"Dangerous" streets? when the media tries to crunch its own numbers
OpenFile has a young and bike-friendly group of journalists writing and researching for it. It has just produced a pretty cool map of traffic "accidents" called OpenRoad where people can choose start and end points of a journey to see the number of crashes along the way that were reported to the police. (I'm going to stick to calling them crashes since we don't know if they were actually accidents or if they could have been prevented). The Globe and Mail created a similar map not too long ago, using the same / similar data. I wrote up my criticisms of that map and most of the same criticisms stand for OpenRoad. OpenFile, in this initiative, is misleading the public more than they're helping. They've failed to provide context for the numbers and the whole project implies that they are helping people find the most dangerous intersections.
OpenFile is aware of the problem with the data and notes that "A route with more accidents isn't necessarily more dangerous for each individual on it." Popular cycling routes will inevitably have more crashes than routes where no cyclists venture. Their map is unable to tell us why an intersection has lots or few crashes. Is it because it gets used a lot by cyclists? Or does it have bad sightlines, heavy traffic, or lots of potholes? We have no idea based on their map.
It begs the question of the usefulness of the map at all if we can't even use it to help us make decisions on our routes.
With the one hand OpenFile explained the limitations of the OpenRoad map, but with the other hand they were exploiting it for headlines like "MONTREAL'S MOST DANGEROUS INTERSECTION FOR CYCLISTS" and a tamer headline "Vancouver’s intersection with the most bicycle-vehicle collisions", which still contained the statement "It will come as no surprise that Main Street at East 2nd Avenue, a hub of both bikes and cars, is the most dangerous cycling intersection in Vancouver with 10 accidents reported to police between 2007 and 2010".
No, that's not how it works.
The media now seems to be more interested in playing with cool new tools rather than consult with experts - in this case bike safety researchers - in order to understand how we can actually measure "safety" and "danger".
We don't have good cycling counts in Toronto. The best we have is the Downtown Cordon Count from 2010 but it will serve the purpose for my argument. Let's compare College and Dundas. The report counted the number of cyclists passing a cordon over the day and the western boundary was Spadina. Using the OpenRoad map I selected a route on College from Bathurst to Beverley and the same from Dundas. I then can make a rough comparison of crashes to the cyclist counts on these two streets.
|Street||Cyclists (per day)||Crashes (2007-2010)|
There are 3 times as many cyclists on College as Dundas, but only 2 times as many crashes. (I am making what I think is a safe assumption that cycling traffic on College tends to always be about 3 times as high as Dundas). From the viewpoint of an individual cyclist which street would you think would mean less risk of a crash? Going by the OpenRoad map and headlines I would venture that OpenFile would proclaim Dundas as the safer street and College as more dangerous. College has twice as many crashes after all. But as soon as we take into account how many cyclists are actually travelling on these two streets College starts to look a lot better than Dundas.
Ideally in such investigative journalism it would be nice to have a map like this, which maps the pedestrian collisions with a denominator of pedestrian traffic. But we can't get that. So what is the point of OpenRoad?
OpenFile and the Globe and Mail have, at best, just provided yet another fun-looking but pointless widget, and, at worst, helped to persuade people to choose cycling routes that are actually more dangerous than their tool would tell them. Anyone up for biking on Steeles? It has a heck of a lot fewer crashes than College Street.