Canadian Study Claims Butterfly Ballot is Flawed

Nov. 30, 2000 -- Voters have complained they’re confusing. Lawyers have argued it in court. Now a Canadian scientist is publishing a study to try and prove the infamous “butterfly ballot” leads to errors.

“The morning after the election I was drinking my coffee and watching TV when they flashed an image of the butterfly ballot,” says Robert Sinclair, a Canadian psychologist at the University of Alberta. “I saw right away there was a serious design problem.”

Creating Canadian Equivalent

Since the Canadian elections were only 20 days away, Sinclair called his former graduate adviser, Melvin Mark, a psychologist at Pennsylvania State University, and the two designed a quick study. They decided to poll Canadians on their preferences among the candidates for Prime Minister using ballots with single columns and others with the two-column design.

His results, which will be published next Thursday in the journal Nature, show that more than 7.5 percent of those using a ballot designed like the so-called butterfly ballot, intended to vote for one candidate, but mistakenly voted for another.

The study addresses alleged problems in the two-column butterfly ballot which was used in Palm Beach County, Florida. More than 19,000 ballots in the Florida county were thrown out when machines determined that voters had marked votes for two candidates. Others have argued that Reform Party Candidate Pat Buchanan received a disproportionately high number of votes in the county because of the ballots.

To tailor his test ballot for Canadians, Sinclair listed the candidates, Stockwell Day and Jean Chretien, from Canada’s two dominant parties, at the top of the ballot in the left column. Their locations corresponded to the placement of Governor George W. Bush’s and Vice President Al Gore’s names on the Palm Beach ballot.

On the right, Sinclair listed candidates from lesser-known parties, including candidate Joe Clark, at the top of the right column, corresponding to the place where Buchanan’s name appeared on the Palm Beach County ballot.

Then Sinclair made hundreds of copies of both ballot designs and took them to mock polling stations at two locations in Edmonton, Alberta.

Students Made No Mistakes

In the first location, Sinclair polled Canadian college students and had them select their choices by filling in circles on one of the two ballots. In addition, the participants wrote in the candidate they intended to vote for so Sinclair could see if they voted correctly.

None of the 324 students who took Sinclair’s poll made mistakes in their selections, although the students did report a higher rate of confusion when using the ballot with two columns.

Assuming that college students might be more practiced at filling out the test-like ballot forms, Sinclair then set up a second polling station at a shopping mall in Edmonton the following day. Half of the 116 shoppers selected their choices on the single-column ballot and half used the two-column ballot.

Among those who used the single-column ballots, no one marked the wrong candidate. But four people who used the butterfly ballot marked their choices incorrectly.

Statistically Confusing

Sinclair acknowledges that four may not seem like a high number, but he argues that statistically, it’s very significant.

He points out that the number of people who made mistakes amounted to 7.55 percent of those 116 polled at the mall. Among 150,000 voters, he calculates that number would accumulate to more than 2,000 incorrect ballots.

He also adds that three out of the four people who made mistakes, voted for Clark when they had intended to vote for Chretien. The mistake was identical, he says, to the one that would-be Gore voters have claimed they made in Palm Beach County.

While Mark says he and Sinclair would have preferred to have included more people in the study, he claims their time was limited.

“The longer we kept data collection up, the more likely the Canadian voters would have heard across the border about the butterfly ballot,” Mark says. “And that might have affected the results.”

Lesson in Design

The butterfly ballot was designed by a Democratic county election supervisor in Palm Beach County. The official, Theresa LePore, has already faced numerous lawsuits over the ballot design and has at least one Web site (LePoorDesign.com) dedicated to lampooning her design instincts.

But not everyone is convinced LePore should shoulder all the blame.

“Yes, the butterfly ballot could be more intuitive,” says document designer Brian Fleming of the Lawrenceville, Ala.-based company, HelpWrite Inc. “But it’s also the responsibility of anyone casting their vote to either understand what they’re doing or ask for help.”

Sinclair and Mark say their study is not intended to be a basis for litigation. (Sinclair adds, “I’m Canadian, I could care less who your president is!”) Instead, they say, it’s intended to make a belated point.

“This study cost me three days of my life and 30 bucks,” Sinclair says. “You’d think that somebody might have had the foresight to do what I did and find that this ballot causes confusion.”