Davontae Sanford was dressed in his pajamas, standing on the front porch of his house on Detroit’s east side on a September night in 2007, when he saw police searching the street with flashlights. A few hours earlier, two blocks away on Runyon Street, four people had been shot to death, and the cops were searching the neighborhood for clues. Sanford asked the officers what was going on; a few minutes later he was telling them he might know who the killers were.
Sanford, who was 14 and attended the special education class at his high school, was taken to the station and questioned for several hours without a parent or a lawyer present. (The officers told him no lawyer would come to advise him in the middle of the night.) Sanford spun some stories that made little sense and was released. But the next night, detectives brought him in again.
“It was very, very tense,” Sanford said by phone on Wednesday.
This time, interrogators said they knew he was guilty and had evidence to prove it—blood on his shoes. They told him he could go home to his mother if he would simply admit to the murders. Police and Sanford’s lawyers disagree on what happened in the interrogation room that night, and there’s no recording of the session. But he recalled this week that “they played good cop, bad cop. Some of them talked crazy.” And a few weeks after the interrogation he told a court psychologist—who was to determine if the boy was competent to waive his Miranda rights—“I kept saying, ‘I don’t know what happened.’ They said, ‘Tell me something.’ ” So he started making up stories. “I didn’t know what was going on,” he told the court psychologist. “I didn’t know I was making statements.” But police presented him with a confession and told him, “Sign it, sign it.” Finally Sanford did as he was told.
Most of the details in the signed confession contradicted the evidence, and there was no blood on his shoes—the cops had invented that. Nevertheless, the public defender assigned to Sanford’s case did not challenge the confession and instead encouraged him to plead guilty to second-degree murder. Sanford was sentenced to 37 to 90 years in prison.
Two weeks later, a hit man named Vincent Smothers was arrested for a different killing and told police that he and another man—not Sanford—had committed the Runyon Street murders as well. His confession, unlike the teenager’s, precisely matched the facts of the crime.
For the next eight years, a team of public interest lawyers worked to get Sanford out of prison. On June 7, 2016, Sanford’s conviction was vacated, and he finally went home to his mother.
To most of us, it’s inconceivable that someone would admit to a crime he hadn’t committed. But since 1989, two hundred twenty-eight people in the United States who made confessions have been exonerated, according to the University of Michigan Law School’s National Registry of Exonerations. Sanford was No. 228. In addition to those who were vindicated, scholars have identified four men who confessed but were most likely innocent and were executed. (The actual number of false confessions is much higher, as exoneration is extraordinarily difficult, even when DNA exculpates a false confessor.)
“The American style of interrogation is guilt-presumptive, accusatory, and confrontational,” said Richard Leo, a professor in law and social psychology at the University of San Francisco and an expert on false confession. The ploys that were used on Davontae Sanford, and remain routine practice, can be so manipulative and intimidating, Leo said, that they sometimes compel an innocent person to admit to something he hasn’t done.
Leo and other experts say that there’s a better way. Nearly 25 years ago, a team of British detectives and academics developed a new method of interviewing suspects. Officers trained in this method don’t insist on the suspect’s guilt; they don’t offer false promises. Instead, they encourage the suspect to tell his whole story, without interruption. Then they ask open-ended follow-up questions. They are looking not for a confession but for the truth.
The method is called PEACE (an acronym for the five steps involved—Planning and Preparation, Engage and Explain, Account, Closure, Evaluation), and it’s standard protocol in England and Wales, Australia, New Zealand, Norway, and on the Canadian island of Newfoundland. A review of scholarly research published last year in the Annual Review of Law and Social Science concludes that PEACE elicits far fewer false confessions than the accusatory approach and significantly improves the rate of true confessions.
Sgt. Det. John Brown, a 22-year veteran of the Boston Police Department, has conducted hundreds of interviews during his years in the gang and homicide units. Even apart from the question of false confession, he said that in his experience, the prevailing mode of conducting interrogations doesn’t fit police needs: “I don’t think it works for us if we want to get these cases done the right way. If you talk to people the right way and you use the techniques that are incorporated in PEACE, I think you’ll get what you are looking for.”
So why do police departments in the United States cling to the old ways?
