The Government claims it has not gone soft on testing seven-year-olds. In its relentless drive for higher standards, the need for an accurate picture of how Year 2s in England are performing matters more to ministers than ever.
But, for the first time, the Government is prepared to concede that the present national tests may not be the best way of obtaining that information. A pilot, which begins this autumn in one-in-four local education authorities, will give far greater weight to teacher assessment in profiling seven-year-olds.
This is what teachers have been arguing for. In its Primary Strategy document, published in the spring, the Department for Education and Skills acknowledged the view of headteachers that "a teacher's overall, rounded assessment of a child's progress through the year (taking into account the regular tests and tasks that children do) is a more accurate guide to a child's progress at this age than their performance in one particular set of tasks and tests."
This does not mean the Government is willing to do away with key stage 1 tests altogether. Tests for English, maths and science will continue in the same format as before, but teachers will use them to "underpin" their own assessments.
At present, teachers produce two "levels" for each child, one derived from the tests, the other from teacher assessment. In the pilot, teachers will use the test results to "make one comprehensive judgement about a pupil's progress and attainment". The two elements, testing and teacher assessment, will thus be brought together in a slightly different balance, with teacher assessment taking a more central role.
The example of Wales is thought to have exerted some influence. Welsh ministers scrapped compulsory key stage 1 tests in 2002, to ease bureaucracy and place greater confidence in teachers' judgements. Instead of tests, ACCAC (the Welsh curriculum and exams council) now produces "Optional Assessment Materials"; teachers can adapt these to their schemes of work and use them in assessing pupils at any point in the year.
Teachers involved in the DfES pilot will have a similar freedom in deciding when in the year to use tests for different pupils. They will also have a limited choice of which tests to use: next year they can choose from 2003 or 2004 papers, and the following year from 2003, 2004 or 2005. In some instances, teachers may decide a test is not necessary if a "task" performed by a child has already given them enough assessment evidence.
This will mean more flexibility for teachers and a reduction in bureaucracy, argues Jackie Bawden, head of testing at the Qualifications and Curriculum Authority. She says: "We're emphasising to teachers that they need to deliver sufficient of the tasks and tests to apply the QCA mark scheme and obtain a level. Some children, for instance, may not need to do a reading test as well as a reading task.
"We're placing trust in the teacher's judgement, informed by the test. It is the teacher, after all, who knows the youngster, and so much happens in those first years at school."
Instead of moderating test scores as at present, local authorities will be required to moderate a sample of teacher assessments - say, five or six per class - to ensure that teachers' judgements are robust and consistent. An evaluation of the pilot next summer will determine whether the new system adds to or reduces the workload of schools and local authorities, and whether it produces reliable information for parents, teachers and ministers. "The key thing is to inform the next stage of teaching and learning," says Ms Bawden.
But the changes are already attracting criticism. Professor Peter Tymms of the University of Durham, who runs the PIPS project (Performance Indicators in Primary Schools), is even more dissatisfied with the pilot arrangements than with the present key stage 1 SATs. He says the current system does not yield high-quality data because tests for younger children have to be short and are therefore less reliable; with the pilot scheme, he believes the conflation of the two sets of information, from tests and teacher assessments, will make the assessments even less clear.
Professor Tymms argues that giving greater weight to teacher assessment undermines the usefulness of the data in promoting teacher accountability.
"What we need," he says, "is an independent body to monitor students over time, using a sampling procedure - like the model practised by the National Association for Educational Performance in the United States."
The QCA is, on this occasion, quick to defend teachers' judgements: "There is always scepticism from some people about teachers doing the marking," says Ms Bawden. "But we provide a very detailed marking scheme for them, and there is training."