How Anti-Immigration Activists Misuse Rape Statistics
The Internet has brought an enormous mass of knowledge to the fingertips of everyone with a computer, smartphone or tablet. Never before have so many individuals been so close to true scientific facts about the world, from fun facts about animals to the latest crime statistics. Large communities with blogs, forums and social media groups have grown up around a wide variety of special interests and it has become a powerful tool for communication, cooperation and the advancement of human knowledge.
However, this has also led to the creation of ideologically isolated Internet communities, where faulty claims and misunderstandings of statistics and empirical evidence gets repeated in an endless echo chamber and all refutations are either ignored, misrepresented or subjected to ideologically driven rejection, often with stale references to supposed “political correctness”, as if that was a statistically mature rebuttal.
This article will show that according to crime victim surveys, the actual rate of sex crimes has been more or less unchanged in Sweden between 2005 and 2014, despite the fact that immigration has increased during the same time period. Instead, the increasing rates of reported rapes are influenced by expansion of the legal rape definition, an increase in the tendency to report rapes, police efforts to classify each individual rape as a separate crime and their tendency to classify any sex crime that could potentially be rape as rape. It will also demonstrate that reported rates between countries such as Sweden and Denmark cannot be naively compared to do the large difference in legal rape definition and police registration methods.
Primer: how does crime statistics work?
Before we look at the data, here is a short primer on what crime statistics is and different ways to measure crime. This is crucial, since a lot of arguments that commonly occur in mainstream, alternative and social media are based on confusing these different measurements.
When looking at crime statistics, there are two main sources of information:
(1) Data from reported crimes: these kinds of datasets provide information about how many crimes of a certain type was reported to the police during a certain time period (such as a calendar year).
(2) Data from crime victim surveys: these datasets come from surveys sent out to many thousands of people who ask them if they have been the victim of particular crimes during e. g. the past 12 months.
These may look similar, but they are in fact completely different and answer very different questions.
Reported crime rates are influenced by the real crime rates, propensity for reporting, trust in the criminal justice system, victim stigma, crime definitions and registration methods. This is a measurement of how many reports of a certain crime that comes in to the police, and thus a measure of the police investigation load.
Crime victim surveys, on the other hand, attempt to measure how many individuals are the victim of a certain crime during each time period i.e. the actual crime rates. Since these are anonymous, is not done by the police and uses the same crime definitions over the years, it is not influenced by the confounders that reported crime rates are.
To make it crystal clear: reported crime rates is not the same as actual crime rates and changes in one does not need to be reflected in the other. In fact, they do not even have to correlate. This is because not all crime reports are true and not all crimes get reported to the police.
If you want information about actual rates, you have to look at crime victim surveys.
The Swedish Crime Survey show that sex crime incidence is unchanged since 2005
The Swedish National Council for Crime Prevention (BRÅ) does a crime victim survey every year called The Swedish Crime Survey (“Nationella Trygghetsundersökningen”, abbreviated NTU). They send out surveys to a random and nationally representative sample of 20 000 people in Sweden and about 12 000 people respond (60% response rate). They survey asks people if they have been the victim of certain crimes and other questions related to crime and the criminal justice system. They started at 2005 and summary statistics was recently released for 2014.
Here is BRÅ data from 2005-2014 for crimes against individuals, expressed in proportion of total population:
Red is any kind of sex crime, blue is robberies, purple is any kind of physical assault, yellow is fraud, dark-cyan is any form of harassment and green is any kind of illegal threats.
The major result from this graph is that not a lot happens over time. Since 2005, most categories are roughly the same or slightly decreasing, although there is some random fluctuation from year to year.
The Swedish Crime Survey attempts to ask additional details about the seriousness of the sex crime in question by asking if the sex crime was accompanied by violence or threats of violence, if the victim was incapacitated by sleeping, alcohol or drugs and if the victim would label it as rape. However, because it cannot be generally expected that victims of sexual crimes fully understand legal definitions of rape, those estimates are more uncertain. The year-to-year variability in rapes is comparable to that of sex crimes generally.
