As a response to my arguing that Muir cites evidence selectively, Muir claims I do the same thing. This is a strange accusation. In a short comment, while noting the overall conclusion based on all research, I can naturally only refer to research that alters the picture compared to the report’s review. But to show that I take intellectual honesty seriously, I give Muir a freebie and point to another study he missed. It finds a small negative effect of Swedish for-profit schools – which I also noted in a speech at Policy Exchange earlier this year. Readers might also be interested in my takes on it.
Muir also notes Elacqua’s study as evidence of selective reading. Why? Because Elacqua suggests that the positive impact of for-profit chain schools in Chile perhaps is due to their benefiting from scale. But this is partly my argument. If for-profit schools scale up and perform better than other school types, this indicates something about the profit motive. In Sweden, for example, ten large for-profit companies run 30% of all Swedish compulsory free schools – whereas non-profit schools remain small and local. Scale, therefore, is clearly important in the debate on profits in education.
Furthermore, Muir does not mention other research of relevance here: in Chile, only for-profit schools enrolling poor pupils increase in size as a result of improving academic results. This indicates that quality induces for-profit schools to scale up among pupils who most need them. Neither does Muir mention Böhlmark and Lindahl’s new study – which finds no differences between the effects of Swedish for-profit and non-profit school competition on overall achievement– nor the results in my paper’s preferred model. After evaluating all studies, and taking into account that some methodologies are preferred, the picture is not that mixed: there are basically no general quality differences between for-profit and non-profit schools.
And my argument has never been that for-profit schools per se are better than non-profit schools. Rather, the point is that competition can improve achievement, and that for-profit schools produce a supply-side dynamic we cannot achieve without them. Muir claims that ‘Sahlgren quotes some further studies which he claims support his case [for competition]’. This is an understatement. I cite a review article covering all PISA/TIMSS research in which a consistent finding is that private-school competition raises international test scores. This research is not mixed. Although most studies use relatively unsophisticated methodologies, I highlight a paper using a more careful research design. It finds that private/independent-school competition increases PISA scores –among pupils in state and private/independent schools – as well as decreases costs. Since international tests are important in today’s education debate, and since cross-national studies can capture system-level effects better than most within-country analyses, these results are conspicuous.
Instead, Muir repeatedly writes that the official PISA report does not find that competition improves achievement. But all research is not good research– and one has to understand what makes the difference. The PISA report is not an academic paper, and there is no reason why we should discard all proper studies for a report that merely evaluates cross-national correlations. It proves nothing.
It is true that country-specific research on competition and achievement is mixed. Firstly, it is important to note what ‘mixed’ means in this context, namely no or positive results. There are extremely few methodologically sound studies finding negative achievement effects of school competition. Note that even if a policy produces null effects on one metric, it could still maximise welfare. A null impact, as opposed to serious negative effects, is therefore not sufficient to oppose a policy.
Secondly, Muir largely ignores my argument that one has to understand what research analyses and how – while insinuating that I prefer confirmation bias if it favours my argument. This is not true. My upcoming 50,000-word paper evaluates basically all research on school choice and educational achievement, irrespective of results, while also discussing methodologies and analysing system design in different contexts. It also makes the case for comprehensive reform in a way that maximises evaluation opportunities – as I have argued previously, for-profit schools are not enough. Instead, they must be complemented by a coherent reform package that transforms the incentive structure in education.
Overall, the evidence suggests the following: (1) for-profit and non-profit schools perform on a par; (2) for-profit schools scale up more often for the right reason; (3) private-school competition generates better school productivity in international comparison tests; (4) a significant part of within-country research indicates that school choice and competition, properly designed, can improve outcomes; and (5) extremely few studies find negative effects. This does not mean that we should automatically discard (6) research that displays null effects. But we must carefully evaluate systems and studies to understand why there are consistently positive findings in some countries/regions – and on international tests – but more mixed results elsewhere.
Overall, therefore, the case for an outright ban on for-profit schools is simply not grounded in research. My claim against Muir’s ‘normative argument’ was hardly ideological, as he contends, but rather based on my belief that any argument should be backed up by evidence. A key feature of policymaking in my home country has been its reliance on social scientific research – it is time that UK education policy makers, and wonks, also adopt this approach.
Gabriel H. Sahlgren is Research Fellow at the Institute of Economic Affairs. Follow him on Twitter.