The Statistics Commission has rebuffed an attempt by the Department for Education and Skills to persuade it that it was wrong after questioning the extent of one of Labour's biggest education success stories. And it has re-iterated a call for the DfES to produce a statement which would set out any statistical caveats behind the change in test scores.
The Treasury-funded commission created a stir in February after publishing a report saying that although there had been improvements between 1995 and 2000 at key stage 2, the test results substantially overstated the rise.
Scores for 11-year-olds jumped from 48 per cent reaching the expected level in English in 1995 to 75 per cent in 2000. In maths, the corresponding rise was 44 to 72 per cent. The commission said the rise was largely down to teachers drilling pupils for the tests.
Sir David Normington, DfES permanent secretary, wrote to say that the department did not accept the commission's findings.
Ruth Kelly, the Education Secretary, also told the BBC's Today programme that the commission's findings were not being accepted.
However, Richard Alldritt, the commission's chief executive, has now said it will not be changing the substance of its position after a meeting with senior DfES officials last week.
He said: "We still feel that the KS2 results for the period 1995-2000 are likely to have overstated the rise in attainment."
The commission accepted that it would have been better for it to have considered the test improvements in English and maths separately. In its report, maths and English were considered together.
The evidence of overstatement was much stronger in English than maths, Mr Alldritt said. But it was still the case that improvements had been overstated when the two sets of results were considered together.
A DfES spokesman said the department was sticking to its position that international studies showed that primary standards were "high and rising".
* Pupils who take GCSEs a year early, or late, will have their achievements included in the secondary school tables this year, writes Helen Ward.
The tables have previously only included the results of pupils who were 15 at the start of the school year, usually Year 11, in which they took the exam. This year's tables will include a figure to show the percentage of pupils who are aged 15, to guard against pupils being held back inappropriately.
The DfES has said that schools wanted age-based reporting to end because it deterred giving pupils alternative arrangements.
The tables will continue to include a figure to show how many 15-year-olds get five good GCSEs until 2008, to allow results to be compared with previous years.