The Great Rape Joke Controversy™ Part 1

A subject I’ve been meaning to tackle for a while, but never quite got around to, is that of The Great Rape Joke Controversy™, which flared up on Twitter (again!) the other day – and if you missed the action then I’d heartily recommend that you read Ally Fogg’s take on this latest round of Twittericuffs for its sensible, but sadly all-too-rare, discussion of the issues and the background to this latest rape joke shitstorm.

Rather than pitching in to this debate with yet another batch of musings on the subjective nature of humor, I thought I’d take a different approach and use a bit of science to try and unpick some of the more contentious arguments that tend to get thrown around whenever the rape joke issue rears its ugly head, and as an inveterate states geeks, one that immediately grabbed my attention was a tweet directed at Richard Herring by Sophia McDougall, which reads:

@Herring1967 You know there’s a 1/4 chance the woman had BEEN raped? & that some women in yr audience definitely would’ve been? Still funny?

Could this really be true?

Is there really a 1 in 4 chance that the female heckler to whom Herring directed his contentious putdown may have been raped or that there would have been at least some women in his audience that had been raped?

The underlying argument is, of course, that rape should be considered a taboo subject by comedians because there is high probability that their audience include rape survivors who likely not only to be offended by jokes which deal with or even touch on the subject of rape but who might actually have been so traumatised by the experience of rape to the extent that any unexpected exposure to that subject might have, at the very least, a transient adverse effect on their mental health.

That’s the theory, but is an argument that can be supported with any evidence.

Well, for starters, the suggestion that there is a 1 in 4 chance that Herring’s female heckler may be a rape survivor is complete and utter nonsense.

The 1 in 4 figure given is a classic zombie statistic which was originally derived, back in the 1990’s, as an estimate of the lifetime risk of rape based a small, unscientific, survey of American female college students commissioned by a women’s lifestyle magazine, a grade of evidence that ranks only slightly above pulling numbers out of your own arse. So, even if it had ever been validated scientifically – which it hasn’t – it was still derived from an outdated survey of an unrepresentative sample group in a country other than the UK and cannot, therefore, be cited as an estimate of the lifetime risk of rape for women living in this country.

Based on annual prevalence estimates taken from the British Crime Survey, the raw lifetime risk of rape for women living in England and Wales is approximately 1 in 9, not 1 in 4,  but this doesn’t take into account the prevalence of repeat victimisation and when we all for that the figure for lifetime risk drops to between 1 in 11 and 1 in 13, depending on how conservative your estimates for repeat victimisation are. However, ven this figure cannot reasonably be cited as a estimate of the chance that a woman attending a stand-up gig may have been raped – lifetime risk is only a relevant statistic for women who are old enough to have completed a statistical ‘lifetime’, which in terms of rape statistics means that they would need to be aged 60 or over as that is the basis of the standard lifetime risk calculation for rape, which accounts for the fact that the BCS estimate for the prevalence of rape since the age of 16 in its nationally representative weighted sample, is around 1 in 22, around half the lifetime risk derived from annual prevalence rates.

For under the age of 60 then their lifetime risk needs to be modified taking into account both their age and the fact that a woman’s annual risk of being raped varies according to their age, which for women aged between 16 and 19 is around 12-13 in 1,000 but falls sharply once they reach their twenties to between 2 and 3 in 1,000, although women in their mid-late 30s and early 40s exhibit a slightly higher risk of being raped that women in their 20s, after which the risk beings to tail off. So the good news is that if you’re a woman and you’ve made it through your teens without being raped then your lifetime risk has already fallen about around a third.

Age is, however, not the only demographic characteristic that impacts on women’s risk of being raped, even if it is the biggest risk factor.

Income also makes a difference – women who come from low income households are around 3.5 times more likely to be raped than women from households whose income is at, or above, the national average (i.e. median household income).

Marital status is also a significant factor – the annual risk of being raped is around six times higher for divorcees and four times higher for women who are single or separated than it is for women who are married or cohabiting, which probably explains why women in the mid-late 30s and early 40s show a slightly higher risk than women in their 20s.

