States regularly use administrative records, such as motor-vehicle data, in determining whether people have moved to prune their voter rolls. A Yale-led study of this process in Wisconsin shows that a significant percentage of registered voters are incorrectly identified as having changed addresses, potentially endangering their right to vote.

The study, published in the journal Science Advances, found that at least 4% of people listed as suspected “movers” cast ballots in 2018 elections using addresses that were wrongly flagged as out of date. Minority voters were twice as likely as white voters to cast their ballot with their original address of registration after the state marked them as having moved, the study showed.

The findings suggest that states should more clearly communicate the processes they use to update voter-registration files and that a more robust effort is required to confirm whether individuals have moved before they are removed from the voter rolls, said Yale political scientist Gregory A. Huber, the study’s lead author.

“The process of maintaining states’ voter-registration files cries out for greater transparency,” said Huber, the Forst Family Professor of Political Science in the Faculty of Arts & Sciences. “Our work shows that significant numbers of people are at risk of being disenfranchised, particularly those from minority groups.

“Unfortunately, we don’t know enough about the process used to prune voter rolls nationwide to understand why mistakes occur and how to prevent them.”

Regularly updating voter rolls prevents registration files from becoming bloated with individuals who have died, moved away, or are otherwise no longer eligible to vote. When these rolls swell with ineligible voters, it raises concerns about potential fraud (although there is little evidence it causes unlawful voting, Huber says) and creates headaches for political campaigns, which rely on accurate registration records to reach potential voters.

Americans are not obligated to inform local election officials when they move to a new address, but federal law mandates that states identify changes in residence among registered voters. To better accomplish this task, 30 states, including Wisconsin, and the District of Columbia have formed the Electronic Registration Information Center (ERIC), a non-profit organization that assists them in improving the accuracy of their voter rolls.

ERIC uses various administrative records, including motor vehicle data, change of address information from the U.S. Postal Service, and the Social Security Administration’s master death file, to flag registrations that are potentially out of date. It provides states a “movers list” of people who likely have changed residences. The states contact listed individuals, often by sending them postcards they can use to confirm their address. If people do not return the postcards, their registration can be inactivated, starting the process for removal.

Federal privacy protections and ERIC’s agreements with member states prohibit the organization from disclosing who is marked as having moved and on what basis they were flagged as such, making it difficult to examine its process. However, after submitting a Wisconsin Freedom of Information Act request, Huber and his co-authors obtained special “movers poll books” from the state which list all people who were marked as suspected movers and who did not respond to the postcard notification. Individuals in the books who showed up to vote in 2018 signed their names in these books, providing evidence that they voted at addresses that had been flagged as invalid.

The researchers collected movers poll books from a representative sample of election wards and matched their contents against voting records for 2018 local, state, and federal elections. They found that at least 9,000 people — about 4% of those listed in the poll books — voted in 2018 using the address of registration that ERIC had marked as invalid. Minority voters were twice as likely to be incorrectly identified as having moved.

The study likely undercounts the number of registered voters incorrectly listed as having moved, the researchers said, explaining that a significant number of people who did not respond to the postcard might have nonetheless renewed their voting registration before the poll books were published. In addition, the study examined low-turnout elections, making it likely that many people wrongly listed in the poll books weren’t covered in the analysis because they didn’t vote, Huber said.

The researchers are not suggesting that ERIC intentionally targeted minorities.

“There’s no malice here,” Huber said. “ERIC wants to help states, but relying on administrative records inevitably produces mistakes for any number of reasons. This makes the process used to validate having moved, such as mailed postcards, even more important. Without more information, we can’t be certain why the process disparately affects minorities.”

A potential reason for the disparity is that minorities are more likely than whites to live in apartment buildings and large households, which may increase the risk of errors in administrative records, the researchers suggest. In addition, residents of apartment buildings also may be less likely to confirm their address using the postcard since mail service can be spottier in multi-unit buildings than single-family homes.

Huber credits Wisconsin for taking steps to protect people’s voting rights.

“The poll books are a great way to identify mistakes and prevent people from being disenfranchised,” he said. “The state also has same day voter registration, which is another safety valve that doesn’t exist in many states. We suggest that states expend more effort on contacting people at risk of losing their registration.”