« Bike-Friendly Allocation of Space on Wards Island | Main | Gov. Bob McDonnell asked about Bike Trail Funding »


Feed You can follow this conversation by subscribing to the comment feed for this post.

So, if these numbers are accurate, and there is good reason to believe they are not, cyclists are involved in more crashes per mile than either

I would draw the opposite conclusion: given the crudeness and closeness of the numbers -- 752, 781 and 656 -- I would say that there is no evidence that any mode is any safer than any other.

Fans of risk compensation should rejoice.

I pretty much agree w/ Contrarian here. Given the the accuracy of the numbers is a bit fuzzy, I feel like it's safe to say that they're all w/in the margin of error.

Even if all the data were perfect, and these were the numbers arrived at, cycling would not be that much more dangerous that traveling on foot or by car...

Yeah, its' definitely within the range of error. And the report notes that it is a small sample size as well. If you take the numbers as exact then it indicates the different, but if you take them as fuzzy or even very fuzzy, then it shows little difference.

Comparing risk per-mile instead of per-trip is wrong-headed, IMO. When I get in my car or on my bike to go to work or to the store, I do so to complete a task, not to cover a given number of miles. Since biking is fast and convenient for short trips, the average bicycle trip is short. Per trip, bicycling is the safest way to go (based on your numbers).

On top of that, limiting the study to on-road bicycling eliminates a safety advantage that cyclists enjoy: they can use fully-separated bike paths for some of their transportation needs. Motorists, on the other hand, do not have the option of, as some motorists put it, getting "out of my way and onto the %$#! bike path!"

In any case, thanks for compiling and posting those numbers for us.

Interesting calculations even though it may always be impossible to get a close estimate of the true accident rates. At least unless all bikes get hooked up to mandatory RFID tags or something like that. (Not saying that is something I would want!)

Just be careful that Tony Kornheiser doesn't catch word of this. He might twist the numbers and exaggerate them on-air!

Jonathon. I disagree. If you had two cyclists - One who makes one 20-mile trips per day and one who makes two 1-mile trips per day - I would not expect the second one to have twice as many crashes as the first. I'd expect the first to have 10 times as many crashes as the second (all other things being equal).

I also think miles is better than the idea of measuring crashes per measure of time. If cyclist 1 bikes 10 miles to work at 10mph and cyclist 2 bikes 15 miles to work at 15mph, you could argue that they should be in the same number of crashes. But cyclist 2 should go through 50% more intersections and be passed by a larger number of cars - increasing their exposure.

Wash, I think you're missing Jonathon's point, which is that a trip is a better measure of utility than distance. I often pick my destination depending on my mode -- for instance, if I need something at the hardware store I'll go to the corner hardware store on my bike but Home Depot in my car. Same utility, different distance. Similarly, when I bought my house I specifically chose a location where I could bike to work. If my intention had been to drive I probably would have chosen some place that was a greater distance but an easier drive (shudder). Again, same utility -- both ways I get to work -- but different distances.

Congestion and parking are much bigger concerns when you're driving, and distance is a bigger concern when you're cycling.

I did miss that point, and it's a good one. It reminds me a bit of the way that when times are tough the price of some items - like corn - might actually go up, because people eat less meat and so the demand for corn can go up. So its a good point that when someone gets used to riding a bike places, their total VMT might go down as they opt for closer locations.

Still, I'm not sure why that makes the use of crashes per trip better than crashes per mile.

Still, I'm not sure why that makes the use of crashes per trip better than crashes per mile.

It's an attempt to factor in the fact that different modes have different patterns of usage. It's often been observed that regardless of mode, people tend to organize their lives so that their commute takes about half an hour. So if you're comparing the commutes of a walker who walks a mile and a half each way, a cyclist who rides six miles each way, and a motorist who drives 15 miles each way -- all taking half an hour -- what is an appropriate metric for measuring the safety of their commutes? If on a given day each has exactly the same chance of dying, is it really fair to say that walking is ten times as dangerous as driving, and cycling two and a half times? Or would you say that for the same outcome -- traveling from home to work in half an hour -- the risk is the same?

The problem is that choice of transportation mode becomes tied up in so many lifestyle choices that trying to compare modes ends up as an exercise in comparing lifestyles.

Another possible metric would be accident rate per hour of exposure.

Incidents per mile is the correct type of measurement if you want to compare across modes on the same infrastructure.

Contrarian, you make an incorrect assumption here and then try to say that justifies your argument:
If on a given day each has exactly the same chance of dying, is it really fair to say that walking is ten times as dangerous as driving, and cycling two and a half times?

You assume an outcome (that the crash rate per hour of driving/cycling/walking is the same) and then try to say that justifies measuring it in hours because if you measured it in miles it would be different. But you just pulled that number out of thin air.

The reason incidents per mile is more accurate is because you are trying to have a measurement that approximates the number of interactions you have during a trip. A car that goes 5 miles along a piece of infrastructure is presumably going to have 5 times as many opportunities to have an accident as a car traveling one mile, regardless of their speed.

MLD, I have changed my mind - sort of. I usually say that the ideal is to take a snap shot and determine the percentage of vehicles that are bikes and what are cars (I know there are more categories, but I'm trying to simplify). If 5% of the vehicles downtown are bikes, then they should be in 5% of all crashes. But I also used to say that per mile was a better measure than per minute.

What I realized, after thinking about it longer is that the first measure is crashes per minute. Which makes more sense to me.

Imagine a cyclist who goes out into the road for a 1 hour ride at 0 mph. They would cover 0 miles. Would they have had 0 exposure? No, I don't think so. A bike going slowly, as in your example, would have plenty of opportunities to be in crashes because they would be passed by many cars. Though they would go through fewer intersections, they'd spend more time in them.

So I'm changing my mind and saying that crashes per minute is the better measure. I plan to post an update that uses this data and translates it into crashes per minute.

There was an article in the Post a few years ago that showed the speeds of cars, buses etc... in DC. Does anyone have a link to that?

I'm sticking to may claim that "crashes per trip" is the best measure of mode safety. The purpose of a transportation network is not to maximize miles traveled, but to get people where they are going in minimum time and with maximum safety.

Example: A shoddy network gets me to work in 55 minutes (via 14th street bridge). A better network gets me to work in 40 minutes (via Wilson Bridge). My exposure went down, so I'd say my commute got safer. By the measure of danger per trip, by commute got safer. By the measure of danger per hour it did not. I you think that better-connected network is a safer network, then "per trip" is the correct measure.

Jonathon, the point is to get a measure that can be compared across modes, not across trips. We're trying to compare biking to walking to driving. Not route A to route B. As you point out, your exposure went down when it go shorter. Therefore there is a relationship between exposure and time.

The comments to this entry are closed.

Banner design by creativecouchdesigns.com

City Paper's Best Local Bike Blog 2009


 Subscribe in a reader