False Rape Accusations: Is There A Consensus?

When Dara Lind wrote at Vox that there was finally a consensus, it was a huge and critical point of dispute lifted off the discussion.  After all, among the various problems confronting the discussion of whether there is an “epidemic” of rape and sexual assault, most notably on college campuses, but in broader society as well, one of the backlash fears is that the evisceration of due process and shifting of burdens onto the accused is a terrible thing.

But how terrible?  The response of proponents of extreme change to grossly favor rape accusers was that false rape accusations very rarely happen. Almost never.  And so, we should default to believing that every accusation is true, because who would anyone falsely accuse someone of a rape?

Of course, the reaction was that it happens, and has happened in notorious cases, like Duke Lacrosse and more recently Jackie/UVA, Mattress Girl and others. But were these outliers? What was the frequency of false accusations? Dara Lind set out to narrow this down, if not exactly answer the question.

For one thing, research has finally nailed down a consistent range for how many reports of rape are false: somewhere between 2 and 8 percent, which is a lot narrower than the 1.5 percent to 90 percent range of the past.

If so, this could reshape the debate. Not that 2 to 8% was inconsequential, but it wasn’t 90% either, right?   So I immediately turned to my statistics guru, Francis Walker, and asked whether Dara got it right.  As it turns out, maybe not.

My first problem with this is that there aren’t any US studies that I am aware of that actually use the 2-8% range.  The only place I’ve seen that range used is in the The Voice article which, as I’ve previously discussed, isn’t exactly peer-reviewed research.  Even Lisak, who is a listed author of The Voice article says the range is wider at 2-10%. I asked Lind about this and here is her response:

A: […] I almost certainly relied too much on Lonsway, between the interview I conducted with her and her response to Lisak.

So we begin with a problem, that not only is there no consensus of 2 to 8 %, but in fact no study that ever said 2 to 8 % at all.  Rather, the Lisak study suggested 2 to 10%, but the two percent differential is, as it turns out, a relatively small problem.

To her credit, Dara Lind was very forthcoming about the bases upon which she drew, and to critique of her methodology in reaching the conclusion that there is a consensus.  Unfortunately, this also meant that there was no consensus, and that her 2-8% statistic was, at best, a particularly misleading stat.

People think the question these studies answer is “What percentage of rape reports are false?”  In reality, the question they are really answering is “What percentage of rape reports can we classify as false with a high degree of certainty?”  As a result, these studies don’t give us binary outcomes.  This isn’t necessarily a result of flawed design studies either.  Statistics isn’t a magical art form capable of diving absolute truth from thin air. When it comes to sexual assault, unless you physically were able to witness what happened, it can be very difficult to classify a report as either true or false.

To the extent that the 2-10% false accusation statistic is valid, which itself is subject to significant criticism, what it reflects are accusations that can be characterized as false with a sufficient degree of certainty, which is “a pretty high bar to clear.”  In the Lisak study, the specific number attributed to false accusations was 5.9%.

In contrast, accusations are assumed to be true if they weren’t proven false.

Next up is “Case did not proceed” which was the most used classification at 44.9% . . . Next is “Insufficient information to assign a category” at 13.9% . . . Finally, we have “Case proceeded” at 35.3%:

This classification was applied if, after an investigation, the report resulted in a referral for prosecution or disciplinary action or some other administrative action by the university (e.g., the victim elected not to pursue university sanctions, but the alleged perpetrator was barred from a particular building).

It may be tempting to view this category as “true,” but compare these criteria to the extremely strict definition of a false report.

In other words, the “true” accusations category was defined as cases that proceeded, which ignores that the commencement of an action, whether disciplinary or criminal, doesn’t make an allegation true, but merely “true enough” to pursue.  But even so, it still leaves a division as follows:

True-ish: 35.3%

Definitively False: 5.9%

Inconclusive: 58.8%

What this tells us is that there is no consensus, nor anything approaching a consensus, as to the percentage of false rape and sexual assault accusations, and that anyone suggesting otherwise is wrong, whether intentionally or ignorantly.

Don’t blame me. It’s just what the numbers say.

The question is whether this research is going to get acknowledged, or if false accusations are going to continue to be treated as an unknowable X-factor in rape cases. It used to be a genuine mystery — but we now know a few things. We know that police tend to overestimate how many allegations are false, and we’re moving toward a consensus about how frequent false allegations really are.