This also means that the doomsday claims made in alternative media and Internet forums about an avalanche of rapes and sex crimes in the wake of immigration is empirically false. So how could so many political ideologues be so profoundly mistaken on this issue? It turns out that it is because they look solely on incidence of reported rapes and not actual incidence of rape.
If actual incidence of rape is unchanged, why are rape reports increasing?
Anti-immigration activists often point out that the observed number of rape reports per year is increasing. This is true, but does not demonstrate that the actual incidence of rape is increasing. As we saw above, it is more or less constant over time (with some year-to-year fluctuations).
So what is going on? Well, the increase is caused by a couple of different factors:
—> Changing definitions of rape: over the past 20 years, Sweden has broadened the definition of rape several times: actions that are comparable to rape (e. g. forced vaginal fisting) was classified as rape in 1998, sex with someone in a helpless state (e. g. sleeping or passed out due to alcohol or drugs) was classified as rape in 2005 and sex with someone in a particularly vulnerable situation (e. g. victim paralyzed with fear) was classified as rape in 2013. Since a higher proportion of sex crimes is classified as rape, the incidence of rape reports increases over time.
—> Increased tendency to report sex crimes: a doubling of the proportion of sex crimes reported to the police occurred between 2005 (10%) to 2011 (20%) according to BRÅ, but this figure fluctuates a bit from year to year. This statistic may be influenced by increase in societal equality, more attention given to sex crimes, and increased trust in the criminal justice system by victims.
—> Extensive registration efforts: Swedish police has made a stronger conscious effort at registering all potential rapes over time: “a lot of effort is made to register all cases that can be suspected to be rape. As this is done at a very early stage of the process, cases are included that later turn out to be some other sex crime, or even no crime at all.” (according to BRÅ).
etc.
Together, these factors increases the incidence of reported rapes. No one is claiming that any individual factor causes all of it and they are not post hoc rationalizations since crime victim survey data shows that the incidence of sex crimes is more or less constant since 2005.
Why naive comparison of incidence of rape reports between Sweden and Denmark is invalid
When anti-immigration activists are faced with data from crime victim surveys that show that actual rates of sex crimes is more or less constant in Sweden from 2005, they rarely change their minds. Instead they either appeal to conspiracy theories about the government hiding statistics (even though it is freely available online), and if they are more sophisticated, they will try to make faulty comparisons with rape statistics from other countries, such as Denmark. However, this is not a valid comparison.
Why not?
BRÅ emphasize the distinction between actual crimes and reported crimes:
Within established research about levels of crime and crime development, people are agreed that it is not possible to evaluate and compare the actual levels of violent crimes (such as rape) between countries by comparing the number of crimes reported to the police. This is because there are significant differences between the judicial systems of countries and systems for creating statistics showing crimes reported to the police.
BRÅ goes on to cite a couple of reasons why a naive comparison between Sweden and other countries does not work:
Broader definition and extensive registration
Firstly, in Sweden there is a noticeably broad definition of what constitutes rape. This means that more acts in Sweden are regarded and registered as rape than in the majority of other countries. Secondly, in Sweden a lot of effort is made to register all cases that can be suspected to be rape. As this is done at a very early stage of the process, cases are included that later turn out to be some other sex crime, or even no crime at all. In addition to this, all individual acts are registered — not just the latest occasion or the main crime. In many other countries cases like these are filtered out and do not show up in the statistics.
In other words, Sweden has a much broader definition than other countries, police makes an effort to label sex crimes that they suspect are rapes and registers each rape as a separate event even when the perpetrator has raped a victim several times over a longer period of time (such as months or years).
This can be independently verified by comparing the number of rapes with the number of unique rape victims. For instance, the NTU 2013 report states that (my translation):
Of the polled, 0.8% states that they were the victims of sexual crimes during 2012, which corresponds to about 62 000 persons in the population (aged 16-79). The proportion of sex crime victims has been relatively stable during the eight measurement dates (see table 3A). Just like when it comes to threats and assault, most people who were the victim of a sex crime during 2012 were exposed to a single event, but there is a smaller group that are repeatedly victimized. The number of events amount to an estimated 223 000.
