This is in response to a link being passed around by email among Portland bike advocates and planners recently to a news feature about a 2010 study from the Center for Transportation Research in Austin, TX, about installation at that time of experimental "bike boxes". On receiving the link, I obtained a copy of the study and reviewed it, as well as other critiques of it. These are my thoughts.
First, some background reading
My general objections to bike boxes
My own thoughts on the Austin study
The executive summary defines safety "along the following lines: (1) The bicyclist used the bicycle lane to approach the intersection, (2) the bicyclist used the bicycle box after installation, (3) motorists did not encroach on the stop line or bicycle box, (4) the bicyclist departed the intersection before the motorist and (5) the bicyclist did not make an illegal movement, such as running a red light."
Definitions 1 & 2 are assuming the conclusion that proper use of the bike box leads to greater safety. So does the statement a few paragraphs below that "the proportion of bicyclists who used the bicycle lane to approach the intersection instead of riding in the full lane increased substantially". Those definitions are not studying safety, they are merely studying adherance to the bike box.
The stated purpose of bike boxes was given as "reducing the risk of righthook collisions by moving bicyclists in front of motorists waiting at a red light, instead of waiting to the right of motorists." Notice that the safety benefit of the bike box is compared to that of waiting to the right of a line of cars that could turn right, but NOT compared to a bicyclist waiting in line in the travel lane, which provides the same visibility advantage to the following motorist as the bike box does to the front motorist. Yet in several places, the report mentions an increase in bicyclists using the bike lane COMPARED TO the travel lane, not compared to riding alongside the travel lane. Which is it? However, if many bicyclists are still simply waiting in the bike lane rather than entering the box area (the study indicates only 15-25% entered the box), this doesn't seem very successful, since most cyclists are still "waiting to the right of motorists", the behavior the bike box sought to avoid.
Also note that, simultaneous with the bike box installation, right turn on red was prohibited, and this was mentioned several times as being key to the success of the bike box. Yet, just prohibiting right turn on red alone, without installing a bike box, would (one could hypothesize) also help reduce the right hook risk collision. Making both changes at once compromises the conclusion that the bike box is necessary for this reduction. In any case, the study also observed high rates of illegal right turn on red by motorists, 79% at one location. Interestingly, motorist adherence to "no right on red" was not included in the study's definition of safety, despite decrease of the right hook hazard being a stated goal of the bike box.
So, the rate of bicyclists using the bike box as designed were low, 15-25%. Rates of motorists illegally turning right on red were high, 79% at one location. Rates of motorist encroachment at one location were on average 50%. This is success?
Yet the study concludes "The bicycle box markings ... have proven to be an effective method of improving safety of bicyclists and motorists at intersections...." But they have only studied behavior, and even that was not an overwhelming success. No actual crash data was reviewed. In assuming that using the bike box properly provides safety, they are assuming their conclusion, and only really studying how to make sure people use the device properly, not its actual safety.
The study "recommends that bicycle boxes be installed at intersections where the majority of motorists do not turn right during a red phase and the volume of bicyclists is high. Of the intersections that meet this condition, priority should be given to intersections where the bicycle lane does not continue immediately after the intersection and to intersections where a safety hazard (e.g., a history of right-hook collisions) is evident."
Mixed results indeed.
Edited by John Brooking on Apr 18, 2012 10:18 AM