The first litmus test that I apply to these crash studies, kind of like Van Halen's infamous M-and-M contract rider, is this: How does the study break down the crash stats by the age of the bicyclist, if at all? After all, that info is available in the police reports, and it’s critical to any crash study's usefulness. Well, the Boulder study is better than most but still doesn't use this age information in any meaningful way. Its authors tell us only how the overall bike-car crash count breaks down by age, a metric which doesn’t say anything important to them, as it seems to jibe at first glance with the overall demographic of the city. The studiers don't go any deeper into the age information, failing to cross-tabulate age with fault or type of wreck. This is like offering someone a delicious-looking donut, but only allowing them to nibble a few sprinkles off the top. Incredibly frustrating, and useless. But typical. It fails the litmus test.
Anyway, here's what else I can get out of it.
-- Bikes and pedestrians do not seem to be seriously overrepresented in the collision numbers. There is no epidemic of bike-car crashes in Boulder, and certainly no epidemic of bike-pedestrian crashes, with only 7 reported in almost four years (2008-partial 2011). That's impressive considering the amount of pedestrian traffic that mixes with bikes in several places around town. The number of reported car-bike crashes (516) is substantial but not crazy huge. It comes down to about 12 reported crashes per month over the study period.
-- 56% of car-bike crashes occurred in or at a crosswalk! Here's the first flapping red flag in the Boulder numbers. This high percentage points to a lot of sidewalk riding -- or, in Boulder's case, sidepath riding (the path along Broadway is classic sidepath with a few classic associated intersection conflicts). The report doesn't break it down but I have to wonder, with my cap off to the, you know, new skeptics of the new-school infrastructure -- how many of the incidents which were blamed on motorists involved a bike coming into the road from a sidepath or sidewalk? Some quick math -- 56% of 516 is 289 (crosswalk-related crashes). 28% of 516 is 145 (total cited cyclists). So there were probably something like 200 crosswalk crashes for which the bicyclist did not receive a citation (even if every one of the cited riders had been in a crosswalk, there would still be 144 non-cited crosswalkers left over). That's roughly 40% of all reported crashes which were sidewalk- or sidepath-related, and which were not blamed on the cyclist. Yow! Ming! So on the issue of fault or responsibility this study becomes very muddy indeed. It could support an argument in favor of "vehicular cycling" despite the fact that drivers were found liable far more often than bicyclists. Drivers were cited almost twice as often as bicyclists, and only in 28% of all reported bike-car crashes did the cyclist receive a citation, even with the low average age and (no doubt) experience level of the overall Boulder cycling population. But all that goes up in the air if a huge percentage of the bicyclists are coming off the sidewalk, many of them no doubt against traffic.
Let me show how the lack of meaningful age info could affect this report. It appears that about 10% of reported crashes involved kids under 16. Well, we know kids under 16 who are involved in car-bike crashes are very likely to be clearly "at fault" (~85%), probably riding blindly into the road from a driveway or sidewalk, a pattern that is starkly different from that seen in adults. That means about 52 of the reported wrecks involved young kids, and in all probability at least 40 or so of those were the kid's fault. So, if we remove all the kids' wrecks from the picture here is what we might get: 464 total crashes; 105 bicyclist at-fault. The share of bicyclists' at-fault collisions declines from 28% to about 22%. Even with all those college kids still included. So we see that removing the younger kids' wrecks from the study could have a significant effect on the overall picture, even though there just aren’t that many younger kids riding bikes any more. (We have removed 5-10 % of the riders in the study to achieve a 20% drop in the share of rider at-fault collisions.) We still need to refrain from making conclusions about or recommendations for adult riders based on studies in which the wrecks of little kids are quietly but insidiously tucked away, like some unholy statistical turducken.
-- Ah, but wait a second, 30% or so of incidents did not result in any citation. It's not clear if there is any trend or bias apparent there. Failure to cite can benefit the cyclist or the driver. But that is a large dark pool of unknown. It’s possible that many of the 40 or so of the kids' at-fault wrecks are in this category. If the kids' wrecks are tucked away over there in "no citation" land, despite their culpability, removing the kids from the study wouldn’t really have much effect at all, raising the percentage of all cyclist-cited crashes a little bit, and raising the percentage of driver-cited collisions too. In any case it is clear that looking at citations alone is not going to teach us the profound things we want to understand about the crashes in Boulder.
-- The vast majority of reported car-bike crashes occurred at intersections as we would expect. Of the rest, half occurred in a bike lane and half in a "travel lane." So no grist for the bike lane haters or lovers there.
