More details on the new biennial science tests for 11-year-olds to be taken for the first time next year have emerged from the Department for Education.
Around 1,900 schools – or 11 per cent – will be chosen to take part in the sample programme. A random sample of five pupils per school will be selected for the tests, making a total of 9,500 pupils.
Previously it was thought that all schools would have a sample of pupils undertaking the test but this has since been cleared up by the government.
The changes are designed to give a more accurate picture of science knowledge among primary school leavers compared to the previous sampling method of testing 27,000 Year 6 pupils from 750 schools.
There will be 15 tests lasting 25 minutes each created for the new science sample programme compared to two 45-minute tests previously. Each pupil will take three of the new tests.
A spokeswoman for the Department for Education said: “We will be assessing more of the curriculum in a greater number of schools so we are confident that this is a better estimate of national performance than the previous model.”
The science tests were previously part of Sats, compulsory tests taken by every Year 6 child, until 2009 when a review found they gave little information on how well a child would do at science in secondary school.
A sample test was introduced in 2010 and the results showed a large fall with 81 per cent of pupils reaching the expected level 4 in 2010, compared to 88 per cent the previous year. The results rose again to 85 per cent in 2011 and 84 per cent in 2012.
Last year, they were suspended as work began on the new biennial sample tests as outlined in the Standards and Testing Agency business plan for 2013/2014.
Kathryn James, director of policy for NAHT, said: “We’re relieved that it looks likely that not every school will be involved in these tests and that the process should be less onerous on the sample schools taking part. However, we’ll need to look closely when the details are officially announced to ensure that the data produced by the sampling is robust.”