Sweden's immigration taboo

Immigration data kept behind closed doors

In 2021, a study was published with the goal of giving a broad characterization of rape offenders in Sweden (Khoshnood et al., 2021). But this study received some unwanted and unexpected attention.

What was the issue? The researchers had found that the majority of rape convicts had immigrant background (either first- or second-generation). A female graduate student at the University of Gothenburg then filed a complaint to the Swedish government’s ethics review appeals board (“Önep”). In the complaint she wrote:

“Nowhere [in the ethics application] do I see that permission is requested to investigate the hypothesis that immigrants are overrepresented in the statistic for convicted rapists.”

While the authors had an ethics permit to handle and analyze the data, the complaint alleged that they did not have a permit to analyze specifically the hypothesis regarding immigrants. This resulted in a lengthy investigation of the authors.

To be clear, the goal of the study was not specifically to inquire into the perpetrators’ national backgrounds. Instead, the aim was to provide a broad characterization of rape offenders in Sweden. To do this they analyzed a large set of descriptive variables: year of birth, sex, educational attainment, socioeconomic status, number of prior convictions, type of prior convictions, psychiatric diagnoses, resilience, standardized test performance, school achievement, and, finally, country of birth. Immigrant status was but one of many variables; one they incidentally found had significance.

Önep (the ethics review appeals board) alleged that the authors lacked ethics approval to handle the “sensitive personal data” in the paper (Albinsson, 2021). Importantly, in this context, “sensitive data” should not be understood in the usual sense of personally identifiable information (PII). Of course, individuals may not be identifiable, and their anonymity should be secured. But, since their analyses were all of aggregate-level trends, personal identifiability was of no concern. Instead, “sensitive” in this context is more properly understood in the sense of the topic and results being politically sensitive. Indeed, the investigation documents show that one of the critiques of the researchers was that they had insufficiently explained how their research would “reduce exclusion and improve integration” of immigrants.

The fallout

The case dragged on for a long time. But, according to the most recent news in 2023, the chief prosecutor decided to close the preliminary investigation. The authors were not found guilty of anything. Yet lack of guilt is only a partial consolation. The authors had been pulled into a stressful case that dragged on for years.

The authors recently published a new study, similar to their prior research of rape offenders. In this study, they provided a characterization of firearm-related homicide offenders (Khoshnood et al., 2024). They analyzed the exact same explanatory variables as previously, except for one noteworthy difference: immigrant status was not considered.

There is no mention of why immigrant status was excluded as a descriptive variable in this later study. They used the same register data, so it was certainly available in the data. We can only speculate why. Perhaps it was an editorial decision by the publication, or—more likely I think—the authors were simply not willing to face further social backlash, or risk their precarious academic positions as they are being watched with more attentive eyes.

The Swedish government’s role

While immigration is widely considered a controversial subject, how countries have responded varies markedly. Sweden’s relative reluctance to openly sharing information about immigration stands in stark contrast to their neighbor, Denmark. This is despite the fact that immigrants are far more numerous in Sweden, making it an even more relevant public matter.

For example, the Danish Ministry of Finance have published multiple reports regarding the net financial contribution of immigrants to public finances. No such analyses are issued by the government in Sweden.

Annually, the Danish statistical agency (“Statistics Denmark”) publish a report about immigration which includes, among many other things, statistics regarding criminality. The Swedish agency responsible for crime statistics (“Brå”) have published a few reports detailing crime statistics in the context of immigration. The most recent Brå report was published in 2021. But the latest report before that was published more than 15 years earlier, in 2005. They are therefore very few and far between.

The difference between Denmark and Sweden goes well beyond frequency of publication. Unlike the Brå report, the Danish report includes analyses at the level of country-of-origin. Furthermore, publicly available and easily accessible nation-of-origin conviction data is provided by Statistics Denmark, which anyone is free to look at and analyze. Brå, the Swedish agency, do not provide raw crime data that one can freely navigate. They only provide statistics and graphs of some pre-defined metrics, which do not include much in the way of immigration-relevant facts.

The greatest taboo

Conspicuously absent in Swedish immigration research are detailed breakdowns by country of birth. You may see statistics by continent or broad region, but rarely by individual countries. The Brå crime report mentioned above, for example, did not include country-level results. But this reluctance to publicize country-level results did not always exist. As Joakim Ruist (2015) explains:

“The data is from 2007, which was the last year before Statistics Sweden enacted a policy of only providing researchers with less detailed information on individual country of birth in their micro data sets.”

Indeed, if you look at the LINDA database (Longitudinal INdividual DAta for Sweden), country of birth was available in the data set for decades prior to 2007 (see Table A2 in Edin & Fredriksson, 2000). Furthermore, if you go far enough back, the 1996 Brå report did provide country of origin results.¹

This taboo permeates Swedish institutions and is often enforced even when country-of-origin data is available. Andreas Ek (2024) conducted an analysis of the influence of immigrant cultural values on productivity and, in an associated ReadMe file, he describes how the ethics board barred the publication of country-level results:

When applying for data that includes at least one variable deemed “sensitive” (birth country and parental birth country in the case of this project), it is mandatory to have the research project approved by an ethics review board. […] To get this project approved, I agreed to the condition that I do not make any results available at the country level, but only present results related to average country characteristics. Therefore, I omit country names.

Once again, we see that, in the Swedish context, “sensitive” data is about political sensitivity, rather than the usual concerns about personally identifiable information.

Summary

There is a strong reluctance in Sweden to study the impact of immigration in their country. When compared with their Nordic neighbors (especially Denmark), Sweden is far less willing to provide their citizens with information regarding immigration. As far as I'm concerned, there is no justification for withholding data like this. In a democracy, citizens should have reasonably free access to such information (provided that anonymity is ensured), and citizens may draw policy conclusions from facts in accordance with their own values.

This piece should not be interpreted as some strawman variant suggesting that no research on immigration is ever conducted and published, or that you will be imprisoned if you study immigration. Just because it’s not a modern manifestation of Orwell’s “1984”, that does not mean there is no issue. Taboos and censorship exist on a spectrum. As I’ve already noted, a few reports have been published by Brå, though the publication history is very sparse. And individual researchers can still decide to study the topic, if they so choose.

However, any individual researcher who wishes to fill the void faces an uphill battle. Ethics review boards in Sweden are highly restrictive with respect to what may be published when you deal with these so-called “sensitive” variables.

Beyond the government’s unwillingness to properly inform the public, and the unreasonable restrictions enforced by ethics boards, the broader Swedish milieu can be hostile to unwelcome findings. Even if you have done nothing wrong, and you have been granted an ethics permit, you may face backlash from the broader public or students taking offense at your findings. The authors of the study characterizing convicted rapists in Sweden experienced first-hand that the “wrong” results can lead to social furor, and even lengthy legal battles.

This state of affairs undoubtedly has the effect that many researchers in Sweden refrain from studying the topic, even if it is not strictly disallowed. Why study a topic with additional barriers and restrictions, and where the social risk is so much higher than many other interesting topics? The real answer is because it’s important, but scientists are people, and surely many deem it too risky territory to tread.

1. Ahlberg, J. (1996). Invandrares och invandrares barns brottslighet: En statistisk analys. The pdf can be found here.