A new report issued by the Justice Department’s Bureau of Justice Statistics provides wildly different estimates of sexual assault than the Obama administration’s frequently cited 1-in-5 statistic. That number, which was calculated in a 2007 Justice Department study on sexual assault, has been offered as justification for a recent intense focus on addressing campus sexual assault, largely under the terms of federal Title IX legislation.
How do campus rape statistics from the surveys compare?
Justice Department Campus Sexual Assault Study (CSA): 20% of female students
Justice Department National Crime Victimization Survey (NCVS): 0.6% of female students
No, that is not a typo! The same government department has produced one estimate 33 times higher than a different estimate. Naturally, partisan opinionators on both sides of the issue have been quick to defend whichever stat fits their narrative best. Some tout the “new,” low estimate as proof that campus rape is a myth while others stand by the original figure and insist campus rape is an epidemic.
Who’s right in the Justice Department?
I was so baffled by this discrepancy that I waded into the weeds to compare the methodology and findings for each report. I doubt it will surprise anyone to learn that this is very much an “apples and oranges” comparison. In the interest of increasing understanding of what is really happening among college students, I offer an objective analysis.
The objectives of the surveys are distinctly at odds.
The objectives of each of the reports are very different. In anticipation of this, each addresses the other, saying that the CSA addresses sexual assault on college campuses from a public health perspective, while the NCVS uses a criminal justice perspective, with the primary aim of comparing students and non-students in the general population, rather than establishing precise estimates of sexual assault and rape.
The NCVS provides a more detailed explanation:
“The NCVS is an omnibus survey designed to collect information on experiences with a broad range of crimes. It is likewise presented to respondents as a survey about criminal victimization.
…The NCVS collects information on nonfatal personal crimes (rape or sexual assault, robbery, aggravated and simple assault, and personal larceny) and household property crimes (burglary, motor vehicle theft, and other theft).
Because victims of rape or sexual assault may not consider their victimization a crime, this context could discourage or suppress recall and reporting of those incidents. Additionally, because the NCVS covers a wide range of criminal victimization, the number of screening questions related to rape and sexual assault are limited.
Methodological differences that lead to higher estimates of rape and sexual assault in the CSA should not affect the NCVS comparisons between groups.”
In contrast, the CSA study focused exclusively on sexual assault in depth:
“The CSA Study was undertaken specifically to document the prevalence of distinct types of sexual assault among university women (with “types” defined by how the assault was achieved, such as the use of physical force or incapacitation of the victim due to drugs or alcohol), as well as the context, consequences, and reporting of distinct types of sexual assault among a large sample of undergraduate women from two large universities.
In the CSA Study, sexual assault includes a wide range of victimizations, including rape and other types of unwanted sexual contact (e.g., sexual battery).”
The Definition of Sexual Assault differs in the two surveys:
The CSA definition of sexual assault:
- forced touching of a sexual nature (includes forcible kissing, fondling, and grabbing)
- oral sex
- sexual intercourse
- anal sex
- sexual penetration with a finger or object
The NCVS definition of sexual assault:
Rape: Penetration with force or threat of force against someone’s will. Includes vaginal, oral or anal penetration.
Sexual Assault: Attacks or attempted attacks with unwanted sexual contact.
- Includes forcible fondling and grabbing, but not forcible kissing.
Only the CSA addresses incapacitation.
Both surveys include “unwanted sexual contact due to physical force.” Only the CSA measures “unable to provide consent due to incapacitation.”
It should be noted that this includes not just voluntary drinking, but also the deliberate administration of alcohol or drugs to render the other party incapacitated for the purpose of having sex, e.g. roofies, Everclear punch. This behavior is not captured in the NCVS.
The surveys word the questions very differently.
The NCVS specifies that it is gathering crime data and asks two questions re rape and sexual assault as part of the larger interview:
1. Other than any incidents already mentioned, has anyone attacked or threatened you in any of these ways: any rape, attempted rape, or other type of sexual attack?
2. Incidents involving forced or unwanted sexual acts are often difficult to talk about. (Other than any incidents already mentioned), have you been forced or coerced to engage in unwanted sexual activity?
Note that no specific definitions of sexual assault or rape are offered – it is left completely up to the respondent.
