|
Edited on Thu Sep-04-08 11:55 PM by tiptoe
2008 ELECTION MODELA Monte Carlo Electoral Vote SimulationUpdated: September 4
Press REFRESH after linking to a graph to view the latest update
Chart State Poll Aggregate + Projection Trend Chart National 5-Poll Moving Average Projection Chart State vs. National: Vote Share Projection Trends Chart Battleground-State Polls Chart Battleground-State Win Probability Chart Obama Electoral Vote Simulation Frequency Chart Electoral Vote + Win Probability Trend Chart Electoral Vote + Projected Vote Share Trend Chart Undecided Voter Allocation + Win Probability Chart Monte Carlo Electoral Vote Simulation Trials 2008 Election Model Fraud Analyzer Uncounted & Switched Votes Chart Effect on Obama Projected Vote Share Chart Effect on Obama Projected Electoral Vote This State National State National Monte Carlo Simulation Update Poll 5-Poll 2-party 2-party Expected 9/04/2008 Aggregate Average Projection Projection ELECTORAL VOTE Obama McCain 46.39 (51.27) 44.09 (48.73) 49.00 (53.96) 41.80 (46.04) 52.10 47.90 54.52 45.48 334 204
15-Poll
End
Sample
Poll
NATIONAL MODEL Pre Undecided-Voter Allocation 5-Poll Mov Avg 2-Party 2-Party Projection (60% UVA) 5-Poll Mov Avg Trend Rasmussen Gallup Hotline/FD USA/Gallup CBS News
CNN CNN Hotline/FD USA/Gallup ABC/WP
FOX CBS/NYT NBC/WSJ Bloomberg Quinnipiac
RV vs LV Poll Averages
Post-Convention Pre-Convention Total Change Share of Other
Date 9/03 9/03 8/31 8/31 8/31
8/31 8/24 8/24 8/23 8/22
8/20 8/19 8/18 8/18 8/17
Size 3000 LV 2771 RV 805 RV 1835 RV 781 RV
927 RV 909 RV 1022 RV 765 LV 1000 RV
900 RV 869 RV 1005 RV 1248 RV 1547 LV
LV avg RV avg Total
8/31–9/04 8/17–8/24
MoE 1.79% 1.86% 3.45% 2.29% 3.51%
3.22% 3.25% 3.07% 3.54% 3.10%
3.27% 3.32% 3.09% 2.77% 2.49%
Obama 50 49 48 50 48
49 47 44 48 49
42 45 45 45 47
48.3 46.8 47.1
49.0 45.8 3.2 92.1
McCain 45 42 39 43 40
48 47 40 45 43
39 42 42 43 42
44.0 42.3 42.7
42.8 42.6 0.3 7.9
Other 5 9 13 7 12
3 6 16 7 8
19 13 13 12 11
7.7 10.9 10.3
8.2 11.7 -3.5
Spread 5 7 9 7 8
1 0 4 3 6
3 3 3 2 5
4.3 4.4 4.4
6.2 3.2 2.9
Obama 49.0 48.8 48.4 47.6 47.2
47.4 46.0 45.6 45.8 45.2
44.8 44.6 44.6 44.8 44.4
McCain 41.8 42.4 43.4 43.6 44.0
44.6 42.8 41.8 42.2 41.8
41.6 43.0 42.2 42.4 41.4
| Obama 54.0 53.5 52.7 52.2 51.8
51.5 51.8 52.2 52.0 52.0
51.9 50.9 51.4 51.4 51.7
Win Prob 100.0 100.0 93.9 97.0 83.7
82.3 86.1 91.8 87.1 89.2
86.7 70.5 81.0 83.5 91.5
Obama 54.5 54.1 53.3 52.9 52.5
52.2 52.7 53.2 53.0 53.0
53.0 52.0 52.5 52.5 52.9
McCain 45.5 45.9 46.7 47.1 47.5
47.8 47.3 46.8 47.0 47.0
47.0 48.0 47.5 47.5 47.1
Spread 9.0 8.2 6.6 5.8 5.0
4.4 5.4 6.3 6.0 6.0
5.9 4.1 5.0 5.0 5.8
Win Prob 100.0 100.0 97.0 99.3 91.7
91.0 95.0 97.8 95.1 97.1
96.2 88.5 94.5 96.0 98.9
The Obama surge in the national polls continues. He leads the National projection model by nine points, 54.5–45.5%. He leads the State model by 52.1–47.9% with 334 expected electoral votes. Since state polls lag the nationals by 2–3 weeks, the State vs. National vote share projection Trend should tighten. Obama could expect 400 electoral votes within the next few weeks — just in time for another October Surprise.
