This article contains graphic descriptions of sexual assault.

Three years ago, Kiyona Phillips, now 31, bumped into a Facebook acquaintance in the early hours of a Sunday morning, at the tail end of a rare night out on the town. Phillips was leaving the since-shuttered Tuscana Lounge in downtown Washington, D.C., where she had gone with friends who had persuaded her to join them to celebrate Howard University’s homecoming weekend. Outside the club, Phillips saw Daniel (not his real name), with whom she had exchanged some awkward messages about a year earlier. In person, he was charming and warm, so when he asked Phillips for her phone number, she gave it to him. And when he texted her shortly afterward, at around 4.30 a.m., inviting her to meet him for breakfast, she was flattered.

Daniel picked up Phillips and took her to the Denny’s on Bladensburg Road in Northeast D.C., where they shared a romantic breakfast. After breakfast, Daniel drove Phillips back to her house in Southeast D.C.

Parked outside Phillips’ house in Daniel’s two-door, the two started making out.

The circumstances of what happened next is where Phillips’ and Daniel’s accounts dramatically diverge, lapsing into the “he said, she said” quagmire that dooms so many sexual assault investigations and prosecutions in the United States.

The “he said” version of what happened in that car was that Phillips was “all over” Daniel; he said that she masturbated his penis, that she wanted the sex that followed.

The “she said” version is that, after an initial, mutual kiss, Daniel grabbed Phillips’ breasts and, even after she told him to stop, forced himself across and into the passenger seat, tugged off her pantyhose and underwear, pinned her knees to her chest, and vaginally raped her. She struggled, but he overwhelmed her. She couldn’t shove him off.

Phillips promptly reported the rape to police—a rare action for most rape victims—and within hours, found herself submitting to a lengthy sexual assault forensic exam at MedStar Washington Hospital Center, an experience she told Rewire was painful and humiliating. It was also ultimately futile: Based on responses to a public records request that Phillips later submitted in frustration over the lack of action in her case, with the help of the nonprofit Network for Victim Recovery of DC, she believes that her kit was never even processed.

(Thanks to sexual assault reforms in D.C., enacted after Phillips’ alleged assault, officials now must notify victims when their rape kits are tested. Furthermore, officials with D.C.’s sexual assault forensic nurse examiner program say there is currently no backlog of rape kits in the District.)

Ultimately, the Metropolitan Police Department of the District of Columbia (MPD) detective tasked with Phillips’ case, William Weeks, as well as the prosecutor to whom Weeks presented Phillips’ case, believed the “he said” version of the story, and Daniel was never charged.

“I honestly understand why a lot of people don’t report rape,” Phillips told Rewire in an interview. “Because if you’re not half-dead or beat in the face or it’s not recorded in any kind of way, you really have to prove that this happened to you.”

Neither Weeks nor the MPD would comment on Phillips’ case for this report. (According to a recent investigation by the Washington Post, Weeks was assigned to the property crimes unit last fall, having been transferred first from MPD’s Youth Investigations Division, and then from its Sex Assault Unit. The Post reported on a family’s accusation that Weeks fabricated the police report of an 11-year-old who was allegedly raped by multiple adult men in 2008 but was jailed thanks to Weeks’ report that she invented the assaults.)

The fact that Phillips underwent a forensic exam should, at least in theory, have helped move her case out of the “he said, she said” cul-de-sac, by including evidence about trauma to her vagina.

However, an investigation by Rewire shows that the fate of Phillips’ complaint fits the pattern of how the “he said, she said” dilemma has outgrown the realm of legitimate legal query, and has instead come to justify the systemic failure of police and prosecutors nationwide to properly process forensic evidence that could lead to more sexual assault convictions, and also to identify serial rapists who otherwise remain at large. Investigators’ unwillingness to push past the superficial appearance of a “he said, she said” case has also frustrated federal lawmakers who have appropriated more than $1 billion trying to reduce rape kit backlogs. Additionally, researchers say that police attitudes toward sexual assault are lagging social and scientific understanding of evidence in sexual assault cases.

“Since so many sexual assailants are serial offenders … the DNA from a rape kit is often the material difference between a sexual predator going to jail or remaining free to reoffend,” said U.S. Sen. John Cornyn (R-TX) at a Senate Judiciary subcommittee hearing he convened last month to discuss the state of rape kit backlogs. “When rape kits remain untested and sitting on a shelf, the consequences can be nothing short of devastating.”

Cornyn has been one of the leading congressional champions for tackling the persistent backlogs of untested rape kits—estimated to be in the hundreds of thousands by sexual assault advocates and experts—that are stored in police facilities throughout the country, and has been one of the loudest voices in supporting additional resources to do so.

