Research that casts doubt on improvements in national test results has remained unpublished by England's exam regulator for nearly two years.
The failure of the Qualifications and Curriculum Authority to publish the research - which suggests that scores have risen partly because some tests have become easier - has raised concerns about its independence from ministers.
Big improvements in key stage 2 test results in the years 1997-2000 for English and maths have been described as the biggest education vote-winner for Labour at the 2001 general election.
One source with close QCA links said that the study had been delayed because it was likely to contain "some very uncomfortable results which should be part of an open debate".
The QCA received the findings of a pound;300,000 investigation it commissioned into the tests taken by seven, 11 and 14-year-olds in January 2002. They have yet to be released.
The TES understands that the report - based on the biggest study of its kind carried out on the key stage 2 tests - suggests that pupil progress in some subjects is not as great as that shown in official results.
The research, which looked at 11,000 pupils, tried to establish an independent view as to whether tests became harder or easier between 1996 and 2001.
The results suggest that, in some subjects, they became easier. The controversy comes as the QCA, under its chief executive Ken Boston, is trying to throw off a long-held reputation for secrecy and closeness to the government of the day.
In 1999, Alf Massey, an assessment expert from the University of Cambridge Local Examinations Syndicate, was contracted by the QCA to carry out the research.
To obtain independent evidence of the tests' difficulty, pupils in Northern Ireland sat versions of them in English, maths and science from four separate years: 1996, 1999, 2000 and 2001.
From 1996 to 2000, results in the high-profile key stage 2 tests climbed dramatically.
Mr Massey would not reveal the details of the findings. But he said: "The research does demonstrate that there were genuine improvements in performance in schools across this period.
"The big question is whether or not the improvements that we were able to detect were as big as, or smaller than, the changes in results in the national test levels."
The TES understands the research emphasises difficulties in maintaining standards between tests in different years, and calls for improvements in the regulator's standard-setting processes.
However, there is no suggestion that the Government deliberately made any test easier. Some got harder, the research found, and there is evidence of genuine improvements over the period.
A QCA spokesman said: "The fact that there's a time between the arrival of a piece of research and its eventual publication does not mean that anything fishy is going on. It can mean that the results are being discussed with researchers to inform policy."
The research is expected to be published by the end of the year.
Two researchers told The TES that such a long delay was unusual. One academic who has worked on contracts for the DfES - though not the QCA - said research which undermined policy was less likely to be published.
Professor Roger Murphy, of Nottingham university, who has done research for the QCA, said: "The boards, the QCA and the DfES are constantly afraid of what's going to be written next. But they haven't done themselves any favours trying to keep the lid on the system's strengths and weaknesses."
news 15 analysis 22