Exit Poll Probability Analysis - state by state

[TruthIsAll]

I have posted this table a number of times.

But I am reposting because: 
1) There are many new DUers who haven't seen it 
2) to clarify the assumptions and calculations. 

3) Kerry's exit poll and vote percentages are based on the
exit poll numbers downloaded by Simon at 12:22 AM Nov 3. 

4) The numbers have been converted to two-party percentages so
as to total 100%. The purpose was to maintain comparability to
other pre-election projection models. The adjustments are
proportional to the exit poll and vote percentages. The
post-adjusted numbers are in proportion to the original
numbers. There is NO BIAS for Kerry or Bush.

For example: 
Assume Bush won the state vote 50%-49% with 1% going to third
parties.
Then Bush's adjusted two-party percentage is 50/99 = 50.505%
Kerry's two-party percentage is 49/99 = 49.495%

These are the relevant statistics from the table:
N = exit poll sample size
Poll = Kerry's exit poll percentage
Vote = Kerry's vote percentage
Diff = Kerry deviation from exit poll to vote

MOE = Margin of error = 1/sqrt(N)
StDev = standard deviation = MOE /1.96 (95% confidence level)

Prob = Probability of vote deviation based on Diff and StDev
Prob is calculated for each state using the Excel Function:
Prob = 1-Normdist(Poll, Vote, StdDev, true)
Odds = odds of occurrence = 1 in (1/Prob)

Analysis Summary
----------------
Bush vote tallies exceeded the exit poll MOE in 16 states.
The probability of this occurence is given by the Excel
formula:
Prob = 1- BINOMDIST(15,51,.025,TRUE)
The odds of this occurring due to chance alone: 1 in 13.5
trillion. 

Deviations without regard to margin favored Bush in 41 states.
The probability of this occurence is given by the Excel
formula:
Prob = 1- BINOMDIST(40,50,0.5,TRUE)
The odds of this occurring due to chance alone: 1 in 356,000. 

Unweighted average Kerry deviation: 1.80%.
Unweighted average state MOE: 2.85%.
MOE was significantly lower than the average (2.85%) in the
battleground states where the sample size was large.
A total of 73,607 were polled nationwide.					


St	N	Poll	Vote	Diff	StDev	MoE	Prob  Odds	>MoE?	Favor
AK	910	40.5%	36.1%	-4.4%	1.7%	3.3%	0.005	222	yes	Bush
AL	730	41.0%	37.0%	-4.0%	1.9%	3.7%	0.017	59	yes	Bush
AR	1,402	46.6%	45.5%	-1.1%	1.4%	2.7%	0.200	5		Bush
AZ	1,859	47.0%	44.4%	-2.6%	1.2%	2.3%	0.015	65	yes	Bush
CA	1,919	54.0%	55.6%	1.6%	1.2%	2.3%	0.909	1		Kerry

CO	2,515	49.1%	47.5%	-1.6%	1.0%	2.0%	0.055	18		Bush
CT	872	58.5%	55.1%	-3.4%	1.7%	3.4%	0.025	41	yes	Bush
DC	795	91.0%	90.9%	-0.1%	1.8%	3.5%	0.480	2		Bush
DE	770	58.5%	53.5%	-5.0%	1.8%	3.6%	0.003	289	yes	Bush
FL	2,846	50.5%	47.5%	-3.0%	1.0%	1.9%	0.001	1,305	yes	Bush

GA	1,536	43.0%	41.4%	-1.6%	1.3%	2.6%	0.112	9		Bush
HI	499	53.3%	54.5%	1.2%	2.3%	4.5%	0.707	1		Kerry
IA	2,502	50.6%	49.5%	-1.2%	1.0%	2.0%	0.129	8		Bush
ID	559	33.5%	30.6%	-2.9%	2.2%	4.2%	0.090	11		Bush
IL	1,392	57.0%	55.0%	-2.0%	1.4%	2.7%	0.072	14		Bush

