Unpublished figures from school inspectors show they are overwhelmingly cautious in their judgments, grading the vast majority of primary school lessons as broadly average for teaching quality and standards.

The figures, taken from internal documents by last night’s Channel 4 Dispatches programme, show that inspectors almost never use grades 1 (excellent) or 7 (very poor) - despite the famously-held view of chief inspector Chris Woodhead that 15,000 teachers are incompetent (later revised to 13,500).

Instead, the grades appear heavily bunched around 3, 4 and 5 which, for example, account for 95 per cent of primary lessons on the crucial measure of pupils’ standards.

No lessons score 1 or 7 for pupil progress or attainment, and no lessons scored the bottom grade for teaching quality.

This has led to accusations from leading statisticians and educationists that OFSTED’s judgments, which cost more than pound;100 million a year, are uninformative, unreliable and could wrongly blight a teacher’s career.

If correct, the criticisms also undermine OFSTED’s massive statistical database. Chris Woodhead recently told MPs on the education select committee that Ofsted data should shape education policy.

Professor Harvey Goldstein, an authority on educational statistics from London University’s Institute of Education, told Dispatches: “Either there’s something wrong with the Ofsted assessment scale or with the way that inspectors are using it. There’s clearly a problem. At some key stages they may only be using three points on the scale.

“We need an independent inquiry into how inspectors reach their judgments.”

He is also concerned that one-third of the inspectors’ assessments at the sensitive grade 45 borderline are potentially unreliable and could wrongly condemn a teacher as below average - or hide the failings of poor staff.

According to OFSTED’s own research, presented to the British Educational Research Association last year, when two experienced inspectors judged the same lesson, one third of the grade 5 (unsatisfactory) assessments made by one were judged grade 4 (satisfactory) by the other.

“That’s very low reliability for such an important boundary,” said Professor Goldstein. The consequences for individual teachers could be serious.

Liberal Democrat education spokesman Don Foster responded by calling for greater openness: “Nobody questions the need for a tough and independent inspectorate. But there have to be major questions about some of OFSTED’s gradings,” he said.

“We urgently need all the information that Ofsted gathers made available. It’s almost impossible to get answers to questions in the form I want.

“I’m also concerned about OFSTED’s failure to keep to its own remit. There is no evidence provided in the annual report on the standards of attainment of pupils based on the inspectors’ own data.”

Colin Richards, formerly head of primary inspection for Ofsted, is also concerned at the omission.

In a letter to The TES, he said the absence “fundamentally undermines not just the report but the whole inspection edifice.

“Why no mention of inspector’s judgments of these standards in the report? There can only be one reason. They are not considered reliable or valid enough.”

Speaking on the programme, Chris Woodhead denied that there is significant bunching.

A spokesman later said: “The figures do not imply there is anything wrong with our grades or the way inspectors use them. Nor do they show that the inspectors aren’t prepared to use the full range.”

The attainment data was not included in the annual report, he said, because it was not comparable with the previous year’s results.