If the answer is affirmative, then the NCVS asks for the following details:
- Physical injuries sustained
- Presence of a weapon
- Offender characteristics
- Reporting to police
The CSA identifies as a survey about public health and “uses behaviorally specific questions to ascertain whether the respondent experienced rape or sexual assault. It asks about an exhaustive list of explicit types of unwanted sexual contact a victim may have experienced, such as being made to perform or receive anal or oral sex.”
CSA Initial Questions:
1. Has anyone had sexual contact with you by using physical force or threatening to physically harm you?
2. Has anyone attempted but not succeeded in having sexual contact with you by using or threatening to use physical force against you?
3. Has someone had sexual contact with you when you were unable to provide consent or stop what was happening because you were passed out, drugged, drunk, incapacitated, or asleep? This question asks about incidents that you are certain happened.
4. Have you suspected that someone has had sexual contact with you when you were unable to provide consent or stop what was happening because you were passed out,
drugged, drunk, incapacitated, or asleep? This question asks about events that you think (but are not certain) happened.
For example, a woman whose boyfriend has forced her to perform anal sex may or may not classify that activity as rape or assault on the NCVS survey. In contrast, in the CSA survey she would respond “yes” to that specific question, and it would be included in the tally for sexual assault.
In short, the NCVS allows the victim to dictate whether a crime has been committed, whereas the CSA makes that judgment call. We can see a very similar dynamic in the work of David Lisak, who has extensively studied sexual assault from the male’s perspective. He found that men will not describe themselves as rapists, but willingly admit to performing behaviors that clearly constitute rape:
Lisak avoided the use of terms such as “rape,” “assault,” and “abuse,” instead describing in detail the behavior in question, without applying labels that the perpetrators might not identify with.
Although the situations described are legally rape, Lisak found the men were not reluctant to talk about them, seeing them as sexual conquests to brag about, and did not think of themselves as rapists; according to Lisak, such men are narcissistic and “like nothing better” than to talk about their “sexual exploits.”
Here are the questions Lisak asked:
“Have you ever had sexual intercourse with an adult when they didn’t want to because you used physical force [twisting their arm, holding them down, etc.] if they didn’t cooperate?”
“Have you ever had sexual intercourse with someone, even though they did not want to, because they were too intoxicated (on drugs or alcohol) to resist your sexual advances, (e.g. removing their clothes)?”
Similarly, in the CSA we see that many of the women whose experiences clearly constituted sexual assault or rape did not perceive it that way. This is evident in the responses women gave for not reporting their experiences:
|Unwanted sexual activity while incapacitated||Unwanted sexual activity via physical force|
|Fear of reprisal||55%||18%|
|Not serious enough to report||67%||56%|
|Did not have proof||79%||26%|
|Unclear crime was committed||35%||35%|
|Did not think police would take seriously||58%||22%|
|Didn’t think anything could be done to assailant||40%||23%|
|Didn’t want anyone to know about substance use||53%||8%|
|Didn’t want anyone to find out||29%||42%|
|Felt responsible for being attacked||50%||n/a|
However, in the NCVS many victims who clearly identified as crime victims also did not report their attacks:
|Sexual assault or rape|
|“It’s a personal matter”||26%|
|Fear of reprisal||20%|
|“It’s not important enough”||12%|
This suggests that women and men both need better education in the area of sexual assault, so that both fully understand what behaviors constitute a crime.
The methodologies varied widely in other ways.
- Cross-section data collection of 6,800 undergraduates (F=5,466, M=1,375)
- Two large public universities (one in the South, one in the Midwest).
- Anonymous web-based survey of 15-20 minutes duration.
The CSA had a low 42% response rate, which they address here:
“The overall response rates for survey completion for the undergraduate women sampled at the two universities were 42.2% and 42.8%, respectively. It is logical to surmise that students who did not participate in a survey about sexual assault may differ from those who did participate.
However, the reasons for nonresponse could affect prevalence estimates in opposing ways. Some nonrespondents (nonvictims) may have chosen not to participate because they felt that they had no relevant experience, whereas other respondents (victims) may have chosen not to participate because they anticipated that taking the survey might be upsetting to them…Other modes of data collection would not have given respondents the same degree of anonymity and privacy…which are associated with more accurate reporting of sensitive behaviors.
…To reduce nonresponse bias and increase sample representativeness, weights adjusting for nonresponse were developed using a Generalized Exponential Model. Cohen’s effect size was used as a measure of the magnitude of the bias, and we added weights for university, gender, year of study, and race/ethnicity, which reduced bias to negligible levels.”
