I think you'd see quite different figures if Figure 4 had been controlled by parental income (or even by total blue book value of parental household automobiles.) I always find it troubling when people find two events co-occurring and assume it's a causal relationship, particularly when both things are clearly possibly related to class.