[Top: John S. Allen's Home Page]
[Up: Research]
[Previous:]
[Next:]

[contact John S. Allen by e-mail]

Site Logo, Track bike (2 KB GIF)Bikexprt.com Web Site


John Allen's comments on Chapter 4 of the FHWA's
Bicycle Safety Related Research Synthesis

Some subtopics -- those covered in depth are bracketed by asterisks.

Introduction
*Erroneous statements about the Effective Cycling program*
The ethics of marketing-driven facilities planning
*Misrepresentation of the 1975 Kaplan study as supporting bike lanes*
Erroneous comparison of crash studies
Some odd arguments in favor of bike lanes
*An example of use of statistically invalid data*
*Corvallis, Oregon crash rate reduction attributed erroneously to bike lanes*
Madison, Wisconsin study shows crash rate increase after bike lanes installed
Misrepresentation of a Danish crash study

This article presents my comments on Chapter 4 of the Federal Highway Administration (FHWA) report Bicycle Safety Related Research Synthesis, a document which government officials and planners frequently cite in support of bike lane construction.

The Synthesis was prepared for the Bicycle Federation of America (BFA). The BFA (renamed in 2001-2002 to the National Center for Bicycling and Walking or NCBW) obtains money through foundation grants, solicits membership from planners in government agencies at the Federal, state and local level, lobbies on behalf of ISTEA and other bicycling-related legislation, organizes the biennial Pro-Bike and Pro-Walk conferences, and fulfills research contracts for the Federal Government and other government bodies. The Bicycle Safety Related Research Synthesis is the product of a Federal contract.

I have very mixed feelings about the NCBW. Clearly, it is an effectively managed organization, which has found its way into the corridors of government. The call in Washington for alternatives to private motor vehicles for transportation is stronger as a result of the NCBW's efforts. I have attended eight of the national Pro-Bike conferences, in 1982, 1984, 1990, 1992, 1996, 2000, 2002 and 2008, as well as a between-years regional conference in 1991. I find these conferences very useful for keeping in touch with other people who are concerned about bicycling issues.

On the other hand, the NCBW is not a membership organization directly representing bicyclists. Neither is the NCBW an academic institution, whose work is subject to academic standards of accreditation and peer review. I am uncomfortable with the NCBW's approach to many of the issues facing bicyclists. I am also uncomfortable with many of the products of the NCBW's contract work. To me, it appears that the NCBW, in its close relationship with the Federal Highway Administration, is establishing a political agenda for bicycling which is based on a marketing vision, rather than on sound scientific and engineering principles. The NCBW shapes its agenda to appeal to reluctant government officials, planners and the non-bicycling public.

No product of the NCBW's work puts my concern into more stark relief than the document entitled Bicycle Safety Related Research Synthesis, which the (then) BFA produced for the FHWA in 1995.

What is most disturbing to me about the Synthesis is its ideological bias -- a stance of "ends justify the means," rather than the scientific impartiality which is appropriate in a review of scientific literature. The end which the Synthesis advocates is an increase in bicycle use. The primary means proposed is the construction of bicycle lanes.

The Synthesis contains numerous errors of fact and outright misrepresentations of the research literature as it rushes to build a case for bicycle lanes. To some degree, these misrepresentations appear to reflect lack of training of the authors of the Synthesis in statistical methods, or else careless, rushed preparation of the Synthesis. On the other hand, some of the misrepresentations, in particular those about the 1975 Kaplan study, are very striking.

An impartial review of the research literature reveals that there is no single, simple answer to the question of whether bicycle lanes are in fact safer and more desirable than other planning alternatives. But many of the research documents which the Synthesis cites are out of print, and not easy to obtain. In that light, I find the misrepresentation of these documents in the Synthesis very alarming. It amounts to a rewriting of history. A new generation of activists, planners and officials may well derive its interpretation of the research literature not from the original documents, but from the Synthesis.

My comments on specifics of Chapter 4 of the Synthesis will document the points I have just made. Another review of the Synthesis has been prepared by John Forester and is available on his Web site. Yet another has been prepared by Alan Wachtel and is on the same site as the review you are reading now. Each of these reviews raises different points. John Forester's is the most comprehensive. Mine covers some issues in more depth. Alan Wachtel's is short and to the point about the evidence for and against the safety of bicycle lanes.

