Year on year we hear the arguments as GCSE and A-level results
come out. Standards have gone up, standards have gone down, too many people are
getting top grades. At exam boards within Higher Education we worry if results in
one module are out of kilter with the rest.
At the institution that I’ve just left we even used a tool that showed
this. We worry too, if the standard deviation changes over the years or is
vastly different from the average standard deviation for a cohort.
Definitions
A driving test is generally criterion-referenced. Certainly
at least the theory part is. You have to have a very high score to pass. Even
in the practical test, there is a recognised minimum performance required. After
all, one is about to let you out with a powerful killing machine. The good
driving instructor will work out what you need to know and which skills you
need to acquire and how to teach you all of that. If that instructor improves her
own teaching skills then she should have lots of students who pass.
Yet there is still often a feeling that the examiners want
to pass a certain number a day and any more or fewer than the average would be
queried.
The old O-level and A-level were norm-related. This means
that a certain percentage of students were going to get each grade. This could have
the distinct disadvantage that if you were in a “good” cohort, then the bar was
raised.
Why we shouldn’t worry about GCSEs
When they were first introduced in the 1986 the big news was
that they were supposed to be criterion-referenced. We were supposed to work
out what needed to be learnt – and this was not just knowledge but skills as
well- and then work out how best to teach it. There was a possibility, then, of
standards rising year on year. Yet when this does happen, it is regarded with suspicion.
My two concrete examples
For much of my time at the University of Salford I taught two
specialist courses: Introduction to
Children’s Literature, and Writing
Novels for Young People. My research has constantly fed into those modules
so that at the end of nine years I was teaching a much more complex and
demanding courses. However, my teaching skills also improved (about time too,
after 42 years!) so despite the courses in effect getting harder results were
also getting slightly better. Certainly, the collective knowledge and skill
competency were increasing.
Perhaps we need both?
Criterion-referencing helps us to really define what is needed.
Norm-referencing helps us to control and investigate. However, if we get a low standard
deviation, or all students obtaining 60+, we shouldn’t just assume that the
course was too easy or assignments too leniently marked. We should scrutinize
the course and the outcomes carefully.