How evidence-based are the Conservative manifesto proposals on grammar schools?

Alice Sullivan. 
The Conservative manifesto confirms Teresa May’s pledge to reintroduce grammar schools, as part of a drive to turn Britain into ‘The World’s Great Meritocracy’. But does this claim stand up to scrutiny?
As has been widely rehearsed in public debate ever since the Conservative government published its schools green paper back in September 2016, the essential reason that grammar schools are such an unlikely tool for promoting social mobility is that working class kids are far less likely than their more privileged peers to attend them. The wider evidence base provides no findings that suggest that selection will help Britain to tackle educational inequality and increase social mobility. Among other sources, a review of the evidence published by Parliament, produced by Parliament’s in-house source of independent analysis, makes this clear.
So, how does the Conservative Manifesto justify the claim that grammar schools will promote social mobility? It makes a startling claim: “Contrary to what some people allege, official research shows that slightly more children from ordinary, working class families attend selective schools as a percentage of the school intake as compared to non-selective schools.”
The source for this claim is a DfE document featuring this graph:

At a glance, this graph in fact shows clearly that pupils from families on below average incomes are far less strongly represented in Grammar schools than in non-selective schools. So, how does the manifesto reach its surprising claim about ‘ordinary working class families?’. This is achieved via some interesting slippage in definitions. The DfE document in effect redefines ‘ordinary working families’. It does so by excluding families who are in receipt of pupil premium funding (which essentially means the most disadvantaged, regardless of whether their parents are in work). So the red and blue parts of the bar are excluded, and only those who are not eligible for pupil premium, but whose families are on below median incomes (the yellow bar) count as ‘ordinary working families’. The manifesto compounds this by extending this definition from ‘ordinary working families’ to ‘ordinary working class families’ (emphasis added). As a result, the bottom third of families no longer count as either ordinary or working class, and are left out of the calculation.
In highlighting this particular example it is important to note that there may be selective use of data elsewhere in the various other party manifestos. But the Conservative manifesto claim that children from ordinary working class families are well represented in Grammar schools seem to go beyond selective use of evidence, and enters the territory of statistical jiggery-pokery. Selective education and grammar schools are particular areas of interest for me, and this is just one example of the fact-checking that academics and others can provide to support informed and evidence-based public debate in the run up to the election. Politicians frequently cite the cause of ‘evidence-based policy’ – they need to live up to their rhetoric.
All the evidence suggests that teacher quality, rather than school structures, is what makes the biggest difference for children’s learning. At a time when Britain faces a crisis in teacher retention and recruitment, I would argue that it would make sense to invest in making the teaching profession attractive rather than putting funds into creating new selective schools.