There aren’t many safe bets in politics these days, but I’d be willing to wager that expansion of free schools will be high on the list of education pledges in the Conservative Party manifesto, if and when it comes to a general election.

The free schools programme is now being placed front and centre of the government’s narrative on improving school performance and tackling disadvantage. While the scheme that was intended to introduce innovation, choice and higher standards took a back seat under the last prime minister, Boris Johnson has wasted no time at all throwing his weight behind the programme; this week’s Queen’s Speech confirmed a new wave of more than 200 further free schools.

Published yesterday are provisional secondary school performance tables for 2019, with some free schools getting their first set of GCSE results. I’d be equally willing to predict that free schools will continue to be the highest performing school group, not because there is strong evidence for their ability to deliver high standards, but because the evidence shows that for many free schools the odds are stacked in their favour to begin with.

It’s another one of those safe bets.

New analysis published today by the Education Policy Institute sets out why this is the case. Our research finds that the strong performance of many secondary free schools can at least partly be explained by the characteristics of the pupils who attend them, and because the free school programme has failed to target areas of the country where education outcomes are poorest.

This last point poses a particularly difficult question for the programme, given that one of its initial purposes was to improve access to good quality school places.

It appears that the Department for Education has already drawn similar conclusions. The new round of applications to set up free schools now gives far tighter criteria than before: proposers of free schools must now demonstrate both a need for more school capacity and a need for higher-quality education in the local area, as opposed to the more lenient either/or of earlier waves.

The new round also encourages applicants to set up free schools in areas that have yet to be reached by the scheme. It may prove a tall order to attract proposals meeting all these criteria, given the paucity of free school applications to date on the grounds of local school standards.

Regardless of these recruitment challenges, policymakers cannot simply expect the same high GCSE results as free schools expand into new areas, particularly if they start targeting the low performing areas they were initially designed to address.

This is because the strong performance seen from some free schools is not likely to result simply from things that can be easily replicated across the county, like specific approaches to curriculum, teaching or behaviour.

Rather, free schools’ high performance is in fact likely linked to their local intakes: over half of the pupils in the top free schools are those who, despite being from economically disadvantaged backgrounds, still achieve higher grades in their GCSEs compared to other, similarly poorer pupils. These are often disadvantaged pupils from urban and ethnic minority communities, who typically perform well.

This context needs to be considered when interpreting the new provisional GCSE results published today. The latest results are likely already being used to advance the case for free school expansion, but it is clear that the areas served by successful free schools are markedly different to those ‘left behind’ areas that the scheme was designed to address, but has so far failed to reach.

Given the evidence that the top performing free schools are likely to be front-loaded for success by the communities they serve, the government should be wary of pushing ahead with trying to replicate its good news story in those disadvantaged areas with poor education outcomes. If the aim in education is to now “level-up” opportunity, then more careful consideration is needed of how to tackle the challenges faced by these parts of the country.

Bobbie Mills is a senior researcher at the Education Policy Institute