How we check reports
September 2026 · v0.1
prisoncheck measures the practical experience of each prison in England and Wales: waits, local rules, and what changed. Not a survey of every family.
Sources
- Community submissions from families, visitors, people who have been inside, and professionals.
- Official prison information from GOV.UK / HMPPS, labelled Official and dated.
- Publicly available information such as occupancy figures and Independent Monitoring Board reports.
Collection
A person picks what they are reporting, names the prison, says when it happened, writes what happened, and says whose experience it was. The original words are stored. A structured observation is derived from those words and from any days, chips, or dates they gave.
Processing
Free text stays as free text. Waits become a number of days. Topics map onto a short internal taxonomy. The live prison page is a view of current observations, not the other way around.
Verification
A new submission is unverified. It becomes corroborated when another family reports the same prison and question, or when someone later marks that fact still true. Official facts are never mixed into community ranges.
Limitations
PrisonCheck submissions are voluntary and self-selected and therefore should not be assumed to represent the experiences of all prisoners, families or visitors. Old experiences are kept, but they do not stand in for the current picture.
Privacy
We ask for the prison, the topic, when it happened, what happened, and whose experience it was. We do not ask for names, prisoner numbers, or addresses. Notes with a phone number, email, link or prison number are turned away, and notes that might name or accuse someone are read by a person before they are published. After a report, two optional questions ask how the person is related to the one inside and whether children at home are waiting for contact, by age group. Those answers are never published; research sees totals of five or more families only.
Versioning
Each observation has an experienced date and a submitted date. Dataset snapshots are versioned by month so later researchers can see what the picture was, not only what it is now.
Taxonomy
Fifteen top-level categories. We will learn the rest from submissions.
Communication · Visits · Money · Clothing · Property · Reception · Healthcare · Food · Regime · Release · Resettlement · Travel · Family · Staff interaction · Other