When I was at uni and was cycling to a 10am lecture, a small van pulled out of a side road across my path. I was in a cycle lane and had right of way. He simply didn't look in my direction until he had already started to pull out. I was unable to stop or avoid him, so hit the front of his van head on, was briefly airborne, and landed on the road on the other side. Miraculously only scuffs and bruises, although my bike was a write-off. At that point I started wearing a helmet(!!!)
The next year, one of my friends went over the handlebars of her bike during a collision, except she landed head-first on the road, thankfully she was wearing a helmet. The helmet was split, she walked away with nothing other than mild concussion. Another friend and I walked out to rescue her bike later from the accident location, it also was a write-off.
Once you've seen a bike accident up-close there is simply no justification for not wearing a helmet. Even if you haven't seen an accident, wear a helmet!
To be clear, I am not saying that it is safer to ride without a helmet, just an interesting study of behavior.
I get that you know people who are harmed by cars (as do I), but your appeal to pathos & anecdote should not have a bearing here.
> The distribution of overtaking events shifted just over one-fifth of a standard deviation closer to the rider – a potentially important behaviour if, as theoretical frameworks suggest, near-misses and collisions lie on a continuum.
Looking at the graph in figure 1, "one-fifth of a standard" looks as unimpressive as expected. The distance from the kerb looks much more relevant.
IIUC all the analysis is based on 2355 data gathered by Walker riding a bike himself a few years ago between two cities. It's not a mix of data from different persons or a mix of city and countryside rides. The most interesting part is that Walker published 5 articles about the same data, and he got a different result in each one.
You're right that perhaps it is a big difference between cyclists, my guess would be the number of cars matters more than the number of cyclists. The study has been replicated elsewhere.
> Walker published 5 articles about the same data, and he got a different result in each one.
Uh, no? The 8.5 cm result has been consistent.
At the margin, these things make a difference, especially given that collisions are rare as a fraction for rides taken.
The study is disputed by one researcher, the paper I linked contains a rebuttal and I think it is convincing.