How Big a Problem Is Sexual Assault on Campus?
A closer look at the numbers we have How many women are raped at college in America every year? The answer matters to parents thinking about where (and whether) to send their daughters off to live on campus,...
A closer look at the numbers we have How many women are raped at college in America every year? The answer matters to parents thinking about where (and whether) to send their daughters off to live on campus, to taxpayers who subsidize universities with hundreds of billions of dollars a year and to university leaders who wonder how much to worry about this problem versus a long list of other concerns like affordability, free speech, AI, demographic cliffs and declining enrollments, and so on.
Right now, no one can be confident they have even a ballpark answer to this question. One imperfect source of data comes from victim reports to campus authorities, which schools then have to report publicly as part of the federal government’s Clery Act . If some victims choose not to report, which some almost certainly do, those figures will be too low.
The other imperfect data source we have is student surveys, which have all the problems of surveys. Having two data sources that are both imperfect wouldn’t be so bad if they agreed with one another, but it turns out the best surveys we have, carried out by one of the nation’s leading survey research firms (Westat) on behalf of the Association of American Universities (AAU) back in 2019, imply a per-capita rate of campus rapes that is over 20 times what’s in the Clery Act data for the same period.
One might have hoped that such a big discrepancy, together with the importance of the topic, would have led to a careful forensic examination of the data to figure out what’s really going on. What we’ve gotten instead is largely a new front in the culture wars. Five minutes on the internet is all it takes to find headlines like “The latest big sexual assault survey is (like others) more hype than science,” and book reviews charging “implicit misogyny ... vastly overstated claims, questionable statistics and quotes massaged beyond their original meaning.” But what would a careful forensic examination of the data suggest about which data source is right (or at least less wrong)?
I try to answer that question here by carrying out a validation exercise of the surveys to make sense of which data source gets us closer to the true number, and what that true number might be. When pollsters try to validate political surveys, they compare who survey respondents say they’ll vote for with some objective “ground truth” — what the final voting results turned out to be.
My version of that exercise is to compare the number of women who report in surveys that they were raped and reported it to authorities — and so, if true, would show up in the official Clery Act data — with the actual Clery Act data. By my estimation, we cannot reject the idea that the AAU survey data are roughly correct.
That is, the number of rapes we’d expect to see in the Clery Act data if the AAU surveys were correct is within the margin of error of what we actually see in the Clery Act, after accounting for the fact that the Clery Act data themselves include some measurement error as well: Reported figures for campus rapes tend to be far higher during years when schools are having their data audited by the U.S.
Department of Education than when they are not , for example. What does this imply about the prevalence of rape on campus? The AAU survey data can only speak to what is going on with the sort of four-year, selective-enrollment university that participates in the AAU survey ( these types of schools enroll around 16% of all college students ).
With that caveat in mind, the surveys suggest that for this type of school, by the time they graduate about 1 in 5 women will be raped, defined as vaginal or anal penetration or oral sex using force, threat of force or inability to consent (for example, because the woman is unconscious).
We could also look at sexual assault, which is harder to validate given the greater room for ambiguity and variability in what types of events are to be included. In any case the AAU survey implies that 1 in 3 women will be either raped or subject to “sexual touching” — for example, pushed up against the wall of a dorm or fraternity house and felt up, kissed or “grabbed by the pussy” against her will — by graduation.
Imperfect data source #1: Official reports in the Clery Act data One way to measure how many women are raped on campus is to look at how many report that to campus officials or the police. Those are the numbers the federal government’s Jeanne Clery Campus Safety Act (“Clery Act”) requires to be reported by any college or university that gets federal funding.
These numbers get a lot of attention partly because they’re available for every school — every school has to report them — and because they seem like “hard data.” The problem is that many of the victims of all sorts of different crimes choose not to report to authorities, which means official reports understate the true number of crimes.
