Sweden

How the measurement is built

Two requests went to all 21 county administrative boards with identical wording: A concerns the most recent inspection reports, B the most recent injunction decisions. A third request, C, concerns statistics from the Swedish Board of Agriculture's animal welfare inspection register. Every answer is handled under a protocol written in advance: the same narrowing when a fee is raised, the same request for a formal decision, the same deadlines. When the questions are identical and the handling predetermined, differences in outcome reflect the authorities, not the questions.

The collection is closed. All 43 requests reached a final status, and every authority concerned received its own figures on 2 October 2026 with three weeks to reply. The authorities' replies stand verbatim beside the figures on the Swedish page Handlingar, and a missing reply is written out as a missing reply. The refusal of request C by the Swedish Board of Agriculture is under appeal before the Administrative Court of Appeal in Jönköping, registered on 10 September 2026 as case 3339-26 and not yet decided.

The same cases in both classes

Point P1 of the protocol names the strongest comparison: cases where the same inspection appears both as a free-standing report in class A and as a document in class B. The authority, the period and the cases are then the same, and what differs is the document's legal status.

6 counties carry the authority's own case numbers in their file names in both classes, and in one of them, Västernorrland, the same cases appear in both. It has 20 such cases: 20 of 20 inspection reports are redacted in class A, against 0 of 20 documents from the same cases in class B. The classes hold different documents from one case, not the same document measured twice.

The comparison cannot be made in the other counties. 177 of 310 files carry no case number in the name, and without it nothing ties a report to a decision.

The annexes

The request covered annexes, and withholding a whole annex is the strongest redaction there is. Annexes are therefore measured on their own and never pooled with the area measure.

6 annex files arrived in the whole collection, from 3 counties. 114 documents refer to content in an annex, and 42 of them do so with one sentence from the national inspection report template: that photographs taken are in an annex. That sentence stands in the template whether or not a photograph was taken.

The pre-registered annex measures therefore cannot be computed, and they are reported as a gap with the reason written out. The material does not separate an annex that was held back from one that never existed, nothing links an annex file to its document, the annex file is read from the file name and is therefore a floor, and 45 files have no text layer to read. Finding out which applies would take a question to the authorities, and it is not asked.

The exact wording of the requests

The requests were made in Swedish and are reproduced verbatim below. The same text went to all 21 county administrative boards on 3 August 2026, one request per e-mail; request C went to the Swedish Board of Agriculture the same day.

Fee-stopped requests

Requests where a fee remained after the protocol's narrowing. Copies are then declined, a written decision on the fee is requested, and the outcome is documented. No fee has been paid. The fee question took 41 written exchanges, spread over the 16 counties that announced a fee, most of them in Jönköping, Kalmar and Örebro with five each. The outcomes: the fee fell away after the narrowing in 12 counties, no fee was announced in 5, a decision was issued in 2, a decision was refused in 1, one county answered with a general policy document, and no county left the question unanswered.

• Länsstyrelsen i Jönköping · A · no copies were collected
• Länsstyrelsen i Jönköping · B · no copies were collected
• Länsstyrelsen i Kalmar · A · the documents arrived later, without a fee
• Länsstyrelsen i Kalmar · B · the documents arrived later, without a fee
• Länsstyrelsen i Skåne · A · no copies were collected
• Länsstyrelsen i Skåne · B · no copies were collected
• Länsstyrelsen i Örebro · A · no copies were collected
• Länsstyrelsen i Örebro · B · no copies were collected

What this page does not show

• The documents themselves. They contain personal data and are never published; the listing with a checksum per file is on the Swedish page Handlingar.
• Any claim that an authority has broken the law. The law is unsettled, the Supreme Administrative Court has not decided the question, and four cases are before the administrative courts of appeal.
• Three counties that declined the copies once a fee remained after the narrowing. They carry a response time and a fee figure but no documents in the redaction measure, and the loss is never counted as a release or as a refusal.
• Pigs as a comparison. Sixteen documents have pigs as their dominant animal, spread across ten counties, so no county has material on both sides.
• What a document is when an authority bundles several into one file. The measure counts the delivered file, and two of the 343 files hold five documents each. One is Norrbotten's class B delivery, which therefore stands at one document where the county released five decisions; the redaction share is unaffected, since all five are unredacted. The other sits outside the class figures.
• The limits of the measure. They are in the method, together with the deviations from the pre-registration and the errors corrected during the work.