Attendance and missing education
Education chapter 6. How many children are not in class, the four different ways that is counted, and why those four numbers must never be added together.
Four separate statistics describe children who are not in a classroom, and each counts something different. This chapter sets out all four, keeps them apart, and is explicit that the most important number — how many children are receiving no suitable education at all — is not among them.
Absence
18.1% of pupils were persistently absent in 2024/25 — missing a tenth or more of their sessions. In secondary schools the figure was 23.4%, and in special schools 35.3%.
2.4% of pupils were severely absent in 2024/25, missing at least half of their sessions.
Persistent absence means missing 10% or more of sessions — roughly one day a fortnight. The threshold was 15% until 2015/16, so the recorded rate rose sharply on the definition change alone. Figures either side of that year are not the same measure.
Special schools
30.530.631.430.33028.429.426.527.526.928.529.628.848.940.438.337.235.306/07/08/09/10/11/12/13/14/15/16/17/18/20/21/22/23/24/State-funded primary
14.214.414.614.113.210.4118.18.48.28.38.78.28.817.716.214.61306/07/08/09/10/11/12/13/14/15/16/17/18/20/21/22/23/24/State-funded secondary
24.922.822.520.919.516.816.513.613.813.113.513.913.714.827.726.525.623.406/07/08/09/10/11/12/13/14/15/16/17/18/20/21/22/23/24/Total
19.318.518.517.416.313.513.610.71110.510.811.210.912.122.521.22018.106/07/08/09/10/11/12/13/14/15/16/17/18/20/21/22/23/24/
Faded bars are years before the definition changed in 2015/16. They are shown because the data exists, and faded because they are not the same measure.
Source: DfE, Pupil absence in schools in England. Percentages are of enrolments.
View as table
| Period | Special schools | State-funded primary | State-funded secondary | Total |
|---|---|---|---|---|
| 2006/07 | 30.5% | 14.2% | 24.9% | 19.3% |
| 2007/08 | 30.6% | 14.4% | 22.8% | 18.5% |
| 2008/09 | 31.4% | 14.6% | 22.5% | 18.5% |
| 2009/10 | 30.3% | 14.1% | 20.9% | 17.4% |
| 2010/11 | 30% | 13.2% | 19.5% | 16.3% |
| 2011/12 | 28.4% | 10.4% | 16.8% | 13.5% |
| 2012/13 | 29.4% | 11% | 16.5% | 13.6% |
| 2013/14 | 26.5% | 8.1% | 13.6% | 10.7% |
| 2014/15 | 27.5% | 8.4% | 13.8% | 11% |
| 2015/16 | 26.9% | 8.2% | 13.1% | 10.5% |
| 2016/17 | 28.5% | 8.3% | 13.5% | 10.8% |
| 2017/18 | 29.6% | 8.7% | 13.9% | 11.2% |
| 2018/19 | 28.8% | 8.2% | 13.7% | 10.9% |
| 2020/21 | 48.9% | 8.8% | 14.8% | 12.1% |
| 2021/22 | 40.4% | 17.7% | 27.7% | 22.5% |
| 2022/23 | 38.3% | 16.2% | 26.5% | 21.2% |
| 2023/24 | 37.2% | 14.6% | 25.6% | 20% |
| 2024/25 | 35.3% | 13% | 23.4% | 18.1% |
The threshold changed in 2015/16
Persistent absence means missing 10% or more of sessions — roughly one day a fortnight. The threshold was 15% until 2015/16, so the recorded rate rose sharply on the definition change alone. Figures either side of that year are not the same measure.
Absence is sessions, not children
A pupil who misses a fortnight contributes twenty sessions to the absence figures, not one. That is why an absence rate cannot be read as a number of children, and why it cannot be added to the home-education or missing-education counts further down this page.
Educated elsewhere, or nowhere
126,000
Electively home educated
Autumn term, 2025/26
34,700
Identified as missing education
Children local authorities have found, not children who are missing
+56%
Rise in home education
Across the published series, which begins in 2022/23
126,000 children were recorded as electively home educated in the autumn term of 2025/26, up from 80,900 in 2022/23 — a rise of 56%.
Elective home education is lawful and requires no permission: a parent may educate at home provided the education is suitable. Local authorities have no duty to monitor it routinely and no register of it existed nationally before this collection, so the series is short and starts from a low base of what authorities knew.
Source: DfE, Elective home education. Three census points each academic year, each labelled with its term.
