Exit Poll Question

[Mistwell]

I've asked this question twice now, with no response. So, I figured I would make it a thread.

According to this site:

http://synapse.princeton.edu/~sam/pollcalc_letters_florida_jimg.html

Bush actually did better in some exit polls than he did at the actual poll.

New Jersey: B +2

New Mexico: B +1

Wisconsin: B +2

And their data shows that Bush was falsely favored in some counties in Florida as well.

Were these somehow disproved at another time? Am I misreading this data? Or is this just old news that was refuted in some other way?

[Salomonity]

1. No, exit polls are sometimes wrong. --nt

[TruthIsAll]

36. THE RED SHIFT DOWNLOADED BY SIMON SAYS OTHERWISE

41 STATES DEVIATED TO BUSH (INCLUDING NJ,NM,WI), 17 WERE OVER 3%
10 STATES DEVIATED TO KERRY, 2 WERE OVER 2% (IN KS AND ND, NO LESS).

I WON'T REPEAT THE ODDS AGAIN.
YOU HAVE SEEN THEM BEFORE.

OF COURSE, SOME WILL SAY ITS SYSTEMATIC EXIT POLLING ERROR - ALWAYS IN FAVOR OF BUSH.

"SIR, I PRESSED KERRY, BUT IT KEEPS COMING UP BUSH. WHY IS THAT"?

[Mistwell]

46. Yeah, see, there is the claim again

"OF COURSE, SOME WILL SAY ITS SYSTEMATIC EXIT POLLING ERROR - ALWAYS IN FAVOR OF BUSH"

See, there is that claim again, despite the fact that you just yourself posted that it was NOT "always in favor of Bush".

Odds say that 50% should deviate Kerry, and 50% should deviate Bush. In this case, you are saying 20% deviated Kerry, and 80% deviated Bush, for a total variance from the middle by 30%.

I don't know the odds of that happening, or if it is unusual in comparison to other exit polls from other countries and from this country from prior elections.

However, I do know that your own post indicates that the claims of "always favoring Bush" are false. They should not be made. That level of extremism, in my opinion, hurts the cause. It's better to be accurate, and present the issue in a truthful and therefore more compelling light, than to exagerate and taint the power of the claim with the exageration.

I know you disagree on this issue, and think it is better to use hyperbole to get people's attention rather than to be straight up in providing all of the data (warts and all). I guess we just have to agree to disagree on the best approach for this issue.

[TruthIsAll]

49. The odds of 41 out of 50 going to Bush are 1 in 356,000. See post 7.

Its like flipping a coin and having it come up heads 41 times.
The odds are equal for a swing to Kerry or Bush: 50%

This has absolutely nothing to do with MOE or sample-size.

So how will you spin that one?

[Mistwell]

56. I'm not spinning anything

The whole point of this thread is to END spinning this stuff and just be straight about it. For example, you yourself spun this issue in this very thread by claiming that ALL deviations swung for Bush, when you know damn well it wasn't ALL, just MOST. That's spin.

Show me your math for the odds, and I'd be happy to look at it.

[TruthIsAll]

59. I NEVER SAID THAT ALL DEVIATIONS WENT TO BUSH.

LOOK AT POST #36.
YOU ARE BEING VERY CARELESS IN YOUR COMMENTS.

[Mistwell]

68. It wasn't careless

You said "OF COURSE, SOME WILL SAY ITS SYSTEMATIC EXIT POLLING ERROR - ALWAYS IN FAVOR OF BUSH"

How is that materially different than "ALL DEVIATIONS WENT TO BUSH".

What else did you mean by "ALWAYS IN FAVOR OF BUSH". The word "ALWAYS" has meaning in that sentence. What meaning did you think it had?

[TruthIsAll]

76. 86 of 88 documented touchscreens switched Kerry to Bush.

Let's stop the word games.

The fact is the VAST PREPONDERANCE of voting "anomalies" worked out in favor of Bush.

Do you deny that?

And 40 out of 50 states deviated to Bush.
Do you deny that also?

You can't.

So you try to deny the math.
But you don't understand the math.
That's obvious.

I WILL SPELL IT OUT FOR YOU ONCE AGAIN.
I WILL TRY TO BE VERY CLEAR ABOUT THIS.
IF YOU PAY ATTENTION, YOU WILL GET IT.
EVEN IF YOU WON'T ADMIT IT.

1. THERE IS 50/50 PROBABILITY THAT THE FLIP OF FAIR COIN
WOULD TURN UP HEADS.

2. THERE IS A 50/50 PROBABILITY THAT IN A FAIR ELECTION ANY GIVEN STATE WOULD DEVIATE TO KERRY OR BUSH FROM THE EXIT POLL.

3. 40 OUT OF 50 STATES(80%) DEVIATED TO BUSH.

4. IT IS ILLUSTATIVE TO COMPUTE THE ODDS.

5. THE BINOMIAL DISTRIBUTION FUNCTION IS APPLICABLE, SINCE WE HAVE AN EITHER/OR, YES OR NO. THAT'S CALLED A BINOMIAL EVENT. THERE CAN OCCUR ONLY ONE OF TWO POSSIBILITIES: DEVIATE TO KERRY OR DEVIATE TO BUSH.

6. THE ODDS ARE 1 IN 356,000 THAT 40 OF 50 STATES WOULD DEVIATE TO BUSH, WHICH THEY DID. BUSH WAS VERY LUCKY.

THE SAME ODDS WOULD APPLY TO KERRY. BUT HE WAS NOT SO LUCKY. IT DIDN'T HAPPEN.

NOW, LETS ASSUME THAT THE EXIT POLL WAS CORRECT.
THAT MEANS IT WAS NOT A FAIR ELECTION.

BUT YOU AND YOUR BUDDIES REFUSE TO EVEN CONSIDER THAT SCENARIO.
YOU SAY THAT THE VOTE TALLIES ARE CORRECT. PERIOD.
OF COURSE YOU SAY THAT.
WHAT ELSE CAN YOU SAY?

[Mistwell]

94. Response

"Let's stop the word games."

I didn't think it was a game, but fair enough.

"The fact is the VAST PREPONDERANCE of voting "anomalies" worked out in favor of Bush. Do you deny that?"

It sure does look that way. Of course, vast perponderance is not the same as always. Being precise is a good thing. This, by the way, is near the top of the list of things that seem fishy to me...the use of words like "Always" when what you mean is "Almost always". It might be a minor quibble to you, but to me it's quite important.

"And 40 out of 50 states deviated to Bush.
Do you deny that also? You can't."

I have to check. There is serious question as to which poll is which, and how the numbers break down as the polls go. Some are adjusted, some are not. You have mentioned "tainted" poll numbers before, and I am not fully educated on that issue yet. You've mentioned MOE's that are different from what I saw (1.9% is what I saw), but that could be because we are looking at different polls. So, I am neither agreeing nor denying until I become more educated on the subject.

"So you try to deny the math.
But you don't understand the math.
That's obvious."

Since I have not made an attempt to do the math, yet, I don't see how you can say that. In fact, given you are unaware of my background, it's pretty big assumption on your part. Is this kind of assumption representative of your analysis on most topics, or unique to dealing with me?

"I WILL SPELL IT OUT FOR YOU ONCE AGAIN.
I WILL TRY TO BE VERY CLEAR ABOUT THIS.
IF YOU PAY ATTENTION, YOU WILL GET IT.
EVEN IF YOU WON'T ADMIT IT.

1. THERE IS 50/50 PROBABILITY THAT THE FLIP OF FAIR COIN
WOULD TURN UP HEADS."

True. However, the question is, is it accurate to use this analogy for exit polling? Let's see, as we get to:

"2. THERE IS A 50/50 PROBABILITY THAT IN A FAIR ELECTION ANY GIVEN STATE WOULD DEVIATE TO KERRY OR BUSH FROM THE EXIT POLL."

False. If the methodology used has empirically shown probability that is different than 50/50, then your assumption would be false. In the four elections prior to this election, all four early raw exit polls tended to overstate the Democratic vote by up to 8%. That would tend to mean that there is at least the possibility that there are factors inherant in the poll that result in something other than a 50/50 odds assumption.

This is a pretty important issue to me. Can you back up your assumption that the poll should come out 50/50, based on simlar historical data from prior elections? So far, the only response from you that I have seen is that there was a massive, 16 year long Republican conspiracy to manipulate the vote even before there was any electronic voting. That, my friend, doesn't cut it with me as mathematical analysis.

"3. 40 OUT OF 50 STATES(80%) DEVIATED TO BUSH."

See above. This may be true, but I don't know yet.

"4. IT IS ILLUSTATIVE TO COMPUTE THE ODDS."

Sure. It might be useful

"5. THE BINOMIAL DISTRIBUTION FUNCTION IS APPLICABLE, SINCE WE HAVE AN EITHER/OR, YES OR NO. THAT'S CALLED A BINOMIAL EVENT. THERE CAN OCCUR ONLY ONE OF TWO POSSIBILITIES: DEVIATE TO KERRY OR DEVIATE TO BUSH."

Shouldn't we account for a statistical dead heat? I would think a tie would be possible...that the actual result actually matches, or is so close as to be statisitically insignificant, the exit poll result.

"6. THE ODDS ARE 1 IN 356,000 THAT 40 OF 50 STATES WOULD DEVIATE TO BUSH, WHICH THEY DID. BUSH WAS VERY LUCKY.

THE SAME ODDS WOULD APPLY TO KERRY. BUT HE WAS NOT SO LUCKY. IT DIDN'T HAPPEN."

Well, I have to do the math on that of course. But, for now, let's assume your math is correct (though, as I said, may be based on a fundamentally flawed assumption).

"NOW, LETS ASSUME THAT THE EXIT POLL WAS CORRECT.
THAT MEANS IT WAS NOT A FAIR ELECTION."

No. That means it probably was not a fair election...but there is still a chance it was fair, even assuming everything you just said is true. Sure, it's long odds...but then, if you compute the odds that the 2000 election would come down to a few hundred votes, or the current Governor's race in Washington would come down to a few hundred votes also only 4 years later, then you will see that stranger (or at least as strange) things have happened. And that is how we should state it - that the odds are extremely low that it was a fair election...but not claim it was impossible. Again, precision is important.

