Exclusive - Prime Minister's GCSE benchmark branded unfair by heads under closure threat.
Forty per cent of secondaries threatened with closure because they are deemed to be failing have above-average scores on the Government's measure of effectiveness.
A TES analysis of GCSE statistics has found that 250 of the 636 "failing" schools have contextual value added (CVA) scores above average, meaning pupils did better than expected. Ofsted has even rated some of the schools as outstanding.
CVA was introduced by the Government two years ago as a fair measure of how pupils have progressed on average since the age of 11. It takes into account gender, age, special educational needs, ethnicity, free school meals status, first language, whether they have moved schools, if they are in care and if they live in a deprived area.
The average score is 1,000. A score of 1,006 means, on average, every pupil at a school achieved the equivalent of one GCSE grade higher than expected in one subject, based on prior attainment.
The 250 schools have all scored above 1,000 but are still threatened with closure or merger - with staff possibly losing their jobs - because Gordon Brown has decreed that all secondaries will be expected to have 30 per cent of their pupils gaining five good GCSEs including English and maths by 2012, which the 250 have not yet achieved.
John Dunford, general secretary of the Association of School and College Leaders, said: "The CVA scores and Ofsted reports of a number of these schools suggest the Government has got it completely wrong. Schools should be supported, not threatened."
Castle Community College in Deal, Kent, has a score of 1,066. This means that, on average, a pupil who could expect to get eight grade Cs elsewhere would get three As and five Bs at the school.
But the secondary modern's "raw score" is 23 per cent, putting it well within the Prime Minister's sights.
Christine Chapman, the head, said: "It isn't fair to be judged as failing because we are below the 30 per cent level. We have been judged outstanding by Ofsted."
The school with the biggest discrepancy is Joseph Rushton Technology College in Lincoln, which has just 13 per cent of pupils getting five good GCSEs but a CVA of 1,057. Steve Davies, deputy head, said: "We run several vocational courses which suit students very well. The idea that the 13 per cent statistic means it is a failing school is misguided. You can't take one set of statistics and draw conclusions from that - 74 per cent of students get five good GCSEs or equivalent."
The 362-pupil school is due to merge with two primary schools.
Lord Adonis, the schools minister, also used raw scores to criticise schools last year, suggesting that those missing the 30 per cent target "lacked leadership and vision". But a TES analysis of Ofsted reports showed that half of the schools he criticised had good or outstanding leadership. Only 6 per cent were inadequate or unsatisfactory.
Although the advent of contextual value added is welcome news for some schools, it could have serious consequences for others.
A law introduced last year says schools in the bottom quarter nationally on CVA scores can be subject to intervention by their local authority. It can issue them with a warning notice, force them to work with other schools, replace their entire governing body, and take back their delegated budget.
So while Mr Brown and others criticise low scorers, authorities may also be taking action against schools which have above average GCSE scores but are deemed to be "coasting". With pupil data under the microscope, few schools can rest easy.
To see the end results of schools and pupils, first look at their starting points.
Working out how well pupils have done at secondary school should be simple - they have either got five good GCSEs or they haven't, writes Helen Ward.
But if you want to measure how well the school has done, it is thought to be fairer to look not only at pupils' end results but also their starting points, and to factor in elements that could affect performance - for example, whether they speak English, have special educational needs, their ethnicity and level of deprivation. So contextual value added (CVA) was invented.
CVA scores can give schools that have low raw scores a chance to shine. Moreton Community School in Wolverhampton has just 30 per cent of pupils gaining five good GCSEs including English and maths, but a CVA of 1,090.5.
It is also possible for some schools with high raw scores to have a low CVA score: for example, in schools where almost all pupils leave with good exam results but where they all started with good key stage 2 test results.
But the real difficulty is that as the numbers are calculated, they become more difficult to read. And if you are a pupil, you either have a C grade in English GCSE or you don't.
CVA includes a "confidence interval", which is dependent on how many pupils take the exams. This is a statistical measure which gives a range of scores within which the "true" school effectiveness lies.
The confidence interval is why Deborah Wilson, of Bristol University's Centre for Market and Public Organisations, said that ranking schools by CVA was "extremely misleading" because the difference in CVA scores between schools is not always statistically significant: for example, there is usually little to distinguish between a school with a score of 1,005 and one with 995.
But in a paper co-written with Anete Piebalga, Ms Wilson concludes that the measure is not worthless.
It states: "Overall, school-level CVA is a more accurate performance measure than its predecessors," she said. "It better isolates school effectiveness from composition and so provides a better measure of the actual impact of a school on pupils' progress."