Consider, for example, the National Crime Victimization Survey (NCVS), sponsored by the U.S. Department of Justice, which is a survey of the general public’s overall experiences with crime. NCVS respondents who say they’ve been the victim of a crime are asked whether they reported it to the police.
While the numbers change a bit year to year, the top-line finding is most victims even of violent crimes (52.1%) choose not to report to the police. For rape specifically (rapes that occur anywhere in society, not limited to campus), more than three-quarters of victims say they don’t report.
Why do so many rape victims not report? Many victims say they’re skeptical any good will come of it. The data suggest there’s something to that concern: If the victim does report to police, the chance the offender gets arrested and sentenced to any time behind bars is about 4% (that is, under 1 in 20. (The figure comes from multiplying the share of sexual crimes that result in an arrest — around 27% — by the share that someone arrested for rape gets convicted and sentenced to prison, which is not more than around 15%.) At the same time, the costs of reporting can be high.
Three-quarters of rape victims have PTSD within a month of being raped, according to a 2023 study by Emily Dworkin, Anna Jaffe and Michele Bedard-Gilligan. Symptoms can include recurrent intrusive memories, nightmares, flashbacks that can make it feel as if the rape is happening again, dissociative amnesia, depression, self-blame, fear, apathy, sleeplessness, bursts of anger and self-destructive behavior.
Many victims abuse drugs or alcohol to “numb out” and avoid thinking about what happened. A year later, nearly half (41.5%) still have PTSD. Reporting requires women to think and talk about their rape again and go through an adversarial process the other side has incentives to make as unpleasant as possible.
It is possible that reporting rates could be even lower to campus authorities if students are skeptical that the university will do anything in response; for example, it was recently reported that 1,200 sexual assault, dating and domestic violence, and stalking cases were reported to Cornell campus authorities between 2018 and 2024 and just seven resulted in expulsion or dismissal.
A third key reason many victims say they don’t report is because what they experienced doesn’t match the stereotypical view of rape as something that is committed by a stranger with extreme force. Women are less likely to consider the rape a rape if the offender is an acquaintance or their boyfriend or if the woman was drunk, since she may attribute what happened partly to her own actions. (These factors are potentially particularly important for rapes that happen on college campuses.) For example, consider the account of one woman who did not consider her experience to be a rape: He took me to his bedroom and I layed down because I was tired but then he got on top of me and started kissing me ...
Then he started going a little farther, and before I knew it he was taking my clothes off. I remember thinking that I didn’t really want to continue, but I couldn’t really do anything to stop him. I was so drunk it just seemed pointless to argue, or put up any fight ...
Note this would be legally classified as rape in most states if she was deemed to have been too drunk to consent, but not in states like New York if she voluntarily became intoxicated. A different problem with the Clery Act data, beyond victim non-reporting, is that many rapes and sexual assaults that should be included in the data don’t seem to be.
For example, one study found that in the year a university’s Clery Act data are audited by the U.S. Department of Education, the number of events recorded goes up by 45% . The year after the audit, the reported numbers go right back to what they were in the pre-audit year.
Presumably this is not about the tendency of victims to report more in audit years than others, but rather more about how universities aggregate and publicly report information on the rapes and assaults that do get reported by victims. Of course there will be some student reports of rape that turn out not to be rape, which all else equal would lead the Clery Act data to be too high.
But probably much more common and important are the various reasons to worry that the official Clery Act data will understate not just the true number of rapes on campus (because of victim non-reporting) but may even also understate the number of rapes on campus that are officially reported.
Imperfect data source #2: Student surveys The only alternative to the Clery Act data is to rely on student surveys, which are subject to all the challenges of carrying out any survey — plus some that are specific to this sensitive topic. The 2019 AAU survey was administered online by Westat, a leading survey research firm.
All enrolled students received an email inviting them to participate (21 schools sent the emails themselves; at 12 schools, the email came from Westat). Students received between three and eight follow-up contacts, depending on the school. Many of the schools offered a modest incentive to participate, like the chance to win an Amazon gift card worth between $5 and $20.
