Risk Per Trip

Travelling In the Realm of Very Small Numbers

Related articles:
There's a lot of controversy over the relative safety of different modes of transport. At least, among people who think about these things there's a lot of controversy. The average Joe and Jane almost certainly believe what the auto industry, the insurance industry, and the highway lobby have been telling us for 60 years: Your Family is Safer in a Car -- and the heavier the car, the better. However, among cyclists and bike/ped advocates, the debate rages on: is cycling (walking) really more dangerous than driving? And if so, by how much? This essay touches on one aspect of the "risk problem" -- oddly enough, one reason that it's hard to make sense of this vexed subject is because road risks are so very low.

Ironically one of the hardest challenges facing the "highway safety" wonks is the fact that road travel is so darned safe. It's hard to get people to modify their behaviour based on such small risks: each person knows, and rightly, that the odds against their survival are altered only infinitesimally by speeding just a little, by running just one stop sign, etc.

Out of a motorised population of 286 million (USA), even a tiny, one-in-2-million risk per trip adds up to 3 figures of fatalities per diem. Which seems large at first glance (35,000 people die in cars each year, the equivalent of several commercial airliner crashes per week!).

On the other hand, in that same year 120,000 people will die, so the actuarists tell us, of medical misdiagnosis and misprescription; and the anti-tobacco lobby tells us that nicotine addiction accounts for 400,000 lives annually. In the same year, fewer than 800 US cyclists will die on the road. Many people honestly believe that riding a bike is far more dangerous than smoking, but in face of these figures I would tend to disagree :-) About 1 in 250 smokers develops lung cancer every year, an annual risk factor that would make me (personally) sit up and pay attention.

In my essay In Defence of the Bicycle, I said
We must also bear in mind that all road risks experienced by Americans are relatively low on whole-pop scales of risk. Let's take a whole-population view of annual risks: there's a 1/600 chance that you will commit suicide, and a 1/11,000 chance that you will die as a car occupant, this year. Your average chance of untimely death from all causes is about 1/115. Your population-based risk of dying while cycling this year is only about 1/130,000.

Here one must rightly object that only 6 percent of Americans actually cycle regularly -- of course the whole-pop risk looks low! Sadly true, and so we must adjust that risk upward by a factor of about 16 to compare it to the whole-pop risk of death in a car. [This factor of 16 is based on an assumption that almost 100 percent of Americans either ride in, or drive, cars regularly.] If almost every person in the US cycled (as many as now travel by car), then projecting from current risk levels, we would expect the cyclist annual fatality risk to be 16 times higher or 1/8,125 ... in the same ballpark as the 1/11,000 risk from driving or riding in a car: less than 1.5 times as risky.

A local cyclist and bike advocate responded:
>>     i also object to this reasoning.  using your numbers,
>> the risk would be about 35% greater.  i don't think that's
>> a negligible amount.

And this is in fact what most people think when they read that cycling is 1.5 times as dangerous, or twice or even 11 times as dangerous, as driving. They think "that's a huge margin, of course I would be smarter to drive."

But when assessing per-trip risk (should I drive, walk, or bike to the store today) we have to remember that we're travelling in the realm of very, very small numbers. Modest ratios -- even steep ratios -- between numbers this small are not strategically relevant.

What I mean is that the risk per unit (mile) is so tiny as to be "imperceptible" to either driver or cyclist. A 35% increase in a very tiny risk doesn't make a large difference for an individual.

To put this more vividly: if I decide to go swimming today, as compared to never going swimming, I increase my risk of drowning infinitely (any positive number is infinitely larger than zero). But the risk of drowning if I go swimming today is so very slight, that the difference between doing so and not doing so is not strategically relevant. So I'll go swimming if I feel like it, rather than agonize over whether I should take such a minute risk. Similarly, taking two glasses of wine a year as opposed to one glass increases my unit (glass) exposure to alcohol by 100% -- doubles it! Yet it does not increase my lifetime risk of cirrhosis of the liver (a risk of alcohol consumption) by any perceptible margin. So I am unlikely to worry about this difference.

Or a third approach: according to national US health statistics, you are twice as likely to die of appendicitis as of meningitis this very year. Twice as likely is a large factor -- a good bit larger than the mere 35% increase mentioned above. But this is not a good reason to be very worried about appendicitis -- the risk of death from meningitis is one in a million, and from appendicitis it's one in half a million. These odds are so slight compared to the overall risks of routine existence that there's no need to worry about the ratio between the two.

This is how I feel about the per-trip risk difference between cycling and driving -- I don't worry about any one particular trip today, more than I worry about dying of meningitis this year, or whether I should or should not go swimming or have one glass of wine on New Year's Eve.

Why? Because the per-trip risk is very, very small, just like the risk per glass or the risk per swim.