And now, my comments. In order to understand them, and in fairness, you need to have a copy of the Synthesis. It is now available online.

I recommend John Forester's Books Effective Cycling and Bicycle Transportation for a contrasting point of view on many of the research documents cited in the Synthesis. Both of Forester's books are published by MIT Press, and may be ordered through MIT Press directly or through any bookstore.

The boldface statements in quotes, below, are from the Synthesis, starting near the beginning of Chapter 4. Following each quote, my comments follow in the normal typeface.

"The AASHTO Guide (and two other publications) culminated an intense period of 6 or 7 years of experimentation and research into many aspects of bicycling."

What intense research and experimentation? Only two studies are mentioned here. They are both good studies, and both of them contradict the Synthesis, as I will show. 

"Forester’s highly influential Effective Cycling was published in 1984..."

The first MIT Press edition, published in 1984, was the fourth edition of the book. Three earlier editions were published by the author’s business, Custom Cycle Fitments, beginning in 1975.

"Effective Cycling holds that bicyclists fare best when they behave like motor vehicles and that bicyclists should not venture onto the roads until they can ride this way."

This is a garbled paraphrase. Vehicles can not "behave." They are inanimate objects. The correct quote from Forester is "Cyclists fare best when they act, and are treated in return, as drivers of vehicles..."

As to the "not venture onto the roads" part of the quotation, that’s 180 degrees from the truth. Here’s Forester’s advice to beginners, from Effective Cycling: "for now, however, ride on easy-traffic roads...then you can cycle enough to get experience, which will allow you to understand and practice the more advanced habits and maneuvers." The Effective Cycling program teaches road riding skills through road riding.

Using the term "motor vehicles" instead of "vehicles" falsely implies that fundamentally different principles of operation apply to bicycles and to motor vehicles, while ignoring minor, but important differences. For example, the traffic law specifically exempts bicyclists, who can easily use hand signals, from the requirement to have turn signal lights.

These errors make it clear enough that the authors of the Synthesis are unwilling or unable to think clearly about vehicle operation or to represent Forester’s ideas accurately.

[Continuing to paraphrase Forester] "Bike paths and special facilities are demanded by people with a ‘cyclist’s inferiority complex’ about riding on the road..."

This statement is largely true. The problem with special facilities is, however, not that any are built but that many of them are built by people with little understanding of bicycling. This leads them to be built in the wrong places, for the wrong reasons, and with serious designed-in hazards. 

"and by people with ulterior political and environmental motives for encouraging bicycling."

There are good political and environmental motives to encourage bicycling, but there is no excuse in a scientific study for an "end justifies the means" approach. It is this approach, as typified in the Synthesis, which is ulterior.

Turning the tables and describing Forester's characterization of environmental and political motives as ulterior insinuates that Forester does not seek to encourage bicycling, and is anti-environmentalist. By extension, the same insinuation is made about education programs based on Forester's work. This is a "spin" on the facts, 180 degrees from the truth. Skillful bicyclists ride several times as far per year as typical daily, casual bicyclists. Forester has described a practical program of cyclist instruction and infrastructure improvement to extend these benefits to larger numbers of Americans.

Forester's recommendations are frequently criticized and ignored by people who understand the environmental benefits of bicycling but who do not understand bicycling; also, by motoring interests who see separate facilities as a way to remove bicyclists from the roads. But Forester's opposition to poorly-conceived bicycle facilities projects is not opposition to environmentalism. He is, among other things, a long-time Sierra Club member -- but I digress.

 "The book has a hard core of followers drawn predominantly from the ranks of experienced and club bicyclists around the country...however, less than 3,000 bicyclists are estimated to have passed an Effective Cycling course..."

A few paragraphs earlier, Forester's book was described as highly influential. Now, by an unattributed estimate, his message reaches less than 3,000 cyclists. How can we have it both ways?

By using only the smallest possible number, and the opinionated term "hard core," the Synthesis attempts to marginalize the Effective Cycling program, and understates its impact by a factor of at least 1,000.

I can go quite a ways toward demonstrating this with my own data. My own pamphlet, Bicycling Street Smarts, which teaches the same road techniques, has sold over 400,000 copies and has become the official bicycle drivers' manual in five states. My revised edition of Glenn’s New Complete Bicycle Manual was published in 1987 and has sold several tens of thousands of copies. My series of articles on road riding techniques published in Bicycling magazine in the early through mid 1980’s reached a readership of approximately 2 million. I am only one author among several who have achieved comparable exposure for our writings on Effective Cycling techniques.

