A single exam result can change a pupil’s life – but grading in England is a lottery

GCSE, AS and A-level grades are important. Students’ destinies depend on them. A B-grade rather than an A-grade for A-level biology might lose Sam a university place to study medicine; a 3 rather than a 4 for GCSE English language requires Alex to re-sit, and severely limits future opportunities.

We all, rightly, expect grades to be reliable, with those few errors that do happen rapidly corrected by an appeal.

Those may be the expectations.

The truth, however, is very different.

Grades are reliable, at best, to one grade either way – and many wrong grades cannot be corrected by an appeal.

If you’re surprised, let me explain...

That grades “are reliable to one grade either way” has an impeccable source: they are words given in evidence to the education select committee on 2 September 2020 by Ofqual’s then chief regulator Glenys Stacey – the senior executive of the organisation that oversees and regulates school exams in England.

Those words are significant for they imply that a certificate showing, for example, “GCSE English language, grade 3” really means “Based on the answers to this exam, the grade the candidate truly merits might be grade 3. Or perhaps grade 2. Maybe grade 4. No one knows which.”

Any decision based on any grade as shown on any certificate is therefore unsafe. Like the decision to deny Sam a place to study medicine. Like the decision to force Alex to re-sit GCSE English.

You might be thinking: “That must all be because of bad marking.”

It isn’t. It’s a consequence of the possibility that different, equally qualified examiners might give (slightly) different marks to the same answer, as can happen for exams inviting candidates to express their thoughts in their own words. One examiner might mark a particular answer 5/10, and another 6/10 – not because either has made a mistake, but rather because of a legitimate difference in academic opinion. When the individual marks are aggregated, the totals might be, say, 44/100 or 46/100.

If grade 4 is defined as “all marks from 43 to 47 inclusive”, the student is awarded grade 4 regardless. But if the grade 3/4 boundary is 45/46, a mark of 44 corresponds to grade 3, and 46, grade 4. The grade on the pupil’s certificate is the result of the lottery of who did the marking.

If this were rare, it would not be a problem. To find out how rare it is, around 12 years ago Ofqual carried out a research programme in which large numbers of exam scripts, in 14 subjects, were in essence marked twice – once by an assistant examiner (as in “normal” marking), and again by a subject senior examiner, whose mark, and hence grade, is designated by Ofqual as “definitive”, the ultimate authority of right.

Each script therefore had two marks, and two grades, one being the grade shown on the certificate, and the other being the definitive/right grade. The results were published in 2018: about 98% of A-level grades as awarded were found to be the definitive grade, or the next higher or lower; likewise 96% of GCSE grades.

Hence the chief regulator’s (nearly true) admission that grades “are reliable to one grade either way”.

The definitive grade could, of course, be identified in a re-mark by a subject senior examiner. Prior to the summer 2016 exams, this is what would have happened if a grade was appealed, or to use Ofqual’s preferred term, “challenged”. But in 2016 Ofqual changed the rules, allowing a re-mark only if a “review of marking” – the process invoked by a challenge – discovers what Ofqual calls a “marking error”, a failure of the original mark to comply with the pre-determined mark scheme.

That sounds very reasonable. But there’s a catch.

In my earlier example, I described the possibility that a script might be marked, entirely correctly, but given either 44 marks, grade 3, or 46 marks, grade 4, depending on who did the marking. Suppose that the script is actually marked 44, grade 3, but would have been awarded 46 by a senior examiner, corresponding to the definitive grade 4.

If the candidate challenges the grade 3, the “review of marking” would discover no marking errors, for the mark of 44 is totally legitimate. The originally awarded non-definitive/wrong grade 3 is therefore confirmed. I find this to be most unfair.

Ofqual clearly disagrees, for in a story posted on its website on 26 May 2016, entitled “Fairness at the heart of proposed changes to marking reviews and appeals system”, it claims that “It is not fair to allow some students to have a second bite of the cherry by giving them a higher mark on review, when the first mark was perfectly appropriate.”

Ofqual’s justification assumes that all legitimate marks are identical in every respect. But they are not. Some correspond to the definitive grade. But some don’t. That difference is critical.

So under the rules that have been in place since 2016, even if the original “bite of the cherry” were poisoned by corresponding to a non-definitive/wrong grade, Ofqual now denies access to the antidote, a thorough re-mark. Wrong grades resulting from differences in academic opinion therefore last for ever, as does the associated damage.

Ofqual should, of course, deliver reliable grades. In the future, once AI marking is trusted, only one “examiner” could mark all scripts in each subject, consistently applying the same standards throughout. This eliminates examiner variability, so solving the problem. That’s a long way away.

But Ofqual could take two, highly beneficial, actions right now.

The first is to ensure that every certificate shows, in bold letters: OFQUAL WARNING: THE GRADES ON THIS CERTIFICATE ARE RELIABLE, AT BEST, ONLY TO ONE GRADE EITHER WAY. For that’s the truth.

I appreciate that this does not solve the fundamental unreliability problem. But it does alert everyone that it is unwise to use the single grade, as printed, as the basis for any decision.

The second is to scrap the deeply unfair “two bites of the cherry” policy by changing the rules for appeals, re-instating a re-mark by a senior examiner on request, at a reduced cost so that there is no barrier.

Dennis Sherwood is the author of Missing the Mark: Why So Many School Exam Grades Are Wrong, and How to Get Results We Can Trust

Photograph by Getty Images