Top: Home Page
Up: Table of contents
Previous: Preface to Web version
Next: Cover of FHWA version
COMMENTS ON THE BIKECENTENNIAL STUDY BY JOHN S. ALLEN
Contents of this page © 2002, John S. Allen
Comments by others are welcome; selected comments will be posted. Please submit by e-mail.
The development of the Bikecentennial trans-America route and the tours along that route in 1976 were landmark events in American bicycling. The organizers of the event are to be commended highly for having collected data and preparated a report as part of their project.
The Bikecentennial study, based on data from 4065 bicycle tourists riding all or part of the trans-America route in 1976, provides much useful information. The Bikecentennial data set, representing 10.4 million miles of riding, is generally robust and supports many of the conclusions well. The population base is diverse; it includes a larger percentage of younger cyclists than, for example the Kaplan report on League of American Wheelmen members.
Nonetheless, many conclusions in the report represent the authors' opinions to a greater or lesser degree, rather than being based on the collected data. The authors are straightforward about this (page 18).
The authors' experience with bicycling leads them to focus well on real issues, and in general greatly strengthens the report. With hindsight, however, some conclusions have been shown to be questionable, or have been disproved.
It is important to realize that the authors often had no choice but to rely on their own experience, because their study was a pioneering one. The very important Cross-Fisher study of car-bicycle collisions, for example, had not been published at the time the Bikecentennial report was being prepared.
My comments below are of three types:
Page references below are to the pages of the original Federal Highway Administration publication, DOT HS-803 206. The page references are bookmarked in the Internet version of the report, so you can go to the part to which a comment refers.
The comments below are organized according to the sections of the report to which they refer.
The riding environment of the BikeCentennial tours was somewhat unusual, in being limited for the most part to low-traffic rural roads and lightly-used state highways. It is notable in this connection that the overall crash rate (80 per million miles) is lower than that for the Kaplan report (113 per million miles), despite the generally lesser experience of the bicyclists (page 2). The authors attribute the low crash rate in large part to good route selection, and back up this conclusion with the evidence that crash rates vary greatly from one state to another along the route.
The conclusions about rates of accidents for different types of bicyclists also are interesting, and agree well with those of other studies.
The fatality rate of 1 per 5.2 million miles (page 3) is, however, not well-established, because two fatalities are too few to establish a firm statistical basis. A count of fatalities for all tours conducted by Bikecentennial (now renamed Adventure Cycling Association) from its inception in 1976 to the present would have greater validity.
The breakdown of all accident causes (page 4) is reasonably consistent with that of other studies. The rate of car-bike collisions is given as 7.8% of the total, and of bicyclist crashes that occurred while evading motor vehicles, as 5.8%. The rate of bike-bike collisions at 20.1% is much higher than that of the Kaplan report on League of American Wheelmen members, certainly reflecting to some degree that Bikecentennial participants usually rode in groups, while the Kaplan report accounted for all types of riding.
The conclusions that downhills are high risk areas (page 4), carrying baggage increases risk (page 4), and crash rate increases at the end of a long day's riding (page 5) are to be expected, but the Bikecentennial report was first to quantify them.
This study is also unusual in determining the level of reporting of bicycle crashes to police, (page 5), verifying that most crashes go unreported.
Conclusions about safety equipment (page 6) do not appear to be based on sufficient evidence to be statistically reliable, except for the statistic on percentage of helmet use. On the other hand, the report's call for better touring equipment in the interest of safety is most interesting, because new types of equipment have in fact come into widespread use since the study was conducted -- better baggage racks and bags, baggage trailers, and above all, all-terrain bicycles and hybrid bicycles with wider tires and sturdier frames.
The report agrees with others in its call for better bicyclist education (page 5) and law enforcement (page 7). The conclusion that bicycle use on roadways is acceptably safe (page 7) agrees with that of other studies. However, the call for a system of designated routes (page 8) raises some political concerns. The safety advantages of riding on well-selected or improved routes are well-supported in the study. However, there are problems with route selection and improvement which this study does not address:
The comments on specially constructed bicycle facilities in the Overall Problem section (page 20) accurately reflect the experience that such facilities are often unsafe for travel at normal bicycle speeds, or impractical as route segments. On the other hand, the statements about visibility which follow do not address the major issues raised by the Cross-Fisher study and other studies. Quoting the Bikecentennial report:
There are, to be sure, situations in which motorists fail to see bicyclists who are well-equipped and riding correctly -- including one of the two fatalities recorded in the Bikecentennial report. It has been shown, however, that bicyclists' principal problem with visibility occurs when they are riding in a position where motorists do not expect to see traffic -- for example, when riding on a sidewalk, or on the wrong side of the road; or at night through the failure to use adequate lighting equipment, or any at all. The Cross-Fisher study was the first to quantify this issue.
