HARTFORD, Conn. (AP) — Hundreds of Connecticut state police troopers falsified information on at least 26,000 traffic stops from 2014 to 2021, skewing reports on the race and ethnicity of pulled-over motorists, according to an audit released Wednesday.

Data analysts with the University of Connecticut said the reports resulted in too many drivers being identified as white. They cautioned, however, that they did not try to determine whether the records were intentionally falsified or were wrong due to carelessness or human error.

Gov. Ned Lamont said he has referred the matter to the chief state’s attorney’s office for investigation, and urged the public to not jump to conclusions.

“There’s no indication that was purposeful,” he said at an unrelated news conference Wednesday. “A lot of it may have been inadvertent.”

The audit was spurred by a Hearst Connecticut Media report last year that said four state troopers in an eastern Connecticut barracks intentionally creating hundreds of bogus traffic stop tickets to boost their productivity numbers. After internal affairs investigations, one trooper was suspended for 10 days, another was suspended for two days and the other two retired before the probe was completed.

The audit found that the number of traffic infractions reported to the Connecticut Racial Profiling Prohibition Project advisory board didn’t match those reported to the state court system, which handles all traffic citations, according to analysts with the Institute for Municipal and Regional Policy at the University of Connecticut.

The analysts “have a high level of confidence that at least 25,966 infraction records were falsified and submitted to the racial profiling systems,” the report said. The audit said the number of falsified infraction records could be as high as 58,553, if certain criteria were included. Analysts reviewed more than 800,000 infractions issues over the seven-year period.

“We are talking about ... a pattern of having records where you cannot find a corresponding record in the court system,” said Ken Barone, one of the UConn analysts. “If you claimed you stopped a car and issued a ticket, there should be a ticket.”

Lamont and the analysts noted that the number of discrepancies between state police data and court system data has decreased in recent years. The report said the most discrepancies were in 2014.

The analysts said they audited traffic stop information submitted by about 1,300 troopers, and 311 of them had a “statistically significant number of discrepancies” during at least one year.

The American Civil Liberties Union of Connecticut called the audit’s findings troubling.

“This audit reveals a breathtaking disrespect for the state’s racial profiling prohibition law by Connecticut State Police employees and, even worse, for that law’s goal of reducing systemic racism in policing,” said Claudine Constant, the group’s public policy and advocacy director.

“Whether intentional or not, the impact of police falsifying and inaccurately reporting records is the same: police have obscured the true information about how often they stop drivers of color compared to white drivers,” she said in a statement.

State police said in a statement Wednesday that the agency has no tolerance for false reporting and it has been working with the UConn institute to prevent it.

“The State Police are deeply committed to ensuring the integrity of Connecticut’s racial profiling data and to maintaining public confidence in the essential public safety services our troopers provide each day,” the statement said.

Police statewide, including local departments, are mandated to submit traffic stop data to the state, under Connecticut’s 1999 law aimed at preventing racial profiling. The UConn institute analyzes the data and submits periodic reports, which have shown that officers disproportionately pulled over Black and Hispanic drivers compared to white motorists.

The institute said Wednesday that the integrity of those analyses was in question because the falsified records “were more likely to be reported as White drivers and less likely to be reported as Black or Hispanic drivers.”

The audit also found that some state police data was underreported to the racial profiling board, and those records were more likely to involve Hispanic drivers.

Falsified police data is nothing new.

In 2010, for example, several New York City officers faced internal charges based on allegations by a fellow officer that they manipulated crime stats. In Texas, state troopers were found to be inaccurately recording the race of minority drivers, according to a KXAN-TV report in 2015.