By any reasonable measure, the Effective Cycling program and work based on it are a success, as measured by the bicycling habits and safety record of people who ride according to its principles, and by the observed improvement in bicyclists’ traffic skills in this country since the program started. If the program had the support it deserves from government, it would succeed much better yet.

If the topic of the Synthesis were water sports instead of bicycling, the Synthesis would disparage the Red Cross water safety program as having "a hard core of followers drawn from the ranks of experienced swimmers, YMCA and swimming club members." It would go on to recommend the construction of wading pools to increase safety and the popularity of swimming. Of course, neither the public nor public officials would buy these arguments: the Red Cross program is too well-known and understood, as are the limitations of swimming in wading pools.

The Effective Cycling program is the bicycling equivalent of the Red Cross water-safety program. There is no difference in their approach. The only significant differences between them are their subject matter, and that the Red Cross program is widespread and well-known enough that its name is a household word.

Andy Clarke, co-author of the FHWA report, and John Forester are engaged in a feud of sorts. Forester sometimes attacks the motives of people who are basically well-meaning. He takes his anti-bikeway stance farther than is considered reasonable by many cycling advocates, even those who sympathize with him. He alienates many people in this way, and Andy Clarke is one whom he has alienated.

Forester is, however, careful about facts. If you want to debate him, you can challenge him on well-defined issues and find documentation for his opinions. Read his books and see. The Synthesis, on the other hand, as shown by the examples in this review, often doesn’t represent facts and research documents accurately.

Selecting and Designating Bicycle Routes...shows a strong inclination towards the attitudes and philosophies of Effective Cycling...as such, it was probably an accurate reflection of the mood of the mid-1980’s, and the reality of what was feasible at that time.

Bill Wilkinson, who wrote Selecting and Designating Bicycle Routes, was Andy Clarke’s boss at BFA. The quoted statement is BFA’s description of its own change in policy. It insinuates that other strategies are superior to the Effective Cycling approach. Furthermore, it insinuates that Effective Cycling is somehow bound up with an undefined "mood", perhaps reflecting the transportation policy of the Reagan administration, which was more focused toward motor vehicles and less toward alternatives than other recent administrations. Unlike the policies of the Bicycle Federation, those of Effective Cyclinghave never blown with political winds, because Effective Cycling has always been a scientifically grounded program of cyclist instruction.

 "While the new AASHTO Guide covers much of the same ground as the 1981 edition, there were some important changes made."

 This overstates the progress between the 1981 and 1991 editions. The changes were minor, as a comparison of the two Guides makes clear. The fundamental emphasis and design specifications are hardly changed. 

"Accommodating current use [vs.] Increasing the number of users"

The central question is: what is a responsible approach? There are several ways to look at this question. Many environmentalists and planners regard bicycling as a tool to improve environmental and social conditions, by reducing automobile use and shaping urban land use to a human scale. For many people, this is the central rationale for more bicycling. This is the approach which the U.S. Government expresses in its national bicycling and walking study, and which Wilkinson advances in his 1992 report cited here.

However, the approach of increasing the number of users by giving them what they think they need steps over the line separating responsible engineering from an "end justifies the means" approach. While some special facilities make bicycling more attractive and also decrease its risks, many people who advocate special facilities are not knowledgeable enough to evaluate facilities designs for safety and utility. Most planners and other government officials who are in a position to construct bicycle facilities also lack this knowledge. Furthermore, excessive emphasis on facilities and too little on education perpetuates the extremely low level of skill which is typical of Americans who ride bicycles.

When they are made aware of the dangers of their proposals, advocates of special facilities often search high and low for evidence that special facilities really are safer. If the advocates can not convince themselves, they may adopt an "ends justify the means" approach, accepting bicyclist casualties as a tradeoff against expected environmental benefits. Wilkinson, as cited here, adopts a third justification: that of marketing, of "selling" people on bicycling. As described here, a facility succeeds if it induces more people to ride bicycles. As paraphrased in the Synthesis,

"find out what potential and infrequent users say would encourage them to ride more often and start to offer them the product."

This is a dangerous approach when public safety is at issue. Potential and infrequent users know very little about bicycling, and a "product" designed to reflect their desires may endanger their safety. If I let my 5-year old son just eat what he wanted to, he would eat candy all day.

