Don’t harm opportunities for other people’s children

When we consider progression to higher education, the focus usually centres on eligibility for free school meals and exam results.

But in my Higher Education Policy Institute (HEPI) Occasional Report, ‘The Impact of Selective Secondary Education on Progression to Higher Education’, I consider a number of measures, including POLAR, households below the median income and ethnicity.

Looking through each of these lenses, we can see startlingly different results. Although selective and non-selective areas performed similarly in terms of progression to HE as a whole, when it came to highly selective HE – particularly Oxbridge – the gap widened. In particular:

  • 39% of pupils in selective school areas progress from state schools to highly selective universities, compared to just 23% in comprehensive areas;
  • a state school pupil from the most disadvantaged quintile is more than twice as likely to progress to Oxbridge if they live in a selective area than a non-selective area; and
  • a state school pupil with a BME background is more than five times as likely to progress to Oxbridge if they live in a selective area rather than a non-selective area.

The report also found that the specialist maths schools run by Exeter and King’s College London also performed outstandingly in terms of progression to HE. It also discusses why public attitudes to grammar school expansion might differ from those of educational experts.

Questions and critiques

Since publication, there have been a number of helpful critiques of the report, some of which I want to address here.

One of the report’s headline statistics was about BME pupils. A number of people rightly noted that the BME grouping is broad, and that some groups within it, such as black children, face particular disadvantage in terms of educational attainment and progression to HE.

This is a fair observation. The question of black pupil progression is addressed on page 33 of the report. A black state school pupil is 3.1 times as likely to progress to Cambridge if they live in a selective area compared to non-selective; an impact lower than that seen for BME pupils overall (5.7 times as likely) but still a highly positive one. To put it another way – had comprehensive areas sent black pupils to Cambridge at the same rate as grammar areas, Cambridge would have had more than twice as many black state-school students matriculating between 2015 and 2017.

Another concern is whether the paper sufficiently takes into account the fact that some grammar school pupils cross local authority areas in order to attend. Again, this is a good question to ask: the most reliable estimate of the number of grammar school pupils who travel in from a different local authority area is 25%. If one takes a maximal approach and reallocates a full quarter of the selective area pupils with outcomes under examination to non-selective areas, the results remain sizeable and significant. For example, POLAR quintile 1 pupils are now only 1.9 times as likely to go to Cambridge if they live in a selective area (down from 2.3 times); selective areas now send 37% of state school pupils compared to 24% in non-selective areas (changed from 39% and 23%). So the findings cannot be explained by this phenomenon.

Some have observed that it is not only pupils eligible for free school meals who less likely to attend grammar schools, these schools are more likely to take students from better off backgrounds generally. This is true, but a large number of students from groups such as ‘below the median income but not formally disadvantaged’ still do obtain access. The report’s findings demonstrate that notwithstanding such differentials, the relative benefit of living in an area with grammar schools – in terms of increasing the likelihood of progression to elite HE – is still positive.

The most severe criticism levied has been misuse of data, in particular the statement that 45% of pupils at grammar schools come from below median income families. It is pointed out that the figure is provisional and suggests flaws with it – though the critique fails to mention that the Department for Education document explicitly takes steps to remedy these. The median figure used by the DfE (and therefore by me) is not £24,800 (the national figure), but instead the median household income measurable in their dataset, which is somewhat lower. It is also suggested that if this figure is not accurate, it calls into question the other findings of the report.

I acknowledge that the figure is provisional, but it is the best estimate currently available. Two of the seven recommendations in my report are to call for better data and further research.

More importantly, the key findings on progression to HE do not depend on the 45% figure and are drawn from independent data sources, as referenced in the report. This includes the fact that 39% of pupils in selective school areas progress from state schools to highly selective universities, compared to just 23% in comprehensive areas, and that England’s 163 grammar schools send 30% more BME pupils to Cambridge than the nearly 2000 comprehensive schools. These are questions that must be answered by champions of the comprehensive system.

I do not expect this to be the last word on grammar schools and social mobility. Disadvantage has many components and academic success can likewise be measured in different ways. Further research, greater alignment of data and deeper analysis would all be welcome. But what the report does show is that for some groups of less advantaged pupils, grammar schools transform the opportunities available. The question of their contribution to social mobility is more complex than whether or not they take sufficient pupils eligible for free school meals.