View as table
| Period | Children recorded as electively home educated |
|---|---|
| 2022/23 Autumn | 80,900 |
| 2022/23 Spring | 86,200 |
| 2022/23 Summer | 97,600 |
| 2023/24 Autumn | 92,000 |
| 2023/24 Spring | 103,500 |
| 2023/24 Summer | 117,900 |
| 2024/25 Autumn | 111,700 |
| 2024/25 Spring | 123,300 |
| 2024/25 Summer | 137,200 |
| 2025/26 Autumn | 126,000 |
Why families say they moved to home education, 2025/26
Shares, not counts. At national level the Department publishes the percentage of home-educated children in each reason category but suppresses the underlying headcounts, so these are the distribution and cannot be converted back into numbers of children.
- Unknown17%
- Mental health16%
- Philosophical or preferential12%
- No reason given12%
- Other11%
- Lifestyle9%
- School dissatisfaction general8%
- School dissatisfaction SEND4%
- School dissatisfaction bullying3%
- Physical health2%
The largest category is that nobody knows
The two largest entries are “unknown” and “no reason given”. Local authorities record the reason a parent offers, where a parent offers one, and there is no duty to give one. A distribution in which the biggest slice is an absence of information is worth reading as a fact about the collection rather than about families.
Three census points a year, not one
Both of these are counted three times an academic year — autumn, spring and summer. Every point below is labelled with its term, because the counts rise across a school year and comparing an autumn figure with a summer one is not comparing like with like.
Missing education counts identification, not absence
A child missing education is of compulsory school age, not on any school roll, and not receiving suitable education otherwise. It is a count of children local authorities have identified at a census point — which means it counts identification, and a child nobody has identified is by construction absent from it.
A child missing education is of compulsory school age, not on any school roll, and not receiving suitable education otherwise. It is a count of children local authorities have identified at a census point — which means it counts identification, and a child nobody has identified is by construction absent from it.
Source: DfE, Children missing education. Three census points each academic year, each labelled with its term.
View as table
| Period | Children identified as missing education |
|---|---|
| 2022/23 Autumn | 24,700 |
| 2022/23 Spring | 24,700 |
| 2022/23 Summer | 28,100 |
| 2023/24 Autumn | 33,000 |
| 2023/24 Spring | 36,700 |
| 2023/24 Summer | 36,600 |
| 2024/25 Autumn | 39,200 |
| 2024/25 Spring | 34,400 |
| 2024/25 Summer | 34,100 |
| 2025/26 Autumn | 34,700 |
Association, not cause
Absence, exclusion and home education move together over this period, and each is associated with the others in the published data. None of that establishes which causes which, and plausible accounts run in both directions — exclusion can precede home education, and persistent absence can precede both.
The Education Visibility Gap
Every chapter in this section runs into questions its data cannot answer. Gathered in one place, those questions stop looking like a series of individual caveats and start showing a shape: the education statistics measure the system’s own activity in great detail, and what happens to individual children considerably less. This register is maintained as data — each chapter declares its gaps, and the build fails if one ships without them.
Is a child actually receiving the provision named in their EHC plan?
No national dataset records delivery. The statistics record that plans exist and what is written in them, not what happens in classrooms. This is the largest single gap in the SEND evidence base: the legal duty in section 42 is to secure provision, and nothing national measures whether it was secured.
What would close it: A national collection of provision delivered against provision specified, at pupil level.
Chapter: Special educational needs and disabilities →Is the underlying prevalence of need rising?
The statistics count plans, requests and recorded needs — the system’s activity. A rise is consistent with rising need, better identification, changed thresholds, or population growth, and the published data does not separate them.
What would close it: A population-level prevalence study independent of the assessment system.
Chapter: Special educational needs and disabilities →How many children are receiving no suitable education at all?
The children-missing-education figure counts children a local authority has identified. A child nobody has identified is by construction absent from it, so the statistic measures the reach of identification as much as the size of the problem, and cannot be read as a total.
What would close it: A cross-checked national register of children of compulsory school age against school rolls and known home education.
Chapter: Attendance and missing education →What did schools stop doing to balance their budgets?
A balanced budget records that a school did not overspend. Nothing national records what was reduced or withdrawn to achieve it.
What would close it: A collection of provision withdrawn, which no administration has proposed.
Chapter: What education costs →How much did children learn before the late 1980s?
National testing, published results and a national curriculum all arrive with the Education Reform Act 1988. Before it, comparable national attainment statistics largely did not exist, so long-run questions about English education cannot be answered — not because the data is hard to find but because it was never collected.
What would close it: Nothing available. The measurements were not taken at the time.
Chapter: Eighty years of education →
How were appeals to the SEND tribunal decided?
The Ministry of Justice publishes appeals received and appeals disposed of, but not the outcome split. The share of appeals upheld is therefore absent from this site even though the cases were decided.
