David Cameron is due to announce a rapid expansion of free schools today. By a strange coincidence, a new report by ‘independent think tank’ Policy Exchange is also timed for release today.

Our analysis provides 10 reasons why this ‘research’ should have no place in shaping government policy. Our conclusion:

“This is a seriously flawed and misleading report, scarcely recognizable as ‘research’ and it is irresponsible for a Prime Minister to base future policy on such scant and biased evidence.” 

 ‘A rising tide: the competitive benefits of free schools’

(N Porter and J Simons, for Policy Exchange, March 2015)

(available online here)

1)  This is not in any sense independent research. It was carefully timed to coincide with the Prime Minister’s announcement about a further expansion of the free schools programme. It was conducted in-house by Policy Exchange, an organisation established by Michael Gove and other leading Conservative Party members, and which describes itself as seeking “free market” solutions to public policy problems (i.e. privatisation). Although a registered charity, it does not release the names of its funders. Moreover, although its financial statement describes most of its spending as ‘research’, much of this appears to be devising and advocating Conservative Party policy. Indeed, its 2012-13 accounts boast of its influence on government policy including increasing chain-sponsored academies, the mandatory work scheme for unemployed people, and Prevent. (This report on free schools was funded particularly by Krishna Rao and Jeremy Isaacs of Goldmann Sachs, who also funded an earlier report on teachers’ performance pay.)

2) The authors have very limited experience of undertaking significant scale research projects, although one has experience of writing and advocating Conservative Party policy. Policy Exchange claims to be ‘independent’ but the report’s authors are by no means neutral: one was involved in founding an academy, and the other a (not very successful) free school.

3) Impact on neighbouring schools. The bulk of the report is devoted to backing up a claim that free schools impact positively on nearby schools, ostensibly through competition and spreading good practice.The statistical evidence provided fails to evidence this (see point 4) and the explanation of how this is brought about is implausible (see point 5).

4) Statistical evidence. Tables 2.1 and 2.2 show that the only set of neighbouring schools which progressed more than the local or national average were neighbours of secondary free schools opened in 2011. There were only 5 of these!

However, because lower performing schools have more scope to improve, the analysis is then conducted in quartiles. This produces 16 different comparisons: the schools located near to free schools do worse than nationally in exactly half of these. (There is a tendency for neighbours in the lower performing quartiles to be better rather than worse.) [For more detail, see Henry Stewart’s analysis at http://bit.ly/PXNoEffect]

5) Implausible causal explanation. Although the report cannot explain how this could occur, it attributes it to competition: ‘Competition from free schools does seem to be driving a response… [particularly] amongst the lowest performing schools’. This fails to consider all the other pressures impacting on low performing schools (Ofsted repeated visits, floor targets etc.) besides which the establishment of a free school seems of little consequence. Even more implausible, since most of the free schools are recent foundations, quite small and only have younger children, how could a small number of 11 and 12 year olds improve the GCSEs of large numbers of 16 year olds at nearby schools?

6) No direct output advantages. The main reason for taking this indirect approach is that there is scant evidence of free schools having better outcomes. The report acknowledges that Ofsted reports on free schools show no advantage (a third are ‘needs improving’ or ‘special measures’). Out of the 69 (non-special, non-alternative provision) free secondary or all-through schools considered in this report, only 8 have GCSE results. These are either small numbers, or (in the case of conversions to free school status) pupils who had most of their education at prestious or elite schools. Altogether just 464 pupils, or less than 1 in 1000 of England’s GCSE candidates – a narrow basis on which to launch a rapid expansion of these schools. The quality of attainment is extremely variable. (link here to free school GCSE results).

7) Other claims that free schools benefit their neighbours. On p9, we read “This report believes [sic!] that high performing Free Schools can offer benefits… through demonstration of good practice, innovation, and competitive effects”. However, it provides no evidence of such good practice and innovation, but merely cites a flawed DfE report (Cirin R, 2014, p5) which relies entirely on the beliefs and assertions of free school heads. Indeed, the innovation they point to is largely in terms of changes to the school day and terms / holidays (which can occur in local authority schools). In terms of curriculum, it is their freedom not to teach some subjects (examples given history, music, ICT, MFL) and in a few cases various subjects which neighbouring schools don’t provide (but many do!) such as Mandarin, classics, environment and engineering. They also claim to have ‘extension programmes’ by which they mean lunchtime and after-school clubs.

8) Providing school places in areas of shortage. The report makes the absurd claim that free schools have taken pressure off local authorities in areas where there is a shortage of places (“they have contributed to alleviating basic need in many Local Authorities”) – but this funding is taxpayers money, and local authorities could have provided these places more efficiently if the money had gone to them.

9) Parent power. The original justification for free schools was expressed in terms of enabling parents and community groups to establish their own schools. This report acknowledges (albeit confusingly expressed) the increasing tendency of these schools to be opened by academies and academy chains, but its graph is designed to minimise and conceal the extent of this. (Page 8, Figures ES1 merges waves 5-7 into one bar, and provides no numbers, thus disguising the higher numbers in more recent waves.)

10) Benefits to disadvantaged communities. The report acknowledges that bids which are successful tend to come from ‘more advantaged communities and professional backgrounds’ who were ‘not groups who were as willing to operate in disadvantaged communities as other groups who were less successful’ (p15, based on R Higham, 2013). Whilst free schools were more likely to open in more deprived neighbourhoods, they tended to take better off children with higher prior attainment (though also a higher proportion of non-white children). (p15 from Green, Allen and Jenkins 2014)

In conclusion, this is a seriously flawed and misleading report, scarcely recognizable as ‘research’ and it is irresponsible for a Prime Minister to base future policy on such scant and biased evidence.