Paul Spella / Katie Martin

An Epidemic of Disbelief

What new research reveals about sexual predators, and why police fail to catch them

Updated at 11:17 a.m. ET on July 22, 2019.

This article is part of our project “The Presence of Justice,” which is supported by a grant from the John D. and Catherine T. MacArthur Foundation’s Safety and Justice Challenge.


Robert Spada walked into the decrepit warehouse in Detroit and surveyed the chaos: Thousands of cardboard boxes and large plastic bags were piled haphazardly throughout the cavernous space. The air inside was hot and musty. Spada, an assistant prosecutor, saw that some of the windows were open, others broken, exposing the room to the summer heat. Above the boxes, birds glided in slow, swooping circles.

It was August 17, 2009, and this brick fortress of a building housed evidence that had been collected by the Detroit Police Department. Spada’s visit had been prompted by a question: Why were police sometimes unable to locate crucial evidence? The answer lay in the disarray before him.

To hear more feature stories, see our full list or get the Audm iPhone app.

As Spada wandered through the warehouse, he made another discovery, one that would help uncover a decades-long scandal, not just in Detroit but across the country. He noticed rows of steel shelving lined with white cardboard boxes, 10 inches tall and a foot wide, stacked six feet high. What are those? he asked a Detroit police officer who was accompanying him. Rape kits, the officer said.

“I’m assuming they’ve been tested?” Spada said.

“Oh, they’ve all been tested.”

Spada pulled out a box and peered inside. The containers were still sealed, indicating that the evidence had never been sent to a lab. He opened four more boxes: the same.

“I tried to do a quick calculation,” he later told me. “I came up with approximately 10,000.”

Spada’s estimate was conservative. Eventually 11,341 untested rape kits were found, some dating back more than 30 years—each one a hermetically sealed testament to the most terrifying minutes of a woman’s life, each one holding evidence that had been swabbed or plucked from the most private parts of her body. And in all likelihood, some microscopic part of her assailant—his DNA, his identity—sat in that kit as well.

Or kits.

Eric Eugene Wilkes was known to Detroit police for robbery and carjacking. Not for rape. Yet Wilkes’s DNA was in boxes scattered throughout the warehouse, even as he walked free. His DNA first arrived there more than 18 years ago, after he raped a woman waiting for a bus on December 26, 2000. It next appeared after another rape four months later. Three days after that, police shelved the untested kit from his third victim.

One can imagine a certain rhythm to the process, as police hoist kit after kit onto the metal shelves, not knowing that they hold in their hands the identity of a serial rapist. Here’s the evidence box from a deaf woman Wilkes assaulted in June 2006. There’s one from a woman he raped in May 2007. The kit from his sixth victim arrived in June 2010. Another a month later. Two more in August 2011. His 10th victim, four months after that. Not until he raped his 11th victim, in January 2012, did the sequence end, because that woman saw Eric Wilkes two days after the assault and called the police, who arrested him. Eleven years, 11 violent rapes—all while Wilkes’s identity was preserved in sealed containers that no one had bothered to open.

The untested rape kits would continue to accumulate for years after Spada’s visit. But that August day became a defining moment for survivors of sexual assault. Spada called Kym Worthy, the county prosecutor, and told her what he’d found. “I was livid,” Worthy recalls. “I wanted to test them all immediately.” She began talking to reporters, and the decrepit warehouse in Detroit with the broken windows became a powerful symbol of police negligence.

Since then, Detroit and other jurisdictions across the country have shipped tens of thousands of kits to labs for testing. The results have upended assumptions about sexual predators—showing, for example, that serial rapists are far more common than many experts had previously believed.

But the rape-kit scandal has turned out to be only a visible symptom, a mole on the skin that hints at a pervasive cancer just below the surface. The deeper problem is a criminal-justice system in which police officers continue to reflexively disbelieve women who say they’ve been raped—even in this age of the #MeToo movement, and even when DNA testing can confirm many allegations. From the moment a woman calls 911 (and it is almost always a woman; male victims rarely report sexual assaults), a rape allegation becomes, at every stage, more likely to slide into an investigatory crevice. Police may try to discourage the victim from filing a report. If she insists on pursuing a case, it may not be assigned to a detective. If her case is assigned to a detective, it will likely close with little investigation and no arrest. If an arrest is made, the prosecutor may decline to bring charges: no trial, no conviction, no punishment.

Each year, roughly 125,000 rapes are reported across the United States. Sometimes the decision to close a case is surely correct; no one wants to smear an innocent man’s reputation or curtail his freedom because of a false report. But in 49 out of every 50 rape cases, the alleged assailant goes free—often, we now know, to assault again. Which means that rape—more than murder, more than robbery or assault—is by far the easiest violent crime to get away with.

“Right there,” Liz Garcia says, pointing to a second-floor window of a modest white house in Cleveland. “That’s the window of the bedroom that I was raped in.” March 23, 2004, she recalls, was a bright, crisp day. With her twin girls in school and her paramedic training almost complete, she decided it was just the day to wash her Ford Explorer. She ran upstairs to the bathroom for a towel. Looking in the mirror, she saw the door swing open behind her. She turned and saw black shoes. Her gaze traveled upward: black pants, black gloves, black jacket, black ski mask.

Over the next two hours, the man dragged Garcia from room to room. She thought of running or jumping out a window, but he was bigger, muscular; he seemed to anticipate her moves. He raped her three times. He was prepared and meticulous. He wore gloves and a condom. He spread a towel on Garcia’s bed, and took it with him when he left. “He had shaved his legs and chest”—she could feel the stubble—“so he wouldn’t leave hair behind. He knew what he was doing.” He ordered her to wash out her mouth, and made her shower as he watched. Before leaving, he told her to count to 500.

“He closed the shower curtain, and I heard him go down the stairs. I am standing there. Do I get out? Do I count? And all of a sudden”—Garcia yanked her hand from right to left—“he opens up the shower curtain. I didn’t even hear him come back up the stairs. It was terrifying.”

Satisfied that Garcia had not moved, the man fled.

Although the police didn’t yet know it, a serial rapist had been stalking Cleveland since the mid-1990s. He’d begun with vulnerable women: women willing to sell sex for drugs or money, an unlucky woman whose car ran out of gas, one teenager who was skipping school, another with a prosthetic leg. This should have put the police on high alert, Tim McGinty, a former Cuyahoga County prosecutor, told me. Vulnerable people—drug addicts, prostitutes, people living in poor neighborhoods—are the “canaries in the coal mine. If you’ve got a serial rapist out there, who does he hit first? He hits the vulnerable people.”

By 2004, the rapist had graduated to home invasions and more prosperous victims. One week after the attack on Liz Garcia, a 55-year-old schoolteacher was raped in her home. Only then, after attacks on two middle-class women, did the police make a public plea for leads. The department received an anonymous tip: an envelope with a newspaper clipping and an arrest record for a former probation officer named Nathan Ford. The police apprehended Ford and swabbed him. As part of a pilot study, the department had sent some 250 rape kits off for DNA testing—and Ford’s DNA matched eight of them. But not Liz Garcia’s. The police tested her kit but didn’t find her assailant’s DNA. “They told me I would never know who the attacker was,” she says.

Liz Garcia was attacked in her home in Cleveland by a serial rapist who had eluded the police for years. (Jessica Dimmock)

At the time, if you were raped in Cleveland and you were poor or otherwise vulnerable, police would likely make a couple of phone calls and move on. You can see this play out in the police files documenting the response to Nathan Ford’s early attacks. All of Ford’s victims who came forward had forensic exams, but detectives were more likely to shelve the kits than send them to a lab. Rarely did a detective visit the victim, witnesses, or the crime scene. If a victim couldn’t come to police headquarters on the detective’s timetable—because she couldn’t find transportation or child care or get time off from work—she was labeled “uncooperative.” The case was closed. In other instances, the detective wrote that he couldn’t locate the victim, and this was enough to end the investigation. Yet when investigators reopened sexual-assault cold cases 20 years later, they almost always found the victim within a few hours.

When the Cuyahoga County prosecutor’s office hired a team of researchers at Case Western Reserve University, in 2015, to pore through police files and other records connected to thousands of untested rape kits in Cleveland, they quickly spotted the same pattern. In a random sample of cases, mainly from the mid-’90s, they found that the notes from many police investigations barely filled a single page. In 40 percent of cases, detectives never contacted the victim. In three out of four, they never interviewed her. Half of the investigations were closed in a week, a quarter in a day. As for rape kits—the one type of evidence that might definitively identify a rapist—police rarely sent them to the lab for testing. Granted, testing a kit could cost more than $5,000 in the late ’90s and 2000s. But during part of that time, the state was paying police departments to send in evidence. And even when the cost of testing a kit dropped to less than $1,000, police still tucked away the evidence in storage. Ultimately, Cleveland would accumulate some 7,000 untested kits.

Nathan Ford’s rampage wasn’t enough to persuade the Cleveland police to begin addressing the rape-kit backlog. What did persuade them was a serial killer. In October 2009, the police discovered the bodies of 11 women buried in the home and backyard of Anthony Sowell, a convicted rapist. Over the years, some of Sowell’s intended victims had escaped and reported his attempts to rape them. But the police had never thoroughly investigated their claims. At least one woman had completed a forensic exam. The police had tested the rape kit—but only for drugs in her system, not for the rapist’s DNA.

The Sowell case became a scandal, and it raised larger questions: Why weren’t attacks on women being investigated? How many rape kits did the police department have in storage? How many had been tested?

Under pressure from then–Ohio Attorney General Mike DeWine, the city’s police department began sending off kits for testing in 2011. Officials called it a “forklift” approach because every box, no matter how old, was shipped to a state lab. At first the progress was slow. But in January 2013, Tim McGinty, who had just been elected Cuyahoga County prosecutor, created a task force devoted to testing the kits and reinvestigating cases. He brought in 25 detectives, mostly out of retirement, and assigned half a dozen assistant prosecutors to the effort. He allowed two reporters from The Plain Dealer to sit in on their weekly meeting.

Within weeks, DNA results started arriving from the lab: More than a third of the rape kits were pinging in the FBI’s Combined DNA Index System, known as CODIS. Created in the 1990s, the database contains DNA profiles collected at crime scenes across the country, many of them linked to the name of a known criminal. Cleveland investigators were soon identifying rapists who had eluded detection for decades. “It was much more fruitful than we ever in our wildest dreams imagined,” recalls DeWine, now the governor of Ohio. Some weeks, Richard Bell, the prosecutor in charge of the task force, would announce 20 new DNA matches.

Investigators sometimes had only a few days to build a 20-year-old case—to locate victims and witnesses and gather their sworn statements—before the statute of limitations ran out. “There was one hit where we turned it around in two days and brought it into the grand jury at 4:15 p.m., before the 4:30 end of day,” Bell recalls. Cases with fewer than 10 days remaining were labeled, in red ink, all hands on deck.

Since Cuyahoga County began forklifting its kits, prosecutors have indicted nearly 750 rapists in cold cases and convicted more than 400 of them. (Detroit, which got a later start, has convicted some 175 men.) “They would never have resurrected the [closed cases] without this project,” Bell says.

Richard Bell, the prosecutor leading a task force in Cleveland that has convicted more than 400 rapists in cold cases. (Jessica Dimmock)

For more than a decade, Liz Garcia had wondered whether her rapist would return to kill her and her daughters, as he’d promised. She suffered panic attacks, sometimes five a day. She avoided answering the door. She showered with the curtain open. She left the light on all night. She slept on the couch, with her back to the wall. “I had knives under my pillows. I hid knives all over the house,” she told me.

Not until she found a detective’s card tucked in her door more than a decade later did she cease to regard the world outside her home like a prey without cover. The lab had retested her rape kit using newer technology; this time it detected male DNA and identified her attacker: Nathan Ford. The police also discovered more victims whose kits had been shelved for years, bringing Ford’s total to 22 rape kits. By then he was already in prison and serving a life sentence. Garcia could put away her knives. She still sleeps with the light on.

When the members of Cleveland’s task force began shipping rape kits to the state lab, they didn’t imagine they’d end up fomenting a small revolution in criminology. Yet those evidence boxes uncovered new clues about the behavior of sexual assailants and overturned some basic assumptions—about how often they offend, whom they attack, and how they might be captured.

Rachel Lovell, the lead researcher at Case Western, reviewed the results of the tests and found herself with a new and superior class of information. In the past, most research on rapists relied on prison records or “self-reports”—that is, surveys of people who answered questions anonymously about their behavior. But here, in her hands, were the biological name tags of thousands of men who had committed a rape and walked away. It was a larger and far more objective sample of sexual offenders. It was the difference between a pencil sketch and a color photograph.

What struck her first was the sheer number of repeat offenders: Of the rape kits containing DNA that generated a CODIS hit, nearly one in five pointed to a serial rapist—giving the Cleveland investigators leads on some 480 serial predators to date. On a practical level, this suggested that every allegation of rape should be investigated as if it might have been committed by a repeat offender. “The way we’ve traditionally thought of sexual assault is this ‘he said, she said’ situation, where they investigate the sexual assault in isolation,” Lovell told me. Instead, detectives should search for other victims or other violent crimes committed nearby, always presuming that a rapist might have attacked before. “We make those assumptions with burglary, with murder, with almost any other crime,” Lovell said, “but not a sexual assault of an adult.”

Another surprise for police and prosecutors involved profiling. All but the most specialized criminologists had assumed that serial rapists have a signature, a certain style and preference. Gun or knife? Alley or car? Were their victims white, black, or Hispanic? Investigators even named them: the ponytail rapist, the early-morning rapist, the preacher rapist.

But Lovell recalled sitting in Cleveland’s weekly task-force meeting, listening to the investigators describe cases. They would say: This guy approached two of his victims on a bicycle, but there was this other attack that didn’t fit the pattern. Or: This guy assaulted his stepdaughter, but he also raped two strangers. “I was always like, ‘This seems so very different,’ ” Lovell said. “This is not what we think about a serial offender. Usually we think of serial offenders as particularly methodical, organized, structured—the ones that make TV.”

