Evidence produced this week by the Department for Education and Employment to demonstrate that comprehensives perform better when they are in competition with grammar schools is based on the GCSE results over two years in 20 schools.
Gillian Shephard, the Education and Employment Secretary, told MPs that there is evidence which suggests that, in areas where there is some selection, standards in all schools are higher. The claim runs counter to the view that grammars "cream off" the most able pupils, thereby having an impact on the results in neighbouring schools.
According to Mrs Shephard,the existence of selective schools is a spur to other schools and is the justification for the Government's claim to be raising standards by introducing greater selection.
The DFEE evidence of the positive impact on other schools of grammars is based on an analysis of the greater rate of improvement at GCSE by non-selective schools in Salisbury, Wiltshire, and in south Birmingham.
The figures show that the four non-selective schools in Salisbury, which compete against two single-sex grammars, have improved their exam results over the years 1993-95 at a rate twice the national average.
The proportion of fifth-formers gaining five or more higher grade GCSEs has gone up by 5.7 per cent, compared with a national figure of 2.4 per cent. The number of pupils who gained five or mores passes at grades A to C rose from 31 per cent to 36.7 per cent. However, the average rate of improvement across Wiltshire schools in that period has also been 5.7 per cent. Nationally, the success rate rose from 41.5 per cent to 43.5 per cent.
The local authority described the study as a fairly speculative use of data. Chris Dale, head of planning and resources, said careful research was needed before any such conclusion could be drawn.
In south Birmingham, the study found that the average rate of improvement in 16 non-selective schools within a three-mile radius of a grammar was 7. 3 per cent, compared with the Birmingham average of 4.3 per cent and the national average of 2.4 per cent. In the 16 schools, the number of pupils gaining five or more higher grade GCSEs rose from 24.3 per cent in 1993 to 31.6 per cent in 1995.
Tim Brighouse, chief education officer in Birmingham, is sceptical that the existence of the grammar school had any impact on the results of the comprehensives. The schools have not been identified by the DFEE, but city officials believe they are a clutch of schools close to either King Edward's boys' grammar in Camp Hill, King's Heath, or King Edward's Five Ways, a boys' grammar in Bartley.
"Overall the biggest improvements at GCSE have been amongst girls, so it is difficult to connect the presence of a boys' grammar with the improved rate of results," says Mr Brighouse. "I suspect the siting of the grammar school is no more significant than the siting of the motorway or the airport."
The more likely explanation, according to the local authority, is the stress that is being placed on school improvement.