Students were asked about rape and sexual assault both since the start of the current year (roughly analogous to how the Clery Act reports by calendar year) and since enrollment in college. One challenge is getting a representative sample of people to respond to the survey, which is of particular concern given that the response rates for these sorts of surveys can be low.
For example, the response rate for the AAU’s 2019 survey was 22% of all students at the participating schools. If sexual assault survivors are more likely to respond than others, the survey results may overstate prevalence rates for the campus as a whole. A study by Westat, the survey firm that carried out the AAU survey, suggests this is true to some extent but that the size of the bias that results may be modest.
Their analysis comes from, for example, comparing late responders to the survey (more reluctant respondents who required more follow-up contact attempts before they answered) with early responders, or comparing people who answer under incentivized versus non-incentivized conditions.
A different concern is the possibility of ambiguously worded survey questions. For example, an overly broad or vague question might lead people to report on a mix of more serious and less serious events that then make it hard to tell how many truly serious events happened. To avoid that problem, the 2019 AAU survey was very specific and defined terms in detail.
Rape was defined as vaginal or anal penetration or oral sex, by force, threat of force or because the victim could not consent. Physical force was itself separately defined. The survey also asked about sexual assault as well, which they defined as including not just penetration but also “sexual touching,” defined as kissing, “touching someone’s breast, chest, crotch, groin, or buttocks,” or “grabbing, groping, or rubbing against the other in a sexual way, even if the touching is over the other’s clothes.” By comparison the FBI’s Uniform Crime Report definition of sexual assault consists of rape, statutory rape, incest and “fondling,” defined as “touching of the private body parts of another person for the purpose of sexual gratification, without the consent of the victim, including instances where the victim is incapable of giving consent because of his/her age or because of his/her temporary or permanent mental incapacity.” A different challenge is respondent misreporting.
For example, previous research by myself and others has shown that a small share of people will say yes to literally anything on a survey question — whether they’re not paying attention, they’re delusional or some other reason, this will lead to false positives. The flip side is if survivors feel that victimization is embarrassing or stigmatizing and leads them to not report honestly, so-called “social desirability bias” will lead to false negatives that contribute to underestimation.
This is all to say: Whether the survey results are, on net, too high or too low can’t be predicted by just reasoning alone. It’s ultimately an empirical question. I try to answer that question next.
Continue reading
Watch a short ad to unlock the full article
The rest stays locked if you skip or close the ad early.
Validation exercise To get a sense of the accuracy or inaccuracy of the surveys, comparing answers to the survey about something observable in the world as a “ground truth” benchmark for the same set of schools provides a true apples-to-apples comparison. I do this by comparing the AAU survey results for students who say they were raped since the start of the current academic year, that the rape happened in a location that means the school is supposed to include the event as part of the Clery Act requirements, and that the student reported the event to campus authorities.
This sounds simple, but there are a number of important details here that are easy to get wrong. For starters, the set of schools included in the AAU survey, just 33 universities out of some 5,700 in the U.S. in 2019, is not representative of all post-secondary institutions. The AAU sample is more likely to include four-year residential colleges.
Because so many students are raped in university dorms and fraternity houses, we’d expect AAU schools to have higher rates of rape victimization than the broader set of schools that include, for instance, those where many students live at home and commute to school. That means comparing the survey results of the AAU schools to the Clery Act data from all schools, which is what I believe was essentially the method of some previous validation exercises such as the one by Stuart Taylor, as reported in a Washington Post op-ed , compares apples to oranges.
Another thing we need to attend to is the fact that the benchmark here — the Clery Act data — are themselves subject to some measurement error. As noted above, a school’s Clery Act figures are 45% higher during years when the U.S. Department of Education is auditing the school’s numbers.
That means during the non-audited years, a student who was raped and reported this to school authorities might not show up in the Clery Act data because the school is, for whatever reason, not always including such in their official Clery Act statistics. A third thing we need to account for is statistical sampling uncertainty.
