A study released in Copenhagen titled “Road safety and perceived risk of cycle facilities in Copenhagen”, is sometimes referenced by local Vehicular Cyclists as proof that cycling infrastructure poses a greater safety risk, though they typically avoid the paper’s conclusion which clearly advocates in favor of infrastructure. Noting this confusion, a BFOC commenter posted the following request:
I’d appreciate a thoughtful post about this study, which is about the effects of bike infrastructure in Copenhagen. I have doubts about the methodology, it not being explained real well, but the conclusions include:
Cycle tracks increase cycling 18-20%
Cycle tracks increase accidents 9-10%
Cycle lanes were less effective at increasing cycling and it was unclear if they raised accidents more than cycle tracks
The paper did not discuss to what extent (if any) all this infrastructure caused cyclists to lose road riding rights they enjoyed previously.
To clarify the paper, we contacted the study’s Danish author, Soren Jensen, for a breakdown, and asked if bicycle facilities would benefit Dallas. We also communicated with Dr. Lon D. Roberts, author of SPC (Statistical Process Control) for Right Brain Thinkers and Professor of the course Gleaning Facts from Figures to provide a better understanding due to confusing concepts in statistical analysis. Before getting started, some definitions are in order…a cycle track (pictured above) is simply a dedicated bike lane, which is physically separated from car traffic, while a bike lane is simply a painted line on a street that has no physical barrier between auto lanes.
Below is our conversations, starting with Mr. Jensen:
Hello Jason,
The cycle tracks (kerb between drive lane and cycle track, and kerb between sidewalk and cycle track) increase cycling by 18-20%, whereas cycle lanes (only a 30 cm wide white marking to drive lane) increase cycling by 5-7%. These figures have been found for streets in Copenhagen. I do not know, what the figures would be in Dallas, but because cycle facilities are seldom in Texas, I believe the figures would be higher (much higher).
The cycle tracks have resulted in an increase of 9-10% in both accidents and injuries. The study accounts for confounding factors (general safety trends, changes in traffic volumes and regression-to-the-mean), i.e. the stated safety effect is truly the result of constructing cycle tracts. There are several reasons for the increase in accidents, however, the most dominating one is that construction of cycle tracks on main roads often leads to a parking ban on these roads, which then leads to many automobiles being driven onto sidestreets and being parked there – and this leads to many more accidents at intersections between main roads and side streets. If parking is not banned on the main road then there most often is no change in safety when cycle tracks are being constructed. The cycle lanes have resulted in an increase of 5-15% in both accidents and injuries – i.e. practically the same as cycle tracks. Both cycle lanes and cycle tracks are most often about 2.0 metres wide in Copenhagen. All investigated cycle tracks and cycle lanes are one-way, i.e. there is one cycle track/lane in each side of the road.
The Traffic Act in Denmark clearly states that if a road has cycle tracks or lanes, then the bicyclists must use them, i.e. they must not ride on drive lanes or sidewalks – only children up to 6 years old may ride on sidewalks.
In the past 5-8 years, the construction of cycle tracks have often been a part of a more comprehensive “campaign” in order to get more people cycling in many communities in Denmark. The size of the possible spil-over or synergy effect on cycle traffic volumes, which might come from mixing new cycle facilities with campaigning, is unknown – and is also may have implication on safety. But I do not know if an increase in bicycle traffic in Dallas will lead to better or worse safety overall (total number of road deaths etc.) – but I do know that it will lead to better safety for the bicyclists.
Kind regards,
Søren Underlien Jensen
Soren,
Could you clarify one part for me? You stated that
“The cycle lanes have resulted in an increase of 5-15% in both accidents and injuries ”
but ended by stating “I do know that it will lead to better safety for the bicyclists.”
If you’re seeing an overall increase in accidents, would this not lead to less safety for bicyclists?
Jason
Hi Jason,
There exists clear relationships between the traffic density and traffic safety for all travel modes (from walking to aviation – probably even space shuttles). This relationship says – higher density, better safety. An example is P L Jacobsens paper http://injuryprevention.bmj.com/cgi/content/abstract/9/3/205
Chers
Søren Underlien Jensen
Got that? I too was a bit unclear. So, I contacted Dr. Lon D. Roberts, author of Statistical Process Control for Right Brain Thinkers to explain in greater detail.
Jason,
Like with most studies involving statistical analysis, one must be careful as a case can be made for and against a particular position, depending on how the users of the data choose to “cherry pick” the results. Looking at the big picture, based on Soren’s study, his comments to you, and the other study he referenced, there is a simple syllogism to use that can be backed up by the numbers and percentages that shows favor for the development of bicycle facilities.
* The likelihood an individual bicyclist will experience an accident goes down as the number of bicycle riders go up.
