Alternatives to the Automobile

Road Safety, Part I: Asking the Wrong Questions

"Ah," said Arthur, "this is obviously some strange usage of the word safe that I wasn't previously aware of."

-- Douglas Adams, The Hitch-Hiker's Guide to the Galaxy

In the last couple of decades, it's been rare to hear cycling or bicycles discussed in the US media -- or by health or traffic authorities -- except as a "safety issue." No matter where you look, the word "bike" or "bicycle" or "cycling" ends up glued to the words "safety" or "safely" or "safer".

I recently participated in the US Census Bureau's national health survey (2001). When they asked "do you ride a bicycle," I started to smile; I thought that at last, a national agency had realized that cycling is good for people's health. I thought they would ask how many miles a week I cycle and what my average pace is. But no... the only thing they wanted to know was "Do you wear a helmet?". The next question, as I recall, was something about how often I had unprotected sex. It was pretty clear that the survey designers viewed cycling only as a "health risk" activity.

In fact, the vast majority of the American public is now convinced that cycling is a dangerous, risky thing to do. When I ask my car-dependent friends why they use their cars for short (one and two mile) errands, I find that their top reason for not using a bike is that "it's dangerous." My CDF's often express worry or even disapproval when I leave a social gathering on my bike, chiding me for daring to ride a whole two or three miles across town after dark.

Last year, a colleague of mine borrowed a Zap-equipped bike from a government agency for a week. He had to sign a loan agreement, one of whose terms read "In signing this agreement, I acknowledge that bicycles are inherently dangerous." Come on now. Hand grenades are inherently dangerous, TNT is inherently dangerous, and battery acid is inherently dangerous -- but bicycles? Please.

Admittedly this is the litigation-mad US, where product warnings such as "Do not attempt to carve roast while rotisserie is turning," or "Warning! danger of electrocution! do not put radio in mouth!" are not uncommon. But there seems to be a special paranoia about cycling; there's an entrenched belief that the only safe way to get around, the choice any sensible person would make, is to drive a car.

We need to re-examine what we mean by highway and traffic "safety."

When CalTrans widens a road, they often claim that the road is now "safer". What they mean is that the incidence of automobile crashes has been reduced. Usually, they mean that people can now drive at a speed which previously was "too fast" and resulted in spinouts or other crashes.

What highway authorities do not discuss is that widening roads encourages people to drive faster, and that faster traffic makes the road more dangerous for pedestrians, cyclists, wildlife, pets, etc. When CalTrans reports that a road has become "safer for pedestrians", what they mean is that there have been fewer fatalities on it this year than last year. They don't ask whether local residents have stopped trying to cross the road at all, because of its danger. Authorities often practise "safety by displacement;" they report that they have improved conditions, when instead they have made a road or intersection sufficiently dangerous that more vulnerable road users are now (sensibly) avoiding it.

To equate "safety" with "fewer fatalities," without taking into account restrictions imposed on the nonmotorized community's mobility, is misleading. Solitary confinement is a very safe place to be, yet few of us would be thankful for it.

In the UK, traditional planners were appalled by the design of a pedestrian walkway which was separated from the roadway by a row of substantial light posts. "Safety experts" moved the light posts to the inner side of the walkway, because they were "dangerous" -- cars might strike them. The planners thus removed a barrier that protected pedestrians from cars, and made the sidewalk significantly more dangerous for pedestrians.

When we talk about "safety" we always need to ask "safety for whom." We need to ask whether safety has been increased for one sector of the community at the expense of another.

For years, people in motorized nations have been told that heavier cars are safer than lighter cars. People with families in particular, and families with teenage drivers, tend to buy the most tank-like car they can afford. Statistics show that they and their kids will be "safer" inside a heavier vehicle. However, statistics also show that heavier vehicles are more likely to kill the occupants of lighter cars, not to mention pedestrians and cyclists. By pursuing individual "safety," motorists increase road danger for everyone.

When we talk about "safety", we need to consider not only what risks an individual takes, but what risks he or she imposes on others in the community.

