STATE RED SHIFT PROBABILITIES: EXIT POLLS TO ACTUALS

[TruthIsAll]

[BlueDog2u]

1. Nice

[TruthIsAll]

33. COMPARE THE STATES ON THE LEFT SIDE OF THE GRAPH TO THOSE ON THE RIGHT.

The states on the LEFT side of the graph have extremely LOW probabilities. Unusual deviations were OUTSIDE THE MOE.

The states on the RIGHT side have HIGH probabilities.
The deviations were NOT that unusual and WITHIN the MOE.

WHAT DISTINGUISHED THE STATES ON THE LEFT (UNDER 3%) FROM THOSE ON THE EXTREME RIGHT?

Most (80%) of the deviations moved in Bush's favor.

[GingerSnaps]

2. Can I ask a stupid question

Will it help change the election to the real winner? I hope so :hug:

[Gman]

3. What does your graph mean?

The probability of what in each state? If you're charting the differences between actuals to exits that's one thing. But you're saying "probability".

(We analysts are always asked these questions and we always tend to feel everyone already knows what our data means. I say "we" because I'm a budget analyst by trade and understand why people ask.)

[TruthIsAll]

8. SEE POST #7

tia

[TruthIsAll]

4. HERE IS THE RED SHIFT DATA

Average State Poll Margin of Error: 3.05% 
(based on published statitistics on the number polled)			

Average Standard Deviation 1.55% 
(input to the Excel Normal Distribution function to calculate
the probabilities)

For example, here is the calculation for the probability that
Kerry's 2-party% in AZ would shift from the exit poll to the
actual by 2.56% (47%-44.44%): Prob = 4.96% 

Prob = NORMDIST(44.44%,47.00%,1.55%,TRUE)
								
	EXIT POLLS	Kerry%	ACTUAL VOTE 	Kerry	Red		
State	Kerry	Bush	2-Party	Kerry	Bush	2-party	Shift	Prob
AK	40.5	59.5	40.50%	35	62	36.08%	8	0.22%
AL	41	59	41.00%	37	63	37.00%	8	0.49%
AR	46.6	53.4	46.60%	45	54	45.45%	2.2	23.00%
AZ	47	53	47.00%	44	55	44.44%	5	4.96%
CA	54	46	54.00%	55	44	55.56%	-3	84.22%

CO	49.1	50.9	49.10%	47	52	47.47%	3.2	14.72%
CT	58.5	41.5	58.50%	54	44	55.10%	7	1.42%
DC	91	9	91.00%	90	9	90.91%	1	47.66%
DE	58.5	41.5	58.50%	53	46	53.54%	10	0.07%
FL	50	49	50.51%	47	52	47.47%	6	2.53%

GA	43	57	43.00%	41	58	41.41%	3	15.31%
HI	53.3	46.7	53.30%	54	45	54.55%	-2.4	78.92%
IA	50.7	49.4	50.65%	49	50	49.49%	2.3	22.82%
ID	33.5	66.5	33.50%	30	68	30.61%	5	3.12%
IL	57	43	57.00%	55	45	55.00%	4	9.85%

IN	41	59	41.00%	39	60	39.39%	3	15.01%
KS	35	65	35.00%	37	62	37.37%	-5	93.72%
KY	41	59	41.00%	40	60	40.00%	2	25.94%
LA	44.5	55.5	44.50%	42	57	42.42%	4	9.03%
MA	66	34	66.00%	62	37	62.63%	7	1.48%

MD	57	43	57.00%	56	43	56.57%	1	38.97%
ME	54.8	45.3	54.75%	53	45	54.08%	1.5	33.43%
MI	52.5	47.5	52.50%	51	48	51.52%	2	26.26%
MN	54.5	45.5	54.50%	51	48	51.52%	6	2.71%
MO	47.5	52.5	47.50%	46	54	46.00%	3	16.66%

MS	43.3	56.8	43.26%	40	60	40.00%	6.5	1.78%
MT	39.8	60.3	39.76%	39	59	39.80%	-0.5	50.92%
NC	48	52	48.00%	44	56	44.00%	8	0.49%
ND	34	66	34.00%	36	63	36.36%	-5	93.64%
NE	36.8	63.3	36.76%	32	67	32.32%	8.5	0.21%

NH	55.4	44.6	55.40%	50	49	50.51%	9.8	0.08%
NJ	55	45	55.00%	53	46	53.54%	3	17.23%
NM	51.3	48.7	51.30%	49	50	49.49%	3.6	12.21%
NV	49.4	50.7	49.35%	48	51	48.48%	1.7	28.82%
NY	63	37	63.00%	58	40	59.18%	8	0.69%

OH	52.1	47.9	52.10%	49	51	49.00%	6.2	2.28%
OK	35	65	35.00%	34	66	34.00%	2	25.94%
OR	51.2	48.8	51.20%	52	48	52.00%	-1.6	69.71%
PA	54.4	45.7	54.35%	51	49	51.00%	6.7	1.54%
RI	64	36	64.00%	60	39	60.61%	7	1.43%

SC	46	54	46.00%	41	58	41.41%	9	0.15%
SD	37.8	62.3	37.76%	39	60	39.39%	-3.5	85.38%
TN	41.5	58.5	41.50%	43	57	43.00%	-3	83.34%
TX	37	63	37.00%	38	61	38.38%	-3	81.40%
UT	30.5	69.5	30.50%	27	71	27.55%	5	2.85%

VA	47	51	47.96%	45	54	45.45%	5	5.31%
VT	65	35	65.00%	59	39	60.20%	10	0.10%
WA	55	45.1	54.95%	53	46	53.54%	2.9	18.15%
WI	52.5	47.5	52.50%	50	49	50.51%	4	9.90%

WV	45.3	54.8	45.25%	43	56	43.43%	3.5	12.01%
WY	29	65	30.85%	29	69	29.59%	4	20.83%

[gristy]

6. How about if you graphed the shift and the probablity of that shift

for each state on the same graph.

[TruthIsAll]

19. It would be too cluttered, plus the scale would be a major problem.

You have the numbers in the numerical chart I have supplied.

[TruthIsAll]

35. NEW! GRAPH OF KERRY'S 2-PARTY% SHIFT TO BUSH AND PROBABILITY OF THE SHIFT

LEFT AXIS: KERRY 2-PARTY RED SHIFT TO BUSH (BARS)
RIGHT AXIS: PROBABILITY OF RED SHIFT (LINE)

Shifts to Bush (MOST states) are NEGATIVE and on the LEFT and CENTER of the graph.

Shifts to Kerry (only NINE states) are POSITIVE and on the extreme RIGHT side of the graph.

[gristy]

49. now we're getting somewhere

I would have expected the probability curve to reach a maximum when the "red shift" is zero, however. A big blue shift (bars on the right) should be just as improbable as a big red shift.

Also, you should "label your axes!", as one of my high school teachers would always say (sadly, I cannot remember which one).

[gristy]

26. Well, I copied this into excel

and I put the "shift" data on a bar chart (like you have the probability data), and now I am trying to add the probablity data to the bar chart in a line format, with its axis (showing the line graph's scale) on the right side.

But I've never done a graph of mixed format like this, and can't see how to do it right now. I think that such a graph (along with more text explanation on the graph itself) would help.

[Trillo]

38. I added electoral numbers and sorted, hope you don't mind

State	Prob	Electors
DE	0.07%	3
NH	0.08%	4
VT	0.10%	3
SC	0.15%	8
NE	0.21%	5
AK	0.22%	3
NC	0.49%	15
AL	0.49%	9
NY	0.69%	31
CT	1.42%	7
RI	1.43%	4
MA	1.48%	12
PA	1.54%	21
MS	1.78%	6
OH	2.28%	20
FL	2.53%	27
MN	2.71%	10
UT	2.85%	5
ID	3.12%	4
AZ	4.96%	10
VA	5.31%	13
LA	9.03%	9
IL	9.85%	21
WI	9.90%	10
WV	12.01%	5
NM	12.21%	5
CO	14.72%	9
IN	15.01%	11
GA	15.31%	15
MO	16.66%	11
NJ	17.23%	15
WA	18.15%	11
WY	20.83%	3
IA	22.82%	7
AR	23.00%	6
KY	25.94%	8
OK	25.94%	7
MI	26.26%	17
NV	28.82%	5
ME	33.43%	4
MD	38.97%	10
DC	47.66%	3
MT	50.92%	3
OR	69.71%	7
HI	78.92%	4
TX	81.40%	34
TN	83.34%	11
CA	84.22%	55
SD	85.38%	3
ND	93.64%	3
KS	93.72%	6

electoral numbers from:
http://www.fec.gov/pages/elecvote.htm

[gristy]

5. It is not clear what you are graphing

[TruthIsAll]

7. THE GRAPH SHOWS THE PROBABILITIES THAT THE RED SHIFT

COULD NOT BE DUE TO CHANCE ALONE IN EACH STATE.

IN OTHER WORDS, THESE ARE THE PROBABILITIES THAT THE ACTUALS WOULD DEVIATE FROM THE EXIT POLLS IN THE GIVEN STATE BY THE RED SHIFT PERCENTAGE.

I CALCULATE THE DIFFERENCES BETWEEN KERRY'S 2-PARTY% FOR THE ACTUAL AND THE EXIT POLL. I USE THESE TWO PERCENTAGES, ALONG WITH THE STANDARD DEVIATION, AS INPUT TO THE EXCEL FUNCTION.

LOOK AT THE NUMERICAL CHART OF THE ACTUAL NUMBERS.

When the actuals are CLOSE to the exit polls, the probability is HIGH. There is nothing fishy here.

when the actuals are FAR from the exit polls, the probability is LOW (near ZERO). We are swimming with the fishes.

[bri_in_austin]

9. To regular people who do not read graphs everyday and do not speak

statistician talk:

What does your graph mean?

[bri_in_austin]

10. Oh yeah,

and what is red shift?

[TruthIsAll]

12. The Red Shift is the magnitude of the shift from the Exit Polls to Bush

I assumed readers would be familiar with prior threads on this topic.

View the SCOOP thread by Althecat. Go here:
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x61892

[bri_in_austin]

14. Gotcha

Thanks.

[BlueEyedSon]

13. What is your "threshold percent" for suspecting tampering?

50%?
10%?

Also, does the red shift prob correlate with something else? Republican SOS? Optical scan? Diebold? Weather?

[TruthIsAll]

22. I leave that to the readers.

Anything under 3.0 percent is...well, highly suspicious.

I have no specific data regarding Optiscam, etc factors.

Keep focused. This is a picture of the forest.
Let others cut down the trees.

KISS

[ewulf]

54. But thats not really what the Graph shows:

The graph shows that the probability is highest for the results from the few states that had Kerry do BETTER in the tabulated results than in the exit polls. It should be redone to reflect that a deviation in either direction (favoring Bush or Kerry) is less likely than a non-deviation; it would make everything more convincing if you're calculation shoed with result.

[Dolphyn]

60. Graph shows probability that red shift is AT LEAST xx by chance ...

So, for Kansas: Red shift is -5, and by chance there is a 93.7% chance that the red shift is at least -5. So, the probability for Kansas is 93.7%.

Montana: Red shift is -0.5, and by chance there is a 50.9% chance that the red shift is at least -0.5. So, the probability for Montana is 50.9%.

At least I think that's what he's saying, LOL.

Confusing? Yes.

I don't think this particular display of the data will win any arguments with Republicans, because it is hard to understand and at first glance it appears to be biased.

[TruthIsAll]

66. THIS SHOULD CLEAR IT UP FOR YOU.

A LOW probability (STATES ON THE LEFT AND MIDDLE OF THE GRAPH) means that it is UNLIKELY that the % decline in Kerry's 2-party percentage was due to CHANCE. In other words, the change was most likely due to either FRAUD or MISTABULATION.

A HIGH probability (STATES ON THE RIGHT SIDE OF THE GRAPH) means that it is LIKELY that the %change was due to CHANCE. In other words, the change was most likely NOT due to either FRAUD or MISTABULATION.

[TruthIsAll]

67. Biased? These are the numbers. The Normal Distribution is the tool.

In fact, I was being conservative in using the 3% MOE.

The MOE was MUCH LOWER in the critical states where over 2000 were polled.

Where is the bias?

Did you see Professor Freeman's study?

Here is a snippet of his updated HYPOTHESIS:

................................
Summary and Implications

In this paper, I have tried to demonstrate that:

• exit poll data are fundamentally sound,

• the deviations between exit poll predictions and vote tallies in the three critical
battleground states could not have occurred strictly by chance or random error, and that

• no explanations of the discrepancy have yet been provided.

The unexplained discrepancy leaves us with two broad categories of hypotheses: the exit poll data are wrong or misleading in ways that have not yet been adequately explained, or the count is wrong. It’s important that we review exit poll data and methodology and soon while the evidence is fresh, but such analyses require NEP’s raw data.40 Particularly useful statistical analyses would compare the “shift” in states, counties and precincts where safeguards are strong vs. those where they are suspect. Even more important, however, are investigation into the multitude of allegations and concerns about the count itself.

<snip>

Widespread assumption of misplay undermines not only the legitimacy of the President, but faith in the foundations of the democracy. That the President did not legitimately win the election is a still a very premature conclusion, but the election’s unexplained exit poll discrepancies make it an unavoidable hypothesis. In this paper, I have tried to make the case that the media, academia, polling agencies, and law enforcement agencies should investigate it with a much greater sense of urgency and responsibility than they have thus far shown.

<snip>

[Be Brave]

76. I agree with this comment. nt

[bri_in_austin]

11. I don't mean to be a pain, but

I still don't know what you mean. (I am an educated person, I promise)

[TruthIsAll]

15. You will have to read the SCOOP thread.

I do not know your level of knowledge in probability and statistics.

Suffice to say, the analysis attempts to show that there were many states with statistical anomalies, all in favor of Bush.