Even the type of property you live in can make a difference; living in social housing doubles a woman’s risk of rape compared to living in private rented accommodation and this also increases her risk by a factor of 4 when compared to owner occupiers.

The key point to take from this is that whatever the population risk of rape may be for women, the actual likelihood of a particular audience containing one of more rape victims, let alone a particular woman in that audience being rape victim, is heavily dependant on the demographic make-up of that audience – and there are other demographic factors that may easily come into play, such as ethnicity, for which we don’t have any population risk or prevalence estimates.

In the case of rape, as with so many other things, the risk that women face is not evenly distributed across the female population but contingent on factors such as age, social class, lifestyle and other characteristics, and so the chance of an audience at a comedy performance containing a rape survivor who could be traumatised by a reference to rape in the show is similarly contingent on those factors.

So the chance of there being a rape survivor in comedy audience is relatively small and anything but a ‘definite’ unless the gig is an arena show.

That, however, is only half the story – remember, the argument here isn’t just a rape survivor might not see the funny side of a joke that includes a reference to rape or even that such an individual might be offended by the such a joke. The issue here is whether or not comedic references to rape  might pose a significant risk of trauma in the somewhat unlikely event that there is rape survivor and it is only in the last five years that an significant research into the prevalence of trauma-related psychiatric disorders in rape survivors has found its way into print.

For our purposes, perhaps the most useful papers to consider are a couple of studies Zinzow et al.; one of which appeared in the Journal of Interpersonal Violence in 2010, the other in Social Psychiatry and Psychiatric Epidemiology as recently as May of this year.

Our first paper by Zinzow et al., “Drug- or Alcohol-Facilitated, Incapacitated, and Forcible Rape in Relationship to Mental Health Among a National Sample of Women.” [1] is fairly typical of the limited range of epidemiological studies that have looked at, at least in part, the relationship between rape and mental health in women and provides some very valuable information in the form of relative risk ratios for Post-Traumatic Stress Disorder in relation to the rape tactics adopted by perpetrators:

Women who reported forcible rape (FR) were over three times as likely as nonvictims to meet lifetime criteria for both psychiatric disorders (Post-Traumatic Stress Disorder and Major Depressive Episodes), even while accounting for other rape experiences and revictimization history. FR was associated with significantly higher risk for PTSD and MDE than incapacitated rape (IR), and significantly higher risk for MDE than Drug/Alcohol Facilitated Rape (DAFR) and IR. Therefore, rape tactics did appear to differ in relation to these two common rape-related mental health outcomes. In addition, this was the first study to look at DAFR and IR with respect to mental health correlates. Women reporting a history of DAFR were almost twice as likely as nonvictims to meet criteria for PTSD. However, neither form of substance-involved rape emerged as a significant predictor of MDE.

In terms of definitions, incapacitated rape differs from drug/alcohol facilitated rape is so far as in the former the victim has voluntarily become intoxicated while, in the latter, intoxication is involuntarily, i.e. the victim’s drink was spiked by their assailant, etc.

This is all very useful and important information but, unfortunately, adds very little to the debate surrounding rape-related humour as what it doesn’t give us is any sense of the prevalence of PTSD and other trauma-related disorders amongst rape survivors. For that we need to look at the abstract of Zinzow et al.’s second paper, “Prevalence and risk of psychiatric disorders as a function of variant rape histories: results from a national survey of women.” [2]:

Women with rape histories involving both substance facilitation and forcible tactics reported the highest current prevalence of PTSD (36%), MDE (36%), and AA (20%). Multivariate models demonstrated that this victim group was also at highest risk for psychiatric disorders, after controlling for demographics and childhood and multiple victimization history. Women with substance-facilitated rapes reported higher prevalence of substance abuse in comparison to women with forcible rape histories. Comorbidity between PTSD and other psychiatric disorders was higher among rape victims in comparison to non-rape victims.

