Justin’s Notes

When Being Right Protects Being Wrong

2026-09-18 · AI-generated insight

Inzlicht's confession names a mechanism with unusual candor: moral conviction functioned as cover for methodological sloppiness. He didn't p-hack because he was cynical — he p-hacked because the conclusions felt so obviously good that scrutinizing the path to them seemed almost churlish. The good guys already believed it; the statistics were a formality.

The Winegard and Carl note supplies what looks like a decades-long case study of the same mechanism operating at the level of a whole field. Lewontin's argument — that within-group genetic variation dwarfs between-group variation, therefore population structure is biologically meaningless — contains a straightforward statistical error: it ignores correlation across loci. Their face analogy makes the error almost embarrassing in its simplicity. No single facial feature separates the sexes, yet faces taken whole classify sex at 95 percent accuracy. Aggregation of weak signals is elementary statistics.

So why did the argument survive for thirty years as a canonical citation? Inzlicht's mechanism suggests an answer: an error that lands on the right side of history doesn't get audited. The community's error-correction machinery — peer review, replication, methodological criticism — runs on the willingness of individuals to attack claims, and that willingness collapses when attacking the claim looks like attacking the moral consensus it supports. Inzlicht's p-hacks and Lewontin's fallacy are the same failure at different scales: one researcher letting a result slide, versus a discipline letting an argument slide.

The uncomfortable synthesis is that moral conviction doesn't merely bias which questions get asked; it selectively disables verification. Errors are asymmetrically durable — mistakes pointing the 'wrong' way get hunted down, mistakes pointing the 'right' way get cited. This predicts that a field's most morally comfortable findings should be its least reliable ones, which is roughly what the replication crisis found in social psychology. Inzlicht could eventually confess because the replication crisis gave him a genre for confession. There's no equivalent ritual for retiring a fallacy that a field still finds morally useful.

Woven from these notes

In brief, Lewontin ignored the fact that differences among human populations are correlated. Although population variation at any one genetic locus tends to be small, global population structure becomes clear if one examines correlated differences across loci. For a simple analogy, consider men and women’s faces. If one takes any particular characteristic, say nose size, there is likely to be more variation within a sex than between the sexes (and it would be nearly impossible to classify faces by sex with any accuracy using only nose size). However, if one considers all facial characteristics together (which is, after all, how we actually experience human faces), then sex differences become sufficiently clear that an observer can guess the correct sex more than 95 percent of the time.

— Bo Winegard, Noah Carl Superior: The Return of Race Science—A Review
My work was uplifting people, or so I told myself. Looking back, I suspect that feeling like I was on the right side of history was a kind of moral cover. Maybe I felt less wrong about the p-hacking because I was promoting social justice. The conclusions I drew were what the good guys already believed, after all.
— Michael Inzlicht My P-Hacking Felt Like Truth