So although there were 62 000 people exposed to sex crimes (not just rape, see above), there were 223 000 events, which highlights that the police reports single events separately.
Another way to independently confirm the statements made by BRÅ is to compare the legal rape definition between Sweden and Denmark.
The Swedish legal rape definition can be found in the 1§ of the 6th chapter of the Penal Code called Brottsbalk (1962:700) (my translation):
6th chapter. About sex crimes
1 § The one who through assault or otherwise with violence or threat of criminal act forces a person to intercourse or to undertake another sexual act that with respect to the seriousness of the violation is comparable to intercourse is convicted for rape to prison for at least two years and at most six years.
The same applies to the one who with a person completes an intercourse or a sexual act that is according to the first section comparable to intercourse by unduly taking advantage of the fact that the person due to being unconscious, being asleep, serious fear, intoxication or other drug influence, disease, bodily injury or psychiatric disorder or otherwise with regards to the context are in a particularly vulnerable situation.
[…]
The first paragraph continues for two more sections to discuss sentencing where the crime is considered less severe and more severe than normal, respectively.
The legal rape definition in Denmark can be found in the Criminal Code (“Straffeloven”) Chapter 26 Section 216 (Danish original):
§ 216. For voldtægt straffes med fængsel indtil 8 år den, der
1) tiltvinger sig samleje ved vold eller trussel om vold eller
2) skaffer sig samleje ved anden ulovlig tvang, jf. § 260, eller med en person, der befinder sig i en tilstand eller situation, i hvilken den pågældende er ude af stand til at modsætte sig handlingen.
The first part of the definition is about violence or threat of violence (cf. Swedish definition above) and the second part relates to victims that are incapable of refusing the act due to e. g. sleeping or heavily intoxicated or under the strong influence of drugs.
Thus, the Swedish rape definition is much broader, encompassing sexual activities that are comparable to rape, individuals that cannot resist due to psychiatric disorder or are in a particularly vulnerable situation. In other words, many sex crimes that are classified as rape in Sweden are classified as a lesser sex crime in Denmark.
So when anti-immigration activists write stuff like:
Danish legislation is not very different from Sweden’s, and there is no obvious reason why Danish women should be less inclined to report rape than their Swedish counterparts.
We can now easily understand why they are mistaken: Danish legal definition of rape is very different from the Swedish legal definition and the tendency to report rapes is an explanation for why rape reports are increasing in Sweden whereas crime victim surveys place sex crimes at more or less a constant rate since 2005 (not an explanation for between-country differences in the number of reported rapes).
Additional claims made by that source have been refuted here and here and several other articles on this website.
Conclusion
The moral of the story is that you should not get your information about crime statistics from shady blogs who make trivial statistical errors (like confusing reported and actual crimes) or Youtube videos. Instead, go to the original scientific papers or crime victim surveys, read them and understand the methodology, results, limitations and conclusions. There are no shortcuts to substantial knowledge.
The increasing rates of reported rape in Sweden is primarily caused by a repeated expansion of the rape definition, the fact that police is consciously reported all rapes as individual reports and an observed increase in the tendency to report rape. The differences between rates of reported rape between e. g. Sweden and Denmark is caused by differences in rape definition, differences in registration methods and efforts to rape all sex crimes that might be rape as rape.
In the end, it is doubly ironic that anti-immigration activists say that they refuse to read mainstream media because of its bias, yet (1) are strongly influenced by the increase in sensationalist coverage of violent crime by mainstream media and (2) are so willing to believe claims made by “alternative” media that are unsupported and often contradicted by actual crime victim survey results.
Follow Debunking Denialism on Facebook or Twitter for new updates.
A short addendum to the post:
To anti-immigration activists:
About a dozen or so of you have asked me to address this kind of topic again over a few months. Some have been very polite, while others have been confrontational and abusive.
Now I have taken the time to research the issue and write up a detailed response. I have only relied on official publications by the Swedish Council for Crime Prevention and legal texts from both Sweden and Denmark.
