About a week ago, I created a post requesting data regarding the exit polls in the battleground states. I appreciated the many attempts of my fellow DUers to keep the thread kicked, but it ended up dying in the archives. Well, my dad, who has been the State Epidemiologist for several states, was able to find state by state exit poll information and perform a statistical analysis of that information. The results of the analysis prove what I have strongly suspected since the night of November 2nd, the election was rigged to ensure bush another 4 years. Please read the following and keep this thread kicked. Nominate it for the home page if you will, this is statistical proof that we did not have a legitimate election. Also, forward this information to anyone you please. This information needs to get out there, the mainstream media MUST cover this. I've included the exit poll data from which the article was written, but I tried as hard as I could and I was unable to convert it into HTML table format. If anyone is capable of doing that, I'd greatly appreciate it.
On edit: Oh, and if anyone would like me to send you the article and table data in Word format, just message me with your email address and I'll send you a copy. Thanks.
Comparison of exit polls by state with official vote tallies –
Presidential election of 2004 – Kerry vs. Bush
by Dale Tavris, using exit poll data collected by Jonathon Simon
Introduction
The 2004 U.S. presidential election has generated a great deal of controversy. Millions of U.S. citizens believe that the election was flawed at best, or stolen at worst. This belief is based on perceived discrepancies between published exit polls and official vote tallies, as well as several instances of documented irregularities in the conduct of the election. This report is an analysis of a comparison by state between exit poll results and official vote tallies.
The analysis of exit poll results for this election is hampered by the fact that it is currently impossible to find an official source for them. CNN posted exit poll results by state on its website throughout the day and evening of Tuesday, November 2nd, 2004. But some time late on that Tuesday evening or early Wednesday morning, CNN changed all of its state exit poll results to reflect the official vote total. So currently there is no official site that contains the actual exit poll results.
For example, shortly past 1:00 a.m. Wednesday morning, CNN’s web site indicated for Ohio that women had voted 53-47 for Kerry, and men had voted 51-49 for Kerry, giving Kerry an advantage of approximately four percentage points in the Ohio exit poll. Presumably this was Ohio’s final exit poll. But then the web site suddenly changed to indicate a 50-50 split in Ohio for women, and a 52-47 advantage for Bush among men. This brought the “exit poll” into conformance with the official vote tally for Ohio. We know that this happened with respect to Ohio because web shots exist of the change from one minute to the next. Presumably this happened with respect to several or all of the other states as well. This is assumed because CNN’s state “exit polls” currently, and as of some time Wednesday morning, were all in conformance with the official vote tallies. Yet shortly before that time many of the exit polls displayed on CNN’s web site varied from the official vote tallies, some of them by substantial amounts. I do not know whether or not web shots of these exit polls exist to confirm that fact for other states, as they do for Ohio.
Methods
The great majority of exit poll data in this report came from Jonathon Simon’s internet article of November 11th, titled “To Those Who Seek Information As A Basis For Action Regarding Bush's ‘Victory’ ”. Between 12:19 and 12:38 a.m. on the morning of Wednesday, November 3rd, he collected state by state Edison/Mitofsky exit poll results. This data included the percentage of the vote for Kerry and Bush, the time that the result was listed, and the sample size of the poll for 46 states and the District of Columbia.
He was not able to get to four states (New York, New Jersey, Virginia, and North Carolina) before the data was apparently altered to conform to the official vote tally. For three of these states (NY, NJ, NC) I used the latest exit poll results that I could find listed on the web by bloggers (4:28 p.m. for NY and NJ, later for NC). I could find no exit poll results for Virginia.
Mr. Simon acknowledges that some of his results from other states might also have been altered before he got to them, but he is not sure. I accepted and used all of his 47 results except for Wisconsin, which apparently was altered before he obtained it. The Wisconsin result that he listed conformed to the official vote tally within 0.3%, and there were other late results listed for Wisconsin that were very different from the official vote tally but consistent with each other. It is also of note that the Ohio results are the only ones that Mr. Simon lists as having been obtained prior to 12:19 a.m. Wednesday morning. He lists these as having been obtained at 7:32 p.m. Tuesday evening. Yet, these results are completely consistent with the CNN web shots of the Ohio exit poll distribution by gender, obtained shortly after 1:00 a.m. Wednesday morning.
Official vote tallies were obtained from CNN’s web site dated November 11th:
http://www.cnn.com/ELECTION/2004/pages/results/president. Percentage of votes for Kerry and Bush were calculated as the percentage of votes that were received by those two candidates (i.e., all 3rd party votes were ignored for this analysis).
