Evidence register
See which publications are current, when each was checked, how revisions are handled and which evidence remains deliberately unavailable.
Publication status
A core release is “current” only when its official source was fetched successfully, its observation period and publication date were validated, and it remains inside the section's freshness window. This is a publication check, not an endorsement of the result.
Register checked
Checking
Original publications
Links go to the publisher or current primary tables, not to an aggregator. Each evidence page provides the more specific series and period used.
Individual Great Britain voting-intention publications. public-data.org does not calculate a synthetic polling average.
Original publications
Three named commercial markets with raw decimal odds and unnormalised reciprocal percentages. They appear only when the complete observation is inside its four-hour window.
Original publication
Monthly GDP, labour market, public-sector debt, central-government receipts, inflation, unemployment and long-term migration.
Original publication
Official Bank Rate in the key economic indicators evidence.
Original publication
Monthly referral-to-treatment waiting-time publication. A&E, GP waits, workforce and life expectancy are not part of this measure.
Original publication
Evidence gaps
These topics remain visible as editorial decisions, not as empty promises. They return only when the input data and method can be reproduced.
No value is displayed until one consistent question and reproducible primary-poll series is available.
Original publications
No score is displayed because the former formula and poll-level inputs were not reproducible.
Original publication
No combined figure is displayed. Crime Survey estimates and police-recorded offences require separate official series.
Original publications
No map, ranking or value is displayed because the former layers lacked row-level evidence, standard geography codes and cross-nation comparability.
Original publications
No coefficient or relationship strength is displayed because the former inputs, variables, weighting and calculation were not reproducible.
Original publication
Publication dates
Publisher calendars show when a release is expected; they do not, by themselves, establish that the latest figures are present here. The current-status register records the separate publication check.
Revision handling
Retain the publication period and identify revised observations rather than silently replacing history.
Revision handling
Keep CPI and labour releases distinct and show revisions or methodological breaks at the point of use.
Revision handling
Store dated published observations and expose later revisions; never simulate a continuously changing debt total.
Revision handling
Decision records are immutable; corrections must be recorded separately from the original decision date.
Revision handling
Retain each poll separately with fieldwork, sample, question and direct tables; expire the latest accepted publication after 14 days and never synthesize an undocumented average.
Revision handling
Retain the raw decimal odds and observation time, display unnormalised reciprocal percentages, and withdraw all three markets after four hours.
Revision handling
Treat provider revisions and missing-trust estimates as part of the named RTT release; do not mix A&E, GP, workforce or life-expectancy series into it.
Revision handling
Preserve provisional status, observation period and later revisions; keep visa and nationality administrative statistics outside this measure.