For decades, the US government has given grants for what are now called Comprehensive School Reform (CSR) demonstration programmes. Using the grant, the recipient school implements one of a wide range of teaching and curriculum models. In 2003, a comprehensive meta-analysis of the 29 most widely implemented models was published. The average effect size - the difference between performance of students in the experimental group and students in the control group, weighted by the standard deviation - was +0.15. It is an effect even smaller than the one observed in the studies of pre-school interventions cited earlier. An effect size of +0.15 means, for example, that a child at the 25^th percentile goes to the 30^th percentile.

To put it another way, no model of school reform that has been tried in the United States has demonstrated, with evidence that withstands scientific scrutiny, the kind of dramatic impact on academic performance that the educational romantics say is possible. I have been unable to find rigorous examinations of experimental learning models in England. I hope this article will unearth empirical evidence of dramatic success that I might have missed - which I look forward to examining with a fine?tooth comb.

American advocates of school privatisation (I am one) fare no better than the advocates of government-sponsored reforms. Vouchers and tax credit experiments have been successful in many ways - providing safer and more nurturing school environments, freeing able students from peer pressure not to perform and installing more substantial curricula. But the reading and maths scores of students in these schools compare favourably with students in government schools only when the comparison is with the worst inner-city schools. When the comparison is with ordinary government schools, the differences have been trivial.

Does the English experience of the past decade refute me? When it came to power in 1997, the Labour Government set out to achieve ambitious increases in test scores, and it succeeded. English Key Stage 2, Key Stage 3 and GCSE scores all bear witness to this achievement.