[bri_in_austin]

17. I'm sorry...

I understood when you explained "red shift", now I just need to delete the threads that I made after that, and I don't know how.

[bri_in_austin]

16. How do you delete a thread?

I want to delete this one and the one I'm replying to.

[TruthIsAll]

18. Very strange request.

.

[elehhhhna]

20. Check this out! The Constitution Tells Us Specifically What To Do

The Constitution Tells Us Specifically What To Do

Reduce Representation in Congress for States that Botched the Election

Deliver Harsh Penalties to States that Can't Count Votes Honestly


ReDefeatBush founder David Lytel gave introductory remarks entitled When Bush Comes to Shove and concluding remarks entitled Righting the Left: Rebuilding the Opposition Industry. He called on citizens to initiate lawsuits that would implement the penalties for vote suppression written into the U.S. Constitution (14th Amendment, Section 2): "But when the right to vote at any election for the choice of electors for President and Vice President of the United States... is denied to any male inhabitant of such state... or in any way abridged, except for participation in rebellion, or other crime, the basis of representation shall therein be reduced in the proportion which the number of such male citizens shall bear to the whole number of ... citizens." If investigations into election fraud conducted by honest state or local authorities turn up 550,000 violations in one of these states then OH, FL, NC, IN and other states should have their representation in Congress reduced for their inability to conduct fair elections and denying citizens the right to vote.
__________________

[gristy]

21. You can edit a "post", but you can't delete a "thread"

If you want to remove the text of a post, click "edit" on the lower right hand corner of the post.

[Goldeneye]

23. stupid question here

The probabilites aren't as percents? So probability of .1 = 10%?

[TruthIsAll]

28. .1= .10% or one-tenth of one percent

tia

[bri_in_austin]

24. That's what I meant:

"delete a post", so if I click edit then I can edit it, but I still can't delete it.

[Cadence]

45. I think you mean you want to delete your posts

not the entire thread. You can go back and click edit on the message after it's been posted. Then you can delete your text and put something like self deleted in the subject.

[BlueDog2u]

25. TIA, like I already said, nice work, but...

I notice some anomolies here, for example Vermont. In a related thread its been discussed that Vermont has paper balloting. Also Vermont has one of the highest wins for Kerry of any state.

But the original "unmodified" exit poll showed Kerry leading 65 to 35 while the final result was Kerry 59 to 39, a swing of ten points. That's why Vermont shows up way on the left end of your graph with a probability approaching zero that the exit poll and the tabulation could be as distant as they are.

But my question is this: doesn't this finding lend credence to those who claim that exit polls were not accurate? If we are claiming that the difference between tabulations and exit poll results are a result of hacking with electronic systems, this explanation wouldn't apply to a state that uses paper balloting, would it? Or is there something I'm missing?

[TruthIsAll]

29. Most moved in the direction of Bush. The fraud was massive.

Even if the state didn't flip to Bush, his popular vote margins were padded across the board. Everyone thinks he won by 51-48%. In actuality, it was the reverse.

See the SCOOP Red State Shift threads.

[BlueDog2u]

30. OK

I can believe that as a proposition, but that's really all it is at this point. How would that have been done in a state like Vermont with paper ballots? I'm not saying it couldn't have been done, I'm asking how it could have been. Are Vermont's paper ballots counted by machines susceptible to the same kinds of manipulations as Georgia Diebolds? Or the Opscans?

And if Vermont has paper ballots, and has a friendly local government (as it mostly does), and the discrepancy is very great here, isn't it an ideal state for some selected recounts? Wouldn't it be very easy, in just one or two counties, to show that the vote was rigged in Vermont.

[glitch]

95. Vermont is mixed - some paper ballot counted by hand, some paper

ballot counted by Diebold Opti-scan.
See http://verifiedvoting.org/verifier/map.php?topic_string=5std&state=vermont&county=Franklin

Click on each individual county to see the mix.

[PATRICK]

27. I get it but

Pretend I am an impatient and challenged reporter such as even Olberman professes to be. The visual must accurately and simply portray the issue and then lead to a more mathematically precise explanation AND be able to throw aside Rovian smoke and highly improbably "explanations". Those simple explanations, no matter how bad or unscientific, are the ones being put forth "out there" by even supposedly smart people. The issue is being buried without being examined and needs to be utterly reversed.

Te statisticians tried with the zillion to one scenarios which made me cringe because a) they are implicitly reluctant to buy the premise to begin with no matter that 2+2=4 b) when they start digging with doubts ingrained they simply lose it and leave it.

I have yet to find a chart dumb enough to hit the nail and the eyes and hold up under the pre-programmed prejudice.

The exit polls being withheld(by those responsible for the fudging and doubt!) are a big issue. They control the simple side of a gamed issue and won't let go. Assaulting with charts and science is about as good as proving Global warming until the oceans wash away their DC homes. personally I like the graph, but it needs to be for Dummies. The numbers and results must be crunched for the purpose of PR PRESENTATION and overall punch and picture. A pattern of many polling discrepancies nationwide could be done with enough researchers and programming of charts. This particular time is again being "Florida'd", backed into a corner, whittled down, not taken at value and denied.

OK, I confess, for me too.

[TruthIsAll]

31. Well, the analysis charts and numbers provided by many have

struck a chord.

Don't be so sure that the info is not having an effect.The process is just beginning. Providing circumstantial, mathematical evidence is step one.

This is not as complicated as it appears to be.But I agree. 99% of the people do not have a clue about probability and statistics.

But you can bet your ass that the lawyers who specialize in fraud do.

[goclark]

32. Excellent as always Truth is All

You have certainly taken many hours of your precious time to point out information for the lawyers.
We are all praying that each thread of data works.

[PATRICK]

34. Lawyers and political situations

Somehow I feel a chill because so little satisfaction is ever received and very late from that tack, no matter how heinous and obvious the crime.
The politic ans are above the law and certainly above retribution. It is just as critical, maybe more so if the election can't be won, to quickly reform public opinion. The media is burying this compared to the effect it should have. It is that incredible ostrich effect of 2000, the blind deaf and dumb bum's rush to a single simplistic conclusion favoring the "winner".

I know what is needed and I hope Kerry provides it. I have real hope he will. For every inroad we make the GOP will be rushing ahead to a new level of fraud with impunity and vast rewards. Some is not enough. A MoveOn ad is not enough. The entire media has to be really focused and held to a crime issue and a Kerry advance to "winning". Otherwise it will be a long heartless slog, IMHO.

