But last Saturday educators heard that this belief could be nothing more than a 'billion dollar hoax'.

This conclusion did not come from the government, but one of the world's leading experts on the class size issue, Professor John Hattie, of the University of Auckland in New Zealand, who claims to have studied 300,000 research papers and articles on the subject.

He was supported by Professor Eric Hanushek, an economist at Stanford University's Hoover Institution, another of the international big guns that the University of Hong Kong fielded for its high-profile conference on learning effectiveness and class sizes. Professor Hanushek has been an authority on the issue for the past 30 years. He argued that from an economic perspective, cutting class sizes across the board did not make sense.

Professor Peter Blatchford, from the University of London's Institute of Education, meanwhile, presented evidence from his research involving more than 8,700 primary school children that reducing class sizes had a strong impact on literacy, in particular during the first years of schooling - reception and Primary One - with the less academically able benefiting the most. But beyond that, it did not.

It was a timely debate as the government faces growing public pressure to move faster than its cautious pilot study involving 37 primary schools in which classes have been reduced to 25. The average class size is now 32.4 students at primary level and 38 at secondary, although popular schools may have more - as many as 40 at primary and 45 at secondary.

Secretary for Education Arthur Li Kwok-cheung has recently indicated some extension of the pilot to schools that serve the more disadvantaged communities, a concession rejected by the Professional Teachers' Union on the grounds it would involve a labeling of those schools and be unfair to others.

However, Professor Hattie warned that reducing class sizes alone could be an expensive policy failure, unless it was accompanied by changes in how teachers taught. Research showed that more often than not, teachers did not change their teaching style with smaller classes, which accounted for why cutting student numbers did not result conclusively in improved learning compared with other initiatives. 'Without changing the teaching and ensuring rigour in the curriculum delivery then the effects of this most expensive policy is likely to be close to zero,' he said.

Among the studies quoted were analysis of standardised tests results such as the Third International Mathematics and Science Study, in which Hong Kong and other countries in this region have excelled, despite their large classes.

Professor Hattie put this success down to the greater rigour in adhering to the curriculum in these countries. Where small classes were prevalent, teachers had more freedom to exercise their own discretion of what to teach and when, which could result in a weaker concept of progress across years, and lower expectations.

Professor Hanushek argued that the economic benefits from improving quality education were large, although it would take years for their impact to be felt, given the time it took to implement changes and for the affected children to enter the workforce. But common approaches to reform - including reducing class sizes and increasing teacher qualifications - had failed, he said, adding that there was no correlation in the US between a master's degree and good teaching.

Quality, he said, was best assured by focusing on the standard of teaching, regardless of class size. 'A bad teacher in a small class is still a bad teacher,' he said. A good teacher might do a little better, but fewer students would benefit from their expertise. Small classes could also leave teachers with more periods to take and less time to plan.

He spelt out initiatives that could improve quality, including policies to select, reward and retain good teachers, giving principals more flexibility to manage their resources, and demanding greater accountability of schools for student performance.

'The problem with the class size discussion is that it is imposing a rigid rule on a school and not allowing for decision making at local school level to take effect,' he said.

But local speakers were not convinced. Professor David Watkins, of HKU's Faculty of Education, questioned the appropriateness of the international research to Hong Kong. Here, the issue was complicated by factors such as medium of instruction and the local culture.

'Hong Kong parents, students and teachers believe academic success is due to effort, effort, effort,' he said. 'That means teachers are expected to give a lot of homework, to mark a lot of homework and give lots of tests. These are things that large class sizes make more of a burden. The result is overworked and overloaded teachers,' he said.

'If education reforms are going to work we have to find some way of reducing the workload.' Reducing class size was one way of doing this. 'The demographic decline is an opportunity to do that,' he said.

Dr Lo Mun-ling, principal lecturer in the Hong Kong Institute of Education's Department of Curriculum and Instruction, urged flexibility. 'Give flexibility to schools, so sometimes they can have mass lectures, sometimes small tutorial groups and small classes.'

Ip Kin-yuen, a lecturer in the Department of Education Policy and Administration at HKIEd, who has been conducting research on small classes in Shanghai, urged the audience to consider the issue in the context of education reforms, which were promoting different ideas about excellence and sought to enhance the capacity of teachers.

Teachers had to be trained for the new style of learning, he said. But that did not mean their working conditions, including the size of their classes, was not important.

Pupil's performance in standardised tests might not be affected by being in large classes. But the tests did not reflect everything. 'Most teachers in Shanghai are not expecting to raise test scores,' he said. What they had found was that students spoke more often and better in smaller classes. 'To them that is a major outcome,' he said.

Professor Blatchford acknowledged that pedagogically, teachers were more constrained with larger classes. They might do well in getting children through tests, but not in nurturing independent learners.

His research showed other benefits. 'Small classes allow for more individualised teaching, easier classroom control, more time for marking and planning, and less teacher stress,' he said.

He acknowledged that much of the research in the west had involved much smaller 'large' or 'regular' classes than in Hong Kong. In the most famous study, Project Star, 6,500 Tennessee children and their teachers were allocated to regular classes of 22 to 26 or small classes of 13 to 17. The positive results have been used to support small class policies, but the data has been challenged by Professor Hanushek.

Professor Blatchford conceded that not enough research had been done on what the upper thresholds were before large classes had a negative effect. 'Usually the debate is whether you need to go below 20 before you see the effects,' he said.

Professor Hanushek dismissed this. He said he had analysed research involving classes of 15 to 40, and sometimes even larger. 'The studies don't show a clear threshold when things get worse,' he said.

He cited Japan as an example. There, classes of 40 were common. But whether they were effective depended not on their sizes but a series of factors, such as student motivation, how teachers were trained and teaching quality, he said.

Professor Hattie urged policymakers not to adopt small classes across the board. 'If you are going to reduce class size here, identify those teachers who are brilliant at teaching classes of 15, and those specialised teachers for large classes,' he said.

For all the passion in the class size debate, among educators at the conference, at least, there was also a good deal of consensus - that there should be flexibility and that any reduction had to be matched by ensuring teachers had the professional skills to make the most of opportunities offered by teaching fewer students at a time.

One principal of a special school called for the debate to move on from the current conflict. 'If frontline people talk only about class size and government talks only about its accounting for the money we will get nowhere,' she said. 'We need an open forum prior to the formulation of policy.'

Saturday's conference opened that discussion.