In the early days of uniformed policing, the usual method of extracting a confession from a suspected criminal was to beat it out of him. But by the 1930s public criticism of the third degree led to a search for more enlightened procedures. The first police interrogation training manual, written in 1940 by a Berkeley, California, police lieutenant named W.R. Kidd, emphasized psychological strategies rather than brutality to persuade an uncooperative suspect. “We trap him into a position from which he sees no way out,” Kidd wrote. “He talks.”
Kidd’s approach, and many of the maneuvers he described, were reworked in a series of handbooks by Fred Inbau, a lawyer and early devotee of the polygraph, and a former Chicago cop named John Reid. Today, the interrogation manual they wrote with Joseph P. Buckley and Brian C. Jayne, Criminal Interrogation and Confessions, is the leading text in the field, and nearly every detective in the United States practices some version of what’s known as the Reid Technique.
The technique begins with the “behavior analysis interview,” a diagnostic tool that is supposed to separate the liars from the truth tellers, the guilty from the innocent. Liars will betray themselves, the theory goes, through verbal cues (qualifying phrases, mumbling, delayed responses) and nonverbal behaviors (lack of eye contact, brushing lint from clothing, inspecting fingernails).
Bill Proctor, an investigative and security consultant and adviser to the Sanford family, addresses the media on June 9, 2016.
While fidgeting may suggest nervousness, it’s impossible to say for sure that it indicates deception. Common sense suggests that anyone being grilled about a crime, innocent or guilty, might reasonably show signs of anxiety. The firm Reid opened to train police and sell his books, John E. Reid and Associates, cites a 2014 experiment in which expert government investigators achieved 98 percent accuracy in detecting lies. But the study was not designed to test the efficacy of the behavioral analysis interview or the diagnostic value of verbal and nonverbal cues, and the study’s lead author, Timothy Levine, a professor of communications at the University of Alabama at Birmingham, told TakePart that his results are an outlier; most research shows that trained investigators can identify a liar roughly half the time—no better than random guessing. He said he is “skeptical” that such cues can be a useful guide to deception.
The Reid training seems to trigger a kind of confirmation bias. According to Saul Kassin, a professor of psychology at John Jay College of Criminal Justice, and Christian Meissner, a professor of psychology at Iowa State University, specialized lie-detection training makes investigators more likely to believe a subject is untruthful. Like the hammer that sees only nails, once they learn to scrutinize people’s behavior for lies, they see evidence of lies everywhere. Kassin and Meissner’s research also suggests that although such training often doesn’t improve questioners’ accuracy, it makes them more confident that they’ve guessed correctly. That’s where the real interrogation begins.
The latest edition of the Reid textbook instructs detectives to start an interrogation by telling the suspect they know he’s guilty. Detectives are taught to “discourage” denials: If the suspect claims innocence, the interrogator should either ignore it or “reassert his confidence in the suspect’s guilt.”
The graphic above illustrates convictions in the U.S. resulting from false confessions that have been overturned since 1989. (Infographic: Courtesy Michigan Law School)
The detective might bluff, saying that a polygraph “proves” the suspect committed the crime or—as Detroit police told Devontae Sanford—that physical evidence confirms his presence at the scene, when no such evidence exists. (The Reid textbook notes that an interrogation by necessity operates on a “somewhat lower moral plane” than everyday discourse.) Deceiving suspects in this way is perfectly legal; the U.S. Supreme Court has declared that lying to a suspect does not in itself make a confession inadmissible.
Once the suspect feels cornered, the detective seems to offer a way out by suggesting a story that presents a moral (but not legal) defense—you lost your temper; you intended to repay that money; you were drunk; anyone can understand how that might happen. The detective might propose two scenarios: Did you plan this, or did you just lose control? Cory Armishaw, who was coerced into confessing that he’d shaken his girlfriend’s three-month-old son to death in 2006, recalled the choice he was given: “Either you’re a nice guy who just snapped, or you’re a baby killer,” he told CBC in 2012.
After many grueling hours of questioning, some people, including some innocent people, accept the less damning alternative. (Juveniles and the developmentally disabled are especially vulnerable to these tactics.) We’ve replaced physical coercion with psychological coercion.
“The innocent person who confesses knows they were not at the scene of the crime; they know it’s false,” said Timothy Moore, a professor of psychology at York University in Toronto who has studied interrogations and has served as an expert witness on false confession. But when an interrogator repeatedly rejects his denials, he may begin looking for a way simply to end the ordeal. “He may think, ‘I don’t have a chance, so I’m going to tell him whatever he wants to hear, and we’ll straighten it out in the morning,’ ” Moore said. “But it doesn’t work that way.”
