2004 Exit Poll "Red Shift" As Seen In Vote Numbers

[althecat]

Please note. I posted this yesterday before realising that some gremlins had gotten into the data. I have now fixed it up however and the conclusions are similar... with the exception of New York. Bush's massive gains in NY can now be seen in their proper light...

http://www.scoop.co.nz/mason/stories/HL0411/S00270.htm

Scoop Images: 2004 Exit Poll "Red Shift" As Seen In Vote Numbers

By Scoop Co-Editor Alastair Thompson with DUer Ed Shalom

The following analysis was sent to Scoop as a response to the Scoop article " Complete US Exit Poll Data Confirms Net Suspicions"concerning the discrepancies between the 2004 US exit polls and the Presidential election's actual results. The exit polls showed Democrat John Kerry winning the election comfortably, but the actual election recorded Republican President George Bush winning by over 3 million votes.

Democratic Underground forum member Ed Shalom decided to illustrate the observed State by State "Red Shift" (the difference between the US 2004 General Election exit polls - conducted by Edison Mitofsky - and the actual results), in terms of actual numbers of votes rather than simply as percentages.

As some US States have considerably more voters than others this image enables the reader to view where and how many votes were picked up by George Bush over and above what he was expected to receive in the 4pm exit polls.

The numbers are obtained simply by multiplying total Presidential votes cast per state by the "red shift" percentage.

When presented in this way the data shows a remarkable conclusion, three of the four biggest vote "red shifts" occurred in the three battleground states of Florida, Pennsylvania, and Ohio. Notably the two giant states in terms of Democratic Party vote bookend the series with New York showing the biggest gain in Bush vote numbers vs the exit polls and California showing the biggest vote loss.

In New York Bush picked up 24% more votes than he did in 2000, or 544,973. California is one of a tiny handful of states where Bush did not gain votes in comparison to his 2000 poll, in the initial count he actually lost 34,000 votes..

Coincidentally Florida, Pennsylvania and Ohio were also the three states addressed in the academic paper "The Unexplained Exit Poll Discrepancy" by Dr Stephen F. Freeman. In this paper Dr Freeman discusses the use and methodology of exit polls extensively. He concludes that the odds of the level of variation from exit polls to actuals occurring in these three states as did so was 250 million to one. (note: Dr Freeman has agreed with critics of his report that this may overstate the case a little, but the odds of this occurring remain very very low and statistically speaking near impossibility.)

Dr Freeman ends his report saying:

"Systematic fraud or mistabulation is a premature conclusion, but the election's unexplained exit poll discrepancies make it an unavoidable hypothesis, one that is the responsibility of the media, academia, polling agencies, and the public to investigate."

To date no major US mainstream media outlet (with the sole exception of Keith Olbermann's Countdown show on MSNBC) has addressed the exit poll issue in spite of the fact that the data used in the exit polls was commissioned by them.

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

THE GRAPHS & DATA

The 2004 Exit Poll – Bush "Red Shift" In Percentage Terms


The 2004 Exit Poll – Bush "Red Shift" In Terms Of Pure Vote Numbers

To view the data for this second graph in tabular form see…
http://www.scoop.co.nz/mason/stories/HL0411/S00270.htm

[imenja]

1. Question, why 4 pm exit poll numbers?

Are there no available exit poll data for the times the polls closed?

[althecat]

2. No... we could do the same thing with the jonathan simon data

I might do that.. as it would be illustrative in showing the difference between the early and late exit polls.

[imenja]

3. I think that would be a good idea

because people could always argue pro-Bush voters turned out later, especially if they had read on the web that the exit polls were favoring Kerry. It is also possible that some likely Kerry voters might not have gone to the polls thinking Kerry was a shoe in.

[Straight Shooter]

4. We can argue back that soccer moms vote early in the day

No matter what kind of damage control the Repubs try to exert on these exit polls, we can counter them. We can argue that working people, union people, also vote later in the day.

They shot themselves in the foot by claiming exit polls aren't reliable, when we know damn well they are.

[imenja]

5. yes but

if you want to make a strong, persuasive argument, you want to present the best evidence possible.
I appreciate your efforts.

[althecat]

9. Done and posted below... see jonathan simon data post n/t

[althecat]

10. Thanks for the suggestion....

I have incorporated the Jonathan Simon data into the analysis fully now... with graphs of both now in the story. It makes it more complete and less easy to knock down.

Cheers

[Yupster]

6. CNN has the end of day exit polls

posted here ...

http://www.cnn.com/ELECTION/2004/pages/results/states/US/P/00/epolls.0.html

They show Bush winning 51-47 %.

Of course they don't prove anything because I've seen some contend that these are just another part of the conspiracy as CNN apparently revised their figures in the middle of the night to show Bush winning.

[althecat]

7. Not CNN... Edison Mitofsky

revised the final figures to match the actual vote by incorporating the actual voting data. These are only useful for demographic purposes.

[althecat]

8. Same analysis using the Jonathan Simon - 12pm Exit Poll data

Note figs for NY, NJ, NC and Virginia have been guesstimated by halving the early poll discrepancy as this roughly fits the pattern.

FIRST GRAPH.. (in same order as the 4pm data shows - this enables you to see the variation between the early and later data.)



