Getting guilty right
A new study shows how the innocent get put in jail – and how we can do better
One morning in 1989, a college student was getting ready for work in her fifth-floor Boston apartment when she heard an odd noise. She opened the door to see what it was. A young black man forced his way inside, holding a screwdriver to her neck. Over the next hour, he raped her. He ransacked the apartment, took a camera and some money, and finally left. She called the police and gave a detailed description of the attacker.
At first, the police had no leads. They showed the victim books of mug shots, but she identified no one. She helped a police sketch artist make a composite drawing. An officer happened to see the sketch and it reminded him of a man named Neil Miller. The victim then identified Miller’s photo in a photo array. Miller was charged and brought to trial. A Boston police crime lab analyst told the jury that blood tests on evidence from a rape kit, while not conclusive, were consistent with Miller being the attacker. The jury convicted Miller, and he was sentenced to 26 to 45 years in prison.
That was where he remained for years, until the Innocence Project became involved and requested that modern DNA tests be performed on the evidence in the rape kit. A laboratory compared the evidence to a sample from Miller, and concluded that Miller was not the rapist. The DNA implicated another man connected with two other rapes, and in 2000, Miller was freed. He had served almost 10 years in prison.
Stories like Miller’s have by now become sadly familiar. As DNA testing has become a standard part of criminal investigations, it is also becoming a powerful tool for testing whether the system got convictions right. And its results are eye-opening. In the past two decades, DNA tests have exonerated and helped free more than 260 innocent people. Eighty of those had been sentenced to life in prison, and 17 had been sentenced to death. Since DNA evidence is available only for a fraction of cases, we have no way of knowing how many other innocent people have been convicted. We can be sure the number is substantial.
Beyond the individual stories, there is something else these cases can tell us: where things are going wrong in the justice system. In each case, the trial record can be read like a kind of airline accident report, a set of clues that points to what sent an innocent person to jail and let the real culprit go free. To amass the first systematic record of how these false guilty verdicts were reached, I located more than 200 trial transcripts, together with plea hearings, police reports, and judicial decisions in the cases of the first 250 convicts freed by DNA tests from 1989 through early 2010. With the help of law student research assistants, I then combed through the voluminous records. The results suggest real cause for concern about the accuracy of types of evidence that our courts routinely rely upon. Time and again, certain things went wrong, starting in the early stages of a case and later shaping what happened in the courtroom: the way eyewitness testimony was elicited, the reliability of forensic evidence, the handling of interrogations, and the way the investigations themselves were conducted. These individual cases of injustice offer American law enforcement the chance to learn from its mistakes, much like the aviation industry does — to reduce the number of life-and-death mistakes that are the result of preventable human error.
The stakes could not be higher. Beyond the suffering that each wrongful conviction brings to an innocent person and his family, it also places burdens on victims and on society, leaving guilty people unpunished and undermining the legitimacy of our criminal justice system. The DNA tests that freed innocent people have also identified 112 of the actual culprits, dozens of whom had committed additional crimes before being brought to justice. A more careful survey of how cases can go wrong, combined with the new scientific insights emerging about the justice process — for instance, how psychology can improve eyewitness identification procedures — offers a rare opportunity to make things better. And with more than 1.6 million prisoners in the United States and more than 7 million under correctional supervision, even marginal improvements could result in untold thousands of unsolved crimes being solved, innocent people going free, and guilty people brought to justice.
Eyewitness evidence is pivotal in many criminal trials, and for centuries it has been at the heart of criminal justice. As US Supreme Court Justice William Brennan wrote, “[T]here is almost nothing more convincing than a live human being who takes the stand, points a finger at the defendant, and says ‘That’s the one!’ ”
Yet a look at the DNA exonerations shows that seemingly strong eyewitness identification was the most common type of evidence that led to these wrongful convictions. In Neil Miller’s case, for example, it was the victim who picked him out from a set of photos. When one examines the records, however, one can see that this identification was not nearly as simple as it seemed in court, and that the process itself played a troubling role in her selection of an innocent man. A month after the crime, a detective visited the victim’s apartment and brought Miller’s photograph in a set of photos. The victim said she wasn’t sure, but picked out two photos as possibilities, one of which was Miller. The detective testified at trial that she then told the victim, “if she had a first impression, that the best thing to do was go with her first impression.” Having picked Miller’s photograph first, the victim then identified him as the attacker. Two months later, she was given another set of photos that included a much more recent photo of Miller. His was the only photo repeated from the first set. She picked out Miller.
At that point, Neil Miller was arrested. His defense lawyer planned to raise with the judge the suggestive way that the photo arrays had been handled. However, just before the hearing was to take place, the prosecutor walked the victim past Miller in the hallway outside the courtroom. Even after being told her attacker might be in that hallway, she still was not sure. She thought he might be her attacker. She was not positive until she followed Miller into the courtroom, where it was obvious that he was the suspect in her case. She looked at him again, and said, “This is him.”
What the jury saw, after all of this, was a victim pointing confidently to Neil Miller in the courtroom and declaring that he had attacked her. And this sense of certainty about the wrong person is common: In about three-quarters of the wrongful convictions, an eyewitness had picked out the culprit. And almost without exception, those eyewitnesses testified they were certain they picked the right person. We now know that poor police procedures can actually generate false confidence in witnesses; social science research has vividly demonstrated how eyewitness memory is so malleable that well-intentioned remarks (“don’t worry” or “good job, you picked out the guy”) and even unintended physical cues can dramatically increase certainty. For example, at the trial of Anthony Powell, another Massachusetts man who was later freed by DNA tests, the victim recalled that after she picked out his photo in an array, the officer told her, to comfort her, “Don’t worry, you picked out the right person.”
