The status of the Neale Analysis of Reading Ability (Revised British Edition) is not so straightforward as Michael Jackson says it is (TES, October 4).
The standardisation took place at a time when there was considerable hostility to the assessment of young children, and of 300 schools invited to take part, only two-thirds agreed. In the case of pupils aged six to eight, 46 out of 80 schools provided returns, and no further information was provided on the nature of the schools which took part.
The National Foundation for Educational Research aimed to test 400 pupils from this age group, but actually tested just 195. I pointed out in my review of the test in The TES at the time that there was no way to be sure that this remained a representative sample, and this is still the case. The sample of pupils under six who took part was smaller still - only 126. Although data from these pupils was available, accuracy and comprehension scores of all pupils under six are described in the manual as having been made "by extrapolation". This flies in the face of all professional practice - the NFER sample of children under six was either suitable as a basis for assigning scores or it was not.
The manual proceeds to give a reading age for accuracy of 5.09 for pupils whose raw score is 16, based on reading the first passage accurately. The Australian standardisation, based on a sample designed to be representative of the whole of society, showed an average error rate of 10 on one form of this passage and 11 on the other. A pupil making 10 errors on the first passage would not have reached the baseline score of 5 on the British standardisation.
The NFER based its scores for children under six on the performance of children aged over six rather than on its own scores, and to this extent it has admitted that its sample was unreliable. Add the low participation rate of pupils under eight, and the reliability of the whole exercise for these pupils is called into question. The truth of the matter is that the 1989 standardisation exercise foundered on the failure to guarantee a representative sample, and that very low scores on the test should not be accepted without corroboration. The most important effect of this episode has been to make problems look worse than they really are, often to the distress of parents and teachers. We can only hope that sufficient care has been taken with the new sample to get it right this time.
JOHN BALD 7 Symonds Lane Linton Cambridge