The big question in the debate on class sizes is not whether smaller classes produce better academic results - the answer to that is, “of course they do” - a more pertinent question is, “by how much?”

In attempting to answer this question by systematically weighing up the likely benefits and costs, Dylan William, deputy director of the the Institute of Education in London has come up with interesting answers.

Reducing class sizes across the entire state sector would involve building new classrooms and increasing the teacher workforce by as much as 50 per cent, which would be expensive and might be politically difficult to justify.

Even then, you could not be assured of success. Apart from the raw material – the kids- teaching quality is the most important variable in any classroom. But Professor Williams fears that bringing in the 150,000 new teachers that would be needed may allow into the profession many who shouldn’t be there.

This is what he has to say on the matter: “When you reduce class size all you do is bring in a lot more teachers who are worse than the ones that you have already got. We would need 150,000 more teachers to reduce class sizes to 20. They will probably be the ones who didn’t get jobs at their first interviews.

“What might be a reasonable size effect will disappear with poor teachers. If you don’t do it carefully, you find that it might even reduce achievement by hiring people who shouldn’t really be teachers.”

Far better (and more cost effective) to stick with the teacher/pupil ratios we have and improve teaching methods, he argues.

On the same basis, incidentally, he also suggests scrapping nationally agreed teacher pay scales.

“Research tells us that the best teachers produce four times as much progress in students than the worst. If we are willing to pay a bad teacher £25,000 a year (which is what we do), why not – if we are serious about all of this - pay the best teachers £100,000 a year?”

But how do you tell which are the best teachers? That's a good deal more tricky. Professor William’s own research suggests that it is sometimes the least popular teachers (often the strictest ones) who produce the most progress in their pupils. But if in the process they switch pupils off learning, however, how good are they really?

Times recommends

• Students abandon the drink for gyms and juice
• Ofsted: third of Baby P-style reviews inadequate
• Top boys' school picks first woman head