Listen to this PodcastMore 60-Second Science
[The following is an exact transcript of this podcast.]
There's a new prescription for communities that want to make their streets safer for bike riders: just add more bikes. A team of international researchers looked at cities from Australia to Denmark to California, and found that more riders meant fewer run-ins with cars. The researchers presented their findings to a cycling safety seminar on September 5 in Sydney, Australia.
What's surprising, the researchers say, is that biker safety doesn't seem to correspond to a city's efforts to cut down on accidents. Run-ins between bikes and cars had little to do with miles of bike lanes or lower speed limits. But if the number of bike riders in a city doubled, the rate of bike-car accidents dropped by a third.
Apparently, motorists learn to share the road better when they have to deal with more bikes on their daily commute. Also, more cyclists means more drivers who also bike, which makes them better aware of fellow bikers. The researchers call it a virtuous cycle—run-ins with cars drop with more bikes on the road. And safer cycling means more people strap on a helmet and join the revolution.
—Adam Hinterthuer
60-Second Science is a daily podcast. Subscribe to this Podcast: RSS | iTunes
See what we're tweeting about
10 Comments
Add CommentInteresting that there is no relationship between number of miles of bike lanes and cyclist safety, while there is one between number of riders and cyclist safety. One would expect that miles of bike lane would correspond to increased numbers of cyclists, and so the two would bike lanes and safety would be related nevertheless.
Reply | Report Abuse | Link to thisIt is a case of empathy. The motorists empathize with the bike riders. However, the science of statistics would be capable of detect a trend about the empathy of motorists toward pedestrians? In the metropolises they (we?) - the pedestrians - are to the millions!
Reply | Report Abuse | Link to thisI wonder whether they have the causality right. That is, do more cyclists ride in safer places?
Reply | Report Abuse | Link to thisSince when does correlation equal causality? This sounds like speculation, not science. It's more likely that more safety leads to more bicycling.
Reply | Report Abuse | Link to thisMaybe it is different from every countries. I think that it is a culture issue rather than mere a scientific issue.
Reply | Report Abuse | Link to this"Safety in numbers," the way it's put forth, is junk science. A quick look at most accident causes shows this. Most accident causes involve clueless riding, and simply increasing clueless riding increases accidents.
Reply | Report Abuse | Link to thisSo why do the numbers show increased safety?
First of all, the math underlying the "safety in numbers" assertion is suspect. The assertion comes from Jacobsen's paper, which uses a bunch of equations to make the point. But the equations in the paper don't prove anything. One researcher plugged random numbers from the phone book into those equations and got a graph displaying results similar to Jacobsen's results. The equations give the desired result, no matter what data you plug into them.
Second, increased numbers reflect increased experience among riders, and more cautious riders entering the ranks of cyclists. Thus, you may see better statistics, but the "safety in numbers" effect is NOT making any one individual cyclist safer.
Why is this an issue? Because "safety in numbers" is used to justify the installation of some really dangerous bicycle facilities, such as some of the bikelanes that have led to fatal accidents in Portland, Seattle, Cambridge and Amsterdam. The proponents of those bikelanes dodge the questions about specific accident causes with the questionable assertion that the facilities draw more cyclists, and that more cyclists will automatically make the accident rate go down. This is sloppy thinking!
Those of us who are annoyed by the "safety in numbers" junk science are certainly not opposed to more bicyclists. We're only opposed to junk science being used as an excuse to build unsafe facilities with demonstrated accident causes built-in. The level of safety engineering in bicycle facility design is often quite poor, and "safety in numbers" is used to make it even more so.
-- John Schubert
Coopersburg, Pennsylvania
So far as I know, all of these claims trace back to Jacobsen's paper. I have been criticizing Jacobsen's hypothesis since he first communicated it to me. Jacobsen is coy about his data: another researcher has finally got Jacobsen's California data and is working with it. I have had only the contents of Jacobsen's paper to work with. However, note that all of his plots that look so demonstrative of a declining power curve are made by plotting two ratios using only three values. Examples are Injuries/Number of Cyclists plotted against Number of Cyclists/Population. The fact that the value for Number of Cyclists appears in the denominator of one ratio and in the numerator of the other guarantees a plot that looks like a declining power curve. I demonstrated the same by using four-digit numbers taken sequentially from a telephone directory organized by subscriber's name in the usual way. See the results as shown on my website, johnforester.com under Safety in Numbers. http://johnforester.com/Articles/Social/Numbers%20vs%20Safety.htm. There may be some relationship between proportion of cyclists and their injury rate, but it certainly is not the simple-minded false relationship that Jacobsen thinks he has discovered.
Reply | Report Abuse | Link to thisthis story could be added to the "argument pool" of GRE
Reply | Report Abuse | Link to this:p
Sure! I've gone through some issues like this when taking a GRE test.:D
Reply | Report Abuse | Link to thisJohn Forrester's critique, which is presumably the test on random numbers that John Schubert mentions, is seriously flawed. Forrester implies--or perhaps even states--that any set of random numbers would yield results of the same pattern that Jacobsen found. That is clearly wrong.
Reply | Report Abuse | Link to thisMost important, Jacobsen conducted regressions on several different data sets and in each case, the regression coefficent was statistically different from 1.0 at a very high level of significance. The regression used injuries/capita as the dependent variable (not injuries per cyclist), and cyclists per capita as the independent variable.
Jacobsen also provides some graphs to help one understand the relationshiop in the data, which shows declining relationships between injuries/cyclist and cyclists/population. Forrester objects to those illustrative graphics, as if they were the heart of the analysis. He asserts the truisim that random numbers might also show a declining relationship when portrayed that way. But this proves nothing: The relation between population and the other two variables is hardly random, so an apt comparison would have created a random data set with properties similar to what one would expect if there were no safety in numbers, i.e., population, accident, and cycling would all be highly correlated. (Moreover, Forrester does not really get a relationship that looks like the Jacobsen figures to my eye, though they evidently seem the same to him)
But the main problem with the Forrester critique is that he ignores the analysis upon which the conclusions rely, and then complains about an illustrative graphic as if that was the analysis.