Have free schools failed in Sweden?
Michael Gove's controversial scheme for the expansion of the academies programme was modelled partly on the Swedish system of "free schools".
"We have seen the future in Sweden and it works," said the Education Secretary in 2008.
Recently, however, Mr Gove's Labour counterpart Ed Balls has repeatedly cited the Swedish example as evidence against the effectiveness of free schools.
In his Conference speech yesterday, Mr Balls called the decision to remove some schools from control by their local authority "a policy which we know from Sweden delivered lower standards and higher inequality."
So what can Sweden really teach us? Full Fact attempts to offer some independent education.
The Swedish reforms, enacted in 1992, allowed independent schools to receive public funding by way of a voucher scheme.
The first part of Mr Balls' claim - that free schools deliver "lower standards" - appears to be supported by the most recent Trends in Mathematics and Science Study (TIMSS) in 2007.
In standardised tests, Swedish pupils performed significantly worse than English pupils in both maths and science.
Moreover, comparative results from Sweden had worsened dramatically since 1995, while England's maths results had improved more than any other country included in the survey.
Conclusions from this should be taken with several healthy caveats: there is no way of knowing exactly why scores changed, and comparative testing may give misleading impressions. Still, it does lend some support to Mr Balls' assertion.
Research specifically directed at judging the impact of free schools gives a more ambiguous picture.
Evaluating various studies assessing their academic impact, a report by Bristol University's Centre for Market and Public Organisation (CMPO), Replicating Swedish 'Free School' Reforms in England, concluded that "the best evidence on Sweden's reforms indicates small improvements in academic achievement in areas with more free schools."
The report shows that while one influential paper apparently showed the schools to be a spectacular success, it was based on very early evidence and was criticised for its methodology.
Two more recent and robust studies (Bohlmark and Lindal, Bjorklund et al) agreed that the schools have had a positive academic impact "in both the private schools and — through competition — in municipality schools."
But they also suggest that benefits so far are small - "too small", in fact, "to persist into long-term gains for young people."
Nevertheless, the CMPO study disagrees with Mr Balls on standards.
It does, however, lend a little credence to his second claim that Swedish free schools exacerbate inequality: "the biggest beneficiaries are children from highly educated families; the impact on low educated families and immigrants is close to zero."
But the most common criticism on the grounds of inequality is that parents more active in their children's education, often better educated and wealthier, are more likely to succeed in getting their children into competitive private schools.
In February, the director of the Swedish National Agency for Education blamed the free schools for increasing social stratification, as well as denying that they improved standards.
But Full Fact has been unable to find data on the comparative social make-up of free and state schools, and there is not a consensus on whether disadvantaged children suffer.
The OECD's report on equity in Swedish education was released in 2005, before some of the most important research was released.
Its tentative conclusion is that "reforms appear to have contributed to more segregation and greater inequalities in initial education".
But it also concedes that many from minorities and disadvantaged backgrounds may have gained from economic liberalisation.
For instance, ethnic minorities are over-represented at private schools, which on average perform better.
This is a double-edged point: it might imply benefits for an often disadvantaged sector of society, but it could also suggest increasing segregation.
An article such as this could continue indefinitely, burrowing ever deeper into buts, howevers and endless qualifications.
The weight of evidence seems to be against Ed Balls as far as standards are concerned, although there are valid grounds for doubting the effectiveness of free schools.
Concerning inequality Mr Balls is on slightly firmer ground. He has support from an Education Minister of the centre-right party who passed the reforms, as well as the Swedish equivalent of Ofsted.
But support from experts and authorities is not the same as evidence and, frustrating though it is, Full Fact must once again adjudge the debate ongoing.
Even if the evidence was more conclusive, the Bristol University report points out that the usefulness of Sweden for comparison is compromised by the fact that free schools were introduced alongside other reforms deregulating teachers' pay and the teaching curriculum.
So while Ed Balls can justifiably refer to some evidence suggesting that free schools bring "lower standards and higher inequality", it is far too categorical to claim that we "know" either of these things from the example of Sweden.
As the debate on free schools continues, the Swedish precedent will no doubt continue to be drawn on for both support and opposition. But both sides should remember that as valuable as Sweden is as a source of evidence, this evidence does not definitively settle the case in either direction.
Edgar Gerrard Hughes
Full Fact is looking to improve how we operate. Please email your thoughts on both content and design to email@example.com