SORTED BY GROSS VOTE NUMBERS

[imenja]

11. Great. Thanks!

[Be Brave]

12. Some unbiased critiques of the plots. althecat, important please read.

Actually, I am biased because I am one of those people who are hoping that massive fraud is proven. But if we're gonna do this, let's do it right, lest the freepers prove us wrong, laugh at us, and shove our analyses to our faces.

I think that the value which is plotted in the original graph, i.e. the "redshift" is a bit misleading. A little algebra will explain what I am trying to say. The redshift, as I understand it, is defined as the difference between the Bush lead given by the actual votes and the Bush lead given by the exit polls. So, for example, in the case of Delaware, using the data listed in scoop, the redshift is:

redshift DE = (46 - 53) - (41.5 - 58.5) = -7 + 17 = 10

However, what I think should be plotted is the difference between the Bush actual votes and the Bush exit poll value, which I will call the Bush gain. The reason why I think the redshift is misleading is because the redshift doubles the Bush gain and, furthermore, contains the third party share of the votes. Here is the proof: Let

VotesBush = Bush actual votes (%)
VotesKerry = Kerry actual votes (%)
VotesThird = Third party candidates actual votes (%)
ExitBush = Bush exit poll data (%)
ExitKerry = Kerry exit poll data (%)

Then,

(1) redshift = (VotesBush - VotesKerry) - (ExitBush - ExitKerry)

Rearranging terms, we get

(2) redshift = (VotesBush - ExitBush) + (ExitKerry - VotesKerry)

Now, looking at the scoop data, the sum of the Bush and Kerry votes does not always equal 100%, presumably because the difference went to third party candidates. However, the Bush and Kerry exit poll data always seem to add up to 100%, presumably because the exit poll survey did not include third party candidates. In equation form, this means

(3) VotesBush + VotesKerry + VotesThird = ExitBush + ExitKerry = 100%

So, from equation (3), we get:

(4) VotesBush - ExitBush + VotesThird = ExitKerry - VotesKerry

Substituting the left hand side of equation (4) into equation (2),
we get

(5) redshift = (VotesBush - ExitBush) + (VotesBush - ExitBush + VotesThird)

In other words,

(6) redshift = 2*(VotesBush - ExitBush) + VotesThird

So equation (6) shows that the redshift is twice the Bush gain PLUS the third party share of votes. In my opinion, this is misleading. Because what we really want to study is the Bush gain in the actual votes compared to the exit polls, and not twice this value. And we also do not want to include the third party share of votes in the redshift. In other words, the redshift plot approximately DOUBLES the actual Bush gain.

My second critique is that the graphs should also plot the margin of error of the exit polls. I think that this is but the responsible thing to do. And I think it is also the responsible thing to do to redo the plots to reflect the correct numbers.

Thanks for your attention.

[althecat]

13. I sort of understand but will have to think about this a bit more...

:think:

[Be Brave]

14. Thank you, that's all I ask. n/t

[Be Brave]

16. Comparing your plots of votes versus that of Jonathan Simon proves my

point. Notice that Simon's vote difference is about half of your plotted votes.

[althecat]

17. Nothing like a good sleep to help one understand...

In plain english. I have calculated redshift as the difference in the margins, but that includes votes that Kerry did not get as well as those Bush did get.

I will recalculate the 4pm redshift's on this basis...

This is good as it shows there is little difference between the 4pm and 12pm exit polls. The thirdparty votes are a bit of a red herring as they have almost no impact.

Cheers
al

[althecat]

18. I have edited the original to take into account your sensible point...

http://www.scoop.co.nz/mason/stories/HL0411/S00270.htm

Here is where Jonanthan Simon explains the red shift calc.

DATA DESCRIPTION
Exit Poll Data Bush% then Kerry%, # of respondents, then time
of poll ET, and "Red" Shift%
Note: Red Shift = <(Btab% - Bep%) + (Kep% - Ktab)>/2
tab= tabulated vote, ep=exit poll
The number is positive with net movement toward Bush, negative
(blue shift) with net movement toward Kerry. I'll take Florida
(early) as an example:

Exit Poll % : B=49.8% K=49.7%
Tab (99% precincts) B=52% K=47%
Red Shift: <(52% - 49.8%) + (49.7% - 47%)>/2 = (2.2% +
2.7%)/2 = +2.5%

[althecat]

19. I have added a note to my report saying

Red Shift = (Bush Final Margin - Bush Exit Poll Margin)/2
Red Shift Votes = Red Shift * Presidential Votes Cast

[althecat]

20. Note if the turnout figures are accurate...

i.e. vote stealing was not accomplished by ballot stuffing but by changing votes from Bush to Kerry then the idea of red shift in terms of the margin difference still has some validity.

[Eloriel]

15. Speaking of exit polls, did ya'll see this?

Zogby Speaks Out
According to Zogby, it would have required "wrong sampling in wrong areas throughout the country," or the purposeful manipulation of data to obtain exit poll results so significantly different from the official totals. He viewed neither as a possibility.
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=64057&mesg_id=64057
LINK: http://www.commondreams.org/headlines04/1118-11.htm

[althecat]

21. Pretty damn cool...

[ccarter84]

22. cbs exit polling data?

I heard rumors that there was exit polling data released by cbs in 5 minute increments...but can't find it anyone else hear such rumors?

[althecat]

23. There were several live exit poll results that had...

...changing results through the night. MSNBC, CNN and Fox had em too... I am not certain how frequently they were updated. Finally at around 1am they incorporated actual voting stats into the exit poll results and rendered them effectively useless for auditing purposes.