Where the official numbers are late, revised or simply missing.
Where the official numbers are late, revised or simply missing, and how to check them yourself. When a stat is shaky, we say so.
9 figures, each with its official source.
Shows the scale of normal revision, driven mainly by better estimates of British-national emigration, a designed feature, not an error.
Provisional for ~12 months; ~5-month lag; not yet accredited Unlike the Home Office quarterly statistics, these are not yet accredited official statistics.
Quarterly; YE March 2026 data published 21 May 2026 Accredited and quality-assured, but the newest detailed breakdowns are always ~2 months old.
Provisional operational counts; exclude undetected arrivals and other routes The Home Office says detections 'should not be used to infer levels of irregular migration'. This is where the 'under-recording' concern is legitimate, undetected arrivals are, by definition, not counted.
Home Office has not reported appeal outcomes since year ending March 2023 Obscures how many refused cases are later overturned at appeal. Tribunal (HMCTS/MoJ) data still exists separately.
Nationality inconsistently recorded; new breakdowns are 'experimental' Producers caveat that nationality is recorded at arrest 'not necessarily accurately' and crime/immigration databases don't link cleanly.
Casey: data 'not good enough to support any statements' on offender ethnicity nationally An official argument for humility about ethnicity-crime claims.
The ~£4.9bn total 'does not capture all costs… particularly those incurred by local authorities' The real cost is INCOMPLETE, not overstated, education, ESOL, integration and social care are often not captured.
Migration Observatory · Full Fact · Refugee Council · Office for Statistics Regulation Impartial, independent analysts useful for triangulating official figures. Not government producers.