> People of lower socioeconomic status drive less safe cars on less safe roads for longer commutes.
But can't you account for 'type of car', 'type of road', 'commute length' as direct variables pretty easily without dipping into social/economic backgrounds?