← All chiropractic programs · Cost disclosure, programme by programme
Data availabilityThree files on cost, and the prerequisites dataset beside them, published as they stand. Everything counted on this site about what the nineteen CCE-accredited Doctor of Chiropractic programmes publish about cost is counted from the first of them, and every count can be re-run.
schools_cost_disclosure.csv — 19 rows, 68 columns, one row per programme. Coded columns carry a controlled vocabulary; the wording columns carry the college’s published sentence together with our note on where it sits and how it was checked.
ABSTRACT_COUNTS.md — every count published on this site, with the column each is taken from and the arithmetic shown where a count has any.
RENDERED_ONLY.md — published documents that exist, return HTTP 200, and carry no text layer we could compare against. No figure from any of them appears in the dataset, in prose, in a footnote, or as an approximation. They are listed so anyone can read them directly.
What each accredited program publishes about its prerequisites, captured 3 October 2026, with each amount in the unit the college printed; a program whose site refused capture is marked unverified. The finding it supports is how the accredited programs publish their prerequisites. The files: programs.csv, pathways.csv, requirements.csv, pointers.csv, disagreements.csv, FINDINGS_COUNTS.txt. The dataset is published under CC BY 4.0. The license covers this compilation; quoted institutional text remains the institution’s.
Every not-published in this dataset means the document we read did not answer the question on the date recorded. It is not a claim that the college holds nothing, and it is not a claim that the college declines to say.
19 programmes — 18 in the United States, 1 in Canada. The list is the union of two registers whose gaps do not overlap, because neither reaches 19 on its own.
USDE DAPIP, the Department of Education’s database of accrediting agencies, checked 17 September 2026: 18 institutions, US-only in scope, so it does not reach the Canadian programme. CHEA, republishing the accreditor’s own report with an as of date of October 2024: 18 institutions, and that snapshot predates one US institution’s appearance. A third source, Canadian Memorial Chiropractic College’s own accreditation page, fetched 17 September 2026, carries the one fact DAPIP structurally cannot.
All 19 of our names match the union, on a normalised key — lowercase, collapsed whitespace, hyphen read as a space — because the union of the raw strings is 21: the two registers spell some institutions differently between themselves.
The paths shown inside ABSTRACT_COUNTS.md are our working copies of those registers. The registers themselves are public, at ope.ed.gov and chea.org, and neither working copy is redistributed here.
published — the document exists and answers the question.
not-published — the document exists and does not answer it.
unverified — a document exists and could not be read.
The third is the one that is usually lost. A cost of attendance published as an image with no text layer is not the same fact as a cost of attendance that was never published, and the dataset keeps them apart. Where a field is unverified, the document is listed in RENDERED_ONLY.md with its address.
No figure in the dataset is a total, an average, a difference, a product or a rate that we calculated. Where a college publishes components and no total, the components are recorded and the missing total is stated. A computed total is indistinguishable from a published one to a reader, so none is computed.
Quoted wording is byte-compared against the bytes the server sent before it enters a cell. A string split across a line break or a block boundary is never quoted as joined text.
A column that holds the college’s quoted sentence and our own note in the same cell cannot be counted by string-matching: a match cannot tell which of the two it hit, so the count measures our writing as much as the college’s.
Two sets of counts have been withdrawn on that ground and are not published anywhere on this site. How many of the fourteen colleges that state what their board figure counts count a six-month window, all parts, licensure as an alternative route, or a Canadian board — four figures, withdrawn 17 September 2026. And how many colleges publish a cost disclosure inside an expanding panel — withdrawn 18 September 2026, because the words it matched were our own markup vocabulary rather than anything a college publishes. In both places the underlying records are unchanged and can be read row by row; what is withheld is the number.
A count is published only where the column it reads is coded with a controlled vocabulary, or carries the source’s words and nothing of ours.
The files above are the whole of what is published from the working directory. The per-college source logs, the question log, the working notes and the held findings are not. The deployment names each file on its own line rather than copying a directory, so nothing else can reach the site by accident.
One cell of the dataset previously pointed at the path of a held finding. The pointer was removed on 18 September 2026; the substantive record — that a finding about one accreditor directory link is held unpublished because it rests on a fetch of a path that site’s robots.txt disallows — stays in the cell.
Tell us and it is corrected in public. Every claim this site has narrowed or reversed is collected on the corrections page, with what we said, what is true, how we found out, and the date. Corrections to our own fields are published on the same terms as corrections to claims about institutions.
What has to be true before anything is published at all is set out on the source rules page. If your institution has an entry and something on it is wrong, the page for institutions says how to correct it.