The “dead heat” claimed by pollsters, bloggers and the media is a canard — unless they are factoring fraud into their models and not telling us. The media desperately wants a horserace, and so they fail to adjust the polls for undecided and newly registered voters. They avoid McCain’s gaffes, flip-flops and plagiarisms, while he supports the most unpopular president in history.
And now they say that the race will tighten, because Palin will rally the base, has a strong personality and the executive experience that Obama lacks. And on top of all that, she understands Putin, because Alaska is near Russia. That’s like saying you’re qualified to be a heart surgeon, because there’s a hospital across the street from your house.
In a true democracy, this election would be a slam dunk.
The Great Election Fraud Lockdown: Uncounted, Stuffed and Switched Votes
The dirty little secret of election fraud has always been off the table in Congress. Click on the state abbreviation link in the projection table below to see why election fraud is a nationwide pastime. Professional associations such as AAPOR, media pundits and election forecasters never discuss Election Fraud. On the other hand, they want you to believe that Democrats always do better in the exit polls, because Republican voters are reluctant responders. But they never consider other, more plausible explanations — such as uncounted votes. In 2000, a net 5.4 million of 110.8m total votes cast (4.9%) were uncounted, of which approximately 4.0m were Gore votes. In 2004, 3.4m of 125.7 million were uncounted (2.7%) and 2.5m were for Kerry. Apparently, election fraud naysayers are either unaware that millions of votes are uncounted (or stuffed) in every election — or they don’t want you to know that they know. That is one reason, but not the only one, why the Democratic True vote (and exit poll share) is always greater than the Recorded vote. Read more here. The final 2004 Election Model projection gave Kerry 337 EV and 51.8% of the two-party vote. Bush won the Recorded Vote by 62–59m with 286 EV. But Kerry won the True Vote. The unadjusted, pristine aggregate state exit poll vote share (based on the WPE / IMS method) was provided by exit pollsters Edison-Mitofsky in their Jan. 2005 report. It indicated that Kerry had a 52.0% share with 337 electoral votes — exactly the same as the Election Model Monte Carlo EV projection. The Election Calculator model (see below) indicates that in addition to the uncounted votes, over 5m Kerry votes (7.9%) were switched to Bush. It wasn’t even close: Kerry won the election by close to 10m votes. Because of the 2004 fiasco, a new feature in the 2008 Election Model provides a sensitivity analysis to gauge the potential effect of two major fraud components: uncounted and switched votes. The model summary table displays Obama’s projected electoral and popular vote, assuming a) 3% of total votes cast are uncounted (approximately 75% are Democratic) and b) 4% of his votes are switched to McCain. Two graphs display the effects of 36 uncounted and switched vote scenarios on the electoral and popular vote (see the links below).
Obama needs a massive Voter Registration and GOTV effort to overcome the fraud.
The Impossible Final National Exit Poll: Vote Scam Equivalent of Three-Card Monte
Election forecasters and media pundits who projected a Bush victory avoid mentioning the overwhelming evidence that the election was stolen. On the contrary, a complicit media has been in a permanent election fraud lockdown, while it relentlessly promotes the fictional propaganda that Bush won BOTH elections.
Repeat a lie often enough, and it becomes conventional wisdom. Although the media commissioned exit polls which indicated that Kerry won by 5%, they never explained why mathematically impossible weights were used in the Final Exit Poll to 'force' a match to the recorded vote count.