"BUT YOU AND YOUR BUDDIES REFUSE TO EVEN CONSIDER THAT SCENARIO."

The hell you say? I am fully considering it, nor do I have any buddies here. Well, maybe a few buddies in the science forum...but not in this forum.

"YOU SAY THAT THE VOTE TALLIES ARE CORRECT. PERIOD."

I have never, EVER, said that. I challenge you to show me where I said that.

"OF COURSE YOU SAY THAT."

Giving you the benefit of the doubt...I think you must have me confused with someone else.

"WHAT ELSE CAN YOU SAY?"

I think I just said it :)

[Salomonity]

58. no.

That's the odds of it going to bush 41 times out of 50, IF IT'S A PERFECT SAMPLE.

It wasn't. They likely underpolled rural voters.

[TruthIsAll]

60. YOU WANT A PERFECT SAMPLE? THOSE ARE 51 EXIT POLLS WHICH ARE AS PERFECT

AS YOU CAN GET.
FROM MITOFSKY
THOSE ARE THE EXIT POLLS DOWNLOADED BY SIMON AT 12:22 AM ON NOV.3.

A PERFECT SAMPLE?
YOU WANT A PERFECT SAMPLE?

LOL

[Name removed]

65. Deleted message

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

[Name removed]

69. Deleted message

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

[TruthIsAll]

72. DID YOU READ MY POST? I GOT YOUR ATTENTION, DIDN'T I?

I HOPE THAT YOU LEARN SONMETHING FROM THIS.

YOU ARE GETTING A FREE EDUCATION IN MATH 101 AND IN
READING COMPREHENSION 101.

I DON'T ENJOY THIS.
I WISH WE COULD HAVE MEANINGFUL, INTELLIGENT DISCUSSIONS HERE.
THAT SHOULD BE OUR COMMON GOAL.

BUT AS LONG AS YOU CONTINUE TO MISREPRESENT THE FACTS,
I WILL CALL YOU ON IT.

IF YOU DON'T KNOW MATH,
PLEASE DON'T ARGUE MATHEMATICS.

MAKES SENSE, RIGHT?

[Mistwell]

78. SO THE GOAL WAS TO SCREAM?

I started the thread. You have my attention already. I am responding to you. I have always responded to you. I want to hear what you have to say. There is no reason to type in all caps to get my attention. If you are doing it for my benefit, as you just said, please stop. It's not helping.

[TruthIsAll]

81. if i promise to stop the caps, do you promise to read my posts?

and to discuss the facts that i have presented as clearly as i can.

given that 41 of 50 states deviated to bush, will you stipulate that the the odds i have calculated are correct?

i would appreciate a straight answer, please.
until you do so, how do you expect anyone to take you seriously?

you believe that the exit polls were wrong and the votes were right. fine.

you must at the same yield to the exactitude of the simple mathematics i have presented here.

any further discussion is useless.

and we can just let it go at that.


cheers

tia.

[Mistwell]

92. Stipulate?

I'd be happy to read them. I am rading them. But I am not stipulating that things you post are correct. I will happily take what you say, and check it for accuracy. If I find it accurate, I will say so. If I find it otherwise, I will say that as well.

All my answers to you have been very straight.

I do NOT believe the votes were fine. Nor do I believe the exit polls were necessarily wrong. I do, however, have my doubts about how persuasive the exit poll claim turns out to be. So far, the deeper I look, the less conclusive the issue turns out to be, from my perspective. I'd be happy to be persuaded otherwise. But that does not mean I think Bush won the election. It just means I find other evidence of fraud a lot more persuasive than the exit poll data.

I do not have to yield to "the exactitude of the simple mathematics i...presented here". Not if I find that your mathematical analysis relies on flawed assumptions. If a basic assumption is flawed, then the analysis is flawed as well. I'm always willing to hear your side of it. Are you willing to reconsider anything, if it is presented that you might have flaws in your assumptions?

[TruthIsAll]

61. Mathematics is HYPERBOLE to you. Your post just revealed your

lack of educatioin in rudimentary probability theory.

41 OUT 0F 50 FLIPS TURN UP HEADS?

DO YOU KNOW HOW TO CALCULATE THE ODDS?
DO YOU KNOW WHAT THE BINOMIAL DISTRIBUTION FUNCTION IS?

OBVIOUSLY NOT, BECAUSE IF YOU DID, YOU WOULD NOT HAVE POSTED

"Odds say that 50% should deviate Kerry, and 50% should deviate Bush. In this case, you are saying 20% deviated Kerry, and 80% deviated Bush, for a total variance from the middle by 30%.

I don't know the odds of that happening, or if it is unusual in comparison to other exit polls from other countries and from this country from prior elections"

I HAVE JUST GIVEN YOU THE ODDS.
BUT YOU DON'T UNDERSTAND THE MATH.
DON'T GO TO LAS VEGAS.
YOU WILL LOSE YOUR SHIRT.

[Name removed]

70. Deleted message

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

[electropop]

2. Don't know which exit polls he's referring to

But it appears he is operating on the post-corruption data, long ago discredited. His figures bear no resemblance to the more generally accepted data which people grabbed in screen shots from TV and from the CNN site. You can find it at

http://joeorgren.com/election_fraud.htm

[TruthIsAll]

77. The CNN data is corrupted. The data downloaded by Simon is not.

.

[TruthIsAll]

79. Here are the state exit polls downloaded by Simon.

This is a deviation/probability analysis based on the Simon
download.

Size refers to the exit poll sample size.
Exit and vote percentages are Kerry's.

I have utilized 2-party exit poll and vote numbers. Third
party voters are allocated to Kerry and Bush in proportion to
the Kerry/Bush corresponding split in the poll and vote.

Prob is the probability of the deviation using the Exit poll%,
vote% and standard deviation as input to Normal Distribution
function. 

MOE for each state poll = 1/sqrt(Size).
StDev = standard deviation = MOE/1.96.

State	Size	Exit	Vote	Diff	StDev	MoE	Prob  	Dev>2%?	Favor
DE	770	58.50%	53.54%	-4.96%	1.80%	3.53%	0.29	yes	Bush
NH	1849	55.40%	50.51%	-4.89%	1.16%	2.28%	0.00	yes	Bush
VT	685	65.00%	60.20%	-4.80%	1.91%	3.74%	0.60	yes	Bush
SC	1735	46.00%	41.41%	-4.59%	1.20%	2.35%	0.01	yes	Bush
NE	785	36.76%	32.32%	-4.44%	1.78%	3.50%	0.64	yes	Bush

AK	910	40.50%	36.08%	-4.42%	1.66%	3.25%	0.38	yes	Bush
AL	730	41.00%	37.00%	-4.00%	1.85%	3.63%	1.53	yes	Bush
NC	2167	48.00%	44.00%	-4.00%	1.07%	2.11%	0.01	yes	Bush
NY	1452	63.00%	59.18%	-3.82%	1.31%	2.57%	0.18	yes	Bush
CT	872	58.50%	55.10%	-3.40%	1.69%	3.32%	2.24	yes	Bush

RI	809	64.00%	60.61%	-3.39%	1.76%	3.45%	2.68	yes	Bush
MA	889	66.00%	62.63%	-3.37%	1.68%	3.29%	2.21	yes	Bush
PA	1930	54.35%	51.00%	-3.35%	1.14%	2.23%	0.16	yes	Bush
MS	798	43.26%	40.00%	-3.26%	1.77%	3.47%	3.29	yes	Bush
OH	1963	52.10%	49.00%	-3.10%	1.13%	2.21%	0.30	yes	Bush

FL	2846	50.51%	47.47%	-3.03%	0.94%	1.84%	0.06	yes	Bush
MN	2178	54.50%	51.52%	-2.98%	1.07%	2.10%	0.27	yes	Bush
UT	798	30.50%	27.55%	-2.95%	1.77%	3.47%	4.78	yes	Bush
ID	559	33.50%	30.61%	-2.89%	2.11%	4.14%	8.60	yes	Bush
AZ	1859	47.00%	44.44%	-2.56%	1.16%	2.27%	1.38	yes	Bush

VA	1000	47.96%	45.45%	-2.50%	1.58%	3.10%	5.66	yes	Bush
LA	1669	44.50%	42.42%	-2.08%	1.22%	2.40%	4.49	yes	Bush
IL	1392	57.00%	55.00%	-2.00%	1.34%	2.63%	6.78	yes	Bush
WI	2223	52.50%	50.51%	-1.99%	1.06%	2.08%	3.00		Bush
WV	1722	45.25%	43.43%	-1.82%	1.20%	2.36%	6.54		Bush

NM	1951	51.30%	49.49%	-1.81%	1.13%	2.22%	5.54		Bush
CO	2515	49.10%	47.47%	-1.63%	1.00%	1.95%	5.15		Bush
IN	926	41.00%	39.39%	-1.61%	1.64%	3.22%	16.42		Bush
GA	1536	43.00%	41.41%	-1.59%	1.28%	2.50%	10.69		Bush
MO	2158	47.50%	46.00%	-1.50%	1.08%	2.11%	8.17		Bush

NJ	1520	55.00%	53.54%	-1.46%	1.28%	2.51%	12.67		Bush
WA	2123	54.95%	53.54%	-1.41%	1.09%	2.13%	9.70		Bush
IA	2502	50.65%	49.49%	-1.15%	1.00%	1.96%	12.41		Bush
AR	1402	46.60%	45.45%	-1.15%	1.34%	2.62%	19.55		Bush
KY	1034	41.00%	40.00%	-1.00%	1.55%	3.05%	26.01		Bush

OK	1539	35.00%	34.00%	-1.00%	1.27%	2.50%	21.63		Bush
MI	2452	52.50%	51.52%	-0.98%	1.01%	1.98%	16.47		Bush
NV	2116	49.35%	48.48%	-0.87%	1.09%	2.13%	21.29		Bush
ME	1968	54.75%	54.08%	-0.66%	1.13%	2.21%	27.80		Bush
MD	1000	57.00%	56.57%	-0.43%	1.58%	3.10%	39.18		Bush

DC	795	91.00%	90.91%	-0.09%	1.77%	3.48%	47.96		Bush
MT	640	39.76%	39.80%	0.04%	1.98%	3.87%	50.72		Kerry
OR	1064	51.20%	52.00%	0.80%	1.53%	3.00%	69.91		Kerry
HI	499	53.30%	54.55%	1.25%	2.24%	4.39%	71.10		Kerry
TX	1671	37.00%	38.38%	1.38%	1.22%	2.40%	87.10		Kerry

TN	1774	41.50%	43.00%	1.50%	1.19%	2.33%	89.68		Kerry
CA	1919	54.00%	55.56%	1.56%	1.14%	2.24%	91.35		Kerry
SD	1495	37.76%	39.39%	1.63%	1.29%	2.53%	89.65		Kerry
ND	649	34.00%	36.36%	2.36%	1.96%	3.85%	88.58	yes	Kerry
KS	654	35.00%	37.37%	2.37%	1.96%	3.83%	88.76	yes	Kerry

Avg	1450	49.18%	47.38%	-1.80%	1.42%	2.79%	21.67		Bush
Med	1507.5	49.23%	47.47%	-1.81%	1.29%	2.52%	6.66		Bush
				

[mdhunter]

83. Question.

Are the MoEs listed here as supplied by the polling organization?

