School Connection / Feature
Similar is not neutral: what Ofsted’s 50-school comparator can and cannot tell a board
A new IDSR comparison can reveal performance hidden by national averages. It is a useful question-generator, but neither a neutral description of a school nor a verdict on its effectiveness.
The answer in brief
The comparator can sharpen a board’s questions, but it cannot answer them on its own.
From 1 September, a mainstream school’s inspection data summary report can compare its achievement with a unique set of 50 schools selected for similarity of context. Ofsted describes this as an additional layer of information for inspection discussion, sitting alongside existing data and evidence gathered on site. It explicitly says the comparison will not be used to make grading decisions. That is the first boundary every board should retain.12
A school below the national average but strong among its contextual peers may have expertise worth understanding. One above the national average but behind similar schools may have unrealised capacity. Neither finding proves cause or effectiveness. It identifies where inquiry may be more productive, while absolute outcomes, trends and direct evidence of pupils’ experience remain essential.12
- Treat the comparator as a prompt for investigation, not an inspection forecast.
- Read peer position beside absolute outcomes, trends and the strength of the match.
- Ask what the model leaves out as carefully as what it includes.
01 / What changed
The comparator is now part of inspection preparation, but it is not a new grading formula.
The 1 September IDSR release introduced the similar-schools comparison for achievement alongside refreshed census, workforce, pupil-stability and alternative-provision data. Ofsted describes its purpose as further contextualising achievement outcomes and supporting discussion about achievement and pupils’ progress.1
Inspectors still consider the wider IDSR and the evidence they gather through visits, pupils’ work and conversations. The Department for Education’s planned school-facing platform will also provide the list of 50 for benchmarking and connection, but it will not be publicly accessible. A board should therefore resist treating the peer result as a new public league table.12
02 / How matches are built
The model identifies statistical neighbours for attainment, not institutional twins.
The model first uses regression analysis to identify contextual factors strongly associated with attainment. It then applies a weighted Euclidean-distance calculation to find each school’s statistical nearest neighbours. Matching uses whole-school characteristics rather than only the examination cohort, a choice intended to make peer groups more stable. The initial calculation uses 2025 data.3
Inputs include prior attainment, school size, pupil stability, disadvantage, neighbourhood deprivation, local higher-education participation, SEND, education, health and care plans, and English as an additional language. Prior attainment carries the largest single weight: 0.46 in the KS4 model and 0.38 at KS2. Schools within one set can nevertheless differ appreciably on individual characteristics.23
03 / Why similar is not neutral
Every model contains decisions about which similarities matter and which do not.
This model is built specifically for attainment. The technical note says that the drivers of attendance or financial health may be different. Attendance and behaviour policy were excluded because schools can influence them. Ethnicity was excluded because it is a protected characteristic and added only modest explanatory power. These are explicit design choices, not a complete description of school context.3
Rural access, specialist provision, local labour markets, staffing disruption or a recent leadership change may matter greatly to a board without determining the official match. Conversely, a factor included in the model may be associated with attainment without causing it. Similarity is therefore purpose-specific and statistical; it should not be mistaken for sameness.23
04 / Model fit is not a verdict
Strong aggregate explanatory power does not validate a conclusion about one school.
The published model explains around four-fifths of the variation in KS4 attainment between secondary schools, compared with around one-third at KS2. DfE and Ofsted caution that the lower figure at KS2 does not make the comparison unreliable; it means that attainment varies more within the matched primary groups.3
Even strong model fit across England cannot explain why one named school achieved a particular result. Association is not causation, and a peer average is not an entitlement, target or predicted score. The promised information about match strength matters: a loose match should support more cautious conclusions than a tight one.23
05 / A disciplined board response
Use three frames together: absolute attainment, contextual comparison and direction over time.
A reversal between national and peer positions is a signal to examine curriculum, teaching, attendance and inclusion evidence, not an instruction to celebrate or condemn immediately. Leaders should show the board the absolute outcome, the distribution among similar schools, the school’s trend and the strength of the statistical match.12
Leaders should also be able to explain distinctive circumstances the model cannot see, while avoiding the opposite danger of explaining away persistent gaps. The strongest response is specific: identify a peer-relative strength to preserve, a gap to investigate, the evidence required before action and the date on which the board will review it.23
Leadership questions
Questions that turn the development into a governing conversation.
Board test
Which variables most strongly shape our peer group, and how strong is the resulting match?
Board test
Where does our national position differ materially from our position among similar schools?
Board test
What important local, organisational or recent factor is absent from the model?
Board test
What practice will we investigate, and what further evidence will we require before changing strategy?
School Connection intelligence lens
What the national Observatory can add, and where it must stop.
The Schools Intelligence Observatory can run a Peer Context Gap test with participating schools’ permission. It would securely ingest the peer list and match-strength information from each contributed IDSR, then calculate national percentile, peer-group percentile and the difference between them for each attainment measure. It would also report the peer median, interquartile range, three-year direction and cohort size, and flag movement of at least one quartile between national and peer views.123
The test should include a sensitivity check that removes the five weakest matches and should analyse phase and disadvantage patterns only in appropriately sized aggregates. The result would be a contextual comparison, not an effectiveness measure or causal estimate.3
Public reporting must not disclose a non-public peer list or identify another school from a contributed IDSR without permission. It must not convert peer percentile into an unofficial grade, and it must suppress small or disclosure-prone cohorts. A higher or lower peer position should be described as a question for investigation, never as an expected outcome.12
Only human-approved, public-safe intelligence can appear here. Private candidates, commercial signals, contacts, opportunity values and internal scores are never exposed through School Connection.
Evidence boundary
What this analysis does not prove.
- Initial matches rely on 2025 whole-school data and use prior-attainment inputs affected by COVID-era assessment gaps.
- Aggregate predictive fit does not establish a causal explanation or expected outcome for an individual school.
- The model covers attainment in mainstream schools; it does not cover every outcome or include special schools, alternative provision or pupil referral units.
What we are monitoring next
Publication is the beginning of the watch.
- Ofsted inspector-training material showing how the comparator is used in inspection preparation and discussion.
- The autumn rollout of DfE’s school-facing platform and the way it presents peer lists and match strength.
- The first annual model refresh, including changes to variables, weights, data availability and peer-group stability.
Approved public intelligence
What the live evidence is showing now.
School Connection continues to monitor standards, curriculum & accountability evidence. New machine-detected signals remain private editorial candidates until a human editor investigates and approves them for publication.
This panel reads only the editor-approved School Connection public feed. It never exposes raw Observatory records, private candidates, contacts, commercial opportunities or internal scores.
Sources and methodology
Evidence used in this analysis
School Connection links to the primary source behind each material claim. Source status, period and limitations are stated so readers can reproduce the evidence trail.
- 01
Ofsted · Updated 1 September 2026
IDSR: news and updates
Confirms that the comparator went live in IDSRs on 1 September 2026, uses a unique group of 50 similar schools and will not be used to make grading decisions. - 02
Department for Education and Ofsted · Updated 16 July 2026
Similar schools model for attainment: overview
Explains the intended use in inspection and school improvement, the contextual inputs, the non-public school-facing platform and the planned annual update cycle. - 03
Department for Education and Ofsted · Updated 16 July 2026
Similar schools model for attainment: technical note
Provides the model variables and weights, exclusions, statistical matching method, coverage, model-fit findings and cautions about interpretation.