dialuk.info

Threshold — how disability is defined, counted and governed in the United Kingdom.

Counts

The Census

Entry in CountsSources named in full

Papers, forms and reference volumes on a desk
Once a decade, near-complete coverage, one question.

A decennial snapshot of almost everyone

The census runs once a decade and, for disability statistics, its great virtue is also its great limitation: near-complete coverage of the resident population produces numbers that no continuous survey can match in precision at local level, but the count rests on a single question asked once every ten years.

In England and Wales, the Office for National Statistics has run the census since 1841. The 2021 census — held on 21 March 2021, during the COVID-19 pandemic — asked a two-part disability question derived from the Government Statistical Service harmonised standard. Respondents were first asked whether their day-to-day activities were limited by a health problem or disability expected to last twelve months or more; if yes, they were asked whether the limitation was a lot or a little. Scotland's census, run by National Records of Scotland, was further delayed to 2022 and used comparable but not identical wording. Northern Ireland, counted by NISRA, ran its census in 2021 with similarly harmonised questions.

Papers, forms and reference volumes on a desk
Every published figure sits in a numbered table with a stated base population and reference period. The headline summarises that; it does not replace it.

Photo: Mikhail Nilov / Pexels

The 2021 census for England and Wales recorded approximately 10 million people — around 17.8 per cent of the usual resident population — as disabled under the harmonised question. That figure carries a precision no survey sample can approach at small-area level: local authorities, parliamentary constituencies and output areas all receive disability counts that the Annual Population Survey, with its far smaller sample, simply cannot resolve.

Grey archive boxes on a metal rack
Fig. 2Coverage, cadence and geography differ by instrument, so two totals for the same subject rarely describe the same population.

What the census cannot do is track change. A decade between counts is too long to observe the effect of a policy shift, an economic shock or an ageing population moving through its seventies. The question wording has also changed across censuses, which means comparability over time is limited even within the official series. The 2001 census used different framing from 2011, and 2011 differed from 2021; ONS publishes guidance on each discontinuity, but the breaks are real.

The census definition is not the legal definition under the Equality Act 2010, and it is not the functional assessment used for benefits. It is a self-report of activity limitation, harmonised across government surveys to permit consistent comparison — comparison across place, rather than across time.

More in Counts