The racial burden of cleaning voter rolls

A new study by Penn political scientists shows that errors in removing people from voter rolls in Wisconsin disproportionately impacted minorities.

An illustration of an American flag shows the stripes separating into a maze and one winding up at a ballot box
A new study by Marc Meredith and Katie Steele looks at how cleaning voter rolls impacts minorities.

When Wisconsin cleaned its voter rolls in 2017 and 2018, people who shouldn’t have been removed were inadvertently taken off, and those errors disproportionately impacted minority voters, according to a new study by Marc Meredith of the School of Arts & Sciences and Wharton School senior Katie Steele.

The study was published in the journal Science Advances and was co-authored by Gregory Huber of Yale University and Michael Morse of Harvard University. The findings detail the racial burden of voter list maintenance errors in Wisconsin’s poll books, but also hold lessons for other states around the country, Meredith says.

“States have to figure out ways to identify people who remain registered to vote but no longer reside at their address of registration, and this is a really tough thing to do,” Meredith says. “There’s a lot of controversy about voter registration and when they can be removed. There’s been a Supreme Court case about it recently and lots of popular press attention on the issue.”

Many states work with the Electronic Registration Information Center (ERIC) to help fix their rolls. ERIC is a nonprofit consortium of 30 states and Washington, D.C., that uses data-matching tools to compare voter lists against other records to determine whether voters have moved or died.

“I think ERIC does a pretty good job and generally helps states, but no one has ever evaluated what they do, in part because it’s really hard to get access to the data necessary to do so,” says Meredith.

Their team came up with a creative way to be able to see which registrants ERIC was identifying as potential movers. They were then able to evaluate how accurate the consortium appeared to be in identifying people who had moved and then went on to vote in other places as opposed to people who they suspected might be movers but in fact weren’t.

“While ERIC is usually correct, sometimes they’re wrong, and it turns out they’re more likely to be wrong in the case where the registrant is a racial or ethnic minority as opposed to a white registrant,” Meredith says.

This isn’t about purposeful discrimination on Wisconsin’s part, Meredith says; however, it highlights how voter registration is particularly challenging for people who move frequently.

“At least part of the explanation is that minorities are more likely to be frequent movers, and it’s harder to manage the voter registrations of people who move frequently,” says Meredith.

In planning to research the topic, Meredith, Steele, and their co-authors sent public information requests to almost every state in the nation that used ERIC seeking access to their data.

“Every state told us no,” he says.

But because Wisconsin had a unique system of poll books which exclusively listed registrants that ERIC identified as potential movers, the team could observe these potential movers by collecting and processing copies of these poll books.

Most states are subject to the National Voter Registration Act, which prevents them from immediately removing a voter on the basis of the type of information ERIC would provide, limiting the circumstances under which a voter registration can be immediately removed.

“This paper really highlights why that is important because it’s hard to be entirely sure that someone has moved given the types of information that election administrators have access to,” Meredith says. “Moreover, if there’s disproportionate errors in identifying racial and ethnic minorities, it might be further making it difficult for people who historically have found it difficult to vote. There’s lots of potential for disenfranchisement when safeguards like the National Voter Registration Act are not in place.”

Steele said she found the study particularly fascinating because no one has ever studied ERIC in this way because the data is so difficult to access.

“I think that election administration sometimes gets ignored in discussions about American politics,” she said. “Certainly, there’s a ton of coverage of polling and horse races and candidates. But I think that election administration—like looking at voter registration practices, access to polling places—can and does 100% change electoral outcomes. Working on this research has shown me those effects and clearly there are issues and problems.”

Recently, some states have attempted to push back on some of the requirements of the National Voter Registration Act, and Meredith says his team’s study shows it helps give people time to fix the situation if they are wrongly taken off the rolls, and the current system is a pretty good one.


“We should be very careful when making policy in the name of reducing voter fraud, which should really be called potential voter fraud because there are often other explanations,” he says. “It’s really important to have policies in place that acknowledge that there’s a lot of uncertainty in data indicating that someone may no longer reside at their registration address and make sure that we don't rush to conclusions that ultimately cause people not to be able to vote.”