New research, however, casts light on why the backlog problem hasn’t been solved despite the money already made available for addressing it (though, as Rewire recently reported, there are issues with how federal money aimed at reducing rape kit backlogs has been appropriated and monitored).

The answer lies in attacking the “he said, she said” quandary. Specifically, experts are now beginning to understand that, in general, all rape kits should be tested, whereas police frequently—and wrongly—only test those where they decide that the case does not turn on consent, or where the assailant’s identity is in question.

These decisions are often influenced by overt or unconscious victim-blaming on the part of police, or ignorance about how trauma influences a victim’s conduct both during and after an attack, experts told Rewire.

The result is that huge swathes of rape kits go untested. And that, of course, permits rapists to go on to assault more victims, and demoralizes and discourages victims from attempting to seek justice.

“Virtually No Investigation”

Rape continues to be what many researchers agree is an under-prosecuted and under-reported crime. Criminology professors Cassia Spohn and Katharine Tellis in their 2012 study on prosecution rates in Los Angeles city and county found that of the more than 5,000 rapes and attempted rapes reported to the Los Angeles Police Department between 2005 and 2009, only 12 percent of these cases—about 590—resulted in the arrest of the suspect. Of those arrested, 390 defendants were convicted; 59 percent of those convicted were sentenced to prison, 37 percent were put on probation, and 4 percent received a jail sentence. Another 2012 study, by criminal justice professors Megan Alderden and Sarah Ullman, found that only about 10 percent of the 465 rape and sexual assault crimes reported to a large, unnamed Midwestern police department in 2003 resulted in the prosecutor’s office approving felony charges (the researchers did not report how many of the charges resulted in convictions or jail time).

Rape is notoriously difficult to prosecute because of the requirement to prove that the victim did not consent to the sex act that occurred. Often, there are only two witnesses to the acts in question: the accuser and the alleged suspect. In addition, in many cases the issue of consent boils down to whether or not the prosecution can prove that the suspect fully understood that the accuser did not consent to the sex. The cornerstones of the criminal law—the presumption of innocence and the insistence that it is up to the prosecution to prove all elements of a crime beyond a reasonable doubt—have the unfortunate result that many rapists escape justice.

Some legal reforms have attempted to improve the prosecution and conviction rates for this common crime. In the mid-seventies, U.S. jurisdictions threw out laws that required victims to prove they attempted to resist their attackers, as well as laws that allowed for third-party testimony to determine whether the victim had consented to sex in previous situations.

Nevertheless, the thorny issue of consent continues to lead many criminal investigators not only to close cases prematurely, but to decline to test the forensic evidence submitted in these cases.

If the cases hinges on whether or not the accuser consented to the sex, the logic often goes, what is the point in spending limited resources to process and test DNA?

Yet, that logic relies on a fundamental lack of understanding about how rape kits can assist in the investigation of an alleged attack.

The contents of a rape kit are wide-ranging and varying and obviously depend on each specific case. Rape kits can include a perpetrator’s semen, skin cells, blood; they can include a victim’s blood or urine showing she was drugged; swabs collected from the victim’s body can show physical trauma in her vagina or rectum; and bruises across her body can show she was strangled or otherwise subdued during the attack, which can help prosecutors bring additional charges against the rapist.

But it’s not just the evidence found on the woman’s body and her clothes and bed sheets that can instruct investigators, but also where the evidence was located. Male skin cells found in a victim’s vagina could support her story that she was digitally penetrated; blood or skin under a victim’s fingernails could corroborate another account; and tears in her or his anus could help prove another. The next step in the process is for a crime lab to try to create a DNA profile of the suspect and upload it to a state or federal criminal database in order to match it to a convict or another suspect, or to a future suspect.

So, while rape kits can help identify suspects—which can solve cases but also give victims peace of mind if their rapist is already in prison—they have other important purposes: They can corroborate the victim’s version of events, and they can also exonerate innocent suspects or convicts. They cannot definitively prove that someone was raped, and were never intended to do so, but that doesn’t negate their value both in individual cases and in identifying repeat attackers.

However, evidence is now emerging that puts hard numbers on how many kits go untested due to the misconception that rape kit evidence is of no use when the case could turn on the issue of consent.

In the summer of 2009, representatives from the Michigan State Police, the Detroit Police Department, and the local prosecutor’s office toured a remote police property storage facility to take stock of all of the evidence in local police custody. During that tour, an assistant prosecutor discovered storage boxes stuffed with what turned out to be more than 11,000 untested rape kits.

