According to this website: https://www.rainn.org/statistics/criminal-justice-system,

out of every 1000 rapes, 994 perpetrators will walk free

statistics of rape conviction

Similar percentages are presented for other types of crimes as well, but I find it a bit strange to believe, since it is not about some failed state or anarchy, but about the US!

Is the above claim true?

  • 14
    The use of "rape" in the poster's really questionable as their cited statistics don't just refer to rapes, but also sexual assaults, where "sexual assault" is loosely enough defined to include threats, drugs/alcohol, being pressured, and "noncontact unwanted sexual experiences, which do not involve physical contact" (study). At first glance, it's unclear if there's an argument for why these statistics might be valid for a claim about rape report rates.
    – Nat
    Mar 7 '18 at 11:26
  • 16
    I note that the above poster, on the face of it, assumes that "reported to the police" directly implies guilt in all cases. It seems to bewail that (among other things) we have not achieved 100% conversion from "reported to prosecutors" to "incarcerated" - something that would only happen in the most transparently unfair of court systems.
    – Ben Barden
    Mar 8 '18 at 22:33
  • 12
    @daniel I'm not saying that our system is perfectly fair and accurate. I'm just saying that the complaint made, as such, displays a rather extreme bias once you start looking deeper. The implied assumption of 100% guilt in all cases is just a nice, blatant indication of that bias.
    – Ben Barden
    Mar 9 '18 at 14:32
  • 10
    The stat should state for every 1000 rape accusations, or 1000 rape trials, 994 of alleged perpetrators will walk free. There's no way to know in a strictly epistemic sense who has committed rape, only the number of accusations or cases that come before the court.
    – Zebrafish
    Mar 10 '18 at 20:34
  • 5
    @BenBarden if only we had a way to evaluate an accusation of rape once it was "reported to the police". Maybe there could be some sort of adversarial system that would culminate in a determination made by an impartial panel of citizens. May 29 '18 at 9:36

//o\\ Work In Progress //o\\

(Preliminary) Conclusion: 4 of the 5 claims check out so far (with caveats based on imperfect data sets), based on referenced crime data provided by the DOJ and FBI. (Still working on the final claim)

This page links and discusses their sources and methods: About RAINN's statistics.

Let's break the graphic down into its individual components:

Claim 1: 31% of rapes are reported to the police (310 of 1000)

RAINN says this statistic is calculated by averaging the 5 most recent years of available data from the US DOJ's "Criminal Victimization" report. Let's see if that stacks up.

Here is what I found for "Percent of victimizations reported to police" for Rape/sexual assault (all links are PDFs):

Averaging these 5 values gives 30.4%, so the calculation seems accurate.

Nitpick: The "Rape/sexual assault" crime category in the DOJ reports is fairly broad (it includes attempted rape, other kinds of sexual assault, and even threats of rape), so this particular source doesn't differentiate between the reporting percentage of each of these smaller sub-groups (if there is any difference).

I found another (earlier) source that does distinguish between them, and the numbers aren't out of line. The BJS report "Rape and Sexual Assault: Reporting to Police and Medical Attention, 1992-2000" reports the following in Table 3:

  • Completed Rape: 36% reported to police
  • Attempted Rape: 34% reported to police
  • Sexual assault: 26% reported to police

31% still seems like a reasonable number.

Claim 2: 18% of reports lead to arrest (57 of 310)

This stat comes from the FBI National Incident-Based Reporting System, which collects crime data from states and local agencies.

This is my speculation, but they might have used the "Incidents Cleared by Offense Category, 2015" report for crimes in the category "Sex Offenses":

  • Total incidents: 70,788
  • Cleared by Arrest: 12,431
  • Cleared by "Exceptional Means": 7,679

"Cleared by Arrests" / "Total incidents" works out to 17.6%, which is close to RAINN's figure.

