Borderline pupils who just get into grammar schools in England do better there than they would have done at a comprehensive, a study suggests.
Borderline pupils 'benefit most from a grammar school'
The results of such pupils are better than those of children with similar abilities who went to comprehensives.
The brightest children do equally well wherever they go, studies for the National Foundation for Educational Research (NFER) found.
Borderline pupils benefit from being around higher-achievers, it is thought.
The report said the main impact of grammar schools appeared to be on pupils at the bottom end of their intake range, who narrowly manage to pass the entrance tests.
The report concluded that the brightest children did well wherever they went.
"Contrary to some claims, selective education appears to have little or no impact on pupils at the very top of the prior attainment range.
"The most able pupils make at least as much progress in comprehensive schools."
It also found that grammar school pupils on average make more progress than those in other schools between ages 11 and 14, but from age 14 to 16 the reverse is the case.
"Over the whole five years, from 11 to 16, grammar school pupils make slightly more progress than others," it said.
One reason for the earlier advantage might be that in the Key Stage 3 mathematics and science tests, pupils are entered for different papers or "tiers", which determine the levels they can achieve.
"Pupils in grammar schools are very much more likely to be entered for higher tiers than pupils of equal prior attainment in other schools," NFER said.
The researchers based their findings on a series of studies comparing the outcome of value-added data about children's performance at the ages of 11, 14 and 16, provided by the Department for Education, school profiles and information from Ofsted.
Initially, they looked at Slough in Berkshire, they said, and in later studies found the experience there was reflected across England.
They used statistical models to measure the performance of children in a year group against a number of variables.
Ian Schagen, head of statistics at the NFER, said the group's work was often seized upon by both sides of the grammar school debate.
"The issue of selection can bring out a lot of emotion. We have found our research quoted by both sides."
He said the group was doing further studies which would take into account new school census material which records social data such as home language and take-up of free school meals.
This he said, would give more information about the intake of various types of schools, allowing researchers to look at the effect this has on school results.