The research on discrimination in hiring is about as solid as social science gets. Send out matched applications, identical but for a name that signals race, and count the callbacks; the gap is real, it has been found for decades, and it survives replication across many studies and several countries. Within Western labour markets, majority-on-minority discrimination shows up in the data, again and again. This is good work answering the question it set, and anyone who waves it away has not read it.
The question it set is narrower than the conclusion drawn from it. The entire apparatus - the studies, the researchers, the funding, the design - sits almost wholly inside Western countries, and it operationalises discrimination as the majority disfavouring a minority within a single Western nation. An instrument built and aimed in one direction can only ever catch what lies in that direction.
The only place in the dock
The accumulated result reads as a sentence on the West: here is where discrimination has been measured, so here is where it must live. But that reads the aim of the instrument as if it were a fact about the world. The audits were never run in the places the method cannot reach - the discriminations that run along lines of caste, sect, region, or clan, in societies nobody thought to send matched CVs into. Their absence from the record is not evidence of their absence in fact. It is evidence of where the researchers were standing.
The one instrument that does reach almost everyone tells a different story. Ask people across the world whether they would object to a neighbour of another race - a crude question, but a global one, and one not designed around Western guilt - and the ranking that emerges is not the one the discourse assumes. Several non-Western societies score markedly higher on that measure than the Western ones routinely placed in the dock. It is the evidence least shaped by Western-aimed research, and it points the other way.
What this does and does not license
This is not a reason to dismiss the audit studies. They are true about what they measure, and what they measure matters. It is a reason to stop treating a locally-aimed instrument as a global verdict - which is a mistake made in complete good faith, by careful people, precisely because the findings are so robust that their boundaries become invisible.
The honest confidence levels are worth stating plainly. High, on the direction: measurement pointed one way will find fault one way, and the global picture is not what the Western-only picture implies. Lower, on any precise ranking of who is most or least anything: much of the world is barely surveyed, self-report is noisy, and a clean league table is exactly the overreach this piece is warning against.
Before accepting that "the evidence shows" something about the world, it is worth asking a plain question about the evidence: where was it allowed to look, and what was it built to be unable to see. The answer usually narrows the claim a great deal, and it does so without anyone having lied.
How this piece was made
How this piece was made. The audit-study literature is stated at full strength and conceded as robust before the turn; the argument is epistemic - about the aim of a measurement instrument - and deliberately frames no racial verdict of its own, which is both the intellectually honest position and the safe one. The World Values Survey neighbour question is cited as the globally-reaching counter-instrument; specific country rankings are deliberately not asserted, and the confidence levels are graded in the text. No individual or group is placed in the dock. A critic should test the central claim that Western-aimed research cannot support a global conclusion, and the reliance on a single survey question as the counterweight - both are flagged in the piece as directional, not definitive.