Carol Fitz-Gibbon, director of the Curriculum, Evaluation and Management (CEM) Centre at Durham University, warned north-east secondary heads last week that intervention could do more harm than good without careful scrutiny of what teachers do.
Drawing on a range of studies, Professor Fitz-Gibbon said: "Some mentoring programmes were worse than useless and wasted time and money." Evidence from 11 studies showed that providing counselling for pupils who had been through traumatic events appeared to damage recovery.
"Leave them alone is quite a good idea," she said. "Good intentions do not guarantee that what you do will be beneficial."
Professor Fitz-Gibbon, one of the most outspoken critics of government education policy, said 600 initiatives by Labour south of the border, mostly introduced without any evidence, had failed to make a difference.
Only in primary maths were there signs of progress because of numeracy hours.
She put improved A-level results down to nothing more than "grade inflation". An analysis of 60 university departments by the Engineering Council had concluded that there had been a steady decline in basic skills in maths, physics and engineering and the level of preparation for university.
The CEM Centre's value-added information gleaned from thousands of students across England supported that view.
Professor Fitz-Gibbon told heads meeting at Aberdeen University: "I find it deeply disturbing that we operate in a country that does not care about reality."
Challenging other totems, she urged schools to spend less time trying to involve parents and more working with pupils whose own aptitudes were far more important. Teachers merely reinforced disadvantage by attempting to bring in parents who could not or would not attend.
Equally, schools should pay less attention to differences between boys and girls if they wanted to make substantial progress.
Fifty per cent of the variation in exam results was the result of innate differences between pupils and other factors were much less important.
Thirty per cent of the variation was due to teachers' classroom practice.
Professor Fitz-Gibbon said that small class sizes did appear to have a positive effect and 65 randomised controlled trials confirmed that cross-age tutoring was by far the best intervention schools could make.
"There is more evidence in its favour than any other intervention. The cognitive benefits are substantial for the tutor and tutee," she said.
When older pupils worked with younger ones for short spells over three to four weeks, results and self-esteem improved for everyone involved. It was not necessarily the most academic pupils who benefited most. "If they feel they are doing well in maths, it has an impact on their self-esteem," Professor Fitz-Gibbon said.
Pupil to pupil tutoring worked best at different ages and got over issues of sexism and prejudice. Discipline improved as did attendance. "It makes kids nice to each other and boys volunteer as much as girls," she said.
Professor Fitz-Gibbon accepted that timetabling was often cited as the greatest obstacle since senior pupils were involved in exams. But pupils enjoyed it.