These studies are not without their limitations due to their retrospective design and use of self-report measures for both rape and clincial indications, and nether study appears to include controls for women’s history of mental health problems prior to being rape despite the high probability of this being a significant source of confounding for drug and alcohol abuse. Nevertheless, we are at least getting somewhere – for women with a rape history that puts them in the highest risk group for developing trauma-related mental health problems, the prevalence of major disorders (PTSD and MDE) is a little over 1 in , which, if you want to take something positive from these findings, tells us that women are, on the whole, much more robust – psychologically speaking – than the common and rather stereotypical image of rape victims might suggest.

So, even if there is a rape survivor in a comedy audience, the probability that they also have  trauma-related psychiatric condition which could potentially be triggered by the use of rape-related humour is, at worst, a little over 1 in 3 but also contingent on the circumstances and manner in which the survivor was raped. That said, we also have to take into the fact that social phobias and social anxiety disorder are also a common symptom of rape-related PTSD and MDE, so this alone may significantly reduce the likelihood that our hypothetical rape survivor at a comedy gig with have either of those conditions.

So what should we take from all this evidence?

Well, in the context of the public debate surrounding rape-related humour, the ‘rape survivor in the audience’ argument is all almost thrown into the debate in the assumption that its a debate-ending argument. Comedians are in the business of entertaining people and making them laugh, not traumatising members of the audience – well most are, although I wouldn’t personally care to vouch for the likes of Andrew Dice Clay on that front – so it follows, automatically, that subjects that could cause harm, rather than just offensive, should be taken off the comedic menu.

What the evidence actually tells us, however, is that that particular argument is nowhere near as strong as its proponents suppose and, in its most commonly used form, based entirely on an exaggeration of the risks derived from unvalidated and wholly unreliable zombie statistics.

More generally, what I’d suggest you consider carefully in terms of the ongoing public discourse around rape is the fact that although the causal link between rape and major psychiatric disorders such as PTSD and MDE was firmly established back in the mid 1970s, it wasn’t until the beginning of this century that epidemiological studies that sought to evaluate the prevalence and risk of PTSD in rape survivors begin to appear and that, if you look at the context in which this gap in the research base occurred, there is more to this delay in properly investigating the relationship between rape and trauma-related psychiatric injury than just cultural misogyny or official indifference toward rape victims. The stereotypical image of the traumatised rape survivor that emerged from the earliest studies of what then called ‘rape trauma syndrome’ presented rape campaigners and rape support organisations with a very potent, emotive and believable image of rape victims on which to hang their campaign activities, demands for changes to the law and improvements in law enforcement practices and demands for increased funding for victim support services, so much so that no one really thought to seriously question just how closely this image might reflect the reality of rape victims’ actual experiences until it became apparent, in the 1990s, that public acceptance of this image was creating a problem in the criminal justice system for women who didn’t exhibit obvious signs of severe trauma.

Only when it became apparent that the credibility afforded to rape victims by investigators, prosecutors and juries was being adversely affected by unrealistic expectations of how a rape victim should behave that were derived, in part, from the trauma stereotype – although pre-existing stereotypes are also, of course, a major issue – did anyone seriously begin to question just how valid this stereotype is and, by extension, how common PTSD and other major psychiatric disorders are amongst rape survivors.

Rape is an extremely complex phenomenon, a fact that is all to easily lost in the ideologically driven black and white noise that takes up far too much of the public discourse surrounding rap, and the drowning out of important nuances, complexities and subtleties in that debate can have serious, if wholly intended, consequences particularly when beliefs, assumptions and unverified theories are allowed to dominate the debate to the exclusion of an consideration of the evidence.

Next time out, I’ll be taking a look at the main bone of contention – the ‘rape joke’ – and examining what psychology of laughter can tell about the nature of rape-related humour and why the blanket assertion that rape can never be a legitimate source of humour may not only be wrong but, in some cases, potentially harmful.


1. Zinzow HM, Resnick HS, McCauley JL, Amstadter AB, Ruggiero KJ, Kilpatrick DG. Drug- or Alcohol-Facilitated, Incapacitated, and Forcible Rape in Relationship to Mental Health Among a National Sample of Women. Journal of Interpersonal Violence 2010 25: 2217

2. Zinzow HM, Resnick HS, McCauley JL, Amstadter AB, Ruggiero KJ, Kilpatrick DG. Prevalence and risk of psychiatric disorders as a function of variant rape histories: results from a national survey of women. Social Psychiatry and Psychiatric Epidemiology. 2012 Jun;47(6):893-902. Epub 2011 May 21.

