Grammar Schools and Progression to Higher Education

Grammar Schools and Progression to Higher Education

My latest paper, published by the Higher Education Policy Institute (HEPI), looks at the impact of selective schools in ensuring progress to higher education.

The Impact of Selective Secondary Education on Progression to Higher Education, HEPI Occasional Paper 19, shows grammar schools increase the likelihood of progression for pupils from the bottom two quintiles of social disadvantage and for Black and Minority Ethnic (BME) pupils. The analysis takes into account both the chances of children from different groups getting into a grammar school and how those children who do get in perform. The report is published in HEPI’s series of polemical papers, which are designed to challenge as well as to inform.

The report shows 45% of pupils at grammar schools come from households with below median income and that:

  • 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 paper describes how most previous research has focused narrowly on eligibility for Free School Meals as a measure of disadvantage, which obscures large differences within the remaining 85% of the population.

The report also examines the performance of the new specialist Maths schools and considers public attitudes towards grammar schools, including exploring why the public are more supportive of grammar schools than educational experts.

Read the full report on the HEPI website, here.

Comments where the poster doesn’t appear to have read or engaged with at least the Executive Summary of the report will be ignored or deleted.

4 thoughts on “Grammar Schools and Progression to Higher Education

  1. Some thoughts. First, the 45% figure is unbelievably high, given what we know about the proportion of children at grammar schools with free school meals and the proportion of those at grammar schools who receive coaching in the entrance exams. Reading the comments of others, it seems this is faulty data rather than your mistake – that 70% of the population falls into the “below the median group” according to the official figures you have used. Not to throw the baby out with the bathwater – the point that grammar schools aren’t just for the rich is well-taken, even if the 45% number is not credible, as is the point that a narrow focus on free school meals is myopic. Secondly, the figures about relative performance are interesting. Obviously correlation is not causation, but the differences are stark. More widely, I’m in at least three minds about this work – on the one hand, I’m happy for you personally that you contribution influenced the national debate, and happy as I generally think selective education is a good thing. But I’m frustrated that what is essentially an op-ed piece and not a serious piece of scholarship can have such an impact on the national debate. I acknowledge that research isn’t something that should be reserved for academics – and this sort of research would be hard for an academic to do as there is so much precident in the area that would paralyze the novelty and impact of the piece (which is what we are judged on, rather than importance – clearly this is an important topic), but this is an unreviewed manuscript that frankly wouldn’t pass peer review. You have compared selective areas and non-selective areas – which is fine as far as it goes – but the unresolved question is whether this difference is due to the presence of selective schools or not, or whether is it due to some other characteristics about the areas. The map in your publication shows clearly that selective areas and non-selective areas are not similar – you are comparing apples and oranges. As I said, the magnitude of the difference is striking, and hence I’m not completely dismissive of the findings, but no attempt has been made to compare like with like here.1 Hence the frustration – it is so hard for us as researchers to catch the national attention, and that an analysis that is deeply confounded and flawed can have such impact.

    I’m not sure what the answer is – clearly it’s wrong to limit who can perform research (and there are clear political biases in play here), but on the other hand, poor quality of data and statistical methodology leads to poor inferences and poor decisions.

    1. Without diving into the complexities of particular analyses that could be performed to try to resolve this question – which would be difficult give the data you had access too – one simple step that could have been performed would be to compare the characteristics of selective and non-selective areas, to see how similar they are and what other factors could be driving the differences you have observed. A further step would be to control for these differences.

    1. You’ve made a lot of interesting points there, Steve, and I’ll try to address some of them.

      First, ‘influencing the national debate’ isn’t something that’s hugely connected to research. The things that influence the debate most are often catchy slogans – ‘Dementia tax’ ; ‘For the many, not the few’ ; ‘ Take back control’ – rather than research. There’s a whole scale of things that do, from catchy slogans, to isolated facts, to correllations, to peer reviewed research, and one of the challenges is getting information up that gradient.

      Think tank reports occupy an important point on that scale. They’re not, and they’re not pretending to be, peer reviewed research of journal quality – though many, including this one, will have had to go through some form of review process (HEPI has an advisory board which includes academics; my original submission got a ‘revise and resubmit’). At their best they should be about finding new facts, new correlations and statistics that point to issues of concerns, and connecting those to important policy issues in a way that politicians, opinion formers and the public can connect with. Often they might point to areas where more research is needed – and two of my seven recommendations are concerned with improvements to data and to more research.

      This is fairly typical of the think tank space. A good example – that has a lot of similarities – is a report last month by the Sutton Trust, Access to Advantage, that led with the fact that the 8 highest schools had the same number of acceptances to Oxbridge as the 2894 lowest schools and also highlighted regional differences. This was also almost entirely correlation, with no evidence of unfair treatment, but it’s still a fair report to publish. More broadly, we should beware of isolated demands for rigour: if viewpoints in tune with academia’s views (like the Sutton Trust) can publish such pieces and be cheered on, but for those challenging then peer review is demanded, there’s a serious problem.

      To look at some of the specific points you raised, I agree that follow up research should look with more rigour at what’s driving regional differences. As you agree, the differences are very large (especially with BME, given London doesn’t have grammar schools…). On the below median income point, the critique you’re referring to is disingenuous. The DfE Report I used acknowledges that 70% of children fall below the national median; they therefore don’t use the national median and instead use the ‘median income of households in their dataset’ – which is lower – in all of their subsequent calculations. (NB: there are obvious good reasons why the median household for state school children might be below the national median). Given this is literally set out in the paragraphs immediately below the 70% quote, it’s hard not to attribute this to deliberate bad faith by the authors of the critique.

      Overall, most of the headline data in my report weren’t particularly hard to find. Some of the effects are large enough that they deserve being looked at more closely and provide supporting evidence that grammar schools support social mobility. There are real questions as to why these were being consistently overlooked by researchers – and the report suggests some reasons for this. The purpose of a think tank report like this one isn’t to be the final word on grammar schools – it won’t be – but to drag these sort of overlooked facts into the light of day, profile them and to try to get people thinking more deeply about them.

      1. About influencing national debate, this is literally something that we are judged on (via REF Impact). It’s a different type of impact to catchy slogans, but it’s certainly something that is part of the wider impact of academia.

        About the data for the 45% claim, while I’m highly sceptical about that number, I don’t think it matters a lot what the precise number is.

        If you dropped the causal language from your report and simply reported on associations, then there’s a lot less to criticize. It’d be better if you were able to control for the some of the obvious differences between selective and non-selective areas to get a better idea of how much of this difference is driven by selective schools, but I understand that there are limitations in data availability. As per the above comment, the associations are still striking and require action, even if a substantial proportion of the associations are driven by other sources of inequality rather than the presence of selective schools.

        1. You’re right about the first. I hadn’t meant to imply that research didn’t aim to influence national debate, simply that as so many things influence the national debate, we shouldn’t expect to see a strong correlation between ‘depth of research’ and ‘influence on the national debate’.

          On the second, you’re effectively saying ‘If your paper was less like a think-tank report, it would be more like a research paper.’ I’ll return to the fact that these are too different media and it’s not the case that one is strictly better than the other: I’m sure it’s true that this and almost all.other think tank papers wouldn’t get into peer reviewed journals, but equally I’ve never seen a peer reviewed academic paper that would be remotely suitable as a think tank report.

Leave a Reply

Your email address will not be published. Required fields are marked *