Competition between schools is not raising exam standards, research by the Open University has found.
If anything, the free market seems to depress exam results, according to a study which also suggests higher scores in grant-maintained schools are largely due to improvements in their intake.
To find out whether competition between schools does "drive up standards" - as ministers in the last government claimed - the OU team surveyed 319 secondary schools in 89 areas.
Not all schools are subject to the same degree of competition. Researchers recognised that choice can be affected by where schools are, and that schools and parents may react differently to attempts to create an educational market.
The degree to which the survey schools were affected by competition was assessed. Factors considered were the extent to which parents exercised choice and what opportunity they had to do so - taking into account other nearby schools and surplus places.
Between 1991 and 1996 the percentage of pupils obtaining five good GCSEs in the survey schools increased by 0.15 a year above the national rate of improvement. But in the 14 areas where competition was stiffest the rate of improvement declined in relation to the national rate of improvement.
Schools in areas with little or no competition averaged an increase of 0.18 percentage points a year over the five years.
Reporting these results to this week's International Congress for School Effectiveness and Improvement in Manchester this week, Rosalind Leva#249;cic#171;, an OU researcher, said school performances varied considerably. Only 13 improved consistently year on year. A total of 76 improved in four years out of five.
She said: "Improvement was worse in areas of high competition. There was no evidence to confirm the claim that competition improves schools. "
The team also tried to find out what did contribute to improved exam results.
Negative factors included high proportions of children on free school meals. Above-average improvements were found in schools with falling numbers of children on free meals.
This suggests either that improving the social composition of a school intake improves its exam results or that improving schools attract more middle-class parents.
A third of the schools in the OU survey were grant-maintained and their rates of improvement were higher on average. But, after the effect of the changing social composition of their intakes was allowed for, GM status was not associated with higher than average improvement.
Similarly, whether a school was large or small, a church school or a single-sex school appeared to have no consistent effect on improvement in exam results once other factors had been allowed for.
Improvement was lower in secondary modern schools in selective areas.
In 1996, Gillian Shephard, the former education secretary, argued that: "The existence of a range of different schools drives up standards for all our children."