Wrongful convictions reported for 6 percent of crimes

A study from Penn criminologists results in the first general estimate for the prison population as a whole

Criminologist Charles Loeffler
Criminologist Charles Loeffler led a team of researchers that looked at wrongful convictions in the prison population as a whole. The results represent the first such estimate for crimes across the spectrum, from retail theft to murder.  

DNA-based exonerations for wrongful convictions tend to make news. New evidence comes to light that wasn’t available during the original trial, and someone previously found guilty of a crime is suddenly cleared.

For capital crimes such as murder and rape, this happens in approximately 3 to 5 percent of cases, but Penn criminologist Charles Loeffler realized there was no such estimate for other crimes, those ranging from serious charges like armed robbery and aggravated assault to more minor crimes such as theft and drug possession. So he shaped a study with Jordan Hyatt, now at Drexel University but formerly a research associate at Penn, and Penn criminologist Greg Ridgeway.  

Surveying an intake population of nearly 3,000 state prisoners in Pennsylvania, the researchers found that 6 percent reported being wrongfully convicted, results the team published in the Journal of Quantitative Criminology in April. This is one of the first estimates of its kind for the criminal-justice population as a whole.

“We view this as an upper-bound estimate. In other words, if people were going to be inaccurate, that would lead us to a true rate that’s lower than, rather than higher than, 6 percent,” Loeffler says. “Before we did this study, a reasonable response to the question of how many people are wrongfully convicted would be, ‘We don’t know.’ Through this work, we’ve reduced that uncertainty dramatically.”

There are two approaches to conducting research like this. One asks prosecutors, judges, and others working in the criminal-justice system for their opinion on this rate. On the one hand, the results aren’t biased from respondent self-interest. On the other, they’re difficult to interpret and generalize; does the percentage someone offers include cases he himself tried, those he heard about, those he spoke about with colleagues, or some combination of those options?

Loeffler, Ridgeway, and Hyatt opted to take a second tack: asking the prisoners themselves. According to the researchers, prisoners and other convicted people know their own involvement in their conviction offenses and can presumably relay that for cases ranging from retail theft to murder.

“It’s hard to find another person who is as well-equipped to provide you that information,” Loeffler says. “The problem is, of course, do we have reason to believe that these individuals are willing to cooperate and likely to report honestly to us what happened?”

Much of the fear that the inmates wouldn’t participate or provide accurate accounts dissipated at the study’s outset, when two-thirds took complete responsibility for their most recent crimes. Another quarter took at least partial responsibility. Just 8 percent maintained that they had zero involvement in the most recent crime for which they were convicted, what Loeffler describes as “factual innocence.”

Previous experience using surveys for this type of work told Loeffler and colleagues that the more they could convince participants their answers would be untraceable to a specific individual, the more forthright participants would likely be. During a six-month period, the researchers administered surveys that the prisoners took in a group setting but anonymously.

Questions ranged from those about non-sensitive information like approximate age to more delicate topics like recent crimes and feelings about the conviction’s accuracy. Those who said they had been wrongfully convicted were asked to explain further via justifications that ranged from “I didn’t do any of the things they convicted me of” to “I did what they convicted me of, but I didn’t mean to do it.” From there, the criminologists assessed the accuracy of the responses by comparing the data to administrative statistics available from the prison itself, then statistically adjusted for any respondents they deemed had provided implausible responses.  

“When we looked at the information reported from our respondents, as well as from the correctional system, they lined up very well, which is good because you don’t know whether that’s going to be the case,” Loeffler says. “We also had a very high response rate. One of the concerns you have with sensitive questions is not that individuals will tell you something that isn’t true, but that they’re simply not going to respond at all.”  

Beyond estimating a quantity that has eluded the field for decades, Loeffler says that having this number for wrongful convictions and the data that lead to it could offer the criminal-justice system an opportunity to understand which case factors predict reported wrongful convictions and how these compare to what’s at play in wrongful convictions of capital crimes.

As for whether the rate is high or low, Loeffler sees both perspectives. “It’s a good question because it’s a hard question. Some people look at this and say, ‘Only 6 percent? Alright, it’s working pretty well,’” he says. “Other people look at it and say, “6 percent? That’s a lot of people.’ Both reactions are reasonable.”

For now, at least, it’s a starting point: The current study took place only in a prison in Pennsylvania, but to see whether the results hold for other populations, further administrations will be needed. The research team is currently analyzing results from a recently concluded replication of the survey.  

Charles Loeffler is the Jerry Lee Assistant Professor of Criminology in the criminology department in the School of Arts and Sciences. Greg Ridgeway is an associate professor of criminology and statistics in the criminology department in the School of Arts and Sciences.