IN	926	41.0%	39.4%	-1.6%	1.7%	3.3%	0.169	6		Bush
KS	654	35.0%	37.4%	2.4%	2.0%	3.9%	0.883	1		Kerry
KY	1,034	41.0%	40.0%	-1.0%	1.6%	3.1%	0.264	4		Bush
LA	1,669	44.5%	42.4%	-2.1%	1.2%	2.4%	0.048	21		Bush
MA	889	66.0%	62.6%	-3.4%	1.7%	3.4%	0.024	41	yes	Bush

MD	1,000	57.0%	56.6%	-0.4%	1.6%	3.2%	0.394	3		Bush
ME	1,968	54.7%	54.1%	-0.7%	1.2%	2.3%	0.282	4		Bush
MI	2,452	52.5%	51.5%	-1.0%	1.0%	2.0%	0.170	6		Bush
MN	2,178	54.5%	51.5%	-3.0%	1.1%	2.1%	0.003	316	yes	Bush
MO	2,158	47.5%	46.0%	-1.5%	1.1%	2.2%	0.086	12		Bush

MS	798	43.3%	40.0%	-3.3%	1.8%	3.5%	0.036	28		Bush
MT	640	39.8%	39.8%	0.0%	2.0%	4.0%	0.507	2		Kerry
NC	2,167	48.0%	44.0%	-4.0%	1.1%	2.1%	0.000	7,613	yes	Bush
ND	649	34.0%	36.4%	2.4%	2.0%	3.9%	0.881	1		Kerry
NE	785	36.8%	32.3%	-4.4%	1.8%	3.6%	0.007	136	yes	Bush

NH	1,849	55.4%	50.5%	-4.9%	1.2%	2.3%	0.000	54,025	yes	Bush
NJ	1,520	55.0%	53.5%	-1.5%	1.3%	2.6%	0.132	8		Bush
NM	1,951	51.3%	49.5%	-1.8%	1.2%	2.3%	0.059	17		Bush
NV	2,116	49.4%	48.5%	-0.9%	1.1%	2.2%	0.218	5		Bush
NY	1,452	63.0%	59.2%	-3.8%	1.3%	2.6%	0.002	458	yes	Bush

OH	1,963	52.1%	49.0%	-3.1%	1.2%	2.3%	0.004	282	yes	Bush
OK	1,539	35.0%	34.0%	-1.0%	1.3%	2.5%	0.221	5		Bush
OR	1,064	51.2%	52.0%	0.8%	1.6%	3.1%	0.695	1		Kerry
PA	1,930	54.3%	51.0%	-3.3%	1.2%	2.3%	0.002	504	yes	Bush
RI	809	64.0%	60.6%	-3.4%	1.8%	3.5%	0.029	34		Bush

SC	1,735	46.0%	41.4%	-4.6%	1.2%	2.4%	0.000	11,036	yes	Bush
SD	1,495	37.8%	39.4%	1.6%	1.3%	2.6%	0.892	1		Kerry
TN	1,774	41.5%	43.0%	1.5%	1.2%	2.4%	0.892	1		Kerry
TX	1,671	37.0%	38.4%	1.4%	1.2%	2.4%	0.866	1		Kerry
UT	798	30.5%	27.6%	-2.9%	1.8%	3.5%	0.051	20		Bush

VA	1,431	48.0%	45.5%	-2.5%	1.3%	2.6%	0.032	32		Bush
VT	685	65.0%	60.2%	-4.8%	1.9%	3.8%	0.007	144	yes	Bush
WA	2,123	54.9%	53.5%	-1.4%	1.1%	2.2%	0.101	10		Bush
WI	2,223	52.5%	50.5%	-2.0%	1.1%	2.1%	0.033	31		Bush
WV	1,722	45.3%	43.4%	-1.8%	1.2%	2.4%	0.069	14		Bush
WY	684	30.9%	29.6%	-1.3%	2.0%	3.8%	0.251	4		Bush

Avg	1,443	48.8%	47.0%	-1.79% 1.46%	2.85%	0.11	9		Bush
Med	1,495	49.1%	47.6%	-1.81% 1.32%	2.59%	0.07	14		Bush
Min	499	30.5%	27.5%	-4.96% 0.96%	1.87%	0.00	1		
Max	2,846	91.0%	90.9%	2.37%  2.28%	4.48%	0.91	54,025		

[starmaker]

1. thanks

biggest obvious

largest diversions
new hampshire for primary relief
south carolina to pick candidate

it looks like major inroads in a lot of states

[TruthIsAll]

6. And we must also mention the National Exit Poll

Kerry led by 50.8% - 48.2% in the National Exit Poll

The sample-size was 13,047.
The margin of error was 1.0%.

