I have collected my initial thoughts after reading the Ofqual consultation, released on the 15th January 2021, over GCSE and A-Level replacements for this year. Alongside many others, I submitted proposals for 2020 which I felt would have avoided some of the worst outcomes we saw in Summer last year. My hope is that, this year, some of the suggestions will be given greater weight.
The basic principle underlying the Ofqual consultation is that teachers will be asked to grade students, that they can use a range of different evidence sources to do so, and that exam boards will be asked to produce mini tests / exams as one such source of evidence. This is not unlike the approach used in Key Stage 1 assessments (“SATs”) in primary schools in recent years. The actual process to be used to come up with a summary grade based on various sources of information is not being consulted over now, and it appears this will come from exam boards in guidance issued to teachers at some undetermined time in the future. This is a significant concern, as the devil really will be in the detail.
Overall, I am concerned that the proposed process is complex and ill-defined. There is scope to produce considerable workload for the education sector while still delivering a lack of comparability between centres / schools. I outline my concerns in more detail below.
Exam Board Papers – What are They For?
Ofqual is proposing that exam boards provide teachers papers (‘mini exams’) to “support consistency within and between schools and colleges” and that they “could also help with appeals”. However, it is very unclear how these papers will achieve these objectives. Papers might be sat at school or at home (p.18), they might be under supervision and they might not. Teachers might be asked to ‘remotely supervise’ these tests (p.18). These choices could vary on a per-pupil basis. The taking of tests may even be optional, and certainly teachers “should have some choice” over which questions are answered by their students. Grades will not be determined by these papers, so at best they will form one piece of evidence. If consistency is challenged, will the grades on these papers (when combined in some, as yet undetermined way) overrule other sources of information? This could be a cause of some confusion and needs significant clarity. The scope for lack of comparability of results between centres is significant when placing undue weight on these papers, and I am left wondering whether the additional workload for teachers and exam boards required to implement this proposal is really worth it.
If tests are to be taken (there are good reasons to suggest that they may be bad idea in their currently-envisaged form – see below), then I agree with Ofqual that – in principle – the ideal place to take them is in school (p.18). However, it is absolutely essential that school leaders do not end up feeling pressured to open to all students in an unsafe environment, due to the need for these tests. This is a basic principle, and I would resist any move to place further pressure on school leaders to fully open their schools until it is safe to do so.
Quality Assurance by Exam Boards
The main mechanism being proposed to ensure comparability and fairness between two centres / schools is random sampling (p.20-21). The exam board will sample the evidence base of a particular school for a particular subject, and query this with the school if they feel there is inadequate evidence to support the grades (it is not clear in the consultation whether the sampling will be of all pupils or individual pupils at that centre). This is a reasonable methodology for that particular subject / centre / student, but there is a major piece of information missing to enable judgement of whether this is sufficient for quality assurance of the system as a whole: what proportion of student grades will be sampled in this way? My concern is that the resources available to exam boards will be too small for this to be a large enough sample and that therefore the vast majority of grades awarded will be effectively unmoderated. This approach appears to be motivated by avoiding the bungled attempt at algorithmic moderation proposed in 2020, but without adequate resourcing, comparability between centres is not guaranteed to be better than it was under the abandoned 2020 scheme, and may even be worse.
Moreover, the bar for changing school grades appears to be set very high: “where robust investigation indicates that guidance has not been followed, or malpractice is found” (p.21), so I suspect we are heading towards a system of largely unmoderated centre-assessed grades. In 2020, centres were not aware at the point of returning CAGs that these would end up being given in unmoderated form, and therefore many centres appear to have been cautious when awarding high grades. Will this still be the case in 2021?
It is acknowledged throughout the consultation that centres / schools will have been unable to cover the entire curriculum in many cases. There appear to be two distinct issues to be dealt with here:
A. How to assess a subject with incomplete coverage
There are many ways this could be done. For the sake of argument, consider this question in the simplest setting of an exam. Here, the most direct approach would be simply to assess the entire curriculum, acknowledging that many more students would be unable to answer all questions this year, but re-adjusting grade boundaries to compensate. This may not be the best approach for student wellbeing, however, and in any case the proposal to use non-controlled assessment methods opens up much more flexibility. My concern is that flexibility almost always comes at the cost of comparability.
Ofqual are proposing that teachers have the ability to differentially weight different forms of assessment (e.g. practicals in the sciences). Is is unclear in the consultation whether this is on a per-student or on a per-centre basis – either brings challenges to fairness and transparency, and this point needs to be clarified quite urgently. They are also effectively proposing that teachers can give zero weight to some elements of the curriculum by choosing not to set / use assessments based on these elements. It is as yet undecided whether past work and tests can be used, or whether only work from now on – once students are aware it can be used for these purposes. It is opaque in the consultation how they are proposing to combine these various partial assessments. One approach I would not like to see is a weighted average of the various pieces of evidence available. A more robust approach, and one which may overcome some objections to using prior work, may be to allow teachers to select a number of the highest-graded pieces of work produced to date – a ‘curated portfolio’ approach. This may mitigate against both incomplete curriculum coverage and different student attitudes to summatively-assessed work versus standard class / homework.
