Girls' schools do well in exam league tables because they have bright pupils, not because they are single-sex, according to new research to be published next week.
With girls' schools expected to excel in this year's A-level and GCSE results tables, the findings from Manchester University will prove controversial.
Since the advent of league tables, girls' schools have used their exam results to argue the case for single-sex education.
However, a study of research evidence over the past 30 years carried out by Professor Alan Smithers and Dr Pamela Robinson challenges their claim.
A report commissioned by the co-education group of the Headmasters' Conference of top public schools, most of which now admit some girls, says that girls' schools do well not because they are single sex but because of their pupils' ability and class.
Once these factors have been taken into consideration, there is no significant difference between the performance of single-sex and co-educational schools.
The researchers say that, although girls temporarily improve their performance when they are taught separately for subjects such as science, the gains are short-term and are probably because of the novelty and excitement generated by the experiment.
They point out that most girls' schools are selective either because they are fee-paying or because they are grammar schools. There are only 169 girls' state schools, 54 of them selective.
Various studies have examined the advantages of single-sex and co-education for girls. One, by the late Professor Desmond Nuttall of London University's Institute of Education, found a slight advantage in teaching girls separately.
But the Manchester researchers say the difference is no longer statistically significant when other factors in school performance are taken into account.
Work by Gerard McCrum of Hertford College, Oxford, released earlier this year attributes the declining performance of state school girls at A-level to the big increase in co-education. Fewer state school girls get the top grades than 20 years ago, whereas the performance of girls in independent schools has improved.
However, the Manchester researchers believe today's results cannot fairly be compared with those in the Seventies because a much smaller proportion of girls took A-levels 20 years ago. In addition, more parents are now sending their daughters to fee-paying schools.
Since the introduction of league tables, a number of schools such as Shenfield School in Essex have begun to teach girls separately.
In Bristol, a research project in which girls in a mixed comprehensive were taught separately for science showed that their GCSE results and their confidence improved, but the study ended after three years.
Professor Smithers and Dr Robinson say research into school effectiveness suggests the size of a school's contribution to exam performance that can be identified is quite small. Whether the school is single sex or co-educational plays no part.
The second phase of the research will look in detail at individual schools and of teaching girls separately.Reuse content