* So, if an increase in bicycle ridership leads to a lower likelihood that a particular individual will have an accident, how do we increase bicycle ridership? According to Soren’s study, the way to do this is to construct more cycle tracks and cycle lanes.
* Conclusion: construct more cycle lanes and tracks.
In formal logic this would be expressed as follows:
Major Premise: The per capita likelihood of a bike accident decreases as the number of bike riders increases.
Minor Premise: The number of bike riders increases as the number of bike lanes and bike tracks increases.
Conclusion: Increasing the number of bike lanes reduces the per capita likelihood of a bike accident.
Keep in mind that “inferences” can be drawn from statistics, but this is not the same as saying the statistics “prove” a particular conclusion. (Every statistician will tell you that you can’t “prove” anything with statistics.)
While the bike lanes do not seem to have an effect one way or the other, if someone tried to use Soren’s study to “prove” that an increase in cycle tracks increased accidents by 9%, they’d be guilty of cherry picking the numbers. The accident rate may have increased by 9%, but the number of bicyclists increased by 18-20%. (This is consistent with the Major Premise cited above.)
Using Soren’s percentages, here’s an example starting with the assumption that 10 bicyclists out of 10,000 will experience an accident over a certain period of time if there are no bike tracks:
1. On an individual basis, there’s a 10 out of 10,000 (or 0.1%) chance that an individual biker will experience an accident if there are no bike tracks
2. When the bike tracks were added, the accident rate increased by 9%. In other words, if there are 10 accidents without the tracks, the number of accidents increases 10.9 (or approximately 11). On the other hand, the number of bike riders increased by 18%, from 10,000 to 11,800. Therefore, on an individual basis the likelihood of an accident with the tracks added is now 11 out of 11,800, or 0.09%, as opposed to 0.1% without the lanes/tracks.
Even if you assume there are 1000 bike accidents out of 10,000 without the lanes/tracks, the likelihood on an individual level is 9% with the lanes/tracks rather than 10% without the lanes/tracks.
So to sum up, individual accident rates dropped when bicycle infrastructure was added, and taking that a step further, Soren’s follow-up correspondence recommends that if Dallas added cycle tracks, ridership would be “much higher”, and “that it will lead to better safety for the bicyclists.” He even recommends maintaining parking on streets to further decrease accident rates.
Special thanks to Søren Underlien Jensen, Dr. Lon D. Roberts, and our commenter, Steve A., for requesting the post.
Bike Friendly Oak Cliff 
While I don’t buy into the Jacobsen conclusions (see url for why), that does not invalidate the other information.
I will have to ponder this very interesting post for a while for the nuggets whithin. One thing that DOES worry me is the notion of 2m bike lanes, at least on roads with only two lanes of motor traffic. Why? I’ve seen many motorists use those “wide” bike lanes as “auxiliary passing lanes” to pass other motorists waiting to make a LH turn. I would not want to be a cyclist in such a bike lane at such a time. I imagine that motorists in Copenhagen are better disciplined. In North Texas, though it might seem counterintuitive (and anathema to some), where bike lanes are provided, we might not want to make them appear to be “extra” motoring lanes. Paths do not pose the same potential problem, even around here. The paths gain further to the extent they avoid crossing hazards. I think I preach to the choir on this, however.
Again, very interesting & thought-provoking post. Thanks, and I, for one, will study it further.
BTW, poles, such as in your photo, would reduce the motorist perception of 2m bike lanes as “auxiliary passing lanes.” Paint alone isn’t gonna do it, but nobody wants to hit a pole at 45mph. As I said, the post is complex and bears careful consideration…
[…] friendly to bicyclists, the infatigueable Jason Roberts, the man behind Bike Friendly Oak Cliff, got in touch with Danish urban planner Soren Jensen the author of an influential report on the impact on bicycles and […]
I went ahead and forwarded your article to Jacobsen to get his take on the doubts you held to try and get an explanation. I’ll let you know when I hear back.
Click on the url for a GoogleMaps link to the “auxiliary passing lane” NRH bike lane if that helps. Traffic is moderately heavy, typically about 45mph. The motorists don’t slow down a whole bunch when they sweep through those bike lanes.
Certainly, I’m sure this isn’t anything like you have envisioned for Oak Cliff, but “almost car lane width” bike lanes seem to me to have a hidden danger when just painted down the side of the road.
You’re correct…this is nothing like I’d envision.
There’s two keys for a successful paradigm shift in community or “urban” planning. one is focusing on safety (AND the perception of safety…as noted by Soren). The other is building for people over cars.
What we’re advocating, as I’m sure you can tell, is for both.