If so, why do the most common analyses on this board continue to use 2% for determining whether or not each state poll was innaccurate? Why not use the MoE supplied? In that case it appears that the exit poll vote was outside the MoE in favor of Bush on 14 occasions, and outside the MoE in favor of Kerry not once.

[TruthIsAll]

84. Those are MOE's calculated by yours truly based on sample size.

The formula is 1/sqrt(N) where N is the sample size. It is a well-known formula.

I have used the MOE as the criteria in prior analysis.
Bush exceeds the MOE in 16 states.
The probability of that was calculated as 1 in 13.5 trillion.

This analysis was a follow-up of the original to show that since exit polls are more accurate than pre-election polls (the MOE's are given for pre-election polls) than if we assume an exit poll MOE of 2% (this is still very conservative), then Bush exceeds the 2% exit poll MOE in 23 states.

The probability of this occurrence is less than 30 zeros to the right of the decimal point, which Excel cannot display.

If you want to assume the pre-0election MOE as shown, rather than the 2%, be my guest. It's right there - 16 states.

[Name removed]

89. Deleted message

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

[euler]

97. Yes, why ?

[FogerRox]

118. yes TIA 23 states, I saw your earlier post and counted the 23 states

back in NOv. Historically: Exit polling with 2 white guys running is usually 1% or maybe 1.5% off--from the results. If you had a white male and black female running then you might these kind of numbers---5% --8% 0r even 10%. But that didnt happen 2 white guys were running.

An Incumbant gets his May/June Approval rating Nov. Bush was at 44%.
Kerry would get 56%--take 1% away for 3rd party votes and call it 55% For Kerry. With about 112 million votes cast--thats what we were told-- Kerry wins by about 12 million votes. This is a historical precident. Peace time or war time---Bush was not going to win.

-Kerry killed Bush in the debates.
-The last week of the campaign most polls showed the Kerry momentum and indicated a Kerry win by 5% to 10%.
-ZOgby showed Kerry winning 317 to 221 in the EC.

[mdhunter]

127. For what it's worth, I think you're using an inappropriate technique

to determine significance.

What we have with these data are essentially dichotomous categorical variables. The way to determine single sample significance is via the chi square test, given that we're not actually dealing with continuous variables.

Once that's done you still may be able to use the proceedures you've been using.

I haven't run all the numbers using X^2, nor do I plan to, but I spot checked a few and determined that not all the states you deem to be significant deviations actually are.

For example, Louisianna. The test statistic returns a value of 2.956, well below the critical value of 3.84.

My justification for using X^2 is as follows. Clearly we have a single sample, and the variable in question is a vote for President. For our purposes, this is a vote for either Kerry or Bush. Since the data you posted earlier has Kerry's totals, we have a dichotomous outcome. A vote for Kerry, and a vote for someone else.

Multiplying Kerry's % of the exit by N gives the Observed total for Kerry. Multiplying 1-Kerry's % of the exit by N gives the Observed total for Other/Bush. Multiplying Kerry's % of actual votes by N gives the Expected total for Kerry. Multiplying 1-Kerry's % of actual votes by N gives the Expected total for Other/Bush.

I think that to be the proper way, and I think it reduces by a few the number of statistically significant deviations.

[Gman]

This may be a subtle attempt to discredit the stories on other exit polls

by showing that some polls were wrong in favoring Bush. The bottom line remains in 2004, as it did in 2000, that the exit polls showing Gore winning FL and Kerry winning in OH and other states were right on the money.

[euler]

10. Please provide a link where you saw this proven....

"exit polls showing Gore winning FL and Kerry winning in OH and other states were right on the money"

Thanks.

[Gman]

62. Look it up. It's there

Zogby, for one, called FL for Gore correctly.

[punpirate]

3. Even exit polls have a margin of error...

... so it depends on the sample. Larger the sample, smaller the error.

This likely occurred because of smaller sample sizes in those areas. In the case of NM, I think Bush won with a margin that was right around the MOE.

[Goldeneye]

4. That will happen. Statistics almost never nails it exactly on the head.

If everything were fair and balanced about half the time the deviation between the exit poll and the actuall tally would break for Kerry and the other half of the time they'd break for Bush. In this election the vast majority of states, including the swing states broke for Bush. They did so by a statistically improbably margin. Even worse than that, many of the swing states final tallies were outside the margin of error, and all of them were beyond the margin of error in favor of Bush.

[Mistwell]

5. NM and NJ

I thought both NM and NJ *WERE* battleground states?!

[Goldeneye]

7. NJ

and NM are swing states. The fact remains that the states discrepencies largely favored Bush (and by that I mean by a statistically improbable margin), like I said in the original post.

Think of flipping a coin 13 times and having it come up heads only twice...

[Mistwell]

8. Just trying to get at the truth

I'm just trying to get at the truth. People get upset when I question this stuff, but believe me when I say my questioning is nothing compared to someone NOT on our side.

I've seen it posted here DOZENS of times that ALL the exit poll data showed Kerry getting more votes than the actual results showed. That was wrong, if this data I have posted here is correct. I've seen it poster here dozens more times that all exit poll data in battleground states showed Kerry getting more votes than the actual resultes showed. Well, if this data is correct, that statement as well is wrong (unless you use a very selective definition of battleground states meant to only prove your thesis).

So, I want to hear the real answers. If the exit poll data is "compelling but not conclusive", I am fine with that. Particularly since the evidence of ACTUAL fraud is more compelling to me in general than exit poll evidence. However, I want to know how powerful the exit poll claim really is. If it is weaker than I've heard, but still useful, I want to know that. It's better to have our facts straight, and know where our weaknesses are, than to pretend our case is stronger than it is.

[spooked911]

11. we really don't know how conclusive the exit poll data is because

Mitofsky hasn't released his raw numbers or his methodologies.

For instance--if we could see that an exit poll from a specific precinct was clearly different from ballots that were hand-counted, this would show the exit poll was wrong.

Right now, the exit polls look very compelling because they are so far off in many cases from the vote tally-- well outside the margin of error. But unfortunately they are not proof of fraud yet.

[Mistwell]

43. I agree

I agree 100% with you. We need the raw data to truly come to a conclusion about the exit polls.

[Goldeneye]

12. All right I'll try a different explanation

I've seen it posted here DOZENS of times that ALL the exit poll data showed Kerry getting more votes than the actual results showed. That was wrong, if this data I have posted here is correct.

All right, I don't remember reading that there were no swings in Kerry's direction. It don't know what posts you're refering to, but if there were posts that said that they would be wrong. I think the real number of swing states was 13. 11 of them showed swings towards Bush. The probability of 11 of 13 states actually shifting in Bush's favor, assuming there was no fraud, is very small. Something like 1 in 1000. I don't remember the exact numbers, but bear with me. Second, a few of those states shifted outside the margin of error. Exit polls are very accurate. As I'm sure you have read, they are based on people who have already voted. These pollsters have been doing this since the 1970s and have only recently had trouble with their polling.

As for the exit polls being inaccurate, its possible, but anything is possible. The only reason I have heard is that they over sampled women. But the final poll shows the same percent of female voters as the exit polls used. If you know of anyother reasons they've given, I'd like to hear them.

i think the poster you may be refering to is TIA.

[Lurker321]

18. Goldeneye - "exit polls are very accurate" is a myth

it really is.

http://www.washingtonmonthly.com/archives/individual/2004_11/005178.php

Four previous elections' exit polls:

1988: Dukakis, 50.3 Bush, 49.7       wrong by: 8.3%
1992: Clinton, 46   Bush, 33.2       wrong by: 7.2%
1996: Clinton, 52.2 Dole, 37.5       wrong by: 6.2% 
2000: Gore,    48.5 Bush, 46.2       wrong by: 1.8%

Note: this is from raw exit poll data. When it was massaged, weighted and adjusted, the exit polls were a lot more accurate.

Want to try exit polls in Ukraine? Here are the three exit polls from the last election (that is, the re-vote)
(link: http://www.myrtlebeachonline.com/mld/myrtlebeachonline/10499250.htm)

1.  58.1%  to   38.4%   MOE=2
2.  56.5%  to   41.3%   no MOE given
3.  56%    to   41%     MOE=2

Official results: http://tinyurl.com/3ufxp

    51.99% to   44.2%    

As you see, all three exit polls were wrong, way outside of margin of error.

[euler]

28. I just got around to reading the Lakoff book last night -

don’t think of an elephant!

I think I've discovered why you, me and others can't get through to anyone in the forum when we try to point out problems with exit poll theories:

We beleive that "if we just tell people the facts, since people are basically rational beings, they’ll all reach the right conclusions. But we know from cognitive science that people do not think like that. People think in frames....To be accepted, the truth must fit people’s frames. If the facts do not fit a frame, the frame stays and the facts bounce off."

[TruthIsAll]

63. You have just proven our case. Exit polls in this country

always have the democrats doing better than their vote tallies.

Think.
Why is that?

Why do the Dems always lose from the exit polls to the votes?
Don't you find that suspicious?

Or is it just another coincidence, theorist?

[Lurker321]

64. First of all, glad you finally admitted

that exit polls are NOT always accurate, as you keep claiming, and, in fact, raw presidential exit polls have always been INaccurate. See, we're making progress.

As for the answer to your question, I gave it before but you, as is your custom, ignored my answers completely since they do not fit your biases. Here they are again.

There can be two answers to your question:

1. There has been a massive conspiracy that fraudulently falsified (to the tune of up to 8% in 1988) the final election results from the "correct" raw exit polls in every election since 1988.

2. The exit polls have a systemic bias in them toward Democrats in federal elections.

Which one are you picking?

[TruthIsAll]

67. Both. They are accurate. But a combination of spoilage AND fraud

in the VOTE COUNTING causes the discrepancies.

And the Repukes know that the spoilage occurs predominantly in minority precincts. So they have a great head start. Read Palast.

Thought you got me?

Once again, YOU are the one who inadvertantly confirmed that throufg a conbination of fraud and spoilage, vote tallies deviate to the Repukes.

Thank you for the talking point.

[Lurker321]

74. So your theory is that Republicans have been

fraudulently skewing the elections to the tune of up to 8% of the vote (let's see - that's 5-6 million votes) in EVERY election. And even with such massive fraud they stii have never been caught - even though they have been doing this for at least 16 years.

I am surprised this is the first time I learn of this theory of yours. Funny how you never even mention that in CAPITAL LETTERS in any one of your posts? Why are you hiding this theory of yours?

[euler]

102. He is saying that, but it can't proven....

....unless you are TIA. He has mastered the proof by proclamation technique.

His 'red-shift' calculations have always been baseless, but now he admits the baselessness.

[euler]

101. You're not making sense again.

[davidgmills]

71. Not a Democratic bias at all

Freeman calls it systemic and chronic proof of democratic undercount. The polls were right; the count was wrong.

[Mistwell]

73. Well, you tell me which is more likely

Which is more likely:

1) For the past 16 years there has been a vast conspiracy by Republicans to fraudulently alter votes. This conspiracy was in effect before touch screen voting machines were around...indeed at a time when all voting was by paper ballot (so it cannot be one company doing it electronically with software). Despite the monumental illegal and criminally risky effort of these Republicans in thousands of precincts over those 16 years, none of whom have ever spoken about it in public to this day, the Republicans STILL lost 50% of the time anyway, and essentially tied in one of the 2 remaining races.

2) The methodology for exit polling is flawed in a manner that tends to favor Democrats over Republicans.

:tinfoilhat:

Really, until that Washington Monthly article, I was leaning towards thinking the exit polls were evidence (though not conclusive) of fraud. Now...I guess I am back to thinking the exit polls are useless for predicting fraud in close races - and our best evidence is, as always, the incidents of actual fraud witnessed by people (of which there is plenty).

[Yupster]

99. Ummmm

because Republicans include a lot of grouchy old white men who brush off the exit pollsters?

[euler]

100. You knocked a big hole in your work.

Honestly, I'm shocked you acknowledge the democratic bias in past presidential exit polls. I'm shocked because you can't credibly make the case that there has been systemtic fraud across all these presidential elections. This means you cannot credibly continue arguing your 1 in a 45 bazillion red shift thesis. Did you think about that before making this confession ?

[Nederland]

105. Thank you

I'm glad to see that you have are finally addressing the real issue here. Its never been about the math, its always been about the validity of the numbers that get plugged in. The crux of the issue is whether or not we can trust exit polls to be accurate.

So far, I've never heard anyone explain why we should trust them, but many have explained why we shouldn't.

[emcguffie]

32. are you using the right data?

Sure you don't have the "corrected" exit polls?

[Mistwell]

44. I mostly agree

I don't know the odds of 11 of 13 swinging one way, rather than 6.5 of 13 swinging that way (which would be an exactly equal swing). Perhaps someone could calculate it?

As for margin of error...I really don't know. I don't have sufficient data to know if any prior EARLY exit polls were off by more than the margin of error. I know they have been accurate for a final predictions in races that were not particularly close...but I really don't know for MOE issues concerning close elections.

I do know that the Ukraine exit poll for the SECOND election (which was one of the most watched and counted elections I know of) DID vary beyond the margin of error, though it was accurate for the final prediction of who won.

As for what went wrong with this one (if anything), I have heard the following factors: Too few western states includes; too many women included; too many Democrats included; too many early-day voters included. I have no idea if any of those were really a factor, though it is something I would like to know more about.

As for TIA...yeah, I've tried to discuss these issues with him. I appreciate his work, and think his heart is in the right place. However, I don't think he is interested in discussing any potential claims that the exit poll itself was flawed. That's fine - everyone does things their own way. But it isn't as useful for me, if I am trying to really pin down the exit poll portion of this issue.

[TruthIsAll]

126. I have never said exit polls were perfect. But we know the sample

size and we know the margin of error so we can apply classic probability and statistical theory to calculate the odds that the margin of error would be exceeded due to chance.

But let's forget about MOE for the time being. Instead, let's focus on the FACT that 41 out of 50 states deviated in favor of Bush.

The odds are 50/50 that any ONE state would deviate,if we assume a fair election. Just like the odds are 50/50 that if we flip a fair coin, it will come up heads.

So far, so good.
Now, lets carry this logic to the 50 states.

The Binomial Probability Model tells us that for 41 states to deviate in favor of Bush, the odds are 1 in 356,000.

So, what does this all mean?
It sure does NOT mean that the exit polls were wrong.

Mitofsky/Edison put a substantial effort into the design of these polls, based on their 25 years of experience, to get it right.
So it's unlikely they would get it wrong.

That leaves us with this:
It is more likely, based on this 41 state anomaly, that there was something wrong with the VOTE count. Not the poll. The count.
It may very well be that the polls were RIGHT.

They do not have to be EXACTLY right. Just right enough. If 28 states deviated to Bush, this would not be an issue.

BUT 41 STATES DEVIATED. THAT MAKES IT A VERY BIG ISSUE.
SOMETHING WAS WRONG. VERY WRONG.

Why is it so difficult to entertain THAT possibility?

Especially when the odds are 356,000 to one.

Not 1 out of 10. That's possible.
Not 1 out of 100. That's unlikely, but possible.
Not 1 out of 1,000. That really gets us wondering.
Not 1 out of 10,000. That makes us VERY suspicious.
Not 1 out of 100,000. We KNOW something is wrong.

No. It's 1 out of 356,000. We KNOW something is VERY, VERY wrong.

[euler]

13. I want to know how powerful the exit poll claim really is

Please read every thing written at these links including comments. But, don't take the authors at their word - some are liberal, some conservative, some unknown affiliation, and one is a Christian (some here at DU consider being a Christian automatic disqualification). Instead, use the content in these papers, as well as the sources sited as input into your own Google searches, to help you find yet more information about this exit poll.

I found that the exit polls claims made here on DU are very weak, both because of the data being used, and the incorrect statistics techniques applied. But that's just me, you can draw your own conclusions.

Basics, you may not need this link:

http://www.exit-poll.net/faq.html#a5

Methodology:

http://www.exit-poll.net/election-night/MethodsStatementStateGeneric.pdf

Purpose for exit polls in the US:

http://www.ou.edu/policom/1302_2003_spring/kosicki.htm

Sampling error:

http://www.leemizell.com/freedocs/election_free.pdf

Search for the paragraph starting with: "Exit polls have an undeserved aura of infallibility"

http://hurryupharry.bloghouse.net/archives/2004/11/24/arsenic_and_old_lace.php

Scroll for the paragraph starting with: "Now to the central issue: the claim that exit polls, which never lie, showed Kerry winning."

http://www.alternet.org/story/20934

http://www.washingtonpost.com/wp-dyn/articles/A64906-2004Nov20.html
http://www.mydd.com/story/2004/11/30/171641/64
http://www.cpsa-acsp.ca/papers-2004/Docherty%20and%20others.pdf

http://www.mysterypollster.com/main/2004/12/have_the_exit_p.html
http://www.mysterypollster.com/main/2004/09/just_how_diverg.html
http://www.mysterypollster.com/main/2004/11/the_difference_.html
http://www.mysterypollster.com/main/2004/12/the_raw_exit_po.html
http://www.mysterypollster.com/main/2004/12/what_about_thos.html
http://www.mysterypollster.com/main/2004/12/what_is_the_sam.html

http://stones-cry-out.blogspot.com/2005/01/simon-and-baiman-on-exit-polls.html
http://stones-cry-out.blogspot.com/2005/01/simon-and-baiman-continued.html
http://stones-cry-out.blogspot.com/2004/11/what-went-wrong-with-exit-polling.html
http://stones-cry-out.blogspot.com/2004/11/exit-polling-2004-significantly.html
http://stones-cry-out.blogspot.com/2004/12/exit-poll-2004-update.html

[euler]

14. On more blurb

This is just one example of why exit polls may not be as accurate as we would like:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=97506&mesg_id=97575&page=

[Joe Chi Minh]

15. No, no, no, no, no.

Statistical evidence, in principle, can amount to effective certainty, so that "conclusive" may, as in this case (i.e. in the context of a total preponderance of statistical probabilities of fraud ranging from extreme probability up to infinitely high probability), be closer to the mark than "compelling" (admittedly, itself, binding on a reason mind).

"Evidence of ACTUAL fraud more compelling to me in general than exit poll evidence"(?) Think "grassy knoll". I hope we can help you get closer to the truth.


I was just wondering whether to edit my introductory sequence of negatives, when an African American woman on the box started singing "No, no, no, no, no!", so I'll go with the flow.

[Igel]

17. Statistical evidence can amount to

effective certainly when the assumptions used in taking measurements and in setting up the equations are matched by reality.

I've seen far too much "science" in peer-reviewed journals tossed out because later measurements showed that the assumptions and reality were at odds. The junked research had results that the stats said were highly significant, but still ... the results were wrong.

When I read a stat-based article, I first skim the conclusions in the abstract, then flip over to check the section on methodology, and third I check whether the stats seem properly applied. If the latter two seem reasonable after a few minutes' examination, I bother to read the article; otherwise I don't waste my time.

If you don't know the assumptions underlying the measurements and the statistics, there's no argument possible.

[Joe Chi Minh]

31. Very astute, igil,

if I may say so. Speed reading taken to a different level.

