WEIRD Samples

What it is

The critique is not that the research is bad. It is that "people do X" is a claim about humans while the sample supporting it is a claim about a sliver of them. The uncomfortable version of the argument goes further: that this sliver is not merely unrepresentative but genuinely unusual, so the discipline has been building its account of human nature on its strangest available case.

Adrian Furnham devotes a chapter of The New Psychology to it and credits Henrich's 2020 book with the strong thesis, that WEIRD people are psychologically distinctive, being more individualistic, analytical, impartially rule-following and cooperative toward strangers than most human populations, and that the pattern has a historical cause in Western Christianity's marriage and family programme breaking up extended kin groups.

Two things he does around that thesis are worth keeping. He flags the historical account as open to challenge rather than presenting it as settled, which is the right calibration. And he raises an objection of his own, that WEIRD lumps together Western cultures which differ from each other substantially, so the acronym flattens on one axis while sharpening on another. He also supplies the historical note that about forty years ago the label WASP performed the same critical function in the same argument.

In effect

The verdict has to be stated in parts rather than averaged. The descriptive claim is solid. The historical claim about kinship and Christianity is an interesting and unresolved hypothesis, which is where Furnham lands too.

One claim should be handled at arm's length. He speculates that WEIRD sampling may be part of the cause of the replication crisis. That is his speculation, hedged in his own words, and it is not the standard account. The usual diagnosis is incentives, small samples, analytic flexibility and publication bias, and sampling narrowness is not normally counted among the causes, since a homogeneous sample tends to make a result easier to reproduce in a similarly homogeneous one rather than harder.

The attribution failure in that chapter is worth recording as an apparatus finding. The chapter is named for the acronym and never names the people who coined it in the prose. Henrich, Heine and Norenzayan appear in the bibliography only, so a reader following the argument from the page has nobody to follow.

The mirror image is the reason this concept is more than a scolding. Margaret Mead's Coming of Age in Samoa is the same failure run backwards. She studied sixty-eight girls in three villages on one island, tabulated fifty and had sexual data on twenty-five, then used them across some forty pages to prescribe for American schooling. The sample is as far from WEIRD as a sample gets, and the claim it was made to carry was about the West. Two things follow. Widening the sampling frame does not on its own fix the inference, because a tiny non-Western sample generalised toward the West has the same structure as a large Western one generalised toward everyone. And the discipline's own founding counter-example to universalism was itself a single small non-representative case, which is worth remembering before the WEIRD critique is used as though breadth were automatically rigour.

What it does not say

It does not say that findings from WEIRD samples are false. It says the population they describe is narrower than the write-up implies, which is a question about scope rather than about validity.

It does not say the historical explanation is established. Henrich's account of kinship and Christianity is contested, and Furnham says so.

It does not explain the replication crisis, and the speculation that it might should not be carried into an article as an explanation of it.

---