UW News

December 9, 2004

Who did voters pick on Nov. 2? In some cases, we’ll never know

As citizens of Washington state wait out a third count with 42 votes separating the candidates for governor, new research shows that Washington was not the only state where the voters’ true choice may never be known.

In three other states, the margin of voting error was greater than the margin of victory for the U.S. Senate winners, according to a University of Washington white paper to be released today.

And in three more states, the margin of error was larger than the winning presidential candidate’s victory margin, the researchers found. This means that John Kerry conceivably deserved a dozen more electoral votes than he received — almost enough to swing the election his way.

In all, the study found seven states in which the margin of error in that state’s voting process was greater than the margin of victory for the candidate declared a winner.

“This means state officials certified an electoral outcome that may not have reflected the intent of voters,” said one of the authors, Philip Howard, a UW assistant professor specializing in political communication. “In really close races, the electoral process can’t tell us what the voters want.”

The study, which Howard and his students conducted as part of a project called CampaignAudit.org, documents the U.S. voting system’s inability to determine with certainty the intent of the electorate.

The researchers drew on a 2000 Caltech/MIT study that calculated the known error rate for various kinds of voting processes. Average error margins, according to that study, were 1 percent for punch cards, 1.2 percent for optical scans, 1.6 percent for touch screens or other e-voting, 1.7 percent for levers and 2 percent for hand-counted paper ballots.

The UW group mapped these error percentages onto the 3,140 U.S. counties for which the 2004 voting system was known, and produced statewide voting-technology error rates that could be compared with the margin of victory in key races.

States where President Bush’s victory margin was smaller than the state’s overall error margin were Iowa and New Mexico, while Kerry’s victory margin was less than the margin of error in New Hampshire.

In each case, the majority of voters’ actual intent theoretically could have been the reverse of what was certified. Under the most extreme scenario, both margin-of-error states that went for Bush could have been Kerry’s. If that happened, and if Kerry held onto all the states certified for him, the Democrat would have gained 12 electoral votes — but still not the White House.

For U.S. Senate races, the UW study found Republicans winning by smaller than margin-of-error outcomes in Florida, Kentucky and South Dakota. If they had been declared for Democrats, the G.O.P. would now be looking forward to significantly less powerful control over the Senate.

Washington state, currently undergoing a hand recount after an initial recount tallied a 42-vote difference out of 2.8 million cast, had the country’s only governor’s race within the margin of error.

Howard said the findings call for an urgent national discussion about election reliability.

“Elections officials of all the states should standardize the way they collect and report error statistics,” he said.

Howard said the nation also should consider a system of runoffs for those cases “where it is known that endlessly recounting the sample of votes will not provide a clear outcome.”


For more information, contact Howard at (206) 612-9911 (cell), (206) 221-6532 or pnhoward@u.washington.edu. The full white paper is available at www.CampaignAudit.org.