Eric Beauregard, a criminologist at Simon Fraser University who has interviewed 1,200 sexual offenders, says profiling may fail because a predator’s reality falls short of his fantasy. Most offenders tell him that they do hunt for a certain type of victim, but “what they had in mind and what they selected did not match at all,” he says. “If they are looking for a tall blonde with big breasts, at the end of the day, it was: She was there, she was available, she was alone. Those were the criteria.” Nathan Ford’s victims, for example, were black, white, Hispanic, and Asian; 13 years old and 55; on the west side of the city and on the east.

“Thank God we have DNA,” Dan Clark, one of the Cleveland investigators, says. “Because trying to put together a pattern where there is no pattern is impossible. It’s no wonder we didn’t catch that many people.”

Most rapes, of course, are not committed by strangers. Eighty percent of the time, the rapist is someone a woman knows—they met at a party or a bar; he’s her colleague, friend, mentor, coach. So police saw little reason to send off those rape kits: The man’s identity was never in doubt. But the Cleveland study illuminated another insight—one that shows the tragic consequences of failing to test “acquaintance rape” kits. Historically, investigators had assumed that someone who assaults a stranger by the railroad tracks is nothing like the man who assaults his co-worker or his girlfriend. But it turns out that the space between acquaintance rape and stranger rape is not a wall, but a plaza. When Cleveland investigators uploaded the DNA from the acquaintance-rape kits, they were surprised by how often the results also matched DNA from unsolved stranger rapes. The task force identified dozens of mystery rapists this way.

The Justice Department has awarded $154 million to 54 jurisdictions for rape-kit testing, but the effort hasn’t yet resulted in many convictions. Right: Tim McGinty, the former Cuyahoga County, Ohio, prosecutor who created the task force there. (Jessica Dimmock)

The Case Western research also showed that the great majority of rapists are generalists, or “one-man crime waves.” “They will steal your car, they will steal your watch, and they will steal sex, so to speak, if they can get away with it,” says Neil Malamuth, a psychologist at UCLA. “They are antisocial folks who will commit all sorts of antisocial behavior, including but not limited to sexual aggression.” And eventually, experts say, generalists slip up and get caught.

Consider the story of Natasha Alexenko. She was raped in the stairwell of her New York City apartment building in 1993. The investigation turned up no suspects, and CODIS did not yet exist. Ten years later, the database was up and running, though sparsely populated. When police plugged in the rapist’s DNA, they found no match. With the statute of limitations about to expire, prosecutors were able to indict the John Doe whose DNA was found in Alexenko’s rape kit. Then they waited, hoping he would commit another crime. In 2007, Victor Rondon was stopped for jaywalking in Las Vegas and, in an impulse he surely regrets, punched the police officer. The DNA from Rondon’s swab matched that of Alexenko’s rape kit. Rondon was convicted of rape, sodomy, sexual abuse, burglary, and robbery.

“These are not the Napoleons of crime,” Tim McGinty told me. He paused, reflecting on those 7,000 rape kits sitting in storage in Cleveland while the perpetrators were free on the streets. “They’re morons. We were letting morons beat us.”

Liz Garcia considers herself fortunate. At least the Cleveland police submitted her rape kit for testing, even if they weren’t able to identify her assailant until they retested it 12 years later. But what about the other victims, the ones who endured an invasive forensic exam, expecting the police to marshal that evidence to catch their assailants? “How can somebody just let them sit there?” Garcia asks. “You know, the women calling and calling, trying to find answers. You’re giving them some story, and all along this rape kit is sitting there, not even being tested. ‘No, we don’t have anything further.’ But you could have if you would have tested that kit! You could have avoided other rapes if you would have tested that kit.”

This is the question that haunts every advocate, researcher, and enlightened detective or prosecutor I spoke with: How many rapes could have been prevented if the police had believed the first victim, launched a thorough investigation, and caught the rapist? How many women would have been spared a brutal assault?

The federal government estimates that police departments have warehoused more than 200,000 untested sexual-assault kits. But no one really knows, because cities and states fight to keep those numbers secret. The Joyful Heart Foundation, an advocacy group started by Mariska Hargitay, who stars in Law & Order: Special Victims Unit, has identified more than 225,000 kits through public-records requests. But given that 15 states and many large cities have declined to even count the untested rape kits in their possession, the group believes there may be several hundred thousand more.

In 2015, the Obama administration launched the Sexual Assault Kit Initiative (SAKI) to encourage cities and states to send untested kits to labs, open new investigations, and prosecute the assailants who had slipped under the radar for years or decades. So far, the Justice Department has awarded $154 million to 54 jurisdictions. “It has exceeded our expectations,” says Angela Williamson, who has headed the program since its inception. When she sees the dedication of the detectives and prosecutors who are working long hours on cases that can date back decades, she is overwhelmed with gratitude: “You want to cry.”

“There is no money better spent than the Justice Department spends here, dollar for dollar,” Tim McGinty said. “I don’t think there will ever be another time in history when so many criminals can be arrested so easily, so quickly, so inexpensively, and with such certainty.”