[flintdem]

36. NYT Article suggests reason for Exit poll error

NYT has a lengthy article about Ohio and election turnout:

http://www.nytimes.com/2004/11/21/magazine/21OHIO.html

"Why wasn't it enough? In the days that followed, theories circulated claiming that Republicans had stolen votes from Kerry by messing with the results from electronic voting machines. But the truth was that the Bush campaign had created an entirely new math in Ohio. It wouldn't have been possible eight years ago, or even four. But with so many white, conservative and religious voters now living in the brand-new town houses and McMansions in Ohio's growing ring counties, Republicans were able to mobilize a stunning turnout in areas where their support was more concentrated than it was in the past. Bush's operatives did precisely what they told me seven months ago they would do in these communities: they tapped into a volunteer network using local party organizations, union rolls, gun clubs and churches. They backed it up with a blizzard of targeted appeals; according to the preliminary results of a survey done by the Center for the Study of Elections and Democracy at Brigham Young University, one representative home in Portage County, just outside Cleveland, received 11 pieces of mail from the Republican National Committee.

This effort wasn't visible to Democrats because it was taking place on an entirely new terrain, in counties that Democrats had some vague notion of, but which they never expected could generate so many votes. The 10 Ohio counties with the highest turnout percentages, many of them small and growing, all went for Bush, and none of them had a turnout rate of less than 75 percent.

For Democrats, this new phenomenon on Election Day felt like some kind of horror movie, with conservative voters rising up out of the hills and condo communities in numbers the Kerry forces never knew existed. ''They just came in droves,'' Jennifer Palmieri told me two days after the election. ''We didn't know they had that room to grow. It's like, 'Crunch all you want -- we'll make more.' They just make more Republicans.''

In hindsight, it seemed significant that Bouchard, months before, felt constricted enough by ACT's legal and financial realities to shift its focus, moving canvassers out of more contested counties and precincts and away from the business of trying to convert undecided voters. In the end, these were the voters Kerry needed. But Bouchard and his troops ran smack up against the inherent limits of a 527 in a presidential campaign. They could turn out the vote, but they couldn't really alter its shape.

Therein, perhaps, lies the real lesson from Ohio, and from the election as a whole. From the days of Franklin Delano Roosevelt, and especially after the disputed election of 2000, Democrats operated on the premise that they were superior in numbers, if only because their supporters lived in such concentrated urban communities. If they could mobilize every Democratic vote in America's industrial centers -- and in its populist heartland as well -- then they would win on math alone. Not anymore. Republicans now have their own concentrated vote, and it will probably continue to swell. Turnout operations like ACT can be remarkably successful at corralling the votes that exist, but turnout alone is no longer enough to win a national election for Democrats. The next Democrat who wins will be the one who changes enough minds"


If you read the whole article two points come up. One is that the Kerry campaign's internal exit polls showed Ohio turning toward Bush by the afternoon. Second the Bush votes were piling up in smaller counties with high turnout- counties less likely to be given large representation in the exit polls. Before I get bashed over this check my other limited posts. I am just trying to look at all of the evidence and help other DU'ers see what is out there. Proving fraud also means looking at the other reasonable alternatives and debunking them. In social science the most parsimonious answer is the best and we have to be able to prove that fraud makes the most sense over more acceptable electoral answers.

[TruthIsAll]

37. NICE TRY. Where were you before your FIVE posts?

tia

[BlueDog2u]

42. TIA

I don't think you answered the poster's challenge. Is it true that the Kerry campaign's own internal exit polls indicated a shift toward Bush in Ohio late in the afternoon?

[TruthIsAll]

43. I doubt it. They know. Kerry won. As sure as the suin rises.

Zogby. Harris. Massive registration. Young voters for Kerry 2-1. United Party. ABB. 40 Republican dailies turning away from Bush to support Kerry.I could go on and on.

And exit polls are VERY accurate. Ask Dick Morris.

I don't believe the crap that Repugs vote later than Dems. Pure Bullshit. Or that Repugs won't talk to exit pollsters. Pure Bullshit.
Or that the women and minorities voted for Bush. Pure Bullshit.
Or that Kery did Worse than Gore. Pure Bullshit.

I'm done.

tia

[BlueDog2u]

46. I am waiting to see the results of the recount

in New Hampshire before I decide how accurate or inaccurate exit polls can be. Meanwhile, I still think widespread fraud is a likely possibility and I appreciate your efforts, TIA, in bringing forth the numbers which tend to substantiate that proposition. The Caltech study appears to be crap -- there is no doubt that the work you and others have done is far superior. But a great deal does depend, doesn't it, on the contentious question of how accurage exit polls must be. Zogby is still on your side, and its interesting that he went to Europe to make his views known. Obviously that shows what we are all up against here in trying to find the truth at home. But dogma doesn't help us as much as rigorous analysis and logical discussion.

[flintdem]

44. Correction

If you read the NYT article it says that the raced moved to dead even by evening. Having re-read it I'm not sure who's exit poll numbers they were using (I'm not sure they say), but the numbers were moving toward Bush long before any 1am correction (if these are exit consortium numbers). Sorry for the confusion.

[Nederland]

61. Nice Try

Even people with thousands of posts disagree with you. Try addressing the posters points for a change.

[TruthIsAll]

62. For a change? If you check my replies, I do the best I can to indicate

what is behind my analysis.

But I cannot be drawn into a long discussion about the NYT, who won't even consider the POSSIBILITY, much less the PROBABILITY, of fraud.

[Nederland]

63. TIA

You are like a broken record. Your entire premise rests upon the assumption that the early exit polls were done correctly. Given that the company that produced the exit polls has stated that they were NOT done correctly, that's a really, really dumb assumption.

Garbage in, garbage out. Its not that hard to understand.

[TruthIsAll]

64. They were NOT done correctly? You BELIEVE that?

They only say that since 2000, when Bush is running. Or the Senate needs to be stolen, like in 2002. Remember that one?

Mitofsky has been doing exit polls for 25 years. When will he get it right, like he ALWAYS did prior to 2000?

Nederland, you are a cognitive dissonant.

[flintdem]

65. Even "done correctly" will produce a consistent error

Error for polling can all run one direction. We expect error to be random and in effect equally distributed. In a perfect world (which doesn't exist) that would be true. But sometimes for whatever reason errors all go one direction. The Washington Post reported yesterday that the 1992 Exit poll added 2.5% to Clinton's actual percentage (very similar to the 1.9% added to Kerry). The 1988 exit poll had Dukakis and Bush (senior) dead even when Bush actually won by 7%.

The article also notes one of the lesser know points of this election that Mitofsky warned the networks about the apparent Democratic bias mid-afternoon on Election Day and the networks ignored it.

Here is the link:

http://www.washingtonpost.com/wp-dyn/articles/A64906-2004Nov20.html

If we want to overturn the election we need real evidence (witness testimony, paper trails, proof of hacking) not statistical analysis of poll data or even the Berkley study. I know how social science works. Berkley could get its study reviewed and published and then a year later someone else publishes a study showing they missed an important variable(s) or an interaction term or used the wrong methodology and negate their results and then a year later Berkley publishes a rebuttal and on it goes. That's why MSM is ignoring this issue. The smoking gun needs to be more than statistical probability. That may be very frustrating to some but to say to tens of millions of Republicans AND Democrats (for whom this issue is completely off their radar screen right now) that a study shows the election was stolen (p=.05) will get us nowhere.

Just a note: I don't think responses in this forum should be personal attacks. We're all just trying to get at the truth. We just disagree at how we arrive at it.

