Research in Hampshire primaries casts doubt on the Government's formula for giving schools an effectiveness rating taking account of the social class of pupils.
The Government is planning to produce more sophisticated rankings of schools alongside the tables based on exam results to show the value-added element.
However, the research by Harvey Goldstein, professor of statistical methods at the Institute of Education at London University, suggests there are problems in producing an overall value-added score for individual schools.
Analysis of data on 6,400 pupils in 161 primaries suggests that schools can have different value-added scores in maths to those in English. Schools may be particularly successful with children assessed at five to be below average, but may be less successful with average or above-average children.
The research used baseline assessments of four-year-olds matched against achievement at seven and also progress made by pupils from the age of seven to 11.
The results suggest that baseline assessment is a relatively accurate predictor of performance at seven. Levels of success at seven are strongly associated with pupils' family background. The schools with the highest levels of free school meals have lower aggregate test scores. However, positive influences are the number of terms of schooling before the tests at seven and the age of the child.
Gender is also a factor - girls tend to do better on reading and writing, but worse in maths and science.
A seminar presenting the results of the study, along with a similar study in Surrey will be held at the Institute of Education on October 22. Anyone wishing to attend should contact Sue Cranmer on 0171 612 6651