We all have memories of particularly good - and bad - teachers, and it is difficult to imagine that teachers are not the most important factor in a good education. Unfortunately for those of us who are teachers, the study we have conducted at Monkseaton Language College doesn't support this view.
Over the past four years we have developed a sophisticated IT-based system to track students and provide value-added analysis. The Centre for Educational Management at the University of Durham has also been comparing our GCSE performance with that of 40,000 students in hundreds of schools throughout England who are monitored by the Year 11 and A-level Information Services (better known as YELLIS and ALIS).
The data show that there were no significant differences in teacher performance - in value-added terms - in most of our subject departments. The school has more than 50 teachers, but there were only two instances where teachers clearly made a difference. In each case the teacher concerned appeared to have raised GCSE performance by almost a grade.
I found the lack of difference in most teachers' performance hard to believe: my head of maths, Dave Cross, is a far better teacher than I, and yet my students' results were not significantly different from his.
Looking at the raw scores, the differences between teachers seem very real, but they disappear once factors such as prior attainment are taken into account.
If differences between teachers are not usually a statistically significant factor in student performance then we must rethink many school-improvement issues. This is particularly important when the Green Paper on pay and conditions is making teacher-impact on student outcomes a key factor in appraisal, pay, and even promotion. It may well mean that the assessment of teachers by performance is fatally flawed as it currently stands.
But if teachers do not have the impact we thought, does anything? The surprising answer is "yes"; several things do. The BBC Bitesize revision materials (books, television, and interactive Website) were especially effective. We gave more than 100 students these materials four months before their GCSE examinations and the results were staggering - up to a grade higher on average in all the subjects covered.
We were not as surprised by the impact of Bitesize as we might have been because of our recent experience in changing teaching and learning in modern languages. Mike Butler, the head of department, has developed new courses based on a heavy input of ICT. This included the use of integrated learning systems, CD-Roms, and video-conferencing student-to-student with schools in France, Germany, and Spain.
Again, the impact on students has been very significant, with average GCSE performance running at almost a grade higher than expected for the past three years.
But new learning materials or ICT are not the only solutions. Simon Thomson, charged with raising achievement at Monkseaton, launched a free Easter revision week for students during their holidays. The week focused tightly on the key principles in the main GCSE subjects, with taught and self-study modules. The presentation was modelled on sixth-form and university learning, and again the results with 120 or more students for the past two years have been impressive, with huge gains across subjects varying between a half to a whole grade on average.
These results are remarkable, but there is a need to go further, using the kind of random control trials employed in medical research (this method assigns students to experimental and control groups through an external random process). At Monkseaton such trials are being carried out to test the effect of video-conferencing on language learning, and Bitesize for students who are disaffected.
What are some of the emerging answers? Our research suggests that the way forward may lie with students (not schools and staff), and selling education to them. It implies that costly, quality learning resources, such as Bitesize, are part of the answer. It suggests that changes in the way we organise and present learning to students may produce important gains. Most important, it argues that the attempt to escape from our pre-conceived certainties will, in the end, transform learning as we now know it.
At present education is still in the dark ages. We teach in certain ways because we believe it is effective, not because we have any scientific evidence to show that it works. If we want to improve education, then we have to adopt an approach that builds upon reliable evidence.
* Professor Carol Fitz-Gibbon of Durham University, commenting on the study, said: "Dr Paul Kelley makes important points based on carefully-analysed data. How students were taught seems to have had a bigger impact on results than who taught them.
"However, there is an interpretation problem in studies such as this when students choose the extent of their participation. To give a general example: if students who spend more time on computers do better is this due to the computers or would these students have done better anyway? Carefully controlled trials are needed and Dr Kelley has already begun to do this.
"So where does this get us? The availability of good data clearly helps schools to find out what is working. In addition, Dr Kelley's observations raise questions about performance-related pay, an initiative introduced without evidence for its effectiveness.
"The Government could put PRP to the test. It could go ahead with it in some local education authorities and let the same money be spent in others on well-designed trials of various initiatives. The trials could be conducted by teachers, remunerated for the extra work involved.
"The 'evidence-based LEAs' would have a blossoming of sound knowledge about effective teaching and learning. We might also have happier, more enthused teachers."
Dr Paul Kelley is headteacher of Monkseaton Community High School Language College in Whitley Bay, Tyne and Wear (tel 0191 200 8717). Website: www.ncl.ac.ukschoolszdclouindex