Eight measures, two eras
The table below is the most compressed version of the change Sweden has gone through. Each row is sourced from official statistics published by Swedish authorities or international research institutes. The metrics were chosen because they are reported consistently across both periods and are widely used in academic research.
| Mått | Then (1990s) | Now (2020s) |
|---|---|---|
| Fatal shootings per year(Brå) | approx. 10 (1996) | 53 (2023) |
| Reported bombings(Police) | < 30/year (1990s) | 363 (2024) |
| Share foreign-born(SCB) | 9 % (1990) | 20.5 % (2024) |
| Share with foreign background(SCB) | 13 % (1990) | 27 % (2024) |
| Police 'vulnerable areas'(Police) | 0 (term did not exist) | 59 (2023) |
| PISA reading literacy(OECD/Skolverket) | 516 (2000) | 487 (2022) |
| Interpersonal trust (high)(SOM Institute) | approx. 62 % (1996) | approx. 55 % (2023) |
| Women feeling unsafe out late(Brå NTU) | 27 % (2007) | 34 % (2023) |
Two opposing readings
Swedish criminology contains two main camps. Jerzy Sarnecki has long argued that crime levels are stable when measured per capita and that media narratives exaggerate the change. Ardavan Khoshnood and Amir Rostami argue the opposite: gang violence in particular has reached levels without historical precedent in a Western democracy in peacetime.
The level of organised gang violence we now see in Sweden has no equivalent in Western Europe. We have to stop comparing ourselves to ourselves and start comparing ourselves to our neighbours.
Why the comparison matters
Most of the metrics here are tightly correlated with one another in international research: declining trust, declining school results, rising violence and rising segregation tend to move together. The Swedish case is unusual not because the direction is unique, but because the speed is.
Sources
Read the full investigation of how Sweden has changed.
Back to the overview →