Go forward and not Bac

It's how standards are set, not what kind of exam students take, that matters, says Dylan Wiliam.

Whether the standards of awards made at A level this year are comparable to those made last year should soon be clear. But, even if it turns out that they are, huge damage will have been done.

In the short term, individual students assured of an A grade by their teachers will wonder why they didn't get it and heads could roll at the examining groups and at the Qualifications and Curriculum Authority.

But the long-term effects of this summer's events will be far more serious. Originally, the purpose of A-level examinations was to assist in the selection of school students for university education, with the system largely run by the universities. However, as A-levels and GCSEs became ways of measuring the performance of schools in the 1980s and 1990s, the government assumed greater and greater control of the process of examining in the form of "guidance" and "core codes of practice".

This summer's furore, however, is about a quite separate issue. It is about the extent of political interference not just in the procedures of examining, but in the standards to be applied. It does not matter that in this case the interference (wherever it came from) made high A-level grades harder rather than easier to obtain. What matters is that the standards of A levels are seen to be something the government can vary according to its political priorities. It is highly unlikely that evidence of a direct command to the examining groups will be found. All that will be found will be strong instructions to make sure that there is no grade inflation this year.

The irony in all this is that it probably wouldn't have mattered very much if pass rates had gone up this year, as we would have expected them to, given the number of students who were put off pursuing subjects to A2 level by their performance in their AS examinations. This is because universities are relying ever less on the actual grades received by students.

Universities want to recruit the best students they can, but if they recruit too few or too many, they are penalised, often heavily. They make offers to applicants based on the grades predicted for them by their schools. If the students achieve the required grades, they are in, but if they fail, then the university has to make a decision whether to take the student or not.

Some admissions tutors, like airlines, "overbook" places, expecting some to fail, but most try to get it right first time. If a student - let's call him David - was predicted to get three Bs, and the university asks for this but David gets three Cs, what do you do? Take him or hope to get a better student through clearing?

The difficulty with clearing is that it is a very public process - and many universities do not like to advertise the fact that they are still looking for students to fill their quotas. Perhaps more important, it is also by no means clear that there are better students available through clearing. After all, while David got three Cs, his school thought he would get three Bs, so perhaps it's not too much of a risk to take him after allI And this is what has been happening in recent years, with the school's predicted grades being more important than the grades eventually awarded. We have, in effect, reached the "holy grail" of selection for higher education - the post-qualification admissions system - but with the teachers' predicted grades rather than the actual grades as the qualification.

Those who call for a system like the baccalaureat miss the point - exactly the same technical problems of standard-setting arise there. What we need are ways of combining teachers' assessments of their own students (because these assessments are much more reliable than examinations) with the results of externally set and marked exams (to standardise the judgements made by teachers). Our assessment system looks rather like trying to "inspect quality in" at the final stage of the education process rather than ensuring that appropriate standards permeate all aspects.

Dylan Wiliam is assistant principal, King's College London.