"ODDITIES" WHEN COMPARING EXIT POLLS vs FINAL VOTES IN THE 2004 ELECTION
Posted April 9, 2005
This page summarizes a few of the "oddities" that emerged in the voting patterns in the 2004 Presidential Election. What is most odd about these and most other "oddities" is that they invariably favored the candidacy of George Bush.
ABC, CBS, NBC, FOX, CNN and the AP (the "National Election Pool" or NEP) commissioned an exit poll from Edison Media Research and Mitofsky International for the 2004 election. Their weighted national poll predicted a Kerry victory of 3%, while the final vote tally showed Bush winning by 2.5%. This 5.5% percent discrepancy is the largest in the poll's history, which is an "oddity" by itself, given that exit polling methodology has been perfected over the past 40+ years, and this was the most carefully conducted exit poll ever.
In January 2005, Edison/Mitofsky issued a report ("E/M Report") and suggested the hypothesis that this 5.5% discrepancy resulted from Bush voters preferentially declining to participate in the exit polling (the "reluctant Bush voter" hypothesis).
Attached is the " US Count Votes Report," which is a sophisticated statistical analysis of the E/M Report. It shows that the "reluctant Bush voter" hypothesis requires a pattern of behavior on the part of Kerry vs. Bush voters that defies rationality. For example, Table 3 (p13) shows that the "reluctant Bush voter" hypothesis could only be true if inexplicably ~90% of Kerry voters responded to exit pollsters in precincts where Bush received 80-100% of the vote, when the rate was only ~60% in other precincts. This is especially hard to accept given that there is no similar or reciprocal pattern for Bush voters.
Furthermore, if true, the "reluctant Bush voter" hypothesis should also be apparent in Senate races, since voters do not usually vote split tickets. Data in the E/M Report do not support a "reluctant Bush voter" in Senate races (p16, US Count Votes Report).
The E/M Report also provides differences between exit polls vs. final vote tallies for Presidential races in 1988, 1992, 1996, 2000 and 2004 on a state-by-state basis. Both fluctuation and complexity are apparent, but several comparisons stand out as "oddities."
The nearby states of Florida and Georgia , showed a swing from the exit polls to the final vote tallies that favored the Republican Presidential candidate by 2.3% and 1.8%, respectively, on average in the 1988-2000 elections. In 2004, Georgia showed a consistent swing of 2.2%, while Florida --a crucial battleground state--showed an enormously greater swing of 7.6% toward the Republican.
The nearby states of Ohio, Pennsylvania and Kentucky showed a swing from the exit polls to the final vote tallies favoring the Republican Presidential candidate of 2.5%, 1.8% and 0%, respectively, on average in the 1988-2000 elections. In 2004, Kentucky showed a consistent swing of 0%, while Ohio and Pennsylvania --two crucial battleground states--showed enormously greater swings of 10.9 and 8.8%, respectively, toward the Republican.
It is an "oddity" that dramatic and historically unprecedented swings toward a Republican candidate were apparent in the 2004 election--and in key battleground states. It has the appearance of voting fraud. (Note: swings also occurred in some strong Republican states, which could be interpreted as voting fraud to bolster the Bush popular vote.)
Another "oddity is that in the recent election in Ukraine, a discrepancy between exit polling vs. final vote tallies was taken as prima facie evidence of voting fraud, while in the US a similar discrepancy was dismissed as an exit polling error, even though some of the same groups were involved in the polling.
Count Every Vote.