Newly Released EXIT POLL DATA - requires Some TIA Smoking Gun Treatment

[althecat]

I actually obtained this data a while ago but so far it has proved fairly useless. Mainly because what we wanted to look at were state by state polls.

http://www.scoop.co.nz/stories/pdfs/Mitofsky4zonedata/

At the above link you will find the original Mitofsky/Edison reports for the :

- House & Presidential races
- in vertical and horizontal format (not sure what the difference is)
- For each of the East, West, Mid - West & Southern zones.

Each report comes in three versions. The 4pm run the 8pm run and the final 1.30pm Nov 3. corrected run.

I did not release this data earlier as it was provided by Mitofsky in response to a request from me to pay for the data. They subsequently said the state by state data which is what I wanted would cost $10k which was way out of my price range so we decided to scrap the deal.

However in light of the latest discoveries from ClintCooper and TIA (See this thread for details - http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=214060&mesg_id=214060 ) regarding the how did you vote last time around questions - and the Conyers letter to Mitofsky and the NEP's response now seems like as good a time as ever to get the DU number crunching time to have a look.

In terms of my own wish list. I would like initially some of you folks to have a look at this data and perform the TIA SMOKING GUN analsysis on a four zone basis. On the face of things the exit poll data was more accurate in the West than the East... so this should bear that out.

al

[RaulVB]

1. Kick (n/t)

[Helga Scow Stern]

2. See which thread for details? Or am I missing something?

[althecat]

4. Sorry about that.. edited now and link added

This thread
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=214060&mesg_id=214060

[bleever]

3. Ya sure you betcha I'm kickin'.

[Bill219]

5. ???

Do we know if anyone here is working on this yet?

[Firespirit]

6. It's a bit intimidating.

I have to be up at 6:00 tomorrow, and my eyes hurt just looking at the list of PDFs and what's in them.

Bookmarking after this post is sent. If no one takes it by tomorrow noon, I'll do it. I'm an engineer and I've had statistics.

[Bill219]

7. KICK

for anyone that has not had a chance to look this over yet before it disappears onto the next page

[Time for change]

55. This is what I see as the key parts

Some of the files are pretty difficult to understand, as (it appears to me) some of them represent the numbers after they've already been "adjusted", and others seem to be a combination of adjusted and unadjusted numbers. After looking over the files, here's what I see as the salient points:

In the parent directory, there is an article by Jonathon Simons, the guy who originally gave us much of the unadjusted data:
http://www.scoop.co.nz/stories/pdfs/PopularVotePaper181...
This is an excellent article that points out that Kerry won the popular vote, according to the exit polls, by 2.6%, and the probability of that large of a discrepancy between this and the official vote tally (Bush by 2.8%) happening by chance is almost a million to one.

The key exit poll in the list is
http://www.scoop.co.nz/stories/pdfs/Mitofsky4zonedata/U...
This confirms the basic data in Simons' paper. It has Kerry with a national 3 point lead at 7:33 p.m. on November 2, with a sample size of 11,027. It may not be the final poll, but if not it is very close.
It also has lots of intersting breakdowns by demographic categories.

[pacalo]

8. All I can say is...

I bow before you, althecat, RaulVB, & TruthIsAll! Good work!

[americanwoman]

9. Horizontal and Vertical Explained

FYI:

Horizontal figures consider a characteristic as a "whole" (100%) and break down the percentages for each candidate. e.g. of 100% men, 41% went to * and 57% to Kerry.

Vertical figures consider the candidate's votes as the "whole". e.g. of 100% Kerry votes, 45% were from men and 55% were women.

(These are actual figures from Region=EAST.)

[Bill Bored]

10. Thanks!

I am interested in looking at those job-loss questions to see if the recently unemployed, who voted strongly for Kerry, also voted earlier in the day. This might have skewed early results toward Kerry.

So you say the first 2 are "raw data" and only the latest one has been tweaked to match vote tabulation?

[ClintCooper2003]

11. This also proves that Mitofsky is a FUCKING LIAR. He said....

repeatedly that those "early polls" weren't weighted. Lying bastard.

[ClintCooper2003]

12. Holy Shit! The exit poll from the South shows Kerry won 64% of the

Hispanic vote!!!

[ClintCooper2003]

15. Another thing these exit polls show: Kerry's margin of victory was...

INCREASING, not decreasing as the day grew later.

[EFerrari]

24. Of course he did. We odiamos el mono! n/t

[EFerrari]

26. kick

[georgia10]

46. Question

Do you have a link to where he said they were not weighted?

Thanks!

Edit: my earlier question was how to tell they were weighted. i see where it says they were weighted now :)

[berniew1]

98. Weighting is part of the standard methodology in doing Exit Polls

[ClintCooper2003]

13. kick

[bleever]

14. Kick.

[understandinglife]

16. HAS SOMEONE DOWNLOADED ALL THESE FILES????

If I don't have a response by 1230am PST, I'm doing it. Just cannot risk having access denied.

Peace.

[pointsoflight]

18. Got em. n/t

[understandinglife]

21. Great. And, I've got a 'back up'!! Just couldn't bring myself to wait...

....for a response ;-)

Happy New Year!

[Al-CIAda]

28. Downloaded if needed.

[Bill Bored]

20. got all the Pres Ones here n/t

[thanatonautos]

35. Got em. n/t

[garybeck]

17. what is meant by nov3 "corrected" ???? n/t

[ClintCooper2003]

22. It means they changed the data to conform to the results. A big no-no.

[euler]

114. Big no no ?

Refining exit poll numbers by factoring in actual numbers as they are received from the precincts is standard. It's been done that way since 1967 and will continue to be done that way.

http://www.mysterypollster.com/main/2004/11/the_difference_.html

[americanwoman]

32. Where did you see the "corrected" wording? n/t

[ClintCooper2003]

19. kick

[ClintCooper2003]

23. kick

[bleever]

25. Sleepy time kick.

[Stand and Fight]

27. For the morning crowd... I'm off to bed.

:kick:

[americanwoman]

29. Strange Percentages

It's interesting to look at the percentages for just the new
respondents for each revised set of numbers. The second set of
numbers for South and East are wildly off, as are the third
set for each region. Don't know how significant this is, but
...


Survey   Bush     Kerry     Bush       Kerry
                           (just new) (just new)
============================================
east #1  41.00%   58.00%    41.00%    58.00%
east #2  40.00%   58.00%    34.74%    58.00%
east #3  43.00%   56.00%    50.69%    50.87%

midw #1  49.00%   50.00%    49.00%    50.00%
midw #2  49.00%   50.00%    49.00%    50.00%
midw #3  51.00%   48.00%    57.43%    41.57%

south#1	 54.00%	  44.00%    54.00%    44.00%
south#2	 54.00%	  45.00%    54.00%    47.41%
south#3	 58.00%	  42.00%    88.77%    18.92%

west#1   45.00%	  53.00%    45.00%    53.00%
west#2	 45.00%	  53.00%    45.00%    53.00%
west#3	 49.00%	  50.00%    69.15%    34.88%

all #1	 48.00%	  51.00%    48.00%    51.00%
all #2	 48.00%	  51.00%    48.00%    51.00%
all #3	 51.00%	  48.00%    63.56%    35.44%

[Helga Scow Stern]

31. That does look strange. n/t

[mgr]

49. Not really

What the final tally suggests is that new voters in the south region voted overwhemingly for Bush, and would affect the overall outcome. What you need to include is the number of respondants to each question to figure this out. If the total number of respondants, does not work out, then you have something to scream about (of course, they have also been adjusted somewhat by weighting).


Mike

[k8conant]

61. My calculations match yours...

This shows the amount Mitofsky & Co. had to weight that additional data to get it "right".

[rdmccur]

81. Hey, I wonder which state

in the south made up most of that "gain" for Bush?
Give you 3 guesses.

[pointsoflight]

30. Dispelling some myths

Bush gained 6% in the exit polls going from Mitofsky's 7:30 p.m. numbers (n=11027) to his final numbers (n=13660), reversing a 51-48 Kerry lead into a Bush 51-48 lead.

Some common myths dispelled:

1. The earlier exit poll numbers oversampled women.

Wrong. The 7:30 p.m. exit poll numbers consisted of 54% women and 46% men. The final exit poll numbers included the exact same percentages.

2. The earlier exit poll numbers oversampled minorities.

Wrong. The 7:30 p.m. exit poll numbers consisted of 11% blacks and 9% hispanics. The final exit poll numbers included 11% blacks and 8% hispanics.

3. The earlier exit poll numbers oversampled democrats (i.e., republicans voted later in the day).

Wrong. The 7:30 p.m. exit poll numbers consisted of 38% democrats, the final exit poll numbers consisted of 37%.

4. The earlier exit poll numbers undersampled rural voters.

Wrong. The 7:30 p.m. exit poll numbers consisted of 16% rural voters, same as the final exit poll numbers.

[EFerrari]

34. Thanks -- clearest statement on this I've seen! n/t

[althecat]

39. cool

[sandnsea]

73. One switch 45-59

IIRC, the one switch I noticed between these exit polls & the final CNN polls was the 45-59 age bracket. I don't know what that means, but I do know Oregon's 45-59 bracket went Kerry and Oregon finished with a strong vote for Kerry. Just throwing that in the mix.

[Bill Bored]

89. Earlier or Earliest?

I haven't checked them all, but it is true that the earliest one (4 PM) did oversample women (58% vs. 54%). While this doesn't explain the 6% shift for Bush after 7:30, it does explain how someone could take something true out of context and make it into a myth.

[shiina]

33. simply

Kerry's numbers, percent of vote
                                       final
           3pm raw     7:30pm raw     "corrected" 
          ---------   ------------  ----------------- 
East          58           58               56 

Midwest       50           50               48  

South         44           45               42 

West          53           53               50 

ALL           51           51               48 

1st column - the poll is about 60% done. (8347 of 13,660
respondents)
2nd column - the poll is 80% done.  (11,027 of 13,660
respondents)
3rd column - "corrected" to match the official
totals

So looking at columns 2 and 3 give us our best look at the
difference between the exit polls and the official results.  

Notice how columns 1 and 2 nearly match...
The 80% of us who voted before 7:30 and the 20% who voted
after had extremely different voting patters.  This does not
make sense.  

I'm trying to figure out the math on this...what precentage of
the last 20% would have had to have voted for Bush...hold
on...

[americanwoman]

58. I think my post #29 answers your question.

Look at the third posting for each region and the "just
new" columns which show the B/K percentages for just the
new respondents for each survey.  Overall, for the final 2633
respondents, B* would have to have gotten aprox 64%. Breaks
down like this:

          # new responses                          
Region     in final survey   Bsh     Kerry   Bsh%    Kery%
          (from 2nd survey)
=======   ===============    ======  ======  =====   =====
East      810.24             410.73  412.18  50.69%  50.87%
MidW      871.74             500.67  362.35  57.43%  41.57%
South     512.64             455.06   97.01  88.77%  18.92%
West      437.28             302.40  152.55  69.15%  34.88%

ALL      2633.00            1673.64  933.03  63.56%  35.44%

If these numbers are to be believed then that last set of
voters in all regions but the East would have had to favor B*
heavily. 

[passy]

36. kick

[shiina]

37. So, if we play with these numbers...

Now, read this carefully...I can't think of a way to explain it very well.

We have the 7:30 results, when 80% of the poll was completed. Now, you cannot get an accurate final result if your data is wrong when it's 80% done. So in order for the final exit poll results to be accurate, the 7:30 numbers would have to be accurate as well.

Now, we don't have the final exit poll results, but we know what they would have to have been if * really won and the exit poll is accurate: they would have to match the official results. In order for this to happen, the last 20% (2,643) of the voters they polled would have had to have responded like this:

Kerry : 934 votes - 35%
Bush : 1,704 votes - 64%

Of course, if all are accurate, then this reflects the voting patterns of the people all across the nation.

In other words, either the exit poll was completely wrong, the official results are wrong or ... of the people who voted after 7:30, only about 35% of them voted for Kerry.

[althecat]

38. It's worse than that.... THE TIA SMOKING GUN ANALYSIS MEANS

20% (approx) of all voters are new voters

Of these voters according to the exit polls 65% voted Kerry and 35% voted Bush.

According to the actual poll 35% of these voted Kerry and 65% voted Bush.

Either:

1. New voters lied somewhat uniformly as a group about who they really voted for. In which case why?
2. The pollsters accidentally interviewed a sample who were 40% away from reality.

or

3. The poll was right and the result was wrong.

[EFerrari]

40. 20% were new? I had no idea we did that well.

[Peace Patriot]

43. Must of been that fabulous BushCon grass roots volunteer GOTV effort,

spurred by all those wildly enthusiastic, spontaneous crowds of poor black working people who showed up at every Bush whistlestop.

George Will wrote a very erudite column on this demographic long before the election, based on the recently discovered journals of a southern plantation owner circa 1850 who had traveled to Africa, and who enumerates the ways in which the lives of his slaves are far superior in food, shelter, clothing, medical care, useful occupation, literacy and religious orthodoxy, than their African counterparts, and asserts that, all things considered, slavery was a step up for the African race, a heretofore unspoken belief of many African-Americans, passed down through the generations from the original slaves, parent to child, in the manner of their silent acquiescence to slave and post-slave conditions within the United States, and only recently articulated by a new generation of brilliant, young, Harvard-educated, African-American scholars, funded by the American Enterprise Institute's "Project for a New American Century of Raptured Academics," who, among other things, have produced convincing analyses of certain obscure lyrics in African-American work songs that clearly bespeak stolid acceptance, Christian forbearance, and even joy, at having been rescued from the dead end of African tribal society.

Will asserts that African-Americans have only recently been able to acknowledge the benefits of slavery, and predicts that poor, African-American working people will vote for George Bush in numbers that will raise eyebrows among Democratic demographers who have always and erroneously presumed that African-American slaves hated their slave-owners.

The wildly enthusiastic poor, African-American, working class crowds that greeted President Bush at every campaign stop, and the, to some, astonishing, number of votes they cast for George Bush late in the day, after work, on November 2, 2004, have thus confirmed Will's hypothesis, and shows to go ya what raptured scholarship and very large amounts of money can do.

