I went to Brighton Town Hall (2300 Elmwood Ave.) for the Let's Talk Cycling discussion. It turned out to be the featured lecture at The Rochester Regional Group of the Sierra Club meeting that night. After some Sierra Club business, Jean M. Triest was introduced. She's a Traffic Safety Specialist at The Monroe County Department of Public Safety. As a reasonably well-seasoned cyclist, her talk was a bit on the basic side for me — bikes need brakes, reflectors, a bell, and a headlight and taillight for night; bicyclists are recognized as legal drivers; ride in the same direction as traffic; use hand signals; obey traffic-control devices; be visible; be predictable; etc.
She cited a study from The League of American Bicyclists that examined the causes of bicycle accidents. She started out with some myths and the first was surprising: "traffic passing from the rear is the biggest risk to a bicyclist" is a myth. In a chart she showed me afterward, when comparing types of accidents with bicyclists, getting hit from the rear is the least likely kind of accident. The most likely cause of accidents — 25% — was riding in the wrong direction. Poor lighting is another problem cited (and cited separately from rear-end collisions).
I was getting kind of jaded about the whole thing. For the question in my mind was, "if I do all this stuff, how much can I reduce my risk?" As best I could tell from anecdotal experience, even if I'm a perfect cyclist, my odds are not that much better than if I was not that good.
I decided to dig around a bit and found the Crash-Type Manual for Bicyclists by Carol Tan and The University of North Carolina Highway Safety Research Center [although the information resides at The Turner-Fairbank Highway Research Center.] It's a detailed summary of a study (although inconveniently stored in 73 separate PDF files) on the underlying causes of bicycle accidents. Indeed accidents from the rear are comparatively rare (less than 3%), much more likely on roads higher than 60 KPH (37 MPH), and much more likely in unlit areas at night.
In perusing the data, I noted that by using the same techniques to avoid an accident while driving an automobile, I can avoid the vast majority of the accident types between a bicycle and a car. From there, I can further reduce my risk by being seen: in cases where the bicyclist was otherwise not doing anything out of the ordinary, most of the accidents could be attributed to the driver not seeing the cyclist. It's rather obvious to say this, but most drivers don't want to get into an accident, even with a cyclist — and it's clear from watching a busy roadway that they're generally excellent at not colliding with stuff.
And much of the advice Triest gave was along the same lines: be seen, behave like a car, and don't not behave like a car — stand your ground. By following this advice, it's unlikely you'll ever be in an accident on a bicycle, at least with a car. And if you do get into such an accident, it can be traced to not following that advice.