Nitpick: What is a little unclear for this stat is whether the other column, "cleared by exceptional means" should count. According to the accompanying description, these cases were "cleared" (solved) in some other way:

When an incident is cleared by exceptional means, it was cleared in some other way than by arrest, i.e., the death of the offender, prosecution was declined for other than lack of probable cause, the offender was already in custody in another jurisdiction, the victim refused to cooperate in the prosecution, or the offender was a juvenile and the crime was handled without taking him/her into custody.

Some of these scenarios may be viewed as a "positive outcome" by the victim (if the offender is already in jail for another crime, for example), others would not. Strictly speaking, though, if they aren't prosecuted for this crime, the chart may be correct to exclude them (especially since we can't distinguish which is which).

Claim 3: 19% of arrests are referred to a prosecutor (11 of 57)

(Still working on this one, although this data does not appear to be in the FBI NIBRS data. In fact, I found a page that explicitly says that the FBI UCR program doesn't collect data on the number of persons who were prosecuted, convicted, and/or imprisoned.)

Claim 4: 64% of prosecutions lead to a felony conviction (7 of 11)

This one goes back to the BJS data, thanks for pinpointing the source, @BobTheAverage.

The Felony Defendants in Large Urban Counties report for 2009 (page 24, Table 21) shows a felony conviction rate of 57% for rape.

57% of 11 (in the chart) is 6.3, so rounding that up to 7 seems reasonable.

Nitpick: This report is for the 75 largest counties in the US, not for the entire US population. (This isn't cherry-picking, this report is the only one of this type available.) Clearly urban and rural counties will have different crime patterns; the question here is whether the rates of conviction and incarceration are similar, even if the volume is different.

Claim 5: 85% of felony convictions involve jail time (6 of 7)

From the same report (Page 29, Table 24), 89% of convictions for rape result in jail time.

This is very close to chart value of 85%.

Nitpick: (See nitpick on Claim 4)

  • Note that RAINN explicitly cites the NIBRS data when making a slightly different claim (13/1000 rapes lead to prosecution) on this page. As far as I can tell, NIBRS is an "upgraded" version of the UCR system, but it hasn't been universally adopted by law enforcement agencies yet. I also haven't been able to find this number in the publicly available reports, but it's possible that it's in the data somewhere. May 31 '18 at 17:02
  • @MichaelSeifert Thanks, that's a less muddled footnote. The problem is that I just can't find anywhere in either the NIBRS "data definition" or in the 156-page NIBRS user manual that indicates the database has anything about a final case disposition beyond whether it was "cleared", which we've already used to calculate Claim 3.
    – BradC
    May 31 '18 at 18:38
  • (Oops, we've already used the "cleared" percent in Claim 2, not Claim 3)
    – BradC
    May 31 '18 at 19:27
  • 1
    Re claim 2: Are they really using the positive outcome rate of all rape reports to determine the positive outcome rate of correct rape reports? That's a bit like saying "One out of seven people speak English, so one out of seven Americans speaks English". We'd (hopefully) assume that the positive outcome rate for correct reports would be higher tham that for reports overall, no? The only way for justifiying doing it like this is assuming that the difference between actual and overall reports is negligible. Do we know that that's true?
    – sgf
    Jun 4 '18 at 7:04
  • 1
    And Re claim 4, this only works out if we assume that nobody is charged with more than one rape accusation, otherwise we can't get the number of successful rape prosecutions from the number of successfully prosectued offenders.
    – sgf
    Jun 4 '18 at 7:09

This "answer" is too long for a comment, and is definitely incomplete. It is meant to supplement @BradC's answer, which is currently marked as a work in progress.

The analysis RAINN is doing appears to stitch together data from a variety of different sources. They list their sources, but do not include any indication of how they picked and processed their data. Although this is better than the average completely unsourced infographic, it is still frustratingly opaque. When the question was first posted, I spent a few frustrating hours researching this, and came to some of the same conclusions that @BradC did. I also found the following.