[PlayOn]

62. Are you serious?

I'm not sure whether this is meant to be a sarcastic post or what, but the premise of this post is ludicrous and smacks -- hell, reeks -- of the white supremacist notion that Africans were done a favor when whites enslaved them. Blacks considered slavery advantageous? Really? Then why did they try to escape their slavery, and/or kill their slave masters? Were there some blacks who accepted their enslavement and even rationalized it? Probably. But does that signify that blacks generally "liked" their slave masters and felt grateful at being enslaved? Well, I'm sure there are some victims of rape who weren't terribly traumatized by the ordeal -- hell, some may have even liked it. Ergo, rape victims don't dislike being raped as much as conventional wisdom would have us believe. Right? And George Will speaking for blacks -- about any issue -- is tantamount to Jesse Jackson being a spokesperson for conservative white men. You think blacks are now accepting slavery as something beneficial? Okay, try this: approach a black person and say "Slavery was good for black people, right?". Be sure you can deal with the outcome before you approach, though. And "wildly enthusiastic poor, African-American, working class crowds" greeting Bush? You're full of shit. And that's funny how you refer to him as "President Bush", seeing as how he never won the presidency legitimately. He's not the president: he's a thief. Any good progressive knows that. There's something funny about you...

[Lizzie Borden]

103. George Will=historical revisionism.

[euler]

115. Or.

4. The exit poll was not taken from a random sample.

[Truman01]

41. Excuse me but how did they get proprietary numbers?

How can we trust the sourse for the numbers? Or do we care about that since they jive with what we "know to be true."

Silly evidence rules and all.

TC

[lonestarnot]

42. I am again outraged! WTF! I see that people have already sent this

off to Conyers. I knew it was so, but to SEE it in black and white simply boils my blue blood!

[vj68]

44. kick

:dem::dem::dem::dem::dem:

[Amaryllis]

45. Question for all you numbers crunchers: I keep seeing data analysis

threads, but does anyone ever send this excellent work to Conyers or anyone who can make use of it in building a case? Or does it just stay in the DU forum?

[mgr]

47. Some questions.

I think many will agree with me that these numbers give us more questions than answers.

I noted that every page has the term "weighted" on it, I can only take this to mean that these are not the 'raw' exit poll numbers.

Times given are meaningless, since I cannot tell if they are EST, GMT, PST, MST, CST. A 7:30 exit poll EST is a 4:30 exit poll PST, a 5:30 poll MST, and a 6:30 poll CST. When were the majority of votes cast and counted, cumulatively as well as hourly in each of these four regions. When were the absentee and early voter ballots counted?

I also do not know what states are included in Mitofsky's four regions.

I would certainly do a goodness of fit test comparing the election results to the exit polls, and see if the 'raw' sample of ~20,000 voters would plausibly fall with a population of 100 million plus votes within two standard deviations. What has been done with this data to date is dubious since it has already apparently been massaged once, and gets the once over a second time (though I suspect that the correction term is to address the non-election day votes, if what they are trying to do is create a predictor of outcomes for future elections).

Mike

[mgr]

48. however,

Thinking this over a bit more, if the absentee/early voter ballots show a Kerry majority through out the country, then there is one more nail in Bush's coffin, as this would undermine an argument that it was the absentee ballots or early voters that threw the election Bush's way. It's still circumstantial though.

Where I suspect the problem will lie is with the 7:30 exit poll for the Central Time Zone, which would be 6:30 their time. It would be critical to demonstrate that the vast majority of voters that voted on election day voted prior to this time. What I suspect may be the problem with Mitofski's technique that he addresses through weighting is not simply the percentage,which the sampling can capture well, but the volume. In other words, he may have only sampled 800 people after 7:30 EST, but if they represent 30% of the turn out, they need to be weighed differently than the 11,000 that represent 70%.

Mike

[ClintCooper2003]

50. The trend is clear. Kerry's lead was staying the same or increasing

as time went on.

[mgr]

54. you cannot say this w/o more info

I see your point, but that would only be accurate at best with the east coast results that would be in the 7:30 EST window, and it would be highly unlikely that 30% of the vote was in 30 minutes. However, if you open that time window to the Central, Mountain, or Pacific Time you are far open to a greater percentage of the total vote being missed.

Here is how I see it, you have a large number of voters casting votes before they go to work, or after they get off. You may have a trend for Kerry throughout the day that gets swamped when all the after work voters are Bush supporters.

What I would do is look at the adjustment on 11/3 to the Eastern set, and see how that changes. What you may observe as a Kerry trend may in fact reflect the east coast only.

What this all means is that Mitofski seems to have an east coast bias to his sampling methodology, and that may be why he has to correct his data. In other words, he has an outdated (read crappy) sample design (built from the east coast was all days of the 1960s), unless his 7:30 exit poll data is within each time zone (that is a big assumption on my part that it is not!); because he would stand to miss significant voter movement in the more populous states of California, Texas, Michigan, Ohio etc.

I really wish the source had released this earlier, because I think we now had better re-think how exit polling as conducted in the United States may give an indication of the honesty of elections. Consider this the biggest of bum steers.

Mike

[DoYouEverWonder]

96. I did an analysis of the EV and ABS numbers for FL

I was able to get data for 25 out of 67 counties in FL. Unfortunately, I could not find the data for the other counties, but 25 counties is a pretty good sample.

Absentee
Bush 290,918
Kerry 234,311


Early voting
Bush 219,835
Kerry 267,726


So it seems at least in FL, Bush did well in the Absentee voting but Kerry made up the difference in the Early Voting.


Now keep in mind that the Absentee voting is where I think some of the fraud took place. It didn't happen at early voting because you have to show in person. However, according to my SOE, the Repugs were very, very interested in the Absentee and Inactive Voting lists and that they kept coming back for updated lists. There were also a lot of crazy stories in the press before the election about people going door to door to pick up ballots, dead people voting and people complaining that they never got ballots that they requested, or got ballots that they didn't request. These sorts of tricks lead me to believe that the Repugs sent in a fair number a falsified ballots.

[berniew1]

99. When were absentees added into the vote totals??

[DoYouEverWonder]

117. The absentee ballots are counted before election day

except for the ballots that arrive on election day and from overseas. The counting of the ballots started the Friday before the election.

The overseas ballots have up to 10 more days to arrive in order to be included in the final totals.

In regards to the military vote, most of their ballots came back on time because it was so easy for them to vote this time. They could fax, email or use regular mail to return their ballots. This means that most of the overseas ballots were counted in with all the other absentee ballots and it is no longer possible to find out how the military voted.

[Barrett808]

51. A link to the academic research derived from this raw data, same site

The Unexplained Exit Poll Discrepancy
http://www.scoop.co.nz/stories/pdfs/Mitofsky4zonedata/US2004G_3737_PRES04_NONE_V_Data.pdf

Working Paper: The Effect of Electronic Voting Machines On Change In Support for Bush in the 2004 Florida Elections
http://www.scoop.co.nz/stories/pdfs/election04_WP.pdf

Other interesting docs at:
http://www.scoop.co.nz/stories/pdfs/

[Bill Bored]

52. The First Sample Had 58% Female!

Looking at the H National data:

3:59 PM:
_________K B
Total 100 51 48
Male 42 47 51
Fem 58 53 45

7:33 PM:
_________K B
Total 100 51 48
Male 46 47 51
Fem 54 54 45

Yet Kerry got about 51% of the total vote in both versions. How is this possible? Does it jive with the Vertical percentages?

Wish I had more time to work on this but I've been losing sleep as it is.

[rdmccur]

83. The vote percentages have

been weighted already to take that into account

[Darkhawk32]

53. What I can't figure out is.........................

If Kerry's lead was increasing as the day progressed, how could he have lost it with the "West" region haven't finalizing their exit polls.

Didn't Kerry win the west coast plus Hawaii?

Am I missing something here?

If Bush got like 65% of the vote after 7pm EST, wouldn't that contradict the fact that the last of the states to come in were in the West?

