Exit Poll Response Optimization: Confirms the USCV Simulation in
Debunking the rBr Hypothesis
Richard Charnin (TruthIsAll)
Updated:
Feb. 10, 2012
Further Confirmation of a Kerry Landslide
In January, 2005, exit pollsters Edison-Mitofsky hypothesized that the 6.5% average exit poll discrepancy was due to differential exit poll response rates of Kerry/Bush voters (56/50). It was dubbed “reluctant Bush responder” (rBr) hypothesis. But the exit pollsters did not provide any evidence to back up the theory. In fact, independent simulation, optimization and regression analysis indicated that Bush voters were in fact slightly more eager to be interviewed – meaning that Kerry actually did better than the exit polls indicated.
In Sept. 2005, US Count Votes provided a simulation analysis which effectively debunked rBr. USCV noted “However, our analysis below shows that even if the “E/M hypothetical,” is interpreted as referring to average, rather than constant average, partisan exit poll response rates, it is inconsistent with the reported WPE data. There is no configuration of partisan response rates, however varying across precinct partisanship categories, with overall averages of K=0.56 and B=0.5 that can produce the reported values of the actual E/M exit poll data for mean WPE and median WPE”. See below.
The following set of three exit poll response optimization models confirms the USCV simulation.
1. Precinct Partisanship Exit Poll Response
Optimizer
The
pollsters summarized the exit poll results for 1250 precincts in five partisanship groupings. The
groups ranged from strong Bush to strong Kerry. The optimization model used the
Excel “Solver” algorithm to determine a feasible solution: Kerry won the
2-party vote by 52.15-47.85%. The model used the final recorded 2-party vote,
within precinct error (WPE) and response rates as constraints. The resulting
vote shares exactly matched
the 12:22am National Exit Poll “Voted 2000” shares.
These
graphs display various components of precinct exit poll data by partisanship:
Effect
of Incremental Aggregate Alpha on Precinct Category Alpha
Alpha
(Kerry/Bush response) by Precinct Category
Exit
Poll Response by Precinct Category
Sensitivity
of Kerry Vote Pct and WPE to Aggregate Alpha (K/B)
Bush
Percentage of Refusers by Precinct Category Required
for Recorded Vote
2. Location Size Exit Poll Response Optimizer
The exit
polls were categorized into five groupings by size. The input constraints were
location WPE and response (completion) rates provided by Edison-Mitofsky.
Kerry won
the 2-party vote by 52.15-47.85%, exactly the same margin as the 1250 precinct
optimization.
3. State Exit Poll Response Optimizer
The state
exit polls were categorized into five partisanship groups using state exit poll
response rates provided by Edison-Mitofsky. The state
discrepancies were based on average precinct WPE. Kerry won the 2-party
vote by 52.3-47.7%. The model confirmed
the Precinct Response Optimizer: non-response rates were higher in strong Kerry
states, again contradicting the non-response (rBr) hypothesis. A regression
analysis of Kerry state
poll share and recorded vote vs. the state exit poll completion rate confirmed
the optimizer.
1250 Precincts Categorized by Partisanship Groupings
Given:
1-
Recorded 2-party vote (Bush 51.24- Kerry 48.76%)
2-
Partisanship precinct exit poll
response
3-
Partisanship "Within Precinct Error"
(WPE)
Calculate:
Kerry’s
true vote - aggregate and by partisanship category
Kerry
52.15-Bush 47.85%
Assume
Census 125.74m total votes
cast
2-party
Vote% Vote
Kerry 48.76% 60.70
Bush
51.24% 63.78
Total 100%
124.48
Optimizer results:
(2-party)
Kerry Bush
Vote
64.91 59.57m
2-party
52.15% 47.85%
Deviation
-3.39% > 3.39%
Vote Deviation
-5.89 2.47
Dev /2-pty
-6.50% > 7.09%
Vote share
51.62% 47.37%
PARTISAN
ALPHA
Kerry strongholds:
1.052
Other precincts:
1.203
WEIGHTED AVERAGE
(AGGREGATE)
Response:
53.59%
K/B (alpha):
1.165
SHARE OF EXIT POLL REFUSERS
Kerry
44.87%
Bush
55.13%
PARTISANSHIP CONSTRAINTS (1250
PRECINCTS)
HiBush
Bush Even Kerry HiKerry
Precincts
40 415 540
165 90
KERRY
SHARE
Min
0% 20% 40%
60% 80%
Max
20% 40% 60%
80% 100%
RESPONSE
Min
56% 55% 52%
55% 53%
Max
56% 55% 52%
55% 53%
ALPHA
(K/B)
Min
0.5 0.5 0.5 0.5
0.5
Max
2.0 2.0 2.0 2.0
2.0
WPE
Min 10.0% 6.1% 8.5% 5.9% 0.3%
Max 10.0% 6.1% 8.5% 5.9% 0.3%
Actual
E-M 10.0% 6.1% 8.5% 5.9%
0.3%
OPTIMIZER OUTPUT
SUMMARY
Poll Count
Diff Poll
Count Diff
Kerry
52.15% 48.76% -3.39%
64.91 60.70 -4.22
Bush
47.85% 51.24% 3.39%
59.57 63.79 4.22
Diff
4.29% -2.48% -6.77% 5.346
-3.09 -8.43
Pship
HiBush Bush Even
Kerry HiKerry Total
Prcts
40 415
540 165
90 1250
Votes
3.98 41.33 53.78
16.43 8.96 124.48
Pct
3.2% 33.2% 43.2%
13.2% 7.2% 100.0%
Resp 56.0%
55.0% 52.0% 55.0% 53.0% 53.59%
ALPHA
K/B
1.50 1.17
1.19 1.08 1.0 1.17
K/ 50B
75.1 58.7
59.3 54.1
49.8 58.2
VOTE
Kerry
0.79 14.51 24.58
11.87 8.95 60.70
Pct
19.9% 35.1% 45.7%
72.2% 99.8% 48.76%
Bush
3.19 26.82 29.20
4.56 0.02 63.79
Pct
80.1% 64.9% 54.3%
27.8% 0.2% 51.24%
RESPONDERS
Kerry
0.99 15.77 26.86
12.35 8.93 64.91
Pct
24.9% 38.2% 50.0%
75.2% 99.7% 52.15%
Bush
2.99 25.55 26.91
4.08 0.03 59.57
Pct
75.1% 61.8% 50.0%
24.8% 0.3% 47.85%
REFUSERS
Kerry
13.5% 31.4% 41.1%
68.6% 100% 44.87%
Bush
86.5% 68.6% 58.9%
31.4% 0.0% 55.13%
VOTE
DEVIATION
Kerry 0.20 1.26
2.29 0.48 0.01 4.22
Pct
20.1% 8.0%
8.5%
3.9% 0.20% 6.50%
WPE
Calc
10.0% 6.1% 8.5% 5.9% -0.30% 6.77%
E-M
10.0% 6.1% 8.5% 5.9% -0.30% 6.77%
_____________________________________________________________________
Location-size Exit Poll Optimization
Given:
1-
Recorded vote: Bush 50.73- Kerry
48.27%
(2
party: Bush 51.24%- Kerry
48.76%)
2-
Location exit poll response rates
3-
