Before you compare the numbers

What is being counted?

One pupil usually takes several subjects. If a school reports that 60% of its A-level entries were graded A*, that is not 60% of its pupils. Every percentage is a share of some total; where the school doesn’t say what that total is, we say so.

How different qualifications compare

A school’s history may cross several exam systems. To follow it across those changes, we apply the same published grade mapping to every school. For example, Pre-U D1–⁠D3 contributes to A-level A*–⁠A or equivalent. The original grades and the school’s own figures sit alongside our calculation. The mappings and their basis.

An A-level-only chart has a narrower purpose: it shows A-level results. If the school taught Pre-U instead, that chart may have a gap even where the common comparison continues.

Comparable figures across exam systems

The common scale brings together A levels, Cambridge Pre-U and IB Higher Level. It also connects the older lettered GCSE grades with numbered grades. Every school uses the same mapping, and each subject entry has equal weight. We label the result as our calculation and show the original grades and the school’s own headline alongside it.

Sixth form

Cambridge aligns Pre-U D3 with A and P3 with E, and describes D2 as equivalent to A* for most university offers. Our intermediate groupings are at or below the corresponding UCAS tariff bands. These choices give us a consistent, conservative scale. The qualifications still differ in what and how pupils study. Cambridge’s grading and tariff guide.

For IB Higher Level we follow the combined table published by King’s College School, Wimbledon: 7 counts as A*, 6 as A, 5 as B, 4 as C and 3 as D. This is the school’s convention, adopted consistently by TSR; the UCAS tariff agrees at 7 and 6 but scores a 5 like a C. Grades 2 and 1 still count in the entry total, without being assigned an A-level grade. We exclude IB Standard Level, Pre-U short courses and Global Perspectives and Research, AS levels and the EPQ.

GCSE

High grades combines A*–⁠A with 9–⁠7; C/4 or better combines A*–⁠C with 9–⁠4. These bands use Ofqual’s comparison points: the bottom of 7 aligns with A, and the bottom of 4 with C. Top grades combines A* with 9–⁠8 as a practical approximation. Grade 9 is shown separately, using only entries graded 9–⁠1. This gives readers useful comparisons without assigning invented numbered grades to old results.

C/4 or better is close to 100% at these schools, so it appears on school pages but isn’t offered first in comparisons.

Weighting

When a year mixes exam systems, the comparable figure is the number of entries at or above the band, across every system, divided by all the entries. We never average two percentages. In a year with one system, the comparable figure is the figure published for that system, at the precision it was printed. For some years, an editorial review selects a published result for a particular qualification. The explanation names that publication and the qualifications it covers, and links to the original records.

When a school printed only percentages

We recover the entry counts the percentages allow. If rounding leaves a little doubt we show the range, for example 82.7–⁠82.8%, rather than a decimal we can’t vouch for, and a comparison says whether its answer holds at both ends of the range. If the doubt is more than a point, or an exam system has no entry count, we cannot calculate a combined figure across those systems. A percentage from a selected single-qualification publication can be shown directly at its printed precision, without inventing counts. If neither the records nor an explained editorial choice tells us which qualifications a result covers, we leave that year out of comparisons.

What we don’t do

We don’t estimate how many A*s would have been 9s, spread a school’s grades across a scale it didn’t use, or claim more precision than the school printed. Results from before A* existed (A level before 2010, GCSE before 1994) keep their own scales. 2020 and 2021 were teacher-assessed years: their figures are shown but left out of comparisons unless you ask for them.

The headline, the calculation and the original grades

Each year’s detail identifies TSR’s comparable figure and the school’s own headline, and lets you inspect the original grades used in the calculation. The details show the grade band and qualification scope of the selected publication, together with any calculation used to place it on the common scale.

In a few years a school’s own tables don’t cover every exam system it entered. Where the government’s tables for that year are complete, the year’s details also show their figure, labelled as the Department for Education’s count of pupils completing A levels in England. It is never used in charts or comparisons.

Offers, acceptances and destinations answer different questions

A table of university applications describes one admissions round. A school’s list of where leavers went describes one leaving year. The groups and dates can differ, so a count of offers can’t be set against a count of destinations.

University pages show the school publications that name each university. A mention on its own doesn’t tell you how many pupils went there, or how likely a future applicant is to get a place.

How the comparison is calculated

To calculate a yearly difference, we need two figures that count the same thing, on the same grading scale, for the same kind of group and the same year. If publications disagree, we leave the figure out until the evidence resolves the difference or an explained editorial choice identifies the publication to use. The rules for settling a difference explain the distinction. Charts and calculated rates use the figures as recorded and never replace the figures printed in the source.

The common scale connects grades across qualifications and years. It cannot turn one group or outcome into another: offers, for example, remain separate from places taken up. The labels make those differences visible.

Compare’s verdict checks every range of at least three comparable years measured the same way. Its additional checks show the full series, the latest three and five years, and each year left out in turn. Test a claim lets you require at least three or five comparable years, and checks every range that meets that requirement. Both show how far an answer depends on the years chosen. Neither is a test of statistical significance or a probability, and neither shows that teaching caused the results.

How we word verdicts

A verdict is our judgement on a comparison, so it is never stronger than the figures behind it. It rests on two things: the typical gap (the middle of the yearly gaps, in percentage points) and how many of the tested year ranges point the same way. A verdict always names the years it covers, and never calls one school better.

The point thresholds apply to percentages; for counts, such as offers, only the number of years and how consistently the ranges agree decide the verdict. An “every year” claim holds only if the school was higher in every year we can compare. Scale adopted 23 September 2026.

When figures are missing or disagree

The trail may begin in a results table, a school magazine, an annual report or a university’s own statistics. We keep the source and its definition with each figure. A document can cover many years; the check beside a figure says how far we have followed it.

Missing years

A gap invites a question. Have we yet to find the result? Is the archive unavailable? Or did the school publish something that answers a different question? The school profile explains what we know. An unsuccessful search does not tell us that a result was never published.

When publications disagree

First results, revised results and figures for different groups stay separate and named. Counts we made from a published list are kept apart from totals the publisher printed.

Since 23 September 2026, a difference between two publications is settled only when one of the rules below is met. Each settlement is dated and shown in the revision log of the figures it affects, and earlier versions stay visible. A newer page or a more precise number doesn’t settle a difference on its own.

Sometimes an editorial review chooses a particular published version for a comparison. We record the figure, the qualifications it covers, who approved the choice and when. That is our choice of source; it does not mean the publisher has revised its earlier figures.

What can change a figure or its label

A headline can be labelled as an A-level result while counting Pre-U or IB grades as well. A figure’s value or label changes only on positive evidence: the school’s own re-issued table, the publication’s own wording, or an exact reproduction from the school’s own tables. Not finding a source online is never evidence that a figure is wrong or mislabelled. Many figures come from archived prospectuses, magazines and documents that are no longer online, and they stay as recorded. A figure shown to count a different group keeps its place in the record with its own label.

Corrections and earlier versions

Automated checks flag possible errors; they never change a figure. A correction needs a source and an explanation, and the earlier value stays visible.

Check a figure · Corrections and differing figures

What these figures can’t tell you

Published results can help you question a claim and understand a school’s academic record. They can’t tell you whether a school is right for a particular child, or predict that child’s results or admission.

They don’t allow for prior attainment, how selective a school’s admission is, subject choices, the make-up of a year group or a school’s aims. So they can’t isolate the quality of teaching, or rank schools overall.