At a meeting last fall of several hundred university administrators who were gathered to discuss campus sexual assault, a Department of Justice official repeated a frightening statistic. “We know that the majority of rapes are committed by serial rapists, and those folks are very unlikely to be reached by any prevention messages that we’re going to be sending out, or education about rape,” said Bea Hanson, who works in the DOJ’s Office of Violence Against Women and served on a White House task force on the issue of campus rape.
Hanson was referring to an influential study, published in 2002 by David Lisak, then a professor of psychology at the University of Massachusetts Boston, and Paul Miller, then a clinical psychologist at Brown University School of Medicine. After surveying nearly 2,000 male students at a midsize, urban commuter university in Boston, Lisak and Miller found that of the approximately 6 percent of men who admitted to rape or attempted rape, a startling 63 percent reported committing more than one rape, with an average of six rape acts each. These numbers, Lisak wrote later, point to a “reality in which the vast majority of [campus] rapes are committed by serial, violent predators.”
Over the past few years, the data from Lisak and Miller’s 2002 study has become ubiquitous, cited in countless news reports and advocacy briefs, and even appearing in a report by the White House Council on Women and Girls. Recently, it has been used to argue for harsher punishments — and even jail time — for student rapists, whose cases have traditionally been handled through university judicial systems. Some high-profile universities have implemented “zero-tolerance” policies — where expulsion is the mandatory or preferred punishment for sexual assault — as a way to crack down on offenders. In an op-ed in The New York Times last fall, Yale Law School professor Jed Rubenfeld argued, “Even expulsion is radically deficient. It leaves serial rapists free to rape elsewhere, while their crimes are kept private under confidentiality rules.”
But new research suggests that this “serial rape assumption” may need some rethinking. In a paper published Monday in JAMA Pediatrics, a group of sexual assault researchers led by Kevin Swartout, an assistant professor of psychology at Georgia State University, used longitudinal data to track more than 1,000 male students at two southeastern universities over four years. Using the FBI’s definition of rape, the researchers found a higher proportion of men — 10.8 percent of the total sample, nearly twice as high as the Lisak/Miller study — who would be considered rapists. This suggests that the problem is far more widespread than the older study indicated.
Among the men who reported acts of rape in the new study, only about 25 percent said they had committed those acts over multiple college years. That’s radically different from the 63 percent who admitted to multiple acts in Lisak’s study. The men were also unlikely to have reported committing rape before entering college, suggesting there’s little evidence for the existence of a large group of offenders who perpetrate rape over a long period of time. Instead, the men who committed rape fell into three distinct categories: those who committed rape before entering college but were unlikely to rape once they got to campus; those who were unlikely to rape before entering college and began to rape once they arrived; and a much larger group of men who didn’t commit rape at all, or who committed one rape.1
Lisak didn’t respond to requests for comment for this story, but we did correspond last April for a different story I was writing about sexual assault. He told me over email that picking apart the limitations of a single survey was a distraction. “Whatever the precise percentage of sexual assaults being committed by serial offenders, the fundamental math of this phenomenon dictates that serial offenders represent a very significant part of the problem,” he said. “It is important to recognize this phenomenon because a serial offender does represent a different kind of risk to a community.”
Swartout contends, though, that the heightened attention on Lisak’s research has created a dangerous myopia. “There’s an assumption that if we can stop one particular group of serial rapists, that will solve the problem of campus sexual assault,” Swartout said. “Our findings aren’t in line with that strategy. We found that there are multiple, distinct groups of men who are perpetrating sexual assault on campus.”
Swartout and his colleagues’ research was designed to investigate serial rape specifically, not instances of rape overall. It seems like a strange distinction, but the most important question is how many men are committing rape repeatedly over time. Social scientists disagree about the definition of “serial rape” — especially how many rape acts an offender needs to commit to merit the label — but from the standpoint of prevention, pinning down the specific number of rape acts is less important than figuring out how many men are perpetrating rape, and when.
To look into that, Swartout and his colleagues used two sets of longitudinal data drawn from public universities in the Southeast. For both groups, they used the same methodology and the same survey, the Sexual Experiences Survey, which was created as a tool for measuring rape victimization in the late 1970s. The first set of data was gathered on one campus from August 1990 to April 1995, with a sample of 851 men, and the second was gathered on a different but demographically similar campus from March 2008 to May 2011, with a sample of 795 men.2
Importantly, though, instead of classifying the men by the number of specific acts of rape they reported, as Lisak did, each man was coded based on whether he had committed rape at all during each year of data collection.
