One more big one related to your first point: the demographic driving Teslas is hardly a random sampling of the population, which probably impacts driving behav

One more big one related to your first point: the demographic driving Teslas is hardly a random sampling of the population, which probably impacts driving behavior _but also_ implies the driving done (e.g., where) is probably quite different from the population as a whole.