Adding bike lanes, but maintaining the “cars first” approach, accomplishes neither of the above keys. PM’s recent photo of poorly planned bike lanes is an excellent example of lack of safety, along with maintaining a “car first” model. You might see a slight increase in “perception of safety”, but it still won’t get women out on the roads (that is probably the best gauge for successful bicycling environment). Remember, in the Netherlands, more women cycle than men. Why? A complete separation of vehicle and bicycle infrastructure on majority of streets.
You have to remember one thing we noted in a prior post…how you develop for a suburban zone is vastly different than how you develop for an urban zone. What Europe has done is created a hierarchy of modal priority: first ped, bicycle, bus, car, then truck. How do they do this? Wider sidewalks, bicycle tracks, bus only lanes, small car lanes. This slows car traffic, puts people first and makes streets safe (increases perception and actual safety)
In the US we’ve created the following hierarchy:
car, bus or truck, ped, bicycle…Vehicular Cycling is merely an adaption tool to this model. The reason people aren’t bicycling here is because they know they aren’t given priority. Trying to convince people that a “swim with the sharks” approach is fine, but creates neither riders or a perception of safety. It’s capitulation…and for someone like me, advocating community building and people first planning, thinking that way is pessimistic and defeatist, which are natural traits that come with old age. That’s why most urbanism advocates and planners will tell you is that it’s useless attempting to indoctrinate people beyond mid-life.They have zero buy in, and have reached a point in life that “things will never change”.
The reality is Davis, and Portland have begun implementing the “people-first” hierarchy. Portland is in the mid-stages, and hasn’t completely flipped yet, but once they do, you’ll see the already high ridership rates jump exponentially. The DOT (and Portland’s tourism industry) clearly understands this. That’s why Ray LaHood came out and stated “Portland is THE model for transit planning in the US”. Since this model is now showing high ROI’s in economic development, it’s quickly being transplanted to places like Minneapolis, New York, Boston, Boulder, and more.
So what we’re advocating for in the OC is very attainable, as it’s THE nationwide trend. It’s undeniable, but extremely confusing to those so accustomed to the suburban model. Fortunately, the people-first plan advocates for young, old, disabled, female, and male.
[…] with Danish Engineer, Soren Jensen, to help clarify the findings from his study titled, “Road safety and perceived risk of cycle facilities in Copenhagen”. At the conclusion of the article, Soren summarized that if Dallas added Cycle Tracks to its […]
Here’s the link to Jacobsen’s response to Forester, and your disputes of his study, “Safety in Numbers”:
https://bikefriendlyoc.wordpress.com/2009/07/13/ask-the-experts-paul-l-jacobsen-and-dr-lon-d-roberts-phd/
[…] overall safety by focusing on “people first” road calming. We’ve already noted Soren Jensen’s study of cycle track installation increasing ridership by 18-20%, and Peter Jacobsen’s study of […]
I just read Dr. Roberts’ analysis where he clarifies the difference between absolute numbers and “rates”, suggesting that the absolute number of accidents and injuries went up but the rate went down due to the increase in bicyclists.
The rate of injuries for a stretch of road, of course, is the number of injuries divided by the number of cyclists using that stretch. So if, after a facility went in, the number of injuries went up more than the increase in bicyclists using that facility, the rate of injuries has gone up and most people would conclude that facility made the road less safe for the bicyclists.
A couple of comments.
First, Mr. Jensen stated that the data accounts for confounding factors, including changes in traffic volume. It’s unclear if he means just changes in car traffic or also bicycling traffic, although in context of the question he seems to mean both. If volume changes of bicycling are accounted for, than the increase in bicycling has already been taken into account in estimating the accident and injuries, and these can be taken as indicating a significant increase in the rate of accident and injuries.
Even if bicyclist volume was not adjusted for, cyclists’ injuries increased 22% on the cycle tracks, according to the study, while the increase in the number of cyclists was between 18 and 20%. So there was a slight increase in the cyclist injury rate even if these numbers weren’t adjusted for volume, but probably within the statistical error of the study.
It would be very difficult to conclude from this study that cycle tracks would decrease the rate of bicyclists’ injuries.
Regardingthe assertion that “most people would conclude that the facility made the road less safe” if the number of injuries went up at a faster rate than the increase in bicyclists using the facility. Seems plausible, but not necessarily what “most people” would conclude. Personally, I wouldn’t expect the two variables to track in a linear fashion, whether or not a a facility was installed. In other words, if you put twice as many bicyclists on the road, the accident rate would likely more than double, with or without the facility, simply due to the fact that there is more opportunity for interference between the riders. Not a strong case either way for increasing bicycle ridership . . . unless other safety measures are installed, including rider education. But isn’t that what we did with cars? . . . We instituted laws, rules-of-the-road, driver education, better roads, etc., rather than limiting the use of automobiles to the “elite class.”
