In 1975, an engineering graduate student named Jerrold A. Kaplan produced a thesis entitled "Characteristics of the Regular Adult Bicycle User" in partial fulfillment of his MS degree at the University of Maryland. Over the years, this report has been used by others to support a variety of contentions, the most common being that riding on separate bicycle trails is "2.78 times as dangerous as riding on the roads." Generally, the people making this claim have never seen the Kaplan study and rely on secondary sources. But the question is one that is important:
The short answer is that we don't really know. (I give a modest attempt at a slightly longer answer elsewhere.) And the Kaplan study actually does little to help us out, despite the claims of some advocates. It is worthwhile, then, to ask where does the "2.78 times as dangerous" claim come from and what, in fact, does it mean?
First, let's look at the study. According to the author, "the primary objective of this study [was] to determine the habits of the adult bicycle rider (16 or older), who uses his bicycle on a regular basis, in order to identify characteristics of the bicyclist and his trips" (p.5). To accomplish this task, he conducted a mail-out survey of members of the League of American Wheelmen (now called the League of American Bicyclists). In addition, he did a similar survey of the Washington Area Bicyclist Association for purposes of comparison to the national study.
In surveying the LAW members, Kaplan did NOT intend "to represent the typical American bicyclist of today. This would be a gross misrepresentation of the facts" (p.12). He surveyed LAW members because they represented the largest bicycle group he could reach efficiently; their membership had recently grown dramatically, and they likely included new cyclists, who were perhaps similar to the growing number of Americans who were at the time taking up the bicycle.
The survey Kaplan sent out consisted of 27 questions, which asked about the respondent's sex, age, city, state, zip code, and other personal information. It also asked, among other things, about the types of bicycles and equipment that people owned, their cycling experience and habits, how much riding they did, the types of roads (and other facilities) where they did their riding, and their accident experiences for the previous year.
Kaplan got a good response to his survey. Overall, roughly 52 percent of the 8405 surveys came back. He ended up using 3270, or 38.9 percent, for his analysis. Interestingly enough, although all 50 states were represented in the returns, states near Chicago were "over-represented in terms of each state's population" (p.19). This makes sense when you realize that the organization was then headquartered in Palatine, Illinois, where the then-editor of the LAW Bulletin lived. Whether this affected results in any important way is hard to say. But consider that Illinois accounted for 505 returns whereas California accounted for 313 (p.109). With two to three times as many people at the time and being a hotbed of cycling, California ought to have done better but that's what happens when you survey the membership of a particular organization.
Kaplan asked respondents how many miles they had ridden during 1974. The average reported was 2,332. As might be expected, respondents from fair-weather states like California reported more miles ridden than did those from colder climes. Kaplan also asked people how many trips per month and how many miles per month they rode for different trip purposes (e.g., work, school, shopping, recreation, racing). And for the one trip purpose for which they recorded the greatest number of trips per month, he asked what percentage of their mileage was ridden on four classes of facility (p.108):
[These descriptions, by the way, are presented just as they were presented in the survey. I am not paraphrasing them in any way.]
In total, respondents reported riding 34.9 percent of the miles for their most common bike trip on major streets, 58.4 percent on minor streets, but only 3.2 percent on on-street bikeways, and 3.5 percent on off-street bikeways (fig. 5).
Kaplan also asked people if they had had a "collision or serious fall" in the past year. Some 21.4 percent responded "yes." In analyzing the data by facility type, Kaplan divided the crashes into two categories: "All Accidents" (which included all four of his levels of severity from "bicycle damage only" to "major injury") and "Serious Accidents" (which included only those crashes where someone needed emergency treatment) (p.47).
He then divided the number of crashes reported on each facility type by the number of miles reported ridden on that type. The result was a rate per million miles. Among "all accidents," the rates he derived in this manner were as follows (Table 13):
For "serious accidents" he reported the following rates:
We can now see where the "2.78 times as dangerous" claim comes from. If you divide Kaplan's "all accidents" rate for off-street bicycle facilities by the rate for minor streets, you get 2.78. But what does this really mean? Interestingly enough, those who use this number so widely never claim that on-street bikeways are only 55 percent as dangerous as minor streets and 51 percent as dangerous as major roads. Overall, I believe that the numbers are pretty shakey anyway and none should be seriously used to support or oppose facility options.
One of the main problems with using this data to attack off-street bicycle facilities is the broad definitions Kaplan used. In his survey, an off-street bicycle facility can be a sidewalk bikeway adjacent to a busy arterial or it can be a bike path through an uninterupted greenway. Similarly, a minor street can be a quiet rural backroad or an urban residential street. An on-street bikeway can be a simple signed bike route, a striped bike lane, or a barrier-separated bike lane.
It is reasonable to assume that the various facilities lumped into these definitions might harbor different types of hazards, lead to different types of conflicts, attract different types of users, and have different levels of crash dangers.
Beyond the mix of separate trails and sidewalk bikeways, there is another issue that confounds our ability to come up with one generalized "accident rate" for trails. As someone who has made it his business to study trails throughout much of the U.S. and elsewhere, I know that some are designed, built, and maintained far more carefully than others. I also know that trail user mixes vary widely. And the presence or absence of geographic features like steep hills and mountains varies from place to place.
I've seen some trails that mix bike and dense pedestrian traffic on a surface five feet wide with sharp blind corners and hazardous edge conditions. And I've seen some park trails that are 16-feet wide, with centerline striping, generous sight distances, gentle curves, and smooth hazard-free shoulders. Logic would suggest such vastly different trails would have vastly different crash rates.
In response to a one-time mail survey regarding their past year's bicycling experiences, some people may well be able to say how many miles they average per month and what percentage of the monthly mileage for their most likely trip type occurs on different types of roads and paths. However, there is no "test" of this assumption. In other words, we have no way of knowing how accurate such statements may be. Quite likely, the accuracy varies depending on such factors as whether a person uses an odometer and keeps records of his or her rides. Undoubtedly, some people will generate very good estimates while others will produce wildly innacurate guesses but we have no way of knowing how this might have effected the results.
There are undoubtedly other factors that could confound the accuracy of the crash rates in the Kaplan study. Conceivably, some particularly hazardous trails in the heavily-represented Chicago area might have influenced the results. The fact that the study was done in the mid-70s, when U.S. trail design was in its infancy might have affected the numbers, particularly in terms of their validity today. The focus on the most common trip type might have done something as well, although it is difficult to suggest just what. Bicyclist age certainly could influence the types and circumstances of crashes experienced (the mean age was, after all, 37.7 years and people under 16 weren't counted).
Male riders represented about 88 percent of the respondents and, interestingly enough, the study showed a significantly higher crash rate for women; is it likely more women rode on trails? Hard to say. Also consider that those with many years of experience had quite low crash rates while newcomers had comparatively high rates. Is it likely that the less experienced--and more crash prone--riders tended to ride on separate facilities? On the other hand, is it possible that members of the LAW might ride in such a way as to exacerbate a trail's inherent problems? For example, could someone riding at a relatively high average speed be more likely to come to grief than someone riding slowly?
All in all, the Kaplan study gave an interesting look at the characteristics of members of a national bicycle group in the mid 1970s. That was, after all, the stated purpose. However, people pushing a particular agenda (anti-trail development) have used it to "prove" their point. A close look at the study, itself, does not truly support such a view.
Ultimately, the dangers of trails must be considered in light of such
factors as design, construction, maintenance, and use. The likelihood that
a generalized "accident rate" for facilities with significantly
different particulars could have much validity is very low. Clearly, any
assertion that trails are "2.78 times as dangerous as roads" should
be considered with at least several grains of salt.