Value-added tables are unreliable. So why is Charles Clarke using them to sack heads? Jon Slater reports
When Charles Clarke promised to "take out" bad headteachers he was prepared for a row. The charges of control freakery, bully-boy tactics and political interference in schools that followed were all so much water off a duck's back.
His message was clear. Underperforming secondaries need fresh leaders. That means getting rid of the existing ones. You can't make an omelette without breaking eggs.
He went into battle armed with pound;525 million in leadership incentive grants with which to buy off unwanted school bosses so they would go quietly.
Mr Clarke aims to target heads by using value-added league tables, published for all secondary schools in England for the first time last year. These allow the Government for the first time to distinguish between schools whose poor results are due to their difficult intake and those who are failing their pupils.
As an added bonus they also identify "coasting" schools - these are further up the league but should be doing even better given their intakes.
Value-added is a typical new Labour "third way", between those on the left who would scrap the publication of results and those on the right who champion league tables based on "raw" results, which give no indication of the different challenges schools face. But the Government has tilted more towards the latter.
The Department for Education and Skills developed value-added solely "for accountability purposes", education minister Margaret Hodge told Professor David Jesson, an expert in performance evaluation at the University of York. In other words they exist to judge and rank schools.
Indeed it was only at the last minute that ministers decided not to use value-added data to identify failing schools as part of the launch of the Government's London strategy last month.
But as Mr Clarke increasingly relies on value-added scores, doubts are growing over their validity. The problem is that unless the data is accurate, good as well as bad headteachers could be forced out.
Value-added scores in the official secondary performance tables show whether a school has performed better or worse than expected at GCSE given pupils' results in key stage 3 tests.
Professor Jesson, who produced his own league tables for the Specialist Schools Trust, argues that official calculations contain three major flaws.
First, the Government's calculation does not measure pupils' progress from the age of 11 through to 16. Two separate sets of tables for 11 to 14 and14 to 16 are published. This means schools whose pupils progress quickly at one key stage are penalised in the other. Some schools do well on one measure and poorly on the other.
Second, the tables are unfair to schools with more boys than girls, as they ignore gender differences in performance.
Third, Professor Jesson also complains that the Govern-ment's tables are based on a measure that schools do not understand. While this makes clear to schools where they need to concentrate their efforts, it means the overall picture of a school's performance can be unclear.
The DfES measure gives points for each GCSE grade. Each pupil gets a score based on their best eight GCSEs - and these are used to judge a school's performance. This score, say critics, is "almost impossible to understand" as it is not a measure widely used in schools.
These differences produce big variations in how the same school does in the Government's table and Professor Jesson's (see box). "While our measures may be in agreement about the 'best' and 'worst' schools there is much less agreement about schools with less extreme evaluations," he says.
The TES picked 14 schools from different levels in the Specialist Schools' Trust value-added tables and compared their standing with their rating on the DfES's value-added measure.
Although some schools' performance was similar on both measures, for others such as Hugh Christie technology college in Kent, there was a big difference (see table). Jane Prideaux, acting head of Hugh Christie, said:"We have queried why the results are vastly different with the DfES.
"Obviously we feel the school is nearer to the trust's assessment than the DfES's. It could be a concern, but you have to hope that the LEA will look at other measures before they come marching in."
But Professor Jesson's methodology also has its detractors. By using the number of pupils gaining five Cs or better as his measure of success at 16, he ignores all achievement below that level, they say.
As heads pointed out at the public launch of his work, this would give nervous heads an incentive to focus teaching effort on pupils just below the five A*-C standard at the expense of their peers.
Carol Fitz-Gibbon, professor of education at Durham university, has a further concern about both the Government's and Professor Jesson's tables: both make the cardinal error of judging all GCSEs to be of equal value.
But, as she recently told MPs on the Education Select Committee, it is harder to get a C in physics, maths or languages than in a humanities subject (Opinion, 27).
She runs her own value-added project (called YELLIS) with selected schools which weights subjects by difficulty. It measures progress from 11 to 16.
She normally discourages publication of her results, but in 2001 made an exception.
Professor Fitz-Gibbon wrote to 17 schools officially deemed to be under-performing, but which, on YELLIS figures were doing better than expected. They duly published their figures.
It seems schools who encourage pupils to take exams in "hard" subjects such as maths are penalised by the Government's system.
If heads' jobs depend on their value-added rating they could be tempted to encourage pupils to drop science and languages.
Professor Fitz-Gibbon accuses the Government of providing "fictional data".
"There are foolish incentives in the system which penalise schools that encourage students to take the very subjects that industry says it needs," she says.
Professor Fitz-Gibbon also believes Jesson's method and the Government's give a distorted picture of performance.
The Government's use of the eight best GCSEs encourages schools to enter pupils for many GCSEs knowing their lowest grades will be discarded.
Heads' organisations have seized on these disputes to argue that value-added data is still not good enough to be used to select heads for the sack.
John Dunford of the Secondary Heads Association said: "I am worried about the whole agenda of taking out heads, it is the wrong language and the wrong approach.
"In theory, value-added is a better indication of school performance but unless we get the methodology right, the basis for decisions will be as bad as if they used the raw scores."
With senior government figures admitting privately their method may not necessarily be the best, there will be changes. Although the 2003 tables will be unchanged from last year, a pilot project will test value-added tables at key stages 3 and 4 in selected schools. If they prove a success, they will be produced for all schools from 2004.
Shadow performance tables taking account of qualifications not currently included will also be produced, but not published.
A DfES spokewoman said that value added would be"one of a range of measures" used to assess how schools are doing. Others would include raw GCSE scores and the deprivation of the surrounding area.
The DfES points out that while Professor Jesson's measure only gives credit to pupils who get at least 5 A*-C grades, official value-added includes the achievements of all.
In the meantime, the stark truth is that for some heads the method used to calculate value-added league tables could make the difference between a pat on the back and the sack. There are, it seems, lies, damned lies and value-added statistics.