Decomposing Autism: Liability vs Drift

The author proposes a simple, reproducible method to apportion the rise in autism diagnoses into true liability change versus diagnostic drift using a latent‑liability threshold model. By placing diagnosis rates on the probit scale and anchoring to symptom-score distributions, one can compute a liability‑only counterfactual and estimate each share. — A clear, testable decomposition can resolve ‘autism epidemic’ claims and reorient policy, research, and media coverage toward causes supported by data rather than inference from raw diagnosis counts.