Before going to my polling place to vote in the November 2023 local elections, I searched online to find a sample ballot. After clicking around, I found one on a non-government site and reviewed the list of contests and ballot measures. I had to read the instructions a few times to clearly understand how many candidates I should select in each contest. Similarly, I had to re-read one ballot measure several times—about how city councilors were paid—to understand what the question was asking. After reviewing the sample ballot, I researched the candidates online and went to my polling place on Election Day confident that I knew exactly what was going to be on my ballot.
I was wrong.
Several judicial races hadn’t been listed on the online sample ballot. Standing in the voting booth, I asked myself how it was possible that someone like me, who has devoted my career to studying political science and elections and who does research ahead of time, was encountering such difficulties just trying to cast my vote?
A big part of the answer is poor ballot design.
This Super Tuesday, millions of Americans across sixteen states are going to cast their votes. The question isn’t if we will see uncounted votes but rather how many and whose votes are most likely to go uncounted?
Understanding ballot design
For a brief and puzzling moment in 2000, after the hanging chads debacle in Florida, Americans and election administrators considered the importance of how ballots are designed. Unfortunately, since then, this core element of election administration has been most often overlooked despite its impact on how Americans vote and whether their vote gets counted.
Simply put, ballot design is how electronic and paper ballots look—including the font used, spacing, the placement of contests and candidates, and the wording of instructions. Scientists have shown that all these things matter, and that bad ballot design, or errors on the ballot, can result in uncounted residual votes. Bad ballot design can lead to people voting for too many candidates in a contest (overvotes), failing to select any candidates in a contest (undervotes), or neglecting to vote in contests listed lower on the ballot (ballot roll-off).
Nationally, since 2004, the evidence shows that the number of over- and undervotes averages about one percent. While that sounds low, a big problem is that certain districts have higher residual vote rates than others. Districts with higher populations of Black, Hispanic, low-income, less educated, and older voters have higher rates of unrecorded votes. While some undervotes are intentional, confusing ballot design and voting technology has been linked to over and undervotes rates. Hundreds of thousands of votes go uncounted every election season because of poor ballot design and voter education materials.
Bad ballot design in action
Common flaws in ballot design include poor layout, poor voter instructions, and confusing absentee and mail-in ballot envelopes. The first flaw, bad ballot layout, led to more than twice as many uncounted votes in East St. Louis than occurred statewide in 2008 when election administrators there failed to clearly differentiate between contests. When Ohio paper ballots in ten counties in 2008 split the US Presidential contest across two columns instead of listing candidates in one continuous column, it resulted in 50 percent more uncounted votes in those counties than in those that didn’t split the contest.
An often-cited example of the ramifications of badly written ballot instructions is ballots in Miami-Dade County in Florida. In 2008 and 2010, the county didn’t provide details on how voters could correct any errors they made on their ballots and instead told voters to refer to a separate instructional packet to learn how to fix their mistakes. In 2008, these incomplete and uninformative instructions resulted in an overvote rate 2.5 times higher than the rest of the state in 2008 and five times higher in 2010.
Not only does ballot design matter, but the increasing number of Americans using absentee and mail-in ballots makes ballot design even more crucial.
Already steadily rising since the early 2000s, the percentage of voters who mailed in their ballots was 43% in 2020. Across racial groups, nontraditional voting (voting early or by absentee ballot) is exceedingly popular—more than 65% of non-Hispanic White, Black, Asian, and Hispanic voters reported using nontraditional voting in 2020. In fact, Asian and Hispanic voters had the highest rates of nontraditional voting at 82.4% and 76.7%, respectively. This primary season, some 934,000 absentee votes have been cast in the Michigan primary election so far.
Given the increasing use of absentee and mail-in ballots, eliminating ballot design flaws is more important than ever. While absentee ballot rejection rates are low in most of the country, rates vary across states. For example, in 2020 New York, New Mexico, and Arkansas had more than 2.5% of their absentee ballots rejected. In total, over 530,000 votes were rejected in primaries across the country in 2020.
In my home state of Pennsylvania, 17,000 mail-in ballots—or nearly 3% t of the total number of mail-in ballots cast—were rejected and, therefore, left uncounted in the 2023 primary election.
Ballots were uncounted for a myriad of reasons such as arriving after election day, missing a date or signature, having the incorrect date, or the lack of a secrecy envelope. In response to the high rejection rate, election administrators redesigned mail-in ballot envelopes, rewrote the instructions, and made other design alterations to reduce voter error in the 2024 election.
Bad ballot design affects BIPOC communities unfairly
These changes are especially important because flawed absentee/mail-in ballots disproportionately squash the voices in certain communities. For example, an analysis of absentee rejections in the 2022 midterm election in three Pennsylvania counties found that flawed mail-in ballots were more likely to come from districts with higher concentrations of people of color. For example, in Allegheny County, where I live, voters who submitted flawed ballots were about six percent more likely to come from a zip code with a higher concentration of people of color.
Fortunately, well-designed ballots can reduce these kinds of unfair discrepancies that keep too many from having their votes counted. Districts that use voting equipment programmed to inform voters when they have overvoted significantly lowered number of uncounted ballots, eliminating the correlation between higher uncounted ballot rates and districts with higher Black and Hispanic populations. Moreover, studies indicate that well-designed ballots substantially decrease the number of unrecorded votes in districts with higher Black populations.
How science can improve ballot design?
UCS’s Center for Science and Democracy is working with our recently empaneled Election Science Task Force of experts, election administrators, and elected officials to draft a list of science-based recommendations about the best ways to design ballots and educate voters about the election process.
Among the recommendations we’re exploring are the adoption of pre-election usability tests and public comment periods. Pre-election usability tests allow voters to review election materials and ballots by simulating the voting process. This practice can help identify ballot errors, confusing instructions, and issues with voting technology. It is important for these usability tests to include voters of specific communities, including disabled people and those for whom English is a second language to ensure that the ballot design materials are fully accessible and correctly translated. Another potential best practice is for election administrators to receive public input by holding public comment periods for feedback on sample ballot and mail/absentee materials before they are put into use.
It’s 2024—way past time in our country for our election ballots and voter educational materials to be widely accessible and easy for everyone to understand, regardless of their language, educational attainment, or disability. Bad ballots and inaccessible or incomplete voter education materials shut people out of elections, especially people of color, people with lower incomes, and older people.
The good news is that we can use science to improve our elections and create a healthier democracy and, together, we’re taking steps to make it happen.