A University of Mississippi analysis found that the likelihood of a false confession grows with the amount of time individuals spend in the interrogation room. Researchers warned that this study is only a predictive model, and that no single study should influence regulation or policy around policing. Graphic by Clara Turnage/University Marketing and Communications
Criminal justice professors use statistical modeling to examine wrongful-conviction risk.
by Clara Turnage
A University of Mississippi study is adding nuance to a debate over how often police interrogation tactics can produce false confessions leading to wrongful convictions.
In a recent simulation study published in the Journal of Criminal Justice, Ole Miss criminal justice and legal studies professors Mary Catlin and Thomas Bryan Smith found that the predicted likelihood of a false confession grows with the length of a police interrogation.
“When it comes to false confessions, it is acknowledged that they are probably a relatively rare occurrence,” Catlin said. “Generally speaking, if the police get a confession, it is from a guilty person. Our concern is that when it goes wrong – even if that is less common – there is a massive impact.
“We’re talking about ruining that person’s life, while potentially the person who actually committed this crime is still out there.”
Recent studies indicate that wrongful convictions account for some 4% of all capital cases in the United States, and people who are wrongly convicted spend an average of eight years imprisoned before being released or exonerated. Many reasons exist for wrongful conviction, but the National Registry of Exonerations reports that 13% are tied to false confessions.
A study published earlier this year estimated that police interrogation tactics during interrogations were associated with false-confession wrongful convictions in about 1% of cases.
“I had a few concerns with the analysis, primarily that the vast majority of cases involved may only be short, routine interrogations that do not employ many interrogation tactics,” Smith said. “Imagine if you take every interrogation and throw it into a single bucket, guilty confessions from common short interrogations will outnumber false confessions from rare long interrogations.
“So we ask: What if you arrange interrogations into separate buckets based on the length of the interrogation?”
The researchers used data from previous studies to run thousands of simulations, allowing them to approximate the likelihood of a false confession-wrongful conviction at different interrogation lengths.
While it was true that at typical interrogation lengths of about 1.5 hours, the risk of false confession wrongful conviction is only about 1%, by 12 hours, that risk could reach 25%. By 34 hours, the risk might be as high as 50%.
Interrogations, particularly in more serious crimes, can run that long. Studies show interrogations that yield false confessions average 16 hours, with some interrogations running longer than a full day.
“The exonerations that we see are typically from very prolonged interrogations,” Catlin said. “And duration is really only one part of this. You have to account for stress, for the mental health of the individual, social persuasion techniques and so many other factors.
“Many exonerees tell us they confessed because they thought that meant they would be able to go home.”
The researchers stressed that both studies represent modeled probabilities, not observed rates.
Smith and Catlin’s goal was not to prove the validity of their statistical model over the previous publication. Rather, they wanted to explain that with the available data, estimates of false confession are inexact, and no single simulation should be used to inform policy or policing, the researchers said.
“We are arguing that these simulation studies are too preliminary to be used to inform decisions in individual cases,” Smith said. “This study should serve as a guidepost for what future research needs to do more so than what current policy needs to do.
“It’s a simulation, and while it can be informative, it is still only a simulation. That’s why we argue that this research shouldn’t be informing policy or be used in the courtroom.”
Instead, the researchers point toward decades of studies that suggest evidence-based approaches to interrogation yield more accurate results than accusatorial tactics.
“We know, based on a long history of science, that there are tactics we can identify, approaches we can identify that reduce the risk of false confession, while in some cases increasing the likelihood you’ll get a true confession,” Catlin said.
“If we know those exist – and they’ve been adopted in some places – why would we not use them?”
About the Author