Even a perfectly designed and executed survey will not always get a perfectly representative set of people to participate just by the luck of the draw. That’s why every presidential opinion poll, for instance, always has some margin of statistical uncertainty (like “candidate Jones is up 3 points over candidate Smith, plus or minus 2 percentage points”).
The second figure, “Adjusted for under-reporting, Clery Act counts overlap the range the AAU survey implies,” shows what happens when we carry out a validation exercise comparing the number of rapes we’d expect to see in the Clery Act data compared to what we actually see in the Clery Act data, accounting for an apples-to-apples comparison (comparing AAU school surveys to AAU school Clery Act data), measurement error in the Clery Act data themselves, and statistical sampling uncertainty.
The first row shows my upper bound for the number of rapes the AAU survey implies we should see in the Clery Act data, together with its 95% confidence interval. This upper bound assumes that every program or resource that the student says they reported to is a campus safety authority that forwards the information for Clery Act reporting purposes. (The AAU survey is imperfect at capturing this; see the appendix.) The second row in the figure is my lower bound for this figure, which assumes that every student who reported to a “counselor” was excluded from the Clery Act data given the counselor exception. (I realize not all counselor reports will be exempt, but the data do not allow me to distinguish between those that are versus are not, which is why this is a lower bound.) The final two rows in this figure show the Clery Act data for these AAU schools, first taking the results at face value, and then adjusting for underreporting using the results from Yung (2015).
Intuitively, readers can interpret the Clery Act data taken at face value essentially as something like “what the schools report in years they are not being audited by the U.S. Department of Education,” while the adjusted figure can be taken as “what schools would report if they were being audited.” Figure 2 shows that there is overlap between what I think is the most defensible range for the Clery Act data (the upper and lower bounds of the 95% confidence interval for the Clery Act data adjusted for university underreporting) and the range of the number of reports we’d expect to see in the Clery Act data according to the 2019 AAU survey (defined by the upper end of the 95% confidence interval around my upper-bound estimate and the lower end of the 95% confidence interval around my lower-bound estimate).
Put differently, the total number of rapes in the AAU survey (whether reported or not) was 20 times higher than the number of officially reported rapes in the Clery Act data. The comparison in Figure 2, suggesting the number of officially reported rapes in the AAU survey is in the ballpark of the actual number of officially reported rapes, makes me think the AAU surveys are far closer to the true number of rapes on campus every year than the Clery Act data are.
Implications What do the AAU surveys imply about the number of women raped every year at the type of four-year school included in that survey sample? I find cumulative risk easier to think about myself than annual risk, partly because annual risk numbers mix together the extremely high-risk period of a woman’s first year in college with the (relatively) lower risks of subsequent years.
The AAU survey results suggest that the share of female undergraduates who will be raped by graduation at this type of school is 18% (with a 95% confidence interval from 17% to 19%; see Cantor et al., 2020, Table 15). If we expand the definition to sexual assault generally, the AAU survey implies 33% of women will be either raped or subject to sexual touching against their will by graduation (95% confidence interval from 32% to 34%; Cantor et al., 2020, Table 14).
So what now? For starters, a useful thought exercise may be to ask: Suppose everyone involved in the higher-education enterprise — students, parents, policymakers, taxpayers, university presidents and provosts — knew these numbers and understood they are likely to be close to the true number of rapes and sexual assaults that happen every year to students at schools like this.
Would people start acting differently? If the answer is yes, then several practical questions naturally follow. For example, how do we get these numbers into wider circulation so everyone knows them?
Imagine, for instance, they had to be included in the list of statistics schools give out to families on the college tour (in addition to the number of majors the school offers and its student-faculty ratios, etc.). Or imagine these figures were incorporated as another factor into the U.S.
News rankings. I suspect everyone would know these numbers fairly quickly. The harder challenge is how to do good surveys at every higher-educational institution in the country, not just the well-resourced sort of schools that are in the AAU.
That’s a hard challenge partly because a bad survey is much easier to do than a good one, and partly because one might wonder about how incentivized universities in general are to collect good survey data that might make the safety environment on their campuses look bad. This might be a case where there could be high returns to some federal government requirements.
Thanks to Abigail Adams, David Cantor, Bridget Collier, Elizabeth Glazer and Eric Heath for helpful comments. Thanks to Kate Daugherty, John Greer and Javier Lopez for excellent assistance with the data. Any errors and all opinions are my own.
Appendix This appendix provides details on my comparison of: The 2019 AAU survey’s estimates for the number of female students who say they were raped and reported that to campus authorities, so we would expect that to show up in the Clery Act data for these schools (the survey results are reported in Cantor et al., 2020).
The number of rapes reported in the Clery Act data by these schools. Details on 2019 AAU Survey Results The 33 schools participating in the 2019 AAU survey were: Boston University; Brown; Caltech; Carnegie Mellon; Case Western; Georgetown (the only non-AAU-member school in the sample); Harvard; Iowa State; Johns Hopkins; MIT; Northwestern; Ohio State; Rice; Stanford; Texas A&M; University of Arizona; University of Chicago; University of Kansas; University of North Carolina at Chapel Hill; University of Wisconsin-Madison; University of Florida; University of Michigan; University of Minnesota, Twin Cities; University of Missouri; University of Oregon; University of Pennsylvania; University of Pittsburgh; University of Rochester; University of Southern California; University of Virginia; Vanderbilt University; Washington University in St.
Louis; and Yale. It is important to emphasize that this AAU survey sample is clearly not representative of all 5,700-plus institutions of higher education in the U.S. 1 For starters, the AAU sample consists disproportionately of four-year residential colleges, which is relevant because a large share of all the rapes and sexual assaults that happen to college students occur on campus (locations like dorms and fraternity houses).
So it will be very important to compare the implications of the AAU survey reports to Clery Act data just from those AAU survey schools, rather than to Clery Act data from all colleges (which would create an apples-to-oranges comparison). The details of my validation exercise are as follows.
For the convenience of readers who wish to re-do this exercise in different ways, I include the table numbers from Cantor et al. (2020)’s report on the AAU survey findings for each of the results that I draw from. Estimating annual prevalence of rape to undergraduate and graduate students Because the Clery Act data do not distinguish either the victim’s gender or enrollment status (undergraduate or graduate), I focus on calculating the fraction of all students (male and female, undergraduate and graduate) who report being raped since the start of the 2018-2019 academic year in the 2019 AAU survey.
The AAU survey questions about rape are as follows: Since you have been attending [University], has someone used physical force or threats of physical force to do the following with you: Sexual penetration. When one person puts a penis, fingers, or object inside someone else’s vagina or anus, or Oral sex.
When someone’s mouth or tongue makes contact with someone else’s genitals Since you have been attending [University], has any of the following happened to you while you were unable to consent or stop what was happening because you were passed out, asleep or incapacitated due to drugs or alcohol: Sexual penetration.
When one person puts a penis, finger, or object inside someone else’s vagina or anus Oral sex. When someone’s mouth or tongue makes contact with someone else’s genitals The AAU survey avoided using potentially ambiguous terms like “rape,” and defined terms like Physical force : “...someone holding you down with his or her body weight, pinning your arms, hitting or kicking you, or using or threatening to use a weapon against you.” The survey clarifies: “The sexual behavior may have been performed on you or you may have been made to perform the sexual behaviors on another person.” Those questions seem to capture events that align closely with how the Clery Act defines rape, which is based on the FBI’s definition for its Uniform Crime Report system: “The penetration, no matter how slight, of the vagina or anus with any body part or object, or oral penetration by a sex organ of another person, without the consent of a victim.” The AAU survey also asked about sexual assault as well, which they defined for respondents to include not just penetration but also “sexual touching,” defined as kissing, “touching someone’s breast, chest, crotch, groin, or buttocks,” or “grabbing, groping, or rubbing against the other in a sexual way, even if the touching is over the other’s clothes.” 2 Data on the number of students by gender and enrollment status come from Cantor et al. (2020) Table 2, while data on the annual prevalence of rapes for female students come from Tables 5 (undergraduate) and 6 (graduate) and for men, Tables 7 (undergraduate) and 8 (graduate).