-- The most common car-bike crash types included right hook, left cross and the various pull-outs, into bicyclists who were deemed by police to be operating legally. Pretty consistent with patterns all over the US and Canada. It seems to be the classic looked-but-failed-to-see pattern. Again, however, since a large chunk of these in the Boulder report involve sidewalk or sidepath riding these results are very fogged up. More work is needed, to sort the sidepath/-walk wrecks by fault, type and direction of travel.
-- Here's another doozy. Only about 4% of car-bike collisions resulted from a bicyclist blowing a stop sign or red light, despite the widespread lawbreaking among all kinds of bicyclists in Boulder. (That's about 20 wrecks out of 516.) No further study needed there. That's another torpedo to the myth that scofflaw bicyclists are responsible for most, or many, or a significant number, or whatever you want to say, of collisions with cars, due to their own reckless disregard for stops signs and red lights.
Even among bicyclists, the conventional wisdom has held that light-running scorchers are very, very irresponsible; they cause great injury and damage not only to themselves, but somehow to other road users, and occasionally to innocent housewives in the Ukraine. This happens through the insidious mechanism of chain reaction. One skippy little light-runner on a Chinese fixed gear thinks he got through the red light without incident, but lo, blocks away, tanker trucks explode and babies shoot into the air.
Ah, come on now. The truth is out. While there may be widespread disregard for the law, it's far from reckless disregard, or the numbers would be entirely different. Let's get over this and move on: Bicyclists' running lights and stop signs results in very few collisions relative to the number of violations. It's just not a super dangerous form of scofflaw-ism. You're more likely to get hit rolling through a green light than running a red one. We could argue about why that is so, but we can't argue any more about it being so. Perhaps we see these numbers (not just in Boulder mind you, but all over) because the light-runner is generally eyes-up, watching out, head in the game at that moment of his or her egregious anti-social act. Collisions occur during a different type of cycling altogether.
I'm just throwing that out there as a possibility, something to consider. The statistical safety of this scofflaw behavior is not a possibility, however, not a theory. It is real. We need to grow up and deal with it. If light- and stop sign-running is primarily a PR issue for bicyclists, rather than a safety issue, it would be good to somehow get our minds to that place where we are finally working together to bring the Idaho Stop to communities like Boulder, if not everywhere. Legalize it, and I will advertise it. After all, they're already doing it in Paris, and everybody loves Paris. So far I haven't heard any persuasive arguments to the contrary.
I should point out, for those who don't know, that the Idaho Stop does not legalize blatant reckless scofflaw riding. Violations of right-of-way would still be illegal, and, in theory, could be enforced much more stridently than they are currently, as the situation would finally be clarified for law enforcement. As it is, many police officers don't feel any need to crack down on the harmless violations, so they end up letting more heinous violations go as well. If you could pick one word to describe police enforcement of bicycling violations today that word would be arbitrary. And that's not good. Boulder is classic that way. Idaho Stop. Do it.
-- Back to the kid thing. It appears (hard to tell exactly) from the single graph in the report which addresses the age of bicyclists that roughly 10% of reported car-bike collisions involved kids under 16. This seemingly small number is likely a massive overrepresentation anyway, in terms of total hours or miles ridden, or even in terms of the number of bicyclists in the city. I contend that percentage would have been significantly larger in past decades. This 10% is consistent with the ongoing loss of child cycling in the United States. (Although Boulder is a special case, pod-city, university town biosphere type thing.) I don't have the Boulder numbers from the 1980s or the 1970s but it would be very interesting to compare. Adult bicycling was huge in the past, too, maybe even bigger at times than it is now, but a lot more little kids rode bikes in their neighborhoods. Further research should endeavor to show not only the ages of those involved in car-bike crashes, but the types of crashes that are most prevalent in each age group (a striking pattern), the pattern of fault in each age group (which will also be striking as heck), and any demographic trends in the city's cycling population over the decades (likewise, striking). Failing that, take five minutes to cull the kids' wrecks out of the study so they can’t contaminate the whole thing. Otherwise, you're really just pissin' up a rope.
Let's sum this baby up. Boulder is still a nice pod-city to peruse on a bike. But bicyclists there aren't any more visible than bicyclists anywhere else, and have to stay just as aware to avoid typical looked-but-failed-to-see errors. It's the same way it was 20 years ago: Lapsing into semi-consciousness while rolling into University Boulevard on the sidepath, or sprinting for a yellow light at 28th and Canyon without accounting for potential desperate left-turners, still leads to an all-you-can-eat body-panel buffet.