Is that crashes per mile driven - as NHTSA figures it - or is it raw crashes among the age group? Because if fewer people in that age group are driving, then a decline in the total number of crashes doesn't mean they're driving more safely.