Devolution

Four administrations, four sets of figures, one UK headline — and the joins rarely line up cleanly.

Four administrations, one question

The United Kingdom contains four separate statistical administrations. The Office for National Statistics publishes figures for England, and — through the Annual Population Survey and its successors — for Wales, Scotland and Northern Ireland too, but the devolved administrations also commission and publish their own analyses. Scottish Government statisticians produce disability breakdowns drawing on the same survey frame but presenting them under their own release schedule. The Welsh Government does likewise. The Northern Ireland Statistics and Research Agency operates largely independently, running surveys such as the Continuous Household Survey alongside its use of shared UK frames.

The headline UK disability employment gap, for instance, is built from a pooled Labour Force Survey dataset. Disaggregate it by nation and the confidence intervals widen sharply, because the sample for Northern Ireland or Wales alone is small enough that confidence intervals around single-year estimates can exceed the gap being measured.

Where the seams show

Devolution creates a specific comparability problem. The devolved administrations can — and do — ask supplementary questions or run separate surveys using different wording. When Scotland or Wales uses a differently framed disability question in a standalone publication, that figure is not directly comparable with the ONS headline, even if the underlying population is drawn from the same survey instrument. The wording decides the number, and different wording means a different number.

Responsibility for which statistic is designated official, and under what code, sits with the UK Statistics Authority for ONS outputs, but the devolved equivalents are overseen by their own chief statisticians and operate under devolved arrangements — Statistics for Wales, for example, has its own professional head. A number published by the Scottish Government carries its own designation status, which may or may not match an equivalent ONS release.

The result is that a researcher citing a UK-level figure and one citing a Scotland-level figure from the same reference year may be working from instruments that differ in question wording, sample design, weighting methodology and publication frequency — without the headline numbers advertising any of that.