I chose my words carefully. I didn't say that people are good at perceiving risk, I said they were not terrible at it. In the context of this discussion, being the relative levels of risk within different moments of a single driving journey. Most people aren't navigating a touch screen while steering a complex intersection.
People are famously bad at assessing disparate risks. The unit micromort is useful for clarifying mortality risk of various behaviours.
With respect to crash statistics, obviously we're not going to ask people why they crashed. I would have thought this is so self-evidently flawed as a methodology that it didn't warrant pointing out. It's probably safer to assume that most people will lie about why they crashed.
Fortunately, the statistical question is much simpler than it might seem: are people crashing newer cars more than older cars? Where answering this question becomes complex to answer accurately is when isolating this from both environmental change and vehicle age, because you can't isolate both variables at the same time.
People are famously bad at assessing disparate risks. The unit micromort is useful for clarifying mortality risk of various behaviours.
https://en.wikipedia.org/wiki/Micromort
———————
With respect to crash statistics, obviously we're not going to ask people why they crashed. I would have thought this is so self-evidently flawed as a methodology that it didn't warrant pointing out. It's probably safer to assume that most people will lie about why they crashed.
Fortunately, the statistical question is much simpler than it might seem: are people crashing newer cars more than older cars? Where answering this question becomes complex to answer accurately is when isolating this from both environmental change and vehicle age, because you can't isolate both variables at the same time.