The question of who will get in lies at the heart of the debate about the new generation of grammar schools that the government has proposed.

As we, and plenty of others, have pointed out, children eligible for free school meals (FSM) – a commonly-used proxy for disadvantage – are disproportionately unlikely to get into existing grammar schools.

In recognition of this, in their green paper the government has suggested that new grammar schools could be required to take on quotas of disadvantaged kids [PDF], along with a number of other proposals such as feeder primary schools in disadvantaged areas, and selective schools working closely with non-selective schools.

Why poor children don’t go to grammar school

There are two factors that explain why so few disadvantaged kids go to grammar school.

One is that poor children have lower attainment at age 11 than their peers. This in itself is a big issue, and one that needs tackling. But straight off it means that disadvantaged kids are less likely to pass the 11-plus and go on to grammar school.

The second factor is that, even where a poor child and a child who is not poor have equal attainment at age 11, the former is considerably less likely to go to grammar school than the latter.

1-non-fsm-current 2-fsm-current

This second point, given sufficient political will, would probably be solvable.

The first factor – lower attainment at age 11 among disadvantaged kids – is an issue that, as we say, needs tackling, but which will be considerably harder to fix.

So if the government wants to see more poor children go to grammar school, we should ask the question: how many poor children do they want to go to grammar school, and how far would they be willing to go to achieve this?

Changing the entry requirements

This analysis focuses on four selective local authorities (LAs): Kent, Medway, Buckinghamshire and Lincolnshire1.

Together, grammar schools in these areas admitted 10,196 children in 2014/15.

We have Key Stage 2 (KS2) scores for 9,504 of these children – most of the remainder presumably having gone to independent primary schools where KS2 tests weren’t taken2.

As we don’t really have any information on the children who didn’t take KS2 tests, we’re just going to have to set these nearly 700 places aside, and think about the 9,504 places that went to children who had taken KS2 tests.

The National Pupil Database records that 306 of these 9,504 children were eligible for free school meals, and 9,198 weren’t – a rate of 3.2 per cent, compared to around 10.1 per cent of children in general in these LAs3.

These are LAs that are already fully selective. So we might wonder: how different would things need to be in these LAs for a representative number of disadvantaged children to go to grammar school?

Let’s be optimistic

If the grammar schools in these LAs were to get up to a representative rate, that would require 958 of the 9,504 places we’re focussing on to go to FSM-eligible kids.

Data from 11-plus tests isn’t available, so instead for this analysis we’re going to have to look at KS2 results (specifically, an average of children’s maths and English test scores).

As discussed elsewhere, KS2 test scores aren’t perfect for doing this kind of analysis – they’re not particularly highly correlated with 11-plus results, for example. In the absence of anything else, though, they’re the best option we have.

Let’s assume that 75 per cent of FSM-eligible kids above a certain KS2 score get in to grammar school. This itself would not happen easily – for non-FSM-eligible kids, you currently have to get up to a KS2 score of about 5.6 before we get that sort of proportion going to grammar school – but let’s be optimistic and assume there’s a way to achieve this.

Counting back from the highest-scoring pupils, we’d get to a KS2 score of 4.8 before we had 1,004 FSM-kids who could go to our grammar schools – a little more than the 958 spaces we were hoping to fill. (This does place more stock in a relationship between KS2 scores and 11-plus than we ideally would, but this isn’t the only assumption we’re having to make in this modelling.)

That’d leave exactly 8,500 places for non-FSM-eligible kids.

For non-FSM-eligible kids, counting back from the highest-scoring pupils – here counting those who actually got into grammar school, rather than having to apply a 75 per cent figure – we’d get to a KS2 score of 5.

That actually gives 8,587 non-FSM kids, taking our total to 9,591 versus the 9,504 we were hoping to recruit – so it will be a bit of a squeeze.

This would make 10.5 per cent of the grammar population eligible for free school meals – a little over the 10.1 per cent who are eligible in general in these areas.

What would this mean in practice?

On the face of it, those KS2 scores – 4.8 and 5 – don’t sound too far apart.

But these are the minimum scores which we’d let a child into grammar school with.

When we look in more detail at the results of these two groups, it becomes quite obvious how different their attainment actually would be.

Among the 1,004 FSM-eligible children who would be going to grammar school, the most common – or modal – KS2 score they would have is 4.8.

For non-FSM-eligible children it would be 5.4.

Similarly, comparing means, the FSM-eligible children attending grammar school would have an average KS2 score of 5.1, while for non-FSM-eligible children it would be 5.5.

3-non-fsm-proposed

4-non-fsm-proposed

These are not small differences. (Considering the modal scores, we can think of it as a child in one of the groups having achieved Level 4a, and a child in the other group having achieved Level 5b. If children aren’t told what sub-level they get, the difference sounds even starker – Level 4 versus Level 5.)

Would this be acceptable to those who favour a return to selectivity – specifically, would the label of selective be recognised by the parents of kids with straight Level 5s at KS2 (most of whom would not be FSM-eligible)?

These most definitely would still be selective schools, but it’s debatable whether parents would accept the label.

So what would the government choose to compromise on: the number of FSM-eligible children who go to grammar school, or the schools’ perceived selectivity?

We would argue that this conundrum means the idea of a selective system that is truly socially representative is, in practice, not going to be workable.

Read more of our coverage of grammar schools.

Image: Elizabeth Ellis CC-BY-SA

Notes

1. These areas have been focussed on as they are fully selective, and, due to their size, are less affected by cross-border flows than some other fully selective LAs.
2. This isn’t the total number of children who had been at an independent primary, as children at some independent schools – and particularly those in heavily selective areas – sit KS2 SATs. It looks like a total of 1,001 children out of the 10,196 who went to grammar school in these areas, or 9.8 per cent, were at an independent primary school.
3. This likely understates the general FSM rate, as it counts those who attend school in these four LAs. Cross-border flows are not trivial, and it seems reasonable to think that children are more likely to travel from a different area to attend a grammar school in one of our four LAs, than to attend a secondary modern. And as we’ve seen, the proportion of children who attend grammar school who are FSM-eligible is low.