Typically people will take longer journeys in motorised vehicles (because they go more quickly) so this adds a significant bias to the data. A far better representation of risk is derived if you use “Time spent travelling”. For example, most people will have an upper limit on the time they want to spend travelling (e.g. 1 hour each way commute) and this will affect mode of transport choice.

Also, comparing individuals within each group you will see a significant spread of distances covered, which to a large extent represents the experience of the individual. So an individual with a higher exposure may not have a (proportionately) higher risk (e.g. an adult vs a child cyclist)

> I might speculate that if you are a racing cyclist, cycling along at 20mph, it’s more inconvenient for a motorist to pass you. If you’re doing 10mph, it’s easier.

Very much this — your relative velocity is less, so it needs a longer clear stretch of road to manage safely, and is compounded if there are more than one of them to pass.

It cuts the other way, too — when I’m cycling doing my usual <15mph amble on the way home from work, the worst thing to get passed by is a pack of grimly peddling no-fun-allowed sport cyclists. Buses and farm vehicles are much more inclined to give a wide margin when passing.