The attainment gap between rich and poor pupils in grammar schools is virtually zero.
The gap between disadvantaged pupils and their peers at GCSE level is around three percentage points on average in grammar schools, compared to around 23 in other mainstream state schools. This might reflect more about the types of children that go to grammar schools rather than the effectiveness of the schools themselves.
“When we look at the impact of grammar schools, if you look at attainment, for disadvantaged and non-disadvantaged children, the attainment gap in grammar schools is virtually zero, which it isn’t in other schools”
Theresa May, 14 September 2016
While selective systems widens the educational gap between rich and poor, perhaps grammar schools themselves improve the results of less advantaged children who do get in.
The Department for Education told us that Mrs May was referring to the average gap between the percentage of disadvantaged and non-disadvantaged pupils who achieve five or more GCSEs at A* to C level, including in English and maths.
Non-disadvantaged pupils did slightly better than disadvantaged pupils in English grammar schools in 2015 by this measure. The proportion of non-disadvantaged children achieving this level was on average three percentage points higher than the proportion of disadvantaged pupils reaching this level in each school.
Nationally (mainstream state schools excluding grammars) the difference is around 23 percentage points on average. So the gap in grammar schools is a lot closer to zero, but it’s not quite zero.
This analysis doesn’t account for variation between schools, or between pupils.
Some schools have only a few disadvantaged pupils while others have over one hundred. But the gap is similar—four percentage points—if you compare the performance of all disadvantaged pupils in grammar schools to all non-disadvantaged pupils.
Data isn’t given for schools with very small numbers of disadvantaged pupils, so 49 grammar schools are excluded.
Accounting for pupil characteristics, research published by the Sutton Trust in 2008 found that “pupils eligible for FSM [free school meals] appear to suffer marginally less educational disadvantage if they attend grammar schools”. But it said the effect wasn’t completely clear cut because there could well be further differences between grammar school pupils and their peers not accounted for in the study.
Whether rich or poor, grammar school pupils do well in exams and in life. “There is robust evidence that attending a grammar school is good for the attainment and later earnings of those who get in”, as the Institute for Fiscal Studies puts it.
But as we’ve established, there tends to be a trade-off for society as a whole, as there’s “equally good evidence that those in selective areas who don’t pass the eleven plus do worse than they would have done in a comprehensive system”.
Update 28 September 2016
We updated the article, including the conclusion, with the statistics the government said the claim referred to. We also changed the wording of the claim to better reflect the quote.
Update 24 February 2017
We added "This might reflect more about the types of children that go to grammar schools rather than the effectiveness of the schools themselves." into the conclusion. We also slightly expanded on why the Sutton Trust said the effect wasn't clear cut.
This factcheck is part of a roundup of Prime Minister's Questions. Read the roundup.