Mary Hilton, of Cambridge University, says England's performance in the Progress in International Reading Literacy Study has been overstated because the design of the questions favoured English pupils. The tests, taken by 10-year-olds in 35 countries, were overseen by the National Foundation for Educational Research, which writes our national curriculum tests, to a similar format, she says. And a lower proportion of lower ability children took the tests here than in other countries, also limiting the study's usefulness, she says.
Her claims, in a paper for next month's British Educational Research Journal, are potentially devastating for ministers, who seized on England's good performance to argue that primary standards had been transformed and the national literacy strategy was working.
The Pirls tests were last taken in 2001 and the results released in 2003, when Charles Clarke was the education secretary. England finished third out of 35 countries, after Sweden and the Netherlands and ahead of the United States, Germany, Scotland and France.
Pupils were given a booklet with fiction and non-fiction texts and asked a mix of short answer and multiple choice questions. English pupils would have been prepared for these conventions, because "teaching to the test"
was rife in classrooms here and encouraged by the Government, Ms Hilton says.
"It is hardly necessary to point out that this put English children in a position of peculiar advantage when they came to do the international test," she says.
The paper also said more lower ability and non-native English speaking children were barred from taking the tests here than in all but three other countries: Israel, Greece and Russia.
A Department for Education spokeswoman said the study was carried out independently and had gone through detailed development to remove any possibility of bias towards any country.