NCVS Data related to claim 1

Data from the National Crime Victimization Survey (NCVS) lumps rape and sexual assault together, and defines sexual assault as:

A wide range of victimizations, separate from rape or attempted rape. These crimes include attacks or attempted attacks generally involving unwanted sexual contact between victim and offender. Sexual assaults may or may not involve force and include such things as grabbing or fondling. Sexual assault also includes verbal threats.

RAINN takes these two together and reports it in their infographic simply as "rape." This seems seriously misleading; Rape includes things far more heinous than grabbing or fondling. I find it baffling that the NCVS does not report these separately.

Claim 4: 64% of prosecutions lead to a felony conviction (7 of 11)

This government report presents data on felony defendants in America's 75 largest counties. These counties contain large cities accounting for 37% of the total population, and are not nationally representative.

In 2009 in the 75 largest urban counties there were 412 felony rape defendants, of which 68% were convicted of something, 57% (235) were convicted of a felony, the rest a misdemeanor. The infographic claims that of the 11 cases referred to a prosecutor, 7 of them will result in a felony conviction. The two sources have a similar felony conviction rate; The difference could be explained by rounding errors in the 7 and 11 or it could be that RAINN took the average of several years of data. (Source: Table 21)

Claim 5: 85% of felony convictions are incarcerated (6 of 7)

According to the same source, 89% of convicted rapists are incarcerated. (Source: Table 24) This is very similar to the numbers presented by RAINN.

BJP Data related to claims 4 and 5 These statistics describe only residents of the biggest cities, 37% of the total population. Big cities and tiny towns have different culture, different police forces, and different patterns of crime. Although extrapolating data from big cities to the whole country is terrible analysis, I can't be sure that RAINN actually did this. The source is called out in RAINN's sources, and the numbers approximately match, but that is not conclusive.

Lack of a conclusion

After looking at all of this data, I still do not know what to think. The data sources RAINN looked through are credible and authoritative, but the sources use a variety of definitions, methodologies and sample populations. The infographic was an attempt to weave these threads of evidence into a clear cut story. After reviewing the evidence, I still believe the broad strokes of the story, but I am skeptical of details and specifics. I believe that a distressingly low number of rapes end in incarceration, but the final number could be very different from 6 out of 1000.

  • Thanks for helping look up those other sources. Regarding the "lumping together" of rape and other kinds of sexual assault, it's not that RAINN is conflating them or ignoring the distinction (it points it out in their data source page), its simply making an assumption that the rate of reporting of the entire group is a usable approximation for the reporting rate for a specific portion of that group. You can certainly argue with that (and I intend to see if other studies support the 31% number), but I don't think it undermines it nearly as much as you seem to think.
    – BradC
    May 30 '18 at 18:30
  • 2
    @BradC I think that that assumption is trying to paper over a hole in their data. It may be correct, but I will remain skeptical. May 30 '18 at 18:36
  • Updated my answer, including adding a second source to Claim 1 that has rates that are +/- 5% of the chart's value, so 31% doesn't seem implausible. I think I understand the concern, though: don't think of it as "but not all 1000 of those were actually rapes!", think of it as "when you select 1000 actual rapes from that larger collection of sexual crimes, how can we estimate the percent of victims who will report to the police?" You can see that even if the "real" reporting rate is, say, 35%, that doesn't change the overall chart much.
    – BradC
    May 30 '18 at 21:47
  • 1
    I agree that is their motivation, but unless you can find a source of arrest/prosecution/conviction data for rape that doesn't have these (or similar) caveats, it seems to me RAINN is using the best data they could find. Every study has scopes and sample sizes and error bars and footnotes, that doesn't automatically invalidate its results.
    – BradC
    May 31 '18 at 19:23
  • 2
    It seems to also be assuming that 100% of rape accusations are true? Does it take into account rape accusations which (a) the police or prosecutors discovered were false during investigation or (b) in which the court found the accused validly not guilty? Or does the data correct for this? (couldn't find anywhere that it did, unless I missed it)
    – Jacob
    Jun 5 '18 at 16:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .