8/28 Senate Forecast Simulation Model - Richard Charnin (TruthIsAll)

Richard Charnin (TruthIsAll)     source: http://richardcharnin.com/2010SenateForecastSimulationModel.htm

Aug. 27, 2010

The 2010 Senate Forecast Simulation Model determines a forecast probability distribution of the GOP gains using sensitivity analysis and Monte Carlo simulation. The model is based on the latest Senate polling and assumes that the election is held today and will be updated periodically. To determine a probability distribution for net GOP gains, four election simulations were executed. The simulations were based on the latest polling win probabilities.

Assuming a fraud-free election and current polls, the best-case scenario for the Democrats is that the GOP will gain five net seats. The worst case scenario is that the GOP will gain nine.

But assuming election fraud (i.e. Democratic vote share is reduced by 3%), the Democratic best-case scenario is that the GOP will gain nine seats. The worst case fraud scenario is that they will gain eleven. The model indicates that the states where fraud is most likely to occur are IL, KY, NC, NV, PA and WI.

The most important factor in any election is voter turnout. High turnout is always good for the Democrats. Pre-election registered voter (RV) polls are superior estimates of the True Vote (not the recorded vote) in high turnout elections like 2004, 2006 and 2008. In 2010, Democratic turnout is expected to be low. The GOP is poised to make significant gains in the House and Senate. Because of the low expected turnout, the model does not currently factor whether the polls are RV or LV.

The percentage breakout of undecided voters is always a key variable in any election forecast. The base case assumption in the model is that the Democrats will win 60% of the undecided vote. The rationale is that most undecided voters are “Professional Progressives” who may stay home on Election Day.
The simple formula: Projected share = Poll share + undecided voter share.

Election fraud is a key factor that is overlooked in polling models. The Fraud Factor is defined as the percentage deviation from the True Vote.
The formula is Recorded Vote share = True Vote share + Fraud Factor

The model assumes that the True Vote is equal to the Projected Share.
Therefore, the Recorded Vote share = Projected share + Fraud Factor

Because of the uncertainty in the base case assumption, a sensitivity analysis table displays a 40-60% Democratic share of undecided voters. The other base case assumption is that the elections will be fraud-free (i.e. vote switch = 0%). Based on the historical record, this is unrealistic. The sensitivity analysis displays the effects of 0-4% vote switching from the Democrat to the GOP. There are a total of 25 scenarios in the sensitivity table, 5 UVA * 5 Vote Switch.

Assumptions:
Fraud: 3% reduction in Democratic share
Case 1: 50% of undecided voters break to the Democrats (UVA)
Case 2: 60% UVA

No Fraud
Case 3: 50% UVA
Case 4: 60% UVA



Sensitivity Analysis