The cyclist poses a far lower risk to every other road user (even pedestrians) than the driver. The cyclist also emits no carcinogenic or allergenic toxins en route, consumes only a tiny amount of petroleum for lubrication, and takes up very little road space. By contrast, the petroconsumption, pollution, mass, and size of cars present measurable risks to the community. In all these senses, the cyclist is much safer.

But today our discussions are based on a very selfish definition of safety, a definition which allows the driver to impose increased danger on others while perceiving him/herself as wise, sensible, and "safety conscious." This definition of safety is at the heart of all arms races: in the US, the average weight of passenger vehicles has been steadily climbing over the last decade as the population of SUVs, minivans, and truck-like vehicles increases. As average vehicle weight and size rise inexorably, people desire ever larger and heavier cars in order to feel "safe". There is no end to this spiral -- we now have a couple of families driving their Hummers around Santa Cruz. It's all very profitable for the auto manufacturers of course, and "good for the economy." But it is generally destructive to communities, as well as to ecosystems. The only winner in an arms race is the arms dealer.

When people describe cycling as "unsafe", what they really mean is "more vulnerable," which is not necessarily the same thing.

Even regarded solely as an individual, the cyclist is avoiding risks, not just running risks. The cyclist enhances his or her own personal health by cycling; regular moderate exercise can free us from many of the diseases and discomforts of a sedentary life. (The British Medical Journal published a paper last year in which the lifetime health benefits of cycling were calculated to outweigh the lifetime risk by 20 to 1.) Habitual car-dependence, by contrast, exposes driver and passengers to more airborne toxins than are present in outside air, and deprives them of the healthy exercise of walking or cycling. Health researchers today even ascribe a large chunk of America's chronic back problems to excessive time spent in badly-designed car seats.

Suppose we accept, for argument's sake, a definition of safety that is rigidly individualistic and oversimplified, restricted to the risk of immediate bodily harm through falls or collisions. Even then, we find that both authorities and the public greatly exaggerate the risks of cycling. From the amount of public attention devoted to cycling "safety" one would get the impression that cycling was a leading cause of death and injury, and that any cyclist who dares to use the public roads is (a) a daredevil and (b) probably doomed. Fortunately, it isn't so; it's the cultural bias of a car-centered society that creates and perpetuates this illusion.

If cycling is to recover from a quarter-century of scaremongering, Americans will have to think some new thoughts about risk and safety.

First, we will have to get a better, more realistic grip on relative risk. Cycling is far less dangerous than many other activities which are less stigmatized by "safety experts." If we keep discussing and promoting "cycling safety" rather than promoting cycling itself, Americans will continue to hear (and heed) the implicit message: Cycling is Dangerous. And they will continue to be frightened out of cycling. In Part Two of this article I'll take a look at risk, how risk is measured, the relative risks of common daily activities, and how we might more realistically assess cycling risks.

Second, we will have to come up with a new usage of the word "safe," and make our authorities and planners aware of it -- a definition which takes into account the aggregate safety of whole communities, which gives the safety of the people outside cars equal consideration. In Part Five I'll talk about carcentrism and how cultural bias affects perceptions of risk and danger; in between, we'll explore the wonderful world of risk and statistics.

Road Safety, Part II: Asking the Right Questions

Transit accidents and assaults tend to receive considerable media attention, giving an exaggerated sense of transit risks. In one 8 month period newspapers published 40 stories with headlines linking "transit" and "death," but only 14 linking "auto" or "car" with death, despite the much greater number of fatalities caused by automobile accidents.

--Victoria Transit Policy Institute

Discussions about "safety" are always discussions about risk (whether risk is really the opposite of safety is an open question). Unfortunately, risk is a statistical concept and statistics are difficult for most people to understand and analyze -- many social critics even say that the population of the US is unusually "innumerate" (the numerical equivalent of "illiterate") and that most people don't understand numbers, period.