[Quakerfriend]

56. Exactly what I was just thinking! n/t

[mgr]

57. South results would swamp pro Kerry East and West

I don't have the pdfs in front of me, and am going by memory, so fire away if I have this wrong.

If I remember the margin between Bush and Kerry in the east and west is about 3% points for Kerry; in the mid-west its reversed for bush; in the south it is strongly tilted to Bush, 57% to 43%, or something like that. That is where the Kerry popular vote lead vanished. A good portion of the south is in the CTZ and portion of Texas I think are even in the MTZ

Mike

[americanwoman]

59. Not necessarily

The regional proportions at each interval were kept in
more-or-less the same range. (Probably accomplished through
weighting.) Here are the regional proportions for each of the
3 reporting periods:

        east    midw    south   west
#1      20.91%  24.78%  33.38%  20.92%
#2      18.84%  25.43%  35.76%  19.98%
final   21.14%  26.91%  32.62%  19.33%

[mgr]

60. Those look cooked, and badly at that

These would be the percentage of total votes, and are these from the reports or your calculations? These make no sense.

As a thought experiment, California alone has ~20% of the US population, and had a 70%+ turnout, so these look a little queer. Last I remember California was in the West, so no matter what happened with the other western states (e.g. no one bothered voting at all) 20% is the bottom most share of the West, if California was in a region to itself. WA, OR, NV, AZ, NM would just add to that share. I can see 22 to 23%.

The only possible explanation would be that this is only for the election day results, and not absentee or early voter. The other question would be how Mitofski divvied up the states into regions.

Mike

[americanwoman]

63. Yup. We need to know how the regions were divided so

we can compare the regional percentages used with the actual voter turnout.

And yes, these were my calculations. (They can be double-checked by anyone who downloaded the data. I might have made an error somewhere.)

Also, I'm pretty sure that the Mitofski data includes phone calls to absentee voters. I don't know at what point they were folded in.

CA represents about 10% of the total US vote.

[Bill Bored]

66. americanwoman, the data only shows sample size per region.

It does NOT show each region's proportion of the total vote, does it?

If you sample more voters in the south, to capture something like the Dixiecrat effect for example, you get more responses. It doesn't mean there are more actual votes there. Am I missing something?

You also sample more voters in the swing states because their margins before the election were closer.

Also, is FL in the East or in the South?

And what about TX?

We should ask for a list of states in each region. Otherwise, we could just add up the state totals on the CNN site like a jigsaw puzzle and place each state in its proper region. I'm sure someone will take the time to do this, but unfortunately, I don't know if I can. It's "hard work!"

[k8conant]

70. I can't compare state totals to region totals...

to figure out which states are in which region. (Mitofsky doesn't list them either anywhere that I can find).

I listed all the state respondant totals and they add up to 75,079.

The national total given is 13,660 (which equals the sum of the 4 regions).

Who didn't they count nationally?

[Bill Bored]

71. Bad News

Sorry to send you on a wild goose chase, but then this illustrates the depth of the problem of relying on these polls for our evidence.

If they had so many bloody responses from the state polls, why didn't they just use them to produce a more accurate national poll? The state ones didn't ask as many questions as the national, but they all asked the most important one which is "For whom did you vote?"

Anyone else care to opine about this?

[althecat]

74. That is something I was wondering about.

Are you saying there were two exit polls. A national one and the state by state ones?

One thing I noticed when I got this data was the 13000 sample size... compared to the 60,000 roughly sample size in the state by state polls. Until now I did not understand why though.

And you are right. If they asked everybody in the state by state poll who they voted for that data could have been fed into this.

This interests me because.

If there were two distinct polls. And they both showed the same result - i.e. kerry winning then that eliminates quite a few possibilities in terms of systemic errors. I.e. if there was some kind of systemic error it would have to be in both polls.

However as I point out in my intro on the basis of TIA's analysis of yesterday I think we now have the clearest possible evidence that the exit polls were correct.

In a nutshell 20% of the exit pollees were new voters. According to the exit poll these broke 65%/35% to Kerry. The actual result is the reverse. We are not talking about a 5% or 10% or even 3* the MOE discrepancy here... the exit poll among these 20% of the voters was out by 100% (recording bush's vote at half of what it actually was) no systemic error can explain this.

al

[Bill Bored]

85. 2 Sets of Polls?

Al, I am very interested in this, but I always get the feeling that it's futile because of Mitofsky's own comments. That said, I like numbers as much as the next guy, and time permitting, this is something to try and sort out, if possible. But even if it DOES prove the election was stolen, it does not describe the mechanism or the remedy, each of which are more important in the long run and perhaps even the short run.

But anyway, you can read all about these polls at:
<http://www.exit-poll.net/index.html>

According to their methodology statements, the national poll surveyed 250 polling places while the state polls used a total of about 1500, with less than 100 voters per polling place. They say the national poll is a subset of the state polls.

So why not just take ALL the state polls, add them up, and see if the percentages for the popular vote are any different from the National sample?

I have to think more about your statement that "20% of the exit pollees were new voters. According to the exit poll these broke 65%/35% to Kerry. The actual result is the reverse."

[berniew1]

100. What do you mean by the actual result?? I don't understand distinction?

which breakout goes with what??

[Bill Bored]

109. I was quoting Al and asked that question in this or another thread. n/t

[americanwoman]

84. Bill Bored,

You asked:It does NOT show each region's proportion of the
total vote, does it?

No it does not. But if Mitofsky is doing their job, the
regional proportions should match fairly closely with voter
turnout. That's why I'd like to know what states went into
what regions. (You can't tell just by looking at Mitofsky
state data because the proportionals definitely are not right.
See my post #67.) 

I took a crack at dividing up the states. Using published
state-by-state voter turnout data, I came up with the
following regional percentages. If Mitofsky used the same
states I did, the percentages match up fairly closely. 

       americanwoman's     Mitofsky's
       calculated          regional percents
       regional %          (from 'final' survey set)
East    21.60              21.14%
MidW    25.30              26.91%
South   31.99              32.62%
West    21.10              19.33%

I broke the states out as follows:
East   : CT, DC, DE, ME, MD, MA, NH, NJ, NY, PA, RI, VT
MidWest: IL, IN, IA, KS, MI, MN, MO, NE, ND, OH, SD, WI
South  : AL, AR, FL, GA, KY, LA, MS, NC, OK, SC, TN, TX, VA,
WV
West   : AK, AZ, CA, CO, HI, ID, MT, NV, NM, OR, UT, WA, WY

 

[Bill Bored]

88. americanwoman, What about this:

And Happy New Year BTW!

Since we know the national totals are a subest of the state totals, what about dividing the national total of 13,660 by the state total ~76,000 I think, and then taking that .18 x each state to get new state numbers that would add up to the national total. You could then use these to map states to regions and check your work, right?

I was just trying to find this out to see if there was any region other than the west in which the later data might have swung the popular vote, like TX maybe. And don't rule out CA either because CA shifted Red even though Kerry won the state!

[ClintCooper2003]

64. althecat, I honestly cannot understand why you didn't send this out...

earlier!

[althecat]

68. Yes... sorry

I did send the data to several researchers... but you are right I should have released it earlier. Probably as soon as Conyers received a kiss off.

[Cheney Killed Bambi]

76. Are you going to need legal help?

Is it possible Mitofsky will sue you for releasing the data?

I'm sure there are several lawyers who will represent you pro bono.

[althecat]

78. Let's hope not.

I would be surprised if it came to that. ** fingers crossed **

[ClintCooper2003]

65. I would have pitched in $50 of my own money to pay for state by state...

you should have started a fundraising campaign immediately. Why didn't you?

[Bill Bored]

67. We have the Nov 3 state by state though.

We can get the sample sizes from there and figure out which states go in which regions of this data. The only thing we'd be missing is the state-by-state first and second polls on Nov 2. But at least we'd be able to see where the states fit into this regional data. Yes?

If no one does it first, I can do it later after my wife and kids are asleep. There are probably only a few ambiguous states near the borders of each region that will either add up correctly or not. We can be pretty certain where most of them are, unless they happen to have sample sizes exactly the same.

[americanwoman]

82. The Mitofsky state-by-state total doesn't match up with

their regional total, and probably for good reason.

For state results, you need to be sure you get a good sample for that state. You can't just add up all of the state results and call them regions. Why? Because you'll likely not have the correct percentages.

For example, the sum of Mitofsky state survey responders = 76,240. The sum of their 4 regional survey responders (east, midwest, south and west) is 13,660. I have to believe the regional survey is either a separate survey or the state figures get weighted.

Mitofsky's sample size for CA is 2390, which represents only 3.13% of all of the total state Mitofsky sample size. But according to published state and national data, CA needs to represent about 10% of the voters.

Hope that makes sense.

[Mojorabbit]

102. I don't know if this helps

All this is over my head.
http://slate.msn.com/id/2109053

[althecat]

69. We needed $10,000 and there was some discussion of raising funds...

BUt that is a lot of money for something that should have been publicly available. We ran several email and letter writing campaigns to try and get the data released too. All to no avail.

[DoYouEverWonder]

95. He should have started?

Don't you think that should have been up to us Americans? althecat has done tons of stuff to help the cause, he's half way around the world and he has a great website to maintain.

I just hate when people think someone else should do something.

[althecat]

72. Talk of the town....

From our referrer logs...

/stories/pdfs/Mitofsky4zonedata/
1559: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215277
1114: http://www.democraticunderground.com/
770:
372: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=102x1112148
370: http://www.dailykos.com/story/2004/12/31/16138/255
296: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=215277&mesg_id=215277
140: http://www.scoop.co.nz/stories/pdfs/
110: http://www.democraticunderground.com
102: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x214701
92: http://www.scoop.co.nz/stories/pdfs/Mitofsky4zonedata
70: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=214701&mesg_id=214701
51: http://www.smirkingchimp.com/viewtopic.php?topic=51119&forum=9
38: http://dailykos.com/story/2004/12/31/16138/255
30: http://www.commongroundcommonsense.org/index.php?showtopic=12949
24: http://www.rawstory.com/
19: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=215277&mesg_id=215277&page=
11: http://rawstory.com/
10: http://www.bbvforums.org/forums/messages/9/1258.html?1104509411
10: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=215277
10: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=214701&mesg_id=214701&page=
8: http://www.scoop.co.nz/
6: http://www.commongroundcommonsense.org/index.php?showtopic=12949&hl=
6: http://www.dailykos.com/section/Diary
6: http://smirkingchimp.com/viewtopic.php?topic=51119&forum=9
6: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x216264
5: http://fullerton.dailykos.com/story/2004/12/31/16138/255
4: http://p079.ezboard.com/fthecaboosefrm17.showMessage?topicID=9123.topic
4: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=214701
4: http://www.dailykos.com/story/2004/12/31/14258/320
3: http://www.astroworld.us/archives/000459.html
3: http://www.DemocraticUnderground.com
3: http://www.democraticunderground.com/discuss/duboard.php
3: http://www.planetthinktanks.com/comm-thread.asp?thread=5348&forum=4
3: http://www.commongroundcommonsense.org/index.php?s=7b70ad674d4db5d9c4bbf402426829c6&showtopic=12949
3: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215277#215293
3: http://www.metafilter.com/mefi/38253
3: http://groups.yahoo.com/group/CASE_OH/message/8352
2: http://www.metafilter.com/contribute/post_comment_preview.cfm
2: http://www.scoop.co.nz/stories/pdfs/Mitofsky4zonedata/
2: http://www.dailykos.com/story/2004/12/30/223723/91
2: http://www.commongroundcommonsense.org/index.php?showtopic=12896&hl=
2: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x21527
2: http://www.sho.com/site/message/thread.do?pagenum=27&groupid=12&topicid=105711&boardid=3014
2: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=216096
2: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=102&topic_id=1112148
2: http://p084.ezboard.com/feqhorizons65958frm10.showAddReplyScreenFromWeb?topicID=329.topic&index=2
2: http://www.bbvforums.org/forums/messages/9/1258.html?1104522603
2: http://www.commongroundcommonsense.org/index.php?showtopic=12896
1: http://tabletalk.salon.com/webx?[e-mail]7aY8ypa7.3@.773a529d/1846
1: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=216264&mesg_id=216439&page=
1: http://tabletalk.salon.com/webx?13@87.0RwGaZgdp3Y.7@.773a529d/1848
1: http://tabletalk.salon.com/webx?[e-mail]2pCu.2@.773a529d/1846
1: http://www.democraticunderground.com/discuss/duboard.php?az=printer_friendly&forum=203&topic_id=215277&mesg_id=215277
1: http://tabletalk.salon.com/webx?[e-mail]9WaZn4ph9.31@.773a529d/1846
1: http://www.commongroundcommonsense.org/index.php?showtopic=12896&st=0&#entry120816
1: http://tabletalk.salon.com/webx?[e-mail]3a4drpok.10@.773a529d/1843
1: http://tabletalk.salon.com/webx?[e-mail]9ah1QpJS.45@.773a529d/1847
1: http://207.179.94.22/forums/forums4us/private.php?do=showpm&pmid=2109
1: http://tabletalk.salon.com/webx?12@79.4RNLaqCtpXb.28@.773a529d/1847
1: http://tabletalk.salon.com/webx?[e-mail].33@.773a529d/1848
1: http://tabletalk.salon.com/webx?[e-mail].17@.773a529d/1844
1: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x216096
1: http://www.democraticunderground.com/discuss//duboard.php?az=show_mesg&forum=203&topic_id=215277&mesg_id=215277
1: http://p084.ezboard.com/feqhorizons65958frm10.showMessage?topicID=329.topic
1: http://www.democraticunderground.com/discuss/duboard.php?az=printer_friendly&forum=203&topic_id=216264
1: http://us.f537.mail.yahoo.com/ym/us/ShowLetter?box=Inbox&login=1&uid=16483
1: http://www.Democraticunderground.com
1: http://www.dailykos.com/story/2004/12/31/16138/255#3
1: http://www.democraticunderground.com/discuss/duboard.php?az=printer_friendly&forum=102&topic_id=1112148
1: http://bartcopnation.com/dc/dcboard.php?az=show_mesg&forum=2&topic_id=345492&mesg_id=345535&page=
1: http://us.f538.mail.yahoo.com/ym/ShowLetter?MsgId=357_2028217_72662_2456_1093_0_7257_2624_868481926&Idx=11&YY=20414&inc=25&order=down&sort=date&pos=0&view=a&head=b&box=Inbox
1: http://www.google.com/search?hl=en&q=http%3A%2F%2Ftinyurl.com%2F499ah
1: http://www.berthall.com/forums/showthread.php?s=dac667e5300f4c6dba35e8a0547bf31b&threadid=12929&perpage=15&pagenumber=4
1: http://webmail.aol.com/msgview.adp?folder=SU5CT1g=&uid=10978552
1: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=216096&mesg_id=216354&page=
1: http://www.democraticunderground.com/discuss//duboard.php?az=view_all&address=203x215277
1: http://webmailb.juno.com/webmail/2FD5B75D/8?block=1&msgList=0000MOG0&folder=Inbox&destFolder=Inbox&command=next&msgNum=0000MOG0&fromPage=read&id=&count=1104529983200
1: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215658
1: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215277#215301
1: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=216264
1: http://www.democraticunderground.com/discuss/duboard.php?az=printer_friendly&forum=203&topic_id=215277
1: http://mail.du.edu/frame.html?rtfPossible=true&lang=en
1: http://www.dailykos.com/story/2004/12/31/16138/255#34
1: http://www.commongroundcommonsense.org/index.php?showtopic=12896&pid=120808&st=0&
1: http://tabletalk.salon.com/webx?50@251.8RX8alKapJ3.8@.773a529d
1: http://us.f500.mail.yahoo.com/ym/ShowLetter?Search=&Idx=1&YY=22403&order=down&sort=date&pos=0
1: http://webmailb.juno.com/webmail/2FD5B75D/8?position=1&msgNum=0000MP00&folder=Inbox&count=1104529787&block=1
1: http://www.berthall.com/forums/showthread.php?s=99375bb8349af2907b0d2db9f3b1657c&postid=175880
1: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215277#215319
1: http://us.f538.mail.yahoo.com/ym/ShowLetter?MsgId=357_49131580_4231914_2434_1093_0_55679_2624_868481926&Idx=6&YY=97396&inc=25&order=down&sort=date&pos=0&view=&head=&box=Inbox
1: http://www.democraticunderground.com/main.html
1: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=215658&mesg_id=216390&page=
1: http://mail.fuzzybud.com/email/scripts/view.pl?EV1=11045200743690959
1: http://tabletalk.salon.com/webx?14@109.8RVlaWqxpG1.32200@.773a529d
1: http://www.dailykos.com/story/2004/12/30/223723/91#270
1: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=215277&mesg_id=215277&listing_type=
1: http://us.f309.mail.yahoo.com/ym/ShowLetter?Search=&Idx=1&YY=16152&order=down&sort=date&pos=0&view=a&head=b
1: http://tabletalk.salon.com/webx?[e-mail]90.4@.773a529d/1847
1: http://tabletalk.salon.com/webx?13@87.0RwGaZgdp3Y.2@.773a529d/1847
1: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x215277#215437
1: http://tabletalk.salon.com/webx?[e-mail]6a2EDpHO.0@.773a529d
1: http://www.rawstory.com
1: http://us.f309.mail.yahoo.com/ym/ShowLetter?MsgId=3547_12224112_265600_3462_936_0_230546_1672_3649564736&Idx=0&YY=41532&inc=50&order=down&sort=date&pos=0&view=a&head=b&box=Inbox
1: http://www.astroworld.us/mt/mt-comments.cgi?entry_id=459
1: http://tabletalk.salon.com/webx?[e-mail].0@.773a529d
1: http://www.google.ca/search?hl=en&q=http%3A%2F%2Ftinyurl.com%2F499ah&btnG=Google+Search&meta=

[IndyPriest]

79. SO, Mr Professional pollster: we got 'em. You know we got 'em....

We know you know we got 'em. EVERYBODY knows we got 'em, and by now EVERYBODY who wants 'em, gots 'em. You wanna talk about it, eh? That's OK, just lean into the microphone....

[Cheney Killed Bambi]

75. Kick

:kick:

[mikelewis]

77. This data also shows that Gore won Ohio in 2000

Did you vote in the presidential election
in 2000? (n=359)
No, I did not vote 13
Yes, for Al Gore 47
Yes, for George W. Bush 35
Yes, for another candidate 4


Hmmmmm, must be a lie, Bush won, didn't he?


AHHHHHHH!!!!!!! I really hate them, I really really do.

[althecat]

80. extracted SMOKING GUN QUESTION data from new Exit Poll PDFs

extracted SMOKING GUN QUESTION data from new Exit Poll PDFs

Number crunchers... have a look at this... what do you see....


4pm DATA FOUR REGIONS

VERTICAL
Did you vote in the presidential election
in 2000? (n=383)
-----------------------Tot K B N
No, I did not vote 16 19 11 -
Yes, for Al Gore 45 70 13 -
Yes, for George W. Bush 36 6 74 -
Yes, for another candidate 4 5 2 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=383)
-----------------------Tot K B N Rep2000
No, I did not vote 16 69 31 - -
Yes, for Al Gore 45 88 12 - -
Yes, for George W. Bush 36 9 91 - -
Yes, for another candidate 4 - - - -

WEST

VERTICAL
Did you vote in the presidential election
in 2000? (n=428)
-----------------------Tot K B N
No, I did not vote 15 20 8 -
Yes, for Al Gore 40 70 4 -
Yes, for George W. Bush 40 6 87 -
Yes, for another candidate 6 5 1 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=428)
-----------------------Tot K B N Rep2000
No, I did not vote 15 73 23 2 -
Yes, for Al Gore 40 95 4 1 -
Yes, for George W. Bush 40 8 91 0 -
Yes, for another candidate 6 43 10 28 -

SOUTH

VERTICAL
Did you vote in the presidential election
in 2000? (n=632)
-----------------------Tot K B N
No, I did not vote 16 19 13 -
Yes, for Al Gore 35 69 6 -
Yes, for George W. Bush 47 6 81 -
Yes, for another candidate 3 5 0 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=632)
-----------------------Tot K B N Rep2000
No, I did not vote 16 54 43 2 -
Yes, for Al Gore 35 90 9 - -
Yes, for George W. Bush 47 6 92 - -
Yes, for another candidate 3 - - - -

MIDWEST

VERTICAL
Did you vote in the presidential election
in 2000? (n=473)
-----------------------Tot K B N
No, I did not vote 15 16 14 -
Yes, for Al Gore 39 69 6 -
Yes, for George W. Bush 44 11 79 -
Yes, for another candidate 3 4 1 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=473)
-----------------------Tot K B N Rep2000
No, I did not vote 15 54 46 - -
Yes, for Al Gore 39 93 7 - -
Yes, for George W. Bush 44 14 86 0 -
Yes, for another candidate 3 - - - -

****************

TOTAL – (NONE IN PDFS 4pm DATA)

VERTICAL
Did you vote in the presidential election
in 2000? (n=1,916)
-----------------------Tot K B N
No, I did not vote 15 18 12 -
Yes, for Al Gore 39 70 7 -
Yes, for George W. Bush 42 7 80 -
Yes, for another candidate 4 4 1 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=1,916)
-----------------------Tot K B N Rep2000
No, I did not vote 15 62 37 1 -
Yes, for Al Gore 39 91 8 0 -
Yes, for George W. Bush 42 9 90 0 -
Yes, for another candidate 4 61 12 16

****************

TOTAL – (NONE IN PDFS) (7.30pm DATA)

VERTICAL
Did you vote in the presidential election
in 2000? (n=2,545)
-----------------------Tot K B N
No, I did not vote 17 19 14 14
Yes, for Al Gore 38 68 7 27
Yes, for George W. Bush 41 8 79 11
Yes, for another candidate 4 5 1 48