Marketing-driven engineering led to the Challenger space shuttle disaster. The Martin-Marietta Corporation ignored an engineer's warnings of a serious technical problem, and gave the Federal Government what it wanted: lots of space shuttle flights which made NASA look good, and made money for Martin-Marietta. Then, the Challenger blew up.

Planning for bicycling is not rocket science -- but in some ways, it is more complicated. Mistakes in planning for bicycling can persist much longer, because failure is not as spectacular, and the reasons for failure are not as obvious. Though many more people die from poor bicycling policy than died in the Challenger disaster, they do not all die at once in a spectacular explosion which is broadcast worldwide on television.

Wilkinson (as paraphrased in the Synthesis) does qualify his statement, by saying that "special facilities for B/C cyclists [less experienced cyclists] are additions to the existing system, not substitutes for shared use of the roadway." This statement is appropriate in a discussion of bicycle paths, and indicates that bicyclists should not be prohibited from the roadway because a path provides an alternate route. This statement is more problematic in a discussion of bicycle lanes, which redefine roadway use rather than substituting for use of the roadway. Unfortunately, however, once any type of special facility is provided, alternative bicycle-related improvements are usually neglected.

Wilkinson as cited here declares an armistice between the "Elite Cyclists Syndrome" and "Bicycling is Dangerous syndrome." Both terms are uncomplimentary, but the term "elite" is inaccurate. Those who are regarded as "elite" by most Americans are merely competent.

Let's put this into perspective. What is an elite cyclist? Well, for example, Greg Lemond is an elite cyclist. He won the Tour de France three times. But in some ways, amazingly, he is not competent. The descriptions of road safety and use of the brakes in his book, for example, are incorrect. As a truly elite cyclist, he has led a sheltered life. A magazine article a few years ago described how he rode a mountain bike on a flat tire because he was not equipped to change it.

The section of the Synthesis on "Separation versus Integration" is relatively fair in pointing out that the safety of the Dutch model for bicycle facilities has been called into question. However, there are other inaccuracies. For example:

 "Proponents of Forester have subsequently extended his theory to include bike lanes."

 The source of opposition to bike lanes and the scientific underpinnings of it are both misstated here. Forester himself extends his theory to bike lanes, at length. The discussion occupies 5 pages in Effective Cycling and 11 pages in Bicycle Transportation. Forester himself conducted experiments and reviewed research data on bike lane safety, and also cites operational traffic theory.

"The Kaplan study actually found bike lanes to be slightly over half as dangerous as major and minor highways and approximately five times safer than separate bike paths."

This is a serious misrepresentation of the findings of Jerrold Kaplan's 1975 study, Characteristics of the Regular Adult Bicycle User. There is no way that the Kaplan study can be cited either to support or to condemn bike lanes. Kaplan refers not to "bike lanes" but to "on-street bicycle facility (lane, route)." (Kaplan, Characteristics of the Regular Adult Bicycle User, p. 76, table 13). Kaplan’s data table, and his questionnaire item about crash locations, which I have reproduced on this Web site for reference, make no distinction between a bicycle lane and a bicycle route. The entire Kaplan study, which had long been out of print, is now posted on this web site, so you may read it and draw your own conclusions.

When Kaplan collected his data, bike lanes were rare. There were, however, a number of designated bicycle touring routes around the country, mostly on lightly-traveled rural roads. Almost certainly, most of the riding which Kaplan described as on "on-street bicycle facility (lane, route)" was on designated routes rather than bike lanes. 

Furthermore, the amount of riding and number of crashes on "(lane, route)" reported by Kaplan are too small to have scientific validity, even before splitting them up into two separate categories for routes and lanes.

"The Kaplan study is dated, based on a small number of accidents reported by a narrow sample of bicyclists, and is, at best, inconclusive."

This statement misrepresents the relationship between sample size and statistical validity. The Kaplan study reports 754 crashes from a sample of over 3,000 bicyclists. It is statistically robust. A sample of 1000, as commonly used in pre-election polling, is sufficient for a 2% confidence interval.

Most of the 754 crashes reported in the Kaplan study occurred on roadways not specifically designated for bicycling. 81 of the crashes occurred on bicycle paths, but only 14 on "(lane, route)." Only the "(lane, route)" category is so small that the results are statistically inconclusive. Both for this reason and because it makes no distinction between lanes and routes, the Kaplan study does not support the pro-bike lane conclusions which the Synthesis says that it does.