If aggregate numbers and a drumbeat of positive news stories are the proper measures, then the SAKI program has been a huge success. The Justice Department reported that some 61,000 rape kits have been inventoried, and nearly 45,000 tested. Police have opened (or reopened) 5,500 investigations, and prosecutors have won 498 convictions or plea agreements.

But drill down a bit, and you see that two places account for most of the progress. In response to a Freedom of Information Act request, the Justice Department reported that of the 41 SAKI sites that began receiving money in 2015, Cleveland and Detroit accounted for 38 percent of all new investigations in the first three years. (Numbers for 2018 are not yet available.) When a suspect was charged, it happened in Detroit or Cleveland 69 percent of the time. As for winning a conviction at trial or securing a plea agreement, 82 times out of 100, the prosecutors in Detroit or Cleveland were the ones going out for a celebratory beer.

Everywhere else, the distance between aspiration and accomplishment is startling. When I looked at the performance of the 41 sites, my first thought was: All those zeros! Did the $5.1 million awarded to Wisconsin really buy only four charges and zero convictions? What about Connecticut, which received $3.3 million in that time: not a single reported charge or conviction? Or Iowa, which can’t show a single charge or conviction for $3 million of federal largesse? Mobile, Alabama; New Orleans; Delaware—collectively, they received $6.3 million, but can boast only four arrests and not a single conviction or plea deal.

Williamson says the numbers are “not a fair snapshot of the hard work that’s being done.” She says it can take months to inventory kits and send them off for testing; more time still to receive a name from the lab and then launch an investigation; and possibly years to find the suspect and victim and make an arrest, much less convict him at trial. Detroit and Cleveland just got a head start. “The longer that these grants are out, and the longer these sites have been funded, you’re going to be seeing numbers start to pour in,” Williamson told me. I was briefly persuaded: In the past few months, several of those sites have launched new investigations. But then I remembered that Cleveland had moved from testing kits to securing indictments in less than 10 months. Once the task force received a lab result, it could gather enough evidence in a 20-year-old cold case to arrest a suspect in less than 10 days—because it had the will to do so.

Meaghan Ybos calls the SAKI program “a big charade.” Ybos was raped in her family’s home in 2003, when she was 16. The Memphis police shelved her rape kit for nine years. In the meantime, five other women and a 12-year-old girl were raped by the same man. When Ybos discovered, in 2012, that her kit had never been tested, she began a public fight with the city, talking to reporters, questioning officials at public hearings, and finally suing the city in 2014. The Memphis police admitted to having 2,000 untested kits, a number they later raised to more than 12,000. Still, SAKI officials and women’s-advocacy groups have praised Memphis: With the $4.5 million the city received from the SAKI program, it has sent all its kits for testing, and according to numbers submitted to the Justice Department, it has opened more than 1,000 new investigations, and won more than two dozen convictions or guilty pleas—more than most other SAKI grantees. Ybos counters that city officials for years misled the public about untested kits and continue to fight victims seeking compensation. The funding and praise Memphis has received—“that could be seen as rewarding the worst actors,” she says.

It’s true that the national backlog of untested kits is shrinking. States are also passing laws to ensure that rape kits don’t languish in storage rooms in the future. But you can test every kit in the country and not solve a single case if you don’t follow up on the leads. Rebecca Campbell, a psychologist at Michigan State University who analyzed the lapses by police in Detroit and now trains detectives at SAKI sites, says that officials in some of those jurisdictions have told her they intend only to test kits—not to actually prosecute the men who are identified. Rachel Lovell, of Case Western, has heard this too. “If you’re not investigating or following up on the testing of those kits, what’s the point?” she asks. “It simply becomes a piece of paper in a file.”

Why would officials decide not to pursue these cases? Campbell and Lovell point to the same factor: law enforcement’s abiding skepticism of women who report being raped. This is a problem with no easy fix, says Dan Clark, the Cleveland detective, who conducts training programs across the country for SAKI. Clark tries to teach investigators to take a woman’s allegation of rape as seriously as they would a report of assault or robbery. But in private conversations afterward, he says, it becomes clear the message didn’t sink in: Officers continue to tell him they think that many women lie about being raped, and that their claims aren’t worth investigating. “This is that sort of intractable belief that we could not seem to shake.”

Campbell compares her training sessions to the movie Groundhog Day: “I’m hearing the same things I heard in Detroit back in 2010, 2011. It’s just in a different city, but it’s the same basic idea over and over again.”

She stresses that these are the departments with good intent—the ones that have applied for money and committed to addressing the rape-kit crisis. After many of her sessions, Campbell feels euphoric, sensing that at least in that room and on that day, people have signed on to the message. “Then when that feeling wears off, I think about: Who isn’t at this meeting? Who didn’t apply for SAKI funds? And that’s the majority of law-enforcement agencies and prosecutors’ offices throughout the U.S.”

