Bicycle Use and Hazard Patterns in the U. S.,
and Options for Injury Reduction

Consumer Product Safety Commission of the U. S.
16 November 1993

home                     safety

1 General

The original version of these comments was dated Tuesday, 4 January, 1994

In the following comments, I distinguish between cyclists and bicyclists. Bicyclists are the average bicycle users who do what the CPSC considers to be recreational riding, while cyclists are the enthusiasts and the transportational cyclists.

1.1 Use Patterns

The use patterns probably reasonably reflect the pattern of bicycle use by the U.S. population, but it has no relevance to the pattern of cycling by those who consider themselves cyclists. The CPSC considers the only significant use to be recreational. We consider that transportation is a significant use for cyclists, and also that even our recreation is carried on in a way that is far more comparable to transportation than what is commonly considered recreational bicycling. While the CPSC, considering its mandate to reduce total casualties, is undoubtedly correct in considering plans that are geared for the average bicyclist, two things must not be forgotten. First, cyclists constitute a different population from bicyclists, and what may work for bicyclists definitely won't work for cyclists. Second, when significant social effect is considered, that effect, either today or tomorrow, will come from cyclists rather than bicyclists, and if growth in socially significant cycling occurs, it will involve many bicyclists becoming cyclists through either training or experience.

The U.S. bicyclists (definition: ride at least once per year) are heavily concentrated in the 10 years from ages 5 to 14 (40% of total).

The CPSC considers that hours is a better measure of cycling than is miles. Well, that may be OK for recreation, but it is incorrect for transportation. Furthermore, it is absolutely wrong for considering any learning effect. People who spend many hours at 4 mph on bike paths won't learn anything about cycling. Using hours instead of miles also seriously understates the effect of the better and faster cyclists on the cycling population and on society at large. Using hours instead of miles effectively dismisses the societal effect of bicycle transportation.

The use patterns show a significant contribution from the many-hours cyclists, as is shown by the fact that the median annual riding time, 105 hours, is only 44% of the mean time. Furthermore, the low values in annual hours shown for cyclists over 50 years does not tally with our information. Club cyclists over retirement age have high annual mileages, which naturally require many hours on the road. They do so because they then have the time.

The CPSC tried to discover how much cycling was done on each type of facility, but didn't ask the appropriate questions to discover that information. All it got was answers to Always, More than Half Time, Less than Half Time, Never. This defect has seriously crippled the later analysis.

2 Comparing Bicycle Usage (CPSC) VS Motor Vehicle Usage (USDOT)

It is most illuminating to compare the bicycle usage statistics given by the CPSC with reality and against those for motor-vehicle usage given by the US Department of Transportation

2.1 Bicycle Usage

The CPSC report gives the following statistics:

Cyclists = 67 million

Average hours/year/cyclist = 236

Adult cyclists = 33.5 million

Average hours/year/adult cyclist = 176. (This is calculated weighting the numbers in each age group of those over 20 years of age.)

Therefore:

Cyclist hours per year = 1.58 x 10^10

Adult cyclist hours/year = 0.59 x 10^10

Child cyclist hours/year = 0.79x 10^10

Adult cyclist miles/year (assuming 10 mph) = 5.9 x 10^10.

Child cyclist miles/year (assuming 5 mph)= 4.0 x 10^10.

The CPSC refers to the Rodale Press study of adult cyclists. This gives 34.8 miles per month as the average for cyclists over the age of 17 during warm-weather months. If 8 months are used as the year this is 278 miles per year. If 12 months are used, this is 418 miles per year.

2.2 Motor-Vehicle Usage

The USDOT gives the following statistics:

Automobiles = 144.2 million

2-axle, 4-tire trucks = 39.5 million

Auto miles/year = 11,000

2-axle truck miles/year= 12,000

Therefore, average is 11,260 miles/year.

At 30 mph this is 375 hours/year.

Therefore:

Motor-vehicle hours/year = 6.89 x 10^10

Motor-vehicle miles/year = 207 x 10^10

2.3 Comparing CPSC and Rodale Data

Using the CPSC hours per year and the Rodale miles per year, we get the average speed for adult cyclists. This is 1.6 mph for 8 month year, 2.4 mph for 12 month year. Quite clearly, there is something seriously wrong in either or both of the CPSC's and Rodale's data collection to produce such an absurd result.