What would close it: MoJ to publish disposal outcome by jurisdiction, as it already does for some other tribunals.
Chapter: Special educational needs and disabilities →What did local authorities actually spend on high needs, and did it balance?
The high needs national funding formula tables publish allocations. Outturn and any deficit carried on the high needs block sit in each authority’s section 251 return and audited accounts, not in a national comparable series.
What would close it: A published national series of high needs outturn and reserve position by authority.
Chapter: Special educational needs and disabilities →How many children are home educated, by reason?
At national level the Department publishes the percentage in each reason category but suppresses the underlying headcounts. The distribution is available; the numbers behind it are not. The two largest categories are “unknown” and “no reason given”.
What would close it: Publication of national reason counts, or a collection in which giving a reason is part of the process.
Chapter: Attendance and missing education →Where do school leavers go next?
Destination measures are published only at institution level, in a single file of over half a gigabyte. National totals are not separately published, and summing institution rows would silently drop suppressed small cells and produce a national figure that is wrong.
What would close it: A national-level destinations dataset alongside the institution-level file.
Chapter: Are children learning more? →How many educational psychologists are there in total?
The workforce census counts those employed centrally by local authorities for 28 days or more. Those working privately, through agencies or in other arrangements are outside it, so the published figure is a subset of the profession and not its size.
What would close it: A register of the profession rather than a count of one employment route.
Chapter: Teachers and the workforce →How is the academy sector structured above school level?
The census records each school’s type but not, in this dataset, which multi-academy trust it belongs to or how large that trust is. The governance question most often asked — who actually runs a group of schools — is a level above what this collection carries.
What would close it: Trust-level structure published alongside school-level type in the same release.
Chapter: Who runs our schools? →
How does England compare with Scotland, Wales and Northern Ireland?
Four devolved frameworks with four legal definitions. Scotland’s Additional Support Needs is far broader than an English EHC plan; Wales is mid-transition from SEN to ALN and its recorded count more than halved on the definition change alone.
What would close it: A common measure none of the four administrations has an incentive to adopt.
Chapter: Special educational needs and disabilities →How does exclusion in England compare with the other three nations?
Four devolved frameworks with different terminology and different legal tests. StatsWales, which carries Welsh attendance, attainment, workforce and ALN data used elsewhere in this section, publishes no exclusion dataset at all — 786 dataset titles searched, no match.
What would close it: Welsh exclusion data as a machine-readable dataset, and a common definition none of the four administrations has adopted.
Chapter: Behaviour and exclusion →Are children in England learning more than they used to?
The headline measures were rebuilt four times inside the period they cover: Attainment 8 introduced in 2016 and rescored in 2018/19, GCSE grading moved from letters to 9–1 across 2017–2019, and two years of grades awarded without examinations. A trend line through them traces assessment policy at least as much as attainment.
What would close it: A measure held constant across the period, or a formal concordance between the successive measures that DfE has not published.
Chapter: Are children learning more? →Has the definition of disadvantage stayed the same?
It has not. Free-school-meal eligibility rules changed during the period, including transitional protection during the move to universal credit. A change in who counts as disadvantaged moves the index without anything changing for any child.
What would close it: A back-series recalculated on a single consistent eligibility definition.
Chapter: Who gets left behind? →Did school spending rise or fall in real terms?
Deflating cash spending requires choosing an index, and the GDP deflator, CPI and schools-specific cost indices disagree about the direction over the last decade. This is a live dispute among people who do it professionally, not an oversight, so this site publishes cash and names the institutions that publish real-terms series with their deflator stated.
What would close it: Agreement on an education-specific deflator, which does not exist.
Chapter: What education costs →Are more schools in deficit than a decade ago?
The share rose from 6.0% to 17.8%, but the population being measured shrank throughout: schools converting to academy status leave the collection, and the schools that left were not a random sample of those that stayed. The rise is real in the published series and cannot be cleanly attributed.
What would close it: A consistent financial series spanning maintained schools and academy trusts on the same basis.
Chapter: What education costs →
Do local authorities that decide faster produce better outcomes?
Timeliness is published per authority; outcomes are published for school pupils by SEN status nationally. The two are not joined at a level that would let one be regressed on the other, and the populations differ — the EHC-plan release is all-age 0–25, the outcome data is school pupils.
What would close it: Linked pupil-level data across the two collections, with the population difference handled.
Chapter: Special educational needs and disabilities →What happens to a child after a permanent exclusion?
The exclusion release records the decision. Whether the child reached alternative provision, how long that took, and what followed are not joined to it in any published dataset.
What would close it: Linked pupil-level data across the exclusion, alternative-provision and destinations collections.
