Research on False Rape Reports:

I've been doing some reading on the debate about the incidence of false rape reports. I've looked at a lot of studies on this, and hope to blog some more about it later (short summary: estimates range from under 2% to 40+%, though I have no opinion about which is right). But in the meantime, I thought I'd mention one observation that may be helpful for thinking about other debates as well.

Many people who believe that false rape reports are a tiny fraction of all rape reports argue that very few women would make such false reports. The common line is that women don't lie about rape, which must really mean that very few women lie about rape.

But even if this is true — and I strongly suspect that it is — this is entirely consistent with the possibility that a substantial fraction of rape reports are false. Let's say, for instance, that only 2% of all women age 16-19 would ever lie about rape; and that any particular year, only 2% of that tiny fraction actually do falsely report a rape to the police. So 98% of all women (including relatively young and not very mature women) would never lie about rape, and even of those who might under the right circumstances, most never will. (I use the 16-to-19 age group because the risk of rape is highest there; the same analysis could apply, though, to other age groups.)

There are, however, about 8 million women in the 16-to-19 age group in the U.S., and 2% x 2% x 8 million = 3200 false rape reports per year. The National Crime Victimization Survey (2002 data, see table 3) reports that 2.7 out of 1000 people age 16 to 19, which means 5.4 out of 1000 women age 16 to 19, are raped each year. This is an estimate based on a survey, not on police reports, and it may well be low (the actual rate may be higher); but in any event, we know that the rate of rapes reported to the police is roughly half that estimated to the NCVS (compare the Uniform Crime Reports data, and remember that the UCR data aggregates rapes and attempted rapes, while the NCVS breaks them out). This means that roughly 2.7 out of 1000 women age 16 to 19 report an actual rape each year, for a total of 2.7/1000 x 8 million = 21,600 true rape reports per year.

Under this model, then, 13% of all rape reports to the police would be false (in the 16-to-19 age group), even though only 2% of all women in that age group would ever make a false rape report, and only 2% of those actually make a false rape report each year. Ninety-eight percent of all women may be completely truthful on this subject, and yet we may still have a substantial false rape report rate.

This, of course, is just a model, based on numbers picked out of thin air. Maybe, for instance, the fraction of women who'd ever make a false rape report is much lower than 2%, or maybe it's higher. We can't know for sure.

But the model does illustrate that it's perfectly possible to believe that (1) only a tiny fraction of women would ever lie about being raped, (2) a huge fraction of rapes are unreported (quite possibly even more than 50%, so that rape may be a highly underreported crime by many women, as well as overreported by a few), and yet (3) a substantial fraction of rape reports to the police are false.

Some people who worry about false rape reports may in fact believe that women are psychologically wired to lie about such things; I'm certainly not one, but historically that has been the view of some, to which others have understandably reacted with hostility. That may be why some people take the opposite view: Instead of "women often lie, so the false rape report is very high," they say "women very rarely lie, so the false rape report is very low." But that doesn't follow. False rape reports, however rare they may be as a fraction of all women might still be substantial as a fraction of all rape reports.

UPDATE: Just to make it clear, I am not talking here about reports of ambiguous situations, where the complainant sincerely believes the action was rape, but it turns out that it wasn't. I'm discussing in this post outright fabrications — claims of rape that the complainant fabricated because of a desire for revenge, sympathy, money, an explanation for consensual but adulterous or otherwise socially unacceptable sex, or something else. (In some such claims the complainant may have had consensual sex with the accused, and in others there may have been no sex, but the important point is that the complainant knows there was no nonconsensual sex.) The Kanin study, which I hope to blog about in a few weeks, reports on police accounts that assert such behavior has taken place. Likewise, unless I'm mistaken it's generally believed that the infamous Socttsboro boys case involved a fabricated rape account, though the motive there may have been different from the ones identify.

I stress again that I'm not making any assertions about how common such false accounts actually are. They may be an extremely low fraction of all rape reports, or they may be a substantial minority. My point in the original post was simply to say that one can believe that very few women would ever lie this way, and yet have such false reports be a substantial fraction of all rape reports. My point in this update is to make clear that the problem I discuss here is one of fabrication — not something that can be solved, for instance, by clarifying the definition of rape to remove possible ambiguities.