THIS IS A VERY IMPORTANT POST: THE EXIT POLL TIMELINE

[TruthIsAll]

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x216653#219198

POST # 42

[bluestateguy]

1. One question I want answered about the exit polls

How were the precincts chosen for the exit polls? Was the process used to select exit poll precincts one that guaranteed maximum randomness?

[DoYouEverWonder]

2. I've been trying to get the answer to that question too

I can tell you in my county, which is a liberal stronghold, they picked only one precinct and it was the most conservative one they could find.

[bleever]

4. The process was compromised; why wouldn't that extend to choosing

the precincts.

And remember: ROVE HAD ACCESS TO COUNTY AND PRECINCT-LEVEL RESULTS! (caps for TIA!)

And we all were wondering, what the hell does a political strategist need a frickin' missile launch control room to watch the results from every little hamlet and borough when the damn campaign is over! They already spent all their money; what the hell was he moving around?


WTF?

[McCamy Taylor]

3. So someone floated misinformation in the Internet?

:grr:

[k8conant]

6. They floated misinformation and misvotes across the board...

the exit polls changed to incorporate the misinformation and misvotes.

[bleever]

5. Post also here:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x223436

A Must Read.

[Blue22]

7. The reweighting changed the national results not the last 2000 interviews.

TIA,  I think your analysis is correct.  It is impossible for
the final 2,000 interviews to have turned the prediction to
Bush.  Nevertheless, other posts have noted that it is not
necessary for the final 2,000 exit poll interviews to have
gone overwhelmingly for Bush.  It is only necessary that the
re-weighting based on final tallied votes reflected a dramatic
surge in turnout in republican districts, which is consistent
with some Ohio and Florida reports.  This is demonstrated in
the analysis below.  In this hypothetical case that conforms
to how the exit polls were drawn, Mitovsky picked two
precincts to represent the national popular vote, one heavily
dem and the other heavily rep.  He conducted 45 interviews in
the two precincts by 7PM.  Based on past turnout history in
the two precincts he predicted that Kerry would end up winning
51-48.  Later that night, the tallied votes showed a much
greater turnout for the Rep precinct.  So Mitovsky had to
re-weight his survey results to give more weight to the 25
interviews in the Rep precinct resulting in a predicted Bush
win of 51-48.

The question is why Mitovsky, the father of exit polls, was so
wrong on predicted turnout.  We need to look at the raw
samples, which precincts he chose, and why his predicted
turnout was so wrong.  Was it padding (adding votes)? 
Changing undervotes to full votes (default issue)? Or did this
reflect true turnout?

B-)

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

How re-weighting can change the Predicted Exit Poll Results
without adding any new survey interviews.					     
	            		      Precincts			Total
Precinct		        	1	2		
Exit Poll Sample			20	25		45
Expected Vote			        200	100		300
						
Sample Results 7PM	Kerry		55%	43%		51%
                	Bush		44%	57%		48%
Expected Votes 7PM	Kerry		110	43		153
                	Bush		88	57		145

Tallied Votes 9PM			200	250		450
Re-weighted Sample	Kerry		55%	43%		48%
                	Bush		44%	57%		51%
Re-weighted Votes 9PM	Kerry		110	108		218
                	Bush		88	143		231

[Igel]

10. Well, I'm glad somebody crunched some numbers.

Again, if you (and I don't mean you, Blue) don't know the assumptions underlying your stats, you don't know what they mean. Excel makes the calculation of stats trivial, but it's choosing the tests and setting the parameters that takes smarts.

[berniew1]

11. Remember there was a lot of systematic suppression of minorities

would that do it? Both in registration and counting votes; much longer lines in Dem precincts due to shortage of machines.

[newyawker99]

14. Hi Blue22!!

Welcome to DU!! :toast:

[euler]

15. You're new, TIA has been here forever, but....

...TIA needs to be asking you questions, not the other way around. Last I heard, TIA still thinks re weighting for the actual vote is, by itself, evidence of fraud (although he may have recently seen the light.) You're a breath of fresh air. Thanks.

[bleever]

8. Kickola.

[snot]

9. kick

[papau]

12. kick for freedom - and the hope that some in the media grow balls

:-)

[Al-CIAda]

13. kick...and thanks for all your work -eom

[BreakForNews]

16. That IS important !!

What's your own take on it right now TIA?

(esp. for folks who have not followed this closely)

Original post:
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x216653#219198