Women Men % (std error) # students # rapes % (std error) # students # rapes Undergrad 0.047 (0.001) 272,939 12,828 0.012 (0.001) 257,458 3,089 Grad 0.012 (0.001) 152,119 1,825 0.003 (0.001) 148,450 445 Total 14,654 3,535 Note these results imply that the annual prevalence of rape to students every year is equal to (14,654+3,535) / (830,966) = 18,189 / 830,966 = 0.022 (or 22 per 1,000).
Note that in principle one could do some adjustment to these figures to account for the fact that they are trying to capture the prevalence of rape during the 2018-2019 academic year but come from surveys that were administered mostly in March-April 2019. That means the survey recall period does not include events that happen at the very end of the school year, and so one could potentially inflate these figures somewhat to account for this omission to get a full calendar-year estimate.
But the available data (see for example Krebs et al. 2016) suggest that the prevalence of rape to postsecondary students is disproportionately concentrated at the start of the school year, rather than the end of the year, especially for first-year students, the group that has the highest risk of rape victimization.
So it seems likely that the AAU survey misses only a modest share of events that happen over the course of the entire academic year. Estimating share of rapes that happen in Clery Act geography We next calculate what share of the rapes reported in the AAU survey occurred, according to the survey respondent, in a geography that is covered by the Clery Act — that is, occurred in a place where the school is required to count and report this (on campus or affiliated property).
Table 17 of Cantor et al. (2020) pools together graduate and undergraduate students, but we are able to focus on acts of penetration specifically (so specific to our definition of rape) and calculate this separately for men and women. Women Men % (std error) # rapes # in Clery Act geography % (std error) # rapes # in Clery Act geography Ugrad + Grad 0.545 (0.007) 14,654 7,986 0.481 (0.019) 3,535 1,700 Estimating share of rapes in Clery Act geographies that get reported Table 21 of Cantor et al. (2020) also pools together graduate and undergraduate students, but allows us to calculate what share of students who report they were raped (penetration) reported that to any sort of campus resource or program.
We initially assume here that the events that get reported to literally any campus resource or program will wind up in the Clery Act data; we relax this assumption on the next panel of my calculations. Part of the challenge in figuring out whether the program or resource the student reported to would be counted as a campus safety authority and so have to report for Clery Act purposes comes from the fact that the AAU survey question allows people to say they reported to more than one program (so the percentages for each program or resource listed add up to more than 100%), and we cannot determine which students reported to multiple programs or resources and which combination they reported to (Cantor et al., 2020, Table 44).
Another limitation is this figure pools together different types of offenses, and pools both men and women together as well as undergraduate and graduate students. Women Men % (std error) # rapes in CA geo. # in CA geo. & reported % (std error) # rapes in CA geo. # in CA geo. & reported Reported?
0.295 (0.006) 7,986 2,356 0.178 (0.013) 1,700 303 Summing the figure here for men and women together yields an upper bound estimate for the number of rapes we should expect to see in the Clery Act data if the AAU survey reports were accurate. This is an upper bound because it assumes that every campus program or resource that any student reports to would forward the information to campus authorities for inclusion in the Clery Act reporting, even though we know that some programs and resources are exempt from that type of requirement (for example, pastoral counseling or professional counselors).
We can also calculate a standard error for this estimate using the standard errors for each of the component survey results together with the formulas for the variances of the sum of random variables and the product of random variables (assuming each survey result is independent).
