Some people claim that most crashes involves sleep deprivation, but I agree with Vladimir that we have no idea.

What is suggestive is two studies of one hour differences in school start times (americans drive to school at 7:30 at age 16). One compared Virginia Beach to Chesapeake. Better was when Lexington changed its start time. The student crash rate went down 16%. The state rate went up by 8%, so these are very volatile numbers and this could be data-mining. Also, day light savings causes crashes to spike. What's great about the school studies is that they're about practical interventions and not just assigning blame.