Asssumptions are crucial, aren't they. They're almost always the weakness of the trolls' posts, here. Though more than likely to be deliberate. It's mostly in their most basic, unquantifiable world-view assumptions that they are unspeakably crass and sorry specimens.

[Chi]

40. I gotta ask 8)

Heya KCabotDullesMarxIII
Ya got my curiosity up.
What is the end to this?
"Sometimes, there is simply truth and falsehood, so that a person who pretends to portray a balanced".

[Joe Chi Minh]

45. Absolutely.

Chesterton made the crucial point that the purpose of an open mind is to close on the truth.

It is interesting, isn't it, that Christ spoke so baldly. No sweet-talking, this is the way it is. Well, admittedly he should know. But it absolutely incensed the worldling scribes and Pharisees that he didn't sweet talk them or give any hint that he was interested in what they had to say. "Give and take" is the way we all get along with each other in the normal order of things.

In this context, the psalmists frequent references are very illuminating. They frequently refer to the smarminess of the evil man. "His words are softer than butter, but war is in his heart", and many similar words. Again, "...his words are sweeter than honey, but they are naked swords". How often that hits me when I read the posts of the many trolls, and of course, their pretence of sweet reason, genuinely enquiring minds, the utmost good faith, and how it makes me see red.

Although your MSM are beneath contempt as undiluted organs of propaganda, and ours in the UK, not that much better, all news services in the West and doubtless elsewhere push the line of their respective Establishments. The further to the right they are on the political spectrum, the less their purpose is to inform. What turkeys are going to vote for Christmas if they know the score? (I could be wrong about Scandinavia, which of course in any case, has generally been left-leaning and socially responsible for many years).

As someone pointed out on one of these threads, the contortions our MSM sometimes have to go through, to *purportedly* give a balanced view, but actually pander to the powers that be, actually amount to disinformation.

Another of the ruses that the press in particular deploys is to "run with the hare and hunt with the hounds". Running with the hare of course is one of their primary functions, but making money for their louche owners would even trump that, since if the rag didn't sell, its political influence would become academic.

Consequently, the Daily Mail, for example, in this country, realising when the people are peed off with the right wing, because they've finally rumbled that they've been comprehensively done, will write a less than complimentary piece on Thatcher or her policies. Usually, though, in the same paper, they will write an item praising her "greatness". No don't laugh...

There was an extraordinarily well-known, all-girl pop group, who few few people had heard of who became an overnight sensation. Guess why. One of the girls said she admired Thatcher. Normally, you'd have to say she sold her soul to the devil with those few words, but the fact is, of course, she would be as politically naive as she was immature in terms of morality. Amoral rather than moral. The zeitgeist of recent generations.

The trolls on here are wont to use both: a pretence of balance, good faith and honest opinion; and running with the hare and hunting with the hounds - the hounds here often being represented by the politically naive and impatient among ourselves. So a third ploy of theirs is "divide and rule". They know that the only way to get at us is by honeyed words, so don't expect the operatives to sound anything but charming in their posts - at least the vast majority of the time.

I'm sure you now all this, chi, but you seemed to want my take on it.

[Chi]

50. Wow!!

"I'm sure you now all this, chi, but you seemed to want my take on it."

Well...actually...I wasn't looking for your take on the Dark Lord and his EvilEmpire (including the minions). But none the less, I did enjoy it, and agree with it!
Thanks for your response, sorry to impose on your time.

I was, though, truly curious about the end of that sentence in your bio, it seemed cut off. 8)

It's good to meet others solely focused on truth, regardless of where it might lead.

<bow>

[Joe Chi Minh]

55. Sorry, Chi.

I've just rumbled what you were on about, (though I still don't understand the 8))!

It looks like I thought I'd quoted a tag line in my profile that I'd been impressed with. Actually, I think I referred to it in my post above. The absurd and dishonest contortions the MSM sometimes go through, in order to nullify an unambiguously valid point amounts to deliberate disinformation - purportedly, of course, to provide "fair balance". Incidentally, I envy like mad the technical understanding you blokes have of the figures and all the psephological jargon.

[Chi]

57. Not a problem

"(though I still don't understand the 8))!"

8-) same as 8)
Heh, was a lazy attempt at a smiley face.

[TruthIsAll]

66. Let me say that I love your post. And I hope that you

will continue to provide us with intelligent, logical and sane analysis.

We need more of that.

[Igel]

82. It's not all that astute.

It's triage, and not uncommon. Lots of stuff in print, not all reviewers catch everything, and I have limited time. If I read everything from start to finish that bears on my work, whether the methodology and stats are sound or not, I'd never get my own stuff done. In all fairness, sometimes the methodogy's brilliant, but the stats suck, or I learn something about statistics that I didn't know but the methodology is horrible. Usually both are sound, or sound enough.

Had a friend that was a stats TA before moving into psycholinguistics. There was a fairly well-received article, conclusively arguing a controversial point and nailing it with the stats. Nobody could find a serious error in the methodology, which is where most of the problems usually turn up. A colleague tried to replicate the article's stats, and couldn't. The ex-stats-TA looked it over, and realized that the "conclusively argued" article didn't use the right number of degrees of freedom ... A brilliant article. It just didn't show anything at all.

[Yupster]

98. My thesis on the exit polls mistwell

Is that Republican voters are as a rule angrier and less friendly than Democratic voters, and are more likely to tell the exit poller to go screw himself when he/she asks them to take the poll. This comes from a large distrust of the media which has been hammered into them by the likes of Rush Limbaugh for more than a decade.

That is why year after year, election after election, Republicans always score lower on exit polls.

That's my thesis. I don't know if it's correct, but I think it's a lot more likely than some uber-vote-stealing conspiracy taking place all over the country.

[TruthIsAll]

41. Flip a coin 50 times. Comes up heads 41 times.The Odds: 1 out of 356,000.

Of 50 states, 41 flipped to Bush (his reported tally exceed his exit poll).

Here are the odds for various N states (heads)

N Prob at least N heads Odds 1 out of
25 0.55613758632961 2
28 0.23994383084916 4
30 0.10131937553227 10
35 0.00330022398341 303
38 0.00015293200080 6,539
40 0.00001193066584 83,818

41 0.00000280705005 356,246

42 0.00000058177790 1,718,869
44 0.00000001621870 61,657,214
46 0.00000000022309 4,482,522,805
48 0.00000000000113 882,712,588,665
49 0.00000000000004 22,295,047,660,250

50 0.00000000000000 #DIV/0!

prob = 1- binomdist(41, 50, 0.5, true) = 0.00000280705005

The odds of at least 16 states deviating beyond the MOE for Bush is almost the same ase getting 49 HEADS out of 50 flips.

Odds of at least 49 heads = 1 out of 22.3 trillion
Odds of at least 16 states beyond the MOE = 1 out of 13.5 trillion.

[Chi]

42. Dang good timing.....I have a question for you.

Heya TIA <waves>
I have a question for you.
If I were to flip a coin 50 times....
A) What is the chances it would come up tails 11 times straight.

B) what is the chance it would come up 11 times straight in a particular position (say from the 11th flip through the 22nd flip).

BTW, I'm using 11 to represent the Battleground states (you prolly already figured that 8-) )

Thanx in Advance
Chi

PS If this takes more than a couple seconds, don't bother just for me.

[TruthIsAll]

47. Part B I can do quick. Part A is harder.

The odds of a "run" of 11 heads (or tails) at any point is

1 out of 2048 =1/2*1/2*1/2*1/2*1/2*1/2*1/2*1/2*1/2*1/2*1/2

That's like the odds of a .500 baseball team winning 11 games in a row at any point during the first 50 games of the season.

To win the first 11 games, its 1 out of 2048. In other words, once every 2048 years.

[Chi]

53. Thank you very much!

Thanx TIA
It will be useful in discussions where people refuse to look at state breakdowns, and only regurgitate national averages.

Thanks again
Chi

[mdhunter]

80. If winning baseball games were a random event based on chance

which it's not.

[TruthIsAll]

87. Not true. Each team has a built-in probability of winning.

Good teams have a 60% probability of winning, bad teams have a
40% probability. Average teams have a 50% probability.

We can calculate the probability of n-consecutive wins based
on these probabilities.

A good team has a much better chance of winning 11 in a row
than a poor team. That is just common sense.

The binomial distribution is the appropriate way to calculate
the probability.

Here are the odds of a good, average and poor team winnin 11
in a row.

Team quality      Prob 11 in a row		or 1 out of
Good (60%)	        0.00362797		276
			
Average(50%)	        0.00048828		2,048
			
Poor (40%)   	        0.00004194		23,842

	        =BINOMDIST(11,11,0.6, FALSE)


Here is the Excel function description:

BINOMDIST
Returns the individual term binomial distribution probability.
Use BINOMDIST in problems with a fixed number of tests or
trials, when the outcomes of any trial are only success or
failure, when trials are independent, and when the probability
of success is constant throughout the experiment. For example,
BINOMDIST can calculate the probability that two of the next
three babies born are male.

BINOMDIST(number_s,trials,probability_s,cumulative)

Number_s
is the number of successes in trials.

Trials
 is the number of independent trials.

Probability_s
 is the probability of success on each trial.

Cumulative
 is a logical value that determines the form of the function.
If cumulative is TRUE, then BINOMDIST returns the cumulative
distribution function, which is the probability that there are
at most number_s successes; if FALSE, it returns the
probability mass function, which is the probability that there
are number_s successes.

Number_s and trials are truncated to integers

[mdhunter]

106. Understood.

You're numbers were correct for a coin flip, but then you said that it was the same as baseball team's probability of winning 11 games in a row.

Which, again, it's not because baseball teams have an inherent probability of winning that it unlikely to be .50. A good team may win 70% of its games, and it may not be at all noteworthy that they win 10 or so games in a row.

That's all I'm saying, flipping coins is not like playing baseball, and running elections may be more like playing baseball than flipping coins.

[Yupster]

103. I think you're wrong TIA

The odds of flipping a coin and getting heads 49 of 50 times is not a statistical question.

If that happens, put away your calculator. It's not a statistical question.

One of two things is happening.

1. The coin is not a good one (exit poll not accurate)or
2. The person is not recording the flips correctly (vote tally not accurate)

To try to test which it is 1 or 2, you can watch a tape of five flips and see how they were recorded. If all five flips came up heads and they were recorded heads, you can probably conclude the problem wasn't in the recording.

This is what happened when they looked at New Hampshire where the exit poll showed Kerry running 12 % better than his vote tally. Were the votes tallied right?

Yes, the recount showed the tally was right. It was the exit poll that was wrong which didn't surprise anyone as everyone knew New Hampshire was not going to be a Kerry double digit win.

Anyway, that leaves choice 1 as the likely culprit. There's some reason why the exit polls are not measuring the voting correctly. It's happening election after election in state after state.

It seems that's where the energy should be put - trying to figure out why exit polls consistently overestimate Democratic strength.

[TruthIsAll]

107. Self-delete

.

[TruthIsAll]

108. You have just proved my point without realizing it.

You have just proved my point without realizing it.

You agree that 49 of 50 turning up heads is impossible.
Well, not quite, it's 22 trillion to 1 odds against.

So we must conclude something is wrong with the coin. It could not be a fair coin and come up 49 heads.

Bush exceeding the MOE in 16 states is impossible.
Well, not quite, it's 13.5 trillion to 1.

So we must conclude something is wrong with the election. It could not have been a fair count.

So maybe now you see why circumstantial evidence based on statistical probabilities is used in a court of law, even if it is kept hidden from the court of public opinion.


Thanks for proving my point.

[TruthIsAll]

110. You have just proved my point without realizing it.

First, only YOU would conclude that 5 flips would prove anything. Those are 1 out of 32 odds.

You have just proved my point without realizing it.

I'm talking about odds of 356,000 to 1 against Bush gaining in 41
states.

I'm talking about odds of 13.5 TRILLION to 1 against Bush gaining beyond the individual exit poll MOE's in 16 states.

Second, the recount in NH was not a complete hand recount.

Third, exit polls DO NOT overestimate democratic strength. That is your interpretation. The vote UNDERCOUNTS democratic strength because of any of the following.

1. Spoilage of the votes. Such as in punchcards, like the 65,000 undervotes and 110,000 overvotes in FL 2000. And the ones in Ohio.
This may have been due a combination of machine malfunction OR deliberate overpunching. Oh, don't be shocked.

2. Manipulation of the touchscreen software. Plenty of anecdotal evidence confirms this: 86 of 88 pressed Kerry, and came up Bush.

3. Manipulation of the software for counting the optical scanners.

[mgr]

113. Oh jesus, here we go again

A simple back of the envelope calculation would demonstrate this to be inaccurate.

This issue with exit polls compared to the national election is a matter of accuracy and precision. For the national exit poll to be accurate the order of magnitude of fraud or ballot counting irregularities would be on the level of 1,000,000s.

Ballot spoilage in predominantly black areas would result in irregularites on the order of 10,000s to 100,000. It is insufficient.

Switching of touchscreens are not reported in all precincts, one cannot argue based upon the existential a universal. Addressing only where the default pattern is reported, again this is on the magnitude generously of 100,000s.

There exists no concrete evident that tabulators were in error in presenting results Patterns are suspicious, but until precinct level analyses are performed, this does not apply.

Add it up, it does not get you there.

Mike

[Chi]

128. I think you might have erred

Heya mgr
"There exists no concrete evident that tabulators were in error in presenting results Patterns are suspicious, but until precinct level analyses are performed, this does not apply."

Most mathematicians would not remove part of an equation because it is unknown, they would insert an 'x'.
Remove part of the equation, and then say it doesn't add up, seems a bit illogical.
IMHO

Quick question.... do you think the Ohio tabulators should be verified?

Thnx in Advance

[euler]

121. You've made a science out of making statements that....

.... are demonstrably false.

"overestimate democratic strength" is not his 'interpretation.' It is a verifiable fact.

Mitofsky and Lenski have reported Democratic overstatements to some degree in every election since 1990. Moreover, all of Lenski and Mitofsky's statements were on the record long before Election Day 2004....in some states, the errors in 2000 were still quite large...including five states that had stupendously bad exit poll estimates

The exit poll is a blunt instrument. The polls are getting less accurate. (Mitofsky and Lenski) recommended 'raising the bar' on projections made from exit polls: 'The proposed changes result from a belief that exit polling is 'less accurate than it was before' and that 'we should take exit poll data with caution in making calls,' said Lenski

Read it all:

http://www.mysterypollster.com/main/2004/12/have_the_exit_p.html

[RaulVB]

115. ONLY IN THE USA

Point that you conveniently ignore...

[k8conant]

6. I've read your link and it doesn't say "exit polls" just polls...

although it may indeed be referring to exit polls. Of course all the other analyses have stated that the red swing in the vote count far exceeded any blue swing.

[Mistwell]

9. It is exit polls

It's coming from (and in response to things posted at) this site:

http://election.princeton.edu/

It's off the part labelled "Exit Polls" I believe.

[emcguffie]

34. says state polls the week before.

Calculations were based on all available state polls in the week before the election, which were used to estimate the probability of a Bush/Kerry win, state by state.

[Mistwell]

48. Some of the main page says that

But if you go to the main page, and follow the link that leads to the page I linked to at the top of this thread, you will see it is coming off the Exit Poll portion of the main page.

I could be wrong, but it looked like it was an exit poll response. And, since TIA confirmed that 10 states DID deviate for Kerry, I'm willing to go with that.

[Imnottelling]

16. Let's see.

I checked your link and don't know what polls the person is using, the polls margin of error, or the sample size.

However....

From http://www.scoop.co.nz/mason/stories/HL0411/S00142.htm#a they given exit poll results for Wisconsin as K:B = 49.2%: 48.8% with a sample size of 2223 which gives a margin of error of ~1.1%. The reported election results were K:B = 50.2% : 49.8%.

So it seems that Bush and Kerry did exactly as expected in Wisc. from the exit poll within the margin of error....Bush didn't do BETTER in the exit poll than the reported result. You alays have to consider the error, e.g. if the reported result would have been K:B = 49.2%: 48.8% it would've been coicidence that it was exactly right. I suppose the exit poll should be written more correctly as K:B = (48.2 to 50.2)%: (47.8 to 49.8)% with ~68% confidence. Or in physics, we sometimes write K:B = 49.2(1.1)%: 48.8(1.1)% where the number in the paratheses is one standard deviation.

According to scoop we also have,

New Mex. with K:B = 50.1% : 47.5% with 1951 respondents which gives a MOE ~ 1.2. The reported results, according to cnn.com, is K:B = 49.6% : 50.4%. We see that Kerry's total was within is margin of error. However, Bush's total was 2.42 standard deviations higher than the exit polls suggested....that's a very large deviation, meaning that it is highly improbable, i.e. less than a 1/100 of randomly happening.

New Jersey results essentially match the exit polls, however, scoop states that "data was obtained too late and highly contaminated with actual tabulation results" as it was obtained at 12:50am EST on Nov. 3rd. The results I'm quoting for Wisconsin and New Mexicon were obtained at 12:21am EST and 12:24am EST on Nov. 3rd, respectively.

So, I would say that, according to the polls that I have seen, Bush did NOT do better in the exit polls than he did in the reported results for Wisc, and New Mexico. New Jersey, as far as I know, is unknown.

[tex-wyo-dem]

19. Dick Morris seemed to think the exit polls were cooked...

but in the opposite direction (i.e. Dems cooking it early to suppress votes in the West on election day).

http://www.hillnews.com/morris/110404.aspx

I'm definitely no fan of Dick Morris, in fact I think he's a slug, but his opinion on this matter does hold weight since he has been in politics and elections for a long time.

Note: I don't agree with him on the Dems being the exit poll chefs :)

My questions are:

#1 If exit polls are inaccurate and not meant to predict an outcome or be used to verify and election was run cleanly, then what is their purpose?

#2 What is the purpose of adjusting the exit polls to match the actual vote after midnight? This is like giving a school kid a test to find out if he/she learned anything and then "adjusting" the test answers to match the cheat sheet...in other words, pointless.

#3 With question #1 and #2 in mind, why in hell would networks shell out $millions$ to have something that is worthless?

[Lurker321]

21. Answers

#1 - exit polls are not meant to be used to verify election - that is according to Mitofsky, who ran exit polls in 2004 and who is credited with inventing the concept of exit polls. Their purpose - analyzing voting trends, being able to tell who voted how etc. Any time you see stuff like "African-Americans voted 89 to 11 for Kerry", this information is from an exit poll.

#2 - exit polls are adjusted using actual vote numbers in order to reduce the errors introduced into the raw data by clustering, incorrect sampling, invalid precinct distribution in the precincts picked, etc. Raw exit poll data is very skewed - see my post above. If you went by raw exit poll data, Dukakis would have been elected president.

#3 - because networks need the data that exit polls provide in order to talk about it endlessly during the election coverage and for years to come. In terms of air time filled, they get their money's worth.

[Chi]

20. Who is Jim G.?

Heya Mistwell
I have no idea who this Jim G. is, but I'm assuming you do since you take these numbers as accurate.

The following numbers (exit polls) are from the DailyKos.
These are supposed to be the Battleground states.
Jim G.'s numbers don't match. Nor does NJ make the list as a battleground state.
It might be a good idea to verify Jim G.'s numbers before you bank on them. I would be interested to see what you find.

Zogby has cited 10 Battleground states in this 10/24 article, which doesn't include NJ either. http://www.zogby.com/news/ReadNews.dbm?ID=904

Kerry/Bush
4PM Exit Polls...Final.......Change.....Favors

PA.. 53..46.....51..49.....-2..+3....+5 Bush
FL. .51..49.....47..52.....-4..+3....+7 Bush
NC. 48..52.....43..56.....-5..+4....+9 Bush
OH. 51..49.....49..51.....-2..+2....+4 Bush
MO. 46..54.....46..53......0 ..+1....+1 Bush
AR.. 47..53.....44..55.....-3..+2....+5 Bush
MI.. 51..47.....51..48......0 ..+1....+1 Bush
NM. 50..49.....49..50.....-1..+1....+2 Bush
LA.. 43..56.....42..57.....-1..+1....+2 Bush
CO. 48..51.....46..53.....-2..+2....+4 Bush
AZ. 45..55.....44..55.....-1..0 ....+1 Bush
MN. 54..44.....51..48.....-3..+4....+7 Bush
WI. 52..47.....50..49.....-2..+2....+4 Bush
IA.. 49..49.....49..50......0 ..+1....+1 Bush

See any patterns that wouldn't jive with statistical probability?

NOTE-These numbers were recorded BEFORE the exit poll numbers were 'corrected' to match results.

Thanx in Advance

[Lurker321]

22. See post #18 -

how do the results from the last 4 presidential elections "jive with statistical probability"?

[Chi]

24. Break it down...

Heya Lurker <waves>
Break it down to battleground states VS. non battleground states and I'll be happy to look at them, since that is the context of my comment.
Apples to apples..ya know.
8)

[Lurker321]

25. Actually, Chi,

the national exit poll should be a lot more accurate (that is, its MOE should be a lot less) than the state exit polls, because of the much larger sample size. So if raw national exit polls are that much off historically, state exit polls can be a lot more off.

[Chi]

27. I believe...

I believe the two sides of opinion are separated by a difference of overview.
One camp says 'how can all (or most) battleground states favor Bush'.
(which has a sound statistical base)

And the other camp says 'the national numbers are well within historical margins'
(which is also correct.)

I think, if one were properly motivated, some good answers would be..
1) If this is usual, what is the next worst election that has the battleground states so far out of wack with the results.
And what were those numbers.

2) Is raw data normally withheld from the public, or is this unusual.

3) There is also a question about a Bush/Kerry percentage margin that changed beyond mathematical possibility, according to the increase in sampling size. (percent of change in votes was beyond the amount of new samples).

BTW I'm a Kevin Drum fan myself. Back when he was Calpundit and cat blogging on Fridays 8) Him and Josh Marshall are my favorites.

[Chi]

39. Found answer to #2

"The exit poll data is available for purchase by TV-affiliates, newspapers, internet sites and other news outlets. The polling data is delivered through a secure web application directly to subscribers on election day. Political candidates may subscribe as well, but the data is available one week past voting day."
http://www.exit-poll.net/
election-night/exitpollsystem.html

So I guess the Big question is, WHY are they not releasing the raw data as promised, considering one week has passed 9 times.

[Carolab]

23. Margin of error

Bush's "actual totals" exceeded the MoE by such an extent in 16 of the 51 states as to be statistically impossible.

A shift of +2 or +1 would be tolerable here and there, but that is not what we are talking about in the critical states.

[Mistwell]

51. Statistically impossible?

Okay, show me why it is statisically impossible.

The impression I get is that it is not rare to go beyond the margin of error. Indeed, regular polls (and I do understand that this is not a regular poll) go well over the margin of error all the time. We saw many MANY weeks go by with regular polls showing DRASTICALLY different results that made SOMEONE'S margin of error fautly for sure.

In addition, someone posted to this thread that the margin of error for several of the prior Presidential elections were also blown for those actual exit polls as well, sometimes by a wider margin than this exit poll. Not to mention the recent second election in Ukraine had it's exit polls blown out of the water (though the raw result of choosing a winner was accurate).

So, I need to see why margin of errors for early exit polls are sacred cows. There is more to this than simply running a ststatisical analysis on this single poll. I need to see how those other (unadjusted) polls in earlier years and other locations faired in comparison to this (unadjusted) exit poll. If it is common for exit polls to blow the MOE, then no statisical analysis of a single poll in a vaccum will render an accurate representation of what happened.

[emcguffie]

54. there still could have been cheating in Ukraine

I don't know what the results or exit polls were, but the mere fact that it was a re-vote doesn't mean in all areas it was legit. Probably less likely it was in the eastern part of the country.

So, could have been off of the exit polls because of fraud, once again.

[waz_nc]

75. multiple comparisons

I'll start this by saying that I think exit polls are probably correct and actual vote totals were manipulated. Still I have some questions about the methods applied in calculating the margins of error for the 50 state exit polls.

When you're looking at exit polls in 50 states, I think you need to adjust for the fact you're actually conducting 50 separate statistical tests. Although I don't know anything about polling, as far as I can tell, the margin of error is actually a 95% confidence interval and the probability of something falling outside that is 5% (i.e. p<0.05). But it is a 95% confidence interval for a single test. If you want a family confidence interval of 95% for all 50 tests, you have to set the confidence intervals for the individual tests much lower. Since I don't teach statistics I pasted something below

Methods of adjusting the cut-off p value to control the effects of multiple comparisons:

The standard cut-off for considering a statistical test statistic to be statistically different has been p<0.05. For a single comparison, this indicates that the observed differences would occur by chance only 5% of the time. When multiple comparisons are being made within an experiment, however, the probability of observing a sizable difference for one of the comparisons increases with the number of comparisons being made. If you make 21 independent comparisons, you would expect that one of them generate a test statistic with a significance level < 0.05. There are several ways to adjust the p value to control for the effect of multiple comparisons.

1. Bonferroni: This is a standard multiple test comparison described in most statistics textbooks and is the most stringent of the methods. The p value used for a cut-off is divided by the total number of comparisons to be made. Any single comparison in a study making 100 comparisons would have to generate a test statistic that would have a p value of 0.0005 in a single comparison to be considered significant at the level of p<0.05.

Here is the link http://medir.ohsu.edu/~geneview/education/Multiple%20test%20corrections.doc

Here's another link.

http://www.cmh.edu/stats/ask/bonferroni.asp

And here's a website where you can download a program for calculating adjusted p-values http://sagebrushpress.com/PEPI.html (download the program ADJUSTP).



So if you use Bonferroni, you would need to set your confidence intervals for individual tests at 0.0001 to maintain a 95% family confidence interval. Bonferroni is overly conservative (or less powerful than some other methods that the ADJUSTED P program calculates. I actually took numbers from one of TIA's earlier posts on this subject and put them into the ADJUSTP program to see what I came up with. Here are the results. With the exception of Finner's adjustment, the other adjustments, really reduce the number of states with p-values less than 0.05 (i.e. fall outside the margin of error.)

TIA;s HOLM's HOMMEL's FINNER's
P adj-P adj-P adj-P
DE 0.0029 0.1218 0.1131 0.0170
NH 0.0000 0.0000 0.0000 0.0000
VT 0.0060 0.2340 0.2100 0.0248
SC 0.0001 0.0049 0.0048 0.0025
NE 0.0064 0.2432 0.2176 0.0248
AK 0.0038 0.1520 0.1444 0.0172
AL 0.0153 0.5508 0.3978 0.0501
NC 0.0001 0.0049 0.0048 0.0025
NY 0.0018 0.0810 0.0720 0.0159
CT 0.0024 0.1056 0.0936 0.0170
RI 0.0268 0.9112 0.5628 0.0768
MA 0.0221 0.7735 0.4975 0.0675
PA 0.0016 0.0736 0.0656 0.0159
MS 0.0329 1.0000 0.6251 0.0843
OH 0.0030 0.1230 0.1170 0.0170
FL 0.0006 0.0282 0.0270 0.0075
MN 0.0027 0.1161 0.1053 0.0170
UT 0.0478 1.0000 0.7877 0.1101
ID 0.0860 1.0000 0.9135 0.1484
AZ 0.0138 0.5106 0.3726 0.0484
VA 0.0566 1.0000 0.8896 0.1165
LA 0.0449 1.0000 0.7633 0.1085
IL 0.0678 1.0000 0.9135 0.1265
WI 0.0300 0.9900 0.6000 0.0811
WV 0.0654 1.0000 0.9135 0.1265
NM 0.0554 1.0000 0.8864 0.1165
CO 0.0515 1.0000 0.8240 0.1132
IN 0.1642 1.0000 0.9135 0.2380
GA 0.1069 1.0000 0.9135 0.1717
MO 0.0817 1.0000 0.9135 0.1460
NJ 0.1267 1.0000 0.9135 0.1924
WA 0.0970 1.0000 0.9135 0.1613
IA 0.1241 1.0000 0.9135 0.1924
AR 0.1955 1.0000 0.9135 0.2671
KY 0.2601 1.0000 0.9135 0.3272
OK 0.2163 1.0000 0.9135 0.2829
MI 0.1647 1.0000 0.9135 0.2380
NV 0.2129 1.0000 0.9135 0.2829
ME 0.2780 1.0000 0.9135 0.3414
MD 0.3918 1.0000 0.9135 0.4629
DC 0.4796 1.0000 0.9135 0.5491
MT 0.5072 1.0000 0.9135 0.5693
OR 0.6991 1.0000 0.9135 0.7525
HI 0.7110 1.0000 0.9135 0.7560
TX 0.8710 1.0000 0.9135 0.8973
TN 0.8968 1.0000 0.9135 0.9058
CA 0.9135 1.0000 0.9135 0.9135
SD 0.8965 1.0000 0.9135 0.9058
ND 0.8858 1.0000 0.9135 0.9054
KS 0.8876 1.0000 0.9135 0.9054

[TruthIsAll]

90. This is the way I calculate the probability for all the states.

Note that .025 is the probability that the Bush vote tally in any given state would exceed it's exit poll margin of error.

The MOE = 1/SQRT(n), where n is the exit poll sample size

SIXTEEN (16) out of 51 states exceeded the exit poll MOE for Bush.

The probability P that 16 OR MORE states would EXCEED the MOE for Bush is equal to P = 1 - the probability that 15 or FEWER states WOULD exceed the MOE for Bush.

************* P = 1 - BINOMDIST(15, 51, .025, TRUE) ********

************* P = 1 out of 13.5 TRILLION **************

[RaulVB]

26. Tread attempting to deny validity of exit polls. Nice try! (n/t)

[euler]

29. Don't THREAD on me : ))

[euler]

30. By the way, thanks for the encouragement

[RaulVB]

33. No problem. Is an empty attempt. Have fun! (n/t)

[Joe Chi Minh]

37. Yes, we're

all very impressed by their attempts to joust with the grown-ups, aren't we, Tuco...? Adversely impressed...but impressed.

[RaulVB]

38. That's right (n/t)

[Mistwell]

52. Not deny validity - just pinning down HOW valid

I don't deny that exit polls are both useful in general and useful specifically in this case for proving fraud.