- Nationally representative sample of households, sample size not provided.
- Includes both students and non-students. Students included those enrolled in colleges, vocational schools and trade schools.
- 7 interviews per person over 4 year period.
- First is in-person and 6 follow-ups are conducted by phone every six months.
The overall response rate was 74%. The NCVS acknowledges the difficulty of non-anonymous, in-person reporting:
“The NCVS may be more subject to interviewer effects than the CSA.
…Victims may not be willing to reveal or share their experiences with an interviewer. The level and type of sexual violence reported by victims is sensitive to how items are worded, which definitions are used, the data collection mode, and a variety of other factors related to the interview process.”
Each of the surveys has its strengths and weaknesses.
1. The CSA was designed to better understand sexual assault on campus and recommend policy initiatives for combating it.
It takes the lack of education around sexual assault into account by defining assaultive behaviors.
The NCVS report splits off two questions from a much larger survey about various types of crime, and does not provide legal information.
2. The CSA was not nationally representative.
It may be that different types of colleges or schools in different areas could have different rates of assault.
3. The NCVS required the reporting of highly sensitive and personal information in a face-to-face interview, and continued over a period of 4 years.
Given the reasons women do not report sexual assaults, it seems likely this would have suppressed reporting considerably.
4. Many (including myself) would argue that “forcible kissing” does not belong in the same category of assault as rape.
Unfortunately, I was unable to retrieve CSA data pertaining to this specific accusation, so we don’t know what portion of assaults fall into this category.
5. Given the indisputable fact that alcohol consumption on campus increases sexual assault, it is reasonable to address it in any study of SA on college campuses.
The CSA wording specifies the victim being “unable to consent” or “stop what was happening.” In that sense, it appears to eliminate sexual encounters where both parties may be drunk but still capable of decision-making.
In contrast, the NCVS study does not include assaults committed after victims are plied with drugs or strong alcohol to render them incapacitated, a common practice at fraternity social events.
6. It is important to determine whether a crime has been committed even if the victim does not perceive it as such.
If a woman feels guilty after being raped while blackout drunk, has she not been victimized? If a woman is reluctant to report her long-term boyfriend for forcibly penetrating her anus, has she not been raped? These assaults would have been likely to be reported in the CSA, and very unlikely to show up in the NCVS.
The goal of the CSA study is to determine the prevalence of assault, which is completely different from the prevalence of discipline for assault. It is well understood in both studies that most victims do not report sexual assault.
In addition, setting incapacitation as the bar for sexual assault likely means that men on campus are certainly being assaulted by women, even though such reports by men are extremely rare.
Wandering into the murky waters of who was incapacitated vs. who was drinking but not incapacitated is unlikely to yield good or workable policies. In my view, the only solution is to stop binge drinking, something schools (and students) have little stomach for. One step that could be taken immediately is the outlawing of any drinks using Everclear or other potent alcohols, which are always offered in order to get girls drunk.
Neither of these surveys stands alone as a comprehensive estimate of the prevalence of sexual assault on college campuses. The NCVS was not designed to do so, and is informative in an extremely limited way. Conversely, the CSA is too inclusive, and should at the least split out more serious offenses from less serious ones like forcible kissing.
By the way, there are three red herrings showing up in partisan articles about this subject, all of which should be dismissed here. The first is the finding that reported sexual assault has been decreasing over time. That is true, but we don’t know why. It may be part of a trend of decreasing violent crime. It may also be that victims feel so hopeless about receiving justice they don’t bother. Until there are systems in place for investigating claims fairly and disciplining offenders, we can’t really use reporting data to measure sexual assault. This is especially true in view of the NCVS interviewing techniques.
Other articles are quick to point out that non-students are assaulted more than students, proving that campus rape is a myth. Again, this is not relevant to the comparison. The NCVS was designed to compare students and non-students, while the CSA focused only on students. (And the two surveys defined students differently.)
My own sense is that sexual assault is certainly not limited to college campuses. Hookup culture and fraternity parties may encourage it, but there may be other forms of encouragement in the non-student population. That’s something we don’t yet know.
Finally, the issue of false rape claims is irrelevant to both of these surveys, as neither of them deal with the consequences to attackers.