18 thoughts on “The Great Rape Joke Controversy™ Part 1

  1. Pingback: Unity
  2. Hi Unity. Many thanks for the link and compliment, genuinely flattered.

    Not entirely sure I entirely agree with your analysis though. 

    You’re right that the 1 in 4 stat is probably bogus, and something like 1 in 9 seems more sound. And you’re also right that this masks wide differences by social class, lifestyles, ethnicity etc. 

    But if, for sake of argument, we assume that a comedy club audience consists entirely of affluent of middle classes with abstemious lifestyles (big assumption) the  proportion would still be unlikely to drop below maybe one in 15 of the women (wild top-of-the-head calculation) not to mention 1 in whatever of the men.  

    Since most comedy clubs will have a hundred people minimum, that makes the possibility that there WOULDN’T be someone there who was a rape victim extremely low. It’s much more likely there would be several. 

    And of course, while there may be one comedy club where there will be relatively few rape survivors in the audience, by the same token there might be others where the audience is disproportionately likely to include many victims. 

    So while the numbers you quote are indeed hyperbolic, I think the principle still stands.

    I’d add too that just because someone doesn’t meet the clinical definition of PTSD, doesn’t mean they are not “traumatised” in layman’s terms, and someone doesn’t need to have PTSD in order to be deeply personally disturbed by a rape joke with echoes of their own experience.   

    That said, I completely agree that rape-related stats are immensely complex, widely misunderstood and people are given very false impressions of risks and proportions by sloppy factoids, so thank you for keeping an eye on the ball, it is important. 

    Thanks again


  3. Pingback: Ally Fogg
  4. “And of course, while there may be one comedy club where there will be
    relatively few rape survivors in the audience, by the same token there
    might be others where the audience is disproportionately likely to
    include many victims. ”

    Based on a few quick and dirty calculations, I suspect that you’re right here even to the extent that the probability of there being rape survivors in an audience may vary considerable not only from venue to venue but from night to night as even relative minor changes in the demographic profile of audience seem capable of producing wildly divergent outcomes.

    That said, I think my point holds that the ‘rape survivor in the audience’ argument is nothing like as strong as those who use it believe due to the uncertainties and complex mix of dependent and independent factors.

    1.  It’s like the classic birthday statistics problem. There’s almost 100% probability of getting two people with the same birthday in a group of about 60 people. There’s also greater than 50% in a group of 30.

      It requires a lot of order to keep out the unlikely events in large samples, and generally speaking there isn’t overmuch statistical organization in gathering groups of people for, say, a comedy club. Just as classrooms don’t select for particular birthdays, comedy clubs tend not to select for trauma victims. Well, more strongly than schools select for birthdays but the point stands.

      If you wish to discount the offended survivor argument, I’m not sure statistics are on your side. There are perhaps other more fruitful angles of attack, some explored in your article. In fact, the “uncertainties and complex mix of dependent and independent factors” strengthen the argument by bringing us closer to random, Normal statistics and closer to certainty that X many members of the audience have been victims of this or that crime.

  5. Brilliant to read someone pointing out there is no such group as rape victims, simply people with experiences they will all react differently to, the expectation of how someone should behave after sexual assault fuels the good victim/slut dichotomy that still informs so much of our debate around this subject.

  6. “It requires a lot of order to keep out the unlikely events in large
    samples, and generally speaking there isn’t overmuch statistical
    organization in gathering groups of people for, say, a comedy club.”

    This is true if you deal with the problem purely in terms of demographics and normal statsistics, so yes, like the birthday problem, randomness would tend to move the probability of rape survivor in the audience towards one.

    However – and after giving is some thought, I think this is the key issue – a comedy club audience is not a random selection of people but a collection of agents, so in addition to factoring in prevalence and demographic variation you also need to build feedback into your statistical model and this is where you very quickly start to generate divergent outcomes from even very minor changes in your starting conditions.