Location "Within Precinct Error"
(WPE)
Calculate:
True
2-party Vote share (aggregate and by location)
Kerry
52.15-Bush 47.85%
Assume
Census 125.74m total votes
cast
WPE WPE
Size Weight
Votes Mean Median Prct Response
Big City 13% 16.35
7.9 5.9 105 0.52
Suburbs 45% 56.58
8.1 7.9 487 0.53
Rural 16%
20.12
3.6 3.6 296 0.55
Total
125.74 7.17 6.68 1250 0.54
EXIT POLL (2-party)
Kerry
Bush
Vote
64.91
59.58
Vote share 52.15%
47.85%
Deviation -3.39%
3.39%
Vote Dev
-5.88
2.47
Dev/2-pty
-6.50%
7.08%
True Vote
51.62% 47.37%
Deviation 9.06%
-4.15%
WEIGHTED AVERAGE (AGGREGATE) RESPONSE:
53.98%
PERCENT OF EXIT POLL REFUSERS REQUIRED TO MATCH RECORDED
VOTE
Kerry
44.83%
Bush
55.17%
Rural Town Suburb
City Big City
Prcts
296 126
487 236
105
23.7% 10.1% 39.0%
18.9% 8.4%
Votes 20.12
10.06 56.58 22.63
16.35
16% 8%
45% 18%
13%
Kerry NEP share
12:22am 43%
52% 50%
53% 64%
Final 40%
48% 47%
49% 61%
TRUE 40%
50% 49%
59% 73%
RANGE
CONSTRAINTS
KERRY WIN%
Min
35% 45%
45% 45%
60%
Max
55% 55%
55% 60%
100%
RESPONSE
Min
55% 57%
53% 54%
52%
Max
55% 57%
53% 54%
52%
ALPHA
(K/B)
Min 0.50
0.50 0.50
0.50 0.50
Max 3.00
3.00 3.00
3.00 3.00
WPE
Min 3.6% 4.9%
8.1% 8.5%
7.9%
Max 3.6% 4.9% 8.1% 8.5%
7.9%
E-M 3.6% 4.9% 8.1%
8.5% 7.9%
OPTIMIZER OUTPUT
SUMMARY
...... Poll Count
Diff Poll
Count Diff
Kerry 52.15% 48.76%
-3.39% 64.91 60.70 -4.21
Bush 47.85% 51.24%
3.39% 59.57 63.78 4.21
Diff 4.29% -2.48%
-6.77% 5.34 p; -3.09 -8.43
..... Rural
Town Suburb City
Big City Total
Prcts
296 126
487 236
105 1250
2-pty 29.48
12.55 48.50 23.50 10.46
124.48
Pct 23.7%
10.1% 39.0% 18.9%
8.4% 100.0%
Response 55.0% 57.0%
53.0% 54.0% 52.0% 53.98%
ALPHA
K/B 0.792
0.800 0.840 0.835
0.800 0.82
K/ 50B 39.6
40.0 42.0
41.8 40.0 41.0
VOTE
Kerry 12.69
6.10 21.82 11.75
8.34 60.69
Pct 43.0%
48.6% 45.0% 50.0%
79.7% 48.76%
Bush 16.79
6.45 26.67 11.75
2.12 63.78
Pct 57.0% 51.4%
55.0% 50.0% 20.3% 51.24%
RESPONDERS
Kerry 13.22
6.41 23.79 12.75
8.75 64.91
Pct 44.8%
51.1% 49.1% 54.3%
83.7% 52.15%
Bush 16.26
6.14 24.71 10.75
1.71 59.57
Pct 55.2%
48.9% 51.0% 45.8%
16.3% 47.85%
REFUSERS
Kerry 40.8%
45.4% 40.4% 45.0%
75.4% 44.83%
Bush 59.2%
54.6% 59.6% 55.0%
24.6% 55.17%
VOTE
DEVIATION
Kerry 0.53 0.31 1.96 1.00 0.41 4.21
Pct 4.0% 4.8% 8.3% 7.8% 4.7% 6.49%
WPE
Calc 3.6% 4.9%
8.1% 8.5% 7.9% 6.77%
E-M 3.6% 4.9%
8.1% 8.5% 7.9% 6.77%
Diff 0.0% 0.0%
0.0% 0.0% 0.0% 0.0%
_____________________________________________________________________
State Exit Poll Response Optimizer
Given: Recorded vote
2-Party Total
Kerry 48.76%
48.27%
Bush
51.24% 50.72%
2- State exit poll average response
rate
3- State exit poll average
WPE
Calculate: TRUE VOTE
2-Party
Total
Kerry 52.30%
51.77%
Bush 47.70%
47.21%
WPE 6.77%
7.08%
WEIGHTED AVERAGE (AGGREGATE)
Response
53.32%
REFUSER VOTE SHARE REQUIRED TO MATCH
RECORDED VOTE
Kerry 44.86%
Bush
55.14%
CALCULATED TRUE
VOTE Deviation from Recorded
State
Votes Weight RESP.