If this is true, it's a problem: many of the policies and decisions that shape our daily lives, permit or restrict our freedoms, etc. are justified by risk-assessment studies, and increasingly in the last few decades by a methodology called "Cost/Benefit Analysis". If citizens want to assess the wisdom, justice, or benefit of laws, policies, public information campaigns, and so forth, we need the skills to analyze and understand arguments and claims about risk. Without them, we are suckers for scare campaigns, urban legends, and soundbite tactics.

Discussions of "bicycle safety" and road safety in general can be particularly tricky. Let's take a simplified example: a tale of two towns in California (we'll call them Normalia and Santa Peligrosa). The police department of Santa Peligrosa approaches the city manager with a report that contains some very bad news: Santa Peligrosa has ten times as many cyclist injury accidents as Normalia. It's clear, says the report, that there's a serious bike safety problem, and we need to fix it ASAP!

The local press picks up on this report, and soon a headline appears in the hometown papers: "Santa Peligrosa Cyclists at Risk!" and "A Dangerous Place To Ride". The residents understandably feel alarmed -- their town is ten times more dangerous to bike in than nearby Normalia.

But is it?

Let's say that a local bike activist who has ridden in both towns finds this hard to believe. It just doesn't match her personal experience. She will most likely search the newspaper article in vain for the answer to her first essential question: what are the relative populations of Normalia and S.P.? If Normalia's population is one-tenth that of S.P., then the disparity in cyclist injury accidents would not be at all remarkable. On asking for a copy of the police report, she finds that Normalia was chosen for comparison precisely because its population is similar in size to that of her hometown.

Now the cyclist says to herself, "Yes, but when I rode in Normalia I never saw another cyclist. When I ride here I see other cyclists all the time." In other words, although the populations of the two towns are similar, the population of cyclists may be quite different. We would not be surprised if there were very few surfing injuries in Montana, and we would not be surprised if there were very few tobogganing injuries in the Bahamas. If most people in Normalia drive instead of cycling, it would not be surprising that they have few cyclist injuries.

It is very difficult to get a census of cyclists, because bicycles are not registered and regulated as tightly as automobiles. But our cyclist can do some basic guesstimating. For a start, her hometown is a University town, with a population of 20,000 students, and she knows that college students use bicycles more than the average American. She also knows that her hometown contains several parks which are popular with mountain bikers, and that she often sees surfers riding bikes to the beach. She knows that Santa Peligrosa was founded almost 200 years ago, and she knows that its street plan is fairly tight and its development fairly dense -- so lots of shopping, dining, and entertainment opportunities are within biking distance of people's homes in the original city limits. She sees bike parking racks crowded with bikes when she visits Santa Peligrosa's historic downtown area.

When she visited some friends in Normalia, she noticed that it was a recently created bedroom suburb. It was quite distant from the nearest shopping center, which was a strip mall on a six-lane urban boulevard. She didn't see any bike racks there. On her journey to visit some friends at their Normalia home, she noticed that some of Normalia's streets had no sidewalks at all, and that there were no parks in the development. She saw children riding trikes and Huffy-bikes in suburban cul-de-sacs, but she saw no other cyclists actually going anywhere on Normalia's streets. She herself felt a bit crowded by cars on Normalia's major connector roads, which had been "widened" in some cases from two broad lanes to three tight ones, the final curb lane being too narrow for her taste. On most other streets there was street parking which also crowded the curb lane. Drivers in Normalia didn't seem to be keeping a lookout for bikes. Twice during her visit she had to honk her bike horn at a driver who looked right through her and pulled out of a driveway into her path.

All in all, she didn't find riding in Normalia very pleasant. She did see some Normalia residents driving out of the suburb in their SUVs and minivans, with bikes on the roof or on rear racks; they were heading for parks 20 or 30 miles away.

Our cyclist thought about all this, then sat down and wrote a letter to the local newspaper. She said that she was upset that they would run such a scare story, which might discourage people in town from cycling. She explained that Normalia, which she had personally visited, was a car-centered, sprawling suburb with very few amenities within easy biking distance, and that its streets showed no signs of any attempt to accommodate cyclists or pedestrians. She pointed out that Santa Peligrosa had a student population which during the school year numbered almost 1/7th of the total population; and that just from personal observation she guesstimated there must be at least 100 times as many cyclists on the road, on any given day, than there were in Normalia.