Juries tend to treat a confession as nearly infallible. Once an innocent person confesses, the chances of “straightening it out” are very slim.
Police interviews in Newfoundland are carried out quite differently. Newfoundland is a windswept island in the North Atlantic covered in pine and birch forest. With a population of around half a million, it’s the more thickly settled portion of Canada’s easternmost province, Newfoundland and Labrador. In spring, icebergs sail down from the Arctic and settle along the island’s rocky shore.
Until recently, officers in the Royal Newfoundland Constabulary received little instruction in how to conduct an investigative interview. Some officers took a Reid course at the Canadian Police College on the mainland, but mostly they got by with a mixture of secondhand lessons and on-the-job training. There was no comprehensive philosophy and no scientific research guiding their work.
A decade ago, John House, then a sergeant in the RNC Criminal Investigation Division, and Brent Snook, a professor of psychology at Memorial University of Newfoundland in the provincial capital of St. John’s, set out to change that. House had learned about a new method of police interviewing called PEACE while pursuing a master’s degree in investigative psychology in England in the early 1990s. Snook, who was raised in a small town on Newfoundland’s south coast, also had encountered PEACE in England, where he received his Ph.D. When Snook returned to Newfoundland in 2004, he started researching psychologically based investigative practices that could have everyday, practical applications, such as interviewing. He and House began talking about bringing PEACE to Canada.
Then in 2006 a government inquiry into three wrongful convictions in Newfoundland revealed a tangle of investigative failures, among them “inappropriate” and “skewed” interviewing practices. A report on the cases by Antonio Lamer, the retired chief justice of Canada’s Supreme Court, provided ammunition for Snook and House’s reform efforts. When they approached the new chief of police about bringing PEACE to the RNC, he readily agreed. In 2009 Newfoundland became the first police department in North America to convert to PEACE, and it’s still the only jurisdiction on the continent to banish Reid and adopt an alternative grounded in scientific research.
Today, House explained, “the whole organization has been trained, from specialist child interviewers right through to frontline officers.” Every recruit takes a PEACE class designed by Snook as part of a yearlong police studies program at Memorial University. Officers who are promoted to an investigative unit take an additional two-week course. Every new constable is also required to take Snook’s introductory class in forensic psychology, which covers false confessions, the dangers of manipulative tactics, and the ethics of police work.
Suspects brought in to the RNC for questioning are not subjected to the kind of intimidating, guilt-presumptive interrogation that is typical in much of Canada, and in the United States, because suspects are questioned pretty much the same way as witnesses: Ask them to tell their story, don’t interrupt, and save any challenges until the end. If there’s physical or other evidence that implicates the person, that too is saved until the end and introduced one fact at a time (it’s more difficult to explain away evidence when it’s presented like this).
People think it’s soft and sort-of touchy-feely but it’s not. It’s very systematic. It requires that people do their research, and know their case very well.
John House, former superintendent, Royal Newfoundland Constabulary Criminal Investigation Division
A few miles from the busy harbor in downtown St. John’s, in a conference room in Memorial University’s sprawling science building, Snook cued up a DVD to show what a PEACE interview might look like. Not yet 40, Snook wears rectangular glasses and stylish shoes. The video, he explained, shows a newly trained constable trying out his skills on a volunteer who claims he was not involved in an (imaginary) assault the night before. The task of the trainee might be summarized this way: Presume nothing, and listen.
The PEACE interviewer asks the volunteer to tell his story—What did you do last night?—from start to finish. The trainee has been taught to “transfer control” to the subject, so he says, “I wasn’t there, so I’m relying on you to explain what happened.” He asks the subject to provide as many details as possible, even if they seem irrelevant. Then the interviewer lets the volunteer talk.
Once the mock suspect is through telling his story, the officer asks “probing” questions, looking for “checkable details.” He says things like “Tell me more,” like a good dinner-party companion would. (Ideally, a PEACE interviewer should do no more than 20 percent of the talking; with a highly uncooperative subject, the strategy would shift to “conversation management,” which allows for more targeted questions.) He asks the subject to draw a picture of the scene. Finally, the officer reveals prepared evidence that contradicts his story and invites him to explain the discrepancies.
“People think it’s soft and sort of touchy-feely, but it’s not,” House remarked. “It’s very systematic. It requires that people do their research and know their case very well.”
PEACE is modeled on the cognitive interview, which was developed by psychologists Edward Geiselman and Ronald Fisher in the 1980s and draws on decades of psychological research on how we store and retrieve memories. The research shows we provide more details when allowed to tell a story from beginning to end, without interruption. (A staccato question-and-answer format produces more spare responses.) We offer more accurate descriptions when asked open-ended questions (“What did he look like?” rather than “Was he tall or short?”), and recounting trivial details can prompt us to remember important ones. Sometimes witnesses are asked to tell a story backward, which can likewise shake loose other memories. Reversing the chronology also adds to the “cognitive load” of the task, which means that guilty storytellers will struggle to keep the details straight and reveal their deception.
Davontae Sanford responds to a question about what advice he has for wrongfully convicted prisoners.
“The bottom line is we get a lot more information” from witnesses, said Sgt. Colin McNeil, who heads the RNC’s investigative interviewing unit. “A lot more information. It’s not about confession chasing.”
Information collected using this method is more reliable as well. PEACE involves less risk of introducing suggestive statements that can distort a witness’ memories. In the training courses, Snook said, “we tell them to treat memory like a crime scene. They don’t contaminate it; they don’t move things around. Just secure it, and keep it pristine.”
Advocates say questioning witnesses as PEACE instructs protects against tunnel vision and other errors that lead to wrongful prosecutions. Because investigators must gather a lot of information before they consider challenging the subject’s account, they are less likely to make premature judgments about guilt or innocence. It’s the basic principle of scientific inquiry, Snook explained: “You collect data first, and then you make your decision.”
There’s evidence that even with good training, dramatic improvements in practice can be elusive. A field evaluation published in April found that although trained RNC interviewers asked better questions in general (fewer leading ones), they didn’t ask for a free narrative each time and still posed a lot of less effective yes-or-no questions.
Working with local law enforcement agencies to change practices is a terribly slow process.
Christopher Kelly, assistant professor of sociology and criminal justice, Saint Joseph's University
Nevertheless, 25 years after PEACE was first deployed in Britain, dozens of studies have demonstrated that on the metrics that matter most, the method is a success. Multiple studies confirm that when compared with a Reid-style interrogation, a data-collecting approach significantly reduces the odds of a false confession and maintains or even increases the odds of a true one. If Davontae Sanford had been questioned under a PEACE protocol, perhaps he would not have spent eight years in prison for a crime he did not commit.
PEACE beats Reid in another area too, according to Inspector Todd Barron, who headed the RNC’s investigative interviewing unit and led PEACE training for six years. “Whether you agree with the Reid model or you don’t,” Barron said, “here’s a fact: It was designed for suspects, for people who were determined to be culpable in a crime.” It has little to say about questioning witnesses.
Yet talking to witnesses is the core of a police investigation. “For every suspect interview that a police officer does, they probably do eight or nine witness or victim interviews,” Barron explained. It’s crucial that investigators learn “how to accurately capture information from victims and witnesses and more importantly, to capture it in a proper way that isn’t tainted.” Newfoundland’s constables are among the few in the world who are systematically taught this skill.
Barron recalls that when the PEACE course was introduced, some of the most experienced officers were skeptical that they would learn anything useful. By the end of the first week, that changed. “The overwhelming sentiment was ‘Where was this 20 years ago when I started doing policing?’ ”
PEACE is slowly spreading to other districts in Canada. Several law enforcement agencies, including the Niagara Regional Police Service in Ontario, have sent envoys to St. John’s and started running PEACE courses of their own. But so far no police department in the United States has done the same.
“I would think most police officers in the U.S., or here, are simply interested in doing things the right way and open-minded enough to take on constructive change,” House wrote in an email recently. “I think most police officers are not fixated on a particular method, and are therefore open to evidence on what is truly best practice.” Yet experts on interrogation say that U.S. law enforcement is peculiarly resistant to change.
Military intelligence has shown more interest in reform than police, according to Melissa Russano, an associate professor of criminal justice at Roger Williams University. Russano helped train the U.S. Air Force Office of Special Investigations (the counterintelligence wing) in a method of interrogating “high-value detainees”—terrorism suspects—that Russano says is largely modeled on PEACE.
Russano would like to bring similar instruction to local police, but she believes the Reid technique is entrenched in the culture of U.S. law enforcement. Many years of tradition and practice, she noted, and Reid’s books and courses being a “booming business,” are “big hurdles to overcome.” The Major Crimes Division at the Los Angeles Police Department is in the process of converting to the Air Force system, according to Det. Mark Severino, and is not using Reid. But there has not been a department-wide policy change. “It’s going to take a while,” Severino said.
Christopher Kelly, an assistant professor of sociology and criminal justice at Saint Joseph’s University in Philadelphia, helped nudge the LAPD’s reforms along when he concluded, after analyzing LA cops’ videotaped interviews, that suspects were more cooperative when detectives worked to build rapport and less so when they were confrontational. But Kelly doesn’t see widespread reform coming anytime soon. “Working with local law enforcement agencies to change practices is a terribly slow process,” he said. “It’s like trying to turn the Titanic.”
Joe Buckley, the president of John E. Reid and Associates, is not worried about competition from PEACE. He thinks it is fine as far as it goes, but it doesn’t get the job done. Police departments that adopt it, he predicted, “are going to learn down the road that a lot of times people are not going to make an acknowledgment that they did it without more persuasive efforts.”
A lot of police departments, USF law professor Richard Leo believes, are simply “in denial” about the failings of current practice. He points to the case of the Central Park Five, whose convictions for raping and beating a jogger nearly to death in 1989 were overturned 13 years later. In spite of DNA and other evidence pointing to a different perpetrator, the New York Police Department refused to accept that the teenagers’ confessions were coerced and that they put the wrong guys in jail. Though the actual rapist insisted he acted alone, an NYPD panel concluded in 2003 “it is more likely than not” that the five teenagers “participated in an attack.”
Even in Newfoundland, the revolution has stalled slightly. Recently, the constabulary has begun to consider allowing some of the tactics that House and Snook hoped they’d banished for good. Last year, the Royal Canadian Mounted Police—Canada’s national police force—released its own protocol for suspect interviewing, which looks a lot like PEACE but allows officers, in cases where there’s solid evidence, to resort to some of the ploys that are typical of Reid.
“That stuff is quick and dirty,” Brent Snook said, “rather than the painstaking, slow, methodical, cerebral process of working with people to get information.” The Mounties’ hybrid program is being promoted across Canada, and now that House has retired, he worries that there could be a partial relapse at the RNC. “But we keep fighting the fight.”
The Newfoundland reformers have started proselytizing in the United States and have won a few converts. One of them is Boston Sgt. Det. John Brown. A year and a half ago, he and a couple of homicide detectives attended a seminar on PEACE that Snook and Barron presented at the Suffolk County District Attorney’s Office. New BPD recruits now learn some of the elements of PEACE before they earn their badge and weapon. Brown would like to see a nationwide switch to PEACE, but in the meantime, he says, detectives should stop using “ruses and trickery” to get confessions. “At the end of the day I don’t think it’s really necessary,” he said.
Just a few miles south of St. John’s is Cape Spear, a narrow peninsula reaching into the Atlantic Ocean. Brent Snook stood there on a wet afternoon in mid-May, looking out at a roiling gray sky and dark sea. The wind pelted rain across the hillside. “Nobody in Newfoundland uses an umbrella,” he called out cheerfully. “The rain comes in horizontally.”
Back in town, a different sort of storm was brewing. The RNC was investigating a homicide—a rare event on the island, where three or four a year is about average. The week before, the body of a man had been discovered outside an apartment building on a quiet street in St. John’s, his red hoodie soaked in blood.
Residents were alarmed when a woman who’d moved in a few days before, and who’d just been released from a psychiatric hospital, was charged with the murder. A date for a preliminary court inquiry is expected to be set next week.
A grisly, well-publicized crime, a psychologically unstable suspect: These are some of the risk factors that can lead to an aggressive interrogation, and sometimes a miscarriage of justice. Canadian law says the RNC can use face-saving excuses and other persuasive tactics to urge the suspect to confess. But RNC investigators don’t operate simply according to what’s allowed. They operate according to “evidence on what is truly best practice,” as John House put it. And according to “what’s right.”