PROBABILITY TABLE: Exit Poll Margin of Error vs. Vote Deviation

[TruthIsAll]

For example, if the Bush National exit poll deviation was
3.50% in going from from 47.0% in the preliminary poll to
50.5% in the recorded vote, then the odds are 1 in 288 billion
for this result to have occured by chance alone. 

The 1.0% MOE stated by Edison/Mitofsky in the notes at the
bottom of the exit poll summary screen (WP/NEP) is used for
the calculation. 

If the MOE was 1.50%, the odds are much higher that the
deviation could have been due to chance: 1 in 416,000.

We use the Normal Distribution Function, calculated over a
matrix of the two variables: Deviation and MoE (1.96* standard
deviation at the 95% confidence level). The standard deviation
= MoE/1.96 is entered in the Excel function.

The probability P is given by:
P = 1 - NORMDIST(Actual vote, Exit poll, MoE/1.96,TRUE)

P = 1 - NORMDIST(.505, .470, .01/1.96, true) = 1/288 billion.


		Probability Matrix						
	Deviation from preliminary exit poll to the final vote						

		Exit Poll Margin of Error						
Dev.	1.00%	1.50%	1.75%	2.00%	2.25%	2.50%	2.75%	3.00%
		          1 in 						
1.00%	40	10	8	6	5	5	4	4
1.50%	609	40	22	14	10	8	7	6
2.00%	23k	223	80	40	25	17	13	10
2.50%	2.1mm	1.8k	391	140	68	40	27	20
3.00%	485mm	23k	2.6k	609	223	107	62	40
3.50%	288bn	416k	23k	3.3k	871	330	159	90
4.00%	450tr	11.5m	267k	23k	4k	1.1k	459	223

[Mistwell]

1. Just a question

I know I usually disagree with you TIA, but this one is an honest question. I wouldn't blame you for not believing me on that, but it's the truth.

Mitofsky says that early in the afternoon on November 2nd, preliminary weightings for the national exit poll overstated the proportion of women in the electorate. The problem was caused by a programming error involving the gender composition that was being used for the
absentee/early voter portion of the national exit poll. The target proportion for the national absentee age-race-sex responses was not being computed correctly. The weighting problem that led to the original waves of the national exit poll showing 58% of the national voters were female.

My question is, is the early exit poll you are using for your calculations based on the early poll mentioned above, or is it from one of the ones that happened after they corrected the above-mentioned problem? I actually suspect it's the later, but I wanted to check.

[TruthIsAll]

2. The final exit poll had it at 54/46, as did the preliminary. Never changed

	CNN	7:58pm	11027	Sample		WP	12:22am	13047	Sample	CNN	2:04pm	13660	Sample
     
	Mix1	Bush	Kerry	Nader		Mix2	Bush	Kerry	Nader		Mix3	Bush	Kerry	Nader
GENDER
Male 	46.0%	51%	47%	1%		46%	52%	47%	1%		46%	55%	44%	0
Female 	54.0%	45%	54%	1%		54%	45%	54%	1%		54%	48%	51%	0
	100%	47.76%	50.78%	1.00%		100%	48.22%	50.78%	1.00%		100%	51.22%	47.78%	0.00%

See the post below for the links...

[Mistwell]

4. hmm?

Huh. So, what the heck is Mitofsky talking about?

[TruthIsAll]

3. Here are links to the exit polls and a comparative study.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x340029

Preliminary I
11/02 7:38pm CNN 11027 respondents
http://www.exitpollz.org/CNN_national2.htm

Preliminary II
11/03 12:22am WP 13047 respondents
http:
//media.washingtonpost.com/wp-srv/politics/elections/2004/graphics/exitpolls_us_110204.gif

FINAL
11/03 2:04pm CNN 13660 respondents
http://www.cnn.com/ELECTION/2004/pages/results/states/US/P/00/epolls.0.html