Last summer, Marc Jerome, president of Monroe College in New York, noticed something peculiar about the way the Education Department tracked student loan repayment rates. The measure of students who repaid at least a dollar toward their principal balance is a critical component of an employment regulation, a debt forgiveness rule and the government’s College Scorecard, yet the calculations differ.

It mattered to Jerome, in part, because the private for-profit college that he runs is subject to all three, and he had been advocating on repayment rates since 2014. And though the Scorecard more than doubled the repayment rates at Monroe, he alerted education officials.

That was August. This month, the Education Department said it corrected a “coding error” in the Scorecard that undercounted people who did not pay down the balance on their undergraduate student loans. As a result, the repayment rates for most colleges and universities were inflated.

“After discovering the coding error, the department worked to get accurate, refreshed data out as soon as possible, not waiting until the next annual Scorecard update to do so,” said Lynn Mahaffie, a senior official in the Education Department, in statement identifying the mistake.

Jerome said he found that statement a bit incredulous, considering that he told the agency something was up with the data months earlier. In a letter to the department reviewed by The Washington Post, the college president questions whether the Scorecard data are accurate, citing research to the contrary from the Brookings Institution. He also points out large variations in the three-, five- and seven-year repayment rates listed on the Scorecard. Instead of rates improving each year as borrowers’ earnings increase, they went down and then back up, he wrote.

The letter was submitted as a comment in response to the proposed overhaul of borrower defense to repayment, a statute that wipes away federal loans if a school used illegal or deceptive tactics to persuade students to borrow money for college. One of the requirements of the proposed rule called on for-profit colleges with poor repayment outcomes to warn current and prospective students, a stipulation that Jerome argued should be extended to all schools and should at least be based on accurate information, not the flawed data from the Scorecard.

“There’s a counterintuitive anomaly with the data that needs closer inspection and verification by the Department before it implements penalties associated with a five-year repayment rate,” Jerome wrote, in a comment letter dated Aug. 1. He said he followed up with the department in October to no avail. When the final rule came out in late October, it used a repayment metric that still largely aligned with the Scorecard.

In an interview, Jerome said he suspects the department made errors in the data “in its rush to push out a well-intentioned but ill-conceived rule before time ran out on the administration.”

He added: “The corrected data makes clear that the problem of repayment rates is universal, with less than half of all borrowers paying down principal.”

Education Department officials declined to comment for this article.

Some say the error speaks to the growing pains of the Obama administration’s college search tool. The Scorecard is an invaluable resource for families researching colleges and loan repayment rates, according to the department, are an important measure of quality. But for months, anyone comparing schools got a distorted picture of student outcomes, and that distortion was pretty significant.

Robert Kelchen, an assistant professor of higher education at Seton Hall University, found a roughly 20-point decline in the overall national repayment rate three years out, a 14 percent drop at the five-year mark and a 9 percent decline seven years out.

“This error is significant, because the College Scorecard received millions of page views with incorrect information for more than a year,” Kelchen said. “The Department of Education should have had procedures in place to dig into a few colleges’ data to check for larger errors.”

In announcing the error, Mahaffie, at the Education Department, said the agency has added a number of quality assurances and reran some of its prior tests to ensure the new rates are correct.

Want to read more about college outcomes? Check out:

Colleges are using big data to identify when students are likely to flame out

This chart tells a fascinating story about higher education