When the kits were finally tested, investigators found matches with the DNA of rapists who had gone on to rape multiple women. These serial rapists included “stranger rapists,” who rape people they have never met before, and “date” or “acquaintance” rapists.

The discovery of Detroit’s untested kits sparked a four-year study funded by the U.S. Department of Justice (DOJ) to establish what had led to this backlog. The study, published in March, found that many kits languished simply because police officers did not believe the victims, and chose not to test their rape kits. (A similar DOJ-funded study was conducted in Houston, which around the same time revealed a backlog of more than 6,600 kits.)

The discovery that victim-blaming attitudes dissuaded police and prosecutors from testing kits was not shocking to the study’s lead researcher, Rebecca Campbell, a professor of ecological-community psychology at Michigan State University. But what did strike her was how slim many of the Detroit police officers’ investigation reports in these sexual assault cases were. Campbell told Rewire that in many of the cases her team studied, there was “virtually no investigation” conducted by police, particularly in cases where the suspect said the sex was consensual.

“We could hold reports up that were a single page, and that was it,” she said. “There was no investigation ever taken on the case.”

Campbell said her work confirmed that many police officers neglected to test rape kits because they did not fully understand how sexual assault kits could aid investigations. The department was plagued by limited resources, and rather than spend the hefty expense to have kits tested and analyzed at the crime lab, officers reserved testing kits for when they were sure the cases would go to trial, rather than primarily using the kits to bolster investigations of sexual assault cases.

“When you look at the police reports associated with the kits that were not tested, you see pervasive and rampant victim-blaming, assuming that victims were prostitutes, blaming them for what happened, calling them derogatory names,” Campbell said. “They didn’t test the kits because they didn’t believe the victim, because the victim didn’t act ‘right,’ didn’t behave in a way that they thought they should have if this were a real sexual assault. … The problem was, they didn’t think the victims were credible the vast majority of the time.”

This turned out to be a fatal flaw. The 1,595 Detroit rape kits that were tested yielded 455 hits in a federal criminal database to crimes around the country that included but were not limited to rapes, according to the study’s final report. Researchers were able to identify 127 “serial sexual assaults” (cases involving an offender who sexually assaulted more than one victim), and significantly they found that these serial rapes involved perpetrators who knew their victims, as well as so-called stranger rapists. In fact, the researchers were able to determine that these rape kits were just as likely to produce a hit for an offender or crime scene in both stranger and non-stranger cases.

Campbell said that at least for the Detroit sexual assault kits, it proved to be more economical and effective to test all of the kits, rather than to prioritize stranger kits or another category. But this is not only true for Detroit: Now that more cities are testing their backlogs due to increased public pressure, researchers are finding that both stranger and non-stranger rapists tend to be serial rapists. Most recently, this has proven to be the case in Cleveland, Ohio, where Cleveland Plain Dealer reporters Rachel Dissell and Leila Atassi’s years-long investigative work led police to start counting their untested rape kits. In 2011, according to an interview with the reporters, Cleveland police discovered about 4,000 untested rape kits, and began sending them out for DNA testing. Authorities have so far identified more than 200 serial rapists, whom they believe are responsible for at least 600 rapes, according to the Plain Dealer. The paper’s reporting led to a new state law, enacted last year, which mandates testing for all rape kits, old and new.

New Insights Help Explain Victim Behavior

Part of the reason behind victim-blaming attitudes that persist throughout America’s criminal justice system, experts say, is the ignorance of how trauma affects victims’ behaviors—both during and after the assault. The reality is that many sexual assaults are much more nuanced than the type of stranger rapes in the woods that are the default on popular media. In the real world, many rape victims do not fight back, and many give inconsistent accounts of what happened. There is extensive research on the neurobiology of trauma, which attributes these types of behaviors to various chemicals released by the brain, which can cause victims to fight or to freeze (what’s known as “tonic immobility”) or to experience the assault in hazy fragments, especially if they are intoxicated.

Explaining this phenomenon to juries and judges has become an important tool to try to quell victim-blaming bias, said Jennifer Long, the director and co-founder of AEquitas, a global organization based in D.C. providing resources to prosecutors worldwide in an effort to improve justice in crimes that primarily affect women, such as sexual violence, intimate partner violence, stalking, and human trafficking.

A former Philadelphia prosecutor, Long told Rewire that expert testimony on the neurobiology of trauma is admissible in sexual assault cases in every state now. Pennsylvania was the last state, in 2012, to enact a law allowing this type of testimony, and is currently awaiting the results of a constitutional challenge from the Pennsylvania Supreme Court. But despite the extensive research on the science of trauma, Long said victim-blaming is a persistent contributor to what researchers say are high rates of attrition when it comes to sexual assault cases in America.

