Election 2004 Turnout and National Exit Poll Sensitivity Model
This model calculates Kerry/Bush vote scenarios based on the
MAXIMUM POSSIBLE 2000 voter turnout in 2004. It also analyzes
the effects of INCREMENTAL PERCENTAGE DEVIATIONS from the
National Exit Poll of 13047 respondents. The results were
WEIGHTED AND POSTED on the NEP/WP web site at 12:22am (Kerry
won 51-48%), after the election.
It was a full 12 hours later (1:25pm), after the demographic
weights and vote percentages were changed to match the
recorded vote (Bush won 51-48%) and the final 660 respondents
were added to the total, that the official National Exit Poll
results (13660 respondents) were posted on CNN.
The sensitivity analysis clearly indicates that the Bush
victory was extremely implausible, even assuming MAXIMUM Bush
2000 voter turnout.
By applying actual 2000 votes/2004 votes as weights in the
Voted 2000 category, the Reluctant Bush Responder hypothesis
is effectively eliminated as a possible cause of exit poll
discrepancies from the recorded vote.
We have already shown that the published Final Exit Poll of
13660 respondents, which has Bush winning by 51-48%, is in
error. The reason is simple: the weightings are absolutely
impossible.
There is no way that 43% of the 122.26 million who voted in
2004 could have been Bush 2000 voters since the Bush 2000 vote
was 50.456 million and 50.456/122.26 = 41.27%. Furthermore,
approximately 1.77 mm of these Bush 2000 voters have passed
on, assuming the annual U.S. death rate of 0.87%. Therefore,
the maximum percentage of Bush 2000 voters who could have
voted in 2004 is 48.69/122.26 (39.82%).
The corresponding Gore percentage of 37% must therfore be
incorrect as well. The same calculation applied using the Gore
2000 vote of 50.999 million gives a maximum of 49.21 million
(40.25%). For Nader, the maximum is 3.206 million (2.62%).
The following is clear: The recorded 2004 vote is the sum of
returning Bush, Gore and Nader voters plus new voters (those
who did not vote in 2000):
V = BV +KV + NV + New
This is a very powerful equation, due to its simplicity and
application. Unlike questions of exit poll accuracy, margin of
error, reluctant Bush responders, etc., there can be no
argument about this formula. It’s the equivalent of the
accounting identity:
Assets = Liabilities + Equity.
Since we already know the total vote V (122.26 million) and
the maximum number of Bush, Gore and Nader 2000 voters, we can
proceed to calulate the MINIMUM number of New voters (those
who never voted as well as those who did not vote in 2000 but
voted prior to that).
V = 122.26 = 48.69 + 49.21 + 3.21 + New
and
New = 122.26 - 48.69 – 49.21 - 3.21 = 21.15 million.
This is the MINIMUM number of New 2004 voters, as it assumes
100% voter turnout for Bush, Kerry and Nader. The number of
New voters must have been greater than 21.15 million.
Given this information, we can “stress test” the results of
the National Exit Poll of 13047 respondents to determine how
Bush could win by 3 million votes.
The base case for the analysis assumes 100% turnout of Bush,
Gore and Nader voters.
*****************************************************
National Exit Poll:
13047 respondents @ 12:22am
Using the actual weightings and the exit poll percentages,
Kerry wins by 5.91 million votes (51.94%-47.10%).
Gore turnout: 100.0%
Bush turnout: 100.0%
Total Voted 2000
Mix Kerry Bush Nader
21.150 No 17.30% 57.00% 41.00% 2.00%
Consisting of:
13.448 New 11.00% 56.00% 43.00% 1.00%
7.702 Old 6.30% 58.75% 37.51% 3.75%
49.214 Gore 40.25% 91.00% 8.00% 1.00%
48.690 Bush 39.82% 9.00% 91.00% 0.00%
3.206 Nader 2.62% 71.00% 21.00% 8.00%
122.26 Total 100.0% 51.94% 47.10% 0.96%
122.26 63.50 57.59 1.17
*****************************************************
The following sensitivity matrix gives the Kerry margin for
alternative Gore Voter turnout scenarios (assuming 100% Bush
2000 voter turnout) and deviations from the base case exit
poll voting percentages. Following the table are examples of a
few of the scenario calculations.