Our upper bound estimate is equal to 2,659, with a standard error of 80 and a 95% CI that ranges from 2,502 to 2,816. Estimating share of reported rapes potentially exempt from Clery Act reporting Finally, we can estimate a lower bound for the number of rapes the AAU surveys imply we should see in the Clery Act data by assuming that everyone who says they reported to a counselor (Table 44, Cantor et al.
2020) would be exempt from Clery Act reporting requirements. This is a lower bound estimate because some respondents who report to a counselor also reported to some other campus authority (like campus police) that were required to report as part of the Clery Act, plus some of the counseling resources themselves that students reported to may have served in roles that make them campus safety authorities and so had to report (for example, counselors who also serve in other roles on campus).
The share of students who say they reported to a counselor as one of the programs or resources they reported to is 0.468 (standard error 0.004), pooling genders, types of misconduct events, and graduate/undergraduate enrollment status. This implies at most 1,244 reported rape events would have been exempt from Clery Act reporting requirements.
The implication is our lower bound estimate is 1,414, with a standard error of 44 and a 95% CI that ranges from 1,328 to 1,500. Clery Act Data Estimates How do these survey results compare to the officially reported Clery Act data? One issue to deal with for starters is that the time period covered by the AAU survey is the 2018-2019 academic year, while the Clery Act data are reported by calendar year instead.
Our default is to compare the AAU survey figures for the 2018-2019 academic year with the Clery Act figures for calendar year 2018. That comparison assumes that the prevalence of rape and sexual assault on campus in spring 2019 (the one academic semester or quarter of non-overlap in the AAU survey) is roughly similar to the prevalence in spring 2018 (the non-overlap period in the Clery Act data).
For the 33 schools in the AAU survey, there were 847 total rape incidents reported across all these campuses together for calendar year 2018 (the number is roughly similar for 2019). That means for starters that the annual prevalence of rape to students in these schools implied by the Clery Act data is 847 / 830,966 = 0.001 (or 1 per 1,000).
This comparison to the AAU survey result presented above is the basis for our claim that the AAU survey results are 22 times as high as the official Clery Act-reported data. The challenge is the Clery Act data may understate the number of rapes reported by students to campus authorities that are required to forward the information to be included in the official Clery Act totals.
Some evidence to this effect comes from the study by Yung (2015), who draws on data from a sample of 269 four-year postsecondary institutions, 31 of which had their Clery Act data audited by the U.S. Department of Education at some point over the study period (2001-2012). The study finds that relative to the year before the audit is conducted, during the audit period the number of serious sexual assaults included in the school’s official Clery Act report increases by 45% (the 95% confidence interval that I approximated from Yung, 2015, Figure 5 ranges from 0% to 90%).
The year after the audit, the figure drops again to the same level as it was the year before the audit. This is not literally a “smoking gun” of malfeasance by these schools, but malfeasance is certainly one plausible interpretation (and perhaps the most straightforward possibility in an Occam’s razor sense).
These results, if taken literally, imply that the number of rapes that should have been included in the Clery Act data at the set of schools in the AAU survey equals 1.45 x (847) = 1,228, with a 95% confidence interval that ranges from 858 to 1,608. Notes 1 https://nces.ed.gov/fastfacts/display.asp?id=1122 2 By comparison the FBI’s UCR definition of sexual assault consists of rape, statutory rape, incest and “fondling,” defined as “touching of the private body parts of another person for the purpose of sexual gratification, without the consent of the victim, including instances where the victim is incapable of giving consent because of his/her age or because of his/her temporary or permanent mental incapacity.” Jens Ludwig is the Edwin A. and Betty L.
Bergman Distinguished Service Professor at the University of Chicago and Pritzker Director of the Crime Lab. He co-directs the Education Lab and NBER’s working group on economics of crime.
Article text via FreeNewsAPI. Rights remain with Vital City.
Read on publisher site → Opens Vital City in a new tab
