Do maths schools add value?

Every maths school's marketing, and U-Maths' own network-wide figures, point to the same set of facts: high A-level grades, strong Oxbridge and Russell Group progression, Outstanding Ofsted ratings across the board. All of that is true, sourced, and worth taking seriously. It's also, on its own, an incomplete answer to the question a lot of families are actually asking: will this school make my child do better than they would have done anyway?

The selection-effects problem

Every maths school selects its intake specifically on mathematical aptitude — a GCSE grade profile, an aptitude test, an interview, sometimes all three. That's the entire point of the model. But it also means every headline outcome — the A* rates, the Oxbridge percentages, the Outstanding ratings — comes from a population that was already unusually strong at maths before they arrived. A school that takes the most mathematically able 16-year-olds in a region and sends a lot of them to Oxbridge has demonstrated that those students were capable of getting to Oxbridge. It has not, by itself, demonstrated that the school caused an outcome those students wouldn't otherwise have reached at a strong comprehensive sixth form, a grammar school, or an independent school.

This isn't a criticism specific to maths schools — it's the standard problem with evaluating any selective institution, from grammar schools to Oxbridge colleges themselves. But it's rarely said plainly in this niche, because almost everyone writing about maths schools is either the schools themselves, their university sponsors, or their own network body.

What the data on this site can and can't tell you

The DfE's own "progress score" measure, which several schools cite (Cambridge Maths School's +0.65, Exeter's +0.23 to +0.44 across recent years, Liverpool's +0.76), is the closest thing to a value-added metric in wide official use — it compares a student's actual A-level grades against a prediction based on their GCSE results, which does at least partly control for prior attainment rather than just reporting raw outcomes. That's a meaningfully better signal than a raw A* rate. It still can't fully separate "this school teaches unusually well" from "this school's students were unusually motivated and supported enough to apply, test well, interview well, and get in" — motivation and family support are correlated with the kind of prior attainment the progress-score model is trying to control for, not independent of it.

The closest thing to independent evidence

In October 2025, The Observatory for Mathematical Education at the University of Nottingham published an evaluation, commissioned by U-Maths, finding higher A-level attainment and value-added scores, and higher progression to selective STEM universities and Oxbridge, among maths-school students — with the largest apparent benefit among under-represented groups. This is the most credible evidence this sector has produced on the value-added question, because it comes from a dedicated research body rather than the network's own marketing. It comes with one important caveat, though: it was commissioned by U-Maths, the body being evaluated. That's a normal way for sector research to get funded, and it doesn't mean the findings are wrong — but "commissioned by" is not the same as "conducted fully at arm's length," and we haven't been able to review its full methodology to see exactly how it handled selection effects. Read it, weigh it, and don't treat it as the last word.

What the earlier research said

Iain Mansfield's 2019 paper for HEPI, "The Impact of Selective Secondary Education on Progression to Higher Education", examined maths schools alongside grammar schools and found both associated with higher progression to highly selective universities among disadvantaged students specifically — and recommended expanding maths schools from two to ten, a target the network has since passed. It's also worth knowing that the paper's related grammar-school findings were disputed shortly after publication by researchers at UCL's Institute of Education and the University of Bath, who argued the underlying statistical approach was flawed. We're citing that critique because a site built on being more careful than the marketing shouldn't quietly drop the part of the story that complicates a favourable finding.

Where this leaves you

• The outcomes are real. Students at these schools do get strong A-level results and go on to strong universities in large numbers.
• How much of that is the school, versus who applied and got in, is genuinely unresolved by the evidence currently available.
• The best available research (the Observatory's 2025 evaluation) leans toward the schools adding real value, particularly for disadvantaged and under-represented students — but it was commissioned by the sector it's evaluating, so treat it as a strong data point, not a settled verdict.
• If you're choosing a school for a specific child, the honest answer is that fit, subject enthusiasm and commute all matter as much as the network-wide numbers — see our guide on whether a maths school is worth it for the fuller decision-making picture.