVE HIM A LIE-DETECTOR.


SHAWN, WHY DON'T YOU JUST GIVE IT UP AND STICK TO YOUR TV SHOW?

I

[Name removed]

83. Deleted message

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

[reality_bites]

99. No. For the 2000 presidental election, they were off

outside the margin of error - 2.8%.
In 1988 they were off 7.1% in favor Dems
In 1992 they were off 7.2% in favor Dems
In 1988 they were off 5.8% in favor Dems

See:

http://www.washingtonmonthly.com/archives/individual/20...

and

http://www.tcf.org/publications/pow/nov17_2004.pdf

[TruthIsAll]

101. I will address 2000. Very Quickly. Now listen carefully.

The election was stolen. The exit polls were right. The number of votes counted was wrong. Got that? Florida. Florida.

Gore won by a lot more than 537,000 votes. But the votes were spoiled - as they have been for the Democrats since time immemorial. The Dems always lose millions of votes in minority precincts in various and sundry ways.

Welcome to the real world of perennial election fraud.

[Nederland]

123. Interesting

The poster provides a link demonstrating that exit polls aren't always correct. He cites four different elections where they were off, but you address only one. What about the others? Are you conceding the point that exit polls aren't always correct?

[Freddie Stubbs]

76. The probability could be quite high

if the exit poll methodology was faulty.

[Heewack]

78. This is true.

There is a probablity affixed to the exit polling itself. It could be 10%, 50%, 90%, 99% accurate, but it will never be 100% accurate.

[slackmaster]

89. BLASPHEMY!

How dare you try to introduce reason into this thread?

[foo_bar]

106. my powerpoint slide says you're a lyin' yellow-belly fleeper!

[slackmaster]

122. My solar-powered hard slide rule case says yer a Halliburton stoogie!

[bleever]

80. Kinda funny that only one side here has any empirical evidence.

Fortunately we don't need to find irrefutable evidence. We're not looking to get published or win the Nobel in math here. We're just trying to get the votes counted.

And ladies and gentleman, that train has left the station.

Thanks, TIA.

[foo_bar]

103. a binomial dist on crap is still crap

Exit polls aren't random occurrences (neither are coin tosses for that matter, unless they're virtual coin tosses on a "truly random" number generator).

You can punch the accuracy of your pre-11/2 predictions into Excel, then marvel at the inconceivable odds of being wrong every single time. A 0.00 batting average only impresses in the creationist and telepsychic industries.

[TruthIsAll]

109. Ok, sage, tell us now that Bush won fairly. The exit polls were wrong.

Go ahead. Say it.

[foo_bar]

110. fallacy of a false dilemma

Example

Either you are with me or against me.

http://changingminds.org/techniques/argument/fallacies/...

You're wrong, and one firm's midday exit polls were wrong. These two facts aren't mutually exclusive.

[TruthIsAll]

114. Of course, you failed the test. No specifics. Just grandiose verbiage.

I want to see something substantive. Your flowery prose is nothing but ether.

[foo_bar]

117. if you can't grasp conditional logic, there isn't much to explain

http://www.democraticunderground.com/discuss/duboard.ph...
http://www.democraticunderground.com/discuss/duboard.ph...

I'll await more projective, unspecific attacks on the messenger in lieu of cogent responses. Not unlike a certain political figure incapable of admitting fallibility.

[TruthIsAll]

115. Answer the question. Answer the question.

From now on, play your word games with yourself.

Keep reciting your googled references.

And avoid answering the simple questions I pose.

[foo_bar]

119. plonking uncomfortable truths doesn't make your handle less Orwellian

These "word games" are called discussion and peer review, yet they pose an inimitable threat to a faith-based worldview.

[bleever]

113. Still waiting for some empirical refutation to go with that sniping.

[TruthIsAll]

116. You won't get any from these freepers.

tia

[foo_bar]

120. Freepers are characterized by a "with us or against us" world view

that treats dissent and critical thinking as enemies of "the cause". Seem familiar?

[foo_bar]

118. the burden of proof is on the asserter

I can't disprove the proposition of cold fusion in TIA's bathtub, or electoral fraud by Bigfoots. What I can prove is that TIA's base assumptions change every poll to support pre-ordained conclusions.

http://www.democraticunderground.com/discuss/duboard.ph...