This deserves its own thread.
Has there ever been any objective data about a person's measurable driving habits that correlates well with their overall risk of being in a crash?
Note: I said driving habits. Not age or gender.
I'm thinking about electronic data that would be possible to measure. For instance, the radar tech that tells you how far you are from the wall behind you should also be able to tell how closely you tailgate the driver in front. It seems intuitive to say that people who habitually drive close to the car in front will usually crash more often, although I'd need to see proper field data to reach that conclusion.
But a simple recording of speed would also provide good data, I assume.
Has this ever been seriously tried? Or even proposed? Imagine: an insurance system where your premiums are based on how you drive, as measured by unbiased electronic devices, not your age or gender.