2011 Census figures for the eleven councils formed since 2011
Eleven councils formed after 2011 had no 2011 point in their ethnic group timeline, because the site held their 2021 Census figure where the 2011 one belongs: Bournemouth, Christchurch and Poole; Buckinghamshire; Cumberland; Dorset; East Suffolk; North Northamptonshire; North Yorkshire; Somerset; West Northamptonshire; West Suffolk; and Westmorland and Furness. Each is a merger of whole 2011 districts, so its 2011 figure is now the sum of those districts' counts in Census 2011 table KS201EW: Cumberland's 274,549 residents are Allerdale's 96,422, Carlisle's 107,524 and Copeland's 70,603. Their place pages gain a 2011 point in the timeline and in the composition table, and the 318 areas' 2011 totals now add up to the England and Wales total of 56,075,912 exactly. The SEND finding's rank correlation had counted Bournemouth, Christchurch and Poole and Dorset as having no change in the White British share; with their 2011 figures it is -0.05, not -0.04, across the same 123 authorities, and the finding carries a dated update. The shift-share breakdown still covers the 307 areas with a 2011 Census row of their own; extending it to all 318 follows separately.
Constituency profiles weighted by postcode share alone
Constituency pages build their White British and foreign-born shares from the local authorities each constituency overlaps. Each authority was weighted by the share of the constituency's postcodes in it multiplied by the whole authority's population, which pulled every split constituency towards its larger authority: Rossendale and Darwen showed 71.5% White British in 2021, weighted towards all of Blackburn with Darwen, and now shows 78.0%. The weight is now the postcode share alone. The 229 constituencies that span more than one authority are recomputed, and 175 of them move by 0.05 points or more on the 2021 share; the 345 inside one authority do not change. These remain authority-level figures apportioned by postcode share, not constituency counts.
Six findings corrected after the October data refresh
Every figure in the findings was recomputed from the data files the site now publishes, and six findings carry a dated correction. The adult social care and SEND findings had used a 25-authority file typed in by hand in April 2026 that matched no published table and gave figures for district councils, which run neither service. On DHSC's 2024-25 figures, spend per adult is £687 for Lancashire County Council, which covers eight of the ten councils the adult social care finding compared, so its comparison does not hold. On DfE's 2025/26 figures, pupils with an EHC plan in England rose by 65.4% in five years, not the 42% the SEND finding gave, and growth shows no relationship with demographic change (rank correlation -0.04 across 123 authorities). The births finding had 2024 figures for five local authorities, four national counts and Burnley's split by the mother's region of birth that did not match ONS Tables 1 and 6a; Tower Hamlets, at 65.1%, was not past 70%. The 86-areas and Birmingham findings carry the corrected cohort-component figures, 54.7% and 20.6%, and the 86-areas finding no longer says the two models agree to within half a point. The Blackburn finding corrects a 2011 share and an asylum figure for Preston that matched no Home Office quarter. The adult social care and SEND findings are retitled to say what the published figures show, with the old headlines kept in their corrections.
Two-model spread recomputed after a double count
The cohort-component model behind the two-model comparison on place pages read its population envelope, the ONS 2022-based subnational population projections, twice over: it added each area's "All ages" row to the rows for each single year of age, so every target was double the published figure (Burnley in 2031: 199,651 against the file's 99,825). It has been re-run on its own, with the fix and on the same inputs as model v8.0, and only its own fields change: the cohort-component 2051 White British share and its scenario range in 315 of the 318 areas, and the spread computed from them. The Hamilton-Perry projections, which are the headline figures, do not change. Across the 269 areas the current Hamilton-Perry model projects, the median 2051 spread is 6.7 percentage points, not approximately 8; the largest is 25.2, not 32.3; and 8 areas have a spread above 20 points, not 17. The spread changed in 266 of those 269 areas, by a median of 2.1 points and at most 7.1. The comparison is no longer shown for the other 49 areas, which carry projections from an earlier model run: in 28 of them the Hamilton-Perry figure was the cohort-component model's own earlier output, so the comparison set the model against itself. The methodology page now reads these figures from the data.
