The Unsafe Curves of Sweden

Category: Best data-driven reporting (small and large newsrooms)

Country/area: Sweden

Organisation: Sveriges Television

Organisation size: Big

Publication date: 26/09/2019

Credit: Helena Bengtsson, RIckard Andersson, Fredrik Stålnacke, Oskar Jönsson, Linnea Carlén, Lena ten Hoopen

Project description:

The data team at SVT used methods not previously used by journalists – or experts in the field – to reveal extensive flaws with Swedish roads.They found almost 16 000 curves on Swedish roads that are unsafe to travel on. The roads are constructed in such a way that it’s impossible to keep the car on the road if you travel at the set speed limit. The team was also able to show that more accidents happen in these curves than those who are constructed correctly. A description in English can be found here: https://medium.com/the-svt-tech-blog/finding-16-000-unsafe-curves-427889afde8b

Impact reached:

To start with, the findings were used in a 45 min documentary that was broadcasted at Sep 25th. There the result of the analysis was only mentioned briefly. On the morning of Sep 26th, there were broadcasts and articles published on the web, in both national and local news. All 24 local news outlets did local stories of their own. There was a panel at the national morning show and news segments in all broadcasts from morning until night. At the nine o’clock news that night the Minister for Infrastructure, Tomas Eneroth, called for change and that there should be signs posted along the unsafe curves. We’ve also had a follow-up question in the parliament from a member of the parliament to the minister. There were over 100 articles written about the story in mostly local media – and the team was also invited by a group of analysts from the Swedish Transport Administration (the agency that are responsible for road safety in Sweden) to discuss how the data was used and the methods of the project. 

Techniques/technologies used:

Every state road in Sweden is measured by a measuring car that drives around and collects all kind of data about the road: the radius of the curves, how uneven the surface is, how much (or how little) the road leans to give a few examples. The data is collected meter by meter, but it’s stored with an average of 20 meters. This data is stored in a database that is publicly available through an API. (PMSV3) A selection was made from the team to look at roads with at least one accident per year for the last five years. We excluded parts of the roads that had a speed limit lower than 70 km/h and also roads that were separated in the middle. In all we looked at 26 000 km of roads and we did this in 100-meter segments. For each segment we first determined whether it was a curve or not using another formula found among guidelines for constructing roads. (VGU 2015:086, s 106) Once we concluded that the segment was a curve, we calculated a formula for friction for each segment. This formula calculates the friction needed to keep the car safely on the road using the speed limit on the road, the radius of the curve and the cross slope of the road. 

We then had to turn these segments back to curves, and in all we did analysis on about 1.1 million segments. We found a problem with the data for round abouts and some cross roads. The measuring car either lost data or had data that was abviously wrong. We had to use OpenStreetMap to identify and exclude those “curves” that were not really curves. 

See a longer description in English here: https://medium.com/the-svt-tech-blog/finding-16-000-unsafe-curves-427889afde8b

What was the hardest part of this project?

Knowing that our method was ok. We talked to several outside experts. We read a lot of scientific reports on the subject. We started to write a methodology very early on in the project. We also shared that methodology with all outside experts and also with the agency. We did this early on and asked for feedback – long before we did our interview. 

Having a mixed team of journalists and developers was crucial for this project. Just dowloading the data and making all the calculations needed would have been hard without the expertise from the system developers involved in the project. Getting data and findings to our local news outlets was another challenge. We ended up creating unique pdf-files for each outlet, containing a description of the method and chosen roads from their area. We also spent a lot of time on the phone with our local journalists, explaining what we’ve done and suggesting angles and stories for them to do. This was also what really made the project, since the roads in question are mostly in the country side and through the reporting we heard voices from people affected by these roads.

What can others learn from this project?

To work closely with developers – also for analysis and data gathering – not just for visuals. To make sure you write a methodology early and keep revising and adding to that through out the project. Not to be afraid to reach out to experts in the field – they can help you get things on track. 

Project links: