the green box report

okay, so i went to this PBoT "bicycle brownbag" program yesterday, where these two PSU professors reported on their study of the so-called "bike boxes" that were installed at twelve intersections not quite two years ago.

jennifer dill directs the center for transportation studies at PSU, and christopher monsere is an assistant professor in civil engineering there, with a specialty in transportation. the study was funded in part by the city and in part by something called OTREC, which gets its money from the federal DOT.

the idea was to compare the behavior of bicyclists and motorists at these twelve intersections before and after the bike boxes were installed, to see whether the treatment actually reduces "conflicts."

nine of the twelve boxes are painted green, three are not, and they included two intersections in the study that do not have bike boxes, as a "control." they took a lot of video at these intersections -- over nine hundred hours --, and they paid some graduate students to go through a very small fraction of it, counting things like motorists (or cyclists) encroaching on the pedestrian crosswalks and so on. and looking for right hooks.

and before i get all snarky, let me say (1) i respect the effort, (2) these people have a lot of specialized training i do not have in how to design studies and analyze data, (3) etc.

there are deficiencies in the data, and dill and monsere readily admit this, and they readily acknowledge that they could not really draw any substantial conclusions from it, at least not yet. so it is unclear what the $72k in grant money has bought, apart from a lot of as yet unexamined video.

one threshold problem is that the "before" videos were taken january through march, while the "after" videos were taken april through june. obviously a lot more people on bikes in spring than in winter, and dill and monsere have attempted to correct for this through some weighting of the data, but what they really cannot adjust for with any accuracy is the (probable) fact that a higher percentage of people biking in winter know what the hell they are doing, while spring brings out the amateurs. just a thought.

another problem that could turn out to be a major difficulty in analyzing the data is that the videos do not show the signal phase, so in many cases the person who is tasked with coding "conflicts" cannot determine whether these are occurring as people are starting out of the light as it turns green or as they are rolling through on a light that has been green for awhile.

and a central problem, at least for me, is that the key metric, "conflicts," has not been defined. in response to a question i put to him, monsere confirmed that they just handed these videos to these grad students and said "flag anything that looks like a conflict to you," and then he and dill sifted through the segments the students had flagged and decided what was and what was not a "conflict."

and apparently the primary criterion was whether a motorist or a cyclist had had to brake suddenly or take evasive action. a lot of stuff the reviewers had identified as potential "conflicts" turned out, in monsere's view, to be a motorist and a cyclist "negotiating" who had priority.

anyway. a couple of not really "findings" but observations that dill and monsere have thus far derived from the limited data.

one: for the most part, cyclists and motorists seemed to understand what the boxes are "for" and how they are supposed to be used, though relatively few cyclists actually positioned themselves in the box proper, in front of through traffic -- and a fair number (myself included) simply asserted the travel lane, ignoring the box altogether.

two: there seemed to be a lower incidence after the boxes went in of motorists crossing into the bike lane ahead of the intersection, but a much greater tendency to cut the corner tight. to me, this says the boxes may actually be exacerbating that particular problem (if it is a problem), by positioning the motorist further back from the intersection and in effect inviting her to start the right turn from back there.

the box at southwest broadway and taylor seemed to present some special difficulty. this is the corner out in front of columbia sportswear. the "after" counts showed a lot more motorist encroachment on the pedestrian crosswalk. i am going to guess that this anomaly has something to do with the fact that the person reviewing the videos was unable to determine when the light was green, so you might see a motorist rolling forward to make a right and then stopping to allow the north-south crosswalk to clear.

but it could be something else. the right turn at that intersection is rather tight to begin with, as is the right turn two blocks earlier, at morrison. at both these intersections, i routinely pass on the left while a motorist is waiting for the crosswalk to clear. it is just possible that the only "solution" at these two locations (and probably a number of others) is simply to forbid motorists to turn right, period.

in any event. and i mentioned this to dill and monsere after the presentation. they need to define what exactly they are looking for in these videos, so that a third party (that's me) can look at their final report and say, okay, they identified the relevant stuff, or nope, they should have been looking for "x" instead.

and for me, the "x" they should be looking for is not just whether there were significantly fewer (or more) evasive maneuvers after the boxes were installed, but whether these mostly occurred during the green phase. which frankly seemed to be the case in the video clips they showed.

in particular, one clip shows a delivery truck starting into a right turn, presumably just after the light has changed, then hesitating as the driver sees a cyclist approaching in the green bike lane, but then the cyclist also hesitates, and the two of them go into a "you first," "no, you first" routine, with the cyclist finally pedaling through. this was presented as humor, but with the inference that this is actually how the box is "supposed to" work.

[note: in the same situation, i would have moved left (from a position already well into the travel lane), allowing me to pass on the left as the truck went into the turn. the advent of the box has somewhat complicated this rather ordinary maneuver for me, because the motorist will often go into the hesitation mode anyway. however, i am not really complaining, because hey, the motorist was forced to become aware of his surroundings.]

the intended function of the box is to move cyclists to the front during the red phase, taking them out of the right hook. but of course the box is still there during the green phase, with the striped lane leading in and a dashed lane leading out. so the box does nothing to discourage a cyclist from coming up alongside the line of stopped cars, arriving at the head of the queue just after the signal changes and the first car cuts the corner.

in other words, the right hook is to some extent an artifact of the far to the right dogma, here enforced by the striped bike lane itself directing cyclists to the right of the motor traffic, even where a right turn is permitted. if you must have a bike lane, the line should be dashed for the last hundred feet or so before the intersection, to allow motorists to merge right before making the turn and encourage cyclists to merge left, out of the right turn lane. and the far to the right and mandatory sidepath laws should be repealed.