This simulation model analyzes exit poll response rate
scenarios. I wrote it to better understand the USCV model and
used the same input parameters: 1250 precincts, from strong
Bush to strong Kerry, as well as a range of response rate
combinations for Bush and Kerry strongholds.
The model simulates 10 response rate combinations (53%
weighted average). At 10,000 trials per simulation, a total of
100,000 trials were run.
The results confirm what Ron Baiman at USCV has said:
THERE IS NO PLAUSIBLE SCENARIO WHICH CAN REPRODUCE THE BUSH
50.73%-48.28% WIN. ON THE CONTRARY, ALL REALISTIC SCENARIOS
USING E-M EXIT POLL RESPONSE DATA SHOW THAT KERRY WON THE
ELECTION EASILY.
Strong Bush Strong Kerry
1250 Precincts 40 415 540 165 90
Kerry% 10 30 50 70 90
StdDev 0.15 0.05 0.05 0.1 0.15
Kerry Kerry
Sim Median Mean Response Rate WtdAvg
1 53.30% 53.31% 51% 52% 53% 53% 54% 52.68%
2 53.95% 53.97% 52% 52% 53% 54% 55% 52.91%
3 53.72% 53.72% 52% 52% 53% 54% 54% 52.84%
4 53.50% 53.50% 53% 53% 53% 53% 53% 53.00%
5 52.93% 52.92% 51% 53% 53% 53% 51% 52.79%
6 53.62% 53.60% 50% 54% 53% 53% 52% 53.16%
7 53.79% 53.76% 54% 53% 53% 53% 54% 53.10%
8 54.15% 54.15% 52% 53% 54% 53% 52% 53.33%
9 53.87% 53.88% 52% 53% 53% 54% 53% 53.10%
10 53.85% 53.86% 57% 52% 53% 52% 57% 52.95%
Avg 53.67% 53.67% 52.40% 52.70% 53.10% 53.20% 53.50% 52.99%
**********************************************
This is an example of the kind of implausible scenario Bush
needs to to win 50.73%. He needs extremely high response rates
in his strongholds AND weak response rates in the Kerry
precincts.
Even then, the following scenario only generates a 52.43%
total response rate, so Bush needs higher response rates in
his strongolds - but that would cause his 50.73% win
constraint to be exceeded. The vote constraint AND the 53%
average response rate constraint make it virtually impossible.
Bush 50.73% 50.73% 55% 53% 55% 46% 45% 52.43%
**********************************************
Here are the first 30 of 10,000 trials for simulation 5:
Trial Kerry p40 p415 p540 p165 p90
1 51.97% 0.12 5.41 13.35 4.30 2.81
2 52.73% 0.18 5.04 12.62 5.27 3.26
3 56.70% 0.12 6.16 12.13 5.64 4.31
4 53.21% 0.17 5.22 12.62 5.71 2.88
5 56.25% 0.22 5.41 12.86 6.08 3.56
6 48.82% 0.13 5.60 11.65 4.67 2.36
7 49.07% 0.17 5.60 10.92 5.12 2.73
8 52.83% 0.20 5.60 12.13 5.49 3.00
9 49.89% 0.17 5.41 10.92 5.27 3.18
10 56.83% 0.15 6.16 12.62 6.23 3.26
11 48.62% 0.18 5.04 11.16 4.89 3.03
12 51.04% 0.17 6.16 11.16 5.64 2.40
13 51.09% 0.15 6.16 11.16 5.27 2.81
14 49.70% 0.20 5.41 12.13 4.45 2.66
15 52.04% 0.18 5.04 12.13 5.86 2.81
16 50.22% 0.18 5.41 11.41 5.34 2.77
17 54.79% 0.17 5.78 13.11 4.67 3.67
18 49.71% 0.12 5.22 11.41 5.71 2.40
19 52.03% 0.20 5.04 13.11 4.38 3.30
20 53.89% 0.20 5.60 13.11 4.38 3.67
21 58.63% 0.17 5.97 13.35 5.56 4.27
22 50.23% 0.18 5.41 12.38 4.45 2.70
23 49.57% 0.17 5.41 11.16 4.45 3.60
24 51.65% 0.13 5.97 11.65 5.27 2.81
25 53.85% 0.17 5.22 12.86 5.19 3.48
26 52.77% 0.17 5.97 11.16 5.19 3.90
27 55.61% 0.13 5.04 12.62 5.71 4.31
28 56.01% 0.20 5.22 12.86 5.56 4.16
29 51.74% 0.20 5.78 10.92 6.01 2.96
30 57.74% 0.12 6.16 13.11 5.64 3.86