The most important action at a stop sign is to yield to traffic in the cross street. It is usually possible to be prepared to yield without stopping. Only 40% of respondents reported that they always obeyed stop signs (page 25), but few motorists come to a full stop either. A bicycle's generally slower speed and lack of a protruding hood more often allow a bicyclist to move forward slowly and still be able to stop if necessary. The bicyclist who does not come to a complete stop and put down a foot is also able to restart more quickly, reducing the length of the gap in cross traffic necessary to avoid conflicts.
Bicyclists, even more than motorists, can exercise caution while not coming to a full stop. Other countries, in particular France, use far fewer stop signs than the US, requiring only yielding. In this light, the statement that bicyclists will "soon abandon the caution required by law" when faced with too many stop signs, is off the mark, except, commonly, as it applies to sidepaths which have a stop sign at every cross street, and where yielding rules are confused by the requirement for both motorists and bicyclists to cross same-direction traffic streams while turning.
Bicyclists avoid signaling when turning (page 26) for the reason given in the report, that they would have to take a hand off the handlebar -- but for other reasons as well. When traffic is heavy, a bicyclist's signal serves as a request for a driver approaching from behind to let the bicyclist into line. This signal must be made before merging, not when turning. When the bicyclist's maneuver would not affect anyone, there is no purpose in signaling. The report makes no mention of merging, and fails to elucidate whether bicyclists understand that they can and should merge before turning. The report also implies that bicyclists should be expected to signal whether or not the signal serves any purpose.
The discussion of bicyclists' lane position (page 28) correctly states that "[b]oth bicyclists and motorists are confused on proper sharing of lanes." It is mentioned that there are some bicyclists who always ride in the middle of the lane. It is possible that bicyclists who always rode in the middle of the lane are simply discourteous and selfish.
However, John Forester's book Effective Cycling, the first to describe correct lane use principles, had appeared in its first edition one year before the Bikecentennial tour of 1976. It is possible that some of the bicyclists who always rode in the middle of the lane had adopting an exaggerated and distorted version of Forester's advice. Forester advocated riding in the middle of a lane when it would increase safety: in a lane too narrow to share, in order to discourage motorists from overtaking -- particularly on blind right curves where a bicyclist riding nearer the center of the roadway is visible from a greater distance. The discussion unfortunately does not clarify what was the motivation for bicyclists' riding in the middle of the lane
It is stated correctly that some bicyclists always ride at the right side of the lane, and that these bicyclists tend to be "more cautious about traffic, but sometimes are annoyed that motorists fail to see them." According to Forester, these bicyclists are being overcautious, because they fail to understand that assertive behavior is sometimes necessary to maximize their own safety.
The authors state that "[o]n uphills, bicyclists sometimes use a weaving motion across much of the entire lane. Although this is a hazardous practice in areas of traffic, it is done and should be anticipated." Such behavior was common in the 1970s, when a road-racing mentality dominated the bicycle industry. Very few bicycles were equipped with low gears suitable for climbing steep hills with baggage. Since approximately 1985, bicycles with low gears have been widely available, and "slaloming" up hills has become rare.
The 27% helmet use figure (page 28) was high for 1976, though it is not clear from this discussion what types of helmets were used. The first bicycle helmets that provided good protection had been introduced in 1974. Traditional, ineffective racers' "leather hairnet" helmets were still common in 1976, and some of the newly-introduced hard-shell helmets were of poor design.
Non-accident and accident rider profiles
The comment that "[a]ccident riders made more frequent repairs to their bikes" (page 31) suggests that their bicycles were of lower quality. That these bicyclists had more flat tires suggests that they were less skillful at avoiding road hazards and/or tended to ride near the right edge of the road even when it was in poor condition. Poor maintenance can lead to more frequent repairs: for example, underinflation of tires predisposes them to "snakebike" flats, which result from a tire's bottoming out on the rim and pinching the inner tube. Other descriptions of the accident bicyclists' behavior in the report also are indicative of a relatively lower level of skill.
The lack of correlation between number of years of active riding and accidents (page 33) is surprising; other studies, notably Kaplan's study and the Cycle Touring Club [U.K.] study show a great decrease in accident rate with years of experience. The Bikecentennial study does show a decrease with the bicyclist's age. It might be asked what is considered "active riding". It is possible to ride a bicycle for many years without learning good technique; exposure to skilled bicyclists, typically in the context of a bicycle club, greatly speeds the learning process.
That having taken a course of instruction made no difference in the likelihood of an accident (page 33) is not surprising; in 1976 there were no courses which taught scientifically-based accident avoidance for bicyclists.
The statement that "[I]n conclusion, it is unclear whether hazard recognition is a factor in accident prevention. Both groups had a similar awareness (or lack of awareness) of the real threats facing them" (page 33) fails to make an important distinction. Ability to describe categories of hazards when answering a questionnaire is not the same as the ability to recognize and avoid them while riding.
Note that the description of hazards in eastern Kentucky (page 38) includes road conditions themselves, as well as the presence of overloaded coal trucks traveling at high speeds. It is not stated here how the percentages of accident types in this section compared with those for other sections.
"A heavily weighted rear wheel tends to unweight the front wheel, reducing the tracking power and steering action of that tire." Strictly speaking, this statement is incorrect. Rather than the front wheel's being unweighted, the rear wheel is more heavily weighted. The bicycle becomes more unstable primarily because of its increased angular momentum in quick steering, and because the mass of the baggage at the rear causes the bicycle frame to flex in torsion in response to steering, resulting in slow steering response, increased lateral force at the front wheel/road contact patch, and sometimes a shimmy which is difficult to control.
A factor exacerbating this problem is the poor quality of baggage racks available in the 1970s. The common Pletscher rack was very insecurely attached to the bicycle, worsening the torsional flex problem. Also, as stated in the report, bicycle frames typical in the 1970s were not well-designed for touring use.
As of the year 2001, much better equipment is available. Subsequent to the preparation of this study, work by Jim Gentes for Blackburn Designs demonstrated that low-mounted front pannier bags result in a more stable ride than rear pannier bags. The front pannier bags do, however, carry a different risk: if one falls into the spokes of the front wheel, the bicycle stops short and the bicyclist is catapulted forward. A jammed rear wheel is much less likely to lead to a serious crash. Rear racks also are much strudier than was usual in 1976.
The percentages for bicycle-motor vehicle accidents (page 41) -- 10.4% for collisions and 7.1% for near-collisions resulting in the bicyclist's crashing -- are not the same as those in the Executive Summary, which are 7.8% and 5.8% respectively. This discrepancy may result from the present chapter's only examining the most serious accident, when a bicyclist had more than one accident.
The increase in accidents near the end of a long day's ride (page 42) is an interesting phenomenon revealed by this study. Fatigue is the most credible explanation. Lighting conditions were an unlikely cause of the increased accident rate in the late afternoon, as the rides occurred in the summer, with daylight savings time in effect, and the sun did not set before 7:30 PM on any day when the tours were taking place.
Bicycle parts failure (page 43) accounts for only 7% of accidents. This finding is in line with those of other studies. The Kaplan study, of a generally more experienced group of bicyclists, found an even lower rate.
The typical bicycle of the 1970s had poor brakes compared with the typical bicycle of today. Problems included rims with splayed sidewalls which caused the brakes to grab unless the wheels were trued to perfection; steel rims which had poor braking performance in wet weather; and brakes with a relatively low mechanical advantage. These conditions were considered normal in the 1970s and might not have been reported as mechanical failure. As of the year 2001, braking (except on the lowest-price discount store bicycles) has improved greatly, with parallel-sidewall aluminum alloy rims and more powerful braking mechanisms.
The statement that "an unusually high number of motor vehicle-related accidents go unreported, especially the incidents where a motorist crowded a cyclist off the road" makes a comparison with motor vehicle accidents as a whole, not with bicycle accidents as a whole. The reporting rate for bicycle-motor vehicle accidents is higher than for other types of bicycle accidents, as is shown elsewhere in the report, and in many other reports.
Bicycle-motor vehicle accidents
Type of involvement of motor vehicles (page 49) reflects the predominantly rural TransAmerica route. The Cross-Fisher study, which examined the differences between rural and urban accident type distributions, had not yet been published at the time the Bikecentennial study was written.
The table of types of vehicles (page 52) would be more meaningful in connection with information about the percentages of vehicles of each type using the roads. An analysis would be possible in connection with traffic counts from the year when data was collected.