Chapter: Behaviour and exclusion →What is the pupil–teacher ratio, on a basis this site would stand behind?
It needs the pupil count and the teacher count on the same census date with the same school coverage, and the two collections do not align cleanly enough. The Department publishes its own ratio; this site does not compute a competing one.
What would close it: Aligned census dates and coverage between the school census and the workforce census.
Chapter: Teachers and the workforce →
Would this child have done better with different support?
This requires a counterfactual that no observational dataset contains. Comparing children with plans to children without measures the difference between the groups — the groups are defined by the assessed presence of need — and not the effect of the support.
Chapter: Special educational needs and disabilities →Did any particular policy narrow the disadvantage gap?
The gap index is a single national series with no comparison group and no policy variation to exploit. Attributing movement in it to a named intervention would need a design this data cannot support.
Chapter: Who gets left behind? →Did becoming an academy improve a school?
Schools were not assigned to academy status at random. Sponsored academies were converted because they were judged to be underperforming; converter academies chose to convert, often from a stronger position. Comparing the two groups compares populations selected on opposite criteria, which is why serious evaluations disagree with one another.
Chapter: Who runs our schools? →Will any of these proposals work?
Precedents narrow the range of reasonable assumptions and do not settle the question. Secondary technical schools were never built at scale in the 1940s and university technical colleges have a mixed record, but neither determines what a new technical route would do.
Chapter: Where education goes next →
A gap is listed only where a reader would reasonably expect an answer. Questions nobody would think to ask are not gaps, and neither are questions answered elsewhere in the section. The last category, “not answerable by observation”, is kept separate from the rest on purpose: a question requiring a counterfactual is a limit of what any dataset could show, and filing it alongside a genuine collection failure would let the collection failure hide behind it.
24 entries currently registered across the section.
Method
Every figure is computed, none is typed in
Each statistic is generated at build time from the Department for Education’s published datasets, through a claims layer recording the inputs, the calculation, the counting unit and the date verified.
View as table
| Claim | Statement | Unit | Calculation |
|---|---|---|---|
| persistent-absence-latest | 18.1% of pupils were persistently absent in 2024/25 — missing a tenth or more of their sessions. In secondary schools the figure was 23.4%, and in special schools 35.3%. | sessions | Published percentages of enrolments, quoted directly |
| severe-absence-latest | 2.4% of pupils were severely absent in 2024/25, missing at least half of their sessions. | sessions | Published percentage of enrolments |
| home-education | 126,000 children were recorded as electively home educated in the autumn term of 2025/26, up from 80,900 in 2022/23 — a rise of 56%. | children | (latest − first) / first × 100 across the published series |
| children-missing-education | 34,700 children were identified as missing education in the autumn term of 2025/26. | children | Published count at census |
| permanent-exclusions | There were 9,906 permanent exclusions in 2024/25, a rate of 0.12 per hundred pupils. In 2006/07 there were 8,678. | pupils | Published counts and rate; change across the series is (latest − first) / first × 100 |
| suspensions-concentration | There were 913,000 suspensions in 2024/25, but they fell on 326,674 pupils — an average of 2.8 suspensions each for those suspended at all. | episodes | suspensions ÷ pupils suspended at least once. Both are published; the ratio is not |
| exclusion-top-reason | The most common reason recorded for a permanent exclusion in 2024/25 was persistent disruptive behaviour, cited 5,254 times. | pupils | Largest published reason category |
Absence is measured in sessions — half-days — against enrolments. Children missing education and elective home education are headcounts of children at a census point. The three are different units over different populations and must never be added together: a child can appear in more than one.
Last verified 2026-08-17. Data: DfE, Pupil absence in schools in England; Children missing education; Elective home education.
Related chapters
- In this sectionBritain vs EuropeWhat 15-year-olds across Europe reported about skipping school — a pupil-reported measure, not a register, and the only absence figure that compares across countries.
- In this sectionBehaviour and exclusionSuspensions and permanent exclusions — the other route out of a classroom.
- In this sectionSpecial educational needsAbsence is markedly higher in special schools and among pupils with a plan.
- In this sectionEducationThe other chapters, and how this section is built.
- Leaves this sectionPublic financesWhere education spending sits in total government spending.
Scope and definitions
England, not the UK
Education is devolved. Every figure on this page is England only. Scotland, Wales and Northern Ireland operate different frameworks with different legal definitions, and their figures are not directly comparable.
Four different units, never one total
Absence is measured in sessions — half-days — against enrolments. Children missing education and elective home education are headcounts of children at a census point. The three are different units over different populations and must never be added together: a child can appear in more than one.