She therefore suggested that her hometown might actually be ten times safer than Normalia, not ten times more dangerous as had been reported. Since this is a fictional tale, we'll say that the newspaper published a prominent retraction and that Santa Peligrosa's citizens were relieved and glad to realize that their town was not, after all, a deathtrap for cyclists. Happy ending.

Our cyclist's empirical experience and rough guesstimates were pretty good as far as they went; but there is much more she could have asked (or thought about). She caught on to the most fundamental principle: that risk must be assessed in terms of exposure. Exposure cannot always be assessed by gross population -- what if not all of the population is equally exposed? If we had been talking about cancer rates in a community containing a facility that emits dioxins vs a community without such a facility, then calculating cancer diagnoses per 100,000 population might be a very reasonable metric: most people are exposed more or less equally to contaminants in air and groundwater. But if we are talking about the risk of a voluntary activity, such as cycling or skydiving, we need to compare apples with apples.

A real understanding of risk (and the statistics by which risk is assessed and expressed) requires a longer discussion, and more math, than most popular media sources are prepared to publish. This is why we read so many easy (and misleading) soundbites about risk in our media. In the next article in this series, I'll talk about how risk and exposure are calculated, and how cycling rates on various risk scales.

Road Safety, Part III: Measuring Risk

Never tell me the odds!

-- Han Solo, "Star Wars"

Once we agree that risk has to be calculated by actual exposure, there is still much room for disagreement over what "exposure" really means. How do we quantify exposure? Obviously exposure is not expressed in simple units like inches: there is no obvious standard measure for us all to use.

One metric much beloved of highway "safety" experts is Risk Per Mile. You will encounter this metric in almost every public statement about road safety. It "proves" that driving in a car is much safer than riding a bike, and that air travel is extremely safe compared to almost any other mode of transport -- based on the number of fatalities per mile travelled.

It should be pretty obvious that Risk Per Mile is a metric strongly biased in favour of the fastest mode of transport. To demonstrate this, we only need to push the metric a bit to make it cover an extreme case. Suppose I have recently opened the first commercial interstellar transport service. I am offering single-passenger flights, in a plush individual capsule, to Alpha Centauri: that's 4 light years, or about 25 trillion miles, away. You are a potential passenger. So far, my prestigious new company (Big Dog Luxury Space Tours) has made two round trips. The first one went fine; but the second time, the passenger died on the return trip.

In my sales office, during our personal pre-trip consultation, I assure you sincerely that our safety record is marvelous: we've had only one fatality in 100 trillion miles! That's far, far safer than car or airline travel. Nevertheless, if you are a savvy passenger and have investigated the history of my company, you may not be too eager to purchase a ticket and board the spacecraft. To you, the odds may look more like one in two (50 percent). Quite logically, you are assessing Risk Per Trip, not Risk Per Mile.

We might also add that "Risk Per Passenger Mile" (a relative of Risk Per Mile) also presents some problems as a metric. Suppose I have a bus line which is very popular. It has only one accident per thousand trips, and each trip is an average of 30 miles. My buses carry 40 passengers, so I can claim to have only one accident per 1000x30x40 passenger miles, or 1/1,200,000.

Now, if my ticket sales decline for some reason -- for example, because more people are now driving -- and my buses are running half empty, my risk per passenger mile appears to have doubled! It is now 1/600,000. However, we are having no more accidents than we had before, and for each invdividual person contemplating a trip on the bus, nothing has changed. The risk per trip has remained the same. A highway safety official (or automobile manufacturer) might be able to "prove," however, that bus travel is now twice as dangerous as it used to be.

These are only two of several strong arguments in favour of using RPT instead of RPM for road safety assessment. Cyclist and pedestrians travel far shorter distances than drivers. But each person regardless of their transportation mode has to make a certain number of "trips" per week: typical trips are a commute to work, grocery shopping, children going to school, and so forth.

American drivers have been spending more and more time in cars, driving more and more miles (2.5 trillion miles in 2000), and driving faster (many speed limits have been raised, and others are routinely disregarded). One "trip" is getting longer and longer for many drivers -- it's not uncommon for people to live 20, 40, even 100 miles from their workplace. Children are living further from our larger, more centralized schools. Local markets and corner stores are going out of business as their prices are undercut by "box stores" located further from people's homes.

The habitual cyclist or pedestrian, by contrast, tends to choose a home on the basis of proximity to work, shops, and so forth. This person may make almost as many trips per year -- errands, excursions, commuting, what have you -- as the driver, but each trip will be much shorter.

To make this simple, let's pretend for the moment that the risk of being injured in a road accident, per mile, is one in 100,000 (one such injurious accident in 100,000 miles). The motorist driving over 100 miles each day might expect it in 3 years or so, the motorist doing 60 miles a day in 6 years, and so forth.

If the pedestrian travels 1/25 as far in a year as the motorist, and the cyclist travels 1/10 as far, then to experience the same lifetime risk as the motorist, the pedestrian's travels would have to be 25 times more dangerous per mile, and the cyclist's would have to be 10 times more dangerous per mile. So if someone claims that being a pedestrian is "30 times as dangerous as riding in a car, per mile", that may not be as impressive as it sounds -- 30 times more dangerous per mile may very well mean "not a whole lot different, over a lifetime." Few pedestrians will ever walk as far in their lifetimes as an habitual driver will drive. For one thing, many "pedestrians" actually cover many miles using public transit, which in most cities is far safer than the private automobile.

Another way of looking at risk is in terms of exposure in hours, rather than miles or trips. This may actually be the best balanced metric of all. There's a curious phenomenon that many transit analysts have noticed: most people spend about the same amount of time commuting to work regardless of the mode of transport. That amount of time is somewhere between 30 minutes and just over an hour. There are people on both outer edges of the bell curve: the artisan who lives right over her shopfront, or the auto-addict who wants to live 100 miles out in the country but work in the city. But the Average Joe and Jane adjust their lives so they don't have to spend much more than one hour getting to, or from, work.

This means that if we restricted our risk analysis to commute travel, we could actually compare pedestrians, cyclists, and drivers on a common scale of risk per hour. We would know a priori that they all have a more or less constant time exposure. Unfortunately, when statistics are gathered on collisions and injuries, the purpose of the person's trip is not recorded. So much for that idea.

However, we also know that in dense urban and town centers, the actual average speed of an automobile is reduced so that it does not much exceed, and sometimes falls short of, the cyclist's average speed. We could then infer that cyclists and drivers take about the same amount of time to traverse the limited area of an urban core. If we also knew how many cyclists there are, and how many drivers, we would have a fairly normalized set of data for making a risk comparison. But alas, we almost never know how many cyclists there are.

Another way to look at risk is "lifetime" risk, that is, extrapolating from risk/hour statistics (however good those are), we ask ourselves what would be the risk of, say, playing basketball for the rest of your life, and how does that compare to the ordinary mortality rate of day-to-day living? We could attempt this by gathering statistics on injuries to basketball players, by hours of exposure on the court; then we could figure out how many people are injured every year (from hospital emergency room records) and what the average lifetime is at the moment, and we can figure out what the "ouch" rate is for life in general vs that for just playing basketball.

Another way to look at risk is by comparing raw numbers. This is often done for polemical purposes. We know that Americans drive about 2.5 trillion miles a year, and we know that about 30,000 adults die in automobiles every year as a result of crashes. We also know that American casualties during the decade or so of our involvement in Viet Nam totalled about 50,000. We could then say quite truthfully that every two years more Americans die in cars on our highways than died in active combat in Viet Nam during our whole involvement there.