[gandalf]

98. Some of your statements are not understandable. Please explain.

"Error for polling can all run one direction." What does that mean? That is not what statistics tells us if the sample is big enough (which is the case here). Your statement is simply not true.

"We expect error to be ... equally distributed"? What does that mean? I think we expect the error to be normally distributed.

"The smoking gun needs to be more than statistical probability." If statistics can prove that some results are not plausible, that this is a smoking gun that unknown influence factors come into play which have to be investigated.

[flintdem]

68. More Evidence of consistent error

Mitofsky writing in Public Opinion Quarterly in 2003 found that exit polls were twice as likely to overestimate the percentages for democratic candidates as to under estimate them. Even in 2003 he thought the reason was because republicans were less willing to respond -the same reason he is giving now. Here is a quote from his article:

"An inspection of within-precinct error in the exit poll for senate and governor races in 1990, 1994 and 1998 shows an understatement of the Democratic candidate for 20 percent of the 180 poll in that time period and an overstatement 38 percent of the time...The most likely source of this error is differential non-response rates for Democrats and Republicans" (p. 51).

I don't mean to beat a dead horse but we need more than exit polls and statistical analysis. These studies may show us where to look but then we need to find hard, courtroom type evidence (like a recount, testamony from conspiracy members, lines of coding, evidence of a hack). Without that there are too many plausable explanations to explain away the reports and studies.

Maybe its the nature of the beast but I also don't believe we have been visited by aliens or that there were any gunmen on the grassy knoll. Our evidence has to be strong enough to win over skeptics (like MSM).

[TruthIsAll]

73. I agree. Probabilities and statistics are a start. A very important start.

The initial clues are found in the numbers. That is where forensic auditors look.

The mathematics shows a clear case of either fraud or mistabulation - or both.

I am proud that my analysis is in agreement with the Berkely study and Professor Freeman.

And even that toe-sucker, Dick Morris, who said Exit Polls are virtually never wrong. He was probably so shocked he actually told the truth.

[Nederland]

69. Yes he has

Mitofsky has been doing exit polls for 25 years. When will he get it right, like he ALWAYS did prior to 2000?

Two points:

1) He didn't always get it right in the past.
2) He says he got it wrong this time. Why don't you believe him?

[flintdem]

39. Lurking

I didn't even know about this site until September. I tend to be moderate democrat (I supported Bonior in the 2002 primary because he was pro-life). If that doesn't make me democratic enough then I'll just keep lurking and be quiet. But I would argue that a search for the truth must take into account all alternative and all possibilities or it will not be taken seriously. Fraud to overturn a presidential election must be ironclad or it will always be consigned to the fringes!!!

[jdog]

40. Absentee ballots

have been discussed in other threads as having the potential for mischief as well (could this help explain Vermont?). I would imagine more than one method could have been used simply to make it more difficult to determine what happened. TIA's analysis is excellent toward once again proving that Bush's victory is highly suspect. When it's all brought together by someone with integrity AND financial backing, which the MSM USED to represent in this country, it will all come together and help throw Bush OUT.

[TruthIsAll]

52. Fine. So let's count ALL the votes. And let's get ALL the exit Poll Data.

Your tone is condescending. Your arguments are weak.

The fringes? Who are the fringes? Those of us who want ALL the votes to count. And who expect that all the votes have an audit trail? Thse are the fringes?

We know who is in the mainstream: the media which refuses to go after the story? Those who fight against transparency in voting.

Let's have the Whore Media entertain the possibility of Fraud.
Let's have a paper ballot to audit the votes.
Let's rebuke every Republican who has fought the paper audit -that means ALL of them..

Never forget: Gore won in 2000.
The Supreme Court stopped the count.
Tom Delay's thugs stopped the count in Dade County.

Finally, let's cut out the homilies which we hear all the time from those who can never face the truth - the cognitive dissidents.

Time to go.

[foo_bar]

55. project much?

Your tone is condescending. Your arguments are weak.

We project our own unpleasant feelings onto someone else and blame them for having thoughts that we really have.
source

[TruthIsAll]

80. Thanks, Sigmund. But cold, hard numbers have no feelings.

tia.

[foo_bar]

86. especially after they're pulled from you-know-where

[mrdmk]

41. The problem I have with this, if Bush's numbers were played with

in every state. There will be evidence turning up everywhere. Of course the vote is not completed yet and there is a clamp on the media. I really do hope that Kerry is President this year, but we do need hard evidence to turn-up. Keep digging!

[Cadence]

47. Ummmm there IS evidence turning up everywhere.

[ccarter84]

48. not positive, but CA example?

I posted this earlier but not sure if it made it above the
fold, so here is the summary (plain text)

Kern County CA

Pres. Votes         bush     gore/kerry 
2000               110,663  66,003 
2004               136,005  65,676 
2004 vote(gains)   25,342   -327 

From:
http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=201&topic_id=3530#4276

[recentdemocrat]

50. Sorry if this has been already asked and answered

but I have not seen it. I've been lurking here for a bit and registered just so I could clear up something. The original analysis of the red shift was posted by TruthIsAll a while ago, based on 4pm exit poll data. My question is where did this exit poll data come from? As far as I know, NEP has not released any data (other than to the subscribing members). This analysis has provoked a bit of controversy over at dKos where I usually am, and one of the biggest questions has to do with where the original exit poll data came from.
Again, if this had been answered in a previous thread, I apologize, but it's really driving me buggy. Any information would be most appreciated. Thanks.

[TruthIsAll]

51. It's from this Scoop thread

http://www.scoop.co.nz/mason/stories/HL0411/S00142.htm

[IndyPriest]

53. Also from Berkley and Freeman and...!

TIA and Scoop aren't the only sources making this case. Steve Freeman did. Berkley just did. How many other sources do people need??? There's no question about the "red shift" anymore. In courts of law, these types of statistical correlations are accepted as PROOF of fraud. The questions are: who did it, how did they do it, etc.

[TruthIsAll]

71. Welcome to DU! You are a breath of fresh air.

We need more of your kind.

You will sometimes wonder: who are the Democrats at DU?

[recentdemocrat]

72. Not disputing that there is something funny

around all these numbers. Berkley examined discrepancies between Evote/ non evote counties and documented the sources from where they drew the data to analyze. Dr. Freeman also documented where he got the exit poll numbers for his analysis. The only post I had for TIA's analysis didn't show where the exit poll numbers came from, so that's why I asked. Just trying to understand, that's all.

[recentdemocrat]

70. Thank you for the link (n/t)

[Be Brave]

56. I'm afraid the given interpretations of these probabilities are incorrect.

Dead giveaway: Why would a shift of +2.4 for Kerry have a probability of 93.7%, while a shift of -2.5% away from Kerry have a probability of only 5.3%? If the expected distribution of the shifts is normal, wouldn't equal shifts in opposite directions, i.e. positive and negative, have the same differential probabilities?

