Whenever a high-profile account of alleged campus sexual assault comes crashing down – such as Rolling Stone's gang-rape accusation – activists predictably fall back on the claim that only 2 percent of rape accusations are false.

This isn't accurate. First, the 2 percent figure refers to false reports made to the police. Making a false police report carries a penalty, which exists to deter people from doing so (although sometimes that penalty isn't enforced, such as with the Duke Lacrosse rape hoaxer). No such penalty exists on college campuses. (Indeed, even the accuser in the Rolling Stone article, though proven to have lied, did not face any punishment from the school.)

Not having that penalty is meant to make accusers feel more comfortable coming forward, although it's difficult to see how being punished for lying would make truthful victims fearful. Regardless, a lack of consequences makes it easier for student accusers to come forward and punish fellow students who may have hurt them or with whom they had a previous regretted encounter that has come to be seen as assault.

But let's look at that 2 percent figure as if it applied to all cases. If it were true that 2 percent of rape reports were proven to be false, that would indicate that 98 percent were proven to be true, and therefore, accusers should be believed as a matter of default.

The 2 percent figure is the low end of a range (usually up to 8 percent or 10 percent) from multiple studies conducted over the past few decades. The Washington Post, in an article fact-checking a graphic about rape statistics, mentioned a study from the "Making A Difference" Project, which the National Center for the Prosecution of Violence Against Women claimed was the "only research conducted in the U.S. to evaluate the percentage of false reports made to law enforcements."

Assuming that's true (it's not), it would be the only relevant study to use as a basis for false accusations. But that study actually found 7.1 percent of rape reports to be "unfounded/false." But one can't assume that means 92.9 percent of reports were true. The study found an additional 8.5 percent to be classified as "unfounded/baseless." End Violence Against Women International, which produced the MAD study, defined baseless reports as "those that do not meet the elements of the offense and those that were improperly coded as a sexual assault in the first place." This includes sexual assault reports in which a "follow-up investigation reveals either that no crime occurred or that some other type of crime was actually committed (or attempted)." In other words, a false report of rape.

That would bring the number of false rape reports up to 15.6 percent. Another category of rape reports in the MAD study (making up 17.9 percent of the cases) was "closed as an informational report." This categorization occurred when a person reported a sexual assault that did not meet the criteria of a crime. It doesn't mean the person made up the entire story, but what they reported wasn't a crime. This wouldn't be counted as false, but it also isn't in the "true" category, as a crime wasn't actually committed. "Suspended/inactivated" constituted another 28.6 percent of the cases studied by MAD. These are the cases in which there isn't enough evidence to say whether a crime actually occurred.

The next category was "exceptionally cleared," making up 17.9 percent of the total cases studied. EVAW defines these cases as those where an offender is identified, there is enough evidence for arrest but something "beyond the control of law enforcement precludes arresting," such as the death of the suspect or a lack of victim cooperation after the identification.

The final category, 20 percent of the total cases, was those "cleared by arrest." This would seem to indicate that at least 20 percent of rape reports were true, but remember, an arrest does not equal guilt. Another chart in the MAD study has the breakdown of outcomes after the case was referred to prosecution. A whopping one-third (33.2 percent) of cases were rejected by the prosecution, another 20.9 percent were dismissed with all charges dropped and 2.3 percent were dismissed but the charges weren't dropped. Two percent of the cases referred to prosecutors resulted in a trial and a "not guilty" finding. Another third (33.2 percent) resulted in a guilty plea, and just 5.9 percent went to trial and received a "guilty" finding (2.6 percent were classified as "other").

So what does this all mean? Francis Walker, who runs an exceptional blog taking down statistics such as the MAD study, has done the math. He found that from all the statistics listed above, just 7.8 percent of rape reports in the MAD study could be classified as true.

"From this we can see that 39.1% of the cases end in either a guilty verdict or a guilty plea. Multiplying that by the 20% of police cases that result in arrest, we are left with the 7.8% I used at the start of my first post," Walker wrote. "Even this isn't a good number to use though. If a confession that a report is false isn't enough to classify the report as false, then the corollary is that a guilty plea isn't enough to classify a case as true. After all, it would not be difficut [sic] to imagine a scenario where, for any number of reasons, someone pleads guilty to a crime that they didn't commit. 20% x 5.9% leaves us with a 'true' rate of just 1.2%. Even if we decide to be generous and include not just the 20% arrested, but also the 17.9% exceptionally cleared, the number still only goes up to 2.2%."

From all of this one could determine that 15.6 of reports could reliably be determined as false, another 17.9 percent weren't actually crimes and just 1.2 percent (or 2.2 percent) could be reliably determined as true. The remainder would fall into a "we'll never know for sure" category.

Perhaps only 2 percent of rape reports are false. That doesn't necessarily mean that 98 percent are true. But assuming that is the case, one can't possibly know which category a report falls into until after a proper investigation (which would include due process). And to pretend that false reports don't happen just because they are rare minimizes the impact such reports have on those falsely accused.

And beyond all of this, none of this data can be applied to reports of campus sexual assault. There is no data available on the number of campus sexual assault accusations that turn out to be false, as it hasn't been studied. And again, without a penalty associated with false police reports, false accusations made to campus administrators are likely to increase.