According to the Final, 43% of 2004 voters ( 52.6m) were former Bush 2000 voters, while just 37% were former Gore voters. But Bush only had 50.5m votes in 2000. Approximately 2.5m died, and another 2.5m did not return to vote. Therefore, only 45.5m former Bush 2000 voters could have returned to vote in 2004. The Final National Exit Poll overstated the Bush 2004 vote by approximately 7 million; it had to in order to match the corrupt miscounted vote. In the Three-Card Monte con, the mark is tricked into betting that he can find the money card among three face-down cards. A rigged election is the Vote Scam equivalent of the Three-card Monte. What you see in the exit polls is not what you get in the recorded count; the Recorded vote is never equal to the True vote. In this con game, the voter is the mark. Any model which correctly calculates the True vote is doomed to fail in a rigged election.
Allocating Undecided Voters: Sensitivity Analysis In the 2008 Election Model, Obama is considered to be the challenger, since McCain is running for Bush’s third term. Typically, challengers win 60–90% of the undecided vote (UVA) if the incumbent is unpopular. In 2004, final state and national Pre-Election Polls had the race nearly tied at 47%. Bush had a 48% approval rating. That’s one reason why the Gallup poll projected that Kerry would win 88% of the late undecided vote. The 2004 Election Model allocated 75% of the undecided vote to Kerry as the base case of a five UVA sensitivity analysis. The base case scenario projected that Kerry would have an expected 337 electoral votes with 51.8% of the two-party vote. His electoral vote win probability was over 99%. The 2008 Election Model includes a sensitivity (risk) analysis of five Obama undecided voter (UVA) scenario assumptions ranging from 40–80%, with 60% as the base case. This enables one to view the effects of various projection assumptions on the expected electoral vote and win probability. Electoral vote forecasting models which do not provide a risk factor sensitivity analysis are incomplete.
Calculating the Expected Electoral Vote: A Simple Summation
It’s hard to understand why election forecasting blogs and academics and the media, who employ the latest state polls as input to their models, don’t use basic probability, statistics and simulation concepts in forecasting the electoral vote and corresponding win probability.
A meta-analysis or simulation is not required to calculate the expected electoral vote. Of course, the individual state vote projections depend on the particular forecasting method used.
These are the steps used in the 2008 Election Model to calculate the expected electoral vote:
- Project the 2-party vote share V(i) for each state by applying an undecided voter allocation (UVA):
V(i) = PS(i)+UVA(i), for i=1,51 states
- Calculate the probability P(i) of winning each state. A 4% polling MoE (95% confidence) is assumed:
P(i) = NORMDIST ( V(i), 0.5, .04/1.96, true )
- Calculate the expected electoral vote ExpStateEV(i) for each state (win probability times EV):
ExpStateEV(i) = P(i)* EV(i)
- Calculate the total expected electoral vote ExpEV as the sum of the expected state EVs:
ExpEV = Σ ExpStateEV(i), for i = 1,51 states
Calculating the Probability of Winning the Electoral Vote: Monte Carlo Simulation
The Excel-based Election Model is very straightforward as shown above. After updating the database for the latest state polling data, the vote shares are projected. The normal distribution function calculates the corresponding state win probability. The expected state EV is the product of the win probability and electoral vote. The sum of the 51 state expected EVs is the total expected EV. The final step is to calculate the EV Win Probability. The Election Model uses a Monte Carlo (MC) simulation. MC is widely used for analyzing complex systems, when an analytical solution is prohibitive due to the virtually infinite number of possible combinations of risk-based variables (i.e. state win probabilities). A random number generator (RND) is used in the simulated election trials. The EV win probability is just a simple division: the number of winning election trials divided by 5000 (total trials). The Monte Carlo mean and median EV of the election trials match are always within one of the EV summation formula. This proves that 5000 election trials are sufficient to derive a theoretically accurate win probability. The simulation illustrates the Law of Large Numbers (LLN). With all due respect to Professor Sam Wang, his Meta-Analysis program is an unnecessarily complex combinatorial algorithm when compared to Excel and Monte Carlo simulation for calculating the expected Electoral Vote and Win Probability.
