At least 17 million people were removed from the voter rolls between the 2016 election and the 2018 midterms, according to a new report published by the Brennan Center for Justice this week .

The study updated an in-depth 2018 analysis on voter purges and offered further evidence that certain places continued to more aggressively remove people from the voting rolls following the Supreme Courts decision in Shelby County v. Holder in 2013 .

In that decision, the Supreme Court essentially gutted Section 5 of the Voting Rights Act, which required jurisdictions with a history of racial discrimination at the polls to clear their voting changes with the federal government before those changes went into effect.

The number of voters removed from the rolls between 2016 and 2018 was similar to the number removed between 2014 and 2016, researchers found. But between 2006 and 2008, prior to the Supreme Court decision, roughly 4 million fewer voters were removed from the rolls.

The Brennan Center researchers also found that jurisdictions previously subject to Section 5 removed voters from the rolls at a median rate of 40% higher than other jurisdictions. If those places had removed voters at the same rate as other places, about 1.1 million fewer people would have been removed from the rolls.

The Supreme Court argued in 2013 that racial discrimination in those areas had abated and that the Voting Rights Act formula was therefore outdated, but the higher removal rate suggests past discrimination still has a stickiness in those jurisdictions, said Myrna Prez, director of voting rights and elections programs at the Brennan Center.

There is something to their shared history that made sense for them to be covered back then that is still durable, that still exists, Prez said in an interview. Its not random, its not haphazard. Its not happenstance. Theres something to their history and their institutional and structural barriers that [they] still have that today.

Voters may be removed from the rolls for a number of legitimate reasons, such as death or new addresses. The federal data the Brennan Center used for its analysis didnt make it clear how many removals were accurate.

Were not in a position to know how many of them were legitimate, how many of them we would contest, Prez said. The Brennan Center researchers also noted that not every jurisdiction had reported its voter removals, but said they were working off the best available data.

Federal law sets out specific procedures that states have to follow if they want to remove someone for the rolls. The law explicitly prohibits a state from removing someone from the rolls solely because they dont vote. But last year, the U.S. Supreme Court approved an Ohio process in which people were removed if they didnt vote for two years, didnt respond to a voter confirmation notice and then subsequently didnt vote again for another four years. The plaintiffs in the suit argued that Ohio was essentially removing voters because they werent voting.

Before the Shelby decision, places covered by Section 5 that wanted to change their processes for removing voters from the rolls would have to get that change cleared by the federal government. Under that supervision, the counties had voter removal rates that were in line with the rest of the country, the Brennan Center analysis notes. But since 2014, after the Supreme Courts decision, there has been a clear gap in removal rates.

This is of particular interest because this continued and even widening gap debunks possible claims that certain states would experience a one-time jump when free of federal oversight, but then return to rates in line with the rest of the country, Kevin Morris, a Brennan Center researcher, wrote in the report. They havent.