Posted April 1, 2005
By Stephen Dyer
Beacon Journal staff writer
Write an email about this article
Go to the original site of this article (may require registration).
There's a one-in-959,000 chance that exit polls could have been so wrong in predicting the outcome of the 2004 presidential election, according to a statistical analysis released Thursday.
Exit polls in the November election showed Sen. John Kerry, D-Mass., winning by 3 percent, but President George W. Bush won the vote count by 2.5 percent.
The explanation for the discrepancy that was offered by the exit polling firm -- that Kerry voters were more likely to participate in the exit polling -- is an ``implausible theory,'' according to the report issued Thursday by US Count Votes, a group that claims it's made up of about two dozen statisticians.
Twelve -- including a Case Western Reserve University mathematics instructor -- signed the report.
Instead, the data support the idea that ``corruption of the vote count occurred more freely in districts that were overwhelmingly Bush strongholds.''
The report dismisses chance and inaccurate exit polling as the reasons for their discrepancy with the results.
They found that the one hypothesis that can't be ruled out is inaccurate election results.
``The hypothesis that the voters' intent was not accurately recorded or counted... needs further investigation,'' it said.
The conclusion drew a yawn from Ohio election officials, who repeated that the discrepancy issue was settled when the polling firms Edison Media Research and Mitofsky International disavowed its polls because Kerry voters were more likely to answer exit polls -- the theory Thursday's report deemed ``implausible.''
Ohio has been at the center of a voter disenfranchisement debate since the election.
``What are you going to do except laugh at it?'' said Carlo LoParo, spokesman for Ohio Secretary of State J. Kenneth Blackwell, who's responsible for administering Ohio's elections and is a Republican candidate for governor. ``We're not particularly interested in (the report's findings). We wish them luck, but hope they find something more interesting to do.''
The statistical analysis, though, shows that the discrepancy between polls and results was especially high in precincts that voted for Bush -- as high as a 10 percent difference.
The report says if the official explanation -- that Bush voters were more shy about filling out exit polls in precincts with more Kerry voters -- is true, then the precincts with large Bush votes should be more accurate, not less accurate as the data indicate.
The report also called into question new voting machine technologies.
``All voting equipment technologies except paper ballots were associated with large unexplained exit poll discrepancies all favoring the same party, (which) certainly warrants further inquiry,'' the report concludes.
However, LoParo remained unimpressed.
``These (Bush) voters have been much maligned by outside political forces who didn't like the way they voted,'' he said. ``The weather's turning nice. There are more interesting things to do than beat a dead horse.''