These graphs display the effects of uncounted and switched votes on Obama's projected EV and 2-party vote share. ?click">Effect of uncounted and switched votes on the projected vote share ?click">Effect of uncounted and switched votes on the electoral vote
Polling data source: Electoral-vote.com RealClearPolitics.com THE 2008 ELECTION MODEL
Last State National State National Monte Carlo Simulation Update Poll 5-Poll 2-party 2-party Expected 9/04/2008 Aggregate Average Projection Projection EV 60% UVA Obama McCain 46.39 44.09 49.00 41.80 52.10 47.90 54.52 45.48 334 204 2004 Final 75% UVA Kerry Bush 47.88 46.89 47.80 46.60 51.80 48.20 51.77 48.23 337 201 Projected Recorded Vote (assuming fraud) Obama Vote Share after adjustment for: Popular Vote Electoral Vote 49.3% 259 3% - 4% - Uncounted Votes Switched Votes True Vote Sensitivity Analysis I — Undecided Voter Allocation (UVA) UVA Scenario Base Case Obama 40% 50% 60% 70% 80% Projected 2-Party Vote Share Obama McCain 50.2 49.8 51.1 48.9 52.10 47.90 53.1 46.9 54.0 46.0 MoE Obama popular vote win probability 2.0 % 2.5 % 3.0 % 57.6 56.1 55.1 87.0 81.6 77.3 98.0 95.0 91.5 99.9 99.2 97.7 100.0 99.92 99.56 Obama Expected Electoral Vote Average Median 285 284 308 308 334 334 363 364 389 387 Maximum Minimum 372 187 391 197 414 254 438 287 461 318 Obama Electoral Vote Win Probability Trial Wins Probability 3650 73.0 4779 95.6 4994 99.88 5000 100.0 5000 100.0 95% Confidence Level Upper Lower 329 240 352 263 377 290 406 319 430 347 States Won Obama 24 25 28 31 35 Sensitivity Analysis II — Projected Vote Share & Win Probability Obama 46.9 47.9 49.0 50.0 51.1 52.10 Electoral Vote Win Probability 176 0.0 215 1.1 248 17.7 278 63.7 307 95.2 334 99.88 MoE 2.00% 2.50% 3.00%
0.1 0.7 2.1
2.1 5.3 8.8
Popular Vote Win Probability 15.8 21.1 25.2
50.7 50.5 50.4
85.0 79.7 75.6
98.03 95.02 91.51
2008 POLLING ANALYSIS AND PROJECTIONS National Model — see atopState Model(2-party vote shares) L A T E S T P O L L S OBAMA vs KERRY MONTE CARLO EV SIMULATION Pre-Undecided Voter Allocation Projection Projection JK Exit Poll Recorded Diff Diff Obama Obama EV WTD EV EV Flip(*) EV 538 9 3 10 6 55 9 7 3 3 27 15 4 4 21 11 7 6 8 9 4 10 12 17 10 6 11 3 5 5 4 15 5 31 15 3 20 7 7 21 4 8 3 11 34 5 3 13 11 5 10 3 Obama 46.39 % 36 36 36 37 53
46 51 90 50 45
44 61 37 50 43
55 35 37 38 49
53 49 45 53 42
43 47 32 49 45
50 51 53 43 42
47 37 50 48 51
40 43 32 41 31
63 47 52 37 49 40 McCain 44.09 % 51 48 44 47 40
47 36 9 41 46
53 31 53 37 48
40 58 55 55 36
41 37 40 41 53
50 47 50 44 44
39 42 37 46 45
45 42 43 43 30
53 47 56 50 55
29 45 42 45 43 53 Diff 2.30 (15) (12) (8) (10) 13
(1) 15 81 9 (1)
(9) 30 (16) 13 (5)
15 (23) (18) (17) 13
12 12 5 12 (11)
(7) 0 (18) 5 1
11 9 16 (3) (3)
2 (5) 7 5 21
(13) (4) (24) (9) (24)
34 2 10 (8) 6 (13) Obama 52.10 % 43.8 45.6 48.0 46.6 57.2 50.2 58.8 90.6 55.4 50.4 45.8 65.8 43.0 57.8 48.4 58.0 39.2 41.8 42.2 58.0 56.6 57.4 54.0 56.6 45.0 47.2 50.6 42.8 53.2 51.6 56.6 55.2 59.0 49.6 49.8 51.8 49.6 54.2 53.4 62.4 44.2 49.0 39.2 46.4 39.4 67.8 51.8 55.6 47.8 53.8 44.2 Final Kerry 51.80 % 42.0 39.8 48.8 51.0 55.8
50.8 56.5 86.3 57.8 52.3
46.5 52.5 38.3 57.0 41.3
54.5 39.3 42.8 49.0 58.3
56.3 70.8 54.3 55.0 47.3
49.3 41.3 37.3 50.5 51.5
56.0 50.5 60.0 49.0 42.5
52.3 36.3 54.5 53.8 62.0
44.3 46.5 49.2 40.0 29.3
58.3 48.5 55.0 50.0 54.8 33.5 IMS WPE 52.49
42.3 40.6 45.0 45.7 60.7
50.6 62.9 91.5 61.9 51.5
42.4 58.7 32.6 57.1 40.8
51.2 37.5 40.3 44.0 56.1
60.2 66.4 55.0 56.3 49.5
49.5 37.6 37.4 53.4 57.8
58.1 53.6 65.1 50.0 35.0
54.6 34.2 51.9 55.7 62.7
46.2 36.3 43.6 42.4 28.4
67.2 50.3 57.4 40.7 52.6 32.9 Kerry 48.76
37.2 35.9 44.8 45.0 54.9
47.5 54.9 90.1 53.9 47.6
41.8 54.6 30.6 55.4 39.7
49.7 37.0 40.1 42.6 54.1
56.5 62.6 51.7 51.6 40.2
46.6 39.0 33.0 48.4 50.7
53.5 49.5 59.0 44.0 35.9
49.2 34.8 51.9 51.4 60.0
41.3 38.8 43.0 38.6 26.3
59.5 45.9 53.4 43.6 50.2 29.4 Projection 0.30
1.8 5.9 (0.8) (4.4) 1.5
(0.5) 2.3 4.3 (2.3) (1.9)
(0.7) 13.3 4.8 0.8 7.2
3.5 (0.0) (1.0) (6.8) (0.3)
0.4 (13.4) (0.3) 1.6 (2.3)
(2.1) 9.4 5.6 2.7 0.1
0.6 4.7 (1.0) 0.7 7.3
(0.4) 13.4 (0.3) (0.3) 0.4
(0.0) 2.5 (10.0) 6.4 10.2
9.6 3.3 0.6 (2.2) (0.9) 10.7 Exit Poll (0.39) 1.5 5.0 3.0 0.9 (3.5) (0.4) (4.1) (0.9) (6.5) (1.1) 3.4 7.1 10.4 0.7 7.6 6.8 1.7 1.5 (1.8) 1.9 (3.6) (9.0) (1.0) 0.3 (4.5) (2.3) 13.0 5.4 (0.2) (6.2) (1.5) 1.6 (6.1) (0.4) 14.8 (2.8) 15.4 1.4 (2.3) (0.3) (2.0) 12.7 (4.4) 4.0 11.0 0.6 1.5 (1.8) 7.1 1.2 11.3 Exp EV 334.2
0.0 0.0 1.6 0.3 55.0
4.9 7.0 3.0 3.0 15.6
0.3 4.0 0.0 21.0 2.4
7.0 0.0 0.0 0.0 4.0
10.0 12.0 16.6 10.0 0.0
0.9 1.8 0.0 4.7 3.1
15.0 5.0 31.0 6.3 1.4
16.2 3.0 6.9 20.0 4.0
0.0 0.9 0.0 1.3 0.0
3.0 10.5 11.0 0.7 9.7 0.0 Win Prob 99.88
0.1 1.6 16.4 4.8 100.0
53.9 100.0 100.0 99.6 57.8
2.0 100.0 0.0 100.0 21.7
100.0 0.0 0.0 0.0 100.0
99.9 100.0 97.5 99.9 0.7
8.5 61.6 0.0 94.2 78.3
99.9 99.5 100.0 42.2 46.1
81.1 42.2 98.0 95.2 100.0
0.2 31.2 0.0 3.9 0.0
100.0 81.1 99.7 14.1 96.9 0.2 Sprd <9 100%
1.0% 7.0% 21.1% 4.3% 6.6% 2.1% 2.6% 1.9% 3.1% 8.8% 1.8% 13.6% 2.7% 1.4% 8.2% 1.5% 8.9% 0.5% 2.9% Sprd <9 201
10 9 27 11 17 11 3 5 4 15 3 20 7 7 21 3 13 5 10 to Obama 8
AL AK AZ AR CA CO* CT DC DE FL* GA HI ID IL IN IA* KS KY LA ME MD MA MI MN MS MO MT* NE NV* NH NJ NM* NY NC ND OH* OK OR PA RI SC SD TN TX UT VT VA* WA WV WI WY The Election Calculator Model This model uses prior election votes cast, mortality and estimated voter turnout to calculate the True Vote. It was originally developed to determine the 2004 True vote after the fact. It does not calculate the corresponding Electoral vote. But we can estimate the Electoral vote and win probability from the popular vote. National Exit Poll vote shares of returning voters were key inputs.
