Overrepresentation in crime — what Brå 1996, 2005 and 2021 actually show

Brå 1996:2, Brå 2005:17 and Brå 2021:9 are the three official research reviews that measure excess risk of criminal suspicion among foreign-born and their children.

All three show the same qualitative pattern: excess risk exists, varies with crime type and origin region, remains after age-standardisation, but does NOT disappear entirely when controlling for socioeconomics.

After age control, the excess risk of suspicion for all offences is 2.5 times higher for the first generation compared to native-born with two native-born parents. Robbery: 4.1 times. Sexual offences: 2.1 times.

For the second generation (two foreign-born parents), the excess risk is 3.2 for all offences and 5.7 for robbery. This is the most important and most misunderstood figure: the second generation has HIGHER excess risk than the first in some categories.

Non-European origin has higher excess risk than European. This gradient is clear in tables 5.1–5.6.

Suspicion is not the same as conviction. Reporting propensity can vary between groups.

Age-standardisation does not capture all age-related differences since crime propensity is not linear.

Socioeconomic control: when Brå controls for parental income, neighbourhood and own education, the excess risk falls by 30–50%, but does not disappear.

Individual prediction: the statistics say NOTHING about individuals. A first-generation non-European is statistically more likely to be suspected — but still ONE non-suspected individual in their 20s (based on baseline figures).

Excess risk has been relatively stable between reports. Absolute suspicion levels have followed general crime trends.

Prison Service statistics: the share of foreign citizens in prison rose from 25% (2010) to 32% (2024). This reflects both excess risk and immigration volume.

Share of suspects who are foreign-born: 38% (Brå 2021:9). Share of population: 16%. This is robust overrepresentation.

Overrepresentation is a group average, not an individual property. That a group has higher excess risk does not mean an individual member is dangerous.

It explains part of local safety developments — particularly in areas with high concentration of foreign-born with short education.

Politically the numbers are hard to handle. Denying them undermines credibility. Overinterpreting them undermines the rule of law. Brå's own conclusion: 'Differences between groups should be analysed to understand causes and design interventions — not to allocate blame.'

Criminologist Jerzy Sarnecki has consistently argued that socioeconomic factors explain most of the overrepresentation — controlling for parents' income, education and neighborhood shrinks the gap by 60–80% (Sarnecki 2021, 1985 cohort). Brå's own reports 2021:9 and 2024:5 confirm the direction but land at 40–50% reduction — meaning a remaining ethnic component.

Another reading from Kristina Sundquist (Malmö University) is that upbringing conditions in the receiving municipality (segregation, school support, overcrowding) drive more than the country of origin — i.e. cohorts born in Sweden to foreign parents should track Swedish-born once upbringing conditions are equalized. There is some empirical support but it is contradicted by the persistence of differences even in cohorts with high education.

Brå's overrepresentation studies build on convicted persons, not reported ones. That means differences in clearance rate, willingness to prosecute, and trial outcomes affect the figure. If two groups commit equal amounts of crime but one is prosecuted more often, 'overrepresentation' arises without an actual behavioral difference.

The studies cover register data from the 1985 and 2004 cohorts respectively. Current cohorts (2015–2020) will not be reported until 2028–2030 when sufficient follow-up time has been reached. Conclusions about 'today's immigration' therefore cannot be drawn directly from Brå's existing material.

The analysis does not say that an individual person of foreign background is more prone to commit crime. Group-level statistics can never predict individual behavior — attempts to do so ('ethnic profiling') are both methodologically wrong and legally impermissible.

Nor does it say what should be done politically. That an overrepresentation exists is a factual question; how society responds is a normative question requiring balancing of several values (integration, rule of law, effectiveness).