The Kaplan study does represent an atypical sample group (League of American Wheelmen members and Washington Area Bicycle Association members), but this is the most important thing about it. In comparison with other studies, the Kaplan study shows how astonishingly lower the crash rate of competent bicyclists is. In this way, the Kaplan study shows that good riding skill is highly effective in reducing bicycle crash rates.

The Kaplan report surveyed bicyclists directly, rather than seeking them out through police reports. There is systematic error in police reports: for example, crashes involving motor vehicles are much more likely to be reported to the police than are others. Only studies which survey bicyclists directly reflect proportions of different types of crashes accurately. The Kaplan report is unusual among studies of bicycle crashes in that it is based on accurate recording of mileage.

The Kaplan report is indeed dated, though none of the basic factors (except perhaps for helmet use) that would influence rates of serious crashes have changed enough to weaken its strong conclusions. The 1996 Moritz survey of League members reached very similar conclusions.

 "According to Cross and other crash studies..."

We have another apples-and-oranges comparison here. The comparison here fails to point out that the Cross study examines only bicycle-motor vehicle collisions. That is why the percentages of the crash types listed in the Synthesis for the Cross study are about twice as high as for the other two studies cited side by side with it in the Synthesis. This is a gross error.

Furthermore, all three studies cited here include only crashes reported to the police. As I have already stated, crashes involving motor vehicles are more often reported to the police. Therefore, the percentages given here for car-overtaking-bike collisions are much too high, even in the studies which include non motor-vehicle crashes. Cross and the other authors of the cited studies acknowledged the limitations of their methods of data collection, but the authors of the Synthesis do not.

 "During the past 20 years, much has been learned about the planning, design, operation and maintenance of many bicycle facility types....the debate between separation and integration is looking increasingly contrived."

Separate facilities are indeed useful in some places [see example of how they may, for example, provide neighborhood connectivity], but this statement constitutes a "best of all possible worlds" interpretation rather than the reality of substandard facilities which continue to be constructed, or the reality of cost/benefit comparisons between different facility types in different locations.

Designing and Selecting Facilities

 I have little to disagree with in the wording of this section. The problem is one of interpretation. Inappropriate facilities are very often constructed, and appropriate facilities that might be provided -- especially as part of highway projects -- are often neglected, with the result that progressive deterioration of bicycling conditions accompanies improvements in motoring conditions.

Shoulders

I agree with most of this. Shoulders differ from bike lanes in that there is no expectation that bicyclists are required to use them to the exclusion of other parts of the roadway, or motorists to stay out of them regardless of whether bicyclists are present. Permission of both types of vehicle to use both parts of the roadway reduces potential problems at intersections (see below).

Wide Curb Lanes

This section gives lukewarm support to wide curb lanes, claiming that they "accommodate only the group A cyclist and have a limited benefit in encouraging increased use of bicycles." That statement has no basis in scientific fact.

The clear implication is that shoulders or bicycle lanes are preferable. The distinction in terms of safety from overtaking traffic, if any, is one of perception rather than of scientifically-determined fact. I have already discussed the ethical issues of catering to the desires of uninformed bicyclists.

Bike lanes generally fail to channelize traffic properly at intersections, an issue of which the authors could not be unaware, but avoid discussing; that issue is a point in favor of wide outside lanes rather than bike lanes -- or shoulders. Another important issue is the sweeping of debris from the travel lanes by motor vehicles. I will bring that up in connection with another section of the Synthesis.

John Forester (in a response to my inquiry by e-mail) has described a couple of the other studies cited here:

 "You asked some questions. The Maryland study about wide curb lanes was Steve McHenry's. He showed that motorists overtake cyclists reasonably well with either wide curb lanes or with bike lanes, with a bit less diversion from the direct path with bike lanes. However, I do not know whether the motorists initially drove closer to the curb without the bike-lane stripe, so that they simply moved to the usual overtaking position. Nothing else there.

