Cross-checking shows it is still too easy to misinterpret levels and abilities
UP to one in four pupils at any given time are probably at the wrong 5-14 level in reading and mathematics, Stirling has discovered after conducting cross-checks through standardised tests.
Teachers both underestimate and overrate pupils' abilities, undermining much of the target-setting agenda that is driving higher attainment in primaries. Major discrepancies underline the need for more rigorous and valid testing methods that give teachers more accurate information about their pupils' abilities, the council says.
The authority has used the Edinburgh Reading Test (ERT) since 1998 to verify 5-14 tests and last year deployed the NFER-Nelson test in maths. In both subjects, there is a significant mismatch between levels teachers say pupils are at and evidence from standardised tests.
Ian Liddle, principal psychologist, who this weekend will explain the findings to the Scottish Educational Research Association conference in Dundee, said there were great variations within and between schools in assigning levels to individual children.
"If we are using 5-14 tests for target-setting, they are not a robust enough baseline," Mr Liddle said.
In maths, there are "considerable overlaps" at each of the different levels. "At P5, one-quarter of children at level A score between 20 and 25 on the test, but so also do around half of the children assigned to level B. Similarly at P7, one-quarter of children assigned to level B score between 23 and 40; so also do the middle half of children who are on level C of 5-14," Mr Liddle said.
In an analysis for the authority, he believes there are "individual discrepancies in scores which are frequent enough and marked enough" to warrant further investigation by class teachers and school managers. Reading tests revealed similar findings, although discrepancies have been slightly reduced after teachers compared the evidence from the standardised tests with their own use of the 5-14 tests.
Mr Liddle found that the range of scores was "surprisingly large" on the Edinburgh Reading Test. "At P5, students classified as A-scored between 0 and 93; students classified as A scored between 5 and 85; students classified as B scored between 8 and 106; while those classified as C scored between 22 and 113. This pattern was repeated at all three other stages."
In 1999, some 40 per cent of pupils were inappropriately placed in reading in P7. "This shortfall is mainly due to the large number of children at P7 who appear from the ERT test to be capable of level D work, but who are still classified as level C," Mr Liddle says.
He believes it is still too easy to misinterpret 5-14 data and schools should use a mix of methods.
Testing announcement, page 4 Leader, page 22