The point of this would not be to prove that our highways are as dangerous as fighting an unpopular war in a dense jungle! It proves nothing of the kind, since about 60 percent of our population drive cars and nowhere near that number were recruited and sent into combat in Indochina. Such a claim would be utter nonsense. But the gross numbers might serve to illuminate a difference of attitude (in the media, and among the public), a difference in the amount of publicity or the degree of controversy caused by a comparable number of deaths.

Cost/benefit analysis often focuses on gross numbers to calculate the "social cost" of some particular risk. If we know that 400,000 people per annum visit emergency rooms because of an injury involving living-room furniture (I'm not making this up), and we know the average cost of an emergency room visit, we can calculate the "social cost" of being careless with your Barcalounger. See the URLs for more on CBA. Often, official agencies will describe a risk as "major" in terms of the cost it imposes on "society" for the medical treatment of injured persons, the hours of productive work lost because of such injuries, etc.

One thing that makes risk analysis very difficult for most of us is that the numbers range bewilderingly from very small (risk factors) to very big (population). The population of the United States is now around 286 million. About 2.3 million people die each year, on average. About 6.5 percent of those, or about 150,000 people, die "before their time," that is, from murder, road crashes, poisoning, burns, drowning, falls, fire, machinery, and so forth.

For a sense of scale, we could compare this number with the number of Americans who die from medical misdiagnosis or misprescription (about 120,000 per year), or from nicotine and tobacco-related illness (about 400,000). Another 20,000 die of skin cancer, 30,000 commit suicide, and so on. It's hard for most people to "get a grip" on such large numbers. We can imagine one or ten or even a hundred people dying, but a hundred thousand deaths is hard to grasp.

Unfortunately, the number of cyclists who die on our roads each year is about 800-900. This number is easily grasped by any person and "feels large", even though it is a tiny number in the whole picture of national mortality and a small number in the context of annual road fatalities. Many an essay on "cycling safety" commences with "Almost a thousand cyclists die on America's roads every year!" If you don't know the rest of our mortality stats, you don't realize that this represents just two-thirds of one percent of all untimely death.

So, in this bewildering jungle of data how can we assess the risks of cycling, how can we get a grip -- for our own peace of mind, and to acquire a better foothold in debates with others? Can we defend cycling as relatively safe, or is it relatively dangerous? In the next article in this series, I'll consider cycling from the various "risk perspectives" I've outlined above.

Road Safety Part IV: How Does Cycling Compare?

"In signing this agreement I acknowledge that bicycles are inherently dangerous..."

-- terms of bicycle loan agreement, California, 2000

In the last article I discussed various ways in which risk is measured. In this installment I'll discuss various ways of assessing cycling risks according to these metrics, in an attempt to compare cycling to other common activities.

We know, approximately, how many people die while riding bicycles each year in the US. That number has hovered around 900 for the last several years. About 500-600 of those are adults, in any given year, and 300 or so are juveniles.

In the last ten years, about 5000-6000 adult pedestrians have died each year on our public roads; and about 30,000 to 32,000 adults have died while inside motor vehicles. Juvenile pedestrian deaths have declined steadily to a recent low of around 650 per year, but juvenile motor vehicle occupant deaths are holding steady around 2200 per year.

These are all very small numbers in terms of total national mortality. However, they are not an insignificant component of total "untimely" mortality. If we add up all the "road victims" (peds, bikes, mvo) they total around 40,000 deaths. We know that about 150,000 people die violent or untimely deaths; so road deaths are about a quarter of all this type of mortality.

However, cycling deaths in particular are only 1/40th of that quarter, or 1/160th, of all untimely death. That's .625 percent, or a bit more than half of one percent.

We could compare total cycling deaths to other road deaths (gross numbers). We know that about ten times as many adult pedestrians as cyclists, and about 30 times as many adult occupants of motor vehicles as cyclists, die each year on American roads and highways. For juveniles, we know that about twice as many pedestrians as cyclists die on the roads, and about seven times as many motor vehicle occupants. Cycling is not a leading contributor to the gross figures, so it does not represent a major social cost as compared to other road fatalities.