2.4 Comparing Bicycle and Motor-Vehicle Usages

Comparing the CPSC bicycle data and the USDOT motor-vehicle data gives us the following:

On an hours basis, bicycles ridden by children are 11.3% of the vehicles and bicycles ridden by adults are 6.8% of the vehicles. In other words, as you look at the normal typical roadway and sidewalk, at any one time you should expect to see that child cyclists are more than 1 out of 10 motor vehicles, and that adult cyclists are 1 out of 15 motor vehicles.

If, on the other hand, you choose to count miles traveled, then bicycles ridden by children produce 2.2% of the vehicle miles, and bicycles ridden by adults produce 2.7% of the vehicle miles.

Since bicycles are not used on interstates to any great extent, the above figures for bicycle proportions need to be amplified by dividing by the proportions of motorist time and miles that are done on interstates.

Just on the basis of casual observation, the postulated visual proportions of bicycle traffic to motor traffic are absurdly high. The postulated proportion of adult cyclists for the nation, either visibly or on a mileage basis, will be seen only in the very few cities with universities at which many students cycle.

I think that both the CPSC's figures for time spent are too high, and the figures for average miles/month (from Rodale) are too low, as shown by the absurd average speed calculations. The total probable error is approximately 5 times, which is sufficient to make the data useless. I have no estimate as to which factor is most erroneous.

3 Accident Patterns

The accident rate per hour is high for children and for the elderly (whose typical accident appears to be being hit by a car while crossing the street). Because children have both large numbers, high individual hours per year, and high accident rate per hour, they occupy a disproportionate high place in the accident numbers. Reducing the childhood accident rate would be a socially significant action, but it must be remembered that the measures that would work for child bicyclists won't work for adult bicyclists and may well be harmful for cyclists of any age.

The CPSC study confirms previous statements about the low significance of car-bike collisions in the accident mix. Car-bike collisions constitute only 10% of the accidents in the study. Studies by Kaplan and by the National Safety Council showed carbike collisions to be 17% of accidents for adults, and those by several other agencies showed 10% for children. Since children constitute so large a proportion of the U.S. cycling population, the CPSC's figure of 10% appears to be reasonable.

Under causal factors the CPSC lists the following:

Table 1:

Causal Factor

Percent

Uneven surface

27

Excessive speed

22

Collision with moving object

15

Mechanical

13

Collision with stationary object

13

Object caught in spokes

6

Total

110

That's 110% right there. The mechanical reported frequency of 13% doesn't square with the figure of 41 out of 463 (9%) elsewhere reported.

Even though an accident may well have multiple causes, these are probably all the big ones. The countermeasures available for accidents caused by these causes aren't big governmental facility programs; they are programs of individual development of skills and responsibility.

4 Hazard Ratios

The CPSC combined its accident data with its crudely divided data on use of various facility types to produce what it calls hazard ratios. These are the ratios of the accident rates per hour on the different types of facility. I quote some of these:

Children on streets/children on paths=8.02/1

Children on sidewalks/children on streets=1.65/1

Adults on streets/adults on paths=6.93/1

Adults on unpaved surfaces/adults on streets=8.84/1

Adults on highways/adults on streets=2.45/1

These values are the most politically significant of any in the study. Most other data in the study are expected and have been known for years. With publication of this study, undoubtedly many people will use these hazard ratios for political purposes. The question is whether they have any accuracy at all.

Consider that 10% of the accidents are car-bike collisions. Consider that the principal attributed difference between streets and paths is that on streets one has the risk of car-bike collisions while on paths one does not. To get all the 10% of car-bike collisions on the streets when the accident rate on the streets is 8 times higher than on the paths, requires that something like 90% of the cycling is done on paths. However, the demographic data show that 65% of the cycling population never ride on paths and only 17% usually do. The only way that the data could be internally consistent is to assume that streets are much more dangerous than paths in all the other ways as well as in car-bike collisions. That is, the streets have more potholes, more slippery places, are more dangerous at higher speeds, and attract more objects to get caught in the spokes than do paths. That picture is ludicrous. Therefore, the risk ratios are thoroughly inaccurate. The cause of the errors is probably the crude classification of time spent on each type of facility together with a statistical program that listed each person as using only the one type of facility that he used most frequently. (That is, setting a switch ON for that type and OFF for all other types.)