Other people’s children

Many people reading this article will be opposed to the idea of grammar schools. That’s not surprising – after all, their children don’t need them.

For an academic or senior administrator, it doesn’t matter if the local school is only adequate: with their education, they can always make up the difference at home. And if their son or daughter applies for Oxbridge, well, they’ll know plenty of people who can give them a practice interview or prepare them for the Sixth Term Examination Paper should Poppleton Comp not know how to teach it. A few might even (whisper it) hire a tutor or go private.

But now consider the warehouse supervisor who wants her son to be a doctor, the shop assistant who wants his daughter to be an engineer, or the immigrant parents who don’t speak good English but are passionate about their child achieving the very best. None of these people’s children would be eligible for free school meals but their opportunities, privilege and understanding of UK higher education are worlds away from those of the children of a lecturer, banker, senior civil servant or lawyer. These are the ordinary working families for whom grammar schools transform lives.

So the next time you get squeamish about selection, remember that grammar schools aren’t there to benefit the children of the top 20% – those children will do ok in any school system. And think twice before being so keen to remove opportunities from other people’s children.

8 responses to “Don’t harm opportunities for other people’s children

  1. You ignore another serious problem with your methodology – as you are using DfE KS5 Destinations Data for school sixth forms you exclude all young people who either don’t do A-levels (or equivalent) or do them in FE Colleges or Sixth Form Colleges. As a result, you exclude around half the cohort! How can you draw any serious conclusions?

    Your defence of ignoring cross-border transfers – up to 75% at age 11 in some areas – does not stand up to scrutiny. I fear you also misunderstand what POLAR is measuring: there are – it’s not an individual-level measure of disadvantage, (essentially) excludes London and other big conurbations, 86 out of 326 LADs have no-one at all living in POLAR Q1 and half of uni entrants from POLAR Q1 with a known social background are middle-class.

    Finally, you misunderstand the DfE income measure. It looks at median equivalised income after housing costs. Many below median on this measure will include a higher-rate tax payer. Two thirds of children in England are below median income on this measure. A statement that 45% of Grammar pupils are not in the top third richest (with very few in the poorest third) isn’t as powerful as you think it is and it certainly does not follow that there are many working-class kids in grammar schools (as lots of other research shows).

    It’s a shame as there’s some interesting stats in your report that merit further exploration controlling for these serious flaws in your methodology. Perhaps using UCAS data including the entire cohort of 18 years olds in an area cut on a place of residence basis would allow you to look at the issue properly?

  2. Of course it’s great that grammar schools work well for the children who manage to secure places – from whatever background they come. (And if they’re not there for the top 20%, should those children be excluded entry? Would that have any impact on the culture of those schools?) But this does nothing to overturn the fact that, for most children in the position to benefit from grammar school areas, it is a barrier that can only be attempted once, at a very early age and stage in their development. ‘Just’ getting their child in to a grammar school can require parents to have an extensive knowledge and understanding of the UK education sector, and involve tutoring and other preparation. It’s unfair and divisive and it’s not the children who do go to grammar school that we should be concerned about, but those who do not. I’m not aware that your paper addresses the proportion of non-grammar-school-attending students in grammar school areas and their access to HE compared to their comprehensive school peers.

  3. A worked example on the cross-border transfer point is instructive.

    Let’s assume that there are two LAs – Grammar-shire and Comprehensive-shire. Each has 1,000 children living there. Everyone goes to a state school. 200 children living in each authority go to a selective university and the school someone attends makes no difference to this chance. Everyone who passes the 11+ is able enough to go to a selective university. 70 children in Comprehensive-shire pass the 11+ and go to a Grammar school. 70 children in Grammar-shire fail the 11+ and go to a comprehensive.

    The impact of this is that:
    * Of the 1,000 children who go to school in Grammar-shire, 270 get into a selective university or 27%.
    * Of the 1,000 children who go to school in Comprehensive-shire, 130 get into a selective university or 13%

    This is despite only 26% of grammar school attendees being from Comprehensive-shire and only 7% of children in Comprehensive-shire attending a Grammar school.

    It’s obviously a simplified example but clearly shows the assertion that “If one takes a maximal approach and reallocates a full quarter of the selective area pupils with outcomes under examination to non-selective areas, the results remain sizeable and significant” is completely wrong – the cross-border transfers are not randomly chosen!

