This week's GCSE results are a salutary reminder of the vagaries of exams.
How can a cohort of 11-year-olds whose test results leapt five years ago have notched up only a small improvement in their GCSE scores? A look at research by Durham university's Peter Tymms, profiled on page 9, may offer some explanation. His study of primary performance suggests standards of reading have scarcely risen since the mid-1990s. He thinks the improvements are mainly down to better preparation for the tests. Authoritative, unchallenged research by Professor Dylan Wiliam argues that the tests are so unreliable that at least 30 per cent of pupils are awarded the wrong level at the end of key stage 2.
But we cannot be certain that flaws in the tests are the reason for the discrepancy. The challenges facing secondary teachers as they battle with increasingly obstreperous pupils and an imperfect curriculum may explain why teenagers' progress is slower than that of primary children. Professor Wiliam suggests it makes little sense to compare results of one age group with those of another when these are based on different programmes of work.
But ministers might pause to reflect on the test and target-driven regime that has led us to make the comparison. Tests are a necessary and inevitable part of any school system. Put the results in high-stakes targets and league tables and you suggest that they have more importance and reliability than they deserve. If teachers, rather than tests, had a bigger role in assessing pupils, we might have a better idea of how much progress pupils are really making. Teaching to the test is boring for teachers and pupils and narrows the curriculum. Teachers and government advisers have made a powerful case for teacher assessment checked by some external tests. Ministers, looking nervously over their shoulders at the tabloid press, have not yet proved brave enough to agree. Meanwhile, the present regime continues to send out the depressing message that if we have tested children, we have educated them.