We also know, from national road fatality databases like FARS, that in almost all cyclist fatalities, the cyclist was hit by a motor vehicle; so we know that most "cycling" deaths, like most pedestrian deaths and all motor vehicle occupant deaths, are motor-vehicle-related. The World Health Organization once estimated that if the risk imposed by motorized traffic were eliminated (that is, if bikes had the roads to themselves), cycling would be 500 times safer than driving.

Already, starting from these raw numbers, we have some grounds for wondering why we so often hear that bicycles are dangerous. The common risk factor for road fatality seems to be the motor vehicle. We might more accurately say that motor vehicles are dangerous, whether you are inside or outside one; but that fewer cyclist lives are claimed by motor vehicle collision than any other type of traveller.

This might lead us to ask whether riding in a car is really safer than riding on a bike (most people are absolutely certain that it is). We might measure this risk variously, depending on how we define exposure. We might ask what our chance is of being killed or injured per hour of riding in a car -- per mile -- per trip -- or lifetime risk. We could then compare those risks to the risks of cycling.

I don't have enough space here to reproduce all the math, but Ken Kifer has done a lot of this research and calculation (see URLs below). His conclusions:

"Assuming that people are at least passengers from the cradle to the grave (75 years), the driver/passenger has a 1/60th chance of dying in an automobile... Let's say the average cyclist rides 250 hours per year, say 3,000 miles. And we'll say that this person rides 60 out of the normal 75 years of life, or 15,000 hours and 180,000 miles total. Using the Failure Associates figures, this person is going to have to have a 1/256 chance of getting killed while cycling during his lifetime."

If Ken has done his math right, then the lifetime risk of regular cycling is about a quarter of the lifetime risk of regular riding in cars.

If people cycled as many miles in a lifetime as they drive, the lifetime risk of cycling might be higher; but as we said above, cyclists and pedestrians are not hypermobile as drivers tend to be; their choice of transportation voluntarily limits their exposure.

We could also compare the risk of other activities of which we have some intuitive "risk perception". What's the risk of dying per million hours of exposure to a wide range of activities? According to Failure Associates, a professional risk assessment firm, the numbers look like this:

	Skydiving 	128.71
	General Flying 	 15.58
	Motorcycling 	  8.80
	Scuba Diving 	  1.98
	Living 		  1.53
	Swimming	  1.07
	Snowmobiling 	   .88
	Motoring 	   .47
	Water skiing 	   .28
	Bicycling 	   .26
	Airline Flying 	   .15
	Hunting 	   .08

It seems a little odd that cycling should be safer than living! But all this means is that if you were magically immune to every other kind of risk, and you did nothing but ride a bicycle 24 hours a day, you would live far longer than the average person. In reality, some other higher risk (or inevitable old age) will eventually catch up with you. It's pretty intuitive that skydiving should be a hazardous activity, but most people would be surprised to find that cycling rates as safer than swimming! And most people would hotly deny that cycling is safer than driving (motoring).

But perhaps the risk of death is not what people mean when they say cycling is dangerous. Often, safety experts stress the frequency of injury while cycling (as opposed to driving) as proof that cycling is very dangerous. Obviously, we are more exposed to minor bumps, bruises, strains and sprains while doing something active like cycling (or running, swimming, sports, etc) than while sitting strapped into a padded seat in a metal box. But how does cycling compare to other active pursuits?

An Australian study compared the number of injuries per million hours of juvenile participation in active sports. An "injury" for the purposes of this study was any injury that resulted in a clinic or hospital visit. The I/MH figures, ranked in descending order, were: Football 1,900; Squash 1,300; Basketball 1,100; Soccer 600; Bicycling 50. Cycling, then (even for juvenile cyclists who lack many of the skills adult cyclists have acquired) is about 38 times safer than football, 26 times safer than squash, 22 times safer than basketball, and 3 times safer than soccer.

If we take the aggregate injuries from all other sports versus cycling, the other sports sum to 4900 injuries per million hours, as opposed to the 50 for cycling: a ratio of almost 100 to 1. We might expect, for any exposure period, that on average 100 participants in all these other activities, as a group, would be injured for every one cyclist who was injured.