It turns out that your beliefs did not survive critical scrutiny and this critical scrutiny cannot be dismissed outright with references to “political correctness” or “leftist ideology”.
Although I understand that it is unreasonable for me to expect you to give up this anti-immigration activism, can we at least move on to more productive issues like how to increase the number of houses built, improved integration, creating more jobs etc.? Why do we have to rehash the same statistically flawed arguments about immigration and crime? This has been done to death for the past 30 years.
To skeptics who “don’t like politics”:
Yeah, I get it: you do not like politics and think that skepticism should not deal with political issues. However, this post has been exclusively about debunking abuses of statistics, and I think this is fully within the scope of scientific skepticism.
To social justice activists who do not like the term “actual rapes”:
I have avoided the terms “actual rapes” or “actual rape rates” when I mean rape rates as revealed by crime victim surveys in contrast to police rape report rates. This is because I did not want it to be confused with the bullshit terms “real rapes” or “rape rape” that is often used by people who want to distinguish some forms of rape (i.e. alcohol or drug facilitated rapes) from perceived “real rapes” (rapes that match public perception of rape, such as rapes involving direct violence) in an effort to make rape statistics look less serious. This is not at all what I mean. I am only referring to rates from crime victim survey because those are closer to reality.
An agitated reader going by the name of “Katherine Woo” sent me an angry email about this post.
This looks like a fake name since the social media profile linked to it is empty, the IP is anonymous and cannot be resolved and the name is the same name as a celebrity (a common method so that the comments made by the person will drown out in websites and news items about the celebrity).
But, let us give the Devil his due. Notice that there is no attempt at a critical rebuttal whatsoever of the arguments I actually put forward. The emailer does not even make an attempt to engage with the statistics from BRÅ. Why? Because the arguments I made above are solid and I think the emailer knows it.
Statistics from Swedish Council for Crime Prevention is a “painful bit of sophistry”? This goes to show that anti-immigration activists have no respect for careful data analysis and only appeal to statistics when it appears to support their position, but reject it outright when it contradicts their ideology.
No, I actually linked two blog posts where I have destroyed this line of argument in the above article. These can be found in the sentence just before the conclusion section.
The general idea is that (1) the over-representations are small in the criminological context, and groups such as men, unemployed and the poor are much more over-represented, but anti-immigration activists do not bat an eye at this fact and (2) over-representation is the wrong statistic and that if you want to make pronouncements about immigrants as a group, the relevant statistic is not the probability of being an immigrant given that you are a rapist but the probability of being a rapist given that you are an immigrant. This latter probability is very, very low — in Sweden, over 99.7% of immigrants have never been suspected or convicted of a sex crime (see the BRÅ report references in the above two links), let alone rape specifically — and so the anti-immigration complaint disintegrates.
If Woo wants to seriously engage me in a discussion about these issues, a minimum requirement is to have read the post. Knee-jerk reactions do not qualify.
No, It looked specifically at sex crimes. These have more or less been constant for the past 10 years (as measured by crime victim surveys), despite a heavy increase in the amount of immigration to Sweden. This directly contradicts the anti-immigration position that more immigration equals radically more sex crimes. Looking at more long-term trends is no help to anti-immigration activists as we are currently in a long-term decline in more or less all forms of violent crime. Woo should read Steven Pinker’s Better Angles of Our Nature for statistics backing this up.
I have. I linked two such blog posts in the very article Woo objects to.
Yawn. Woo should read my post carefully before writing agitated emails. Saves the embarrassment on Woo’s part.
The subject is not uncomfortable. Woo and other anti-immigration activists are simply wrong. Dangerously wrong, in fact.
Here Woo deploys some version of poisonous M&Ms gambit that I have debunked here and here already.
If Woo is so concerned about “potentially heighten the danger to all women in Western societies”, Woo should advocate the selective abortion of male fetuses, since males would, by the over-representation logic of Woo, “potentially heighten the danger” to women. Of course, Woo would never do this and should thus logically not believe the argument in the first place.
“Concrete fear of being killed or tortured”? Well, there is currently a civil war in Syria with the government fighting several hundred separate groups, including the terrorist organization Daesh. This has led to 300k deaths and about 10M refugees (internal and external combined), perhaps more. This looks like a “concrete fear of being killed”. Thus, according to Woo’s own logic, there should be no issue here.
Finally, as argued above, immigration does not “impinge” on the right to bodily integrity anymore than giving birth to male babies do. It is a ridiculous claim.
Yawn.
An individual men’s rights activist (MRA) took issue with the following section in the above article:
Presumably because they think that societal equality has been on a steady decline over time, and that men are suffering considerably from it, either much more than women or in a different way from them.
However, these factors are the ones mentioned by BRÅ and therefore it makes sense to mention them, even if some disagree with the specific label used for the phenomena. For those MRAs who disapprove of that phrasing, they can mentally replace “increase in societal equality” with “more power to women” or similar.
At any rate, I think we all know what this article means when it says “increase in societal equality”. Thus, this objection seems to relate more to semantics than to the overarching issues discussed in the article.
Doug sent in some of his perspectives on this article. I will go over them, one by one, paraphrasing his considerations.
(1) “Some of the changes in the rape definition occurred prior to the start of the crime victim survey and are thus irrelevant.”
The purpose of going over the changes to the Swedish rape definition was to highlight that changes in definitions have a marked impact on the number of reported cases. This is no different from highlighting how changes in the definition of autism (not vaccines) impacts the number of reported cases over time. Doug quotes me out of context by starting his quote on force vaginal fisting, which was merely an example of the effects of the 1998 inclusion of comparable acts. In reality, there are many more acts that are deemed comparable to rape that does not involve vaginal penetration and this was just an example.
Since the crime victim survey does not show any appreciable change, his objection is not relevant since the point was not to explain changes between 2005 and 2014, because there were no appreciable change in the first place. Bringing up the 1998 definition change is to highlight that how we define rape matters for how many police reports are classified as rape.
(2) “The 2005 change was at the start of the survey and thus also unlikely to be relevant”
Again, the crime victim survey is not about reported cases. The definitions have the largest effect on reported cases. It is the changes in the number of reported cases that are brought up by anti-immigration activists and this change is to a large degree explained by changes in rape definition. Again, having sex with a sleeping person was yet again just an example. It covers all other helpless stages as well.
(3) “How do we know that cases that do not fall under 2013 definition were reported as rape before 2013?”
Simple. The sexual act did not involve violence, the threat of violence or the victim being in a helpless state. Thus, it was not classified as rape under 2005 definition. This is also clear from the many court cases prior to 2013 where people who would have been rapists under the 2013 definition were acquitted because the act did technically not fall under the 2005 definition. Contrary to popular belief, the criminal justice system does pay attention to what criminal definitions are used. Furthermore, a lot of people do not know the rape definitions clearly, so it is not at all clear that they would believe that what they had been the victims of was classified as rape in the legal system.
(4) “The victim being paralyzed with fear is a weird category”
The added category in 2013 was “particularly vulnerable state”, and again, being paralyzed with fear was just an example. The reason this is used is that the legal system does not want to let obvious cases of rape fall between the cracks due to technicalities. Therefore, they believe it is better to take into account whether the victim was in a particularly vulnerable state and judge the case based on the context and evidence for that specific case instead of overly relying on black-and-white cutoffs.
(5) “You claim, without evidence, that there has been an increase tendency to report sex crimes”
Nope, we actually have data for this. The tendency to report sex crimes (“anmälningsbenägenhet”) is calculating by taking the number of reported rapes and dividing it by the number of rapes from crime victim surveys. This ratio has gone from ~10% in 2006 to ~20% in 2011, essentially a doubling. I explained this in the post and even cited a reference. But perhaps Doug did not spend enough time reading my post or the references to become fully aware about this.
(6) “You just cited a single study!”
No, I cited 10 studies. A crime victim survey is done once a year. “All the other data” are from reported rapes from the police, which this very article explains is not a reliable indicator of potential changes in the number of committed rapes.