The states were divided into two categories – battleground and non-battleground states. The battleground states were the ones that were considered to be reasonably winnable by either candidate just prior to the election. There were 11 of these states, and they included Florida, Ohio, Pennsylvania, Michigan, Wisconsin, Minnesota, Iowa, New Mexico, Nevada, Colorado, and New Hampshire. It is important to note that all 11 of these states exhibited differences between the two candidates in the official vote tally of less than 6%, and there was only one other state that did that (Oregon). The percentages of the official vote tallies were figured to the nearest one tenth of a percent for the 11 battleground states, plus eight other close states, whereas for the other states the percentages displayed in CNN’s web site, which were rounded to the nearest 1%, were used.
All comparisons of official vote tallies with exit poll results were made using the Chi Square test. When making these comparisons for groups of states, the official vote tally was weighted for the sample size of the exit poll.
Results
Table 1 shows, for each state, the official vote difference between Kerry and Bush (with Kerry states designated as positive values and Bush states designated as negative values), the best available latest exit poll result, the difference between the two (where exit polls were more favorable to Kerry than the official vote tally a plus designation is used, with a minus designation when the opposite applied), the sample size of the exit poll, and the significance value. The significance (p) value is the probability that the difference between the exit polls and the official vote tally could have occurred by chance. It is calculated exactly for the 11 battleground states, whereas for the other states it is merely stated whether or not the p value is less than .05 (which designates a one in twenty probability that a discrepancy of that magnitude could have occurred by chance).
Of the 11 battleground states, six of them (Ohio, Florida, Wisconsin, Pennsylvania, Minnesota, New Hampshire) exhibited discrepancies between exit polls and official vote tallies that were statistically significant at the p less than .05 level (with Ohio and New Hampshire the most statistically significant – p= .001 and .00006, respectively), and three more were borderline statistically significant (Colorado, Nevada, New Mexico). All 11 of them were more favorable to Kerry than the official vote tally.
Of the 39 other states (including DC, but not Virginia), 12 exhibited statistically significant differences between the exit polls and the official vote tallies. All of these 12 states exhibited exit polls that were more favorable to Kerry than the official vote tally. Of the 27 states that exhibited no statistically significant differences between their exit polls and the official vote tally, 18 of the exit polls were more favorable towards Kerry than the official vote tally, and 9 were more favorable towards Bush than the official vote tally.
Table 2 shows the same information as Table 1 for the group of battleground states combined together, and for all of the other states combined together. The official vote tally in this case is weighted for the sample size of the exit polls. The p value that corresponds to the discrepancies between the exit polls and official vote tallies in the battleground states was 0, taken out to seven digits, for both groups, although the discrepancy is much greater for the battleground states. That means that the probability of that large of a discrepancy occurring by chance was less than one in twenty million.
Discussion and conclusion
This analysis shows large differences between exit poll results and official vote tallies in the battleground states, in all 11 states the exit poll results being more favorable for Kerry than the official vote tallies. The discrepancy is so large that the probability of that happening by chance is less than one in twenty million. The other states also exhibit some discrepancies, in 12 of the 39 states these discrepancies being statistically significant, again all favoring Kerry in the exit polls, as compared to the official vote tallies. But for the non-battleground states, the overall discrepancy is substantially less than for the battleground states, though still highly statistically significant.
Since chance is essentially ruled out in this analysis as an explanation for the discrepancies, there remain two other possible explanations: Either the exit polls are inaccurate, or the official vote tallies do not reflect voter intention because of a flawed election. If the latter is the case and the exit polls did in fact reflect voter intention, that would mean that Kerry would have won Ohio, and perhaps some other states as well, if not for flaws in the way that the election was conducted. With Ohio’s electoral votes alone he would have won the election.
The hypothesis that the election was flawed is supported by several circumstances, including: 1. the uncovering of several election irregularities within the first few days following the election; 2. the fact that many of the voting machines used to count the votes in this election cannot be audited because of the lack of a paper trail, and the software used to run these machines was not open to public scrutiny; and, 3. the fact that some of the owners of these machines are Republican donors, one of them, head of Diebold Walden O’Dell mentioned in a fund-raising letter that he was "committed to helping Ohio deliver its electoral votes to the president next year."
The alternate hypothesis is that the exit polls were not accurate during this election. This would mean that they were heavily biased towards John Kerry. Although it is theoretically possible that this was the case, I am not aware of any evidence to support this. Exit polls have been used for decades to call elections, and I am not aware that they have ever been said to be inaccurate before, or to be biased towards a particular candidate.
It could be argued that the statistically significant discrepancies between exit polls and official vote tallies in the non-battleground states suggest that widespread bias in the exit polls was present. But that wouldn’t explain why the battleground states were characterized by so much more discrepancy than the non-battleground states. And it could just as well be argued that the same election flaws occurring in the battleground states also occurred in the non-battleground states, although to a lesser degree.