  4. I should have made clear on the assumptions:
    * The 11+ is assumed to be a perfect test of ability to enter selective universities – all 200 selective university entrants in Grammar-shire pass and go to the Grammar school (so 200 out of 270 Grammar students are from Grammar-shire)
    * 70 out of the 800 children from Grammar-shire who fail the 11+ cross the border to go to a comprehensive school rather than a secondary modern, none of whom are able enough to go to a selective university

  5. Emma, I would be in favour of multiple entry points to grammar schools, in particular at 13 and 16 (many already have the latter).

    Peter, I’m comparing types of school system – selective and non-selective. Colleges exist in both and no-one is suggesting both. The issue highlighted by the report shows the way non-selective schools (in general) fail to deliver for academically able children in them – unless the parents can make up the difference at home, of course.

    Equivalised household income is an entirely standard measure. It’s used to reflect the fact that a single man earning £30k is better off than a single mum with two children earning the same figure. It’s quite simply a much better measure of household wealth than crude income.

    The numbers in your hypothetical approach don’t really bear any resemblance to reality.

  6. Iain,

    Thanks for the reply.

    You are making a very partial comparison by using the DfE KS5 data and – from your response – deliberately and knowingly so. First, your methodology excludes everyone in secondary moderns (and comprehensives) who does not progress to A-levels. Second, you exclude many children educated in comprehensive schools from 11-16 who progress to highly selective universities purely because of the structure of 16-18 education in their local areas – there are, for example, 42 non-selective LAs where this excludes more than half of entrants to selective universities who were educated in comprehensive schools. I cannot believe you are defending that approach by claiming that comprehensively-educated college students who progress to HE and anyone who doesn’t progress to A-levels is irrelevant to an assessing the relative merits of a Grammar/secondary modern system 11-16 versus a comprehensive system 11-16 – you exclude two thirds of children!

    On equivalised income, my point is that *two thirds* of children are below the median income on this measure and e.g. many who are will be in households with a higher rate income tax payer. They are not from working-class backgrounds as you claim, merely not in the richest third of children.

    On the impact of cross-border transfers, the numbers are of course simplified but it shows how Grammar schools skimming off a large proportion of the most academically able children in a neighbouring local authority and lots of children going the other way to avoid secondary moderns can drive the findings you observe, even if only 25% of Grammar school students are from neighbouring LAs and only 7% of children in a neighbouring LA cross the border to attend.

    I was giving you the benefit of the doubt that these were oversights in your methodology rather than you being engaged in policy-based evidence-making but your response suggests I was wrong to do so. Disappointing!

  7. Hi Iain. Thanks for writing a reply addressing some of the criticisms. However, on the 45% below median income figure, I’m confused by why you continue to omit the fact that your source for that data (the DfE document) reports that the equivalent figure for non-selective schools is around 67%?
    Regardless of the provisional nature of those figures, I don’t see how it’s justified to exclude the latter figure, when in the report you compare 45% in grammars to the “expected” 50%. Your original data source demonstrates that is simply not a valid benchmark.

  8. Hi Iain, Have you taken account of the fact that there is much higher percentage of non-white pupils in grammar schools than in secondary moderns and than in all secondary schools generally? (‘Grammar Schools Statistics 2017’, file:///C:/Users/staff/Downloads/SN01398.pdf) There is also evidence that grammar schools admit ‘fewer pupils from the low attaining ethnic groups, Black African, Black Caribbean, Bangladeshi and Pakistani, than their local area. The gap varied somewhat by ethnic group, but was typically around half the rate in their local area in 2007.’

    Education Data Labs also reported in 2016 that ‘there are striking differences in the propensity of different ethnic groups to gain access to grammar schools. If we look at high achieving eleven-year-olds in the four fully selective local authorities of Kent, Medway, Buckinghamshire and Lincolnshire, just 29 per cent of the white British pupils who achieved a fine grade score of 5.0 on Key Stage 2 (KS2) tests goes onto a grammar school. For Asian, black and other ethnic minority groups, these figures are 56 per cent, 61 per cent and 44 per cent, respectively.’ https://ffteducationdatalab.org.uk/2016/11/ethnic-minority-groups-are-great-at-passing-the-11-plus.

Leave a Reply