INSPECTORS' first impressions of schools have been formed using misleading data, according to a new report commissioned by the Office for Standards in Education.
Before visiting a primary school inspectors look at the number of children receiving free meals and at infants' test scores.
They have been using these figures to put a school's national test results for 11-year-olds into perspective. But OFSTED has now accepted the report's conclusion that current practice was "likely to produce misleading comparisons among schools". From Easter the figures will come with a warning of their limitations.
The researchers, led by Harvey Goldstein of London University's Institute of Education, also criticised the Government's use of league tables - showing 11-year-olds' raw national test scores to compare school performance - saying it did not give a reliable picture.
The researchers warned against using free school meal data or test results for younger pupils alongside raw scores, arguing that such "proxy measures" still produced misleading comparisons.
They argued that only a full value-added analysis - one which adjusts raw data to take account of pupils' achievement at the time they entered school - will allow for fair comparisons between schools.
Professor Goldstein said: "Our results provide little support for the routine use of proxy measures, either for inspection purposes or for publication as 'league-table' rankings of schools.
"Because these measures are not valid indicators of educational quality, it is essential that anyone whowishes to use them, especially for school inspections, must fully understand their nature and limitations. Information about pupil attainment is vital, but it must be handled with care in order to avoid misleading conclusions."
An OFSTED spokesman said: "We are specifically taking Harvey Goldstein's advice and sounding a note of caution to all the people who might use this data - inspectors, heads, governors and teachers. There are clearly significant caveats about this information but it is an interesting starting point of discussion for an inspector to talk to the head teacher about."
The study was carried out by researchers at the institute and at Hampshire County Council, which has devised its own value-added system linking baseline assessment to national test results.
However, most schools do not have access to such complex data and are not always able to analyse it. The researchers therefore investigated whether more readily available information ("proxy measures") would produce sound comparisons.
The team analysed test scores for 11-year-olds in Hampshire using proxy measures, including gender, eligibility for free school meals, age and average test scores for seven-year-olds in the same school.
But these measures still produced misleading comparisons, providing little improvement over the use of raw test scores.
"The use of value-added information in judging school performance," published by the Institute of Education, is available from the institute bookshop, 0171 612 6050, (email email@example.com), priced at pound;6.95.