I just want to pin down how truly accurate they are, and how powerful this one is for evidence of fraud.

At this point, I feel like a lot of people exagerate how powerful exit polls are as a tool for PROVING fraud (and that hurts the cause). I think they are fairly useful, but so far I do not see why people think they are, alone, conclusive of fraud.

[emcguffie]

35. I think those are the "corrected" exit polls.

I'm not positive. but they sure look like they're in line with the outcome, so those are the ones they doctored to match the election.

[Blackbeard]

85. Poll errors

I know it's probably unpopular to suggest the election really wasn't stolen, but if you'll indulge me, I'll explain why the exit polls differ from the actual results.


1) Exit polls use a very small sample size and the actual count uses a very large sample size, in fact, it's 100%.

2) People don't have to answer exit polls. Maybe Bush voters are less likely to answer exit poll questionnaires than are Kerry voters. Think about it -- if you had just held your nose and voted for Bush, would you want to advertise that fact?

3) Exit polls are not taken at every polling place, therefore not every voter is as likely to be polled as every other. They select certain polling places that fit the demographics of "likely" voting patterns from past elections.

4) Since all previous Presidential elections were before 9/11 and Iraq, all of the previous voting patterns are pretty much out the window. Millions turned out to vote who otherwise would not have -- some to vote for Bush, some to vote against him. The point is the polling places selected for exit polls may not have accurately reflected the actual voter demographics.


That, in a nutshell, is why you can't compare exit polls with actual results and conclude that the actual count must be fraudulent. The two measurements are taken from different populations.

[New Earth]

86. the exit poll discrepancy is not the only thing that may imply fraud

it's the 'discrepancies' and 'irregularities' in the vote count, the machine tabulation, the ratio of more votes than voters, voter suppression, not supplying enough voting machines in certain areas, etc.

All these people trying to explain how the exit polls could be wrong are really not going to change anyone's mind that there was fraud. If that is anyone's intention, they are simply wasting their time.

[Name removed]

91. Deleted message

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

[TruthIsAll]

95. Oh, boy. Another one.

1) Exit polls use a very small sample size and the actual count uses a very large sample size, in fact, it's 100%.

Well, duh, you have a point. 100% sample size. I see.
Next.

2) People don't have to answer exit polls. Maybe Bush voters are less likely to answer exit poll questionnaires than are Kerry voters. Think about it -- if you had just held your nose and voted for Bush, would you want to advertise that fact?

Good point. Lot's of boogers were picked in this election.

3) Exit polls are not taken at every polling place, therefore not every voter is as likely to be polled as every other. They select certain polling places that fit the demographics of "likely" voting patterns from past elections.

Right. Mitofsky doesn't know where to poll.

4) Since all previous Presidential elections were before 9/11 and Iraq, all of the previous voting patterns are pretty much out the window. Millions turned out to vote who otherwise would not have -- some to vote for Bush, some to vote against him. The point is the polling places selected for exit polls may not have accurately reflected the actual voter demographics.

Yes. Beacuse of Iraq, people moved to the country. Away from the city.


That, in a nutshell, is why you can't compare exit polls with actual results and conclude that the actual count must be fraudulent. The two measurements are taken from different populations.


Very astute post. I must say, you put a lot of effort into it. And your logic is just impeccable. I hope you testify for the Repukes.

[euler]

104. I missed the report. Please provide a link.

"Beacuse of Iraq, people moved to the country. Away from the city."

[mgr]

109. Actually, his thesis points are right on

If I were grading his argument, he might get a C, possibly a B when compared to the rest of the responses. Using the argument as a straw man to demonstrate a refutation of all the 'naysayers', you would get an F. Your mockery does not stand up to the kernels of truth in his argument. The goal of this thread, is to get at a deeper appreciation of what an exit polls means, this is the fruit of the rejected hypothesis you offered that exit polls demonstrate fraud. This person is just starting to grasp the argument, and should not be put down. Let me recast their argument in a better light:

1.) The fact that total vote count was over 100 million total, and the reported national sample size was around 15,000, would lead any rational person with a moderation of skepticism to look askance at the sampling. This grows ever more ridiculous when 500 absentee voters were sampled for a comparison to a population on the order of 10,000,000 +. The order of magnitude difference would suggest to me that the NEP exit polling is not very robust in addressing any national voting trend, as regional biases may be built in. As those filters are reduced to state and local contests, these concerns may diminish.

2.) If people are not compelled to answer exit polls, and time is money or life itself, it would only be persons with political interests and spare time that would respond. How willing are your parents or co-workers or peers to answer a telephone poll which is far more convenient? If you have left work to go vote, do you have the time to take answering a poll, when you have to get back to work, or home to the kids, or fix dinner,...?

corollary: if exit polling is conducted some distance from the polling place entrance, and between that entrance and where the exit poller may be placed, one may park a car, then who is sampled?

3.) Exit polls cannot be conducted at all polling places. The criteria for precinct selection would be past patterns. The point made is in part addresses the clustering effect, and the fact that possibly uniformitarian principles can reach thresholds where a new pattern of voting may appear. Say, if the preferred bench mark polling place were undergoing a slow gentrification, this may result in slight adjustments within the voting population that would not appear to perturb the exit poll's conclusions (because they are below discernment with the sample methodology), until a certain percentage of the area was yuppiefied. Similar arguments may be made for suburban and exurban places. The problem as I see it, is that a possible explanation for the oddities in Edison Mitofski's work might be that they are trying to develop a new model based upon precincts for which not much previous polling experience is available. I have not seen any evidence that Mitofski conducted the national exit polling in previous elections.

4.) Was the large turnout expected? I don't recall much on this, there is a trend in the country toward the conservative end of the spectrum, and it may just be that a majority share more with the republicans than democrats, and 9/11 was the seminal event causing this. If one were to accept the argument that blind support of the president during war time is needed to show support for our troops, then would this be sufficient to bring out first time republican voters across the country? On the face of it, I don't know, but we are more than thirty years away from our previous unpopular war, and possibly this follows from the conservative rethinking of the causes and possible outcomes of Viet Nam. The election outcome may reflect this simple piece of rationalization, not the deeper discomfort with Bush's polities associated with his poor approval ratings.

We need as many good minds for the cause as we can get, jeering a imperfect argumentation when the gist is on target, denies the ability to grow, and is the antithesis of liberalism.

As a final question, TIA, what in your mind would constitute refutation of the argument that exit polling demonstrates fraud in the national election, persuasive to you that you would abandon your position?

Mike

[TruthIsAll]

111. Your post is totally ridiculous and does not merit a response. n/t

.

[mgr]

112. So, nothing can dissuade you from your position? n/t

[RaulVB]

114. One more "newbie" that doesn't like exit polls...

Interesting how "excited" they get.

We like "tin foil" around here...:tinfoilhat:

Try to "convince" people elsewhere.

[euler]

116. You are once again confusing exit....

....polls with discussions about can and can't be done with exit polls.

He DOES like exit polls. He DOES NOT like what you are attempting to do with them.

[mgr]

117. Have you anything other than ad hominen arguments to make?

I see a poor correlation between my number of posts and amount of time I have spent on this site. I make the effort, unlike many bloggers, to write a coherent, fair, thoughtful, and direct response to the issue. It takes time to do rather than post 'kick' five hundred times. I estimate it takes me anywhere from fifteen minutes to two hours. From what I have seen of your posts, I doubt that the same emphasis on quality and aptness applies.

Apparently from your response, you are adamant in your belief that exit polls show discrepancies from the election results, and it is the election results that do not indicate the overall voters' intent. To irrationally cling to a belief and not even entertain possibilities where that belief may be wrong is ascientific, and not a position I would advertize.

But, you are right, there is no point to continue to discuss this with you, and I am sorry for you.

Mike

[RaulVB]

122. No, I FEEL SORRY FOR YOU...

I really do...

[euler]

124. see this

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=282472&mesg_id=282575&page=

[Blackbeard]

88. One more word on exit polls

The poll mentioned above

http://www.leemizell.com/freedocs/election_free.pdf

has the following breakdown by the voters' 2000 presidential vote:


2000 % 2004
------ -- ------
Bush 43 91% Bush, 9% Kerry
Gore 37 10% Bush, 90% Kerry
Other 4 21% Bush, 71% Kerry, 3% Nader
Didn't Vote 17 45% Bush, 54% Kerry, 1% Nader


According to this poll, Bush won by six points in 2000, while the acutal vote had him losing by 0.5%. Is this evidence of vote tampering in favor of Gore?

[TruthIsAll]

93. Your 2000 exit poll numbers are totally bogus.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=271424

[FogerRox]

120. Tia -it is my understanding that when 2 white guys are running

exit polling is accurate to within 1%--maybe 1.5%.
Now if you have a white male and a black female running that is where you run into exit poll/results discrepancies of 5%--8%--10%.

Also IIRc Mitofsky did the Ukraine and USA exit polling --now with all the money from the directorate of operations (CIA) flowing into the Ukraine--(lets skip to Nov. 5th, thursday-IIRC)and then the NY Times article on Mitofsky's explaination on why the exit polls showing Bush won, were right-Does Mitofsky's legitimacy get impunged by Supporting the Bush thoery --after doing the dirty work for them in the Ukraine? ANd yet the MSM did not offer a valid counter agrument to Mitofsky.

I guess what I'm trying to say is--with the potential conections to CIA money--thusly daddy Bush---I dont think mitofsky is trust worthy.
Follow Mitofsky's actions--do they truly indicate a man of integrity?

[Chi]

96. You might want to check...

Heya Blackbeard
You might want to check the 'Sources' links on that page your taking you info from, neither of them work.
Considering it's a private firm posting these numbers, having dead links to back their facts is less than reassuring.

[mgr]

119. Links seem to work for me just fine. Try typing them in.

Looks like we have the Washington Post data as being preselected as the more accurate than the CNN data from what is reported as the same original source. I thinks this makes the whole enterprise suspect, and the analysis touted by TIA as "cherry picking".

Mike

[Chi]

125. I did type them....

Still didn't work, and I checked it twice before posting
<shrug>
Must be me.

[euler]

123. see this

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=282472&mesg_id=282575&page=