Their real importance lies in their role in the quasi-market system in our schools - a system created by the Thatcher government in the late 1980s.

Restraints on parents’ choice of schools were swept away (though, importantly, the limits imposed by a school’s physical capacity remained), heads and governors gained control of budgets, and funding followed the pupil.

In other words, a school’s success, including the jobs of its staff, was linked directly to its capacity to attract parental custom. As the social policy professor Julian Le Grand has put it, this marked a switch from the century-old view that teachers were “knights” who could be trusted to behave altruistically in children’s interests to the view that they were “knaves”, who needed the sticks and carrots of the market to perform efficiently.

But a market, if it is to provide incentives, requires information.

Otherwise, customers have no basis on which to choose. If you are buying a car, you will rely on three sorts of information. First, the price. Second, straightforward descriptive information about colour, size, engine capacity, and so on. Third, evaluative information about how often the car breaks down and how quickly it starts to rust away.

In state schools, the first point is irrelevant. The second - for example, size of school, location, curriculum, uniform, religious affiliation if any - has always been available. But this is not a rich source of information for most parents. After all national regulations now require all schools to teach the same curriculum.

So all the weight is put on the third category of information: exam and test results. And it’s a good thing too, the advocates of markets would say. Cars (like washing machines and televisions) rarely break down nowadays because customers have more comprehensive information about performance, and manufacturers cannot get away withproducing shoddy products. The availability of exam and test results will do the same for schools.

The argument is flawed. The best way of producing a better-performing car is to improve technical input and workmanship. The best way of producing a better-performing school is to recruit smarter pupils, preferably from middle-class homes.

You can also massage your school’s results through lax supervision of the tests or concentrating efforts on children at the crucial grade borderlines. Teachers, having been told they are knaves, have begun to behave knavishly.

So the problem is not the tests themselves - I favour tests in principle because they help to set priorities - but their role in the quasi-market.

If we must have a market, I would strengthen that second category of information, so that parents have a proper choice between different types of school just as they have a choice between different types of car. That is why I favour specialist schools - for languages, creative arts or sciences, for example.

At present, middle-class parents mostly choose schools on the basis of test results, working-class parents mostly on location. Specialist schools add a new dimensionof choice.

Where practicable, I would ask all parents to express a preference for one type or another (including a generalist option) and, where a school was over-subscribed, the matter should be settled by lottery.

Tests could then perform their proper role of enabling the Government to check that schools are doing their jobs properly.

That is what the NUT should be fighting for. Alas, it has set its face firmly against specialist schools.

Peter Wilby is editor of the New Statesman