Primary school test scores explained: raw, scaled and percentiles

The number of marks awarded before any conversion.

How many marks the pupil gained on that particular paper and which questions were answered successfully.

It does not, by itself, account for test difficulty, age, other pupils’ performance or an expected-standard threshold.

Raw marks expressed as a percentage of the available marks.

The proportion of marks gained on that assessment.

It is not a percentile and does not make papers with different content, totals or difficulty directly comparable.

A raw score converted onto a defined reporting scale for that assessment.

Depending on the test, it may show performance against a threshold or provide a stable reporting scale across test versions.

The number 100 does not universally mean “average” or “expected standard”; the named assessment controls the meaning.

A transformed score interpreted against a defined standardisation or norm sample.

How performance compares with the relevant reference group.

It does not automatically show whether a particular curriculum standard was met. Many tests use a mean of 100, but that convention is not universal.

A standardised result compared with pupils of the same or a very similar age in the test’s norm data.

Whether performance was above, near or below that of the age-based comparison group used by the test.

It is not a universal measure of ability, and unrelated tests are not interchangeable simply because both use age standardisation.

The percentage of the defined comparison group that scored at or below the pupil’s score.

Relative position in the named norm group. GL Assessment describes the 70th percentile as having “performed as well as, or better than, 70 per cent” of that group.

It is not 70% correct and does not mean that 70% of the curriculum is known. Percentiles are not equal-interval scores.

A range around an observed score that reflects measurement uncertainty.

That the pupil’s underlying performance is better represented by a plausible range than by one perfectly exact point.

It is not a prediction of all future results or a permanent measure of the child’s ability.

One of nine broad norm-referenced bands, numbered 1 to 9.

A broad indication of relative performance without encouraging over-reading of tiny score differences.

It does not retain the same precision or interpretation as the underlying standard score.

An approximate age at which the observed raw performance was typical in the test’s norm sample.

A broad description of performance on that particular test.

It does not mean the child thinks, learns or should be taught exactly like a child of that age, and it is less precise than a standard score or percentile.

Statutory end-of-KS2 national curriculum tests in reading, mathematics, and grammar, punctuation and spelling; writing and science are teacher-assessed. KS1 tests are optional.

Raw and scaled KS2 scores, whether the expected standard was met, and teacher-assessment outcomes. Optional KS1 materials use a separate 2026 scale.

The 80–120 KS2 scale and annual conversion tables are England-specific and administration-specific.

Maintained schools administer adaptive online Reading and Numeracy Personalised Assessments to learners in Years 2–9 at least once during the school year.

Skills feedback, progress information and age-standardised information, interpreted in a formative learning context.

Do not apply England’s 80–120 scale or expected-standard threshold to Welsh personalised-assessment reports.

Adaptive National Standardised Assessments are used in P1, P4, P7 and S3, normally once in each school year.

Subject- and stage-specific capacity-band information used as one source of evidence alongside teacher professional judgement.

Do not call these assessments “Scottish SATs” or translate their bands onto England’s scaled-score system.

Assessment arrangements are in transition while a new Year 1–10 statutory assessment pathway is introduced.

Current interim reporting during the transition; the April 2026 policy framework plans scaled scores to replace Levels of Progression.

No settled universal scaled-score range, threshold or conversion method should be stated until current detailed guidance defines it.