"Kroll and Ramey's study was of motorists overtaking on a particular street in Davis, done as part of the FHWA bikeway studies. The street had a lane width of 23 feet, as I remember. They photographed the pair of a stooge cyclist and an unaware motorist at the moment of overtaking and measured the distance between them, using the diameter of the bicycle wheel as the scale in the photograph. They said that with the bike-lane stripe both the stooge cyclist and the unaware motorists swerved less. However, with a lane width of 23 feet, almost anything could be predicted. The motorists had plenty of room, why not move over? That's irrelevant to real life.

"Notice that the studies place considerable significance on the amount that motorists swerve. So far as I know, nobody has recorded any history of car-car collisions because of motorists swerving around cyclists, but this is always one of the items that motorists like to talk about. They think that the lane line keeps cyclists out of their way, so they allow less clearance. Good or bad? So far as I know, it makes an insignificant amount of difference compared with the bad effects of bike-lane stripes."

Bicycle Routes

This section is generally fair except for its emphasis, placing the disadvantages of bicycle routes first. In that way, this section once again expresses the bias toward bike lanes.

Most designated routes have in fact been developed for bicycle touring, as discussed toward the end of the section. A good example is the designated Adventure Cycling Association Transamerica route from Oregon to Virginia. A robust study of tourists on that route showed a significantly lower crash rate on the designated route than for the same population of bicyclists riding off the route.

The section also fails to give sufficient emphasis to the fact that touring bicycle routes are not only on roadways. These routes typically make use of what the route planners consider to be the best available facilities which connect with each other, regardless of their type: shared roadway, bicycle lanes, bicycle path. The type and quality of the route depends on the route planner as much as on the facilities available. Furthermore, the rationale for touring routes is quite different from that for local transportation bicycling. The touring routes are used primarily for longer-distance weekend and vacation travel. Bicycle tourists spend hours at a time in the saddle, covering distances typically ranging from 30 to over 100 miles per day. The object of this travel is recreation, physical exercise, sightseeing and pleasure. An indirect route may often be preferable to a shorter, direct one if it serves these purposes.

This section also includes an illustration of the Denver "Bike arrow" pavement marking which serves as an alternative to bicycle lanes, "offering the encouragement and promotional value of the bike lane without restricting the operation of the bicyclist [should be "bicycle" -- another confusion of driver with vehicle] to the narrow bike lane (which is a concern of opponents of bike lanes)." This description is welcome, but in my opinion, the "bike arrow" deserves more lengthy discussion.

Bicycle Lanes

"Ronkin (1993) addresses these concerns.

The referenced publication by Michael Ronkin is a short article in the March, 1993 issue of Pro-Bike News, the BFA newsletter. The article is not a research paper. It contains Mr. Ronkin's opinions but no research results. The citations which follow are from Ronkin' article, until further notice.

"The difficulties associated with making a left turn in traffic has [sic] nothing to do with the presence of a bike lane."

This statement misses the point, which is that when bike lanes are present, bicyclists are encouraged to make left turns incorrectly, from the right curb, and motorists are more likely to be annoyed if they do not. The statement is also not entirely true, because narrowing the travel lane to provide a bike lane can reduce the width available for motorists to overtake bicyclists waiting to make a left turn.

"Conflicts with right turning movements are dealt with by dashing the bike lane stripe before intersections and dropping the markings altogether across intersections."

This would be nice if it worked. In practice, the dashes do not succeed reliably in encouraging motorists to merge into the bike lane before turning right, and can not be placed at every intersection and driveway when these are closely spaced. In many cases, dashes are not used even when they would be practical.

"Do bicyclists really want to be dependent on automobile traffic to blow the street clean -- or is it more appropriate to schedule more frequent sweeping?"

Sure, it would be nice to have more frequent street sweeping, because there often isn't enough. More of it can't be scheduled with a wave of the hand.

"Bike lanes allow motorists to turn into the roadway without encroaching on another lane, and negates [sic] the need for wide-turning radii at corners..."

What this says is that motorists turn across the bike lane as they enter the roadway. Given the same roadway width, they would do the same thing without a bike lane. [End of statements taken from Ronkin's article.]

"Studies that have reviewed the impact of bike lanes do not support this concern [about safety]."

John Forester has examined some of these studies in his review of the Synthesis. So has Alan Wachtel, in his. Please see their comments. I have already commented on the misinterpretation of the Kaplan study in the Synthesis.

