• Report suggests changes in exit poll methodology Nothing To See here...

[princehal]

http://www.cnn.com/2005/ALLPOLITICS/01/19/exit.polls/index.html

WASHINGTON (CNN) -- Exit polls overstated John Kerry's share of the vote on November 2, both nationally and in many states, because more Kerry supporters participated in the survey than Bush voters, according to an internal review of the exit-polling process released Wednesday.

The report said it is difficult to pinpoint precisely why, in general, Kerry voters were more likely to participate in the exit poll than were Bush voters. "There were certainly motivational factors that are impossible to quantify," the report said.

..snip..

[electropop]

1. just looked at a longer piece of Mitovsky BS

Jeez, he sounds truly desperate. Yes, the error was statistically significant in 26 states (TruthIsAll only claims 20), but it's still a really really good poll, sez Mitovsky. And for some reason, Kerry voters just took the poll more often, unlike every year before 2000. Give me a freakin' break.

[Mistwell]

6. Factual inaccuracy

"And for some reason, Kerry voters just took the poll more often, unlike every year before 2000. Give me a freakin' break."

Exept the opposite is true. EVERY year the unweighted data came out over-favoring Democrats. EVERY SINGLE ONE. And that fact is also mentioned in this very same article, and we've posted the actual pre-weighted data from the Washington Monthly here many times.

Why does this myth continue to pop its ugly head up every time exit polls are mentioned on this board? Come on guys, let's try and get this accurate.

Man, lets hope this starts to put the exit polls are "proof" threads to bed and causes people to focus more on ACTUAL fraud (which there is plenty of proof for) rather than this speculative fraud that doesn't help anything and just distracts good people from the important issues.

People were forced to wait in 10 hour lines to vote. People were discouraged from voting. Ballots changed from district to district. Provisional ballots that should have been valid were disqualified. People were registered and voted in multiple states at the same time. People voted multiple times within the same state. People who are not allowed to vote under the law did actually vote.

THOSE facts are all a lot more important than some shifty and inaccurate statistical analysis of incomplete exit polling data that looks just like every other year and proves nothing other than exit polls are not very useful for predicting election outcomes in close races.

[euler]

7. Bravo

---sounds of clapping---

[skids]

9. While to some extent I see your point...

The truth is that these exit polls were flaky for reasons other than undersampling Bush voters.

Their claim that it was simple survey selection bias is not backed up by their adjustments to account for the fact. For example, in "adjusting" the polls to account for the claimed bias by adjusting upwards the coefficient on Bush voter groups, they managed to imply that two to six million more living Bush2000 voters (depending on mortality) than actually exist showed up again at the polls in 2004 (regardless of who they voted for.)

When you can't fix a demographic poll by adjusting the group coefficients, and the results refuse to mesh to such a drastic extent, then something is very wrong.

While there is perhaps more noise on this issue than is helpful, it is and will remain a valid point of contention.

[euler]

11. This sounds like gobbledeegook.

group coefficients

What do you mean?

[skids]

13. In a demographic survey...

...you group the set of respondents depending on their demographic profile. How that grouping is done is heavy math I won't pretend to understand, probably mixed with a little spit and chewing gum.

Once you have the groups, you weight their results, in laymans terms. In mathematical terms, you change the coefficient applied to that group, which should change the survey results to reflect what would have happened if you had managed to give the survey to more people in that demographic group.

So read again and change "group coefficients" to "weighting" and perhaps it will make more sense to you.

[TruthIsAll]

10. "More Kerry voters participated" DUH! Because of MORE Kerry voters.

They have no reasonable explanation.
Of course more Kerry voters participated.

That's because there WERE more Kerry voters than Bush voters.
They have just convicted themselves.

And they blame the weather? Come on.

It's a sick, sick joke.
And America will fall for it.

Come on, Leno.
Come on, Letterman.

Don't let them get away with it.

[greenmutha]

12. My thoughts exactly, TIA!

That's the very first thing that came to my mind when I read this article!

[Mistwell]

17. TIA - Explain prior years then

TIA, at some point you are going to have to deal with the prior years, from 1988 to 2004, all showing a heavy deviation to favor Democrats in ALL exit polls. Unless you are claiming fraud in ALL elections, then your claim just doesn't hold water.

[sui generis]

2. I just saw that horseshit

the spinmeisters are hard at work.

The transitive theorem of spin: If A <> B and B <> C then CLEARLY the only conclusion is that A = C, just because they said so.

[euler]

8. Exactly

[princehal]

3. It sickens me

That we are asked top buy this crap. And that many Americans do. Dear God, what has happened to this country?

[Goldeneye]

4. If I remember correctly the media believed the exit polls too

...not just the bloggers. Also, didn't they have the same percent of democratic and republican voters in the exit polls as in the accepted results? Doesn't that kill the argument that more Kerry voters were polled? The results for independent voters as well as the % of voters who were dem and voted republican and rep. who voted dem would all have to be wrong, yet they aren't saying they were wrong, or making any changes to the numbers.

This is all we get of a 75 page report?

[understandinglife]

5. I suggest caution regarding 'exit poll' information dated anytime...

....after ~ 1am EST, 3 Nov 2004. It is my personal perspective on any subsequent 'exit poll' data released by any pollster or those who pay them.

Unless all relevant information was retrieved by subpoena including all data, analysis, internal communications at ME, communications with the media customers... -- zero credibility.

Unless all personnel involved in polling, analysis and communications of results were required to make statements under oath -- zero credibility.

And, even under those conditions, would expect all materials to be scrutinized by independent, non-partisan information experts, statisticians, and forensics experts.

In other words, when it was obvious to many following the events at CNN and elsewhere after ~ 11pm EST 2 Nov 2004 that something inexplicable was happening, I decided I'd consider everything suspect until the full force of the law and independent investigative experts were brought into the matter.

Events of the past 2+ months have only strengthened my perception of intentional manipulation of the purpose and veracity of 'exit polls.' However, that's all it is; my opinion and its not what I do for a living so I may be way, way incorrect.

Peace.

ONWARD BUSH OPPOSITION

[euler]

14. So, the well known

...democratic overstatement phenomenon continues.

[euler]

15. Mitofsky and MysteryPollster on Nightline tonight

[Kazak]

16. BS!!!

:puke: