We have now experienced exactly six years of New Labour’s education policies, time enough to assess their effectiveness. Indeed, the ambition of the transformation agenda and government goals for education demand that we assess rigorously what is working and what is not, so that things might be further improved.

Many of the indicators are moving positively, with major improvements in our positions on international league tables, but more modest year-on-year improvement in test scores and GCSEASA-level results. How might we move faster?

The first thing is to proceed beyond the Government’s near obsession with changes at school level. The Ofsted evaluations of Excellence in Cities, education action zones and our general experience all tell us that changing the way a school is organised is not a powerful enough educational lever to deliver great improvements in how pupils do, mostly because it is a long way from pupils’ day-to-day life in classrooms.

All the educational effectiveness literature tells us that teachers in their classrooms matter three or four times more than headteachers and the schools themselves. What we need is a focus on teaching and much information on what is best practice at classroom level, exactly the factor that made the literacy and numeracy strategies so powerful initially.

It really is absurd that probably only 5 per cent of British pupils’ time is spent in collaborative groups or in peer tutoring, given that these are the most powerful tools in the world for developing higher-order skills. It is equally absurd that British teachers have only been given perhaps a quarter of the world’s great knowledge about effective classrooms.

The second thing that would help is to rid ourselves of the notion that the way to improve the system is through school-to-school transfer of good practice, the belief that underlies specialist schools, city academies, beacons, leading-edge schools and pretty much everything else. We need to be blunt: market- based education reforms that deliver more money to the schools that have better practice, to the tune of hundreds of thousands of pounds, mitigate against the transfer of that good practice. Which school is going to give away its trade secrets without much greater monetary rewards than are on offer currently? How can you have markets and then expect schools to behave as if the market rewards don’t exist?

My own impression is that there is pitifully little transfer of good practice going on, and that even with the so-called collaborative schools, there will be precious little in future. This policy cannot work.

What is much more sensible is to focus schools not on being patronisingly dependent on some super school helping them out, but on learning from their own best practice. Every school, no matter how well it is doing overall, has practitioners relatively better than others, many schools will have excellent teachers defined in national terms, and a significant proportion of schools will have world-class people.

Helping schools to learn from their best practitioners is conceptually simple. It is practically easy - there is a much better chance of learning from someone in the next classroom than from someone 20 miles away.

Learning from your colleagues also removes any alibis for poor practice, since it is always possible to find explanations for why other schools are doing better to avoid taking any notice of them. When you are “buddied” with a colleague who is doing better than you with the same children in the same school, that is when the alibis stop. Besides, school-to-school transfer is usually done by the already good teachers, for which school is going to put on show its average teachers? Within-school transfer can find a role for the average helping those who are less than average.

It would be possible at the drop of a policy hat to focus on schools learning from their own best practice, but it is not being done. Why? Perhaps some people simply don’t understand the arguments or the data that generated them. Perhaps talking about variation between teachers takes us to the bad old days of Chris Woodhead, when teachers were seen as being “blamed”. Perhaps it is that we don’t quite know how to get people within the same school or department talking and sharing. Perhaps it is that variation between teachers is our great unmentionable, similar to how variation between schools was in the Seventies before the arrival of the school effectiveness movement.

There is everything to gain from a focus on the classroom and on every school learning from its own best staff. Asking schools to learn from their own best practice has a much better emotional tone to it than expecting them to sit by while goodies are brought to them.

It is time to let schools’ own inner variation be the engine of their own improvement.

David Reynolds is professor of education at the University of Exeter