I have read post #7, and it doesn't really give a clear explanation of how the probabilities were calculated, except that EXCEL did the work. But how did EXCEL calculate the probabilities? Here is what I think it did: The shifts were calculated (Kerry votes minus Kerry exit poll), then it probably tried to fit a normal distribution to get a standard deviation. Note that the fitted normal distribution will be a bad fit and will probably fit well only the shifts up to plus or minus 3%. Or, it could be that the RMS was calculated from the shift data. In any case, given the standard distribution or RMS (note that the two are different), and assuming random errors and therefore an expected normal distribution of shifts, EXCEL calculated the INTEGRAL probability of getting a calculated shift for each state. And the integration was done such that the limits were from minus infinity up to the shift for which the probability was being calculated. That is why by the time the calculation gets to the most positive Kerry shift, the integral probability is close to 100% -- the total integral under the normal distribution is always equal to 100%. If the limits of the integration were instead from a given shift up to positive infinity, then things would be reversed -- the most negative Kerry shift will get assigned a probability close to 100%. This is why I strongly believe that the given interpretations for these probabilities are incorrect. These are not the probabilities we need.

If we are going to do things correctly, we really need the margins of error of the exit polls, state by state. It is from these margins of error that we can properly calculate the probabilities we need. Unfortunately, I have seen the exit poll data posted in these forums and in the scoop article without the margins of error. If we are to do things correctly, neglecting the margins of error is unacceptable. Folks, taking into account the margins of error will not weaken the case for anomalies, but will only strengthen it. Do not ignore them.

[Be Brave]

57. Okay, I just read post #4, and I think I know what happened.

Bottom line is, after reading post #4, I conclude that the calculation of the probabilities for the case where the Kerry shift is positive is wrong. The calculation for the case of negative shift is a start, but why not use the state-by-state margin of error?

The post says the average margin of error is 3.05% (I am assuming at 95% confidence, or 2-sigma confidence). To turn this into a standard deviation, which is a 68% confidence or a 1-sigma confidence, divide the margin of error by 2. Thus the average standard deviation is 1.55%. Like I said, that's a start. But if the margin of error for each state is available, why not use them all? And plot them together with the shifts!

The numbers are then inputted into the EXCEL NORMDIST function. The way I understand it from post #4,

prob = NORMDIST(votes%,exit%,average SD,TRUE).

I gather that the function treats exit% as the mean of the normal distribution, and it treats votes% as the value for which the probabity is to be calculated. This is where the problem begins. The problem is, the NORMDIST function integrates from minus infinity up to votes%, given the exit% as mean and the average SD. So given two states with the same shift, but in opposite directions, say KS and VA, if votes% is greater than exit%, as in the case of KS, the resulting probability will be larger than if the votes% is less than the exit%, as in the case of VA. That just gives the misleading impression that the case for the positive shift is a lot more probable that the case for the negative and equal shift. In actuality, the case of the positive shift should be just as likely as the case of the negative and equal shift. One way to get the correct probabilities is to calculate the absolute value of (votes%-exit%)/state SD, which I will call nsigma, and then modify the inputs to the NORMDIST function, NORMDIST(-nsigma,0,1).

Get the state-by-state margin of error, plot them, calculate the probability correctly.

[TruthIsAll]

58. Good points, but...MY PROBABILITIES ARE ACCURATE.

You say:

"The post says the average margin of error is 3.05% (I am assuming at 95% confidence, or 2-sigma confidence). To turn this into a standard deviation, which is a 68% confidence or a 1-sigma confidence, divide the margin of error by 2. Thus the average standard deviation is 1.55%. Like I said, that's a start. But if the margin of error for each state is available, why not use them all? And plot them together with the shifts"


I could have done that. But I used the average MOE to keep things simple. And it is more conservative, since most of the battleground states had over 2000 polled, which means a 2% MOE. Mitofsky, the head of the Exit Polling group, claims a 3% MOE, which is of course untrue for exit polls. They are much more accurate; in fact, the MOE is LESS than 1%.

SO THE PROBABILITIES ARE MUCH LOWER THAN I HAVE SHOWN HERE. THE CHANCES OF FRAUD ARE MUCH GREATER. BUT NO NEED FOR OVERKILL. LETS BE CONSERVATIVE, TO SATISFY THE NAYSAYERS AND MITOFSKY.

I used the Kerry's two party vote and the actual vote as input to the Normal distribution, which then calculates the probability that the difference between the two would occur. The probabilities make sense intuitively and mathematically.

A deviation of 2% is much more significant where there is a 50/50 split than an 80/20 split. The calculations prove it.

Just look at the relationship between the deviations and the probabilities. Lets focus at the 3% deviations in these CRITICAL states:

State Dev Prob
PA -3.3 1.5 < This is beyond the 3% MOE. Prob is 1.5% this could occurr by chance alone. So there is a 98.5% probability it did NOT occur due to chance.

OH -3.1 2.3 < Like Florida. The gremlins were working hard.

FL -3.0 2.5 < This is at the cusp of the 3% MOE. There is a 2.5% probability that this deviation could occur by chance alone. This is EXACTLY what one would expect: 95% CONFIDENCE +2.5% Bush Tail = 97.5%


I appreciate your comments.

[TruthIsAll]

59. FURTHER PROOF AS TO THE ACCURACY OF THE CALC: LOOK AT MONTANA

There was a ZERO shift. The probability is 50.9%. That's where it should be. Right in the middle of the confidence interval. It did NOT move to Bush; it did NOT move to Kerry. It's right where you would expect it to be.

In other words, there is no case for mistabulation or fraud in Montana.

[Be Brave]

75. TruthIsAll, a zero shift will get a probability of 50%

NO MATTER WHAT STANDARD DEVIATION YOU USE AS LONG AS THE DISTRIBUTION IS NORMAL.

[Be Brave]

74. Did you address the large probabilities you gave to the positive Kerry

shifts? If you did, then I missed in it your post.

Can you post the margins of error fot all the states if you have that data?

Why do you say a MOE of 3% is untrue for exit polls?

[TruthIsAll]

79. Feast your eyes

"Did you address the large probabilities you gave to the
positive Kerry shifts? If you did, then I missed in it your
post.

You mean all NINE of them?

Can you post the margins of error for all the states if you
have that data? I have seen.

