When the Supreme Court badly weakened the Voting Rights Act in 2013, it described the landmark civil rights law as outdated. The formula that Congress had used back in 1965 to decide which areas of the country should have their voting laws placed under federal supervision no longer matched modern patterns of discrimination, Chief Justice John Roberts claimed.

“If Congress had started from scratch in 2006, it plainly could not have enacted the present coverage formula,” Roberts wrote for the majority, explaining why that formula was being struck down.

But a comprehensive new study by a renowned historian and expert in voting discrimination suggests what voting rights advocates have been saying all along: that Roberts got it wrong.

Since 2009, Morgan Kousser, a professor of history and social science at the California Institute of Technology, has been collecting data on voting rights “events” across the country that resulted in a win for minorities. Kousser looked at over 4,100 voting rights lawsuits, inquiries by the Justice Department, and changes to laws in response to the threat of lawsuits, the great majority at the county level, stretching from 1957 to 2013.

Kousser found that over 90% of these events occurred in “covered jurisdictions”—that is, places that, until the Supreme Court’s decision, were required to submit their voting changes to the federal government for “pre-clearance.” The great majority of these jurisdictions were in the South.

The evidence, Kousser writes, “shows that the Chief Justice’s factual assertions were incorrect, that the coverage formula was still congruent with proven violations.”