    1. I will again assert that I don’t think statistics is the proper forum for this argument in the first place. However, in the interest of understanding your position:

      ” I think this is the key issue – a comedy club audience is not a random
      selection of people but a collection of agents, so in addition to
      factoring in prevalence and demographic variation you also need to build
      feedback into your statistical model”

      I’m completely at a loss for what this means. Could you reword it please? Build feedback from what into the model? What is the meaning and significance of club goers not being a selection of people but rather of “agents?”

  7. I should probably add that I really appreciate the detail of your article. The statistics at work are certainly not irrelevant and it’s nice to see them out in the open so the numerical implications of some of the arguments being floated around can be given a framework.

    I personally feel, however, that a lot of both sides are experiencing gut reactions and applying the most logical argument they can find. I don’t think there are many people who, upon close inspection, would insist that rape jokes are bad purely because they hurt the feelings of people who have undergone rape. There’s something else behind that argument, I feel, that isn’t being articulated.

    Part of it is most certainly purely social. We’re told it isn’t ok, so we think it isn’t ok and apply the appropriate rationalizations. But part of it comes from a sense that many people telling these jokes either don’t understand the implications of what they say, are genuinely trying to make people upset for fun, or are otherwise untrustworthy with respect to these sorts of jokes. Trust and comedy is complicated–I don’t feel like I can trust Daniel Tosh for example. I’m not convinced he understands the implications of what he says and/or that he isn’t a mean person.

    Those sorts of ideas, though, are hard to articulate. Especially the “it’s bad because society says so” sorts of ideas. The whole point of those types of taboos, mores, morals, and ethics is that we often feel them more than think or know them. Much easier to articulate is an idea like …

    “Well it could make people feel bad!”
    Why? Did it make you feel bad? Do you have the credibility to say that? Does it matter if it makes you feel bad? What right do you have, anyway? Where do you get off?
    “What if there was a rape victim here right now?”

    Now you’re righteous, you’re a defender. You’re a Good Guy.
    That isn’t to say everyone who argues this sort of thing goes through an analogous process. This is not even to say that it’s an invalid process. It’s an emotional reaction looking for an answer. That’s fine.

    But I feel it’s more productive to address the emotional response. Why do people feel this way? Why this joke, not that one? Why this group of people speaking out against it, not that one? If we treat with the reactions rather than the hasty rationalizations we can get a better idea of why people feel what they feel and how to communicate better both through and around the comedy.

  8. Pingback: asquith
  9. “I’m completely at a loss for what this means. Could you reword it
    please? Build feedback from what into the model? What is the meaning and
    significance of club goers not being a selection of people but rather
    of “agents?””

    An audience is not a random, or even semi random, selection of people drawn from a particular population. It is a collection of individuals each of whom has made a conscious decision to go a gig. As individuals, people have agency, the ability to make a choice as to whether or not they attend a particular comedy gig in a particular location at particular time, and the way in their make that choice may be influenced by any number of factors

    – Have I seen any of the acts before and, if so did I find them entertaining?
    – Is the venue easy to get to, and get home from?
    – Is the venue in a part of time where I feel safe?
    – Do I have to go to work in the morning?
    – Are any of my friends going, or likely to go if I do?
    – Can a I get a babysitter?
    – Have I already committed myself to doing something else at that time?

    And so on…

    Normal statistics will give you a reliable estimate of the probability of their being a rape survivor in a particular subgroup (the audience) or a particular population (people living in the town where the gig is taking place) only if you assume that the composition of the audience is somewhere close to random but this is actually very unlikely because the audience is not made up of people selected from the main population at random but rather of individuals who made an active choice to go to the gig.

    This flips the question on its head. Instead of asking what the probility of an audience (n) containing one or more individuals with particular characteristic (x) is, the question is what is the probability of a particular individual choosing to go to the gig given that their decision may be influenced by a, b, c, d and e. If you model this problem mathematically then the factors that influence individual decisions either for or against going to the gig are typically modelled as feedback because, for the most part, the represent the influence of prior experience or actions.

