Gang related violence in Sweden has been a hot topic over the last couple of years. Within Sweden the discussion tends to polarize quickly. There is very little between “we have no problem” and “we’re going down”, between no discussion and a complete mishmash of information.
This situation is the reason I wanted to find a representation of what the development has been like over a longer period of time. So this blog post shows my attempt to build such a representation and the results from it. The idea is to make the foundation on pure numbers, and not opinionated in any way. To make my task practical I need to simplify it by limiting and defining what to look for.
Framing The Problem
I cannot work with “violence” in general, it is too broad. To limit it but still have something representative I chose to look at shootings only. As far as possible I have tried to work only with confirmed cases perpetrated by non-police individuals. Also, only shootings from firearms not including air guns. They must be public, either outside or in a building open for all.
Sweden as a whole is also too big to cover. Any of our three largest cities should be representative though. Here Malmö (third largest) has gained some notoriety over the years. One hypothesis claims that Malmö’s proximity to continental Europe leads to crime entering and spreading from the city. Anyway, Malmö is my choice of representation. A single city also makes for more interesting map plotting.
Next are reports on individual shootings in Malmö. For this I use articles published by the local newspaper Sydsvenskan (in Swedish). I have found them to report diligently on shootings, also smaller ones. Police reports would be too troublesome for me. I also want the source to be transparent for people to check for themselves.
To find the articles on Malmö shootings I made a simple Python script to scrape Sydsvenskan’s website (in a friendly manner). Often enough information is given on date and location of the shooting, but in cases where details were missing I also made use of local freelance photographer Patrick Persson’s reporting on PPPress, and Swedish discussion board Flashback’s crime thread.
Malmö Shootings Results
At the time of searching, articles on Sydsvenskan were indexed and easy to search from 2001 and forward. I put in some more manual labor to also include the year 2000, and put 1 January 2000 as the starting point for this investigation. Results are available as csv file here.
Looking At The Bare Numbers
The histogram below shows the number of shootings in Malmö per month (thin dark blue bars) and year (wide blue bars) from 1 January 2000 to 31 December 2018. In total I have found 469 shootings in this period.
The second histogram below shows the number of shootings per month in blue bars, and the number of lethal shootings in red. Note that this is not the number of dead, but the number of shootings resulting in one or more deaths.
These graphs show a couple of noteworthy details. In the year 2000 there was no reported shooting, the only clean year in this period.
The surge in 2010 was a result of coinciding events, a serial shooter targeting immigrants (suspected 15 attacks) and increased fighting between two criminal gangs. The gangs were labeled M and K, and their story is covered in a 2011 article in cafe.se (Swedish) and a 2015 article in sydsvenskan.se (Swedish). This period also saw a wave of bus shootings using air guns, but as per my rules those do not count here.
The increase in 2016-2017 again follows from criminal gangs fighting. As of writing (January 2019) the last month without a reported shooting was January 2016, so 3 years ago.
I have not been able to locate all 469 shootings within Malmö, but 467 of them. Here they are plotted on top of a map cropped from OpenStreetMap. I have chosen to gather data from within the surrounding motorway Yttre Ringvägen. Each shooting is represented as a two-dimensional Gaussian function and put down as a marker on the map. When close together, overlapping parts add together. In this way, a single shooting marker has a dark red center, but when overlapping the color becomes brighter. Thus bright red indicates multiple shootings within a limited area.
In order to visualize the development over time I made a video running from 1 January 2000 through to the present day. To see how the situation changes over time the markers have a half-life of 6 months, which means they fade away over roughly 4 years. Deviating from mathematical strictness, you can imagine them as fading from people’s memory.
It is quite obvious from the numbers that the situation have worsened significantly since the early years after 2000. For instance, half the shootings took place over the last 5 years. At the same time, the numbers paint a history over the last two decades but should not be used for predictions. If anything, we can say that we are in a rising trend and that the trend has not yet reversed.
Generalizing a bit from the results we see that the number of shootings in Malmö have risen from a couple per year to several tens per year. So it is not unreasonable to argue that the number of shootings have increased by a factor 10 over the past two decades.
Sydsvenskan published articles twice in 2018 stating that shootings in Malmö are in decline. Somewhat ironically, shootings flared up after both publications. The May article said shootings had dramatically reduced compared to the years before. One month later the largest single shooting to date took place, leaving 3 dead and 3 wounded. The October article claimed that shootings are down compared to 2016-2017, which is correct as we can see in the graphs above. But November and December added another 16 shootings, leaving 2018 the fourth highest on record.
I have no wish to paint in black, but news media tend to judge too quickly. Looking at the graphs above, I think you should wait a year or two before arguing that the trend has shifted. After all, the numbers change all the time.
I am not going to attempt to assess how good or bad the situation is. That is a question for debate, I just wish we could do away with the knee-jerk subjective emotional responses.