At my village primary school more than 50 years ago (which was too small to form more than one class per year group), children preparing for the 11-plus sat in four rows. Though nothing was made explicit, we all knew what the rows signified. Those on the first row were certain to pass; the next row had a sporting chance; the third row were no-hopers, but were willing to go obediently through the motions; and the final row didn’t even bother trying.

The four rows probably matched, with reasonable accuracy, the country’s social structure at the time. The first largely comprised the children of what were then called white-collar workers; the fathers of those on the other side of the classroom tended to be unskilled labourers. That at least is how I remember it and I recall trying to explain to a no-hoper the alien (to him) concept of owner-occupation. Certainly, as the class teacher’s eyes roamed across the rows, he would have seen distinct differences in cleanliness, nutrition and clothing.

I am unsurprised, therefore, to learn that, as streaming and setting return, academics find (as The TES reported last month) social divisions between the streams. Views on grouping children by ability go in roughly 30-year cycles. From the beginning of the century until the 1930s, streaming was unusual. Then that old rogue Cyril Burt claimed to have proved that ability was largely innate. An official report of 1938 insisted that “it is possible at a very early age to predict... the ultimate level of a child’s intellectual powers”.

The trend away from streaming, particularly in primary schools, began in the 1960s, as Burt was discredited, the 11-plus declined and the pressures on teachers to coach their highest performers diminished.

Now, since the late 1990s, ability grouping, with government support, is back. A study of Hampshire primaries found three-quarters setting pupils for at least one subject and a third of those streaming classes across all subjects. Schools have responded to the need to coach children for tests at age 11.

The late Sir Karl Popper, one of the most influential political philosophers of the last century, thought public policy proposals should be formulated and tested as scientific theories. By trial and error, knowledge of how to run public institutions such as schools would then grow rather as scientific knowledge grew. This, he argued, would ensure social progress.

Unfortunately, Sir Karl was writing before educational and social research really got going. There have now been ample opportunities to test theories about ability grouping. The overwhelming conclusion is that streaming, setting or mixed ability make no difference either way. There is evidence that lower-achieving children make less progress under streaming while the high achievers make more. But it is not conclusive. There is also evidence that, at the primary stage, children in lower streams develop more negative attitudes to school than they do if placed in mixed ability classes. But the effect seems to disappear by the time the children start secondary school.

So, far from making progress, as Sir Karl hoped, we go round in circles. It is as though the theory that the Sun orbits the Earth had just come back in, after a period when we believed the opposite. I do not blame researchers for this. I blame Sir Karl, who was sadly mistaken about the possibilities of applying scientific method to public policies.

The truth is that our views on streaming (and on many other aspects of schooling) are determined almost entirely by our political prejudices. I accept that mixed ability classes won’t raise standards - though I don’t think they will depress standards either - and that children from poorer homes will still tend to do badly in school. But I abhor anything that accentuates social divisions, and on those grounds I shall continue to oppose streaming.