B. How to ensure fairness
The consultation acknowledges that students in different parts of the country may have covered different amounts of the curriculum, due to local COVID restrictions. There is an unavoidable tension, therefore, between ‘assessment as a measure of what you can do’ and ‘assessment as a measure of what you can do, under the circumstances you were in’. This tension will not go away, and the Government needs to pick an option as a political decision. Some forms of assessment may mitigate this problem, to a degree, such as the ‘curated portfolio’ proposal made above, but none will solve it.
It is proposed that students are able to appeal to the exam board only ‘on the grounds that the school or college had not acted in line with the exam board’s procedural requirements’ (p.23). I am rather unclear how students are supposed to obtain information over the procedures followed at the school / college, so this sets a very high bar for appeals to the board. Meanwhile, the procedure for appeal to the school (p.23) appears to have a very low bar, and thus could potentially involve a significant extra workload for school staff. There is some suggestion that schools could be allowed to engage staff from other schools to handle marking appeals. If adequately financially resourced, Ofqual may wish to make this mandatory, to avoid conflicts of interest.
It is unclear in the consultation whether students will be able to appeal on the basis of an unfair weighting being applied to different elements of the curriculum (p.14). This could add an additional layer of complexity.
Grade Boundary Cliff-Edges
Grade boundaries have always been problematic. Can we really say that a student one mark either side of a Grade A boundary is that different in attainment? Last year, a bungled attempt was made to address this concern by requiring submission of student rankings within grade boundaries. Centre-Assessed Grades (CAGs) last year were optimistic, but this should come as no surprise – given a candidate I believe has a 50/50 chance of either getting an A or a B, why on earth would I choose a B? This issue will persist under the proposals for 2021, and I believe may be amplified by the knowledge that an algorithmic standardisation process will not be used. I suspect we may see even more complaints about ‘grade inflation’ in 2021, with significant knock-on effects for university admissions and funding. The root cause of this problem appears to be the aim to maintain the illusion of comparability between years for GCSE and A-Level results.
There are very significant workload implications for teachers, for school leaders, and for exam boards in these proposals – far more so than in 2020 arrangements. This workload has explicitly not yet been quantified in the consultation. I believe it needs to be quantified and funded: centres should receive additional funding to support this work, and teachers need to be guaranteed additional non-contact time to undertake the considerable additional work being requested of them.
Private candidates, such as home-educated students, got a very poor deal last year. This must not be repeated, especially since many of the students who would be taking GCSEs and A-Levels this year are exactly the same home-educated students who decided to postpone for one year as a result of the changes last year. I am concerned to ensure comparability of outcomes between private candidates and centre-based candidates, and I am worried that two of the four proposed mechanisms for private candidates essentially propose a completely different form of qualification for these candidates.
Are 2021 (and 2020) qualifications actually GCSEs and A-Levels?
By labelling the qualifications of 2021 as GCSEs / A-Levels, rather than giving them a different title, there is an implicit statement of comparability between grades awarded in 2021 and those in previous years, which is rather questionable. Others made the point that in 2020 it may have been better to label these qualifications differently – the same argument applies in 2021. Even Ofqual implicitly make this point (p.27) when presenting the argument against overseas candidates taking exams as normal “might give rise to comments that there were 2 types of grades awarded”. The reality is that there will at least three types of grades awarded in recent years, pre-2020, 2020, and 2021. Is it time to face up to this and avoid the pretence of comparability between these different systems?
Ofqual seem to believe that if exam boards publish the papers / tests / mini-exams ‘shortly before’ they are taken then that will avoid leaking information but won’t put some students at a disadvantage because ‘students would not know which one(s) they would be required to complete’. I can envisage a situation where some students try to prepare for all published papers the moment they are released online, potentially a much greater number of papers than they will be required to sit, leading to considerable stress and anxiety, with potential equalities implications.
From the consultation, is not clear how exam board sampling will work, but there is the opportunity to bias the sampling process to help detect and correct for unconscious bias, if equalities information is available to exam boards. This could be considered.
On p.29, Ofqual state that ‘The usual assurances of comparability between years, between individual students, between schools and colleges and between exam boards will not be possible.’ This is not inspiring of confidence, but is honest. The question is how we can mitigate these impacts as far as possible. I hope Ofqual will listen carefully to the suggestions for 2021, and publish the approach taken in plenty of time. Releasing the algorithm used in 2020 on the day of A-level result release was unacceptable, and I hope Ofqual have learnt from this experience.