The Importance of Relative Standards in ADHD Diagnoses: Evidence Based on Exact Birth Dates
Elder, T. E. · 2010
grade Bnatural-experimentindependentreplicatednumbers spot-checked
Sample
11,784 children (ECLS-K, kindergarten 1998-99 through grade 8)
Population
US children born just before vs just after state kindergarten-eligibility cutoffs.
Design
Quasi-random relative-age assignment within grade. The decisive test is comparing TEACHER (peer-referenced) with PARENT (non-peer-referenced) ratings.
Key findings
ADHD diagnosis is substantially driven by classroom RELATIVE AGE rather than absolute developmental problems: the teacher (peer-referenced) gradient is 4x the parent gradient. The youngest children in a class are roughly twice as likely to be medicated years later. A labelling artifact with real, lasting consequences.
Counterfactual
Otherwise-similar children who are oldest rather than youngest in their class.
Genetic confound
Birth-date discontinuity.
Replication notes
Replicated internationally; one of the most robust artifact findings in education research.
DOI / URL
10.1016/j.jhealeco.2010.06.003
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| ADHD diagnosis, youngest vs oldest in class | % | 8.4% vs 5.1% — a 3.3pp gap against a 6.4% sample mean | mixed | grade-3 | none | not-applicable | health |
| Teacher-rated ADHD symptoms (peer-referenced) | SD discontinuity | 0.59 | researcher-designed | grade-3 | none | not-applicable | behaviour |
| Parent-rated ADHD symptoms (NOT peer-referenced) | SD discontinuity | 0.14 — much weaker, revealing the artifact | researcher-designed | grade-3 | none | not-applicable | behaviour |
| Stimulant medication use, grades 5 and 8 | relative risk | youngest ~2x as likely — the labelling has persistent treatment consequences | mixed | grade-6+ | none | not-applicable | health |
Cited by
- School starting age, relative age, and academic redshirtingmixedconf: highgc: low