Understanding statistics is hard, but there is an answer to Lind’s question: the research is being acknowledged. It just doesn’t show what she thinks it does.

21 thoughts on “False Rape Accusations: Is There A Consensus?

  1. Sidney Carton

    Even if there were a consensus of 2%, or even 1%, if it’s >0, shouldn’t Blackstone’s formulation hold?

  2. John Thacker

    Indeed, and that’s the reason why, as she noted, the “definitely false” ones were disproportionately stranger rape situations where it could be definitely proven that no encounter took place. She makes a big deal that what she terms “gray rape” or “he said/she said” stories don’t show up as false, but they simply couldn’t by design.

    There were older studies that found a much higher “false” rate, up to 60-80%, because they used an entirely different methodology, claiming that anything where there wasn’t enough evidence to proceed was false. That’s not accurate either, some of those were true, we just don’t know and don’t want to charge such a serious crime without more certainty. Part of the difference too is that all of these studies just measure how things were believed by the police investigators at the time of the study.

    In other words, they assume that the police investigators of today are perfectly accurate in determining which charges are false and which are true (and which should proceed to trial), but then use that data to make a case for changing how the police judge things. If the current judgment is so perfect, why change?

    Even more so, the 2-10% range she gives is not different from the FBI’s statistics for other violent crimes like murder. So even granting the extremely dubious interpretation of the statistics, it only argues for weakening due process for defendants if she were equally “tough on crime” elsewhere.

    1. Patrick Maupin

      Ooooh, numerical masturbation! Can I play?

      Instead of “all unknowns are rape” or “all unknowns are false accusations”, you could make the gross assumption that, in most cases, the reason that you have an unknown is completely orthogonal to the question of whether that particular unknown was actually a rape or a false accusation. Using that assumption, you essentially toss out all the “inconclusives”, and just compare, e.g. the “definitively false” reports against the “true-ish” reports. E.g. 5.9 / (5.9 + 35.3) gives you the result that 14.3% of all rape reports that could be classified as “true-ish” or “definitively false” were “definitively false.”

      Of course, not knowing why rape reports fall into the “unknown” category, I have no idea whether this assumption is anywhere near reasonable, and, of course, statistics are completely meaningless to the actual victims, whether of rape or of false accusation of rape.

      1. SHG Post author

        Bear in mind that the truish are not subjected to the same rigors as the false, just to further screw with the worthlessness of the stats.

        1. Patrick Maupin

          Exactly. But that particular lack of information is also information. For example, if there were to be a consensus (and that’s what it’s all about, right?) that the 35.3/5.9/58.8 numbers are believable, and also that the proposed assumption that “inconclusive” have the same good/bad breakdowns as the “conclusive” is believable, then 14.3 percent becomes the lower believable bound of false accusations. Given the associated weakness of the conditions on the “True-ish” category, the actual value would almost certainly be quite a bit higher.

  3. REvers

    Stats posted as percentages need to be taken with a grain of salt unless the total numbers are included. Even 10% doesn’t sound too bad until you realize it equates to about 8900 false accusations per year.

    Note: I found the total accusations of 89K on a website that purportedly had DoJ statistics on it. I have no idea if it’s an accurate number, but it sounds reasonable to me. And frankly, it need not be all that accurate to make the point, I think. A small percentage of a large number is still going to be a large number.

  4. Rick

    Another issue with false accusation statistics is that they’re usually based on studies of police reports. It’s not clear at all to me that the same accusation truth/falsity rate will play out in campus judicial proceedings or in social circles: perhaps those are lower-cost areas to accuse, so there will be more false accusations in those areas. Perhaps those are lower-benefit areas, so there will be fewer false accusations. I think a lot of people make the inferential leap from police report statistics to the other areas and assume we should treat those accusations with the same (lack of ) skepticism.

    1. SHG Post author

      …they’re usually based on studies of police reports.

      Usually? Is it better when you make shit up than when they make shit up?

      1. Rick

        I’m not sure I’m grasping the target of your question. If you think I’m full of shit and that most of the “2-10%” claims are based on studies of campus conduct proceedings and (for lack of a better term) social accusations, that’s possible but all the false accusations studies I’ve read have focused solely on police investigations. So they’re not really studies of rape accusations generally; they’re (at best) studies of accusations that were reported to police.