Kerry Pct
Bush Pct
Votes Pct WPE
Total 121,056
100% 53.32% 63,314
52.30% 57,741 47.70% 4287
6.77% 7.08%
HIGH
BUSH
UT
905 0.75%
59.6% 270
29.9% 635
70.1% 29 10.7% 6.4%
WY
238 0.20%
66.0% 76
31.8% 163
68.2% 5
6.8% 4.3%
ID
590 0.49%
63.2% 184
31.2% 406
68.8% 3
1.6% 1.0%
NE
767 0.63% 66.5%
285 37.2%
482 62.8%
31 10.9% 8.1%
OK
1,464 1.21%
53.2% 490
33.5% 974
66.5% 14 2.8% 1.9%
BUSH
ND
308 0.25%
63.0% 103
33.5% 205
66.5% 8 7.8% 5.2%
AK
302 0.25%
53.2% 126
41.6% 176
58.4% 14 11.5% 9.6%
KS
1,171 0.97%
64.5% 445
38.0% 726
62.0% 10 2.2%
1.7%
TX
7,360 6.08%
58.3% 3,009 40.9%
4,350 59.1% 177
5.9% 4.8%
SD
382 0.32%
42.7% 141
37.0% 241
63.0% 8 5.7% 4.2%
MT
440 0.36%
63.0% 170
38.6% 270
61.4%
4 2.3% 1.8%
IN
2,448 2.02%
38.6% 987
40.3% 1,461 59.7%
18 1.9% 1.5%
KY
1,782 1.47%
52.6% 712
39.9% 1,070 60.1%
1 0.1% 0.1%
MS
1,130 0.93%
49.6% 522
46.1% 609
53.9% 64 12.2% 11.3%
SC
1,600 1.32%
59.4% 742 46.4%
858 53.6%
80 10.8% 10.0%
GA
3,280 2.71%
63.9% 1,402 42.7%
1,878 57.3%
36 2.6% 2.2%
LA
1,922 1.59%
47.8% 857
44.6% 1,066 55.4%
37 4.3% 3.8%
TN
2,421 2.00%
66.7% 1,043 43.1%
1,378 56.9%
6 0.6% 0.5%
WV
750 0.62%
48.7% 305
40.6% 446
59.4% 22 7.1% 5.8%
NC
3,487 2.88%
52.6% 1,723 49.4%
1,764 50.6% 197
11.4% 11.3%
AZ
1,998 1.65%
57.3% 939
47.0% 1,058 53.0%
46 4.9% 4.6%
AR
1,043 0.86%
60.2% 473
45.3% 571
54.7% 3
0.6% 0.5%
VA
3,172 2.62%
56.4% 1,580 49.8%
1,592 50.2% 125
7.9% 7.9%
MO
2,715 2.24% 47.0%
1,338 49.3% 1,377
50.7% 79 5.9%
5.8%
EVEN
FL
7,548 6.24%
49.0% 3,870 51.3%
3,678 48.7% 287
7.4% 7.6%
CO 2,103
1.74% 55.5%
1,066 50.7% 1,037
49.3% 64 6.0%
6.1%
NV
816 0.67%
49.1% 438
53.7% 377
46.3% 41 9.4%
10.1%
OH
5,599 4.62%
45.0% 3,045 54.4%
2,554 45.6% 305
10.0% 10.9%
NM
748 0.62%
56.9% 400
53.5% 348
46.5% 29 7.3%
7.8%
IA
1,494 1.23%
52.6% 764
51.2% 730
48.8% 22 2.9%
3.0%
WI
2,968 2.45%
55.3% 1,559 52.5%
1,408 47.5%
70 4.5% 4.7%
NH
672 0.55%
44.0% 386
57.5% 286
42.5% 46 11.8%
13.6%
PA
5,732 4.73%
46.8% 3,190 55.7%
2,542 44.3% 252
7.9% 8.8%
MI
4,793 3.96%
50.2% 2,630 54.9%
2,163 45.1% 151
5.7% 6.3%
MN
2,792 2.31%
45.3% 1,575 56.4%
1,217 43.6% 130
8.2% 9.3%
OR
1,810 1.50%
53.0% 943
52.1% 867
47.9% 0 0.0% 0.0%
KERRY
NJ
3,581 2.96%
59.7% 2,085 58.2%
1,496 41.8% 174
8.3% 9.7%
WA
2,815 2.33% 53.8%
1,628 57.8% 1,187
42.2% 118 7.3% 8.4%
DE
372 0.31%
57.5% 230
61.8% 142
38.2% 30 12.9%
15.9%
HI
426 0.35%
53.4% 242
56.8% 184
43.2% 10 4.1%
4.7%
ME
727 0.60%
61.3% 411
56.5% 316
43.5% 14 3.4%
3.8%
CA
12,255 10.12% 50.5% 7,413
60.5% 4,842 39.5%
668 9.0% 10.9%
IL
5,239 4.33%
51.9% 3,007 57.4%
2,231 42.6% 115
3.8% 4.4%
CT
1,551 1.28%
51.0% 979
63.1% 572
36.9% 122 12.4% 15.7%
MD
2,359 1.95% 59.4%
1,430 60.6%
929 39.4%
96 6.7% 8.1%
HIGH
KERRY
NY
7,277 6.01%
57.9% 4,729 65.0%
2,548 35.0% 415
8.8% 11.4%
VT
305 0.25%
53.1% 207
67.8% 98
32.2% 23 11.1%
15.0%
RI
429 0.35%
44.2% 270
62.9% 159 37.1%
10 3.7% 4.7%
MA
2,875
2.37% 56.5%
1,887 65.6%
988 34.4% 83
4.4% 5.8%
DC
224
0.19% 53.5%
207 92.2%
17 7.8%
4 1.8% 3.4%
CALCULATED TRUE
VOTE
Deviation from
Recorded
State
Votes Weight RESP.