“There’s a disconnect between the rapes happening and the criminal justice response to them,” Long said. “And the key piece that comes up again and again and again is victim-blaming, that the victim is being blamed somehow for either inviting the attack or being disbelieved because they engaged in some behavior that is misunderstood.”

Of course, giving evidence in court is one challenge; communicating these complex scientific theories to the hundreds of thousands of individual police and investigators who decide whether or not to test a rape kit presents a challenge of a different order.

Momentum for Reform

Of course, victim-blaming attitudes aside, many sexual assault advocates agree that police departments’ lack of funding is also a huge issue contributing to rape kit backlogs.

After all, processing and testing rape kits is an expensive endeavor, particularly after cities discover huge piles of kits and are then pressured to test them all at once. Unable to process all the rape kits in-house, many police departments end up outsourcing the testing to private labs. Two of the biggest labs that contract with many different cities are Bode Technology in Lorton, Virginia, and Sorenson Forensics in Salt Lake City. Sorenson Forensics actually increased its workforce and lab facilities in 2013 to accommodate the increased demand from cities to test forgotten rape kits. Sorenson Forensics spokeswoman Camilla Green told Rewire that their average rates are $500 to $1,000 per kit depending on the scope of the work required. She said that is the industry standard.

But at last month’s Senate Judiciary subcommittee hearing on rape kit backlogs, the advocates and state law-enforcement officials who were called to testify instructed senators that additional funding to test rape kits is just one among several needed reforms in the handling of rape kits. Witnesses suggested that law-enforcement agencies need to start proactively doing what Detroit’s law enforcement officials did by accident: auditing their police storage facilities for untested and forgotten rape kits.

Several states have passed laws requiring these type of audits, beginning with Illinois and Texas in 2011. New York-based sexual assault advocacy group the Joyful Heart Foundation believes that once more cities start auditing their backlogs, only then will we understand how big this problem really is.

The nonprofit has over the last few years embarked on its own private endeavor to identify backlogs using public records requests and recently uncovered a combined total of more than 9,000 kits in Charlotte, North Carolina; Jacksonville, Florida; Kansas City, Missouri; Portland, Oregon; and San Diego, California.

Advocacy groups like the Joyful Heart Foundation say more money needs to go toward helping jurisdictions deal with these massive backlogs once they are discovered, which is why they worked with the federal government to help create and advocate for a new $41 million federal grant program within the Bureau of Justice Assistance that was designed to help jurisdictions not just test backlogged rape kits, but also use the money to reform how they investigate and prosecute sexual assault crimes. President Obama’s budget request for 2016 includes another $41 million to continue the program next year, as well as $20 million for the Department of Justice’s National Institute of Justice to research ways to reduce the backlog of sexual assault kits. This month the House of Representatives approved a 2016 spending bill that includes the $41 million grant program, plus an additional $4 million; the Senate is still considering the bill. The New York County District Attorney’s Office has also created a $35 million grant program to help other jurisdictions eliminate their backlogs.

“Testing rape kits is only one step toward comprehensive reform,” Joyful Heart Foundation Managing Director Sarah Haacke Byrd told Rewire in an email. “Once the problem is acknowledged and the first kits are sent out for testing, cities are left to grapple with the enormous task of finding a way to test all of the rape kits in their storage facilities, and figuring out how to investigate and prosecute these cases, re-engage survivors in the process and address any systemic failures that led to the creation of the problem in the first place.”

Witnesses at last month’s congressional hearing also stressed the need for laws mandating that police departments test all rape kits (several states have begun passing such laws) and that more cities need to implement multidisciplinary programs often called Sexual Assault Response Teams (SART) or Sexual Assault Nurse Examiner (SANE) programs, which connect victims with victim advocates and specially trained sexual assault forensic nurses. Research shows that these types of programs increase the likelihood that sexual assault cases will be prosecuted and also improve health outcomes for victims. According to the DOJ, there are more than 600 SANE programs throughout the United States and Canada, but they are still absent in many cities.

Amid these reforms, researchers like Rebecca Campbell hope the criminal justice system works on improving rape investigation and prosecutions generally, rather than focusing solely on testing rape kits.

“Money for rape kit testing is important; there’s no two ways about it,” Campbell said. “But the issue here is not just testing. Testing is one piece of an interconnected puzzle. So, you could test a whole bunch of kits, but if your investigators and your prosecutors don’t do anything with it, then people sort of scratch their heads and say, was this a reasonable use of public funds? Well, it would be if we had the follow-up investigation, prosecutions, and support to survivors. Because then we have perpetrators off the street, we’re preventing crime, and that is a significant issue for public safety.”