UC Berkeley Press Release
Researchers debunk conventional wisdom on trial witnesses
BERKELEY – A new study authored in part by a University of California, Berkeley, professor of public policy and law throws cold water on a common theory that a confident witness who errs in trial testimony is still more credible than a less confident witness who similarly slips up.
The researchers concluded that self-assured witnesses who make a mistake - even on issues of little importance - undermine their credibility by raising doubts about their competency, their ability to judge their own abilities and their motivations.
"People giving testimony, or advice, or opinions should therefore be careful to express appropriate degrees of confidence in their assertions," the researchers write in a summary of their report in the January issue of the journal Psychological Science. "Otherwise, the 13th stroke of the clock will cast the other 12 in doubt."
The researchers included Robert J. MacCoun of UC Berkeley, a professor at the Goldman School of Public Policy and at the School of Law (Boalt Hall); Elizabeth Tenney, a doctoral student in psychology at the University of Virginia; Barbara Spellman, an associate professor of psychology at the University of Virginia; and Reid Hastie, a professor of behavioral science at the University of Chicago.
MacCoun said the team's findings challenge the frequent tendency of attorneys to pressure their witnesses to project a strong sense of confidence and to minimize the use of hedges like "I think" or "maybe." Academic experts encounter similar pressures when asked to testify before policy makers, he said. But this first-of-its-kind study shows that such a strategy can backfire if a cocky witness gets caught in a mistake.
MacCoun and his colleagues' research is based on two mock legal judgment experiments.
In the first, 48 undergraduates read various versions of a summary of the criminal trial of someone charged with breaking and entering. The summaries included testimony by a confident witness who insisted that he saw the defendant leaving the crime scene carrying stereo gear at 7:00, while the victim testified that he was home until 8:15.
In another version, a not-so-confident witness said he saw the suspect with the stolen goods at 7:00, but then admitted uncertainty about the time when confronted with the victim's account. Other accounts contained no apparent error by either the confident witness or the one who was not.
When no mistake was made in the trials with confident witnesses, students were more likely to support convicting the suspect. But when an error was made, the preference for conviction was higher when the witness was not so sure.
In their second test, the researchers asked 103 undergraduates to act as jurors for a purported civil trial involving a car accident and witnesses - one confident and one not so confident - who concluded that different vehicles caused the mishap. The mock jurors learned that each witness was correct in describing weather conditions at the time of the accident, but were wrong about what they were doing just before they saw the wreck.
The mistakes caused the credibility rating of each witness to drop, but the confident witness suffered a bigger decline, researchers found.
MacCoun said these findings have been borne out with two newer studies not included in the journal article.
"This comes up all the time with talk shows, where experts like Richard Perl (a federal policy advisor who supported invading Iraq with a minimal number of troops) are so brash in their assertions, and so wrong," MacCoun said. "Then, suddenly, you don't see them (on TV) any more."
He noted complementary work by UC Berkeley's Philip Tetlock, a professor of organizational behavior, psychology and political science and the author of "Expert Political Judgment: How Good Is It? How Can We Know?" (2005). Tetlock concluded in the highly-regarded book that predictions by so-called experts are frequently flat-out wrong, but that they rarely acknowledge the fallibility of their judgments.