"As mentioned before, until the motorist is able to recognize the special problems confronting bicyclists and understands how best he may share the roadway, unavoidable accidents will continue." (page 62). It would be more accurate to say that avoidable accidents will continue.
Bicycle equipment, parts and accessories
The authors' comments about bicycle equipment (page 63 and following) are especially astute. Many of the improvements which the authors suggested have come to pass in the 25 years since the report was published, as a result of product innovation, the expanding market for good-quality bicycle equipment, and the development of all-terrain and hybrid (road/off-road) bicycles. Equipment which was available only on high-priced or custom bicycles in 1976 has become widely available in mass production since that time.
The authors' recommendation of helmets, rear-view mirrors and conspicuity aids as safety accessories (page 67 and following) has not corresponded as well to developments in the years since the report was written.
A serious issue with recommending safety equipment is whether it should be mandatory or not. Advocacy of specific equipment has sometimes led to its being required by law, resulting in presumption of negligence for bicyclists who do not use it. That issue has become especially heated in connection with helmets. On the other hand, the law does need to set minimum equipment requirements. One which is conspicuously not stated in the Bikecentennial report is for a headlight when riding at night. The riders were expected to complete their day's tour segment during daylight; how many also used their bicycles to run errands during the evening is not stated in the report. An examination of accident data trends over the course of the summer might shed some light on this issue, as darkness comes earlier toward the end of the summer.
The questionnaire item about preferred roadway improvements (page 73) is problematic, because it reflects uninformed opinion. In particular, the idea of paths adjacent to the road is popular with uninformed bicyclists, though these paths have been shown to be unsafe (see documentation elsewhere on this site); and the distinction between a designated bike lane and an undesignated striped shoulder would not have been clear to respondents in 1976, at which time bike lanes were rare but undesignated striped shoulders were very common.
The statement (page 77) that "[t]he accident rate for this stretch of highway was the same as the average for the trail--one accident for every 9.4 miles ridden" referring to a stretch of Interstate highway which was part of the TransAmerica trail, is in error. The accident rate reported elsewhere is 79.82 per million miles, or one accident per 12,528 miles ridden. Also, policies on bicycle use of Interstate highways have changed since the report was written. Colorado, for example, has a number of sections of Interstates where bicyclists are permitted to ride on the shoulder. The statement "[c]yclists permitted to use interstate shoulders should be required to exit at each ramp in order to avoid conflict with motor vehicles" prescribes only one solution, while others may be more suitable. Exiting at a ramp is not practical where the interchange is with a divided highway, and may be unsafe under other conditions as well. Bicyclists may have the option to stop where a ramp diverges from the highway, yield and cross the ramp there; or they may continue to the left of the ramp.
Appendix a, bicycling skills
This appendix comes to reasonable conclusions:
This chapter is, however, too short to describe the necessary skills in detail, and it also suffers from having been written before scientifically-based bicycle riding skills instruction was available.
The implication that improvements suitable for pedestrians would also be suitable for bicycling is distressing in the statement (page a-1), "[t]he compounding factors are making use of roadways not designed for pedestrians or other self-propelled travelers and sharing these facilities with drivers of high-speed, heavy-mass vehicles who are not accustomed to the low profile of a bicyclist." The operating characteristics of bicycles and pedestrians have little in common.
The information on bicycle size (page a-2) is outdated, because more sophisticated bicycle fit measuring systems have been developed; and because new types of bicycles (off-road, recumbent) have different sizing requirements. Also, once the bicycle's fit is approximately right, fine adjustments need to be made based on the bicyclist's experience.
"Ankling" (page a-3) has been largely dismissed as a myth since the report was written. Failure to use the ankles correctly in pedaling results largely from incorrect saddle height. Correct use of the ankles develops automatically if the saddle height is correct, and especially if toe clips or clip-in pedals are used.
The information on steering (page a-4) is technically incorrect. The bicycle indeed steers by leaning, but the lean is initiated either by the bicycle's inherent weave to keep balance or intentionally by countersteering. Leaning the body does not initiate a turn, but only makes the bicycle lean the other way.
The advice on emergency braking (page a-4) does not give the essential information on how to modulate the front brake which was first widely published in John Forester's book Effective Cycling. For the shortest stops, the rear brake is used lightly so that skidding of the rear wheel becomes an indicator that force on the front brake lever must be decreased.