Pupils with an EHC plan corrected: the old count was doubled
Every place page's "Total EHCPs" figure was twice the Department for Education's count. Blackburn with Darwen showed 2,768; DfE's 2024/25 figure is 1,384. The script reading DfE's file sen_phase_type_.csv added up every local authority row marked as the total for all phases and all school types, and that file carries two such rows for each authority, one for all schools and one for schools that are not hospital schools, each with the full count. The 132 areas then on the site summed to 649,346, more than DfE's 482,640 for the whole of England. The "EHCPs / 10k" rate was built on the doubled count, so it was twice too high as well; five-year growth compared two counts doubled in the same way, so it was right. The section now reads DfE, Special educational needs in England, academic year 2025/26, published 11 June 2026, data set "Pupils in all schools, by type of SEN provision", taking only the row for all schools. It shows pupils with an EHC plan and DfE's own percentage of pupils, in place of a rate per 10,000 residents of every age that DfE does not use: Blackburn with Darwen has 1,549 pupils with an EHC plan, 5.2% of pupils, and England 538,547, 6.0%. The 21 county councils, left out of the old file, are back, so district pages now show their county's figures.
Homepage 2011 White British share corrected to 80.5%
The homepage gave the White British share of England and Wales in 2011 as 80.0%. It weighted each area's 2011 share by its 2021 population, and it counted the eleven authorities created after 2011, whose earliest figure on this site is from the 2021 Census, inside the 2011 figure. Census 2011 table KS201EW gives 45,134,686 White British residents of 56,075,912 in England and Wales, 80.5%, which is the figure the homepage now shows. Both Census years are now summed from Census counts and weighted by their own population; the 2021 figure, 74.4%, does not change.
Shift-share national effect recomputed from the Census
Every place page's shift-share breakdown, and the finding on what drives ethnic change, gave the national effect as -6.4 percentage points. That figure was not computed from the 2011 Census: the site's 2011 population totals were all zero, because the script reading Census table KS201EW did not recognise its "All usual residents" row, and the calculation fell back to a typed 80% for 2011. With the real totals, the White British share across the 307 authorities that can be compared on the same boundaries fell from 79.7% in 2011 to 73.5% in 2021 (Census 2011 KS201EW, Census 2021 TS021), a national effect of -6.2 points. The age structure and local components move with it on every page, and Barnsley and Sheffield, whose 2011 figures had left out the Other ethnic group, are recomputed in full: Sheffield's total change is -6.3 points, not -9.5. Coventry's leading component changes from the national trend to the local one. The finding carries a dated correction.
One withholding rule for the counts, and a denominator with them
The homepage and the national page counted areas crossing below a 50% White British share, under a label promising the areas already below it. That is a flow counted as a stock, and it published 60 for 2051 while leaving out Birmingham, Leicester, Luton, Slough, Manchester and 22 others that had passed the threshold before the count began. The place pages also withhold a projected year where a residual Census group runs away past a quarter of the population at several times its 2021 share, so the site was counting Enfield in 2051 while Enfield's own page declined to show 2051 at all. Both rules now apply everywhere: 86 areas below 50% at 2051, across the 312 areas where 2051 is publishable, and 99 at 2061 across 255. The counts carry their denominator, because a count without one invites a comparison against the full 318. This is the figure the site's own finding has published since 13 August 2026. See pull request 58 and commit a9a98f0.
Four findings reconciled to v8.0 data
The findings on Birmingham, the Arab population, Roma and Gypsy or Irish Traveller, and the NEWETHPOP validation were still carrying figures from the model run that the 13 August 2026 recalibration replaced. Every number in them has been re-derived from the data the place pages and the homepage render. Birmingham's projection table now reads 32.8% White British in 2031, 22.9% in 2041 and 9.1% in 2061, the last of those 4.4 points above the figure it replaces, and its 2021 Christian share is 36.2% rather than the raw Census Muslim share it had been crossed with. Six of the eight Arab 2051 shares were re-derived, Ealing to 11.2% and Kingston upon Thames to 4.2%, and the national count is now the Census figure at source: 320,203 people in England, 0.57% of the population, from table TS021. The Roma and Gypsy or Irish Traveller national counts could not be derived from the projection file at all, because 49 of the 318 areas carry no 20-group detail, so they are now Census 2021 TS021 for England and Wales, 100,981 Roma and 67,768 Gypsy or Irish Traveller, 168,749 between them. Swale, at 0.61%, leads the Gypsy or Irish Traveller list it had been missing from. The NEWETHPOP accuracy distribution's over-10pp bucket reads 20 areas, 7% of the 296 compared. The 86-areas finding keeps its count and its table; its national-picture paragraph is regenerated on the site's own weighting at 74.4% in 2021 across all 318 areas and 56.3% by 2051 across the 312 where 2051 is publishable.
