British Resilience Index

Methodology changes

• Two indicators change source this release. DCMS retired the Community Life Survey in October 2025 and replaced it with the Community and Engagement Survey, so neighbourhood belonging and regular formal volunteering now come from the new survey. Its first edition combines April to September 2025 collected under the old survey with October 2025 to March 2026 under the new one, weighted towards the larger later sample. Both points are taken because DCMS compares the two years directly in its own headline, and the break is recorded on each indicator's weakness note. Welsh children looked after also moves vintage: the publisher now reports the rate as a whole number and restates March 2024, so the series is taken wholly from the current edition rather than mixing vintages.
• From this release five indicators moved from provisional bounds to percentile normalisation, their verified histories having reached eight points (cancer 62 day standard, prison capacity, avoidable mortality, fuel poverty, neighbourhood belonging), the migration path stated on the methodology page. Publisher revisions adopted: DESNZ restated fuel poverty for 2022 and 2023, and the avoidable mortality history now comes wholly from the February 2026 ONS edition so every point shares one vintage. An undated prison capacity history point that could not be traced to any weekly bulletin was replaced with the verified 11 August 2025 reading. The change was measured against the August release before this one was cut: on unchanged August 2026 data it moves the national score from 51.63 to 51.00, and September's new data then moves it to 50.71. So of the 1 point fall from 52 to 51, the normalisation change accounts for about two thirds and new source data for the rest.
• The scoring model is revised this release, the first change to it since June. Volatility leaves the metric stress blend, which becomes two thirds level and one third trend. On this register volatility was pinned at its maximum for 14 of the 71 scored indicators and was a placeholder for 19, so on nearly half the register a component that only ever pushed stress upward was either saturated or invented. It is still computed and published on every indicator. Second, a component that cannot be measured now contributes nothing and the remaining weight is rescaled, rather than contributing a neutral 50: under the old rule an indicator with a short history could not score below 20 or above 80 while one with a long history could span the full range, which was an artefact of history length rather than a statement about uncertainty. Third, the systemic risk penalty is retired. It added 5 or 10 points once three or five domains crossed 75; the cliff was arbitrary, the index carries no model of how a failure in one domain reaches another and so had nothing to calibrate the magnitude against, and it was never applied to the retrospective trend line, leaving the headline and the published history on scales that could differ by up to ten points. It stood at 0 in this release and the one before, so retiring it changes no published figure. The count of domains at or above 75 is now published as a signal instead. Measured on unchanged September data, the revision moves the national score from 50.71 to 51.43: the headline stays 51 and the band stays Fragile. Domain scores do move, by up to 5 points, and rule of law crosses from Strained to Fragile at 51. Finally the top two band labels are renamed. Critical becomes High stress, and the top band becomes Extreme stress, because the former names implied the index knows where a system breaks down and it does not. Band identifiers in the published data are unchanged.