At a cost that will eventually climb to $14million per year, it is promising to cut the size of every government primary school class to 21. All that's missing is the money and the evidence.

Put another way, what's missing is the evidence that cutting class sizes would be value for money.

The concept of ''value for money'' is hard to grasp when you are an anxious parent like me being offered a gift during an election campaign.

It only makes sense if you compare it to the value that could be obtained if the $14million per year was spent another way.

It could be spent on salary bonuses for teachers who are really good. An extra $10,000 or an extra $20,000 per year might keep and attract the teachers we really need as well as fostering a culture of excellence.

The Liberals say they've got evidence that lower class sizes would improve educational outcomes. But it is so weak as to be embarrassing.

Quoted with approval in their policy document is a study undertaken for the NSW Government that asked parents, principals and teachers ''whether they thought the class size reduction had had an impact upon student attainment in literacy and numeracy''.

Guess what? They did. But there was no objective measure in the study of whether those outcomes had improved and the answers received were a bit like those you would expect to a question phrased along the lines of, ''We have just spent more money on you has it helped?''

Lower class sizes often seem to help. We see them in special classes for poorly performing students and for gifted children. But that doesn't mean they help generally.

Project Star in the United States was the biggest attempt to answer the question. It cut class sizes in a randomly selected group of Tennessee schools and then compared the results of their students to the results of those in schools whose class-sizes hadn't been cut. The results were impressive, but tainted.

The teachers knew about the trial and knew that if it succeeded class sizes would be cut state-wide.

Andrew Leigh of the Australian National University points to a better, more recent study by Harvard University's Professor Caroline Hoxby, who compared the results of students in classes that just happened to be large with those that just happened to be small. As he puts it, her massive study found the effect of class size on performance to be ''precisely nil''. So we won't miss out on much if the Liberals don't win.