Any index of school improvement must focus on the achievement of all pupils or it will distort what schools do, Dr Gerry Gough of Keele University told educationists in London last week.
Dr Gough, who has been working with Professor Michael Barber on the development of a "school improvement index" that can be set alongside raw exam results, said an index that included only GCSE grades A to C would mean that one group of pupils - those likely to get A to C grades - would get more attention while others got less.
He was speaking at a seminar on improving London schools organised by the Association of London Authorities. It was held the day after the Government launched a consultation document asking whether the 1995 league tables should include a school improvement index.
The Keele index would cover all pupils eligible to sit exams rather than just those entered and would allocate points for each grade achieved, from seven points for an A to one point for a G. Taking the base year as 100, the index would then show how results had improved over time. Fluctuations would be ironed out by giving each school's final value for a year as an average of three years.
Dr Gough said the index would be particularly useful for showing an individual school how it could improve its performance, allowing it to compare achievement in different subjects and to track, for instance, whether its teaching policies were equally effective for boys and girls.
He conceded that the index could not take account of changes in intake, although he pointed out that intake was usually "pretty constant". To allow for such changes would require a "value-added" approach that tracked individual pupils through the system - but he was not sure there would ever be a satisfactory method of doing that.
Trying to make improvement happen was even harder than trying to measure it, Dr Hilary Nicolle, director of education of the London borough of Islington, told the seminar.
Since 1993, Islington had tackled poor achievement in three ways: making a push to reduce truancy; introducing a "sticking plaster" project, under which schools could bid for up to Pounds 5,000 per school to improve the achievement of individual children in Year 11 by such means as holiday revision courses; and developing a long-term project to raise pupils' achievement in partnership with five out of the borough's nine secondary schools.
Schools taking part had to agree that the head should lead the project and had to give a commitment to raising GCSE achievement in the core subjects of English, maths and science, she said.
A major issue that had emerged was the low level of language and literacy among native English-speaking pupils. The poor writing performance of boys in comparison with girls was a recurring theme.
Islington was now trying to develop a borough-wide language policy for secondary schools and a four-year primary project to produce 99 per cent literacy at age 11 by 1999.
"When we say literacy we mean at least level 3 of the national curriculum. . ." Dr Nicolle said. "That may not seem a very tough target to you but it is an ambitious target for us."
"If they haven't got to level 3 by secondary school they can't cope because they just can't read the books."
The need to raise expectations among parents as well as teachers and pupils was underlined by Richard Jarman, head of Rokeby School for boys in Newham, East London.
Mr Jarman stressed the importance of putting up the exam rate. Since he became head in 1989, he had overseen a sharp increase in the number of pupils entered for exams, in the teeth of much teacher resistance : from 67 per cent of Year 11 pupils to 100 per cent last year. Against his expectations, this had produced both a drop in the number of pupils failing to show up for exams and an increase in grades achieved across the board, not just in low grades.