Methodology

Listings were compiled from public directory pages (breed association, registry, or trade publication) fetched 2026-07-18 — the 6 sources listed below. Parsing and normalization support maintenance; source citations do not independently confirm a farm’s current operation. The dataset is planned to refresh quarterly through a manual review queue that never auto-publishes.

Sources

Every row cites the exact page it came from. The current sources are:

Data from these pages is re-aggregated as facts (farm, city, state, breed, website); page prose and logos are not copied.

Field policy

Version 1 is privacy-light: farm/business name, city, state, breed, website, and source only. Phone numbers, emails, and owner/person names that appear on source pages are read only to validate a row and are deliberately dropped. Contact channels are added later through claim-your-listing.

Selection

All Ohio and all Iowa association listings with a real farm/business name were kept. The Myotonic Goat Registry is a large single-breed directory, so a round-robin sample (about one per state) was taken to maximize geographic coverage within a small proof dataset. Rows with only an owner name and no farm name were skipped.

Normalization

Breed wording is mapped to a controlled vocabulary (for example “Experimental,” “Recorded Grade,” and “Grade” all map to Recorded Grade); the original wording is preserved in the data. Myotonic rows carry the breed from the registry's single-breed identity, since the directory shows no per-row breed.

Update cadence

The plan (see refresh.py) is a quarterly re-fetch of each source page, a field-by-field diff against the current rows, and a review queue of adds, changes, and removals. Nothing is auto-published: a human approves the queue, then the site is regenerated.

Corrections

Corrections and removals will flow through the claim-your-listing intake. Until that backend exists, the data reflects our interpretation of source records fetched on 2026-07-18.