These supposed hazard ratios do not agree with other information. For example, they say that riding on sidewalks is safer than riding on streets. That disagrees with the three studies that we know: the Palo Alto staff report that showed a 54% increase in accidents when cycling on sidewalks was imposed, my experiment on those same streets in which I faced imminent carbike collisions that I was able to avoid only by using extreme skill (and luck) at the rate of two per mile, and the Wachtel and Lewiston study of those same streets that concluded that sidewalk cyclists had a higher car-bike collision rate than roadway cyclists. I gave up further testing when the eighth such confrontation nearly killed me. The truth is that sidewalks are extremely dangerous places for cycling. The AASHTO standards recognize this by recommending against sidewalk bike paths, the only type of bikeway that they disapprove of.

The same analysis applies to bike paths. In general, bike paths are places where the dangers that cause the majority of accidents to cyclists (see the causal factors cited by the CPSC, as reported above) are more prevalent than on streets and highways. The only danger that is more prevalent on streets than on paths is said to be motor traffic (I say said to be because many paths present great hazards of motor traffic, as evidenced by my sidewalk-cycling experiment). If the other dangers of streets are only as prevalent as they are on paths (to use the most conservative argument), and since car-bike collisions constitute only 10% of accidents to cyclists, streets can be only 10% more dangerous than paths, instead of the 8 times greater factor produced by the CPSC.

The same analysis applies to riding on unpaved surfaces versus riding on streets. The the CPSC says that streets are 3.44 times as dangerous as unpaved surfaces for children, and 8.84 times as dangerous for adults. This means, if the ratios are accurate, that unpaved surfaces are smoother, safer at higher speeds, and collect fewer foreign objects than are streets, a completely absurd conception.

One possible reason for the CPSC's inaccurate conclusions is that it doesn't consider that the populations of people who ride on different types of facilities are different and that the manner of their riding when on different types of facilities also may differ. Riding a bicycle on sidewalks may be reasonably safe for the person who rides in a pedestrian manner, or at least not much more dangerous than walking on sidewalks. However, cycling on sidewalks at normal road speeds is extremely dangerous. Riding a bicycle on bike paths at walking speeds is probably reasonably safe, but cycling on bike paths at normal road speeds is, again, very dangerous. Presumably, those who ride on sidewalks are a different population from those who normally ride on highways.

5 Mechanical Hazards

Because the CPSC is the Consumer Product Safety Commission, it concentrates on casualties caused by products. Only 9% of the accidents discovered were caused at all by product defects, and most of those were caused by bad maintenance and careless use. The accident investigations were pretty poor. For example, in one the investigator concluded that had the cyclist not removed the chain guard, the chain, when it broke or otherwise came off, would not have got caught in the wheel. Considering the typical hockey-stick chain guard, that conclusion is absurd. The CPSC concludes that the accident pattern does not warrant any change in its standard. This is a false conclusion, as is shown under Nighttime Equipment.

6 Nighttime Equipment

In addition to saying that the accident pattern shows no reason to change the requirements in the CPSC's regulation for bicycle design, the CPSC says that its study was not sufficient to come to any conclusions about nighttime cycling equipment. While these statements are, strictly speaking, accurate, they consider only the data of this study in a very restricted manner, not the whole number of facts that describe the situation. The general implications of this statement are discussed below; this section concentrates on the nighttime equipment problem.

The CPSC missed the significance of several points. The first was that its own data showed that only 1 /3 of those who ride at night used either a headlamp or a taillamp. The second was that its own data showed that the accident rate at conditions other than full daylight was considerably higher than during full daylight. The third was that its own data showed that, when riding at night, 50% more people added a taillamp to the existing rear reflector than used a headlamp. For even those who used a lamp of either type, this is exactly the wrong countermeasure. 80% of the car-bike collisions probably caused by darkness occur from the front, where the headlamp is the required safety equipment. In other words, when confronted with a bicycle that has a reflector at both ends, more people choose to supplement the rear reflector with a tail lamp than choose to supplement the front reflector with a headlamp. The facts are that a rear reflector has a very reasonable chance of preventing car-bike collisions from the rear because the car's headlamps shine upon it, while the front reflector has no chance at all of preventing most collisions from the front because the car's headlamps never shine upon it until, if at all, at practically the moment of collision.

The CPSC is now arguing exactly the opposite of its position in the case of Forester vs CPSC. In that case, the CPSC argued that its experts were so smart that they could predict that accidents of particular types were going to happen (even though nobody had yet documented any in the century of bicycle use that preceded their standard), and that therefore they had the duty to protect the public from their imagined accidents by issuing a regulation. The CPSC told the court that it would be absurd to have to produce a "body count' 'before a regulation could be issued. Now, in this study, they argue that because accidents from this known cause are so few, the body count is still too low to require a change in the regulation.

The data of this study show exactly what I have always argued: the all-reflector system required by the CPSC is dangerously deceptive. It misleads people into not using the proper equipment and it misleads even those people who choose to do something into adopting the equipment that provides the least benefit instead of adopting that which produces the most benefit. That is plenty of information to conclude that the CPSC's all-reflector system should be abolished and replaced with one that requires, when cycling during darkness, a headlamp and a bright rear reflector. The CPSC could produce a standard for the amount and distribution of light to be provided by a headlamp and for its mounting system. There is no need for the CPSC to produce a new standard for the rear reflector because reflectors made to the existing SAE standard for use on motor vehicles and highway markers are much brighter than the deliberately dim reflectors that the CPSC regulation now requires.

7 Training Considerations

Confronted with data showing that a very large proportion of the accidents to cyclists are those for which the appropriate countermeasure is a training program, the CPSC chose not to entertain that suggestion. The prime reason for declining, cited by the supposed expert who considered the question, is that children are insufficiently psychologically and neurologically mature to learn and practice safe cycling technique. However, even that expert suggests that children in the third grade can start learning and that children in the sixth grade can learn quite well. He quotes the old but inaccurate saw that it is not children's ignorance of the law that causes them to misbehave, it is their immaturity.

Whether the CPSC should be in the training business is a different question entirely (I think that a training program designed to CPSC requirements might well lead to disaster), but the accuracy of the CPSC's assessment is relevant. The report pays no consideration to what children need to be taught and how to teach it to them. Children are very conscious of rules and, in games, obey them scrupulously and enforce sanctions against those who disobey. They understand that there is a particular way to play each game. But they don't learn the rules first and play the game later. They learn the rules by playing the game. Almost the only consideration of training in this study is based on the idea of learning the rules in the classroom for use when playing the game on the streets. We know that doesn't work. However, my work teaching children from second and third grades shows that they can be shown how to ride properly by doing it themselves, and that in the process they learn why the rules are as they are. After 15 hours of on-the-road training I had groups of children from second and third grades riding better than the average adults in the same communities that are noted as cycling cities.

8 Miscellaneous

The CPSC lists the Bicycle Federation of American as an organization of bicycle users. I will repeat John Allen's crack about that: the BFA is not an organization of people who use bicycles; it is an organization that uses cyclists for its own benefit.

9 Implications for Policy

9.1 Mechanical Requirements

The CPSC argues that the fact that few accidents are caused by mechanical failure shows that its bicycle regulation need not be revised. Insofar as nighttime protective equipment is concerned, this is a false conclusion as has been shown by the discussion above. However, this is only one side of the question. The other side is whether the CPSC regulation has reduced accidents. In other words, would the accident rate rise if the CPSC regulation did not exist? The plain fact is that the CPSC regulation was never addressed to the prevalent causes of accidents to cyclists. The CPSC regulation does not consider smoothing riding surfaces, nor collecting foreign objects from them, or improving obedience to the traffic laws. It does not even consider the prevalent causes of mechanical troubles that cause injuries to cyclists. Its own study shows that the prevalent mechanical troubles are caused by bad maintenance, a matter which the CPSC is powerless to address. But even more significant is the question of whether the CPSC's regulation actually prevents injuries. For example, does the absence of punctures of the skin caused by the unravelled strands at the ends of brake wires demonstrate that the CPSC's requirement for capping the ends of brake wires is effective? It does not, because there weren't many such injuries ever, and, so far as I know, they were never reported and were considered only a minor nuisance by those who incurred them, principally mechanics working on bicycles. Furthermore, the crimped-on caps commonly supplied to meet the CPSC requirement prevent lubrication of brake wires, thereby contributing to brake failures.

It is noteworthy, though not unexpected, that the CPSC relies on a body count when that is in its favor, after saying for so long that the public safety requires that action be taken before the bodies arrive to be counted.

The CPSC remarks on the accident-preventing or injury-preventing characteristics of its requirements for frames and front forks, and notes that it discovered no injuries from failed front forks or frames. While a true statement, this conclusion is utter hokum. Accidents caused by failure of front forks are very rare and their typical cause, metal fatigue, cannot be either detected or prevented by the CPSC's requirements for forks. In actual fact the CPSC's requirements for front forks and for frames are based on the false notion that the ability to absorb energy under deformation will prevent injury to cyclists in frontal collisions. Everybody with any engineering sense knows that this is physically impossible, contrary to the laws of physics. Yet the requirement persists.

The same analysis applies to the absence of accidents or injuries from wheel failures caused by so many spokes pulling through the material of the rim that the wheel collapses. There were no such accidents in the CPSC's data. The point is that there never were any such accidents reported before the CPSC's regulation was issued, and the physical mechanism that the CPSC postulated would cause such accidents has been shown by later research to not exist. In fact, the stresses in the wheel are the exact opposite of what the CPSC presumed would cause such an accident.

The policy question is not whether any new requirements should be added to the regulation but whether many requirements that now exist should be eliminated as useless, and whether the requirement for the all-reflector system based on wide-angle reflectors should be repealed as a public danger.

9.2 Non-Mechanical Policy Considerations

Unfortunately, this CPSC study could be used to affect policy in many other areas of cycling besides the design of bicycles. The assumption that the important aspects of cycling have to do with children and recreation has eliminated many other factors from being considered, to the detriment of the accuracy of the study and producing the appearance of valid grounds for policy in areas other than bicycle design. The conclusions drawn by the CPSC in areas other than bicycle design show that the CPSC has no expertise in these areas, but this reasonable conclusion will not prevent others from quoting the CPSC to suit their own interests. I discuss three areas of concern: nighttime equipment, training, and cycling facilities.

9.3 All-Reflector Nighttime System

The details of the nighttime equipment problem are discussed above. The CPSC considered that its system of wide-angle reflectors provides adequate safety when riding at night, and it still says, in this document, that bicyclists should be urged to make sure that all their reflectors are in good working order. The problem is not whether the reflectors produce the optical effects that are required. On all the evidence, those in current production adequately exceed the CPSC's requirements. The problem is whether the performance characteristics required by the CPSC have any relevance to the prevention of accidents at night. (Nobody is arguing that they have any positive effect in daylight.) That is a traffic engineering question that should have been studied and answered by experts in cycling and cycling transportation engineering, not by experts in testing products. The CPSC has known the true answers to that question, provided by people who know the subject, since before it first issued its regulation, and it still hasn't done what it should about requirements for nighttime equipment. The reason is, obviously, that the CPSC doesn't have the appropriate expertise in traffic engineering and in cycling to produce such knowledge.

9.4 Training of Cyclists

The CPSC dismisses the area of training done by itself (which is probably a good decision, considering the probable design of a training program designed according to the CPSC's ignorance of cycling), but it does so with words about mental immaturity and similar matters that show that it has no expertise in the areas of training or cycling but that give an entirely incorrect idea of what the problems are and what training should be done.

9.5 Cycling Facilities

The CPSC's analysis of the relative safety of different cycling facilities disagrees with the best knowledge in the field and is internally inconsistent, but it agrees with public superstitions and the agenda that is prevalent among those who call themselves bicycle activists. From the CPSC's point of view, as long as all the participants are merely enjoying themselves, they ought to do it in the safest possible way, by riding on bike paths and unpaved surfaces. This begs the question of whether riding on bike paths or unpaved surfaces provides the best enjoyment of cycling. For some it may, for others it certainly does not. I, for one, do not consider that riding on bike paths at 10 mph amid the crowd of children, dogs, and incompetent cyclists while trying to dodge the cars at the intersections is anything more than the cause for acute worry. No study that does not consider the different forms of behavior on the different facilities, the actual accident rates on these facilities for the different forms of use, the different forms of cycling which exist and are enjoyed, the different functions that cycling performs for the user and for society (which range from the prevention of heart disease to the provision of transportation), the different locations in which these are used, and the different skills possessed by and used by the riders can produce a true risk analysis of the various types of bicycle facility.

Even supposing that the CPSC's analysis has some measure of accuracy, the CPSC's conclusions are wrong. It may well be true that riding at 5 mph on bike paths is very safe. However, that is no basis for recommending, as the CPSC does, that many more bike paths be provided, because riding at 5 mph to the locations served by bike paths does not provide the required service for those for whom that style of cycling is unsuitable, who are a considerable proportion (maybe the large majority) of those who use bicycles for either fitness or transportation. The plain fact is that making such recommendations requires expertise in traffic engineering, cycling transportation engineering, cycling itself, and city planning, with additional inputs from human factors, psychology and sociology. The CPSC's study and the conclusions that it draws from its data shows that it has no significant, relevant expertise in these subjects.