State	BUSH	KERRY	#Resp	Time	RShift	K/K+B	Moe	Std
Ala	58.1	40.5	730	12:17AM	4.2	41.08%	3.63%	1.85%
Ak	57.8	38.8	910	01:00AM	4	40.17%	3.25%	1.66%
Az	52.8	46.7	1859	12:19	2.5	46.93%	2.27%	1.16%
Ar	52.9	46.1	1402	12:22	1.1	46.57%	2.62%	1.34%
Ca	46.6	54.6	1919	12:23	(-)1.5	53.95%	2.24%	1.14%
CT	40.9	57.7	872	12:22	3.4	58.52%	3.32%	1.69%
(CT	44.4	54.7	872	12:53)	0.2	55.20%	3.32%	1.69%
DC	8.2	89.8	795	12:22	0.3	91.63%	3.48%	1.77%
De	40.7	57.3	770	12:22	4.8	58.47%	3.53%	1.80%
Ga	56.6	42.9	1536	12:22	2.2	43.12%	2.50%	1.28%
Hi	46.7	53.3	499	12:22	(-)1.2	53.30%	4.39%	2.24%
Id	65.7	32.9	559	12:22	2.6	33.37%	4.14%	2.11%
Il	42.4	56.6	1392	12:23	1.6	57.17%	2.63%	1.34%
In	58.4	40.6	926	12:22	1.6	41.01%	3.22%	1.64%
Ks	64.5	34.1	654	12:22	(-)2.7	34.58%	3.83%	1.96%
Ky	58.4	40.2	1034	12:22	0.9	40.77%	3.05%	1.55%
La	54.7	43.9	1669	12:21	2.1	44.52%	2.40%	1.22%
Me	44.3	53.8	1968	12:22	0.8	54.84%	2.21%	1.13%
Md	42.3	56.2	1000	12:22	0.5	57.06%	3.10%	1.58%
Ma	32.9	65.2	889	12:22	3.7	66.46%	3.29%	1.68%
Ms	56.5	43	798	12:22	3.3	43.22%	3.47%	1.77%
Mo	52	47	2158	12:21	1.5	47.47%	2.11%	1.08%
Mt	58	37.5	640	12:22	(-)0.3	39.27%	3.87%	1.98%
ND	64.4	32.6	649	12:22	(-)2.4	33.61%	3.85%	1.96%
OK	65	34.6	1539	12:23	0.8	34.74%	2.50%	1.27%
Or	47.9	50.3	1064	12:22	(-)1.3	51.22%	3.00%	1.53%
RI	34.9	62.7	809	12:22	3.4	64.24%	3.45%	1.76%
SC	53.4	45.1	1735	12:24	4.4	45.79%	2.35%	1.20%
SD	61	36.5	1495	12:24	(-)1.8	37.44%	2.53%	1.29%
Tn	58	40.6	1774	12:23	(-)1.7	41.18%	2.33%	1.19%
Tx	62.2	36.3	1671	12:22	(-)2.0	36.85%	2.40%	1.22%
Ut	68.1	29.1	798	12:22	2.5	29.94%	3.47%	1.77%
Vt	33.3	63.7	685	12:22	5.2	65.67%	3.74%	1.91%
Wa     44	54.1	2123	12:38	1.6	55.15%	2.13%	1.09%
WV	54	44.5	1722	12:24	1.8	45.18%	2.36%	1.20%
Wy	65.5	30.9	684	12:22	2.7	32.05%	3.75%	1.91%
 

Why do you say a MOE of 3% is untrue for exit polls? 
Here's why:

a) they poll more than 2000 in the battleground states (see
the MOE's above) which brings the MOE for a standard (not an
exit poll) down to 2%.

b) On top of that,  exit polls are INHERENTLY MORE ACCURATE
THAN STANDARD POLLS. They are correct to within ONE HALF OF
ONE percent in the last 3 German elections. See Professor
Freeman's updated analysis. 

[Be Brave]

83. Why do you resort to sarcasm rather than correct you calculations of

the probabilities in the cases where there were positive Kerry shifts? Until you correct those, the fact remains that your posted plot, as it is, misleads everyone who reads your post. Do you really expect anyone to believe that a positive Kerry shift of 2.4% has a chance probability of 93.7%, while a negative shift of -2.5% has a chance probability of 5.3%? If your analysis cannot stand the logical scrutiny of someone who's on Kerry's side, how can it stand scrutiny or worse, spin, from someone on the other side?

And if you do decide to correct the probabilities for the cases with positive Kerry shifts, then you will find that around 21 cases with negative shifts will have probabilities larger than the smallest probability that can be found in the cases with positive shift. This is easily seen in the plot in your post.

According to the data you posted above with the number of respondents and the MOE, many of the states do have a margin of error of 3% or larger, and many of the states have respondents of less than 1000. Why do you argue with that, and why don't you use the state MOE instead of doing an average? If they are "correct to within one half of one percent", then how come there are positive Kerry shifts larger than 1%?

You said in your previous post that your probability calculations are accurate. I strongly disagree. If you insist in saying that the exit polls are accurate while the data-takers' MOE are not, and you are just making your best guess at what the appropriate MOE should be, in other words introducing your own bias, then how can your calculation be accurate? It is at best your best estimate given your own bias on the data. I cannot convince myself that your best estimate is better than the calculation that would result if the actual margins of error are used. If you try to publish your analysis, you will no doubt get the same question.

[TruthIsAll]

89. BE BRAVE,YOU JUST DON'T GET IT: BETTER READ THIS:

Posted by Be Brave
the probabilities in the cases where there were positive Kerry shifts? Until you correct those, the fact remains that your posted plot, as it is, misleads everyone who reads your post. Do you really expect anyone to believe that a positive Kerry shift of 2.4% has a chance probability of 93.7%, while a negative shift of -2.5% has a chance probability of 5.3%? If your analysis cannot stand the logical scrutiny of someone who's on Kerry's side, how can it stand scrutiny or worse, spin, from someone on the

MY RESPONSE:
You do not yet understand the purpose of the analysis, so I'll explain it to you again. You fail to grasp the basic logic of the probabilities.

First of all, look at the totality of the moves - 41 of 50 to Bush.
Nine states moved to Kerry.

For EACH STATE I have calculated the probability that the deviation in the Kerry/(Kerry+Bush) percentage from the Exit Poll to BUSH was NOT due to fraud or mistabulation. Focus on the word NOT.

Where the probabilities are LOW (on the LEFT) then it means that the probability of fraud or mistabulation in FAVOR of Bush is HIGH (e.g. FL, OH, PA).

Conversely, where the probabilities were HIGH (on the RIGHT) the probability of fraud or mistab is LOW, like in KS or OR.

In other words, in the vast majority of states, the probabilities tend toward statistical probability of fraud or mistabulation (again, for Bush).

The purpose of the graph/analysis is to ut a quantitative measure on the probability of fraud on the part of BOTH Bush AND Kerry. It just happens that there are MANY more instances (states) of probable fraud for Bush than there are for Kerry.

I hope that clears it up for you.

[reality_bites]

85. MOE is a function of sample size,

not the outcome of 3 polls in Germany. A poll is a poll is a poll. You don't calculate MOE differently for exit polls. More about calculating MOE here:

http://www.westgroupresearch.com/research/margin.html

Exit polls are NOT INHERENTLY MORE ACCURATE THAN STANDARD POLLS. You mention anecdotally 3 German exit polls accurate to 1/2 percent. So what? I can point to 3 elections here in the US where exit polls were wrong by 3% to 8%. For example, go here:

Unfortunately for your argument, exit polls have leaned heavily to the democrat side for every presidential election since 1988:
http://www.washingtonmonthly.com/archives/individual/2004_11/005178.php

I've been reading your posts and looking at your chart for several days. What you are trying to prove may in fact be true, but you aren't proving, or even advancing your position.