"Lumber, coal and other large trucks need extra space, and a rider must be ready to leave the roadway if necessary. These professional drivers are earning their living on the road, while cyclists are using it for recreation. A friendly wave will usually be returned." (page a-6) Any overtaking driver who fails to wait for a safe opportunity to overtake is in violation of traffic law, and any driver who does this intentionally is committing an act of assault. Bicyclists may not be able to prevent such conduct on the road, but such behavior should not be condoned.
"Time of day is often an important safety consideration. Temperatures are usually best, and traffic lightest, in the early morning. Leaving or approaching major population centers during rush hours or the noon hour should be avoided. One should be off the road before 5 PM--poor light, fatigue, and the drinking driver make this the most dangerous time of day." (page a-7)
"Night Riding. This is not recommended. Riding at night requires wearing bright, reflective clothing, using lights and reflectors front and rear, and listening for traffic. One must leave the roadway when a car is heard or seen approaching."
It makes good sense for tourists to finish their rides during the day, but use of the bicycle for transportation -- including evening errands during a tour -- requires riding during rush hours and hours of darkness. The two statements above fail to stand up for this basic need and right. Leaving the road whenever a car is heard or seen approaching is not practical, and not necessary if the bicyclist's lighting and reflectors are adequate. It is easier for a bicyclist to tell whether a car is overtaking safely at night than during the day, by the bicyclist's looking ahead at the shadow the bicyclist casts in the light of the motorist's headlamps.
Appendix c, tables and maps
All in all, the data analysis provides some valuable results. However, there are some cross-correlations which were not conducted and might answer important questions. One very important table of results is missing from the report. Only the title, relating to style of riding on the day of the accident, is present.
Marital status, amount of education and student/educator status (page c-3) correlate strongly with accident rate. However, they also correlate strongly with age (page c-2). There does not appear to have been an attempt to determine to which degree these variables correlate with each other, in order to determine whether there is an independent effect.
That right-handed people have a lower accident rate (page c-3) has been shown in other studies but this study is unusual in making this finding as it applies to bicycling.
That having taken a bicycle course correlates negatively with accident rate (page c-4) is surprising except that no courses existed at the time of the Bikecentennial tour which taught scientifically-based accident avoidance. But why would a course correlate with increased risk of accidents? Maybe because younger people are more likely to have taken one?
The use of extension levers (page c-5) correlates negatively with the accident rate. Extension levers as manufactured in the 1970s were incapable of applying full braking power, and were placed where hand position on the handlebars reduced stability. So why the negative correlation? Perhaps because respondents without extension levers had difficulty with the levers on their drop handlebars, or because they had become fatigued and did not hold their heads up to look forward when in the drop position to use the brake levers? Note, however, that the correlation between type of handlebars and accident rate is weak.
The long list of positive correlations between parts failure and higher accident risk (page c-6) points to the accident riders' having poorer bicycles or maintaining them more poorly, both of which would increase risk.
Lower accident risk correlates with slower speed and with obedience to traffic laws (page c-7). These results are expected. That frequent mechanical checks correlate with a higher accident rate is possibly explained by the need for them because the bicycle was in poor condition, or by the bicyclist's lack of skill in detecting mechanical problems while riding. The results concerning bright clothing and safety flags are positive but statistically inconclusive; those concerning safety triangles are confusing.
Riding at night correlated with a higher accident rate (page c-9). Riding carefully in wet weather correlated with a lower accident rate, consistent with Kaplan's findings. Positive answers to these questions do not indicate that an accident occurred at night or in bad weather; only that riding was done under these conditions.
Riding alone correlated with a higher accident rate. Reasons for this could have to do with lone riders' learning less about riding skills from companions; with less information's' being exchanged about local conditions; and with increased visibility of groups of cyclists.
Riding while ill increased the accident rate -- no surprise here.
The title COMPARISON OF NORMAL PROCEDURES WITH PROCEDURE FOLLOWED ON DAY OF ACCIDENT FOR ACCIDENT SAMPLE is at the bottom of page c-9 but the table is missing. This table would likely provide some important insights.
Opinions of most hazardous riding conditions (page c-10) show some interesting discrepancies from the actual numbers of accidents under various conditions.
Carrying baggage increased accident rates (page c-33). That rates for bike-inn riders were lower may reflect that they carried less baggage but also that they were likely to be more affluent and so, older.
Locations of fractures (page c-40) and bruises (page c-43) correlate well with hilly conditions (page c-38).
Top: Home Page
Up: Table of contents
Previous: Preface to Web version
Next: Cover of FHWA version