Beyond Protecting the Innocent

When a criminal trial goes wrong, either an innocent person is convicted, or a guilty person goes free. Under the U.S. Constitution, the first error has long been understood to be far worse than the second. The Supreme Court has often quoted the English jurist Sir William Blackstone’s expression that it is better for ten guilty people to escape punishment than for one innocent to be condemned. The presumption of innocence, the requirement of proof beyond a reasonable doubt, and other due process protections all aim to prevent wrongful convictions of the innocent.

But when researchers have asked the public how they feel about these two mistakes, they learned something unexpected: most Americans say both are equally bad.

This puzzle sits at the heart of a research project underway at the Wilson Center for Science and Justice. Duke Law professor and faculty director Brandon Garrett and his longtime collaborator, UVA Law professor and psychologist Gregory Mitchell, along with Wilson Center research scientist Jesse Lopez, and postdoctoral fellow Jessica Gettleman, recently completed a new article, forthcoming in the Vanderbilt Law Review,  “Aversions to Criminal Justice Errors and their Consequences.”

The research focuses on how people weigh these two types of mistakes in the criminal legal system: convicting an innocent person, or letting a guilty one go free. Researchers call this "error aversion,” but put more simply, it measures which mistake people would rather avoid making. As Lopez put it, "The concept reflects a deeper intuition about which risks people are willing to tolerate, and which consequences they find most unacceptable."

In prior work summarizing surveys of more than 12,000 Americans, Garrett and Mitchell found that 63% of respondents viewed the two errors as equally harmful. The pattern held across the political spectrum: majorities of Democrats, Republicans, and Independents alike rated both errors as equally harmful, challenging the assumption that conservatives are more concerned about letting the guilty go free while liberals are more concerned about convicting the . It also held across types of cases. When asked about crimes ranging from rape and vehicular manslaughter to misdemeanor DUI, 50% or more of respondents continued to view avoiding both mistakes as equally important.

What makes this surprising is not that people gravitated toward a middle option. People often do. What makes it surprising is the context. Lopez explained that the criminal legal system is built on the idea of preventing wrongful convictions, so you would expect more people to prioritize that error. Instead, many seem to view the system as a balancing act for justice, even though it was never designed that way.

The team’s earlier research also showed that these attitudes carry real consequences. Mock jurors who saw both mistakes as equally harmful, or who saw false acquittals as worse, were more likely to convict and be more receptive to prosecution arguments. Error aversion is not just an abstract attitude, it shapes behavior that determines life outcomes.

The new project pushes this work further, drawing on data from Time Sharing Experiments for the Social Sciences (TESS), a nationally representative sample of Americans, which allowed the team to ask two new questions.

First, they examined whether anything can break the link between someone's pre-existing attitudes and their willingness to convict. In one study, participants read a burglary trial scenario in which the defense attorney either makes a standard argument or frames the case by invoking Blackstone directly, reminding jurors that it is better for ten guilty people to go free than one innocent to suffer. The goal was to find out if this kind of reminder might shift the jurors who came in primed to convict. So far, it has not. Whether a different framing might succeed remains an open question.

Second, they studied error aversions beyond the courtroom, recognizing that human judgment shapes outcomes long before a trial begins. A person must choose whether to call 911 about a stranger fiddling with a doorknob at night, and a witness must decide whether to identify a suspect from a lineup. Each of these moments carries its own version of the same dilemma, weighing the risk of wrongly drawing an innocent person into the legal system against the risk of letting something real go unaddressed. The team found that people's attitudes about errors shaped whether they would make an eyewitness identification. But the earlier decision—whether to call the police at all—seemed to operate on instinct, largely independent of those attitudes.

What are the implications? For Lopez, one takeaway is that: “we need to reach people with a clearer understanding of how most people actually think about error and risk.” And Garrett adds that “we cannot just make arguments about fairness, although people deeply value due process. We need to understand that fairness and public safety cannot exist without the other.” The authors pointed out that when a wrongful conviction occurs, and the crime was real, the guilty person also went free by definition. That type of error harms both public safety and due process—and making that type of error clear to the public is an important challenge going forward.