Kerry Pct
Bush Pct
Votes Pct WPE
Total
121,056 100% 53.32%
63,314 52.30% 57,741 47.70%
4287 6.77% 7.08%
HBUSH
3,965
3.3% 61.7%
1,306 32.7% 2,659
67.3% 54 5.4%
3.58%
BUSH
39,582 32.7% 55.2%
17,415 42.8% 22,167
57.2% 955 3.6%
3.49%
EVEN
37,073 30.6% 50.2%
19,868 53.7% 17,205
46.3% 1,397 6.8% 7.35%
KERRY
29,326 24.2% 55.4%
17,426 59.2% 11,900
40.8% 1,346 7.5% 9.07%
HKERRY 11,110
9.2% 53.0%
7,300 70.7% 3,810
29.3% 535 6.0%
8.06%
From: US Count Votes:
The
exit polling firm Edison/Mitofsky (E/M), and USCV,
agree that the historically unprecedented discrepancy between the exit polls
and the reported vote count for the 2004
The E/M hypothetical was widely interpreted by
the media and by USCV as a claim that the 2004 exit poll discrepancy was caused
by a pervasive, and on average uniform, shortfall in Bush voter exit poll
response relative to Kerry voter exit poll response that was dubbed the
“reluctant Bush response” (rBr) hypothesis. A recent
clarification by E/M indicates that the “E/M hypothetical” should be
interpreted as referring to hypothetical average (rather than constant average)
partisan exit poll response rates. In this interpretation, average precinct
partisan response rates may vary widely by reported precinct vote shares, yet all of the reported WPE could be
explained by partisan response rates whose average across the sample is K=0.56
and B=.50.
This
interpretation of the “E/M hypothetical” does not depend on the “rBr hypothesis” of constant average partisan response rates
which was shown by USCV to be inconsistent with the pattern of the exit poll
discrepancy. However, our analysis below shows that even if the “E/M
hypothetical,” is interpreted as referring to average, rather than constant
average, partisan exit poll response rates, it is inconsistent with the
reported WPE data. There is no configuration of partisan response rates,
however varying across precinct partisanship categories, with overall averages
of K=0.56 and B=0.5 that can produce the
reported values of the actual E/M exit poll data for mean WPE and median WPE.
Thus,
neither a "reluctant Bush responder" (rBr)
hypothesis interpretation nor an overall average interpretation of the “E/M
hypothetical” is consistent with the WPE pattern shown by the Edison/Mitofsky exit polling data. A larger overall average partisan exit poll
response gap than that proposed by E/M could account for the observed WPE, but
the large differences in mean and median partisan response bias necessary to
produce the actual observed WPE levels across partisan precinct categories
would need to be explained. In particular the very large mean and median WPE
for precincts with over 80% reported Bush vote, and the almost zero mean and
median WPE for precincts with over 80% reported Kerry vote, requires an
explanation.
Ten
months after the election, no plausible explanation of the 2004 exit poll
discrepancy, based on exit polling error, has been provided by E/M. The
precinct level exit polling and official vote count data that would enable
independent investigators to analyze the exit poll discrepancy has not been
provided to the public. Perhaps an exit
poll explanation for the discrepancy does exist. However, a cloud of suspicion
is cast on the 2004 presidential election results because the possibility that
a “vote miscounts” explanation is required to generate the reported exit poll
discrepancies is still open. It is a matter of the utmost national importance
that detailed precinct level exit polling and election data that would allow
for investigation by independent analysts, such as USCV, be publicly released.