For another assessment of cycling and injury we might consider a demographic considered "high risk" in all actuarial tables: young males, ages 15-24. No matter what the activity -- driving, cycling, falling -- this demographic generally manages to rack up more injuries than everyone else. This would give a "worst case" risk assessment: a risky demographic participating in an activity which is said to be inherently risky.

To give the bicycle an even tougher time, we can also pick a specific risk which people particularly associate with cycling: head injury. According to conventional beliefs we would expect these "risky" young males to be particularly vulnerable to this danger, and therefore to be more at risk from cycling than from other activities. About 75,000 young Americans die from head trauma every year, so this is not a very small problem. How large a part does cycling play in head injuries?

If we look at our worst-case risk group, 50 percent of their head injuries are incurred while in a motor vehicle during a crash. 21 percent are caused by falls. 12 percent result from assault (interpersonal violence). These three categories account for 89 percent of all head injuries to this high-risk demographic.

Only 10 percent of such injuries are a result of participating in active sports; cycling is lumped under "sports", and contributes only a small fraction of fatalities in this category. This is consistent with the risk per million hours for various sporting activities, as cited above. Cycling does not stand out as a particularly risky activity -- even when it is done by particularly risky people.

We could then ask whether the risk of head injury, which we believe is particularly acute for cyclists, means that cycling contributes a large share to national injury statistics (and associated costs) for all demographic groups for this type of injury. We would expect surfing, for example, to contribute more to drowning or hypothermia statistics than to sprained ankle statistics, because drowning and hypothermia are risks particularly associated with surfing. We find that in the US, head injuries requiring medical treatment constitute only 1.5 percent of all bicycle-related injuries. And those bicycle-related head injuries in turn constitute only 1 to 1.5 percent of total US head injuries -- and from .5 to .9 percent of head injury fatalities.

Since some fairly good estimates suggest that about 6.6 percent of the US population rides a bike regularly, an "even handed" distribution of injury incidence (i.e. one which assumes that cycling is at least as risky as all other activities) would mean that 6 percent of all head injuries should result from cycling. The low figure of 1 to 1.5 percent contradicts this expectation and suggests that cycling is actually a lower contributor to head injuries than most other activities. By this time, you may not be too surprised to hear that the leading cause of head injury and related fatality, for all age ranges and both genders, is . . . being inside a motor vehicle during a collision.

We might also look at the detailed information about bicycle fatalities which is available from the FARS database. If cycling is inherently risky, in and of itself, we would expect the deaths of cyclists to be randomly distributed. Cyclists should die because bicycles are dangerous -- no matter where they ride or how they ride. In fact, there are some patterns in the data. One suggestive pattern is found in adult fatalities: almost half of them happen at night, and in most of these it is noted that the bicycle did not carry active lighting (i.e., it had only passive reflectors, or no lighting at all).

This identifiable pattern in the cycling fatalities might suggest to us that though cycling itself is not particularly dangerous, cycling after dark -- on roads with fast-moving cars -- without carrying lights, is quite dangerous. This should come as no surprise: it would also be very dangerous to drive a car on similar roads after dark without lights. We might wonder why we enforce night lighting for cars, but not for cyclists, when perhaps as many as half our adult cycling fatalities might have been prevented by adequate lighting. When people operate cars dangerously, i.e. without lights at night, and they come to grief as a result, we describe this as unsafe operation; but when people operate bicycles unsafely and come to grief (albeit in much smaller numbers) we conclude that bicycles are dangerous.

I hope that these interesting statistics will go some small way towards challenging the by now dogmatic assumption that Cycling is Dangerous. Cycling could be made safer -- and so for that matter could walking and breathing -- if we practised some moderation in the use of automobiles. But even in the presence of automobiles, cycling still does not contribute an injury or fatality count proportionate with the percentage of people who cycle.

We are left to wonder why there is such a strong belief, among both officials and the general public, that "bicycles are dangerous".

Road Safety, Part V: The Cultural Construction of Risk

or Why are Bicycles being Dangerised?