Kerry Margin Sensitivity to Gore Voter Turnout
and Deviations from Exit Poll Percentages
(assume 100% Bush 2000 voter turnout)
New Gore Kerry Percentage Vote Deviation from Exit Poll
Turnout 0% -1% -2% -3%
21.151 100% 5.91 3.46 1.02 -1.43
22.134 98% 5.25 2.8 0.36 -2.09
23.119 96% 4.59 2.14 -0.30 -2.75
24.103 94% 3.93 1.49 -0.96 -3.41
25.087 92% 3.27 0.83 -1.62
26.072 90% 2.61 0.17 -2.28
27.056 88% 1.95 -0.49 -2.94
28.040 86% 1.29 -1.15
29.024 84% 0.63 -1.81
30.009 82% -0.03 -2.47
30.993 80% -0.69 -3.13
33.946 73% -2.99
Looking at the table, the following is clear: For Bush to
realize a 3 million vote margin, one of the following had to
occur:
1) Assuming no change to the base case percentages, there
would have had to have been a 73% turnout of Gore 2000 voters
and a 100% Bush turnout. This is clearly impossible.
2) Assuming a 1% decline from the exit poll result in each of
the Kerry subgroups, an impossible 80% Gore turnout was
necessary for a Bush 3.13 million vote margin.
3) Assuming a 2% decline, the Gore turnout must have been 88%.
Again impossible.
4) Assuming a 3% decline, the turnout must have been 94%. This
may seem possible until one looks at the number of New voters
necessary to make up the difference - 24.103 million.
Because of the necessary constraints (122.26 million total
2004 vote, maximum number of 2000 voters alive who could vote
in 2004) the difference had to made up by New voters - 3
million more than the 21.15 million, as confirmed by the (Did
Not Vote) 17% weighting. And more importantly, by the voting
identity equation. Clearly impossible.
*****************************************************
Kerry margin: 3.46 million
Kerry Deviation from Exit Poll:-1.0%
Gore turnout:100.0%
Bush turnout:100.0%
Voted 2000 Mix Kerry Bush Nader
21.150 No 17.30% 56.00% 42.00% 2.00%
Consisting of:
13.448 New 11.00% 55.00% 44.00% 1.00%
7.702 Old 6.30% 57.75% 38.51% 3.75%
49.214 Gore 40.25% 90.00% 9.00% 1.00%
48.690 Bush 39.82% 8.00% 92.00% 0.00%
3.206 Nader 2.62% 70.00% 22.00% 8.00%
122.26 Total 100.0% 50.94% 48.10% 0.96%
122.26 62.28 58.81 1.17
**************************************************
Kerry Margin: 1.49 million
Kerry Deviation from Exit Poll:-1.0%
Gore turnout: 94.0%
Bush turnout:100.0%
Voted 2000 Mix Kerry Bush Nader
24.103 No 19.71% 56.00% 42.00% 2.00%
Consisting of:
15.326 New 12.54% 55.00% 44.00% 1.00%
8.777 Old 7.18% 57.75% 38.51% 3.75%
46.261 Gore 37.84% 90.00% 9.00% 1.00%
48.690 Bush 39.82% 8.00% 92.00% 0.00%
3.206 Nader 2.62% 70.00% 22.00% 8.00%
122.26 Total 100.0% 50.12% 48.90% 0.98%
122.26 61.27 59.79 1.20
***************************************************
Kerry Margin: 0.36 million
Kerry Deviation from Exit Poll: -2.0%
Gore turnout: 98.0%
Bush turnout: 100.0%
Voted 2000 Mix Kerry Bush Nader
22.134 No 18.10% 55.00% 43.00% 2.00%
Consisting of:
14.074 New 11.51% 54.00% 45.00% 1.00%
8.061 Old 6.59% 56.75% 39.51% 3.75%
48.230 Gore 39.45% 89.00% 10.00% 1.00%
48.690 Bush 39.82% 7.00% 93.00% 0.00%
3.206 Nader 2.62% 69.00% 23.00% 8.00%
122.26 Total 100.0% 49.66% 49.37% 0.97%
122.26 60.72 60.36 1.18