Most U.S. citizens are aware that there was an apparent problem with exit polls during the 2000 election, when Florida was called wrong twice in the same night. But this was not the fault of the exit polls. The first call, which was based largely on exit polls and went to Al Gore, was actually correct with respect to voter intent. Analysis following the election indicated that Gore would probably have received approximately 30,000 more votes if not for voter confusion caused by the infamous “butterfly ballot” used in Palm Beach County. That would have easily erased Gore’s deficit of a few hundred votes. The second call of that night, which went to Bush, was based on a computer error, which gave Gore minus 20,000 votes from a single precinct in Volusia County. Because of the magnitude and the character of this error, and because a paper trail existed, it was quickly identified and corrected. But a less obvious error might not have been identified and corrected.
In conclusion there appears to be good reason to suspect serious flaws in our recently held election. We will never know the extent of them unless they are thoroughly investigated.
Table 1 – Comparison of exit polls by state with official vote tallies –
Presidential election of 2004 – Kerry vs. Bush
State Vote dif* Exit poll dif** Discrepancy*** Sample size p value****
Battleground states
IA - 0.9 1.3 2.2 2502 .29
FL - 5.1 - 0.1 5.0 2846 .008
OH - 2.7 4.2 6.9 1963 .001
WI 0.4 5 4.6 2223 .03
NM - 1.1 2.6 3.7 1951 .10
PA 2.3 8.7 6.4 1930 .005
MI 3.5 5.0 1.5 2452 .47
MN 3.5 9.0 5.5 2178 .01
NH 1.4 10.8 9.4 1849 .00006
CO - 5.7 - 1.8 3.9 2515 .05
NV - 2.7 1.3 4.0 2116 .07
Non-battleground states
AR - 9.9 - 6.8 3.1 1402 NS
HI 8.7 6.6 - 2.1 499 NS
ME 8.2 9.5 1.3 1968 NS
MO - 7.3 - 5.0 2.3 2158 NS
NJ 6.3 10 3.7 1520 NS
OR 4.0 2.4 - 1.6 1064 NS
WA 7.6 9.9 2.3 2123 NS
WV -13.0 - 9.5 3.5 1722 NS
AL -26 -18 8 730 <.05
AK -27 -19 8 910 <.05
AZ -11 - 6 5 1859 <.05
CA 11 8 - 3 1919 NS
CT 10 17 7 872 <.05
DE 7 17 10 770 <.05
GA -17 -14 3 1536 NS
ID -38 -33 5 559 NS
IL 10 14 4 1392 NS
IN -21 -18 3 926 NS
KS -25 -30 - 5 654 NS
KY -20 -18 2 1034 NS
LA -15 -11 4 1669 NS
MA 25 32 7 889 <.05
MD 13 14 1 1000 NS
MT -20 -20.5 - .5 640 NS
MS -20 -13.5 6.5 798 NS
NE -35 -26.5 8.5 785 NE
NY 18 26 8 1452 <.05
NC -12 - 4 8 2167 <.05
ND -27 -32 - 5 649 NS
OK -32 -30 2 1539 NS
RI 21 28 7 809 <.05
SC -17 - 8 9 1735 <.05
SD -21 -24.5 - 3.5 1495 NS
TN -14 -17 - 3 1774 NS
TX -23 -26 - 3 1671 NS
UT -44 -39 5 798 NS
VT 20 30 10 685 <.05
DC 81 82 1 795 NS
Table 2 – Comparison of combined exit polls with official vote tallies for battleground states and
non-battleground states – Presidential election of 2004 – Kerry vs. Bush
Group Vote dif* Exit poll dif** Discrepancy*** Sample size p value****
Battleground - 0.7 3.8 4.5 24525 .0000000
Other states - 4.0 -3.8 0.2 47651 .55
* Percent Kerry official vote total minus percent Bush vote total
** Percent Kerry vote total minus percent Bush vote in latest exit poll – all exit poll data are taken from
Jonathan Simon’s internet article, except for New Jersey, New York, North Carolina, and Wisconsin
(Virginia is not included because I couldn’t find any information on Virginia exit polls)
*** Percent difference in exit poll minus percent difference in official vote tally
**** p value is the probability that the discrepancy between the vote difference and the exit poll data
could be as large as it is on the basis of chance. Exact p values are given for the battleground states.
For the other states, it is denoted only whether or not the p value was statistically significant (p<.05)
or not (NS – not statistically significant, p>.05).
Printer-friendly format
Email this thread to a friend
Bookmark this thread
(1000+ posts)
Send PM |
Profile |
Ignore
