WASHINGTON — Former President Donald Trump’s 2020 campaign commissioned an outside research firm in a bid to prove electoral fraud claims but never released the findings because the firm disputed many of his theories and could not offer any proof that he was the rightful winner of the election, according to four people familiar with the matter.

The campaign paid researchers from Berkeley Research Group, the people said, to study 2020 election results in six states, looking for fraud and irregularities to highlight in public and in the courts. Among the areas examined were voter machine malfunctions, instances of dead people voting and any evidence that could help Trump show he won, the people said. None of the findings were presented to the public or in court.

About a dozen people at the firm worked on the report, including econometricians, who use statistics to model and predict outcomes, the people said. The work was carried out in the final weeks of 2020, before the Jan. 6 riot of Trump supporters at the U.S. Capitol.

Trump continues to falsely assert that the 2020 election was stolen despite abundant evidence to the contrary, much of which had been provided to him or was publicly available before the Capitol assault. The Trump campaign’s commissioning of its own report to study the then-president’s fraud claims has not been previously reported.

“They looked at everything: change of addresses, illegal immigrants, ballot harvesting, people voting twice, machines being tampered with, ballots that were sent to vacant addresses that were returned and voted,” said a person familiar with the work who, like others, spoke on the condition of anonymity to describe private research and meetings. “Literally anything you could think of. Voter turnout anomalies, date of birth anomalies, whether dead people voted. If there was anything under the sun that could be thought of, they looked at it.”

The findings were not what the Trump campaign had been hoping for, according to the four people. While the researchers believed there were voting anomalies and unusual data patterns in a few states, along with some instances in which laws may have been skirted, they did not believe the anomalies were significant enough to make a difference in who won the election.

The research also contradicted some of Trump’s more conspiratorial theories, such as his baseless allegations about Dominion voting machines supposedly being used to rig the results and large numbers of dead people voting.

A person familiar with the findings said there were at least a dozen hypotheses that Trump’s team wanted tested.

“None of these were significant enough,” this person said. “Just like any election, there are always errors, omissions and irregularities. It was nowhere close enough to what they wanted to prove, and it actually went in both directions.”

Senior officials from Berkeley Research Group briefed Trump, former chief of staff Mark Meadows and others on the findings in a December 2020 conference call, people familiar with the matter said. Meadows showed skepticism of the findings and continued to maintain that Trump won. Trump also continued to say he won the election. The call grew contentious, people with knowledge of the meeting said.

The research group’s officials maintained privately that they did not come into the research with any predetermined conclusions and simply wanted to examine the data provided by the Trump campaign in the battleground states.

Through a spokesman, Meadows declined to comment.

“President Trump received a record-breaking 74 million votes, the most of any sitting president in the history of the country. Anyone who takes a look at Joe Biden glitching through his presidency knows who really won the election,” Trump spokesman Steven Cheung said. Biden won 81 million votes and won the electoral college, 306 to 232.

Cheung did not answer a question about Trump’s reaction to the researchers’ findings.

“Our experts provide independent and objective factual analysis and as a matter of firm policy, we do not comment on client engagements or on privileged and confidential matters,” a spokesperson for Berkeley Research Group said.

The findings from Berkeley were among the many streams of information after the election that showed Trump he lost. According to testimony presented to the Jan. 6 committee, Trump was repeatedly told by advisers that he did not win the election but continued to cast about for others who would entertain his theories and say that he had won. Dozens of judges – including many Trump appointees – rejected his campaign’s attempts to challenge election results in court.

Trump has continued to spread false claims that he won the election, frustrating some of his advisers who wish he would move to a forward-looking message as part of his 2024 bid to reclaim the presidency.

The Berkeley research came about, according to people familiar with the matter, after Trump as well as some of his advisers became convinced the election was stolen. Others on his team wanted a sober analysis of what they could say and prove, some of the people said. Some of Trump’s advisers even hoped that a definitive report from Berkeley Research Group might tamp down some of the false claims.

“The goal was to find out what actually happened,” one of these people said. “If you remember in that time, there were all sorts of crazy things being said. We wanted to sort it out.”

The Berkeley research was done through a subsidiary company called East Bay Dispute and Advisory. FEC filings show the Trump campaign paid East Bay Dispute and Advisory more than $600,000 in the final weeks of 2020. A person familiar with the matter said there were also other researchers commissioned to help prove electoral fraud from outside Berkeley Research Group. The payments were described as consulting fees.

The states studied by the analysts over a period of several weeks included Georgia, Pennsylvania, Michigan, Wisconsin, Arizona and Nevada, according to people familiar with the matter. All but Nevada had been won by Trump in 2016 but flipped to the Democratic nominee four years later.

The Washington Post has not reviewed a copy of the report, but three people familiar with its contents described the findings.

Those who worked on the report included Janet Thornton, who has about 40 years of experience in accounting and investigations, according to Berkeley Research Group’s website. Others included Craig Freeman and John Auerbach, the people said. Their professional biographies describe decades of experience in accounting, investigating corporate fraud and handling other complicated inquiries.

Auerbach, who is now with a different firm, declined to comment. Freeman, who left Berkeley as well, and Thornton did not respond to requests for comment.