English test results taken by more than 600,000 14-year-olds will be inaccurate when they are eventually released later this term, teachers and local authority advisers across England have said.

Confidence in the key stage 3 results, which has been shaky at best in recent years, appears to have collapsed in the light of shambolic administrative and marking problems this summer.

Seven out of 10 advisers said they did not think this year’s final English results would accurately reflect their schools’ achievements, a TES survey of one in three local authorities found.

Eighty-five per cent said they thought the results would be less reliable this year than last, following the botched launch of new marking arrangements designed to improve accuracy.

And not a single adviser in the 50 authorities questioned was happy with administration of this year’s tests, which saw results released right at the end of last term without having been thoroughly checked.

Separately, The TES carried out a smaller survey of 35 schools. Some 77 per cent said they did not think the test results would accurately reflect their pupils’ achievement this year.

This week, the Qualifications and Curriculum Authority, which oversees the tests, revealed that three of its board members - Mike Beasley, Edward Gould and Sue Kirkham - are to conduct an investigation into this year’s administrative problems.

The inquiry will cover delays in sending out results this summer, and “operational difficulties” with the tests, but not the quality of marking.

Findings from schools included:

* results which had swung from 74 per cent achieving level 5 or above in 2002, to 49 per cent in 2003, back to 71 per cent in 2004, with similar ability pupils.

* a school which appealed for the third year running, found this led this year to a six-point rise in the results of its 76 papers.

* an adviser who reported that 100 pupils had no marks for the same question.

* One of England’s largest authorities reported that many pupils were wrongly given the same level they had achieved three years ago in KS2 tests.

Many schools complained that pupils asked to answer a question on Macbeth, in which they were asked how the concept of trust is conveyed, were disadvantaged compared to those who studied either Henry V or Twelfth Night.

This year’s results have yet to be released after ministers said the Department for Education and Skills’ chief statistician was unsatisfied that they were “fit for purpose”.

This followed computer problems which meant that thousands of results, in which pupils had scored near the borderline of two national curriculum levels, had not been double-checked.

However, the survey suggests that teachers’ and advisers’ worries extend far deeper than this problem.

One adviser said: “It is unforgivable that the Government, which uses statistics for so many purposes, including performance management, has got it so wrong again this year.”

The picture is not uniform. Some schools and a minority of local authorities say they were confident that their results would correctly reflect their pupils’ achievements.

But the findings are damning. The QCA which oversees the testing process, apologised in July after the results were released late but has argued that the current marking procedures are more accurate than those that went before.

The QCA said it was confident that final results for the tests were accurate.