No evidence grammar school systems are best for the brightest, study of 500,000 pupils reveals

The study, which was based on test results and other information from all pupils in England, challenges the common belief that high achievers do better in more academic schools.

Findings, published in the peer-reviewed journal Educational Review, demonstrate those with the highest scores at age 11 were less likely to achieve five top GCSE grades if they went to grammar school than if they attended a comprehensive, once other background factors were taken into account.

Lead author Dr Binwei Lu, a member of Durham University Evidence Centre for Education, and now an assistant professor at Zhejiang University in China, says the results show "no evidence of a superior academic effect of selective systems on pupils' academic outcomes."

"While the general results for the effects of the two systems reveal neither system to be superior, an internal pattern implies negative results from the selective system, from which both high-performing and low-performing pupils may suffer."

Overall, the results show little difference in GCSE pass rates between areas with grammar schools and areas which were non-selective. The paper concludes that expansion of England's 163 grammar schools would be expensive and unwise, given the lack of evidence that it would raise standards.

The team, which also included Dr Jake Anders, from the UCL Institute of Education in London, as well as Professor Nadia Siddiqui, and Dr Xin Shao from Durham University, looked at data on almost 500,000 pupils who attended state schools in England and who took GCSEs in 2016. In addition to their SATS test scores at age 11, it took into account social background, ethnicity, special educational needs, gender and the month in which they were born, all of which can affect exam performance.

It found grammar school pupils had about 90 per cent of the odds of achieving five A or A* grades compared with those in non-selective areas. This remained the case when the sample was restricted to those with high scores at age 11 and was also true for those who attended non-selective schools in selective areas. Since 2016 the grading system has changed and this now equates to grades 9, 8 and 7.

The research also looked at how selective and non-selective areas fared in getting pupils five GCSEs at grades A*-C (now grades 9-4), and found no difference between selective and non-selective areas. While there was a slight positive effect for those attending grammar school, there was an equivalent negative effect for pupils in non-selective schools in those areas.

The researchers concluded that, contrary to popular belief, educational selection was detrimental to high performers. They suggested three possible reasons:

• The 'big fish little pond effect' -- pupils in selective schools might see themselves as less able in comparison to their classmates, and this might be demotivating.

• The stress of a competitive system might have a negative effect on performance.

• There might be wider issues which the research could not identify: for example, high performance in London, which does not have many grammar schools, could pull up the average for non-selective systems.

Meanwhile, they said, there is a risk that pupils who do not get into grammar school suffer from a lack of role models, a sense of failure and a higher concentration of disadvantaged pupils in their non-selective schools.

Co-author Dr Shao, says any further expansion of grammar schools would be likely to disadvantage the able pupils they were designed to help.

"Our study adds to the evidence that the expansion of grammar schools and the selective system is unlikely to raise national academic standards," she explains. "The costs of reorganising our education system to have more selection would be high, and there are much more important priorities for investment to support equalising opportunities for those regardless of their family background: expanding the selective system would not be a wise decision."

Limitations of the research include the linear and logistic regression model used. The authors recommend that future studies add a regional element to this new analysis to "further explore the 'London effect' in the effectiveness of selective schooling."