Posted: October 29, 2008

The Ohio Ballot Experiment of 2008

In preparation for this year’s election, I have been working with Larry Norden and Margaret Chen of the Brennan Center, and Whitney Quesenbery of the Usability Professional Association to focus more attention on improving ballot design, to reduce voting errors and minimize the number of disputed ballots in post-election recounts. Our report, Better Ballots, summarizes research on ballot design and identifies several confusing ballot features that should be avoided. One of the main recommendations in the report is to list candidates for the same office in a single column on a single ballot page or screen. Some voters see the second column as another contest and mistakenly make a selection from both columns, thus overvoting the ballot.

Ohio has eight candidates on the ballot running for president this year. With this many candidates, some Ohio counties using optically scanned ballots have listed the candidates in two columns on the ballot. By the most recent count, twelve counties in Ohio using optical scan ballots are listing the candidates for president in two columns. The counties are Ashtabula, Athens, Auglaize, Champaign, Delaware, Lawrence, Logan, Madison, Ottawa, Seneca, Shelby, and Wyandot, and they include more than half a million registered Ohio voters. Larry Norden has already blogged about this issue, with links to images of many of the county ballots. A directive from the Secretary of State’s office to county election officials included a ballot template that also listed the presidential candidates in two columns (see page 12 of the directive). We communicated with the Secretary of State’s office and with county election officials, urging them to avoid the two-column layout for the presidential contest, but we were not persuasive enough. In addition, at least three Ohio counties using electronic voting machines have placed the candidates for president on two separate pages. These are Hancock, Portage, and Wayne counties, including more than 230,000 registered voters. Larry Norden has also blogged with more details about these ballots.

We have seen ballots like this before, and the results are not pretty. One way to measure the impact of ballot design on voting behavior is the “residual vote rate” – the difference between the number of ballots cast and the number of valid votes cast in a particular contest. In 2002, ballots in Kewaunee County, Wisconsin listed gubernatorial candidates in two columns (see an image of the ballot here). The residual vote rate for the gubernatorial contest in Kewaunee County was almost 12%, substantially higher than the residual vote rate of 1% in the rest of the state where the contest was listed in one column. A similar problem contributed to the presidential election controversy in Florida in 2000. While a lot of attention focused on the “butterfly ballot” and punch cards in Palm Beach County, there were fourteen other Florida counties with optical scan systems that listed the presidential candidates in two columns. The residual vote rate for president was much higher in the two-column counties (6.9%) than in other optical scan counties in Florida that listed the candidates in one column (0.9%).

Voting in the 2008 general election has already begun in Ohio, so it appears that it is too late the change these ballot designs. In effect, Ohio is conducting what social scientists call a “natural experiment” with respect to the arrangement of presidential candidates on the ballot. We will have a chance to see the results after Election Day. If there is a recount in Ohio, expect some extra attention for the ballots in the counties listed above.