The main political parties are scrapping over the issue of class size, and OFSTED's high media profile has given it influence both over policy and public belief. The criticism will be seized on by teachers' leaders and Labour, which is committed to no class of more than 30 for five to seven-year-olds.
Professor Bennett, from Exeter University, believes that the great majority of parents, teachers and chairs of governors believe classes are too large. He said that for them class size was a very, if not the most, important educational issue. His assertion appeared to be confirmed this summer when 80 per cent of 13,000 parents of primary pupils quizzed by market researchers said they believed class size affected achievement.
Ministers have, however, consistently argued that there is no real evidence of the link between class size and pupil achievement. Their claims have been based largely on the report Class Size and the Quality of Education, published by OFSTED last year.
Research by a team from Nottingham University headed by Professor Day for the National Association of Head Teachers claims that the report was flawed in design, execution and presentation.
"It is impossible to be convinced about the validity of OFSTED's conclusion on class size and the quality of education. The researchers' sampling is unreliable, their approach is methodologically suspect, and their analysis is superficial and incomplete.
"The main conclusion - that there is little point in reducing classes for the majority of children and (by implication) no reason to prevent class sizes in maintained schools from continuing to rise - is both unsound and unsafe. "
OFSTED claimed in its report last year that there was no clear link between the size of a class and the quality of teaching and learning.
Its findings were based on evidence gathered from the inspection reports on 594 secondary and 1,173 primary schools.
Professor Day criticised OFSTED for taking data collected for one purpose and using it for another, saying: "Its uncritical use of its own inspection results as research data means that it conducts the debate entirely within its own 'official' definition of what constitutes quality teaching and learning. It selects from the available information to support a narrow, predetermined 'official' definition of quality."
Examination of research from both the UK and America by the Nottingham team showed that large classes led to increased teacher stress, absenteeism and burn-out, and affected pupil behaviour.
And David Hart, general secretary of the NAHT, said: "It is now quite clear that OFSTED's report was a shabby attempt to undermine what anybody with an ounce of common sense knows to be true: class size does matter."
Professor Bennett said the subjective snapshot judgments made by the 2,000-plus OFSTED inspectors said nothing about whether the pupils observed would have achieved, or progressed, better in classes of different sizes.
He urged that their findings be treated with "a good deal of caution" and said British evidence on class size and pupil achievement was sparse, flawed and ambiguous. "The recent OFSTED report is of the same stable."
Class size research and the quality of education from the publications department, National Association of Head Teachers, 1 Heath Square, Boltro Road, Haywards Head, West Sussex. Price Pounds 15. Class size in primary schools from the Research and Information on State Education Trust, 54 Broadwalk, London E18 2DW. Price Pounds 5.
In a report published last November, OFSTED said:
there was no clear link between the size of a class and the quality of teaching and learning within it;
small classes did appear to benefit children in the early years of primary education, pupils with special educational needs and those of lower attainment in secondary schools and pupils learning English as a second language;
teaching methods and classroom organisation had a greater impact on learning than the size of the class;
the use of classroom assistants had an important influence on the quality of teaching and learning, especially in larger classes;
to reduce the average class size in primary schools by one pupil would cost Pounds 170 million.