Full index
All 25 entries
The complete roster, grouped by section. Section pages carry the same lists with their own introductions.
Three at once, and they do not agree.
- Three definitions, one populationA legal test, a survey question and a benefits assessment are three different instruments measuring three different things, and all three get reported as 'the number of disabled people'.
- The legal testWhat the statutory definition actually says, and the parts of it that do the work.
- The survey definitionSelf-reported, and deliberately so.
- The assessmentA functional test built for a different purpose entirely.
- Why they divergeThree instruments, three populations, one headline.
The wording that decides the number.
- The wording decides the numberChange 'long-standing illness' to 'condition expected to last 12 months' and the count moves by millions.
- Filter questionsWhat gets asked first, and who is routed past the rest.
- Self-reportThe count depends on identification, which depends on the question.
- Comparability over timeWhy a series breaks when the question is improved.
Who publishes what, and how often.
- Who publishes whatSeveral bodies publish disability figures on different schedules from different instruments. Naming the source is not optional.
- The censusOnce a decade, near-complete coverage, one question.
- The continuous surveysSmaller samples, more often, with confidence intervals that matter.
- Administrative dataCounts of people in a system, which is not a count of people.
The differences measured year after year.
- The employment gapThe most-quoted single figure in the field: the difference in employment rate between disabled and non-disabled people, and what it does and does not tell you.
- The pay gapMeasured on medians, and sensitive to who is in work at all.
- Extra costEstimates of what living with a condition costs, and why they vary so widely.
- Intersecting countsWhat happens to sample size when two characteristics are crossed.
Who is responsible for which part.
- Who is responsible for whatStatistics, regulation, enforcement and policy sit in different places, which is why no single body can answer most questions.
- The statistics authorityWhat designation as an official statistic actually means.
- DevolutionFour administrations, four sets of figures, one UK headline.
- Select committeesWhere published evidence gets tested in public.
Where a figure comes from, and how to read it.
- Reading a bulletinWhere the caveats live in a statistical release, and why they are the interesting part.
- Confidence intervalsA number with a range around it, and what a change smaller than the range means.
- RevisionsFigures change after publication. How that is signalled.
- Citing a figureSource, table, year. Without all three, a number is not usable.