Exposure index regenerated, with an observed language input
The exposure index published one component that scored zero for every area. It read a projected growth in non-English speaking that no script in the pipeline writes, so the field was null for all 318 areas, the component contributed nothing, and a score described as five components was in practice four. The site models ethnic group and nothing else, so no honest projection of language exists to put there. The component now reads an observation instead: the share of residents aged 3 and over whose main language is not English, ONS Census 2021 TS029 via NOMIS, multiplied by 1.0 and capped at 20 points, so 20 points is a 20% share. Like the asylum support rate, it is a level rather than a change, and the index page says so. The whole file was regenerated at the same time, because the published scores had been computed on the projection vintage that model v8.0 replaced on 13 August 2026 and on the 320-area map that the Barnsley and Sheffield duplicate-code fix cut to 318. Between the three causes, 98 of the 318 areas move band, all of them upward, and the high exposure band grows from 17 areas to 59. The 42 areas entering that band do so on the language component: it gives them between 7 and 20 points, a median of 17, and not one of them would reach 70 on the other four components alone. Barnsley and Sheffield pick up the asylum support rate they had been scoring zero on, because the Home Office file keys them on codes the model does not use. The place pages also drop an item labelled projected non-English growth, which was the fall in the White British share with its sign reversed and measured nothing about language. See pull request 65.
Site audit and corrections
Reconciled the NEWETHPOP MAE figures across findings and methodology. National headline figures (White British share 2011 / 2021 / 2051 / 2061) are now computed live from the projection data rather than hard-coded. Added null-safety to per-place stat blocks so missing values render "n/a" rather than "undefined%". Constituency pages now carry a derived demographic snapshot (White British share, foreign-born share), weighted by postcode share and 2021 population. Removed AI-style writing tells (em-dashes, hype words, shouty caps). Expanded modelling acronyms (WBI, CCR, CWR, TFR, SNPP, IPF, MAE, NEWETHPOP, NINo, EAL, DC2101EW, RM021, TS021 / 22 / 27 / 29) in the glossary. Northern Ireland party colours added to the constituency template. Hardcoded "of 320" rank labels now derive from the dataset. Sources page expanded to list DWP Stat-Xplore, NOMIS and ONS births data.
Per-place MiniMap (Phase 3)
Place and constituency pages now show a thumbnail map highlighting the focal area against neighbouring boundaries. Local authority maps use the indigo brand colour for the focal; constituency maps use the 2024 General Election winner's party colour.
Model v8.0, recalibrated on an out-of-sample test
Cohort change ratios are now shrunk toward the national ratio by cell size and capped at 1.65 growth per decade. Both settings were selected on a new validation that fits ratios on Census 2001 to 2011 and forecasts 2021, scored against the actual Census 2021, so the fitting window never touches the target. On that test the model scores MAE 1.56pp on the White British share with a bias of +0.05pp, against 2.82pp and -2.13pp for the settings it replaces. The previous backcast score of 1.71pp has been withdrawn: it fitted on the same two Censuses it was tested against and was measuring the model's own guardrails, not its accuracy. It also had the direction wrong. The old settings were projecting change too fast, not too slow. Two code fixes land with it: births were being sized off the previous projection step's women for male cohorts, and the stochastic run carried its own separate copy of the ratio construction, which is why confidence bands failed to contain their own point estimate in 71% of area-years. Runaway long-horizon projections fall from 177 area-years to 19.
Coverage parity (P2)
Added 631 Westminster constituency pages with 2024 General Election results, sitting MPs and Personal Independence Payment claimant counts. Constituency-to-local-authority cross-links use postcode-share weights from the ONS pcon-lookup. Three new findings: 24-year NINo arrivals arc, working-and-not-working by passport group (Census 2021 RM021), and 2024 births by mother's country of birth.
Model v7.0
Census 2011 DC2101EW (18 groups, observed) replaces the previous proportional-splitting Census 2011 base, so per-area Hamilton-Perry CCRs now derive from direct Census observations at both ends of the calibration window. DfE School Census 2024/25 independent calibration applied to ages 0-5 (20% damped). Per-area MAE published on the methodology page. Backcast MAE reported at the time as 1.71pp against NEWETHPOP 2.58pp. That comparison was withdrawn on 13 August 2026: the backcast fitted on the same endpoints it was tested against, so it measured the guardrails rather than accuracy. See the v8.0 entry above.
UK Demographics launch
Initial release: 320 local authority demographic profiles with ethnic projections to 2061. Hamilton-Perry v6.0 model, 20 ethnic groups, Census-direct base, 1,000 Monte Carlo simulations with sigma 0.02 generating 80% and 95% confidence intervals.