There appear to be three purposes for exams; they are:
- Indicating the standard each student has obtained in a particular field of study.
- Differentiating between 'bright' students and 'weak' students.
- Measuring a school's effectiveness.
If the sole purpose of the exam system is to indicate the standards each student has reached then the testing system should set out the standard that needs to be achieved for each grade and, if a student meets that standard, they should be given the grade
However, this leads to the problem that if a large number of students reaches the same standard, the second purpose of differentiating between students would not be met.
In order to meet the second purpose, it would be necessary to set limits for each grade so that, even if a student does make the standard for a particular grade they are only awarded it if there are grades of that type left.
My understanding is that the old O-Level was norm-referenced in that the proportion of each grade was fixed and awarded based on your rank position (relative achievement). The GCSE is criterion-referenced and grades are awarded upon the standard achieved (absolute achievement).
Norm-referencing is good for purpose two, in that is generates very clear differentiation between students but it does lead to unfairness when students reach the same standard but are awarded different grades. Criterion-referencing is good for purpose one but, once larger numbers of students start to reach the same grades, differentiation becomes harder.
When you throw into the mix the third purpose, which has the effect of requiring schools and teachers to improve constantly, trouble is bound to occur!
In order to resolve the tensions that exist I would propose the following to cover each purpose above:
- GCSE grades are given by reference to strict criteria and these are checked by Ofqual (or a similar body) comparing assessment methods and pass marks.
- Each student is also given a ranking for each subject based on national norms.
- School effectiveness is measured by looking at student progress (eg KS2 to KS4) rather than by absolute attainment (eg 5A*-C inc EN&MA).