1988-2000 Exit Poll Deviation trend was moving to ZERO - until 2004

[TruthIsAll]

When Clinton ran in 1992 there were three candidates. Clinton beat Bush by 43-38%. Perot ended up with 19%.

And in '92 there were no touchscreens, so a lot of democratic spoilage was from punched cards. Now the central tabulators, touchscreens and optiscans do the work that the spoiled punched cards use to do for the Repubs.

Witness 2000, and how the exit poll matched Gore's final vote.

In any case the trend was headed to near zero - until it reversed in 2004.

Questions:
1. Is it just a coincidence that in 2004 the new touchscreens account for 30% of vote - and the trend changed?

2. Is it just a coincidence that Republican deviations always exceed those of the Dems?

3. Could the absolute deviations of the discrepancies between Rep and Dem be from third party candidates? Otherwise, wouldn't one expect the same percentage deviations, with only the sign changed?

WHAT CAUSED THE TREND REVERSAL?

[funflower]

1. For the graphically challenged, such as myself....

Would you explain in more detail what this chart means?

[TruthIsAll]

6. Explanation.

This chart is showing the deviations from the exit poll to the final vote for Democratic and Republican candidates since 1988.

The straight lines reflect the "best fit" linear trend fitting the all the deviations, including the last set (2004).

Had 2004 been excluded from the chart, BOTH trend lines (from 1988-2000) would have converged near the x-axis (or Y= ZERO) in 2004.

The question remains: WHAT CAUSED THE REVERSAL OF TREND IN 2004?

[liam_laddie]

2. kick!

TIA - congrats on making an easy-to-understand image
of a complex statistic. Another stitch in the proof-of-manipulation fabric. Thanks!
The cause was obviously E/M's electricity being cut off during
the early a.m. of Nov 3. Karl was seen scurrying away with some
bolt cutters in his hand, and burn marks on his feet and skull.
Plus the all of E/M's UPS's went dead. Plus the 613 overnight voters had to be awakened. Or was it the network vans outside beaming change-rays into the computers? Plus...oh, never mind.

[impeachthescoundrel]

3. Vote Manipulation

Pure and simple

[Wow]

4. New methods of manipulation

Oops, I must mean new methods of taking polls. Yeah right.

[On the Road]

5. To Answer the Question

it is necessary to understand what the causes are of the discrepancy and why they changed over time.

Personally, I don't know.

[captain crunch]

7. So if I'm reading this right, except for 2000,

04 election results were closer to the exit polls than any election since before 88. Is that right?

[TruthIsAll]

14. You must look at the results in context.

Clinton won handily in both '92 and '96, with Perot at 19% and 8%.
Spoiled punched cards were the major culprit.

Gore won the popular vote in 2000.
Spoiled punched cards were the major culprit (see FL where 175,000 votes were spoiled, 75% of them Gore votes).

Bush gained from the exit polls, while Gore stayed flat.
Nader hurt Gore.

In 2004, touchscreens were installed in place of many punch cards.
So the exit polls should have been right on the money, right?
Wrong. With BBV came a reversal in trend.

Coincidence?

[captain crunch]

8. Haven't you claimed in the past

that previous exit polls accuratly predicted the winner up to the election of 2004?
Am I missing something? This graph doesn't show that at all.

[TruthIsAll]

9. Good point. The polls which I was referring to were the FINAL exit polls

At the time, I was unaware of the fact that the final, true exit polls were re-weighted to match the vote count.

The graph depicts deviations from the final preliminary exit poll. For example, in 2004, the preliminary exit poll of the first 13047 respondents had Kerry winning by 51-48%.

The final 660 respondents brought the total to 13660, which was re-weighted - and voila, Bush is the winner by 51-48%.

The only problem with that is I have shown that the 13047 National Exit Poll matched the state polls within .06% - AFTER RE-WEIGHTING THE INDIVIDUAL STATE POLLS TO MATCH THE TOTAL STATE VOTE.

SO WHY WAS THERE ANOTHER RE-WEIGHTING?
TO MATCH THE BUSH VOTE, NOT THE TOTAL VOTE.

EXAMPLE:
THE 41% WEIGHTING OF BUSH 2000 VOTERS IN THE 13047 EXIT POLL (WHICH KERRY WON) WAS CHANGED TO AN IMPOSSIBLE 43% IN THE FINAL 13660 EXIT POLL (WHICH KERRY LOST).

I say impossible, because 43% of 122.26 million is 52.57mm, which is 2.11 mm more than the 50.456 million votes that Bush got in 2000. Where did these 2.11 mm voters come from? And what about the 1.75 mm Bush 2000 voters who died? Did they come back to life to vote?

What does it all mean? The original, preliminary exit polls reflect true voter intent. The final poll is massaged to match the recorded vote - and assumes that the recorded vote is correct, that there was ZERO ballot spoilage and ZERO miscounted votes.

[TruthIsAll]

10. Febble, in the spirit of healthy debate....

You referred to the 1992 exit poll discrepancy.

What are your thoughts about the 1988-2000 trend reversal in 2004?

[Febble]

15. My thoughts are here

http://www.dailykos.com/story/2005/4/6/8028/83645

[autorank]

11. The "Truth" is so inconvenient...but it cannot be ignored.

Great post.

[LSdemocrat]

12. Umm, to me it looks like 2000 is the outlier, not 2004

[TruthIsAll]

13. Why do you say that?

/