Eric Vallabh Minikel   ·   11.522 Research Seminar on UIS   ·   2009-11-03

Research Update

 

→ Research area:

Connecting bicycle facility design (including route choice), accident rates and accident severity to form a public policy recommendation about how to best accommodate bicyclists.

 

→Background:

 A reasonable goal for public policy is minimizing the risk of serious accidents (accidents leading to serious injury or death) that cyclists face.  This can be decomposed into two parts:

serious accidents / cycle trip = ( serious accidents / accidents ) * ( accidents / cycle trip )

A wealth of literature examines the effect of various design interventions on the second term, (accidents / cycle trip), sometimes with counter-intuitive results.  Jensen (2007) finds that cycle tracks increase accidents per cycle trip, and that bike lanes have no significant effect, compared with no bicycle facility at all.  And yet as Pucher (2000) points out, all of this is hard to reconcile with the fact that Germany and the Netherlands, which have extensive bicycle facilities, have far lower fatality rates per cycle trip than the U.S. which has almost no such facilities.

One possible explanation lies in the notion of causative safety in numbers: when more people cycle, as in Germany or the Netherlands, traffic is slowed and/or drivers become more aware of cyclists, leading to fewer accidents.  Yet while the phenomenon of correlative safety in numbers has been widely documented, for example by Jacobsen (2003), no studies have established direction of causation, and one could just as easily imagine that the reason more people cycle is because they correctly perceive it to be safer.

Since most cyclist-motor vehicle accidents are minor and go unreported, another possible explanation is that facilities such as cycle tracks draw more timid cyclists onto the road, who are more likely to report a minor accident than seasoned cyclists are, thus increasing the number of reported accidents without increasing the underlying number of accidents.

Another possible explanation is that bicycle facilities do increase either the rate of accidents or the proportion which are reported, but decrease the rate of serious accidents.  Kim et al (2006) examine factors which predict a greater severity of injury, given that an accident occurred—in essence, the first term in the equation at top.  They find, for example, that high speeds and heavy trucks increase the likelihood than an accident will be severe.

Yet little research has been done examining both terms simultaneously.  For example, if Kim et al are correct, one would expect a lower rate of serious accidents per cycle trip on quiet side streets than on major thoroughfares.

 

→Research direction:

Berkeley, California accommodates cyclists primarily on bicycle boulevards—quiet side streets optimized for bicycle use, with little motorized traffic and almost no heavy vehicles—while Cambridge, Massachusetts mostly offers bike lanes on major streets with high motor vehicle volumes and a preponderance of heavy trucks and public buses.  One would expect a rather different profile of accident types in these two places, as well as differing rates of severe accidents.  If it’s true that facilities draw out more timid cyclists who are more likely to report minor accidents, then one would also expect that in both places, over time as facilities have been completed, the proportion of reported accidents which are severe should decrease.

In most bicycle safety research, exposure data is a serious constraint: it is hard to quantify rates per cycle trip, since no one knows how many cycle trips take place.  Cycle count data for particular streets are quite thin.  However, one bright spot is that comparing entire towns to one another makes it possible to experiment with using journey-to-work mode choice data from the U.S. Census, in other words, number of bicycle commuters, as a proxy for the number of all cycle trips.

My intent is to examine the data from both places, looking for any similarities, differences and trends which may help explain the contradictory results found in literature to date.

I have data from Cambridge, MA and from all of Alameda County, California (an area much larger and more populous than just Berkeley) and from the Denver area, Colorado and fatality-only data from Montreal, Quebec.

 

 →Selected bibliography:

Pucher, John and Dijkstra, Lewis.  “Making Walking and Cycling Safer: Lessons from Europe.”  February 2000.

Jensen, Soren Underlien.  Bicycle Tracks and Lanes: a Before-After Study.  TRB 2007

Jacobsen, P.L.  “Safety in numbers: more walkers and bicyclists, safer walking and bicycling.”  Injury Prevention 9 (2003).  pp 205-209.

Kim et al. 2006. “Bicyclist injury severities in bicycle–motor vehicle accidents.” Accident Analysis and Prevention.

Turner, S.A., Roozenburg, A.P., and Francis, T. 2006.  “Predicting accident rates for cyclists and pedestrians.”  Land Transport New Zealand Research Report 289.

Smith, Robert L. and Walsh, Thomas. 1988.  “Safety Impacts of Bicycle Lanes.”  Transportation Research Record 1168.  pp 49-56.

Robinson, Dorothy L. 2005. “Safety in numbers in Australia: more walkers and bicyclists, safer walking and cycling.” Health Promotion Journal of Australia 16 (1).

Allen-Munley, C., Daniel, J. and Dhar, S. 2003. “Urban bicycle route safety rating logistic model.”  Transportation Research Board 2004.

 

Back to discussion list.

Back to 11.522 home page.