Swartout and his colleagues focused their analysis on the second data set, because it is substantially more recent than the first set. They found that of the 105 men who admitted to rape in that sample, 37 percent fell into the “decreasing” trajectory (39 men), which meant that they admitted to rape in high school but didn’t report continuing to rape throughout college (the vast majority — 87 percent — of this group committed rape during one or fewer college years). Eighteen percent of the self-reported rapists (19 men) fell into the “increasing” trajectory, which meant that all but two did not begin committing rape until they arrived at college (nearly all of this group committed rape in multiple college years, primarily as sophomores, juniors and seniors). Most (75 percent) of the men who admitted to rape during college did so during one year.
“Based on this data, if universities tailor their sexual assault intervention strategies to fit the assumption that most rape is being perpetrated by serial offenders, you’ll miss three-quarters of the people who are committing rape on college campuses,” Swartout said.
There are, of course, some limitations to Swartout’s study. The sample size of students who admitted to rape is small — 177 of the 1,645 surveyed (105 of the 795 in the second data set) — and the universities are in the same region of the country, which limits generalizability. And although the sample of male students matched the demographics of the universities they attended, the respondents were overwhelmingly white.
Finding reliable numbers on sexual assault perpetration is notoriously difficult to do. Christopher Krebs, a senior research social scientist at RTI International and the lead author of one of the most frequently cited sexual assault surveys conducted in the past decade, hadn’t seen the new study, but told me that his attempts to pin down a believable perpetration number have always fallen short. “You’re asking a college student about some very serious interpersonal crimes. It would defy reality to think that all of them are going to be honest,” he said. He added, however, that the Sexual Experiences Survey — which he does not use in his research — has been known to produce higher rates of perpetration than other surveys.
The biggest challenge for Swartout and his team, though, was overcoming a shortcoming in the Lisak/Miller study: its inability to tell when — or under what circumstances — the rape acts occurred. This is a familiar problem for most sexual assault researchers. Respondents are unlikely to tell the truth if surveys ask outright whether they’ve committed rape or been raped; often, they don’t think about their experiences in those terms. Instead, researchers use a series of specific, behaviorally based questions — something like “Did you perform oral sex on someone without their consent?” These questions are the best way to get respondents, who may not admit to rape outright, to be honest about their actions, but the surveys include such a wide range of behaviors that several rape acts could be part of the same assault.
“In a single assault, a man could use alcohol and bodily force against his victim,” Swartout said. “Those would get coded as two separate rape acts.” For that reason, it’s impossible to tell whether the average of six rape acts committed by the “serial rapists” in Lisak and Miller’s study are distinct assaults, committed at different times or with different victims, or several acts that were part of the same rape.
Using a longitudinal approach helped Swartout and his co-authors minimize — if not solve — this problem because it took the question of specific rape acts out of the equation. The multi-act rapists in the new survey had to admit to rape at separate points in time, making it clear that they had committed multiple, distinct assaults.
“Longitudinal data is like an accordion,” Swartout explained. “You can unpack what’s happening when.” Collecting data all at once, as Lisak did, is more like a snapshot. “It collapses events that have happened over a long time frame and makes them look like they all occurred at once.”
Andra Tharp, a senior adviser for the Air Force’s Sexual Assault Prevention and Response division and a former research scientist at the Centers for Disease Control and Prevention, said using longitudinal data produces a more nuanced result. “By drawing out the data over time,” she said, “you see all these different degrees of behavior. And what you don’t see is a large group of chronic perpetrators.”
The new findings could have serious implications for how sexual assault educators on college campuses tackle the problem, as well as for the kinds of policies that the federal government might require and support. “If you think about most rapes as these cold-blooded, premeditated acts by predators who can’t be changed, that will logically lead to a criminal-justice-focused approach, basically trying to root out a few bad apples,” Tharp said. “But if it’s mostly sporadic and opportunistic behavior, we need to think more about prevention and intervention — a broader public health approach instead of focusing primarily on a few high-risk individuals.”
From Swartout’s perspective, the new research is just the beginning. “What we want to know now is what’s behind these trajectories,” he said. If administrators and policymakers start thinking about perpetrators as a more diverse group, he hopes future research can begin to determine what’s motivating — and what might deter — campus rapists.
But perhaps the biggest lesson from this new study is the danger of relying on one piece of research for complex policy responses. As alarming as the Lisak research might seem, it offered a relatively simple solution: Isolate the handful of young men who were committing the vast majority of campus rapes, and expel them or turn them over to law enforcement. Swartout’s findings offer fewer easy answers. Instead, they point to a more complicated reality, where more men are committing rape and their behavior is harder to predict. Then again, if researchers can figure out why these patterns are occurring, it will be easier to craft a solution that can influence an entire campus — and not just target a handful of men.
CORRECTION (July 13, 12:15 p.m.): An earlier version of this article incorrectly said the Sexual Experiences Survey has been known to produce higher response rates from rape perpetrators. It should have said the survey has been known to produce higher rates of perpetration.