Stately

For researchers

Civometry holds 390 series from 97 official publications by 17 publishers, keyed to current UK areas. Nothing here is modelled, weighted, smoothed or interpolated. Every figure is either as its publisher published it, or arithmetic on published figures with the sum shown, so any number can be reproduced from the source rather than taken on trust.

Reproducing a single figure

The provenance endpoint returns, for any figure, the file it was read from with its sha256, the sheet, row and column it sits in, when the file was retrieved and the commit of the code that read it. Download the file from the publisher, check the hash, open the cell.

GET /api/provenance/<metric>/<area>?period=YYYY-MM-DD

/api/provenance/council_tax_band_d_authority/camden

Where a publisher has since revised the figure, the earlier vintages come back with it. The same history is on revisions, and Civometry’s own corrections are on corrections.

Working with the whole thing

  • The API — area report cards, comparisons, map layers and full series as JSON or CSV.
  • The data catalogue — every publication with its publisher, cadence, coverage and whether the latest release is in.
  • Releases — every file ingested, newest first, with its hash and the date it was retrieved.
  • Methodology — the rules the site holds itself to, including what it refuses to do.

Citing it

Cite the publisher for a published figure. Cite Civometry as well only for a figure it calculated. The worked examples are on how to cite; for a reference manager:

@misc{civometry,
  author       = {{Civometry}},
  title        = {Civometry: United Kingdom official statistics, area by area},
  year         = {2026},
  howpublished = {\url{https://civometry.com}},
  note         = {Accessed: dd Month yyyy}
}

What is not published, and why

The data gaps register records questions official statistics do not answer, each classified by what exists: derivable, partial, fragmented across bodies, behind a login, restricted for disclosure control, or not collected at all. Each entry names the authority that would hold the record and says whether a request would help. It is meant to be a research agenda as much as an explanation.

If a figure here is wrong

Challenges are checked against the source table, and the ones that hold are published with the original and corrected value. Say which part is wrong — the source, the numerator, the denominator, the mapping, the period, the formula or the reading — and it can be checked.

Every metric page and every data-gap entry carries a challenge form, so an objection arrives attached to the thing it is about. What has been accepted or answered so far is on challenges and corrections.