When it comes to "adding value" to pupils' results, Wales's lowest-performing local education authorities fare poorly - regardless of whether poverty or previous attainment are taken into account, according to Assembly government figures.
But academics are divided over whether the new data can be used to judge LEA performance. The Assembly government has awarded a three-year contract to the Fischer Family Trust to develop value-added measures for schools and LEAs in Wales.
The Trust is working on three models, which take into account pupils'
previous attainment; previous attainment, gender, and month of birth; and all of these factors plus eligibility for free school meals (poverty) and other demographic data.
Last summer, the Assembly government issued secondary schools and LEAs with value-added data about pupils who took GCSEs in 2004, based on the last of these three models.
The data, obtained by TES Cymru under the Freedom of Information Act, shows pupils in Monmouthshire, Wrexham and Denbighshire consistently underperformed against expectations when it came to passing five or more GCSEs at grades A*-C, irrespective of which model was used.
Pupils in Neath Port Talbot, Vale of Glamorgan and Carmarthenshire consistently overperformed at this level, on both models.
Results varied depending on which model was used, and there were some contradictory results. And in a third or more of Wales's 22 LEAs, the value-added results were not statistically significant. This means they could have been down to chance rather than the efforts of schools and LEA officers.
David Reynolds, professor of education at Plymouth university, said generally lower funding levels in Monmouthshire, Wrexham and Denbighshire could help explain pupils' under-performance. Independent schools may be skimming off the most able pupils in Monmouthshire, while the low-skills economy of the two northern authorities could be contributing to lower expectations.
In contrast, pupils in Neath Port Talbot are benefiting from better funding, an authority-wide focus on school improvement, and joint working by the schools.
"Whatever the precise explanation, the general picture is one of considerable variation in the results of the Welsh LEAs - more than 15 percentage points on the GCSE measure," he said. "This variation is likely to dramatically affect the life chances of pupils."
But Dr Robert Coe, director of secondary projects at the curriculum, evaluation and management centre at Durham university, said it was difficult to judge LEAs' comparative performance using value-added data.
Even school-by-school comparisons may not give a fair picture, because of the aggregation of a wide range of results.
"When you look at the value-added of a school, you are looking across a range of subjects. If you look at individual subjects, in almost every secondary some are good and some bad," he said.
"When you look at the whole school, you are pulling together a mixed picture and overall average. The latter is not particularly meaningful, and at the level of the LEA, it's even worse."
But he said it was important for schools to have as much data as possible about pupils' performance to focus on improving results. "It is at the classroom level that you can make a difference."
Schools and LEAs have already received updated value-added data for the 2005 pupil cohort, for all three models. An Assembly government spokesperson said it was seeking feedback from schools and LEAs on all the indicators and models being piloted.
opinion cymru 19