Roche Limit

The exam regulator debacle last spring, when summer exams were summarily cancelled across England and Wales, was an avoidable mess.

From the beginning it was clear that was quite possible to run essential exams even with coronavirus in the ascendancy. A Levels could have been saved by scrapping all GCSE exams, except perhaps English and Maths. GCSES used to be a school leaver qualification, but now that every student has to attend some form of education or training until they are 18 their usefulness has declined until they are mostly just a distraction. Exam boards would have held two sittings for each A Level, making use of the alternate papers they hold onto in case the primary papers are compromised, which would have allow socially distanced exams to take place. Students in exam halls are already spread out for obvious reasons, so only a little extra space would have been needed to keep the whole process covid-safe. But then the government panicked, although it is hard to criticise them given the uniqueness of the situation.

The result of the rushed arrangement was that the national averages were very little different last year, while a very large number of students were awarded grades far from those that their teachers had expected. After a national outcry raw teacher predictions were used unmoderated, leaving final grades inconsistent from school to school.

Was it possible to make a better fist of awarding estimated grades? It was always going to be difficult to award grades that were trusted as much as normal without the evidence of actual exam results, but the government made a big mistake. The nominally independent Ofqual quango which regulates qualifications explained the risks of each option, recommended that exams were scrapped and replaced with a certification of performance, to avoid damaging comparisons with genuine A Levels. This would have allowed students to progress to university on the basis of offers, and it would have allowed for the necessary grade inflation as exam boards sought individual fairness.

But Ofqual came under pressure to issue actual A Level grades from the government which was also keen to avoid politically damaging grade inflation. This was a fine aspiration, but the complexities of the statistical model which was needed to assign grades placed that desire in opposition to the reliability of individual awards. Forcing the grades to fit the same distribution as previous years meant the model risked introducing inconsistencies at the individual level. The government ignored the warnings and went for issuing moderated simulated exam grades, with the instruction to focus on the average results at the expense of individual variations. Students missed their university offers by the thousand. Students who had never received anything other that top grades were awarded several grades lower, and others failed their courses, much to the surprise of teachers who had firmly predicted good passes. Small independent schools had class sizes too small to moderate, so their high predictions went through without reductions, a likelihood also predicted by Ofqual and ignored by a blinkered Education Secretary.

And the rest is history. The government of Scotland cracked and switched to the unreliable, unmoderated Centre Assessed Grades from the teachers, followed inevitably by England two weeks later, introducing an whole new set of irregularities, now brushed under the carpet. One can only hope that if exams are cancelled in 2021 there will be a more informed plan in place.