Skip to main content

How Reliable Is Accomplice Testimony Like Michael Cohen's?

Don't take it without corroboration.
Michael Cohen, former attorney President Donald Trump, testifies before the House Committee on Oversight and Reform on Capitol Hill on February 27th, 2019, in Washington, D.C.

Michael Cohen, former attorney President Donald Trump, testifies before the House Committee on Oversight and Reform on Capitol Hill on February 27th, 2019, in Washington, D.C.

Michael Cohen, President Donald Trump's former lawyer and aide, testified before the House of Representatives' Committee on Oversight and Reform on Wednesday, saying that Trump broke the law in his efforts to win the presidency. For example, Cohen said Trump directed him to lie about payments Cohen made to the adult-film actress known as Stormy Daniels, to keep her quiet about an alleged extramarital affair with Trump. To defend the president, Republicans on the committee cast doubt on Cohen's credibility, underscoring that he has already pleaded guilty to lying to Congress and will be going to prison for it. Trump drew on the same arguments in a tweet early Wednesday:

So how reliable is testimony like Cohen's, where an accomplice to a crime offers information on his former boss or partner?

Lawyers in the United States have long used accomplice testimony to build their cases, so legal scholars have closely examined it. It has its weaknesses, as a 1987 Cornell Law Review article lays out:

Accomplice plea agreements tend to produce unreliable testimony because they create an incentive for the accomplice to shift blame to the defendant or other co-conspirators. Further, an accomplice may wish to please the prosecutor to ensure lenient prosecution in his own case. Nevertheless, courts have held that the testimony of accomplices who receive lenient treatment is not per se unreliable.

Exactly how weak is accomplice testimony? I didn't find any quantitative experiments testing that question, but there is data suggesting "jailhouse snitches"—prisoners already serving sentences for other crimes—are frequently, and dangerously, wrong. The Innocence Project has found that jailhouse snitch testimony was central in 15 percent of cases that were found to be wrongful as a result of DNA evidence. Jailhouse snitches are considered even less reliable than accomplices because they have nothing to lose. They typically tell prosecutors they overheard someone confessing to a misdeed, in exchange for a lighter punishment. An accomplice, at least, faces the risk of implicating himself.

In addition, so few cases are brought for perjury that it's not a big deterrent keeping people from lying in court, as a 1990 Yale Law Review article argues. Presumably that applies to lying to Congress too.

One way to bolster the reliability of accomplice testimony is to require that it be corroborated with other evidence before it's used to convict anyone. California, Texas, and New York all have laws requiring such corroboration.

Of course, this hearing is not a court trial. In his prepared opening statement, the committee chair, Representative Elijah E. Cummings (D-Maryland), framed the hearing as a chance for the public to hear Cohen's story, for all its potential flaws. The core reason that accomplice testimony is unreliable and needs corroboration is that those who agree to testify for prosecutors get a clear benefit in return: some protections or a lighter sentence. What Cohen gets out of talking to Congress now is a bit murkier. It is a chance to influence public opinion about him. Might that ultimately affect his legal fate? In his testimony, he said he's seeking a reduced sentence from the Southern District of New York, though he also said that's not why he's cooperating. Instead, he argued he has nothing left to lose.

Finally, will knowing about the nature of accomplice testimony make a difference to how the public evaluates Cohen's credibility? One recent study suggests not. In the study, college students all read the same case (a real one, with details and names changed to fit the study and to ensure students' knowledge of the actual case wouldn't affect their decisions). Some students also read expert testimony about how unreliable accomplice testimony can be, while some did not. Both groups were equally likely to vote the defendant guilty.