The National Exit Poll must match the official vote count — come hell or high water
TruthIsAll source: http://www.geocities.com/electionmodel/2008NEPSolver.htmJan. 3, 2009Assume you had the full set of the Preliminary National Exit Poll results that showed Obama winning with a 57.8% vote share. As you suspected, more Kerry voters returned than did Bush voters. The poll results differ sharply from the official vote count that has Obama winning 52.34%. But the National Exit Poll Final results must match the official vote count — come hell or high water. It’s standard operating procedure: pollsters alter exit poll data to be the same as the election results. How would you go about it?Preliminary Unadjusted 2008 National Exit Poll (assumed values)There are two possibilities: you can adjust the Mix of returning '04-voters and/or the individual candidate Vote Shares (4195 Respondents, 1.5% MoE).Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
15.8%
19.2
72.0%
26.0%
2%
13.79
4.98
0.38
Kerry
42.9%
52.0
89.8%
8.2%
2%
46.66
4.29
1.04
Bush
40.4%
49.0
18.2%
80.8%
1%
8.93
39.61
0.49
Other
0.85%
1.0
66.0%
24.0%
10%
0.68
0.25
0.10
121.21
70.06
49.12
2.02
100%
57.80%
40.53%
1.66%
You load the Excel Solver add-in program. It will determine the returning voter mix that will force the cross tab to match the recorded vote share — or at least get very close to it. It will find the required mix using a sophisticated
goal-seeking algorithm. It’s just trial and error on steroids.
Solver needs the ranges for the return voter percentage mix and the corresponding vote shares to adjust, until it matches the target (recorded) Obama vote share. It also needs to know the numerical
constraints; the mix and share percentages must all be greater than zero and sum to 100%.
Now that Solver knows the target vote share and the range of percentages to adjust subject to the constraints, it can begin the search for the required mix and vote shares. Within seconds, it indicates that it has found the values needed to match the recorded 52.34% share.
2008 National Exit Poll ('forcing' a match to the Recorded Vote)1. Target the Obama vote share at 52.34%Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
15.8%
19.2
71.0%
27.00%
2%
13.61
5.17
0.38
Kerry
38.5%
46.6
87.1%
10.90%
2%
40.62
5.08
0.93
Bush
44.9%
54.4
15.7%
83.30%
1%
8.54
45.30
0.54
Other
0.85%
1.0
66.0%
24.00%
10%
0.68
0.25
0.10
121.21
63.44
55.79
1.96
100%
52.34%
46.03%
1.62%
Solver has determined that to match the recorded vote, the returning voter mix from the initial Preliminary exit poll must be radically adjusted: an increase of 5.4 million more returning-Bush voters and corresponding decrease in Kerry voters. Solver also reduced the Obama share of new voters (1%), returning Kerry voters (2.8%) and returning Bush voters (2.5%).
But since the exit poll margin of error is 1.5%, the Obama shares of Kerry and Bush voters must be re-adjusted to fall within the MoE based on the original poll results.2. Adjust the Obama share of Kerry voters to 89% and Bush voters to 17%Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
15.8%
19.2
71%
27%
2%
13.61
5.17
0.38
Kerry
36.7%
44.4
89%
9%
2%
39.54
4.00
0.89
Bush
46.7%
56.6
17%
82%
1%
9.62
46.40
0.57
Other
0.85%
1.0
66%
24%
10%
0.68
0.25
0.10
121.21
63.44
55.82
1.94
100%
52.34%
46.05%
1.60%
But the required spread in the mix between returning Kerry and Bush voters has widened from 6.4% to 10.0% (7.8m to 12.2m votes).
Decrease the percentage of new voters to 13.0%. Increase the percentage of returning third-party voters to 4.0%. Adjust the returning-Bush count accordingly.3. Decrease the new vote mix to 13% and increase the mix of returning third-party voters to 4%Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
13.00%
15.8
71%
27%
2%
11.19
4.25
0.32
Kerry
36.62%
44.4
89%
9%
2%
39.50
3.99
0.89
Bush
46.38%
56.2
17%
82%
1%
9.56
46.10
0.56
Other
4.00%
4.8
66%
24%
10%
3.20
1.16
0.48
121.21
63.44
55.51
2.25
100%
52.34%
45.79%
1.86%
Done. But…10.16 million late votes have been recorded since Election Day; the vote count is 131.37 million.
Obama has won 59% of the late votes, increasing his vote share to 52.87%. Recalculate the mix.4. The updated Final NEP Mix and Vote Shares (12/25/08):Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
13.00%
17.1
71%
27%
2%
12.13
4.61
0.34
Kerry
37.34%
49.1
89%
9%
2%
43.66
4.42
0.98
Bush
45.66%
60.0
17%
82%
1%
10.20
49.19
0.60
Other
4.00%
5.3
66%
24%
10%
3.47
1.26
0.53
131.37
69.46
59.47
2.45
100%
52.87%
45.27%
1.86%
But the Final NEP is not feasible. The returning voter mix required to match the recorded vote is mathematically impossible.5. The following result is feasible. But the returning voter mix is based on an implausible recorded 2004 vote:Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
15.85%
20.8
71%
27%
2%
14.79
5.62
0.42
Kerry
40.40%
53.1
89%
9%
2%
47.24
4.78
1.06
Bush
42.90%
56.4
17%
82%
1%
9.58
46.21
0.56
Other
0.85%
1.1
66%
24%
10%
0.74
0.27
0.11
131.37
72.34
56.88
2.15
100%
55.07%
43.30%
1.64%
6. This result is more likely. The returning voter mix is based on the unadjusted 2004 State Exit Poll Aggregate (WPE):Voted
in 2004
2008
Mix
Voter
Count
Obama
Share
McCain
Share
Other
Share
Obama
Votes
McCain
Votes
Other
Votes
DNV (New in '08)
15.85%
20.8
71%
27%
2%
14.79
5.62
0.42
Kerry
44.00%
57.8
89%
9%
2%
51.44
5.20
1.16
Bush
39.30%
51.6
17%
82%
1%
8.78
42.34
0.52
Other
0.85%
1.1
66%
24%
10%
0.74
0.27
0.11
131.37
75.75
53.42
2.20
100%
57.66%
40.67%
1.68%
The Final NEP indicates that 2004 third-party voters comprised 4% (5.3m) of the 131.37 million who voted in 2008. But there were only 1.2m third-party voters in 2004.
• Bush won by an official 62–59 million votes — a 3 million margin in 2004. But returning Bush voters comprised 46% (60.0m) and Kerry voters just 37% (49.1m) — a 10.9 million spread.
• According to voter mortality tables, approximately 3 million Bush voters died prior to the 2008 election. But even assuming 100% Bush voter turnout in 2008, at most 59 million could have returned to vote. Realistically, approximately 95% (56m) voted. Therefore, the Final NEP overstated the number of returning Bush voters by 4 million. And that’s a conservative number, since it assumes that the recorded vote was equal to the True vote.
According to the unadjusted 2004 State Exit Poll Aggregate, Kerry won by 52–47% (63.6–57.5m). If the aggregate reflected the True vote, then the 2008 NEP overstated the number of returning Bush voters by 9 million.
The Final National Exit Poll anomalies should have been reported by the media. But the NY Times, CNN, ABC, the Washington Post commissioned the NEP. The impossible results will remain official. Otherwise, the public would know that Obama really won by more than 20 million votes. That’s too big a mandate; he would be compelled to enact a progressive agenda. More important, the Corporate Media does not want the public to know that Bushco stole both elections, and that the same Final NEP anomalies existed in 2004.