Suppose we accept (pro tem) that the annual risk (equalized) of cycling death is 1/8,000 or so (see the essay referenced above), then the risk per diem (assuming semi-equally-distributed quotidian cycling habits) is of order one in 2.8 million. Here I'm assuming 350 cycling days per year -- there might well be 15 days a year when you don't go anywhere for any reason). One in 2.8 million is a risk so small as to be "way down in the noise".

The difference between 1/2,800,000 and 1/2,000,000 (35 percent greater risk) is not perceptible for any single trip. My bet is that this difference is a far smaller effect (in terms of the relative danger of your bike ride) than, e.g., how much sleep you got last night, or whether you are distracted by worrying about something else while you ride, or whether your brakes are in good shape. Of course the health risk you incur by choosing to be sedentary on this particular trip (by car) rather than get some exercise (by cycling) is also tiny for each given instance.

For perspective, one in 2 million is about your risk (this year) of dying by falling out of bed. Or of being struck by lightning and killed. You are twice as likely (one in 1 million) to die in your bathtub this year. For me personally, a per-trip risk in this ballpark is not even on the radar.

I insinuated above that the "difference that makes no difference" between the risk of a car trip and the risk of a bike trip is more slender than the difference I might make by small alterations in my own behaviour. Basically, I think this is so, and therefore I do pay attention to some factors I can control, like attentiveness, condition of my bike, horn/mirror/lights and that kind of stuff. These, I feel, reduce my bike risk per trip way below average. Something that is important to remember is that the average cyclist fatality may not be the average cyclist :-)

The Bicycle Institute of America says there are 100 million cyclists of which 31 million ride at least once a week and of which seven million ride almost daily.

[nb this is a bit higher than the 6 percent estimate I quoted earlier, and of course these estimates are absolutely crucial in both risk/mile and risk/hour calculations; the errorbar here could be very large indeed]

Those riding the most frequently have a much lower risk per mile, hour, and trip. In fact, an unusually high number of "cyclists" get killed between midnight and 3 AM, which suggests that the average fatality might be very much different from the average bike rider, just as the average vehicular cyclist has a far different risk from the average bike rider.

[Ken Kifer, online discussion, carfree group]

Something like half of our national adult cyclist fatality stats are associated with night riding (and many of those with inadequate lighting). I am not saying that therefore all bike fatalities are the cyclist's fault, by any means. Nor do I assert that riding at night is bad and no one should do it.

Two deaths in CA this very year (Hoffman and Robertson) were pretty clearly the fault of careless (Hoffman) and deliberately hostile (Robertson) drivers. Those drivers were not average drivers, though, thank goodness. The cyclist's risk of encountering a murderous or lethally incompetent driver is rather like your chances of encountering a homicidal sociopath (or an incompetent surgeon, an arsonist, etc) in your non-velocipedal hours. There are jerks, idiots, and sociopaths scattered throughout society. It's true that allowing them to drive gives them too much power to hurt and kill others... but that's a different topic :-)

Most drivers are neither terminally incompetent nor sociopathically hostile. Some are slightly careless, most are slightly rude, but very few are really out to get you (or themselves) in any spectacular way. It's possible to compensate for the limitations of motorists.

Taking all this under consideration, it's my impression -- after the necessarily limited statistical analysis I've done -- that the cyclist who has good road skills, errs on the side of caution most of the time, keeps the bike in good shape, and stays fairly alert, can reduce his/her personal cycling risk far below the average. No one can quantify the exact effect of each component of this "safety kit," but my bet is that in combination, the sum of these small positive effects at least cancels, maybe more than cancels, the differential in risk between driving and cycling.

In other words, the national averages may show that it is 35% more dangerous to ride than to drive -- but what if you can make yourself and your bike 35% safer than the national average?

To me, bike safety is more a social justice issue than a "direct threat to me personally" type of issue. I have little worry about the 1-in-2M odds of getting wiped out tonight on my way home -- I probably worry more about the much scarier odds on getting cancer (skin or otherwise).

But it does really bother me that if by some remote chance I did get nailed by a driver on my way home, (a) I would probably be blamed (just for being a cyclist) and (b) the driver would almost certainly suffer no legal or social penalty whatsoever, unless he or she was DUI at the time. I would be dead, and the driver would be driving home without even a license suspension. With this, I have a real problem. Even if the chances of this happening are very very slight (and national stats indicate to me that they are), the injustice bothers me. I often think that we should talk less about "bike safety" (with its invidious implication that cycling is inherently dangerous) and talk more about bike justice.

Unfortunately when we exaggerate the danger of cycling, I believe this only feeds the prejudice against cyclists. As long as we define cycling as a perilous activity which only reckless people would undertake, cyclists will tend to get the blame when we are hit by cars.

De Clarke