    Where you go from here is a matter of choice – you could use Bayesian inference to estimate the probability of rape survivor going to a gig or you could create a computational model consisting of, say, 1000 agents, each decision score linked to a simple risk perception model based on feedback and then run a series of simulations, adjusting the feeback parameters, to see exactly how the agents in the model behave. This approach, while time consuming, allows you to explore the audience question in terms of conditional probabilities and you could also expand the model by including other feedback models for, say social network/support, etc.

    Admittedly, modelling of this kind is a bit of a sledgehammer when it comes to cracking a very small nut, like the audeince problem, it’s much more useful for serious stuff like, in the context of rape, how social networks, support and isolation influence mental health risks nd build predictive models which can be used for everything from service planning to indentifying complex behavioural patterns and combinations of indicators that may place that particular individuals at an increase risk of becoming suicidal.

    1.  Fair enough. Mathematically, that makes sense. As you said, though, that kind of statistical precision is overkill for the particular kind of problem we’re trying to solve. Because we’re speaking in generalities, all of these complex factors that might apply to one comedy gig, or one comedian, or one city get sloshed together and smoothed out substantially when looking at many different comedy clubs, comedians, and cities.

      Furthermore, without any equally accurate statistical data giving us an idea of the actual makeup of a comedy club and the actual decision map required to end up at one, all of these hypothetical uncertainties are of little mathematical use compared to the uncertainties we do have at least  some data on. We can work with a lot of the uncertainties I mention, whereas the specific, localized ones you’ve mentioned merely tell us that any estimate we try to make would be imprecise–we can’t vouch for the accuracy of those measurements without comparing them to the results of a more detailed study as we cannot prove that there’s enough correlation between the new, unstudied uncertainties and rape survivors being or not being in the audience. That’s what makes them uncertainties. As long as we’re going to bet the farm as to that last point, we might as well jump back to a less accurate statistical model and assume that it is valid. It’s no worse an assumption as far as I can tell.

      Lastly, I’m curious what you think about my other points. About the survivor argument being an appeal to authority or a deferment of responsibility rather than an argument in it’s own right. I suppose there are no numbers to back it up and you seem much more interested in the mathematics, so it’s alright if you’re just not interested in that discussion. Speculation has it’s problems, after all. But I’m curious what brought you to apply rigorous mathematics to this particular issue, what made it seem particularly appropriate.

  10. Really interesting, cheers for going into the numbers.

    Personally I find the strongest argument against telling rape jokes is that there is a relatively high chance that you will make a rape joke in front a rapist and they will laugh and think that you think rape is okay. That is a pretty horrible thought to me.

    1. All of the studies I’m aware of for prevalence of rapists are US ones, so it’s hard to say. Somewhat less likely, in a fully random sample, than there being a rape victim in the audience, since it appears from the US studies that serial rapists are relatively commonplace as a proportion of rapists.

      Probably a near-certainty in a comedy club, though.

  11. A few other points for your analysis:
     – the term “rape joke” is slightly broader than jokes about rape as defined by the SOA2003 and is generally used to refer to jokes about the various other sexual offences too. It’s an international term. Whether or not this means that you should include the BCS sexual assault figures in your calculation I’m not sure.
     – given the increasing chance of being raped with decreasing age, one can probably assume a noticeable gap between the BCS figures and lifetime risk due to rapes occurring before the age of 16. (I’m not aware of any surveys with comparable methodology to the BCS, and the proposals to extend the BCS to children specifically exclude the sexual offences questions)

    I’m not sure the question of probability given a comedy club environment is particularly useful, though, for two reasons:
    – because the nature of that environment is to deter people who don’t like rape jokes. (By analogy: the various groups who say “we don’t have any disabled customers/members so we don’t need to make our stuff accessible”)
    – because I don’t know of anyone arguing that rape jokes are only wrong if told at comedy nights. I think the concern with the 1 in X figure is more about rape jokes in general with “official” comedians being an “on the record” public example which can be more easily be subjected to critical analysis.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.