Do we know how those departments are doing? I ask.

“It’s a black box.”

Amber Mansfield’s story offers a glimpse into that black box. Mansfield, who is 39 years old and lives in Minnesota, admits to having a “colorful” history. After her parents lost everything to their crack addiction when she was 9, she bounced among foster homes, lived on the streets, and spent time in a juvenile-detention center. When she turned 18, she was left to fend for herself. In her early 20s, she picked up a conviction for drug possession (and served one year) and a misdemeanor charge of prostitution. Since 2005, however, she has had no serious run-ins with the law. She earned her high-school diploma, fled her sketchy Minneapolis neighborhood, and, nine years ago, gave birth to a daughter, who is the center of her life. They now live in the town of Mora, population 3,500, in a white-clapboard house with a bicycle in the front yard and an SUV in the driveway.

I visited Mansfield on a rainy day last fall. She greeted me shyly, and as we settled into a deep couch in her living room she began to unspool her story. In 2011, Mansfield started corresponding with Keith Washington, a childhood friend who was then in prison for assaulting a police officer. Or so he said. “Pretty much everything was a lie,” she told me. She would learn, too late, that he had been convicted of raping and beating his girlfriend. After he was released, in May 2015, they began spending time together as friends. He wanted more, but she hesitated; he had a record, and she had a daughter to consider. On the night of July 22, 2015, Washington flew into a rage, took Mansfield’s car keys, and locked her in the bedroom of his sister’s house. He hit her, pinned her to the ground, tightened his hands around her neck. “This is what it feels like for the last breath to leave your body,” he told her before she passed out. When she regained consciousness, he begged her to forgive him, but a few hours later, he raped her.

After a forensic exam at the hospital, two police officers arrived to take her statement. They peppered her with pointed questions; the interaction seemed more like an interrogation than an interview. She read the doubt in the officers’ faces. “It’s my word against his word,” she said. “I mean, a sex offender and a prostitute. You do the math.”

It was, apparently, a quick calculation.

Mansfield assumed they would run a background check on her, as well as on Washington. She was half correct. The officers looked at her record but not his, and sent the report to Lieutenant Michael Sauro, who headed Minneapolis’s sex-crimes unit. I recently met with Sauro, who is now retired, to discuss Mansfield’s case. He recalled seeing that she had a prostitution charge on her record. “I’m thinking, Whoa, wait a second here. How much resources am I going to spend if you’re that—how should I say—careless with your own self?” Sauro told me. “So after reading three or four paragraphs, I said, ‘To hell with this. We’re not going to spend any time on this.’ So that’s probably why I did not even waste my time running his criminal history.”

Sauro developed a hard-nosed cynicism during his years on the job. “People lie,” he reminded me several times as we sat in his living room. He then explained that when prostitutes report a rape, it’s typically just a deal gone bad; they want revenge for nonpayment. But, I countered, Mansfield’s one prostitution charge had been a dozen years earlier. “Yeah,” Sauro said, “but that lifestyle keeps dragging you back.”

Had anyone taken 20 minutes to enter Washington’s name into a criminal database, he or she would have seen that Washington was a Level 3 sex offender, considered the most violent and most likely to reoffend. Instead, Sauro “redlined” the investigation, shutting it down without assigning it to a detective. Washington was never interviewed by the police. But he did hear about the allegation, prompting him to threaten Mansfield by phone and text. “It was all day, every day,” she said.

Mansfield called the station again and again to learn the status of her case. She never received a response. “Finally I just couldn’t take it anymore,” she said. “I called the up-aboves and just told them: ‘Listen, you guys are putting me in more danger than you’re doing any good. I’m done. Fuck all you guys. That’s it.’ ”

After Amber Mansfield was assaulted and raped by a childhood friend, police checked only her record, not his. Left: The knife she leaves in her door. (Jessica Dimmock)

Sauro remembers the conversation. “When she said, ‘Nah, I don’t want to prosecute it anymore,’ I wasn’t going to beg her to follow through,” he told me. He paused. “But if I would have known he was a Level 3, I would have begged her, okay?”

Months later, Keith Washington was arrested for assaulting two women a few hours apart; he had strangled them and left them unconscious and partially undressed on the street. “If they would have done their job and got him,” Mansfield said, “these other two ladies would have been all right.” (Sauro dismissed this claim as “conjecture.”) In the end, Washington was convicted for assaulting one of the women and is serving a 15-year sentence; the other case was dropped because the woman was unavailable to testify. Police questioned him about assaulting Amber Mansfield. He denied the allegation and, given the complications in the case—her history, their history—prosecutors declined to try him for the attack.

More than three years later, Sauro seemed genuinely chagrined that a rapist had slipped by him. But in the next breath, he noted that his six-person sex-crimes unit was handling more than 400 cases a year. “I mean, when you’re so busy, sometimes you miss stuff,” he said, adding that the unit thoroughly investigates “99 percent of the time.”

Perhaps. But how would anyone know? The lapses in Mansfield’s case didn’t come to light until Washington assaulted the two other women; only then did a detective call her, ask about her assault, and persuade her to testify against him in the other woman’s trial. How many other cases have been closed with little or no investigation and locked away in a filing cabinet, leaving the victims with no answers and no recourse?

I learned about Mansfield through a devastating series of articles in the Minneapolis Star Tribune. The paper analyzed the police files from nearly 1,500 sexual-assault cases across the state that had been closed in 2015 and 2016. “You look through these case files and you see a witness name, and you’re thinking, Okay, they’re going to interview this witness,” recalls Brandon Stahl, one of the reporters. “And they don’t. You see evidence that could be tested or collected—and they don’t do that, either.”

In 65 percent of the cases, Minneapolis investigators failed to interview the victim. Even when detectives had the name of the suspect, more often than not, they didn’t question him. In the end, only 9 percent of the cases resulted in a conviction. No doubt, the detectives struggled under a crushing workload. The paper found that sex-crimes detectives juggled three times the caseload of homicide detectives. But these numbers? They were so low, says MaryJo Webster, who analyzes data for news stories at the paper, that the team kept rechecking its methodology. “We kept thinking, Oh, we’ll get more data, and it’ll be wrong, and we’ll find something different,” she says. “The pattern just held and held and held.”

Sauro called the series a “hit job.” “I want to make one thing perfectly clear: There was never a case—misdemeanor or felony—that I thought was prosecutable that we did not investigate.” Prosecutable is the operative word here. Legally, police are supposed to investigate an allegation based on probable cause, not on whether they think a case can be proved to a jury beyond a reasonable doubt. That’s for a prosecutor to decide. But redlining means that police shut down an investigation without informing a prosecutor of its existence, and fail to gather evidence in what might turn out to be a winnable case.

After poring over hundreds of investigations, Stahl noticed a “reverse domino effect.” Prosecutors look down the road to juries, which, studies show, tend to be older than the general population, more conservative, and more skeptical of rape allegations by vulnerable victims such as Mansfield; predicting that the jury won’t convict, they decline to prosecute. Police see prosecutors declining all but the inevitable wins and think: Why investigate this case when it will never see a courtroom? “And that,” Stahl says, “is where the victim is often left out in the cold.”

Sauro sees it differently. “Why would I present a case for charging that’s not going to be prosecuted?” he told me. “If you want paper shuffled, let’s start shuffling the paper. Let’s go cut down some more trees. But that’s not justice.”

It must be an unenviable job for any detective or prosecutor, trying to discern the contours of truth in the half-light of the most intimate of crimes, one in which there are usually no witnesses and no evidence except the woman’s word to prove that the sex happened against her will. And, of course, a person can regret sex the next morning; communication between two people can fail; what begins as a consensual act can take an unwanted turn. All of this makes it tricky to prove assault beyond a reasonable doubt.

But even given these challenges, the skepticism shown by police and prosecutors—who are not juries, after all—is extraordinary. Officials don’t talk about their methods publicly, and rarely reveal their thinking, much less their motives or biases. But two cities—Detroit and Los Angeles—allowed researchers to read thousands of pages of police reports and to interview detectives and prosecutors. What the researchers found is a subterranean river of chauvinism, where the fate of a rape case usually depends on the detective’s or (less often) prosecutor’s view of the victim—not the alleged perpetrator.

Usually only a certain type of victim will see her rapist prosecuted, says Cassia Spohn, the director of the School of Criminology and Criminal Justice at Arizona State University. Along with Katharine Tellis, a criminologist at California State University at Los Angeles, Spohn published an exhaustive report in 2012 that analyzed sexual-assault investigations and prosecutions in Los Angeles County. “We heard over and over detectives use the term righteous victim,” she told me. A woman who didn’t know her assailant, who fought back, who has a clean record and hadn’t been drinking or offering sex for money or drugs—that woman will be taken seriously. Spohn recalled a typical comment: “ ‘If I had a righteous victim, I would do all that I could to make sure that the suspect was arrested. But most of my victims don’t look like that.’ ”

In cases of acquaintance rape, detectives expressed doubt and blamed the women. They spoke skeptically of “party rapes,” in which women drink too much “and make bad choices.” One described “buyer’s remorse,” where a woman who has been out partying has sex with a man “willingly” and later regrets it. “Out of 10 cases,” one detective said, “eight are false reports.”

Rebecca Campbell heard similar language from investigators in Detroit. In her 2015 report (a 550-page postmortem of Detroit’s rape-kit scandal), detectives often said that women “got what they got” if they knew the man. She asked one detective whether a man can rape an acquaintance. “Truly rape?” he asked. “Sometimes. But not most of the time.”

In some cases, police didn’t believe that sex had occurred at all. Consider this report by a Detroit detective, after a 14-year-old girl claimed she was abducted by two men and raped inside a burned-out house. “This heffer is trippin,” the detective wrote. “She was clean and smellin good, ain’t no way that shit happened like she said … The jig was up. She didn’t want to talk no mo. So her mama took her to the hospital, but they got the fuck outta here.” That investigation warranted two pages, which ended: “This case is closed: UTEEC.” Unable to establish the elements of the crime.

To police officers who haven’t been trained to spot signs of trauma, many rape victims appear to be lying. Why was she laughing when she gave her statement? Why was she so flat and unemotional? One Detroit detective told Campbell that a victim should be “a complete hot mess. They should be crying. They should be very, very traumatized.” But research finds that many victims don’t respond in a predictable fashion. This goes for their behavior during the assault as well as after: Why didn’t she fight? Why didn’t she run? Liz Garcia used to tell people that she would fight like crazy if a stranger ever came into her house. “I don’t say that anymore. I could have had all the weapons in the world in my house. But I couldn’t grab a weapon. He was taller, bigger; there was no fighting him.” One survivor told me she offered her assailant a glass of iced tea, hoping her courtesy would dissuade him. Another tried to politely decline the assault: You don’t have to do that. It’s fine. Yet another pretended she was enjoying herself, hoping he wouldn’t kill her afterward.

Michael Sauro, who headed Minneapolis’s sex-crimes unit and oversaw Amber Mansfield’s case, says the unit thoroughly investigates “99 percent of the time.” (Jessica Dimmock)

If detectives blame or disbelieve a woman, their next step is to close the case by persuading her to withdraw the complaint. In Detroit, Campbell says, detectives sometimes opened interviews by noting that the victim would be charged for false reporting if she said anything that was untrue or couldn’t be corroborated. Worried about being prosecuted, the woman would withdraw the allegation, and the officer would walk her to the door. One survivor told Campbell that the entire process seemed aimed at “culling the herd.”

But even when the victim does everything right, even when the police build a strong case against a suspect—even then, a prosecutor might decline to bring the case to trial. Prosecutors, particularly elected ones, are measured by their wins and losses and may be unwilling to spoil their record with problematic cases. “They only allow certain victims to go to trial, where they feel they have really rock-solid evidence,” Campbell says. “They’ve got to have the perfect victim, the perfect crime, the perfect witness—and anybody who deviates from that is not going to have their day in court.” Maybe the woman has a checkered past. Maybe she had too much to drink that night. Maybe she knew the suspect, triggering the nearly bulletproof defense of consent.

Sometimes, even a confession is not enough. One woman told me about a man who had been on a tour she led of her family’s organic farm. Later that night, while her husband was traveling, the man snuck into her bedroom and assaulted her. At first she thought the man was her husband, and she waited a few seconds before kicking him off the bed. Afterward, she called the police, and two weeks later, as officers listened in and recorded a phone conversation between the two, the man apologized for assaulting her. The officers were elated. But the prosecutor had reservations: No jury would believe she mistook the intruder for her husband. He declined to bring charges.

What recourse does a victim have when police or prosecutors refuse to take her seriously? Virtually none, it seems. She can’t force the police to investigate and she can’t make prosecutors try her case, because the state has vast discretion in how it handles criminal cases. Some women—in San Francisco, Houston, and Memphis—have tried to sue in federal court. They claimed that the state violated their due-process rights by failing to test their rape kits and fully investigate their claims, and that government policies discriminated against women by giving rape cases a lower priority than violent crimes more commonly committed against men, such as aggravated assault and robbery. Those lawsuits have been dismissed or withdrawn, although a federal appeals court recently ruled that the Memphis lawsuit was incorrectly dismissed and should be reinstated. A class-action suit in Austin, Texas, may have a better chance of showing gender discrimination based on one striking fact: Of the more than 200 sexual-assault cases police referred to prosecutors from July 2016 to June 2017, eight resulted in plea agreements, but only one case went to trial. The victim was a man.Yet even as police and prosecutors seem stuck in time, our culture is moving forward. This moment feels fundamentally different from previous decades, when a sensational rape trial would trigger a surge of outrage and promises of reform, only to see the scandal ebb from consciousness. Too many women have disclosed their #MeToo moments; too many rape kits have been pulled out of storage rooms. And if the success of the Cleveland task force proves anything, it is this: Rape cases are winnable. Serial rapists could be swept from the streets and untold numbers of women could escape the worst moments of their life, if police and prosecutors would suspend their disbelief.

This article is part of our project “The Presence of Justice,” which is supported by a grant from the John D. and Catherine T. MacArthur Foundation’s Safety and Justice Challenge.

Barbara Bradley Hagerty is a contributing writer at The Atlantic. She is the author of Life Reimagined: The Science, Art, and Opportunity of Midlife.