Continuing to dig into the granularity on cycling deaths and injuries in Los Angeles, I've developed a bit of python using geopandas to determine which districts are most dangerous. Note that this data is not normalized to the population per district. Districts have a pretty consistent population size, varying by about 10%, and effect sizes between districts are large enough to explore the data without normalizing. Again I'm looking at all bicycle accident reports from 2015-2018 and vehicle accident reports from 2018 only. All accidents, according to the datasets referenced yesterday, are labeled with collision severities of 1 to 4, where 1 = fatal; 2 = severe; 3 = moderate; 4= minor. By looking at type 3 versus type 4 bike accidents per district, we can make some interesting inferences. First off, there is quite a bit of variability in the absolute number of accidents by district. That's certainly an angle we can dig into a bit. It's also a result that could change when (and if) we can look at ridership data instead of absolute counts of accidents. There may be far more riders in district 5, which would skew these data even if the per-rider or per-mile risk is not high. Secondly, if we look at the proportion of accident severities as a function of vehicle type, we observe a skew away from minor accidents on bikes: There are a couple of likely factors to explain this skew. One of them is that bike accidents are generally considered less likely to be 'minor'. Indeed, bike and pedestrian-involved accidents account for around 10% of LA collisions but almost 50% of fatalities. But I think we're also seeing something else in the data, and it is probably important: underreporting. Remember that this data is from police reports, and so if there's no report, the incident doesn't appear in the data. Why do I suspect underreporting? By some accounts, ~50% of minor cycling accidents do not wind up on a police report. There are plenty of studies on this phenomenon. Some reasons: (a) bike vs car accidents that only involve property damage to the bike (or minor injuries) might be settled informally. I once went over a motorist's windshield but escaped uninjured. He promptly brought my bike and I to a bike shop and covered damages directly. Such an exchange would be difficult with vehicles. (b) some bike accidents don't involve motorists and therefore are less likely to have an insurance trigger. Most of these will be minor, because major accidents are more likely to require police/EMT support. In contrast, single-vehicle accidents are more likely to be reported even without injury. (c) rates of hit-and-run may increase for cyclists/pedestrians vs motorists. Why? Motorists might not realize they've hit someone. Alternatively, given that only ~1% of hit-and-runs are successfully prosecuted, many drivers may opt to leave the scene of an accident if they don't think they're going to get caught. On a bike or on foot, your chances of catching up with a driver are next to zero. That's probably part of why 80% of fatal bike/pedestrian incidents are hit-and-runs. (d) some cyclists who try to file police reports are unable to do so. This sounds strange--and it should sound strange. But if you speak to LA cyclists, you'll find that it is not uncommon to request--and not get--police support. It is true that LAPD resources are stretched thin. But that doesn't resonate when, for example, a task force is assembled to deal with scooters. I'm not saying that scooters don't annoy me. But public resources are better directed towards public safety, and if I had to choose between being hit by a scooter running a red light and a truck running a red light, I would choose the former. And so I would rather direct enforcement efforts towards the latter. All of these factors would contribute to underreporting. Just to drive this home, let's look at the 'minor' versus 'moderate' accidents in cars, by district. As you can see, there are roughly double the number of 'minor' accidents versus 'moderate' accidents in cars. I.e., these data look the way you'd expect them to: minor accidents are more common. Why should we care if minor bike accidents are underreported? A couple reasons. Here's one, by way of analogy: people in certain communities (lower S.E.S., immigration status, etc.) are less likely to report even very serious crime. This can drive a deflation in incidence statistics. And that deflation ultimately results in a lack of mitigating enforcement, amplifying the risks faced by certain communities. Robust statistics (can) help cities, police departments, etc., make good decisions in how resources are allotted. They can also identify inequities in safety/service that have economic causes. Here's another reason: minor accidents can precede larger ones. You might call the city to report a pot hole on your commute that gave your car a flat tire. And they'll fill it (well, sometimes), because flat tires come with an economic and labor burden for city residents, and also because a minor accident can easily become a major one. This is the case for cyclists as well. A trouble area that accumulates many minor accidents may also experience major ones, and that can result in significant challenges to public safety. It can also get the city sued: for example, buckled and fragmented pavement (on Griffith Park Blvd, eg) has given me more than one flat. And it has also seriously injured/killed cyclists, resulting in multimillion-dollar payouts by the city. In these cases, minor accidents--probably with cheap fixes--likely predicted larger ones. The lesson--and the hope for the future--is that the power in data might be more fully realized, in order to make our streets safer. It's a daunting task. But consider an adjacent issue: recovering stolen bikes is not an enforcement priority for most police departments. Bikes are not expensive enough, they're mobile, and generally bike owners haven't held onto any way to prove ownership of them. In some cases, local PD just won't act, even with documentation (speaking from personal experience here). But the National Bike Registry project has been changing that, by creating a paper trail around bikes, and by acting as liaisons. In the process, they've subtracted reasons for non-enforcement. They've even helped some bikes to get home safely. And at the end of the day, getting home safely is all that anyone is really asking for. Comments are closed.
|