What should happen when publications disagree?

Eton’s tables: four ways two figures can differ

Eton publishes its A-level results twice, in a table signed by its Examinations Officer and in its annual report. The 2022 pair begin by agreeing, both printing 95.4% for A*–B, which we call “Consistent at the published precision”, a long way of saying they agree as far as either goes. For A*–A they part company, the table printing 83.6% and the report 83.5%, and neither says why. That earns “Sources differ”, the only label of the four that admits we don’t know, and we come back to it below.

In 2023 the trouble lay in our own labelling. The figure we once showed as Eton’s A-level A* share, 44.8%, counted Pre-U entries too, while Eton’s A-level table on its own gives 43.7%; under “Different groups counted” both stay, each labelled, and only the A-level one is compared. “Revised by the school” is the simplest of the four. In 2024 Eton re-signed its table after one grade was raised, so the A* share is 38.7% and the first table’s 38.6% is superseded, with nothing for anyone to adjudicate.

The 2025 pair were the oddest, and the most innocent. The annual report gives A*–A as 78.2%, the table’s 78.2% exactly. A*–B is where the decimal places come apart: the table prints it to two, which our record shows to one, as 95.1%, and the report simply stops after the first, where one would expect it to round. The count behind both is the same, so 2025 joins 2022’s A*–B under “Consistent at the published precision”.

The difference first, the story later

Which brings us back to 2022, where the table and the annual report print different A*–A percentages and neither says why. A re-mark, a rounding or a changed group would each make a tidy story, and the obvious versions of the last two have already failed: the table’s count gives its own figure whether rounded or cut, and adding the Pre-U entries gives a different figure again. Until there is evidence, none of them is the explanation. So that school, year and grade band stay out of the comparison they affect; both publications remain available, other bands can still be used where their evidence permits, and the record page sets out exactly what differs and what we checked.

Later, higher, repeated: none of them settles it

A later document may explain an earlier one, but its date alone can’t; nor should the larger figure win for being larger, a habit better left to league tables; and two publications that agree are less reassuring than they look if both copied the same original table. The questions that help are plainer. What did each document actually count, and does one revise the other or merely repeat it? The source link lets you follow that up, though it is no promise that every figure in the document has been checked.

Survival is not a grade

A long run of tables may mean only that a school’s documents survived, stayed accessible and found a researcher, and a short one nothing worse than gaps in our own collection, so the length of an archive, on its own, tells you little about the education on offer. Why a particular figure is missing matters as well. Sometimes we can’t open the old page; sometimes the page leaves the number out, or the publisher has withheld a small count. Each calls for a different explanation, and none of them means the result was zero.

A question the school can answer

Both publications are there to read, and another band may be clear of the difference. Anyone who asks the school does better to name the table, the year and the band, and to ask which entries were included or whether the figure was revised, since a vague question tends to get a general reply and a precise one gives somebody at the school something they can actually answer.

Where these figures come from

Each record gives the full definition, the documents it came from and what we checked.

1. Eton College, A level 2022: 55.4% of 758 entries graded A*

   The signed results table’s figure; it differs from the annual report.

   Publications differ on A*–A. Checked against the original: A*, A*–B and pass.

2. Eton College, A level 2022: 83.5% of entries graded A*–A

   The annual report’s figure. The documents do not explain why it differs.

   Publications differ on A*–A.

• What caused the example’s disagreement is still unknown.
• A historical dossier that claims a completed audit doesn’t vouch for every current field, or for every reading of one.