As of today, the 2008 Election Calculator confirms the Election Model: Obama has 54.1% and will win the True Vote by 71 – 59m. Input consists of 2004 total votes cast (recorded plus uncounted), mortality and 2004 voter turnout in 2008. The vote shares are similar to the 2004 National Exit Poll shares of returning and new voters.2008 True Vote Election Calculator Forecast Estimated vote share (see National Exit Poll) 2004 Turnout Votes Mix Obama McCain Other DNV - 17.2 13.1% 59% 40% 1% Kerry 95% 60.5 46.2% 89% 10% 1% Bush 95% 51.6 39.4% 11% 88% 1% Other 95% 1.6 1.2% 70% 11% 19% Total 113.7 130.9 100.0% 54.1% 44.7% 1.2% 130.9 70.8 58.5 1.6 Election Forecasting MethodologyThese are a few reasons why Monte Carlo (MC) simulation is preferable to election forecasting methods used in the media and academia: - Academic models forecast national vote shares only (months in advance of the election) using regression analysis of economic and political time-series.
- The Election Model projects the popular and electoral vote (and win probabilities) based on the latest state and national polls right up to the election.
- MC does not arbitrarily designate states as being “too close to call”; it automatically factors in the poll-based win probabilities.
- MC is a powerful tool for analyzing risk-based systems (i.e. electoral vote models) when deriving an analytical solution is impractical or impossible.
In each of the 5000 Monte Carlo election trials, the winner of each state is determined using win probabilities calculated from the latest poll-based projection. The winner of the election trial is the candidate who has at least 270 EV. The electoral vote win probability is simply the number of winning election trials divided by 5000. State and national aggregate popular vote win probabilities are calculated using the Excel normal distribution function. The projected two-party vote share and standard deviation (MoE/1.96) are inputs to the function. A constant 4% MoE is assumed for all state poll win probabilities. The National Model calculates a 5-poll moving average projection assuming the base case Obama 60% UVA scenario. The National projection would normally be a leading indicator of the State model aggregate since it is based on up-to-date polling. The normal distribution function calculates the national popular vote win probability assuming the MoE of the latest poll. Read more about Election Forecasting Models and Monte Carlo simulation here. 2004 Registered Voter (RV) vs. Likely Voter (LV) Polls- The national pre-election RV polls were closer to the True Vote than likely voter LV polls.
- The LV polls, after adjustments, matched the RVs — and the unadjusted exit polls.
The Election Calculator Model determined that Kerry won by 66.9 – 57.7 million. Simple arithmetic shows that approximately 5.4m votes (8.0%) were switched from Kerry to Bush. In most states, votes cast exceeded votes recorded — the net uncounted vote. In Florida, Ohio and other states, votes recorded exceeded votes cast — the net padded vote. 2004 Calculated True Vote 12:22am NEP vote share 2000 Turnout Voted Mix Kerry Bush Other DNV - 25.6 20.4% 57% 41% 2% Gore 95% 49.7 39.5% 91% 8% 1% Bush 95% 46.6 37.1% 10% 90% 0% Other 95% 3.8 3.0% 64% 17% 19% Total 100.1 125.7 100% 53.2% 45.4% 1.4% Votes cast 125.7 66.9 57.1 1.7 Recorded Vote (actual) 122.3 59.0 62.0 1.2 48.3% 50.7% 1.0% Deviation from True Vote -4.9% +5.3% -0.4% Unadjusted Exit Poll 52.0% 47.0% 1.0% Deviation from True Vote -1.2% +1.6% -0.4%
|