Confounder Of The Day: How Sexy Your Parents Were

A 'Confounder Of The Day' piece using psychiatric epidemiology to teach causal inference. A large Danish JAMA Psychiatry study finds more psychiatric disease in children of young mothers and old fathers; the old-father effect has a tidy mechanistic story (accumulating de novo mutations / 'selfish spermatogonal selection' raising mutational load — a 45-year-old father's child is 1.5x schizophrenia, 1.8x autism vs a 25-year-old's). But a clever Petersen et al study shows child mental health tracks the father's age at his FIRST child, not his age at this birth — adjust for the former and the latter vanishes. The likely confound: men with dysfunctional traits/genes take longer to find partners, so they reproduce later; the genes cause both the late fatherhood and the disorder. A clean, instructive dissolution of a plausible causal story (with a candid personal coda about Scott, 30 and waiting to have kids).

Quality 65 · Strong. Strong, low (65). A crisp, well-explained worked example of confounding with a genuine 'aha' (the age-at-first-child identification trick). Held to low-Strong by its narrow scope (one study pair) and brevity; note that the paternal-age->de-novo-mutation link is in fact fairly robust (Kong 2012), so the confound complicates rather than overturns it.

Claude’s paradigm shift 48 · Moderate. Moderate (48). The sharp identification idea is Petersen et al's; Scott popularizes and explains it well, modest novelty for him.

Real-world impact 2 · Minor. Minor (2). A within-blog statistics-teaching example; no policy or practice effect.