And you thought it was as simple as ABC

Exam boards have been criticised for being too lenient. Judith Judd observes how one team of GCSE examiners arrived at its results

In committee room 302, at the Southern Examining Group's headquarters, two examiners are using sample scripts to help to draw the line to determine which GCSE business studies candidates should receive grade C this year.

The grade boundary mark varies from year to year, according to how hard the paper was. Should it be 40 or 41 this time? One examiner claims that those with 40 should be given the benefit of the doubt. Another says no - grade C is one of the three top grades at GCSE and the sample scripts with 40 do not suggest they are worth C.

The chairman steps in to remind them that they have only looked at the first paper and that the picture will change as they consider the second paper and the coursework. A board official points out that, because of a rule change dictated by government advisers, it will be less easy this year to give candidates the benefit of the doubt. So the boundary between grades C and D for the first paper is fixed at 41, just one mark higher than last year.

Marking exams is a high-profile business. During the past few years, critics have suggested that it has become easier to get good grades and that exam boards are so keen to gain the reputation for being "easier" than their competitors that they have lowered standards. In reply, the boards point to their long and careful procedures, now standardised by a code of practice.

Throughout the country, hundreds of similar meetings are taking place at the five GCSE examining groups for each subject. It is a long and labyrinthine process which, for this year's Business Studies, begins at 10am and continues until after 5.30pm. Despite the good humoured atmosphere among the eight-strong group of examiners, no one is in danger of forgetting that the grades of 21,000 candidates depend on their decisions.

There are two main ways in which the grade boundaries are decided. The first is through the judgement of the examiners on a sample of this year's scripts. The second is by looking at statistics showing the proportion of candidates awarded each grade. Both are taken into account to determine the boundary. Both are necessary, say the boards, because it would not be right to adopt a purely statistical approach and assign a proportion of candidates to particular grades: candidates' attainment may vary from year to year.

However, changes in the quality of candidates from one year to the next are likely to be small. This means that big differences in the marks achieved on any paper from one year to the next are much more likely to occur because the paper was harder or the marking stricter. Statistics are used to make allowances for such changes.

The examiners, who are teachers released from their schools, have two sets of figures. First, the grade boundaries for last year's candidates and the proportion awarded each grade and, second, statistical predictions for this year's boundaries, which take into account possible changes in the types of candidate entering. GCSE grades range from A*(top) to G. In the course of the day, they have to decide on the mark required to obtain an A, a C and an F.

The chief examiner starts by saying that he doesn't feel the candidates did much better this year and that he's expecting the outcome of the meeting to be much the same as last year. He brings some questions to the attention of his colleagues. "In question 3A, we were trying to help the less able. It gave a few more easy marks so I've asked for slightly higher marks for the F/G boundary." He directs them to Part B as a question "that sorted the sheep from the goats". He suggests 40 as the boundary and as that coincides exactly with the statistical prediction the group settles down to study a selection of scripts, given marks ranging from 35 to 45 marks.

Silence falls as the committee reads the scripts and decides which are definitely a C and which are definitely not - the term they use is "in" or "out". When they are ready, the chairman asks everyone for their conclusion. The relevant marks are chalked up on the board and a tally kept of the number of ins, outs and question-marks for each mark. Some 40s for example, are in and some are out. An odd 39 may be in but most are out. A line is drawn above the mark at which most are out and another below which most are in, leaving a grey area. This year 38 is definitely out and 42 in. The three marks in between are halved and the result rounded up to the nearest whole number. It is 41, just one above both the statistical prediction and the chief examiner's recommendation.

The whole process is then repeated to decide the percentage required for an A and the same for an F. Then it's on to paper two, where the same decisions have to be taken again.

Here the Information Technology is acknowledged to have been difficult and the meeting is asked to take that into account. There are some exam- howler jokes. "Primary industries use first-hand material, secondary industries use second-hand material and tertiary industries a bit of both," is one. Once the provisional boundaries for both papers and the coursework are agreed, the results are placed alongside the statistics and the final grade boundaries are calculated.

Is it all a fix? The final results suggest not - there is no obvious pattern. Although a slightly higher percentage than last year will receive As and A*s, the proportion receiving Cs will be 2 per cent lower. So complex is the procedure that, if there is a conspiracy, it must have been designed by someone with the brain of an Einstein.