In Jan. 2005, pollsters Edison/Mitofsky released the 2004 exit poll evaluation report. The discrepancies were actually much greater than previously indicated by the Simon download on Election Day. The average within precinct discrepancy (WPE) was a whopping 6.5%. WPE is a measure of the difference in margin between the exit poll and the recorded vote.
The pollsters claimed the discrepancies were not due to their choice of precincts but rather to non-response bias and concluded that the exit polls could not prove fraud. But since the sampled precincts were representative of the total electorate, the discrepancies must have occurred in one of two ways:
1) In the exit poll, Kerry voters responded at a higher rate than Bush voters.
2) The vote count was corrupted.
Mitofsky offered a number of possible reasons for the discrepancies: the physical location of the interviewers from the voting machines, interviewer age and education and precinct size.
But the theory that Kerry voters responded at a higher rate than Bush voters (the reluctant Bush responder theory) was undermined by the exit poll data itself. Response rates were higher in strong Bush states than they were in Kerry states. The Final 2004 National Exit Poll 43/37% returning Bush/Gore voter share of the 2004 electorate also contradicted the rBr theory.
The unadjusted exit poll margin of error was exceeded in 29 states in favor of Bush. The MoE was exceeded in 16 of the adjusted "Composite" state polls after midnight - a 1 in 19 trillion probability. Imagine what the probability is for 29 states.
The exit poll discrepancies were categorized by state, region and voting method. The unadjusted state exit poll aggregate had Kerry winning by 52-47% as opposed to the recorded 50.7-48.3% - a near-zero probability.
The pollsters claimed the voting machine discrepancies did not indicate fraud - even though paper ballot precincts had the lowest (2%) error (WPE) rates. Touch-screens had 7%, optical scanners 6%, mechanical levers 11% and punch cards 7%.
The pollsters never questioned the integrity of the vote count. The media parroted the pollster reluctant Bush responder theory that 56 Kerry voters participated in the exit poll for every 50 Bush voters. A more plausible theory would be that the polls were correct: Kerry won 52.8% (56/106) of the two-party vote.
With three degrees in applied mathematics, Richard Charnin began his career in defense/aerospace as a numerical control engineer/programmer. He later worked for Wall Street investment banks as manager/developer of corporate financial and merger analysis software. With the 1981 advent of personal computers, he converted mainframe FORTRAN programs to spreadsheets. As a consultant, he has developed quantitative applications for domestic and foreign financial institutions and major industrial corporations.
Charnin never imagined himself as the Internet poster “TruthIsAll”. But after the 2000 election fiasco, he was motivated to develop forecast models using pre-election state and national polls. In July 2004 he began posting weekly Election Model projections. The model was the first to use Monte Carlo simulation and sensitivity analysis to calculate the probability of winning the electoral vote. The final projection gave Kerry 337 electoral votes and 51.8% of the vote, closely matching the preliminary exit polls. Following the election, he began posting analyses, sparking heated online debates.
Forced to retire in Feb. 2006 from a near-fatal heart attack while posting, he spent four months in two hospitals. While recuperating, he proceeded to develop projection models and exit poll analyses for the 2006 midterms. In 2008, he did an analysis of the primaries followed by weekly Election Model projections of the presidential election. The final projection gave Obama a 53.1% share and 365.3 expected electoral votes compared to the recorded 52.9% and 365 EV. But the True Vote Model determined that Obama had 58%, 419 EV and a 22 million vote margin. The analysis presented in the book proves that the True Vote always differs from the official recorded vote. In 1969, primitive computers took us to the moon. Faulty hardware is not the cause of voting machine “glitches'. It's the fault of the humans who program them to miscount votes.