Measuring inappropriate medical diagnosis and treatment in survey data: The case of ADHD among school-age children.

We exploit the discontinuity in age when children start kindergarten generated by state eligibility laws to examine whether relative age is a significant determinant of ADHD diagnosis and treatment. Using a regression discontinuity model and exact dates of birth, we find that children born just after the cutoff, who are relatively old-for-grade, have a significantly lower incidence of ADHD diagnosis and treatment compared with similar children born just before the cutoff date, who are relatively young-for-grade. Since ADHD is an underlying neurological problem where incidence rates should not change dramatically from one birth date to the next, these results suggest that age relative to peers in class, and the resulting differences in behavior, directly affects a child's probability of being diagnosed with and treated for ADHD.