Before deciding what to teach next from an assessment score, read the publisher's confidence information for that exact measure and edition. Keep the reported score beside its uncertainty, administration conditions and missing evidence. If the report supplies no confidence information, record that gap and consult the manual or publisher. Don't invent a band or turn a result into a fixed judgement about a pupil.
This guide offers a score-reading note for assessment leads, teachers, SENCos and SLT in England. Use it when preparing a pupil discussion or report. It isn't a new testing requirement or a method for calculating intervals. The aim is to make the next teaching decision proportionate to what the evidence can support.
Keep the explanation attached to the result
Ofqual's introduction to reliability explains why performance can vary with the occasion, questions and marking. Measurement error is part of interpreting assessment evidence; a clerical mistake is a separate issue to correct. Its historical examples aren't instructions for today's statutory assessments.
The publisher's documentation tells you how that uncertainty is represented for its measure. For example, GL Assessment's NGRT Digital technical information distinguishes the reliability of a test from the standard error of measurement used when interpreting an individual's score. These concepts answer different questions. A strong overall reliability statistic doesn't make one result exact.
Read any reported confidence band with its stated level, scale, method and cautions. Don't carry a band from another test, edition or subscale onto this score. Don't describe the level as a probability that this particular pupil has a fixed ability within the displayed range. Keep the publisher's explanation available and ask for clarification where its wording is unclear.

A confidence band doesn't cover every possible problem with a result. Incorrect pupil details, an unsuitable test, interrupted administration or support that changes the task still need investigation. The NGRT Digital teacher guide, for example, gives specific administration instructions, including a quiet environment and limits on help with answers. Check the instructions for the assessment actually used.
A score-reading note to complete before reporting
Use these prompts in the school's approved assessment record. They are an editorial template, not a publisher form or Student Radar fields. Keep the original report and its version available to authorised colleagues. Record the actual teacher and assessment lead by name; a job title alone doesn't identify who accepted this review.
- Decision and purpose
- What teaching or reporting question needs an answer? State what this score can inform and what remains outside its scope.
- Source and version
- Record publisher, test, edition or form, test date, report date and version, score type, relevant reference group and the manual section used. Check pupil details and any conversion table.
- Conditions and curriculum
- Record the task, content sampled, teaching already encountered, independence, support and adjustments used, interruptions and departures from instructions. Identify who observed each detail.
- Uncertainty and gaps
- Retain the publisher's reported band and stated level where supplied, with its explanation. State missing responses, unavailable confidence information and unconfirmed conditions. Don't replace missing evidence with a zero or an assumed normal condition.
- Provisional judgement and reviewer
- Write the limited interpretation, its evidence and the reversible teaching action. Record the responsible professional's name, review date, judgement version and status.
- Review and correction
- Set the next review point and the evidence that could change the judgement. Explain how the pupil, parent or another professional can question the interpretation or correct a source. Preserve the earlier account and attribute any amendment.
If confidence information is unavailable, say so in the note and in any explanation that depends on precision. The assessment lead should seek the original report or publisher guidance before making a fine distinction. Meanwhile, teaching can respond to observed classroom needs without pretending that the missing technical evidence has arrived.
Read the note back against the teaching decision
This made-up record illustrates the process; it contains no real pupil or school data. A teacher is deciding whether to revisit vocabulary in the next reading sequence. A standardised reading result has been copied into a meeting summary, but the publisher's confidence information and report version haven't travelled with it. The summary can't support a precise claim about the pupil's standing.
The teacher retrieves the original report with the assessment lead. Curriculum records show that the pupil missed some teaching in the vocabulary sequence; the support record doesn't confirm whether the agreed classroom preparation was delivered. Attendance, curriculum and support evidence change the question: what teaching has the pupil had a fair opportunity to access, and what can the teacher check next?
The teacher records their name as reviewer and keeps the judgement provisional. They arrange the missed teaching and an ordinary classroom check, then review the response with the assessment lead after that sequence. The pupil can explain their experience through their usual communication route. The parent can add context through the school's agreed contact. Corrected report details, confirmed support delivery or different classroom evidence would reopen the interpretation sooner.

Those records don't prove why the score occurred or diagnose a need. A later improvement wouldn't establish that the extra teaching caused it. Curriculum exposure, attendance, delivery, practice effects, maturation, conditions and measurement error can contribute. Classroom support also doesn't automatically confer formal access arrangements; follow the separate rules for the particular regulated assessment where relevant.
Leave the unsupported conclusion out of the report
Ofqual's Handbook connects validity to the interpretation and intended use of outcomes. Its conditions govern regulated qualifications and awarding organisations; this school note is a suggested practice, not a regulatory requirement. A score or band alone can't establish ability, effort, need or future attainment, or authorise automatic grouping, referral or a high-stakes recommendation.
Before reporting statistically significant change, consult the publisher's comparison method and assumptions for those measures. Don't use overlap or non-overlap between displayed confidence bands as an automatic test of statistically significant change. If two results disagree, use the separate assessment disagreement guide to examine their purposes and comparability. This note starts earlier, with the information needed to read one result responsibly.
For cohort reporting, retain denominators, coverage and missing results. Small cohorts can be unstable and identifiable: aggregate or suppress where needed, avoid spurious precision and explain that a pattern may not generalise. Don't publish individual confidence details in a small-group summary that allows a pupil to be identified.
Where enabled and authorised, Student Radar supports recording standardised scores and using publisher conversion tables for supported tests and editions. Keep the original publisher report and this review note in the school's approved record. Don't assume a score-entry screen calculates confidence intervals or retains successive test sittings. The teacher owns the interpretation and correction route. Explore assessment tracking or the statutory and standardised assessment workflow with that reporting boundary in view.
Sources and further reading
Original sources read on 17 November 2026. Publisher examples concern their specific tests.
- GL Assessment: NGRT Digital, Guidance and information for teachers, fifth edition, copyright 2021; still linked by the publisher's downloads page. Administration and report interpretation sections.
- GL Assessment: NGRT Digital Technical Information, 2018 copyright, publisher's 2025-updated file; test reliability section. No interval or numerical threshold from it is reproduced here.
- Ofqual: Introduction to the concept of reliability, published 16 May 2013; explanatory background applying to England, Northern Ireland and Wales.
- Ofqual Handbook: Section D, general requirements for regulated qualifications, published 12 October 2017; manual updated 11 September 2026. Condition D1.2.