@Peter
Here is Dr. Roberts’ response:
“I’m unclear of the point Mr. Rosenfeld is trying to make. I had said that the data shows that with the addition of facilities, the accident rate increased by 9% while the ridership increased by 18% to 20%. My point was just the opposite of the conclusion that Mr. Rosenfeld arrived at.
Also, I still maintain that it is important to be especially cautious when using percentages . . . they are a haven for people who want to manipulate the numbers to make them support their position. Here’s an extreme for instance: If doing (or not doing) “X “(whatever “X “is) resulted in an increase in bike ridership from 1 billion to 2 billion, but the number of accidents also increased from 1 to 2, the rate of increase would be 100% for both the ridership and the accident rate. Also I stand by my earlier point that any increase in bicycle ridership — however that comes about — is likely to result in a corresponding nonlinear increase the accident rate simply due to the fact that there is an increased likelihood for interference between the riders. If the solution is to limit the number of bicycle riders on the road, who gets the privilege and who doesn’t?”
Could you clarify whether the change in accidents was an absolute number ( total number after – total number before / total number before for a fixed period of time) or a rate? By rate, I mean totally accidents/divided by total bicyclists using the road measured over a period of time.
I am unclear as to whether you mean that the number of accidents increased less than the increase in the number of riders, so that the RATE of accidents went down.
Lon,
You have, likely inadvertently, mixed apples and oranges by comparing a 20% mode share for BICYCLISTS and a 10% crash count increase for ALL CRASHES (for all modes: motorists, bicyclists, moped riders and pedestrians). You had some help from Jensen, who wrote the conclusion in a way that described the 20% bicyclist mode share and a 10% increase in “crashes”. You can see this by looking at table 3 at the URL provided [it’s the 4th image of the URL], which is a publicly available FaceBook album I created about this study. You can see that the first line of the talbe shows a 10% increase for ALL Crashes, which includes all modes, MV, BM and Peds. One wonders why Jensen did not tabulate the crash rate for bicyclist involved crashes.
I took his data from table 4 [the 5th image at the URL provided], and summed the counts for bicyclist involved crashes, and used his numbers to arrive at an aggregate bicyclist crash count increase of 30.5% [see the 6th image at the URL provided].
So the proper comparison is a 30.5% increase in crashes for a 20% mode share increase, which means crashes rose 50% faster than mode share for bicyclists when cycle tracks were created. This looks like “Danger in Numbers” in Copenhagen!
a comment from 2012:
As another reader pointed out, the original paper by Jensen states that the figures have accounted for traffic increases. This has showed up in research from the States in the 70’s, 80’s and 90’s, from Finland, from the Netherlands, New Zealand, Germany and Sweden among others. So it seems to be true. Separated cycling facilities decrease safety at intersections while increasing safety along stretches of road.
What’s interesting though is that in safer countries, cycle tracks appear to be safer than separated cycling facilities in countries where it is more dangerous to ride (like the States).
Also, Jacobsen’s paper on safety in numbers has since been supported by studies in the UK, Australia and the US which have all found the same thing.
The reasons for the increased dangers at intersections with separated cycling facilities remain somewhat unclear. The list of reasons seems to be that: the removal of parking increases the number of cars turning across the path of cyclists, that cyclists do not pay attention to drivers when they feel safe, drivers do not pay attention to cyclists when cyclists appear to be safe, cyclists are easily placed in the blind spot of drivers, the separation makes it more difficult for drivers to pay attention to both cyclists and cars that may be crossing their path (when cyclists are on the street, cars can pull in front of a cyclist, move to the right then slow down forcing the cyclist to slow down behind them) and that such facilities attract less experienced riders.
Some of the treatments in Denmark and the Netherlands support some of these views and demonstrate that the Scands have more awareness of the issue than is published in English papers. Some concepts in the Netherlands pull cyclists further away from traffic allowing cars to fully make their turn prior to crossing the cycle track while in Denmark tracks are either shortened, forcing cyclists to ride with cars and become more attentive and noticeable to drivers. The Danes also allow cyclists to stop ahead of traffic, placing cyclists in front of vehicles and presumably making them more visible. Jensen has written a bit on the topic, but I have yet to find any evidence that the Dutch ideas are any safer. They do indeed seem safer.
All this tends to re-enforce Jensen’s idea that safety on cycle tracks can be worse than on roads, but the over all increase in riders across the city (because the cycle track has some flow on effects meaning that people who have been encouraged to ride because of cycle tracks are also going to be riding elsewhere) will tend to make cycling safer in general (safety in numbers) because more drivers will become cyclists and thus be more understanding and cyclists will become more visible and legitimate road users within the community.
^^edit “Have accounted for traffic accident increases”