Assessing Children's Mathematical Knowledge: social class, sex and problem-solving By Barry Cooper and M iread Dunne Open University Press pound;17.99 This readable book carries some important messages. On the back cover, the authors explain: "In England, the majority of the items in (national) tests have set mathematical tasks in such everyday situations as 'shopping'." But, they suggest, such "realistic" questions may present particular difficulties to children from working-class backgrounds.
This argument is presented in the context of the theories of the sociologists Bernstein and Bourdieu about the contrasting linguistic codes of children from a range of social classes. It clearly offers a challenge. But as the authors show, the situation is really more complex. They tested and interviewed children from six schools, using a selection of key stage 2 and 3 mathematics test questions.
The contextualised questions which assessed pupils' understanding of number, as the majority of "shopping" questions are likely to do, showed no significant bias, for or against, any pupils, whatever their social background.
On the other hand, contextualised questions designed to assess understanding of algebra, or of shape, space and measures, were biased against working-class pupils.
The authors could not compare contextualised with uncontextualised quesions on handling data, as "data" that is "handled" has to be about something, so all these questions were in context. None the less, many of their more detailed discussions with pupils focused on this topic.
On the basis of these interviews the authors suggest working-class pupils are likely to be misled by contextualised questions. They argue that twice as many working-class pupils would pass a test composed entirely of questions set out of context than a test with similar questions placed in context - although how they would develop uncontextualised questions to assess handling data is unclear.
Middle-class pupils, on the other hand, would do much worse if all the questions were out of context, although upper-class pupils would be little affected.
This detailed research, including individual in-depth interviews with pupils, is invaluable - but it might usefully be supported by evidence relating to the development of the questions.
Records of early versions of draft questions and statistics from large-scale trials are not readily accessible. Perhaps the time has come for all the data collected on past test questions to be released to committed, bona fide researchers such as the authors of this book.
Tandi Clausen-May is a senior research officer at the National Foundation for Educational Research in England and Wales