How to make sense of tests

I have considerable sympathy for Colin Richards' view on testing 11-year-olds (Letters, TES, October 23 ). My regret is that he does not go far enough.

It would be sensible to undertake national testing of 11-year-olds on a sampling basis, and across a wider range of attainment than reading, writing and mathematics (which must include number). There are problems, even with that limited range.

The longish history of national tests of reading (1948 to the 1970s) shows some of these. Because everyday vocabulary changes (mannequins become models, wheelwrights become as rare as outside rights) the test questions do not represent equal difficulty as time goes by.

The different views of successive generations of testers about what is a good test and what the results signify underline the difficulties. Averaging results is the curse of the age: the spread must be shown.

The root problem is that there are no standard educational units that children's progress can be measured against. Nor should the user of test results suppose that two identical test scores (whether raw scores or scores converted to take account of the children's ages) show identical performance. They derive from the number of correct responses; the items correctly answered may differ substantially from one testee to the other.

There is no need to go further into the uncertainties related to validity (do the test items properly represent the skillachievement being tested?) or reliability (to what extent will this test produce the same results if repeated with identical children?) for it to be clear that tests lack precision. The fact that you can record a child's reading quotient to the umpteenth decimal place is a consequence of the arithmetical process used, not a sign that assessment is accurate.

None of that leads me to think that testing is useless. It does lead me to think that the results should be treated with great caution. Colin castigates telling parents about "levels". I am also against giving reading scores that mislead parents into thinking that their children can be placed precisely in league-order in their class. That is no more right for children than it is for schools.

By all means use the best that the testers can provide, but convert the apparently precise scores into something more realistic, maybe an A to E scale.

Even more important, use the test results against teachers' day-to-day observations to decide which children need extra teaching if they are to achieve the performance that will enable them to cope with the demands of schools and society at large.

The numbers of children in the grades, A to E, would provide information about the school as a whole.

Nationally, it looks as though Colin and I might agree that something like the Assessment of Performance Unit needs to be resurrected.

Norman Thomas 19 Langley Way Watford Hertfordshire

Log in or register for FREE to continue reading.

It only takes a moment and you'll get access to more news, plus courses, jobs and teaching resources tailored to you