I recently asked teachers on an in-service training course: how can a school improve its GCSE examination performance? Among the suggestions was a cynical proposal: select all the pupils you believe will achieve grades A and B and teach them in a class of 45. Put a teacher of average ability in charge.

Take all pupils likely to achieve grades E, F and G and place them, too, in a class of 45. Put a poor, but muscular teacher, with them.

Then identify all CD borderline students. Put them into a class of 15 with the best teacher. Provide them with first-rate resources and give them continuous encouragement to help them achieve a C.

A sad reductio ad absurdum of course, but as we know in reality many schools, like mine, are tempted not to look at overall examination performance but to skew their efforts towards pupils on the CD borderline.

It is not just that the CD borderline is crucial for improvement in the school’s attainment profile. GCSE grade Cs can open doors and more prestigious progression routes for pupils. There is also an argument that a focus on the CD group can raise standards overall.

In reality, many schools adopt a version of this strategy because the GCSE examination league table places a premium on grade C by reporting the percentage of pupils who achieve five or more grades C and above. It is the statistic most reported by the press and the one most parents use to assess a school.

Why was five or more grades ABC chosen for the league table? It would be thought by most schools to be a poor foundation for A-level. It is not a specific qualification to enter a GNVQ course, neither do universities specify it as a minimum qualification. It is completely arbitrary and I suspect it was chosen by a civil servant who, like me, took his O-levels in the mid-1960s and remembered that you then needed five to become an articled clerk to a solicitor, or to enter the Midland Bank.

A sporting analogy demonstrates how nonsensical the chosen statistic is. Imagine a cricket match in which only the scores of batsmen who make 50 or more count towards their side’s total. Scores below 50 don’t count at all whereas any score of over 50 still only counts as 50 whether it is a scrappy 51, or a faultless century. Not the best way to motivate a team and no way to compare two team performances.

The league table records three sets of GCSE percentages: those achieving five or more grades A* to C; five or more grades A* to G; and one or more grades A* to G. The first statistic is flawed, the latter two provide little differentiation when most schools achieve scores somewhere between 85 to 95 per cent.

So what could replace the current system? Why not use the average total GCSE points of all pupils? The calculation is simple. An A* grade scores 8, an A - 7, a B - 6 etc. A school’s points are aggregated and divided by all pupils on roll in January of the examination year. If this system were used, all pupils would be contributing to the school’s published results. The performance of a pupil achieving 10 grade As would be recognised, just as would that of a student achieving a couple of Fs.

How would it affect the league tables? My own authority, Hertfordshire, publishes school’s average GCSE total points scores. In 1996 there were 80 LEA and grant- maintained secondary schools in the county. They can be ranked under the current method from a school where 93 per cent achieved five or more ABCs to one where 16 per cent did. The average total GCSE score ran from 54. 4 points to 21.7 with a county average of 38.4.

What happens to the rank ordering within the Hertfordshire league table if the revised system is used? The change doesn’t affect the “top” and the “bottom” performing schools. Nine of the “top” performers retain their position in the top 10 whichever system is in operation. Eight of the lowest ranked schools are in the bottom 10 under both methods. However with schools achieving around and below the average our perception of some schools’ performance is changed, in some cases remarkably.

When average total points scores are used one school jumps up the county rankings by 22 places. There are a number of other schools whose place in the rankings rises or falls 10 or more places. In my own school, for example, on the five or more ABC criteria we score 36 per cent. Yet when our average total points score is compared with other schools we are only 0.6 of a point behind a school achieving 46 per cent.

The Government should drop from the league tables the percentage of pupils achieving five or more, and one, A* to G grades. Average total points scores should be introduced immediately and run in conjunction with the five or more A* to C grades figure which should be discontinued after a three-year period. A three-year rolling average of average total GCSE points scores should also be introduced to even out blips in performance.

Until we develop a system which really measures school performance by providing information about the progress made by pupils, let’s at least revise the current crude raw score system. We can, as an interim measure, provide better information for parents which recognises that the aim of a state education system is achievement for all children.

Martin Titchmarsh is headteacher of the Nobel School, Stevenage