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=2,545)
-----------------------Tot K B N Rep2000
No, I did not vote 17 59 39 1 -
Yes, for Al Gore 38 91 8 1 -
Yes, for George W. Bush 41 9 90 0 -
Yes, for another candidate 4 65 13 16 -

****************

TOTAL – (NONE IN PDFS) (CORRECTED DATA – 1.24pm NOV 3rd)

VERTICAL
Did you vote in the presidential election
in 2000? (n=3,168)
-----------------------Tot K B N
No, I did not vote 17 19 15 -
Yes, for Al Gore 37 68 7 -
Yes, for George W. Bush 43 8 77 -
Yes, for another candidate 3 5 1 -

HORIZONTAL
Did you vote in the presidential election
in 2000? (n=3,168)
-----------------------Tot K B N Rep2000
No, I did not vote 17 54 45 1 -
Yes, for Al Gore 37 90 10 0 -
Yes, for George W. Bush 43 9 91 0 -
Yes, for another candidate 3 71 21 3 -

[TruthIsAll]

107. Analysis: How did Bush get 63.6% of the final 623 polled?

Kerry was the solid winner everywhere but the south. 				
Kerry won the vast majority of new voters in EVERY region:
 70% East/West; 54% Midwest/South

Bush got 48.1% of the first 2545 polled.
Bush got 63.6% of the final 623 polled.

This brought him from 48.1% to 51.3% of the 3168 total.


Date    Time	N	K	B	K%	B%			
11/02 4:00pm	1916	977	900	52.1%	47.9%			
								
11/02 7:30pm	2545	1295	1198	51.9%	48.1%			
11/03 1:24pm	3168	1536	1619	48.7%	51.3%	<<???		
Change	      623	240	421	36.4%	63.6%	<<?????		
								
.................................................				


4pm	     K	B	K%	B%				
East	   206	165	55.5%	44.5%				
West	   234	180	56.6%	43.4%				
South	   272	337	44.6%	55.4%				
Midwest  239	225	51.6%	48.4%				
								
Total	   951	906	51.2%	48.8%				

.......................................................								
VERTICAL	383							
EAST...... 	  2000	K%	B%	N%	K  	B	K%	B%
No vote     	16	19	11	-	39	18	10.2%	4.7%
Gore	      45	70	13	-	144	21	37.7%	5.6%
Bush	      36	6	74	-	12	122	3.2%	31.9%
Other	       4	5	2	-	10	3	2.7%	0.9%
	      101	100	100		206	165	53.9%	43.1%
								
								
								
HORIZONTAL	2000	K%	B%	N%	K % 	B %	K  	B
No vote	          16	69	31	-	11.0	5.0	42	19
Gore  	          45	88	12	-	39.6	5.4	152	21
Bush	          36	9	91	-	3.2	32.8	12	125
Other	           4	-	-	-				
Total	         101				53.9	43.1	206	165
							      55.55%	44.45%
								
								
.......................................................								
VERTICAL	428							
WEST  	2000	K%	B%	N%	K	B	K%	B%
No vote	15	20	8	-	47	14	10.9%	3.4%
Gore	    40	70	4	-	164	7	38.3%	1.7%
Bush	    40	6	87	-	14	157	3.3%	36.6%
Other	     6	5	1	-	12	2	2.7%	0.4%
	   101	101	100		237	180	55.3%	42.1%
								
								
							428	
HORIZONTAL	2000	K%	B%	N%	K % 	B %	K  	B
No vote	          15	73	23	2	10.95	3.45	47	15
Gore	          40	95	4	1	38	1.6	163	7
Bush	          40	8	91	0	3.2	36.4	14	156
Other	           6	43	10	28	2.58	0.6	11	3
	          101	                 54.73	42.05	234	180
							      56.55%	43.45%

								
............................................................								
VERTICAL	632							
SOUTH 	2000	K%	B%	N%	K	B	K%	B%
No vote	16	19	13	-	52	44	8.2%	6.9%
Gore	   35	69	6	-	187	20	29.6%	3.2%
Bush	   47	6	81	-	16	273	2.6%	43.1%
Other	    3	5	0	-	14	0	2.1%	0.0%
	   101	99	100		269	337	42.5%	53.3%
								
								
							632	
HORIZONTAL	2000	K%	B%	N%	K% 	B%	K  	B
No vote	          16	54	43	2	8.64	6.88	55	43
Gore	          35	90	9	0	31.5	3.15	199	20
Bush	          47	6	92	0	2.82	43.24	18	273
Other	           3	0	0	0	0	0	0	0
	          101			         42.96	53.27	272	337
						          44.64%	55.36%

								
.....................................................								
VERTICAL	473							
MIDWEST	        2000	    K%	B%	N%	K	B	K%	B%
No vote	          15	16	14	-	38	31	8.1%	6.6%
Gore	          39	69	6	-	165	13	34.9%	2.8%
Bush	           44	11	79	-	26	177	5.6%	37.5%
Other	           3	4	1	-	10	2	2.0%	0.5%
	           101	100	100		239	225	50.5%	47.5%
								
								
								
HORIZONTAL	 2000	K%	B%	N%	K% 	B%	K  	B
No vote	          15	54	46	-	8.1	6.9	38	33
Gore	          39	93	7	-	36.27	2.73	172	13
Bush	          44	14	86	-	6.16	37.84	29	179
Other	           3	0	0	-	0	0	0	0
	         101				50.53	47.47	239	225
						            51.56%	48.44%
								
..........................................................								
								
VERTICAL								
4PM-1916	2000	K%	B%	N%	K	B	K%	B%
No vote	          15	18	12	-	176	108	9.2%	5.6%
Gore	          39	70	7	-	684	63	35.7%	3.3%
Bush	          42	7	80	-	68	720	3.6%	37.6%
Other	           4	4	1	-	39	9	2.0%	0.5%
	          100	99	100		968	900	50.5%	47.0%
								
								
							
HORIZONTAL	2000	K%	B%	N%	K % 	B %	K  	B
No vote	      15	62	37	1	9.3	5.55	178	106
Gore	           39	91	8	0	35.49	3.12	680	60
Bush	           42	9	90	0	3.78	37.8	72	724
Other	           4	61	12	16	2.44	0.48	47	9
	          100	223	147	17	51.01	46.95	977	900
								
								
.....................................................								
VERTICAL	2545							
7:30PM	        2000	K%	B%	N%	K	B	K%	B%
No vote	          17 	19	14	14	246	168	9.7%	6.6%
Gore	          38	68	7	27	881	84	34.6%	3.3%
Bush	          41	8	79	11	104	947	4.1%	37.2%
Other	           4	5	1	48	65	12	2.5%	0.5%
	         100	100	101	100	1295	1210	 50.9%	47.6%
								
								
								
HORIZONTAL	2000	K%	B%	N%	K % 	B %	K  	B
No vote	          17    59	39	1	10.03	6.63	255	169
Gore	       38	    91	8	1	34.58	3.04	880	77
Bush	       41	     9	90	0	3.69	36.9	94	939
Other	        4	    65	13	16	2.6	0.52	66	13
	        100			        50.9   47.09    1295 	1198
								
.......................................................								
VERTICAL	3168							
Nov3  	        2000	K%	B%	N%	K	B	K%	B%
No vote	          17	19	15		292	243	9.2%	7.7%
Gore	          38	68	7		1044	113	33.0%	3.6%
Bush	          43	8	77		123	1247	3.9%	39.4%
Other	           3	5	1	48	77	16	2.4%	0.5%
	          101	100	100	48	1536	1619	48.5%	51.1%
								
								
								
HORIZONTAL	2000	K%	B%	N%	K% 	B%	K  	B
No vote	          17	54	45	1	9.18	7.65	291	242
Gore	         37 	90	10	0	33.3	3.7	1055	117
Bush	         43 	9	91	0	3.87	39.13	123	1240
Other	         3	   71	    21	3	2.13	0.63	67	20
	        100			        48.48	51.11	1536	1619
								

[euler]

116. May be 63.6 of the 623

voted for Bush.

Don't roll your eyes. My statement is as valid as any other statement about these numbers, because there is not enough data to infer anything from them. At least not in the world I live in.

[cyn2]

86. kick n/t

[s-cubed]

87. kick & happy new year

[Melissa G]

90. Kick for Althecat and a Happy New Year! n/t

[Bill Bored]

91. Althecat, you said/please explain

"In a nutshell 20% of the exit pollees were new voters."

Well, either new or 3rd-party voters in 2000, but mostly new.

"According to the exit poll these broke 65%/35% to Kerry."

Yes.

"The actual result is the reverse."

Where did you see an actual result that showed that 65% of new voters voted for Bush?

"We are not talking about a 5% or 10% or even 3* the MOE discrepancy here... the exit poll among these 20% of the voters was out by 100% (recording bush's vote at half of what it actually was) no systemic error can explain this."

I don't see where this was recorded. Please explain. Are you perhaps confusing new voters (i.e. those who didn't vote for Bush or Gore in 2000) with the ~65% of the vote Bush would had to have gotten after 7:30 PM (see post 29)? These are apples and oranges.

[althecat]

93. Explaining the comments

Q: Where did you see an actual result that showed that 65% of new voters voted for Bush?

A: In the final result Bush gained 11 million votes, kerry six million odd. This breaks roughly 65% to 35% in terms of the extra vote. Looking more closely at the exit poll data we find that the poll indicates Kerry gained a majority of the new (not voted before) vote - but lost because his base was severely undermined.

Q: "We are not talking about a 5% or 10% or even 3* the MOE discrepancy here... the exit poll among these 20% of the voters was out by 100% (recording bush's vote at half of what it actually was) no systemic error can explain this."

I don't see where this was recorded. Please explain. Are you perhaps confusing new voters (i.e. those who didn't vote for Bush or Gore in 2000) with the ~65% of the vote Bush would had to have gotten after 7:30 PM (see post 29)? These are apples and oranges.

A: Yes... I see the problem. But as there is no explanation for how the corrected exit poll figures were arrived at there is not a great deal of light there. I am sorry I was a little imprecise in my initial remarks.

[Bill Bored]

92. One more thing besides Happy New Year!

Is any or all the Nov 2 stuff raw data?

[jmc777]

94. kick n/t

:kick:

[wiggs]

97. What does "time of weighting" at top of docs mean? nt

[berniew1]

101. what is connection of this data to the Scoop/NZ data??

[jmknapp]

104. SMOKER?

Here's a remarkable data point at least.

In the 11/2 7:33 pm exit poll data for the whole country, the sample size is 11,027 and the breakdown is 51% Kerry to 48% Bush, 1% Nadir.

Therefore, of the 11,027 people interviewed, there were about 5,624 votes for Kerry and 5,293 for Bush.

In the 11/3 1:24 pm exit poll data, the sample size is 13,660 and the percentages are reversed, 48% Kerry to 51% for Bush.

That translates to 6,557 Kerry votes to 6,967 for Bush.

So of the 13,660-11,027 = 2,633 extra interviews between the two polls, disregarding any reweighting, Kerry had (6557-5624)/2633 = 35% to Bush's (6967-5293)/2633 = 64%.

Quite a finish for Bush.

[davidgmills]

110. That homestretch was something, wasn't it

[Nothing Without Hope]

105. What is the relationship, if any, with the exit poll data here

at this site:
http://www.scoop.co.nz/mason/stories/HL0411/S00227.htm

This is for all individual states, rather than for zones. It was posted Nov 17 and at the time was the only complete state exit poll data set available.

I assume the two sets were not collected by the same people, of course. I have two questions, and I'd be grateful if you could shed any light on them:
1. Are you aware of any reason to doubt the validity or value of the Nov 17 data at the link I gave above?
2. Has anyone compared the results of the Mitkovsky polls with those from this other data set? If so, are they consistent?

[Blue22]

106. House National Exit Poll Points to Bad National Sample

Don't know the validity of these exit poll results, but they are internally consistent. Looking at the House national exit poll data from the same source, we can compare the poll trends across two races.

Survey Bush Kerry House Rep House Dem

All 1 (4PM) 48 51 47 51
All 2 (7:33PM) 48 51 47 51
All 3 (Nov 3) 51 48 50 49
Tallied 51.2 48.4 50.1 47.5

Since the same people were sampled for both surveys, the adjustments to the Nov 3 sample shifted Republican in both sets of races. Either the House should have gone Democratic (which I have heard no mention of)or the poll was biased.

Let's keep our eyes on the ball. * won the national vote because that's what Rove targeted. Focusing on the popular vote keeps us in Rove's briar patch. Kerry won the electoral vote but is stymied by bad counts in Ohio, New Mexico, and perhaps Florida. Solving those problems will solve the riddle.

This is my first post. Hope it made sense.
B-)

[Firespirit]

108. Gerrymandered districts?

I haven't looked that closely at anything but the early evening House polls for the East and the South. The South page doesn't look particularly out of line. That's where I am, so I'm going on personal experience.

This poll isn't broken down into Congressional districts--which, we know, are usually notoriously skewed to one party or the other. Unfortunately, unless the data set can be broken down further, it's difficult to draw ANY conclusions about the validity of it with respect to House races. :(

Welcome to DU! :hi:


Edit to say:

Something occurred to me. This is a heck of a project, but it IS possible to verify the vote breakdown for House races by region.

We get the total vote counts for all House candidates in every House race, then get a combined vote percentage for the individual regions and compare it to this. What states are included in each region?

That ought to lay it to rest whether the sample was bad or not.

[Blue22]

111. House Vote and Poll

I have all the House votes for 11/5/04 in a spreadsheet. If someone could identify which states are in the exit poll region, it would be pretty stratightforward to compare polls to House votes by region. There is one technical problem with House votes and that is the number of districts without opposition. I'm working on estimating the missing data (at one-half of the average state vote per district), if necessary.

B-)

[Bill Bored]

113. If you want the states,

just take the popular vote totals for Pres in each state and derive its percentage of the national popular vote. Then take these percentages and apply them to the regions in jigsaw-like fashion to get the states in each region. I.e. x% of national popular vote = x% of 13,066.

This way you don't need 435 totals to get the states!

Does this make sense?

[Bill Bored]

112. Best First Post I've seen so far!

Welcome to DU! Give this man (or woman) a thread! (Except that there are too many of them to keep track of to begin with!)

[Darkhawk32]

118. DO WE HAVE ANY EXIT POLL DATA FOR SENATORIAL RACES? n/t

[Blue22]

119. We need to look for internal inconsistencies in State polls.

Good question. exitpollz.org has several. I think the answer to our exit poll riddle lies in those states where the exit poll was completely wrong in the presidential race, but correct in the Senate race, e.g., Colorado. If we had the individual exit poll survey responses, we would know which specific precincts to examine for fraud.