Peter Robinson, senior economist at the Institute for Public Policy Research, said ministers should re-examine their drive to hit tough test improvement targets at key stage 2.
His comments follow figures released by the institute this week which suggested that impressive gains in primary test scores in the late 1990s were not reproduced when the same pupils took GCSEs in 2002-4.
Although KS2 English scores jumped from 58 per cent achieving level 4 in 1996 to 75 in 2000, the proportion of pupils achieving five or more good GCSEs followed the same gentle improvement trend in 2000-04 as during previous years. This phenomenon was first reported by The TES last September.
Mr Robinson said it was also observable if one compared the improvement of KS2 English results in the late 1990s with English GCSE scores five years later.
Mr Robinson, IPPR senior economist, who gave a seminar last month on the findings, said that the Government now accepted that improvements in primary results had not been reflected in secondary performance.
This called into question whether ministers should continue to push primaries to achieve the target that 85 per cent of pupils should achieve level 4 in English and in maths by next year.
He said: "If it is true, as everyone agrees, that gains at the end of KS2 have not filtered through to gains at GCSE, what does this tell us about the key government targets, like the 85 per cent target for attaining a level 4 or above at 11?
"Do we need to make one more heave in order to achieve that target, with all that involves for primary schools, if the gains just dissipate by the time most children take GCSEs?"
Mr Robinson said there could be two reasons for the disparity: either the KS2 results overstated the true improvement in primaries, or the gains were genuine but pupils were falling back in secondaries.
The think-tank is to launch an investigation this autumn into the effectiveness of the key stage 3 strategy and also to question the purpose of assessment.
It will also look at whether the system of holding teachers to account for test results was having "undesirable distorting effects on learning".
Whereas GCSE and A-level results had a currency for pupils in the labour market, test results were not important in the same way, said Mr Robinson.
He said he believed that the emphasis on test results had gone too far, but that it was important not to prejudge the research.
Primary forum 28