I’ve been semi-following the brouhaha over the replacement of GCSEs with the Ebacc…
…which, by the way, sounds to me more like a hospital superbug than a qualification…
…and I’ve been frustrated that something very simple, very basic and very fundamental to the whole issue of supposed ‘grade inflation’ has not been mentioned. And that very simple, very basic and very fundamental thing is, namely, the whole question of why ‘grade inflation’ has emerged as a supposed problem over the last ~30 years.
In essence, there are two ways you can award grades, one of which relies on a relative grade scale and the other of which relies on an absolute grade scale.
With a relative grade scale, the grade an individual candidate receives depends on how well they do relative to the rest of the candidates taking the qualification at the same time. So you decide in advance that the top, say, 10% of candidates will get the top grade, and then you stick to that come rain or shine. It doesn’t matter how well or how badly those top 10% of candidates perform, it doesn’t matter how much or how little they actually know about the subject in hand – the top 10% of candidates will get the top grade regardless.
With an absolute grade scale, the grade an individual candidate receives depends on how well they do in absolute terms, with a particular grade threshold being tied to a particular point on the marking scale. So you decide in advance that any candidate getting above, say, 85% of the available marks will get the top grade, and then you stick to that come rain or shine. If the number of candidates getting 85% or more increases, the number of candidates getting the top grade increases; if the number of candidates getting 85% or more decreases, the number of candidates getting the top grade decreases.
Now, the old O Levels used a relative grade scale, but GCSEs, when they were introduced, used an absolute grade scale. So that’s why ‘grade inflation’ emerged as a new phenomenon at the same time GCSEs were introduced – because it was literally impossible for more candidates to get a top grade under the old system, even if standards were increasing year-on-year.
Straightforward comparisons between pass rates under O Level and GCSE tell us nothing about the relative difficulty of those two qualifications, because passes were awarded on a different basis. The resolutely stable pass rates for O Level were not necessarily proof of rigorous academic standards, just as the increasing pass rates for GCSE were not necessarily proof of wanton manipulation of the results. They are simply a demonstration of the difference between relative and absolute grade scales. Crucially, they tell us nothing about why grades have increased since this has been a possibility – whether because this has revealed a long-standing trend in educational attainment that the relative grading scale had been obscuring, or because the introduction of an absolute scale has made it possible to manipulate the figures.*
But this wider debate is, in a sense, irrelevant. If Michael Gove’s concern is to get rid of ‘grade inflation’ then we don’t need to go to all the expense and upheaval of introducing a new exam to achieve that. All we need to do is replace the absolute grading scale with a relative one, and the problem will be solved. We will revert overnight to a situation where a pre-determined proportion of candidates get the top grade, and that proportion will be strictly the same in every year. Whether actual standards of knowledge increase or decrease, there will be no more grade inflation (or deflation).
This would undoubtedly solve some problems. For employers or educational institutions trying to differentiate candidates within a single cohort it will obviously make life easier – but only at the expense of making it harder to differentiate between candidates from different cohorts. One of the reasons the relative grading scale was abandoned in favour of an absolute one was that it was felt to be unfair that the same percentage score could lead to different grades, depending on the year in which the qualification was taken.
In the long run there is not much benefit in a system of qualifications that gives two candidates with the same level of academic ability different grades. An exam system can only really be reliable if it’s consistent. And it can only really be useful if the grade a person gets tells you about their ability in the subject, not the relationship between their ability and that of some other people who happened to take the exam at the same time.
* – Despite the number of people who will tell you that it’s obvious exam standards have fallen – and the vociferousness with which they’ll tell you it – this is not a simple question to answer. It’s not enough to demonstrate that some of the questions on early O Level papers were very hard – as they undoubtedly were; some covered material studied at degree level – we also need to know how many candidates got the answers right. If, thanks to the relative grading scale, it was possible for candidates to get an A without answering the very hard questions correctly then the presence of the very hard questions on the paper wouldn’t mean getting an A used to be harder, or that modern day standards have slipped. So far as I’m aware – please correct me in the comments if I’m wrong – no-one has published any research looking at actual attainment in this way.
While it’s not much of a substitute for actual data, it’s perhaps worth noting as an aside that the story told by GCSE results of consistently improving standards is backed up by international measurements of the UK’s educational achievement. While these do show that the UK’s position relative to other countries is slipping (something that is gleefully seized on by commentators looking for evidence to back up a narrative of declining educational standards), they nonetheless confirm that average educational standards in the UK are improving, albeit less rapidly than in other countries. There does seem to be some evidence to suggest that Britons are getting more knowledgeable over time, even if that contradicts widespread presumptions to the contrary.