Elections this year have been good to the Democrats, but a new study from the Kennedy School of Government reveals that the party of Jefferson and Jackson actually faces a systematic disadvantage in tight races.

Democrats—and especially Hispanic, low-income, non-English speaking, less educated, and young voters—are more likely to miscast their ballots. At least one out of every 400 votes are accidentally cast for a candidate other than the one the voter intended to support, the working paper says.

The paper, posted on the Kennedy School’s Web site last week, used information from the 2003 California recall election—in which 135 candidates, including porn star Mary Carey, the diminutive Gary Coleman, and the ultimately successful Arnold Schwarzenegger, vied for the governorship.

California, among other states, lists candidates’ names in an order that changes from district to district. The 157 variations provided a way for the paper’s authors to draw conclusions on miscast votes.

Confused or rushed voters would sometimes vote for the candidate adjacent to the one they meant to support. “An estimate of adjacency misvoting,” according to the study, was the increase in votes a minor candidate would garner when he or she was placed next to a major candidate on a certain ballot.

Voting results were contrasted with demographic data to see which groups were most likely to cast erroneous votes.

The study estimates that, in a close race, Republican Schwarzenegger would have received an 0.06 percent boost over his Democratic rival due to misvotes alone—since those misvotes were more likely to be cast by Democratic-leaning demographics.

In a close race, like the 2000 presidential election, that margin could have meant the difference between victory and defeat.

Though the study was only based on this particular election, the results can be extrapolated to elections nationwide, said Erzo F. P. Luttmer, an assistant professor of public policy at the Kennedy School and one of the study’s co-authors.

“Californian voters are no more or less likely to make mistakes. The technology used was the same,” he said. “There were a larger number of candidates, which might have increased people’s chances of being confused. But we think that it’s very plausible that this type of effect happens in similar elections.”

To fix the problem with miscast votes, Kelly Shue ’06, a doctoral student in economics and the paper’s other co-author, suggests “clearly separating candidates on a ballot, spacing them apart, and moving away from the use of columns on a ballot.”

The study’s authors hope to follow up their work on the national level—though they acknowledge that California’s randomized ballots provided a particularly good source of data that might be hard to find elsewhere.

The working paper is being submitted