Paperonium

Pregnancy genetics · Int J Epidemiol · 2023

“Strong evidence” a baby’s gene version affects birthweight. By the paper’s own rule, only a hint.

1 in 101 in 10,000every time1 in 1,000the paper’s barrarer than1 in 1,000The summary: meets the pass markabout1 in 260Its own results: misses it
1 in 101 in 10,000every time1 in 1,000the paper’s barrarer than1 in 1,000The summary:meets the pass markabout1 in 260Its own results:misses it

How often luck alone gives one test a result this strong

The paper says

A mother’s genes and her baby’s can each affect birthweight; the study separates the two. Its summary reports “strong evidence” that one gene version acts through the baby: luck alone would produce a result this strong under 1 time in 1,000, the paper’s pass mark.

Its own data show

The results section repeats the claim, then two sentences later prints the figure itself: about 1 time in 260, nearly 4 times too often for the pass mark. Two of its tables print the same figure for this analysis, and its other numbers for this effect agree, so it isn’t a typo.

“1 in 260” sounds rare, but the paper ran 50 tests on this gene. The usual pass mark is 1 in 20; the paper made it 50 times stricter: 1 in 1,000. Across 50 tests, luck alone turns up a result this strong roughly 1 time in 5, so this baby-gene finding is a hint, not strong evidence.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

Partitioning genetic effects on birthweight at classical human leukocyte antigen loci into maternal and fetal components, using structural equation modelling · Int J Epidemiol (2023) · doi:10.1093/ije/dyad142