"The introduction of bicycle lanes on 18th Avenue in Eugene, OR in 1979 resulted in an increase in bicycle use and a substantial reduction in the bicycle accident rate. In the 18 months following installation, the crash rate per 100,000 bike miles (161,000 bicycle kilometers) fell from between 4.5 (1977) and 9 (1978) to 3.9 in the first 6 months and 2.6 in the following 12-month period."

The pre-installation crash rates given in the Synthesis are higher than those in the 1979 Eugene study. That study reports bicycle mileage only for the period 1974-1978, and not separately for each year; however, the number of crashes for the five years 1974-1978 in the 2.8 mile segment where the bike lanes were installed was, respectively, 9, 8, 7, 4 and 4. The doubling of the crash rate from 1977 to 1978 which the Synthesis reports would only be possible if there was a major decline in bicycle use between 1977 and 1978. The date given in the study for striping of the 2.8 miles of 18th Avenue  is October 1978, not 1979; two of the 1978 crashes in the part that was striped occurred after the striping. The 1980 follow-up report cited in the Synthesis may possibly include crashes which had escaped reporting in the 1979 study, but in any case, the numbers don't match. The Eugene bicycle coordinator has warned that the raw numbers of crashes on 18th Avenue are small and that the comparison may not lead to a statistically valid conclusion [her comments].

Nonetheless, the conclusion that the Eugene example indicates a major safety increase due to bike lanes had been repeated in subsequent literature. For example, A Comparative Analysis of Bicycle Lanes Versus Wide Curb Lanes: Final Report, U.S. Federal Highway Administration publication No. FHWA-RD-99-034 (page 11 of PDF file; page 4 of printed document) states: 

Two other studies credited BLs with reducing bicycle-motor vehicle crashes by more than half in Corvallis, Oregon, and by two-thirds in Eugene, Oregon (Ronkin, no date; City of Eugene, 1980).

And moving on to Corvallis:

"the City of Corvallis, OR installed 13 mi (21 km) of bike lanes. A newspaper story in 1982 reported that bicycle crashes fell from 40 in the year prior to the bike lane program to just 16 in the 12 months afterward, and of the 5 crashes that occurred on streets with bike lanes, all involved bicyclists riding at night with no lights."

I have investigated the Corvallis information, which is incorrectly described in the Synthesis as a "study." What the Corvallis experience shows is dramatically different from what the Synthesis reports it as showing. I called Steve Rogers, of the City of Corvallis Public Works department, and he was kind enough to write a letter describing the facilities in Corvallis. He sent me the letter along with a copy of the City of Corvallis bicycle map and of the article by Michael Ronkin in the Bicycle Federation newsletter, Pro Bike News, cited in the Synthesis. I have posted Mr. Rogers's letter and enclosures for reference. Now, consider the following:

  • Mr. Rogers informed me that there are 175 miles of city streets in Corvallis maintained by the city and an additional 20 miles maintained by the State as state highways.
  • The installation of the bicycle lanes was one part of an extended public process involving public meetings and newspaper coverage about bicycling, as well as educational and law enforcement efforts.
  • The bike lane streets replaced a system of designated bicycle routes on parallel secondary streets. No data is presented in the Synthesis about the crash rate on those routes.
  • No information is provided in the Synthesis or the newspaper article as to the source of the crash numbers. It is most likely from police reports. There is no description here of any other source of data or of any checking for statistical validity.
  • Typically, only a small percentage of bicycle crashes is reported to the police, except when they involve motor vehicles.
  • Crashes which involve motor vehicles are typically 10 to 20 percent of the total of all bicycle crashes. 
  • Improvement in bicyclist and motorist behavior, in particular, greater obedience to the traffic law, has been shown to result in drastic reductions in crash rates.

Now, let’s do a little grade-school math. Corvallis had 195 miles of streets, of which 13 miles had bike lanes. Even if no crashes at all occurred on the 13 miles of bike lane streets, an crash reduction to 40% of the total on the year before the installation could occur only if 60 percent of the crashes in the year before the installation had occurred in the same corridor.

But that's not possible, because 5 of the 16 reported crashes in the year following the installation of bike lanes occurred on the bike lane streets! That’s 31 percent of the crashes on 7 percent of the city’s street mileage. The reported crash rate per street mile on the bike lane streets the year after the bike lane installation was 5.7 times as high as on the other streets in Corvallis that year, and 1.7 times as high as on all streets in Corvallis in the year before the bike lanes were installed.

