Carnegie Melon and AAA maintain a website with a lot of great transportation safety stats (It works better in explorer than firefox I found).
Calculations are made using information from two widely-used national databases, the Fatality Analysis Reporting System (FARS) and the National Household Travel Survey (NHTS). Combining information from these two sources provides access to travel-risk calculations that go far beyond what can be found elsewhere (either online or in print). Users can explore the travel risks for millions of different combinations of transportation modes, demographic variables, and a host of other parameters.
Comparing travel risks requires the calculation of risk measures, which TrafficSTATS in terms of fatalities per unit of exposure. Just because a particular mode of travel has more fatalities associated with it, does not mean that it is riskier. In 2004, for example, there were 18,819 fatalities for vehicle occupants in passenger cars compared to 3,779 motorcycle fatalities. With five times more deaths in cars, do cars present a greater fatality risk than motorcycles? No! Because passenger cars are driven much more than motorcycles, their risk per 100 million miles traveled is 1.05, whereas there are 32.61 deaths per 100 million passenger miles on motorcycles.
So how does cycling measure up? Well first of all they combine biking and walking into one category. Oh that irks me. The two have little in common. They aren't done in the same place (sidewalk/street). They go at different speeds. Cyclists wear safety gear (or should). Cyclists signal (or should). It's totally different. Anyway, you can break it out. Here's how the modes break down (I've eliminated some).
Mode Fatalities per 100 million person miles
School Bus 0.02
Charter/tour bus 0.12
Commuter bus 0.24
Van 0.44
SUV 0.82
Other truck 1.03
Car 1.04
Pickup truck 1.12
Bicycle 5.58
Walk 19.73
Motorcycle 31.91
That's nationwide for all users. It gets worse when you separate bicyclists from "other cyclists" (Whatever that is. Is that unicyclists? kids riding in carriers?) It doubles to 11.11 F/100mpm. It goes up again to 11.46 when we consider only our region and drops back to 11.16 when we eliminate those under 21. Yikes. That's still bad. If we want to get more people cycling, we have to find ways to make it safer than that (if you believe the numbers - which come partly from an automobile association). England is having similar safety problems:
bicycle fatalities rose 10 percent to 148 in 2005, the most recent statistics available, and were the only mode of transportation in Britain to show an increase in deaths.
Thanks to Marginal Revolution for the link.
Interesting data. On reading the WashCycle post I was frustrated that the risk assessment used a distance-traveled standard, which didn't seem to be an appropriate way to compare motorized vehicles and bicycles.
On visiting the CMU site, however, I was pleased to see they also compare data using a time-of-travel standard, which I would guess is a better way to compare bikes and cars.
That data shows the relative risk of biking to be much less dangerous than using the distance-traveled standard.
Deaths per 100 million minutes traveled were 0.53 for personally owned motorized vehicles and 0.49 for bicycles. A dramatic difference from the 100 millions miles traveled standard (1.03 for personally owned motorized vehicles and 5.58 for bikes).
Still, I have no doubt that uncaged bicyclists and pedestrians are more likely to die in accidents.
Posted by: Wonger | January 30, 2007 at 01:43 PM
The most exhaustive analysis of the risk of cycling I've ever seen is by Ken Kifer:
http://www.kenkifer.com/bikepages/health/risks.htm
Of course, the underlying problem is that in calculating any cycling accident rates the denominator is going to be somewhere between a loose estimate and a wild guess, as there are no reliable statistics about bicycle use.
Posted by: Contrarian | January 30, 2007 at 02:44 PM
I guess I'm confused as to why fatalities per time is better than per mile. When I have to go some place, I never think I'll travel for x minutes and then come back. Is it that it removes highway/interstate travel (which generally is not done on a bike) from the equation?
Posted by: washcycle | January 30, 2007 at 04:40 PM
I think Wonker's comment "I have no doubt that uncaged bicyclists and pedestrians are more likely to die in accidents" sort of sums it up for me.
I agree that stats like this make it harder to convince people to ride to work. (Of course, it should scare them from walking to work too, I guess.) But it does seem obvious (doesn't it?), that if you're not surrounded by a big medal weapon, you're more likely to get killed in an accident.
Posted by: Chris | January 30, 2007 at 05:00 PM
Maybe looking at fatalities on a per trip basis might yield yet another interesting yet confusing stat. The time travled makes sense to me in that it makes bike trips more comparable to car trips. That is, people have only so many minutes per day in which to travel from place to place. If you travel by car, you probably have to travel farther than someone who can travel by bike, but your time spent traveling might be about the same as the time the biker spends.
Posted by: Nancy | January 30, 2007 at 05:05 PM
If we assume all three (pedestrian, bicycle, car) are involved in an accident, perhaps. Bikes and pedestrians should be more capable of avoiding accidents. And, while single-car collisions often result in death, single-bike and single-person collisions are much less likely to (based on empirical evidence only). But if a car hits a pedestrian or bike (or the other way around) the person in the car is safer.
Posted by: washcycle | January 30, 2007 at 05:07 PM
Or, another way to look at it is "Cars, more than any other mode of transportation, kill people, no matter whether the victims are walking, biking or in cars."
Posted by: Nancy | January 30, 2007 at 07:16 PM
Right, Nancy - I think that's a great way of phrasing it.
Posted by: Chris | January 31, 2007 at 09:41 AM