For almost as long as there has been public education in this country there have been public exams. And for almost as long as there have been public exams, there has been fierce criticism about the way they have been used as a lever for accountability.

Around 100 years ago, the venerable schools inspector Edmund Holmes bemoaned the limitations of seeing good results as an end in themselves.

And now the present chief inspector, David Bell, has begun to sound a cautionary note about the downside of current testing arrangements, with their targets and league tables, and their effect on pupils.

It’s hardly surprising that any notion of assessment has become entwined with summative exams. But they are now starting to be unravelled by assessment for learning (see panel below).

Assessment for learning is not a challenge to the need for summative assessment, but it looks forwards instead of backwards. It asks not what a pupil has done so far, but, on the evidence of that, what he or she needs to do to improve.

That is why when standards minister David Miliband announced the Government’s new computerised system, which allows schools to track students against their own prior achievements and that of peers in their own and other schools, and called the innovation assessment for learning, he was wrong.

Assessment for learning is not just elaborate data collection. It is about engineering opportunities in lessons, through classroom interaction and the tasks with which pupils engage, to find out what students have learned and helping them understand how to progress.

We at King’s College wanted to understand what it looked like in the average British classroom. Armed with money from the Nuffield Research Foundation, and with evidence from a review of research into the benefits of formative assessment by Paul Black and Dylan Wiliam (published in Inside the Black Box), we began work in six schools in Oxfordshire and Medway.

Data from exam results showed a significant rise in pupil achievement.

Possibly the chief finding, apart from the marked improvement in pupil performance, was that the participating teachers began to shift their practice. They moved from what might be called a curriculum coverage model to one that emphasised knowledge underpinned by understanding. In turn, their pupils became more reflective about their own learning.

It was not hard to see why a formative classroom would make pupils more reflective. Key strategies include the use of questions and rich tasks, peer and self-assessment, feedback through marking and the formative use of summative tests.

Take questioning, for example: it can simply be an invigorating rapid-fire exchange of recalled facts from previous lessons, or a means of probing pupils’ understanding and extending their thinking. But each requires a completely different type of question.

For if the questions we ask are designed to elicit evidence of whether or not pupils have understood what they have been taught, with a view to doing something about it if they haven’t, they will inevitably be more conceptual than factual. This is because it is easier to hide our lack of understanding behind a well-remembered fact than it is if the question requires some kind of conceptual grasp to answer it.

One teacher, for example, changed from starting a lesson by getting the pupils to recall the word “photosynthesis” to asking them why one plant growing by a window looked healthier than another growing in the shade. An extended dialogue about the process of photosynthesis ensued, with all members of the class involved, rather than a rapid exchange with those who got their hands up the quickest.

Inevitably, perhaps, teachers’ planning altered. Instead of trying to cram every requirement of the exam syllabus into the timetable, they spent their time ensuring everyone had grasped key concepts. In doing so, other practices altered. End-of-module tests, for example, were moved back a few weeks so that teachers could use the remaining lessons to go over those aspects of the course that pupils had not understood.

Pupils need to be involved in formative assessment for the process to be truly successful. This is where peer and self-assessment became important.

I observed many lessons in which pupils of all abilities commented on each other’s work supportively and with considerable insight.

In some cases the teacher had discussed the criteria with pupils; in others the pupils had devised the criteria themselves. But the process of peer assessment gave all the pupils a vocabulary for describing and discussing what a successful piece of work might look like and what they needed to do to achieve it.

Assessment stopped being something that was done to them, and became an aid to their learning.

Assessment for Learning by Paul Black, Chris Harrison, Clare Lee, Bethan Marshall and Dylan Wiliam has just been published by Open University Press, pound;15.99 paperback; mcgraw-hill.co.ukopenup