Bush won the popular vote by 51.2% - 48.4%
That represnts a 3.0% deviation from the exit poll.

The probability of this occurence is given by:
Probability = 1 – Normdist(.482,.512, .01/1.96,true)

The odds are 1 out of 485 million

[MissBrooks]

2. Too bad Exit Polls Don't Count

[New Earth]

4. too bad votes don't count

[chomskysright]

3. thanks have loved your work and been watching...

I have circulated this around.

[ClintCooper2003]

5. TIA, you should really visit uselectionatlas.org and look at each

state to get the final percentages, including absentee votes. In California, Kerry's final percentage was about 54%.

[Chi]

7. Seems you are correct...

Heya ClintCooper
According to Cal. S.O.S. 54.4% went to Kerry.
http://www.ss.ca.gov/elections/sov/
2004_general/formatted_pres_detail.pdf

[TruthIsAll]

8. You won't believe this. I have just updated the numbers and...

Today I updated the analysis to include the current vote
totals, as requested by another DUer. I had used the
preliminary results as of Nov 3, disregarding any changes to
the totals.

The final voting numbers make the results even more
improbable, if that is possible.

I do request any enterprising DUer to check my numbers.
Remember, if you do, that the numbers have been converted to
their 2-party equivalents and add to 100%.

In the original analysis: 
1) 16 states deviated beyond the MOE to Bush.
The latest numbers now show that in fact TWENTY (20) did. 

2) 41 states deviated to Bush, regardless of MOE. 
The latest numbers now show that FORTY-THREE (43) did. 

Kerry's exit poll and vote percentages are based on the
exit poll numbers downloaded by Simon at 12:22 AM Nov 3. 

The numbers have been converted to two-party percentages
so as to total 100%. The purpose was to maintain
comparability to other pre-election projection models. The
adjustments are proportional to the exit poll and vote
percentages. The post-adjusted numbers are in proportion to
the original numbers. There is NO BIAS for Kerry or Bush.

For example: 
Assume Bush won the state vote 50%-49% with 1% going to third
parties. 
Bush's adjusted two-party percentage is 50/99 = 50.505%
Kerry's two-party percentage is 49/99 = 49.495%

These are the relevant statistics from the table:
N = exit poll sample size
Poll = Kerry's exit poll percentage
Vote = Kerry's vote percentage
Diff = Kerry deviation from exit poll to vote

MOE = Margin of error = 1/sqrt(N)
StDev = standard deviation = MOE /1.96 (95% confidence level)

Prob = Probability of vote deviation based on Diff and StDev
Prob is calculated for each state using the Excel Function:
Prob = 1-Normdist(Poll, Vote, StdDev, true)
Odds = odds of occurrence = 1 in (1/Prob)

Analysis Summary
----------------
Bush vote tallies exceeded the exit poll MOE in 20 states.
The probability of this occurence is given by the Excel
formula:
Prob = 1- BINOMDIST(19,51,.025,TRUE)

The odds of this occurring due to chance for 16 states is 1 in
13.5
trillion. 

For 20 states, it is beyond the capacity of Excel to display
the result - it's 30 zeros to the RIGHT of the decimal point.
****** The probability is therefore VIRTUALLY ABSOLUTE
ZERO*****


Poll deviations without regard to the margin of error favored
Bush in 43 states.

The probability of this occurence is given by the Excel
formula:
Prob = 1- BINOMDIST(42,51,0.5,TRUE)
The odds of this occurring due to chance alone: 1 in 2.9
MILLION.


The unweighted average Kerry deviation from the exit polls:
2.24%.
The unweighted average MOE for all states: 2.85%.

