Is the "one in five" campus sexual assault statistic a myth?

As someone who frequently watches CollegeHumor videos, I became interested into the apparent backlash against one of their recent videos about college sexual assault, entitled "What If Bears Killed One In Five People?" At the end of the video we are informed that by the time they graduate college, one in five women will be sexually assaulted. In this post I will try to answer the following questions:

1. Is the survey on which that claim is based flawed to the point where its conclusions are invalid?
2. To what extent, if at all, can its results be generalized to every college in the US?
3. What do more recent studies suggest about the prevalence of sexual assault in college women?

The answer to the first and second questions can be found in an excellent Washington Post article by Glenn Kessler. Other criticisms can be found here and here. With respect to flaws in the study (which was called the "Campus Sexual Assault Study" (or CSAS) and released in 2007), they include:

a) The survey's high non-response rate, which may have led to a bias where those who had been victims of sexual assault were more likely to complete the questionnaire.
b) Using an overly broad definition of sexual assault that includes "attempted forced kissing” and intimate interactions while intoxicated. (See the Time story for both of the above criticisms.) Two researchers who worked on the study also said that if you only consider "unwanted sexual penetration", the number drops to about 1 in 7. 

c) Overly broad wording that defines "incapacitation" too broadly, so that it includes not just being unconscious or barely conscious, but also being under the influence to the point of impaired judgment. This matters because 70% of the reported assaults in the study involved incapacitation by either alcohol or drugs.

With respect to the second question, the study in question only surveyed two (large) colleges, which makes it difficult to generalize to every college in the country or even to all other large colleges, as James Fox has pointed out. In fact, even two of the researchers involved in this study acknowledged that it was "not a nationally representative estimate of the prevalence of sexual assault".

3. A more recent survey by the Association of American Universities suggests the actual figure is 23% for unwanted sexual contact, or, for the more narrow category of such contact including penetration or oral sex, 11% of female college students. However, this survey has also been criticized by, among others, James Cantor, its co-principal investigator, who told Slate that the 27 colleges (hey! more than two, that's an improvement) it surveyed are not nationally representative. This problem is one it shares with the CSAS (see b) above). It also has some other problems in common with the CSAS, such as a low response rate (19%), which, as Cantor himself points out, may lead to just the type of bias discussed above under a). This survey has also been criticized for, like the CSAS, using a too-broad definition of sexual assault--it included in this category “sexual touching: touching someone’s breast, chest crotch groin, or buttocks—grabbing, groping or rubbing against the other in a sexual way, even if the touching is over the other’s clothes.”

Another more recent pair of surveys from both Rutgers and the University of Michigan also reached conclusions on par with "one in five": Rutgers' survey found that exactly this proportion of female undergrads had experienced "unwanted sexual contact" since arriving on campus, while UMich's found that "22 percent of female students said they had been assaulted in the last year." Notably, the Rutgers survey also found that almost 1/4 of its undergrad women were victims of "sexual violence" before they even arrived at college. However, it is worth noting that this survey also defined "sexual violence" very broadly: besides the obvious, um, actually violent stuff, it also included "persistent sexual advances or unwanted remarks about their physical appearance."

In conclusion, what we should all be able to agree on is that citing only the CSAS as proof that 1 in 5 American women is sexually assaulted while in college, as both Obama and Biden have done, is completely unjustified. Now, it is also clear that the available data from other surveys suffers from many and, in some cases, all of the same methodological problems as the CSAS (not being nationally representative, low response rates, and broad definitions of sexual assault). Unfortunately it seems that data that exists is too flawed to support claims about how often women in US colleges in general are sexually assaulted.