10 Review of CPSC study by Mac Elliott, 1/17/94

SOME AREAS TO LOOK AT: CPSC U.S. BIKE USE AND HAZARD PATTERNS STUDY DATED NOV 1993

(Following comments are result of cursory review by Mac Elliott. This CPSC Study contains 196 pages of closely packed text and numbers, could have wide impact & influence, and warrants a detailed review by competent people if it is to be issued. See No. 8)

1: Relative risk reported for streets vs paths etc. (7 or 8 to 1, pg 18 et al), differs dramatically from other studies, and will have far reaching effects if accepted. Items 2 - 6 below list some areas to look at.

2. Very small sample size for injuries. Pg 101 Footnote 9 states 79 total injury observations, with only 3 on bike paths and 2 on sidewalks. Small size makes estimated odds sensitive to small changes. (See # 3 & 4 below)

3. Many (most?) reported bikepath and sidewalk collisions occur at street crossings. Probably counted as street and highway collisions, not bikepath or sidewalk. Particularly in light of #2 above, this could make a large difference in risk ratings of streets vs paths, etc.

3a. Sidewalk riders often swing out onto street and back onto sidewalk at driveways to avoid walkers, curbs, and other obstacles, and may be hit on the street, where they may not have been hit if riding out where motorists could see them. Probably counted under street collisions.

4. Table 10 pg 66 indicates 100% of rider's hours were credited to surface (location) on which he or she rode most, 0% to other locations. If so, this certainly could skew reported risks of accidents per hour at each location.

5. Street collisions from child driveway rideouts while playing will count against streets even though no transportation is involved.

6. Other numbers look odd. e.g., 15 billion riding hrs. per yr. (pg 1 et al) at only 5 mph = 75 billion miles per year. Probably 5 to 10 times too high. Also 37.2 injuries per million hours (pg 83 table 1) at 5 mph is 7.4 injuries per million miles, or 135,000 miles per injury. With riding time of 236 hours/year (pg 60 table 8), average distance at 5 mph = 1,200 miles per year. Average rider time between injuries = 135,000/1,200 = 110 years. Way too long. Whatever caused these results may also have skewed other numbers in the report.

7. Were on-street bike lanes treated as bike paths or as streets and highways in report statistics? Not clear. (Pg 18 recommends network of paths and lanes).

8. Report needs a detailed review by someone familiar with cycling habits and statistics, someone who can spot potential errors or confirms statements made.

Mac Elliott, 1/17/94

11 Conclusions

11.1 Statistical Errors

The data regarding the proportions of types of injuries to cyclists may be reasonable. The data regarding the location of accidents and the types of facility associated with them are suspect, in several ways. Car-bike collisions caused because a cyclist was riding on a bike path are probably counted as roadway accidents instead of as bike-path accidents. The absurd conclusions, reached by analyzing the data, showing that bike paths have to be smoother and cleaner than roadways, indicate serious trouble in data collection. The absurd relationships between the CPSC's data and the data of Rodale Press and the USDOT indicate, again, very serious trouble in data collection.

The CPSC itself collected usage data only on a time basis, not on a distance basis. The very great absurdities produced when comparing the CPSC bicycle usage figures against the USDOT's motor-vehicle usage figures indicate, again, that something is enormously wrong. The fact that the data concerning motor-vehicle use are collected on a systematic basis and are reviewed and used for important purposes by at least thousands of people indicates that these data are pretty reliable. Therefore, by far the major portion of the discrepancies between the CPSC's bicycle usage figures and the USDOT's motor-vehicle usage figures must lie with errors in the CPSC's work.

In conclusion about statistics, any conclusion that is derived from the CPSC's hourly usage data must be considered extremely suspect.

11.2 Non-statistical Errors

11.2.1 Lack of Meaningful Study Design

The CPSC's Bicycle Safety Study is so filled with non-statistical technical errors that it is hard to describe them. The first major error is that the study is directed at matters about which the CPSC can do nothing. That is, matters beyond the scope of the CPSC's "bicycle design regulation," the only official product of the CPSC on this subject. That is because failures in bicycle design cause only a minute portion of accidents and injuries to cyclists.

If, on the other hand, the study was intended to discover the magnitude of conditions about which society could take preventive action, then it was very badly designed for that purpose.

Consider the identified causal factor of rough riding surface. That had been recognized as a major factor as long ago as the Kaplan study of 1976, and anecdotally for the preceding century. Had the CPSC study been designed to identify the kinds and locations of surface defects, it could have provided useful information for society to take corrective action. But the CPSC study did nothing of the kind. Its data are so aggregated that they can't be used for any practical purpose.

Consider the previously identified factor of lack of skill among cyclists. The study was not designed to detect any relationship between skill and accident rate, and, in any case, the CPSC had itself dismissed the improvement in skill as unattainable because of the immaturity of children.

11.2.2 Search for Bicycle Design Failures that Might Be Regulatable

The CPSC gave one object of the study as identification of any new patterns of mechanical failure, if there were any, that would suggest the need for changes to its bicycle design regulation. Naturally, it found none. I don't imply that this was a deliberate closing of an eye. Accidents caused by defective bicycle design are so rare, and new types of failure are so unlikely, that this kind of telephone survey is the inappropriate way to try to detect such new types of failures.

I happen to know of one new type of bicycle design failure that causes accidents. This design was rare when the CPSC made its study, but I doubt that, if the CPSC made its study today, such a study could identify this new failure mode. This failure mode is the presence of "lawyers' lips" on the front fork tips that cause quick-release wheels to fall out of front forks. I think these should be banned, but I doubt that the CPSC would do anything about that. Oh yes, I suppose that these lawyers' lips work; what they do is to make it harder to safely adjust the quick-release mechanism while ostensibly protecting the manufacturer from liability. More accidents, but less liability.

I also have in my current accident case list (this is 2001) two cases of accidents caused by wheels built with insufficient spoke tension. (I think that I had another like that a decade ago.) That doesn't mean, to my mind, that there has been a sudden flood of badly-built wheels; just random chance, so I think today. But even if it were not just random chance, I doubt that such accidents would be so frequent that a study like the CPSC's would catch them today.

11.2.3 The Real Bicycle Design Failure Mode

It is correct to criticize the CPSC's bicycle design regulation as largely useless, in that it addresses bicycle designs that rarely cause accidents, and generally doesn't take appropriate action for those that might. It is also correct to criticize the CPSC's bicycle design regulation for causing accidents, specifically those at night through its required all-reflector system of nighttime protective equipment. The CPSC knows this. It was sued for this in 1976, and, don't you know, being sued usually has a pretty strong effect on a governmental regulatory agency. It escaped liability for its error only by lying to the Court about its activities. It has been criticized ever since for the same dangerous requirement. It has held meetings on the subject of nighttime protective equipment. And, despite all this, the greatest failure (or greatest cause of jeopardy, if you want to call it only that) in its bicycle regulation, the CPSC didn't design its Bicycle Safety Study to work out what changes ought to be made. The only conclusion that it reached was that since the accident rate was higher during darkness than during daylight, further studies ought to be made. Is that conclusion new? It might be to you, if you have only just reached the sixth grade in school.

11.2.4 Recommendations Outside Bicycle Design

The CPSC does make recommendations outside of its area of responsibility and supposed expertise. For nighttime accidents, more bike paths and better street lighting. Not, of course, that the CPSC has any knowledge of traffic engineering, let alone bicycle traffic engineering. It just makes a recommendation that is based on superstition, in the erroneous belief that its superstition agrees with the standard knowledge in the field. For lack of skill, the CPSC recommends studies to determine the age at which children can learn to ride properly. Pretty good, that, except that it had been done ten years before. At least, the CPSC didn't try to say what the children should learn.

11.2.5 Summary

In short, the CPSC discovered nothing that was not already known, but claimed to have discovered a lot that was not there to be discovered.

The CPSC's work in the bicycle regulation shows that it has very limited expertise in those hazards that are directly related to defective products. It has far less expertise in the uses to which products may be put, and practically none about the physical and social milieu in which products are actually used. Particularly when considering a product that has as many different uses and styles of use as do bicycles, the CPSC produces nothing but ignorant superstition camouflaged under its presumed expertise as the government agency most responsible for the safety of the product.

Return to: John Forester's Home Page                               Up: Safety