Getting eyewitness testimony right is crucial: Each year as many as 80,000 eyewitnesses make identifications in criminal investigations. There are simple procedures that police can follow to protect against eyewitness errors. The most important is that police conduct identifications “double-blind,” meaning that the officer does not know which person is the suspect, and the witnesses is told that the officer does not know. Other key recommendations include making sure that the suspect does not stand out in the photo array, telling the eyewitness that the suspect may or may not be in the array, showing photographs one at a time to prevent “comparison shopping,” and documenting how certain the eyewitness is at the time of any identification. Some departments use those best practices. Yet most police still use outdated and error-prone procedures.
Confessions, too, have long been seen as clear evidence of guilt. However, 40 of the first 250 DNA exonerees had actually confessed to their crimes. Many were vulnerable people, including mentally disabled persons and juveniles, and most were questioned by police using high-pressure interrogation tactics over many hours. Despite those red flags, the courts allowed their confessions to be used. Judges are very reluctant to suppress a confession, and may be particularly unwilling to suppress confessions that seem particularly accurate and rich in detail. Such confessions, though still false, can be devastating at trial because they are said to include information only the real culprit could have known.
Jeffrey Deskovic’s case provides an example. He was only 17 when police in Peekskill, N.Y., interrogated him about the rape and murder of a young woman from his high school. Police said he confessed in detail, drawing maps of the crime scene and saying he strangled the victim and hit her on the head with a Gatorade bottle. There was a problem, however: DNA testing done before trial excluded Deskovic. The prosecutor speculated that perhaps the victim was “sexually active” with someone shortly before her murder, which could explain the mismatch, but no one ever did anything to investigate that conjecture. The prosecutor focused on the fact that Deskovic gave key details in his confession which only the killer would know. Years later, new DNA tests excluded Deskovic again, and this time they also matched a convict who subsequently pleaded guilty.
Now that we know Deskovic was innocent, we also know that he could not have known the details that appeared in his alleged confession statements. He now says the police fed him the details while they questioned him. And his case is not unique: All but two of the exonerees who falsely confessed had confessions contaminated by such crime scene details.
Psychologists have studied the reasons why people falsely confess, including because they eventually seek to comply with police to simply end the interrogation. And police have long known that such people may repeat an account the police give them even though they are innocent, so trainers carefully instruct detectives not to disclose key details to a suspect. But in a long and intense interrogation, police can unfortunately lose track of who first said key facts, and it’s not hard to see how, under great pressure, fiction could start to replace the truth.
This kind of contamination can be prevented by recording entire interrogations, so it’s clear what a suspect has been told, and when. This reform is being adopted by some departments and states, but only gradually. Police who now record interrogations have become strong supporters of the idea, because it helps them to show that they handled the interrogation professionally and can cut down on motions to suppress confessions.
Perhaps surprisingly, another contributor to wrongful convictions is forensic science itself. Although it was forensic science, in the form of modern DNA testing, that exonerated the people in this study, I also found that the majority of the exonerees had invalid, unreliable, and erroneous forensic evidence presented at their trials.
In case after case, I found that analysts had exaggerated the evidence when they presented their conclusions to the jury. In some cases, they misstated statistics in a way that helped the prosecution. In many other cases, the trials relied on forensic methods that sound scientific but actually depend on subjective judgment, like microscopic hair, fiber, shoe print, or bite-mark comparisons. Despite the apparent precision of such tests, they have never been adequately researched or validated, and we do not know what frequency of the population shares any particular hair or tooth characteristics. Any analyst who said that a bite mark came from the defendant, or that there was a “high probability” that a hair was a match, crossed an important line, implying likelihood without sufficient data to back it up. It is no wonder the National Academy of Sciences concluded in its landmark 2009 report that forensic practice requires systemic reforms. Senator Patrick Leahy has introduced federal reform legislation, which would be a crucial step forward. In the meantime, judges need to carefully scrutinize forensics in their courtrooms.
We know that the innocent people freed by DNA evidence are only the tip of an iceberg. The majority of criminal cases don’t involve any solid biological evidence at all, and even when it exists, convicts seeking later DNA testing often find the evidence was destroyed after trial. Massachusetts, like many states, still has no statute to ensure such evidence is preserved for later testing. Nor have states universally adopted the kinds of reforms that can prevent future wrongful convictions and improve the accuracy of criminal justice, such as improved eyewitness identification procedures, mandatory recording of entire interrogations, or scientific oversight of crime laboratories. In our fragmented criminal justice system, each jurisdiction can go its own way. Change occurs piecemeal.
Neil Miller’s case, however, shows how change can happen. Miller could not be given back the years he lost, but he received $3.2 million in compensation for his wrongful conviction. And in response to his case and others, Suffolk County has changed the way it handles eyewitness testimony to conform with the current best practices (although the state still has not).
The costs of these wrongful convictions are now clear, as are the benefits of adopting reforms. A decade ago, only a handful of states had relaxed the strict rules that limited convicts’ ability to reopen their cases for new DNA testing, and now all but two states have done so. A decade ago, few police departments videotaped interrogations, and now over 750 do so. It took years to exonerate these innocent people and to understand what these cases can teach us about how to make our criminal justice system more accurate. Slowly but surely in reaction to these troubling cases, we are beginning to see the stirrings of a criminal procedure revolution.
Brandon L. Garrett is a professor of law at the University of Virginia School of Law. His new book, ”Convicting the Innocent: Where Criminal Prosecutions Go Wrong,” was published by Harvard University Press.