www.astralcodexten.com/p/bounded-distrust

Why this is incorrect

The post frames the first allegation as essentially: the underlying ethnicity/immigration statistics were publicly accessible, but Swedish researchers still needed permission to use them.

However, the high-profile Swedish case commonly discussed in this context concerns register-based research using individual-level data linked across Swedish population registers, which is not publicly accessible “statistics” in the normal sense.

Evidence

1. The underlying study describes using linked population registers and an ethics approval
   Swedish rape offenders — a latent class analysis states that it analyzed “individuals from Swedish population-based registers” and that the registers were “linked using each individual’s unique identification number” (then replaced with a serial number), and that the study was “covered by ethical approval from the Regional Ethical Review Board in Lund.” That is incompatible with the claim that it was merely using publicly accessible statistics that should be free to use without special permission.

2. Swedish medical press coverage describes the issue as lack/scope of ethics approval for sensitive personal data and offense data
   Läkartidningen reports that ÖNEP’s referrals involved register studies and that ÖNEP’s position was that the articles involved “känsliga personuppgifter” (sensitive personal data) and “personuppgifter rörande lagöverträdelser” (personal data about legal offenses) for which the researcher did not have the required ethics approval.

Bottom line

Because the relevant research was based on non-public, individual-level register data linked by identifiers and governed by ethics-approval requirements, describing it as “publicly accessible” statistics that merely required “permission to use publicly accessible data” is contradicted by the study’s own methods section and by contemporaneous coverage of the ethics-review allegations.