(Contact for press enquiries: 07870 682442)
Today David Cameron is set to announce an expansion of the free schools programme. According to media reports
he will justify this on the basis that “free schools do not just help the performance of their pupils, but pupils in surrounding schools”.
This is a direct quote from Policy Exchange report, “A Rising Tide
”, (released today) which analyses the impact of free schools on the existing schools that are close to them. But the data in the report actually shows very mixed results and does not provide the basis for this quote or for Cameron’s overall claim.
Do the primaries closest to primary free schools show a faster increase? No
The Policy Exchange report is clear on this: “Schools closest to Free Schools perform in line with national results at primary.” There is no extra increase for schools close to free schools.
An analysis comparing the closest schools with the national average, based on the increase in KS2 results since the year the free school opened, shows virtually no difference. Indeed the only opening year for which the results were different shows slightly worse performance by those closest to a free school.
Do the secondaries closest to secondary free schools show a faster increase? No
The report claims that schools closest to free schools perform “better than … national average at secondary”. However this claim is not supported by the data. If we discount those opening in 2011, where Policy Exchange accepts the sample is too small (just five schools), then there is virtually no difference between the change in results in schools close to free schools and those nationally. For 2012 and 2013 openers, the fall nin GCSE results is exactly the same and for 2014 openers there is just a 1% difference.
The argument for free schools is that they introduce competition that will improve the performance of all schools in the area. This is the basis of Cameron’s argument and is the argument tested in the Policy Exchange document. However the data clearly does not support this. Overall the change in results of schools closest to a free school are remarkably similar to those nationally.
Other factors influence the schools involved. In particular 51% of primary free schools and 29% of secondary free schools in the study are in London, according to NUT figures. Given that London schools have, in recent years, improved at a faster rate than those nationally then schools close to free schools would be expected to benefit from this London effect, but do not appear to.
Note: The data above is taken directly from the Policy Exchange report. Copies of the relevant tables are below in the Appendix.
Analysis by prior school performance: No clear causation
Policy Exchange goes on to analyse the effect on nearby schools, according to whether those schools were previously high performing. For primaries they find that in quartile 4 (the lowest 25% of performers), the closest schools increase at a faster rate than similar schools nationally for three of the four years. However the closest schools do less well than similar ones nationally for the other three quartiles..
If this data is used to argue that the presence of primary schools improve the performance of the least well performing schools, it must also be argued that they reduce the performance of the other 75%, which does not seem an impressive result. This analysis is based on relatively small cohorts of data, just 20 or so schools in each of these year cohorts (out of 16,000 primaries in England). It may be that the presence of a free school is not the key factor. The report itself accepts this:
"It should be obvious – but bears setting out explicitly – that such data cannot demonstrate conclusively that any changes seen are as a response to the new Free School. A school appointing a new Head; a change to Academy status; a glut of teachers leaving; a financial crisis – all of these can affect an individual school for better or worse. It should also be remembered that sample sizes in some of these categories is quite small, and correlation should not be mistaken for causation."
For secondaries the data appears to show that less well performing schools close to free schools improve more than similar schools nationally, this time for quartiles 3 and 4. However again the reverse is true for the other 50% of schools. Again the cohorts are small, 20 or so schools in each cohort. And again the schools may have been affected by other factors, such as the London effect.
Do free schools help nearby undersubscribed schools? No
The Policy Exchange report states in its summary: “Primary schools with surplus places show a bigger increase in results than schools which are oversubscribed in every year apart from those approved in 2013.”
Read that sentence carefully. It is not claiming that surplus place schools improve at a faster rate if they are close to free schools. It is simply stating that surplus place schools that are close to free schools improve more than other schools that are close to free schools. It says nothing about any benefit due to free schools.
Indeed if we compare both types (surplus places and not) to national figures, we find neither type of school outperforms:
The data is clear: Schools with surplus places, that are close to free schools, increase at a similar rate to national levels. There is no apparent benefit from competition from a free school.
For schools without surplus places (termed “basic need” schools in the report), they performed better than schools nationally for the 2014 cohort but worse for the 2011, 2012 and 2013 cohorts – and significantly worse for the 2011 and 2012 schools..
The question of why schools with surplus places may improve at a faster rate is an interesting one. It could be due to the pressure to attract pupils, or due to the benefits of smaller classes. However there is no evidence that the presence of a free school leads to schools with surplus places improving at a faster rate than those nationally.
The coalition has been criticised for opening free schools where there are already surplus places. However this has been more on the basis of the waste of money involved than that it will reduce results in nearby schools. This analysis reveals that the introduction of free schools in areas with surplus places brings no extra benefit to those schools and so still seems to be a waste of valuable resources.
Note: This analysis is purely for primaries. Policy Exchange themselves accept the data for secondaries is “inconsistent”.
“At least they don’t do any harm”
This was the question put to me in a Sky News interview, based on the report’s claim that free schools “do not drag down results of neighbouring schools”. My response was that it comes to something when the main claim for a major government policy is that “at least it hasn’t done any harm”.
In fact the data in the report does not support this claim either. The data indicates that, for primaries, schools close to free schools, that have surplus places, perform in line with national levels. However primary schools close to free schools, that are fully subscribed, saw – in three of the four years – their results improve at a slower level than national. If, therefore, the causation is as described in the report then free schools have done harm to nearby fully subscribed schools.
To be honest, I’m not sure that is the case. I think free schools have probably had little effect on neighbouring schools. Any results above or below the national average, in schools close to free schools, are likely to be down to other factors.
So have they caused harm? Well the free schools that have been closed, those that have had their sponsors removed due to poor performance and where the head has been removed due to financial mismanagement – those have all caused harm and indicate a need for greater oversight.
But the greatest harm is in the use of resources that could have been used elsewhere. There has been a £1.7 billion capital spend on free schools and, according to the National Audit Office, 52% of them have opened in areas with little or no need for new places. Cameron will claim they have benefit local under-subscribed schools but the data is clear – they have not, as those nearby schools with surplus places have seen their results increase in line with national figures.
Do free schools and academies have better Ofsted verdicts? No
The government has also made claims that free schools and academies had far better Ofsted verdicts. The Policy Exchange addresses this for free schools but provides no basis to Cameron’s claim:
“Comparing Ofsted grades of Free Schools to all schools over a similar time period does not produce a reliable judgement of quality either for the benefits or weaknesses of free schools.” (P15)
For academies it is also hard to judge. Converter academies have Ofsted verdicts well above the average, but this is understandable given that the schools that converted were primarily already Good or Outstanding. Sponsored academies have Ofsted verdicts well below the average, but again this can be explained by the fact that it was mainly weaker schools that became sponsored academies.
Summary: The creation of free schools does not "help pupils in surrounding schools"
The data in the report is clear. For both primaries and secondaries, there is no benefit to nearby schools from the creation of a free school. Free schools are still an experiment. The first exam results will not be available until 2016. There is no evidence, in this report or elsewhere, of educational benefit from the considerable resources devoted to them.
Appendix: Policy Exchange Data
All of the data in this post is taken directly from Policy Exchange’s report, “A Rising Tide”, and specifically from tables 2,1, 2,2, 2.3, 2.5, 2.6, 2.7 and 2.8: