Alan Wachtel’s comments on Chapter 4 of the FHWA's Bicycle Safety Related Research Synthesis, chapter 4

The Bicycle Safety Related Research Synthesis is a publication of the Federal Highway Administration, prepared under contract by Andy Clarke and Linda Tracy, employees of the Bicycle Federation of America. The Synthesis is very pro-bike lane, and this stance is most evident in its Chapter 4. The following comments are from an e-mail posting by Alan Wachtel, dated February 14, 1996. The headings and comments in brackets have been added for clarity and are not part of the original message.

First, a question which John Allen asked Alan Wachtel, who is a researcher in California and has done studies of bicycle use and accidents.

<< I read in a recent message from John Forester that you have written comments on the FHWA Bicycle Research Synthesis document….I am particularly interested in what you can tell me about your September, 1994 ITE [Institute for Transportation Engineers] Journal publication's not having been cited in the FHWA document. [This was a statistical study of bicycle accidents in Palo Alto. It is now available online]. Two things are clear to me: your study does not support the FHWA's policy, and your study was published before the FHWA document. Whether there was time for Clarke and Tracy to include your study, however, I do not know. >>

 Now, Wachtel’s reply:

My study doesn't support bikeways on sidewalks or paths; it really doesn't say anything about lanes. [The study found an accident rate for sidewalk riding 1.8 times as high as for riding in the street. It looked into the safety of bike lanes, but found no significant difference in bicycle accident rates between arterial streets with and without bike lanes.] There should have been plenty of time for Clarke and Tracy to include it, particularly as they cited my March 1995 signal timing paper in preprint form, and they also had a preprint of the September 1994 paper.

On January 23, John Allen asked our opinion of the [FHWA document]. The version I have is a draft labeled 1/19/94, but if the authors, Andy Clarke and Linda Tracy of the Bicycle Federation, have finally managed to produce a finished product, I don't suppose it's very different from the draft.

John was concerned that the Cambridge [Massachusetts] bicycle coordinator is using Chapter 4 of this publication, "Channelization and Bicycle Facilities," as her defense of bikeways. I will therefore concentrate on this chapter (which is the only one I've read). It's a curious mixture of evenhandedness and bias, of genuine summary of research ("synthesis" goes too far) and verbose bureaucratic padding. As I see it, the authors' argument goes like this:

A lot of important bicycle research went on in the 1970s, but for the most part ceased abruptly about 1980. (This is true, except for the part about the '70s.) Evidence collected then showed that bike paths and bike lanes were hazardous, but those conclusions were based on poorly planned and designed facilities. We have much better design standards now, and if those standards are followed, any type of facility is acceptably safe.

The question, Clarke and Tracy argue, then becomes one not of facility design but of facility selection. If we want to market bicycling to new or casual riders, we have to build the kind of facilities that those riders say they want (i.e., paths and lanes). The chapter's conclusions rely heavily on Wilkinson's 1992 FHWA report "The Effects of Bicycle Accommodations on Bicycle/Motor Vehicle Safety and Traffic Operations," which is described as a technical supplement to the justly maligned "Selecting Roadway Design Treatments to Accommodate Bicycles."

Although Forester is treated with wary respect and described as "highly influential," his views are misrepresented. His opposition to separate bicycle facilities is said to be based on a single source (Kaplan's study). He is said to "grudgingly" accept bicycling on multi-use trails (which the authors support) at slow speeds, but his acknowledgment that paths are also useful as shortcuts and for recreation is ignored, presumably because it makes his opposition look less flexible.

(The publication uses the undefined term "trail" synonymously with "path," a habit that I find irritating, because the carefree rural connotations of trails are inappropriate for urban paths.)

Clarke and Tracy recognize correctly that "trails" are appropriate only as multi-use facilities, where there is little conflict with cross traffic, and where they are properly designed. They mention that their use should not be compulsory and that they do not replace the highway system. They fail to notice that few bike paths can meet these criteria, but they tend to get built anyway. They also say that trails are good places for novice riders and children to learn the skills and techniques that are necessary when riding in traffic, and I can't imagine where they got that idea.

Turning now to bike lanes, the authors' discussion identifies two concerns expressed by opponents: bike lanes discourage the sweeping effects of cars, and they encourage improper left turns by bicyclists and improper right turns by motorists. I think it's misleading to put these on an equal footing, and especially to put sweeping first. In defense of bike lanes they cite Michael Ronkin, who says:

• The difficulties associated with making a left turn in traffic have nothing to do with bike lanes.
• Right-turn conflicts are dealt with by dashing the bike lane stripe before intersections.
• It is better to schedule regular sweeping than to depend on auto traffic to keep the road clean.

I think highly of Michael Ronkin, but I don't find these arguments convincing. Bike lanes do encourage inexperienced bicyclists--their target market--to initiate left turns from the right side of the street. They also encourage motorists to believe that any other behavior is improper. The dashed stripe, in my experience, has disappointingly little effect, even in California, where right-turning motorists are not only permitted but required by law to merge into the bike lane. Regular scheduled sweeping is certainly preferable, but it often occurs only at long intervals, and in some places it is never done at all.

The authors cite five studies in support of bike lanes, and none in opposition. Lefler (1975) reported that bicycle lanes would have prevented approximately 14 percent of crashes in Santa Barbara. I have no idea how he could have determined such a thing. The introduction of bicycle lanes in Eugene, Oregon, in 1979 resulted in a substantial reduction in the bicycle accident rate. But the statistical significance of this reduction is not reported, and the motor vehicle crash rate also fell substantially, suggesting that other causes were at work. A year later in Corvallis, the introduction of bike lanes was followed by a drop in bicycle crashes from 40 in the prior year to 16 afterward. Even assuming that ridership stayed constant, this information is useless without tests for statistical significance, and the conclusion that the bike lanes were responsible requires analysis of the nature and location of the crashes.

Smith and Walsh (1988) examined the effect of installing bicycle lanes on a 1.3-mile section of a one-way arterial pair in Madison, Wisconsin. Comparing four years before the installation to four years afterward, Smith and Walsh found that the total number of bicycle accidents citywide increased by 16 percent (significant at the 0.05 level), while the bicycle lane corridor accidents did not increase significantly compared with the citywide increase.

But a table is presented from this study, and in fact corridor accidents increased by 24 percent! It's simply that the difference between 24 percent in the corridor and 16 percent citywide is not significant at the 0.05 level. Thus the null hypothesis (that bike lanes don't change the accident rate) is not disproved. But it is certainly not proved either. And the numbers do not suggest that bike lanes improve safety.

Finally, a Danish study reports that even cycle lanes of half a meter width reduce the risk by up to a third, whereas lanes of 0.6 meters reach risk reductions of 70-80 percent. But the comparison is with cycle tracks (paths), not with roadways of equivalent width without a bike lane stripe.

Clarke and Tracy call throughout the chapter for more research, and on this point I heartily agree. The research base is incredibly thin and flimsy. We are forced to argue from experience, observation, and deductive reasoning, which are no substitute for experiment. But the meager "research" that Clarke and Tracy offer is much too weak to support their conclusions.

Alan Wachtel

wachtel@aol.com