Hi Anthony, thanks for the informative analysis and interpretation. It is an important task in getting at a difficult problem in risk analysis and management.

One thing I wasn’t sure I agreed on was toward the end of your post:

“The BikeSpot data comparison with CrashStats confirms, at least amongst the top 10 Unsafe Spots, that the perception of bike riding risk is greater than the real risk of serious injury when riding.”

I don’t think that is true. What we can say is that there is great variation in the ‘real risk’ for a set of sites with high ‘perceived risk’. I don’t think it is possible to absolutely compare ‘real’ and ‘perceived’ risk. What are the units of a common scale? CrashStats are counts through time. The BikeSpot data were counts of qualitative assessments.

This is linked to the question of risk attitude; what is high risk? Without having looked at the data, I would think that the risk of fatality or serious injury per km travelled is much higher for a cyclist than for a motorist. So as a commuter, choosing to cycle over drive to work, I that I am exposed to higher risk. That is a topic ripe for long discussion.

One other broad comment, which you touch on, the ‘real risk’ is conditional upon risk management measures, some of which are by riders managing their riding behaviour.