The MOE was significantly lower than the average (2.85%) in
the
battleground states where the sample size was large.
A total of 73,607 were polled nationwide.					

St	N	Poll	Vote	Diff	StDev	MoE	Prob  	 1 in	>MoE?	Favor
AK	910	40.5%	35.6%	-4.9%	1.7%	3.3%	0.002	524	yes	Bush
AL	730	41.0%	36.9%	-4.1%	1.9%	3.7%	0.014	70	yes	Bush
AR	1,402	46.6%	44.6%	-2.0%	1.4%	2.7%	0.066	15		Bush
AZ	1,859	47.0%	44.4%	-2.6%	1.2%	2.3%	0.014	72	yes	Bush
CA	1,919	54.0%	54.4%	0.4%	1.2%	2.3%	0.645	2		Kerry

CO	2,515	49.1%	47.1%	-2.0%	1.0%	2.0%	0.024	42	yes	Bush
CT	872	58.5%	54.3%	-4.2%	1.7%	3.4%	0.008	130	yes	Bush
DC	795	91.0%	89.4%	-1.6%	1.8%	3.5%	0.193	5		Bush
DE	770	58.5%	53.4%	-5.1%	1.8%	3.6%	0.003	385	yes	Bush
FL	2,846	50.5%	47.1%	-3.4%	1.0%	1.9%	0.000	4,799	yes	Bush

GA	1,536	43.0%	41.4%	-1.6%	1.3%	2.6%	0.112	9		Bush
HI	499	53.3%	54.0%	0.7%	2.3%	4.5%	0.622	2		Kerry
IA	2,502	50.6%	49.3%	-1.4%	1.0%	2.0%	0.091	11		Bush
ID	559	33.5%	30.3%	-3.2%	2.2%	4.2%	0.067	15		Bush
IL	1,392	57.0%	54.8%	-2.2%	1.4%	2.7%	0.056	18		Bush

IN	926	41.0%	39.3%	-1.7%	1.7%	3.3%	0.150	7		Bush
KS	654	35.0%	36.6%	1.6%	2.0%	3.9%	0.792	1		Kerry
KY	1,034	41.0%	39.7%	-1.3%	1.6%	3.1%	0.204	5		Bush
LA	1,669	44.5%	42.3%	-2.2%	1.2%	2.4%	0.040	25		Bush
MA	889	66.0%	62.1%	-3.9%	1.7%	3.4%	0.011	90	yes	Bush

MD	1,000	57.0%	56.0%	-1.0%	1.6%	3.2%	0.262	4		Bush
ME	1,968	54.7%	53.6%	-1.2%	1.2%	2.3%	0.154	6		Bush
MI	2,452	52.5%	51.2%	-1.3%	1.0%	2.0%	0.112	9		Bush
MN	2,178	54.5%	51.2%	-3.3%	1.1%	2.1%	0.001	797	yes	Bush
MO	2,158	47.5%	46.1%	-1.4%	1.1%	2.2%	0.101	10		Bush

MS	798	43.3%	40.2%	-3.0%	1.8%	3.5%	0.046	22		Bush
MT	640	39.8%	38.6%	-1.2%	2.0%	4.0%	0.277	4		Bush
NC	2,167	48.0%	43.6%	-4.4%	1.1%	2.1%	0.000	35,233	yes	Bush
ND	649	34.0%	35.5%	1.5%	2.0%	3.9%	0.773	1		Kerry
NE	785	36.8%	32.7%	-4.0%	1.8%	3.6%	0.013	75	yes	Bush

NH	1,849	55.4%	50.4%	-5.0%	1.2%	2.3%	0.000	94,466	yes	Bush
NJ	1,520	55.0%	52.9%	-2.1%	1.3%	2.6%	0.059	17		Bush
NM	1,951	51.3%	49.0%	-2.3%	1.2%	2.3%	0.026	39		Bush
NV	2,116	49.4%	48.1%	-1.3%	1.1%	2.2%	0.128	8		Bush
NY	1,452	63.0%	58.4%	-4.6%	1.3%	2.6%	0.000	3,468	yes	Bush

OH	1,963	52.1%	48.7%	-3.4%	1.2%	2.3%	0.002	624	yes	Bush
OK	1,539	35.0%	34.4%	-0.6%	1.3%	2.5%	0.330	3		Bush
OR	1,064	51.2%	51.6%	0.4%	1.6%	3.1%	0.601	2		Kerry
PA	1,930	54.3%	51.0%	-3.4%	1.2%	2.3%	0.002	566	yes	Bush
RI	809	64.0%	59.6%	-4.4%	1.8%	3.5%	0.007	147	yes	Bush

SC	1,735	46.0%	41.0%	-5.0%	1.2%	2.4%	0.000	52,161	yes	Bush
SD	1,495	37.8%	38.4%	0.7%	1.3%	2.6%	0.697	1		Kerry
TN	1,774	41.5%	42.5%	1.0%	1.2%	2.4%	0.801	1		Kerry
TX	1,671	37.0%	38.2%	1.2%	1.2%	2.4%	0.837	1		Kerry
UT	798	30.5%	26.0%	-4.5%	1.8%	3.5%	0.007	149	yes	Bush

VA	1,431	48.0%	45.6%	-2.4%	1.3%	2.6%	0.038	26		Bush
VT	685	65.0%	59.2%	-5.8%	1.9%	3.8%	0.002	666	yes	Bush
WA	2,123	54.9%	52.9%	-2.1%	1.1%	2.2%	0.030	34		Bush
WI	2,223	52.5%	49.8%	-2.7%	1.1%	2.1%	0.006	175	yes	Bush
WV	1,722	45.3%	43.2%	-2.1%	1.2%	2.4%	0.047	21		Bush
WY	684	30.9%	29.1%	-1.8%	2.0%	3.8%	0.182	5		Bush
	73,607									
Avg	1,443	48.82%	46.58%	-2.24%	1.46%	2.85%	0.1094	9		Bush
Med	1,495	49.10%	47.08%	-2.09%	1.32%	2.59%	0.0474	21		Bush
Min	499	30.50%	26.03%	-5.78%	0.96%	1.87%	0.0000	94,466		
Max	2,846	91.00%	89.43%	1.62%	2.28%	4.48%	0.8372	1		
								

[Mistwell]

9. TIA

You have yet to post a real response to the pretty convincing evidence that unweighted exit polls in ALL prior years resulted in exactly this same kind of deviation from the actual results, and always favored Democrats over Republicans when compared to the actual results.

Your only response, that I have seen, implies the Republicans attempted to steal the popular vote in all of those years...even though Republicans lost the popular vote in all but this election, and even though electronic voting didn't even exist in some of those years.

Until you deal with that evidence, all you are showing us, over and over again, is that this unweighted exit poll had the same exact deviation it always had.

[rdmccur]

10. Show the data please

[Mistwell]

11. Here you go

http://www.washingtonmonthly.com/archives/individual/2004_11/005178.php

[rdmccur]

14. Two caveats

One is the quote from that website article
"Because while the 1988-2000 results above are completely raw and unweighted, we don't know for sure if the 2004 results that Freeman lists in his paper are also completely raw. They may already be partially weighted, in which case we'd expect them to be more accurate than the stuff from past years." My understanding is that the 2004 exit poll data Freeman used was already weighted. The other is, of course, that the previous results you refer to show no break down by state, particulary swing states where the exit polling is more accurate (due to the larger sample sizes).

[Mistwell]

15. Our discussion is unweighted

All the evidence of exit polls showing fraud in THIS election were unweighted. The weighted exit poll shows a Bush win. So weather or not a particular other study was of weighted data or not isn't relevant. When you test unweighted exit poll data, in all years including this one, it weighs heavily Democratic.

As for state by state, I don't see in any way why that is more accurate. Indeed, it might even be less, since it has a higher tendancy to be more urban-focused than the general poll (since some stats basically HAVE no urban center, giving you at least some rural samples, while other states do have an urban center, automatically weighing that stat Democratic).

[icehenge]

12. kick

[bemis12]

13. I see another pattern

Any shift toward Kerry of any size is a 1:1 probability. Seems right to me.