Any arguments that start from the exit poll and try’s to argue that fraud occurred are a waste of time. As shown by the link above, exit polls are not all that accurate. I don't know how the notion that exit polls are accurate got started but it is unfortunate.

The Berkly study is good since it does not rely on exit polls to make a case for fraud. Please devote your time trying to advance the Berkly thesis. You will have more success if you do.

[Be Brave]

77. BTW, I don't get it.

Their 4 PM exit polls are good enough for you, but their 4 PM margins of error aren't? What is your criteria for picking and choosing?

[TruthIsAll]

78. You sure don't get it. No cherry-picking here. Just the facts.

41 of 50 states to Bush. The odds against that alone are almost 2 million to one.

But I'm just getting started. How many were beyond the MOE?
The MOE is 1% in exit polls, 2% in regular polls of 2000 respondents.

But I'll use a nice fat, conservative 3% just to shut you up.
And the odds of it being fraud are still astronomical.What are the probabilities for PA,FL,OH etc.?

In 2000, they stole just enough to win. This time, they stole more than he needed to get a mandate, and make it appear it wasn't even close. Across the board.

The 51-48 Kerry win was reversed to a 51-48 Bush theft. This will be proven in due course. Or we can just kiss our democracy goodbye.

[missouri dem]

81. Lots of people seem to disagree with you. Must be getting warm

around here. Keep up your work.

[Heewack]

82. Again you are using the same flawed methodology.

The probability of a shift in this case is 50/50. The exit poll numbers cannot be used as the null.

[reality_bites]

88. ditto

[Be Brave]

84. Shut me up? Tsk, tsk. Getting nasty, aren't we?

Try to keep it civil so as not to get this thread locked, ok?

It is not the data I don't get, it's your analysis methods.

Regarding the 41/50 odds, I'll reserve my comments and give them in that other thread.

Having a goal of shutting me up is not a very good one. If I honesty think there is a flaw in anyone's analysis, I will speak up. That is how I was trained. It does no one any good to not criticize something which is incorrect.

Why are there only 34 states in the data with MOE you provided?

Let me again make clear what I think is the correct thing to do: First, you need to correct your probabilities for the cases with positive Kerry shift. Second, use the state-by-state margins of error instead of the average MOE. The probabilities for PA, FL, OH, will probably get smaller (though you supplied no data for these states in post #79), yes, but they will probably be no worse than the probability that you currently have assigned for DE. If you use the state-by-state margins of errror, some probabilities will decrease, and some will increase.

[TruthIsAll]

90. I am not trying to shut you up. PLEASE READ POST #89.

You are misinterpreting the analysis. Go to Post #89.

There are only 34 states with MOE because that is all the data I had. So I used the average 3%. That is being CONSERVATIVE. The Exit Poll MOE's were not available for many big states. Suspicious, no?

The big states in the 34 had an MOE of 2% because there were AROUND 2000 respondents. But the MOE is based on pre-election polls- not exit polls. For Exit polls, the MOE is less than ONE percent. At least it has been historicall, except for 2000 and 2004.

Therefore the probability of mistabulation and/or fraud are significantly HIGHER than those presented here, especially in the battleground states where 2000 respondents were exit-polled.

Do you follow?

Now, do us all a favor. Read post #89. That should clear it up for you.

[nolabels]

92. Wonderful fellow poster TruthIsAll

I don't know how or why you can have the patience with these foolish naysayers(or probably other), but it sure is a comedy act they are putting on, you, TruthIsAll really being hero in my book and thanks for another keeper :thumbsup:

They always say don't feed the trolls, but I guess sometimes they work good for straw-men even though they don't know they are doing it.

Thanks TruthIsAll

[reality_bites]

87. You seem to have no serious input on this issue.

For example,

"The odds of winning 41 of 50 states is 2 million to 1."

Well, Reagan won 42 states in 1980. Reagan won 48 states in 1984. What are the odds of that ? Whatever the odds, IT HAPPENNED, so I really doubt your 2 million to 1 claim for 2004 results.

"The 51-48 Kerry win was reversed to a 51-48 Bush theft."

Please provide proof. Exit polls can't be used to make a case since they are historically unreliable (the claim that exit polls are NECESSARILAY reliable is fiction.)

I get frustrated reading DU. I would say that 95% percent of posts like yours purporting to use statistics to make a point simply aren't valid. Use the Berkley report as springboard. It is the only study I've seen so far that has any weight.

[TruthIsAll]

93. The more posts you write, the deeper the hole you dig yourself in.

Do you realize that you have become irrelevant to the VAST majority of Duers who are following this thread?

Anything you say from here will be discounted - except by your fellow trolls, of course.

Can't argue with you, because of one or more of the following:
1. You are just ignorant.
2. You suffer from cognitive dissonance.
3. You are a mascochist who loves to have his head beaten in every time.

Enjoy your short stay at DU. We have seen many of your kind come and go. Soon you will tire of playing the joker.

Bye, Quixote.

[Nederland]

94. TIA

The problem here is your refusal to admit you are wrong. Ever. Perhaps you need a reminder about just how wrong you can be:

Mark this thread. I said it before. I will say it again. Bush won't run.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1008784


Why don't you just admit your mistake and recalculate your graph? Its the honest thing to do.

[TruthIsAll]

99. Why don't you just give it up, you have lost . Your diversions are

nothing more than that.

They lack substance. They lack meat. They do not refute a damn thing I have done here. Others much more knowledgable than you back ME up.

Nice try, picking an old post of mine. I do not apologize for that post. I was wrong then.

I am right now. The evidence accumulates even as you try to plug the dikes. But you can't stop the truth.

[Kathy in Cambridge]

91. This thread is like troll flypaper

it's awfully smelly around here.

[TruthIsAll]

96. For those who are waiting for Bev Harris on CNN tonight.

See the shift.

[TruthIsAll]

97. Kick for PM crowd

tia

[formernaderite]

100. It's sure depressing how close some blue states were...

...this time. In fact the repubs gained points in too many states. I hope we can reverse this trend before 2006.

[Truman01]

101. These graphs and others prove there is smoke. . .

We have to find the actual fire to win.

[TruthIsAll]

102. Well, Jesse Jackson said on CNN: Bush won it fair ansd square.

Bizarro universe.
Black is white.
White is black.
We are through the looking glass.

It's Nov. 24.
Still waiting for the other shoe to drop.

[super simian]

103. OMFG

Guess Kerry missed the boat since he didn't protest the "day after the election."

http://www.michaelmoore.com/words/index.php?id=372emblin
Ohio Democrats Offer Support for Recount Effort

November 24th, 2004 4:17 am
By Brian Faler / Washington Post

Quote:

A spokesman for Ohio Secretary of State J. Kenneth Blackwell dismissed the candidates' allegations. "We did not experience any significant irregularities," Carlo LoParo said. "Half of the world's media was in Ohio monitoring the election. I would think that if there were significant issues of election irregularities, you would have reported them the day after the election.