        Given that there are substantially different legal and social consequences for reporting to police vs. reporting to campus authorities vs. merely telling your buddy that your other friend raped you, it seems to me that we shouldn’t assume that the results of a study of rape reports to police can be generalized to any time someone makes a rape accusation.

        1. SHG Post author

          …but all the false accusations studies I’ve read have focused solely on police investigations.

          Well, since this is all about you, what else could possibly matter than what you’ve read, what you’ve focused on and what you know? Screw all the other people on the internet, all the studies, all the people who have wasted their time reading and thinking about those studies, because this is really only about some guy on the internets named Rick and the shit Rick focused on, so can the rest of the internet just give a shit only about some guy named Rick and the hell with everything else. Am I making myself clear at all?

  5. EH

    People think the question these studies answer is “What percentage of rape reports are false?” In reality, the question they are really answering is “What percentage of rape reports can we classify as false with a high degree of certainty?”
    Another issue is that these reports (and reporters) are generally using wholly different criteria for evaluating false and true reports. It makes no sense to use a “beyond a reasonable doubt” or “clear and convincing” standard for false accusations, coupled with a preponderance standard for true ones. That just weights one side of the scheme. Even folks with a basic knowledge of stats should be able to see that, though they rarely do.

    After all, if you assume that proof of “true” is “preponderance of the evidence,” and assume that the proof of “false” is also the same (in other words, that there is no big “unknown” gray area in the middle) then the false accusation rate is the same as the non-conviction rate. At least or civil trials.

  6. se

    The underlying assumption that false rape report rates are going to be constant over-time and over-places is ridiculous on itself. I would expect false rape rates go up when the rules are loose (e.g. listen and believe college setting) and down when rules are strict or assumed victims punished.

    When it is easy to falsely accuse someone one hates to get even, people of both genders do just that. And attitudes/policies on rapes did changed in the past.

    Even if you could accurately measure false rape rates now, the numbers are likely to change when the rules change again.

  7. John Neff

    I have reviewed all of the data that I could find on the subject of false rape reports and the number of data sets that can survive critical review is very small (maybe six). There is very little agreement among the sets I think are reliable so my conclusion is that all we can do is estimate the range and the range is too large to be useful.

  8. Paul Thomas

    I have serious problems with the whole framing of this issue, particularly when you’re talking about “grey rape” scenarios. It is often the case that someone will make a report which is, in every possible sense of the term, true, yet does not satisfy the legal definition of rape. Indeed, in many instances, especially those involving mild to moderate intoxication, the facts may be essentially undisputed. “I had sex and later regretted it” would fall into this category too. There’s nothing false about such a report– it just isn’t a rape.

    So where do these scenarios fall into the four categories above? Certainly not “false,” and certainly not sufficient to proceed either. Nor do “insufficient information” or “inconclusive” work: the facts are what they are. What these are are truthful reports which fail to state a legally sufficient claim. I haven’t drilled down into the individual studies, but none of the meta-analyses I’ve seen has ever even acknowledged the existence of such a category.

    And I see commenters here are drawing the same nonsensical conclusions– e.g. EH claiming that in a pure preponderance of the evidence scenario, “the false accusation rate is the same as the non-conviction rate.” Hogwash. The non-conviction rate includes cases where the factfinder credits some or all of the accuser’s testimony, i.e. in no sense a “false” allegation, yet concludes that it did not constitute a rape. (Hell, for that matter, the conviction rate may even include a few cases where the opposite is true– the defendant’s own testimony, credited over the accuser’s, establishes that he committed a rape.)

    Bottom line: You had better damn well be sure– beyond a reasonable doubt, even, pace EH– that an allegation of rape or indeed any other serious crime is truly spurious before you officially characterize it as “false.” It’s one thing to be told that the horrible thing that happened to you isn’t enough to rise to the level of a crime. It’s quite another to be told that it did not happen to you at all. If the police wrongly characterize a rape as “false,” they have just inflicted incalculable mental suffering on a person for no reason other than that she or he exercised the constitutional right to petition the government. Accusers should not be exposed to ridicule and shame merely because they don’t know the technical fine points of the legal definition of rape.

    So to the extent that these studies use a restrictive definition of “false claims” and encourage police departments and other recordkeepers to do the same, that is a Good Thing. But they are not going to tell you how many allegations are “false” in the common usage of the term, “factually untrue.” The police are not going to spend resources (nor should they) trying to parse whether a particular meritless case is truly “false” or merely “inconclusive.”

    1. SHG Post author

      The definitional problem of what constitutes a rape or sexual assault is huge, and one that I’ve addressed here a few times. When words become untethered from definitions, it makes communication impossible, which in turn makes empirical analysis impossible.

    2. Francis Walker


      The scenario you describe is typically referred to as “baseless. Here is how the 3 major US sources categorize the scenario you describe:

      FBI – The FBI uses “Unfounded” not false. This includes both false reports and baseless reports and is one of the criticisms of using the unfounded rate as a proxy for false.

      MAD study – There appear to be two categories where this type of case could show up. First is Unfounded/baseless at 8.5% (“Cases determined to be baseless include those that do not meet the elements of the offense and those that were improperly coded as a sexual assault in the first place.”). There is also “Closed as informational report” at 17.9% which on the tracking sheets is described as “Closed as an informational report (elements of a sexual assault offense not met).”

      Lisak study – These cases are grouped into the “Did not proceed” category, which I don’t like as it is a known outcome being grouped in with the cases where there just wasn’t enough info to determine truthfulness. If I had to guess, this was done to prevent people from adding it to the false category.

      Another issue here is that depending on how the results are categorized, they can’t always be used to answer the question you want.

      If you want to know how many reports are just plain false and made up, you can’t use the FBI data because false and baseless are combined.

      If you want to know in how many cases was a rape reported where no rape occurred you can’t use the Lisak study because baseless is lumped in with “case did not proceed.”

      If you want to know how often a specific person was accused of a rape that didn’t occur you can’t use any of the studies because none of them break out stranger vs identified individual.

    3. Patrick Maupin

      “I had sex and later regretted it” would fall into this category too. There’s nothing false about such a report– it just isn’t a rape.

      It’s a false report as soon as it is reported to the police as rape. SHG has reported on, IIRC, several cases where the female regretted the encounter and then was apparently egged on by “friends” and/or family that it wasn’t at all her fault and that the guy should suffer.

      It’s one thing to be told that the horrible thing that happened to you isn’t enough to rise to the level of a crime. It’s quite another to be told that it did not happen to you at all. If the police wrongly characterize a rape as “false,” they have just inflicted incalculable mental suffering on a person for no reason other than that she or he exercised the constitutional right to petition the government.

      There’s exactly that definitional problem Scott mentioned. It is impossible for the government to “wrongly characterize a rape as false” when there was, in fact, no rape. And the constitutional right to petition the government doesn’t include the right to make shit up to try to get somebody else thrown in jail.

      Accusers should not be exposed to ridicule and shame merely because they don’t know the technical fine points of the legal definition of rape.

      Part of your argument seems to boil down to “rape is so traumatic that we shouldn’t discourage its reporting.” Which is all well and good, but even if you don’t buy into the counter-argument that being falsely accused of rape is so traumatic that we shouldn’t encourage baseless accusations, what happens next? I mean, even if you’re thinking of the “victims”, seriously what’s in it for them to report an after-the-fact-I-regretted-it rape? They make the guy’s life hell for awhile, and then they go to court and embarrass themselves recounting the non-rape in front of an entire courtroom? How is that helpful for anybody?

    4. EH

      You’re selecting a relatively unusual definition of “false,” which is akin to “wholly baseless,” “entirely without merit,” or, perhaps, “maliciously incorrect.”

      But generally speaking “false” means “wrong.” If you get accused of rape, and you didn’t rape anyone, the accusation is not true; IOW, it’s false. See, also, “falsely accused of lying about rape.”

      If you want to discuss the very low rates of, say, “malicious reports given with intention to deceive the tribunal” I’m all for it. But if you want to try to work the social-justice-publicist angle of claiming that “false reports” are astoundingly rare, and pulling back to an esoteric definition of “false” when you get challenged on it that won’t fly.

      As for the preponderance standard: you won’t find me disagreeing that preponderance is a problematic standard to use for rape. I’m happy to see that you think it’s inappropriate to target someone to which a “horrible thing [has] happened” as a liar based on a 50% guess. I’m interested in whether you think that it’s OK to punish folks based on that same standard, and whether you think you can safely assume said “horrible thing [has] happened” without, at least, a preponderance finding.

  9. Pingback: But For Fitbit: A Rape That Never Happened

Comments are closed.