The sources cited provide no information about the rate of bicycle use. This information would be needed to support any statement about the safety of the bike lane streets relative to non bike lane streets. What can be said in the absence of this data is that an alarming number of crashes occurred on streets with bike lanes, and a reduction in crashes occurred citywide.

The most obvious hypothesis to explain the dramatic reduction in reported crashes in Corvallis is that the political process that went along with the installation of the bicycle lanes led to an improvement in bicyclist and motorist behavior throughout the city.

The Synthesis describes the five reported crashes in the bicycle lanes as all due to failure to use lights at night. This explanation does not explain the crashes away, or demonstrate that the bike lanes are safe. Rather, if anything, it suggests that bike lanes may foster an illusion of safety for bicyclists who are not using lights. Unfortunately, no further details are given in any of the cited sources as to how these crashes occurred.

The Corvallis example has been cited as a "study" and the conclusion that it indicates a major safety increase due to bike lanes had been repeated in subsequent literature. The same quote from the same Federal Highway Administration publication cited above makes this claim:

Two other studies credited BLs with reducing bicycle-motor vehicle crashes by more than half in Corvallis, Oregon, and by two-thirds in Eugene, Oregon (Ronkin, no date; City of Eugene, 1980).

"The Johnson Street bike lane ..."

The study cited in this example, described in more detail by Alan Wachtel in his review, shows an increase in crash rates. The only claim for the bike lane is that the crash rate there did not go up as much as it did elsewhere, though it did go up substantially more in the first year after the lane was installed. The Synthesis mentions two specific types of crash responsible for the increase, without saying what they were. The crash rate for the entire city went up; this suggests that education and law enforcement would have been a better way to spend city funds to increase safety than the bike lane.

"Even when university towns are excluded from consideration, cities with higher levels of bicycle commuting have on average 70 percent more bikeways per roadway mile and six times more bike lanes per arterial mile."

The statement is intended to support the idea that bike lanes encourage bicycling, and that encouraging "potential bicyclists" is the central goal of planning for bicycling. But are the bike lanes built because bicycling is popular, or is it the other way around? And if the bike lanes encourage bicycling, are their other approaches which might do better?

"More dramatic claims of the safety impact of bike lanes are made in Denmark"

Notice that the comparison here is with "cycle tracks," sidewalk-type bicycle facilities: not with bike lanes as that term is understood in the US. US and European research has shown sidewalk-type facilities to be extremely hazardous, so it is no wonder that bicycle lanes look good in comparison. Alan Wachtel also has discussed this statement in his review.

Is there any good data on bike lane safety? The Wachtel and Lewiston study published in the September, 1994 ITE Journal is carefully done, and finds no significant difference in crash rates between streets in Palo Alto, California with and without bicycle lanes. It is not cited in the Synthesis. Alan Wachtel has more to say about the absence of a citation in his review.

Bicycle Paths

This section describes the problems with bicycle paths quite well. It is intriguing that paths were described and promoted in the 1970s as having the same effect in encouraging bicycle use as with bike lanes in the Synthesis. Both types of facilities do create the perception of safety for novice riders. If this perception is not matched by reality, then serious safety issues and ethical issues arise. These issues are now widely acknowledged in the case of bicycle paths.

Shared lanes

It is important to note that the majority of bicycling almost everywhere, not just in residential neighborhoods and rural areas, occurs in this type of facility.

This section of the Synthesis puts forward formulas for traffic volume, but fails to raise the question of crash types and numbers, which depend heavily on traffic speed. This section also ignores all hazards not related to motor traffic, hazards which account for the great majority of bicycle crashes.

Bicycle and Bus Lanes, Bicycle Boulevards

These two relatively uncommon facility types support bicycle use according to relatively normal traffic principles. Further information about bicycle and bus lanes and bicycle boulevards is available on this site.

Conclusions

The Synthesis places a political agenda ahead of scientific veracity. The fundamental conclusion that the separation vs. integration debate has become "redundant and futile" is one of wishful thinking. The separation vs. integration debate has not become "redundant and futile" and will not, as long as those who advocate and promote separate facilities do so for the wrong reasons, with the wrong expectations, and with too little knowledge of or concern for issues raised by the research literature.


[Top: John S. Allen's Home Page]
[Up: Research]
[Previous:]
[Next:]

[contact John S. Allen by e-mail]

Contents © 1997 John S. Allen

Last revised 8 June 2009