Only examinations truly pass the test: They cause nightmares and stress, but exams are the only way to measure ability to think, says Lincoln Allison

SCENE: a 'graduands' party, for people whose degree results had just been announced. Enter Sarah, a Nice Girl. 'Ah, Sarah' I say cautiously, 'Are you pleased?' (I now use the same approach in cases of pregnancy and excite an astonishing variety of responses). Yes, she says, but upset about Eddy. He was the most devoted student of the year, in the library 12 hours a day. And he's got a lower second, whereas complete idlers got upper seconds . . .

This is an oft-repeated conversation, and it still enrages me. Sarah's assumptions encapsulate some of my least favourite myths, which are as fallacious when applied to examinations as they are to life. What we, as teachers, are actually trying to do, in any educational system worth the name, is to get people to think. The capacity to think embraces the techniques of inference, conjecture, clear statement, argument and so on.

There is no particular way of learning these things, though slightly drunken arguments that last till 4am and long mind-clearing bicycle rides can play a part. What doesn't work very well is 'work': that is, sharply delineating what you have to do and simply plugging away at it. Eddy may well have sat long hours in the library. He may have regurgitated many essays. But who knows what actually went on between his ears until we sit him down in an examination and ask him to construct an argument that answers a complex question. 'Work' is a particularly damaging myth because it makes the fundamental error of confusing means and ends.

That is the philosophical case for having traditional examinations. There is, as well, a much more banal, but over-riding reason, which is that people will cheat any other system. If you allow people to do work in their own time, they will, in large numbers, get somebody else to do it.

There are many ways of doing this, including the help of kindly parents or girlfriends, paying a swot and using old essays (including those from essay banks). There are also new, higher technology, methods such as electronic-mailing essays between colleges and using a rescrambling programme to re-cycle old essays.

Surveys in the US in recent years have shown different proportions of students who will admit to using these techniques, but never less than a substantial minority. In this country, Professor Stephen Newstead of the University of Plymouth presented evidence to the British Psychological Society in March which suggested that 50 per cent of English students would admit to some kind of plagiarism, with a range of lower percentages admitting to more specific forms of cheating. Very few people had cheated in traditional examinations, and those almost entirely in maths and sciences.

Dishonesty is very corrupting: it undermines the incentive to work honestly and devalues any qualifications awarded. I think most good British undergraduates despise cheating and would rather do their own work. But there can be cultural problems with some overseas students: a particular student of mine from a Chinese cultural background simply didn't understand what we wanted in the way of originality. He had handed in a good essay and thought it odd that we should object that it wasn't his good essay.

A second dimension of cheating involves connivance, where the teacher (and we could be talking about anything from national curriculum assessment to a doctoral supervisor) has a vested interest in good marks for his pupil. A vested interest could consist of personal popularity, the rate of applications to the institution, personal or ideological bias or the return of sexual favours.

A properly constituted examination cuts right across all this. It must consist of a high security system in which candidates attempt timed, unseen papers that are set and anonymously marked by people other than the candidate's teacher. An alternative, used in most British universities, is the 'external examiner' from another institution who oversees the marking procedures so that, for example, the candidate who produces a perfectly reasonable interpretation of a question, which is not what the teacher had in mind, can be properly rewarded.

It is often argued against examinations that they are stressful and therefore performances don't represent candidates' true abilities. Seen from another aspect, though, stress is one of their great virtues. Exams are like life: the ability to think quickly and under pressure is a great deal more valuable than the obverse. In my experience, if candidates actually have something to say, they can say it effectively in 15 of the 45 minutes available, and with a splitting headache.

Most careers are going to require people to put arguments, process information and express ideas with limited time and opportunity, under pressure and in a different form from the one they would prefer. So it is possible to write a much more convincing reference for a serious job about a good examinee than about a bad one.

There are powerful and growing arguments for traditional exams, but there are also growing pressures against them. If you suddenly decide to educate 30 per cent of the relevant age group up to degree level, as we have in this country, and also rapidly increase the number of students for whom English is not a first language, you are bound to find that many people cannot meet traditional standards, but can be given a degree if it is based on course and project work.

This process has a long and bitterly debated history in the US, which is likely to be repeated here as many of the newer degree- giving institutions resort to the argument that their methods and standards are not inferior, but merely alternative. Employers will make their own judgements.

In years of taking examinations, I had one truly dramatic moment. It came in finals philosophical logic in Oxford. Three quarters of an hour after the exam had started, I could still only see three questions, as opposed to the required four, that I could answer.

There was a question, however, on the logic of time, a subject on which I had never read a book. But I had had an interesting pub conversation on the subject. 'Can time stand still?' was the question. 'No; or, at least, not for very long,' I replied. They gave me a very good mark. Which goes to show that truth is generally more amusing than untruth, brevity is better than length. It doesn't matter where you get your ideas from and sometimes you have to take risks. In exams as in life.

The author is reader in politics and international studies at the University of Warwick. He is writing in a personal capacity.