The starting position for the UK government when it comes to the pandemic and education has been that we should pretend, as far as possible, that nothing has changed. Hence academics continue to teach in person, even as the coronavirus races through undergraduate populations. And the madness, this summer, of trying to pretend we had held school exams – even issuing grades for them.
This week, in that vein, the government announced that A-levels and GCSE exams would go ahead in England in 2021. This is an error.
A high-stakes exam series is a dangerous thing to run during a lethal pandemic. We are talking about 16- to 18-year-olds: people who may be better placed to withstand the disease, but are eminently capable of spreading it. Besides, we want pupils and teachers to be able to self-isolate without penalty. The educational benefits would need to be strong to justify the risk from exams. I am not clear they are.
Start at the top: an exam measures an entrant’s ability to cram a certain amount of information and hit a mark scheme. The process is intended to measure a candidate’s recall of a subject on a given day.
But people outside schools use exam grades for something else. We outsiders use them to identify people who will do well in a given subject in future years. To use the jargon, we use grades for their “predictive validity”. We rely on GCSEs and A-levels to find students who will be “good at maths” or “good at history” at higher levels.
You can see this in university behaviour. Generally, the universities have fiercely resisted attempts to get involved in determining the content of these exams. But Cambridge did attack ministers about the abolition of the old AS-level because it had good predictive validity. And when exams are changed, grades are changed or new qualifications introduced, they are checked over to see how well they predict future success.
There are always issues with using exams in this way: poorer pupils need to have a greater aptitude for the same subject to get the same grades as richer ones. And while there are (imperfect) processes in place that attempt to account for this, the noise in the results this year may drown out any signal.
After all, since March, when the schools shut, there has been just over one month or so of normal-ish tuition. What pupils have received has been wildly variable.
And the crisis is not over. Schools in areas with higher prevalence will be more seriously disrupted by the virus. A child whose family or class self-isolate will, similarly, lose ground.
While we can run exams this summer that will accurately measure what each entrant knows on exam day about a domain of knowledge, we cannot run exams that will have robust validity. The top students may just have had a lucky pandemic.
Still, in a letter to the Telegraph, one group of educationalists has expressed the view that exams must go ahead. Such is their iron certainty, they also specify how many As and A*s should be awarded (there is no virus deadly enough to make a certain type of British teacher stop thinking grade inflation is the real peril).
They do call for moderation in favour of children who have suffered particularly from the pandemic. But they would repeat the error of this past summer: the information required to do this does not exist – can we identify the children sharing a laptop with siblings?
This entire scheme will not work. Whatever the grade levels we aim for, we are likely to end up with a barrel-load more grade inflation: issuing a child too low a grade is a non-trivial error, so we will probably respond by boosting grades. It is entirely possible we will end up in the same type of crisis next year as we did this year.
So how about something like this: prepare to abandon mandatory exams for the departing cohorts this year. If they must have grades, get their schools to issue them using different grades – perhaps on a pass/merit/distinction basis – so they cannot be compared with other years. Change the qualification name, perhaps.
This would help in several ways.
First, rather than trying to figure out how to robustly say who would deserve an A*, we can choose new awarding criteria that can be assessed by teachers without exams.
Second, we can be generous without worrying about grade inflation. You can’t get any inflation in a time series with one data point.
And, finally – most importantly – this process would be honest. It does not wish away our problems. If you cannot issue a grade with the same properties as in previous years, do not pretend to. We want outsiders to treat these A-levels differently, and changing the grading forces them to. For the qualification-holder, it is much easier to explain that “my A-level looks weird because it’s a pandemic A-level – you can google it,” than it is to explain an unfair low grade.
There are lots of downsides to this plan: discipline may suffer. But schools will still have control of issuing certificates. And exams may be a half-dead rubber anyway. Universities and colleges would be daft to make offers conditional on mechanically getting certain grades when no-one knows how grades will be distributed.
Without data from the A-level and GCSE, some students will start on inappropriate courses at universities and colleges. Some students will have insufficient subject knowledge to progress. We will need a bit of care about that. But all institutions will need to admit people without the normal levels of preparation. That is why two experienced vice-chancellors have already proposed abandoning the A-level for this year – and using the extra time in the school year to help students catch up.
Robust public exams are a good thing. The GCSE, in particular, is a qualification worth preserving. But that is precisely why I would pause for this year. Nothing will do more damage to the qualification system than a second year of exam disasters.