Pairs of upper primary school children worked on a computer-based language task which required the completion of a short piece of text from which some letters had been deleted. Children worked in single-gender or mixed-gender pairs. Previous studies have shown that mixed-gender pairs are at a disadvantage over both boy-boy and girl-girl pairs, at least with this language task. (When a spatial navigation task has been used, girl-girl pairs are at a disadvantage.) The mixed pairs may perform less well as a result of a failure to co-operate, and so the present study observed two organizations of pairs - one explicitly instructed to discuss the task and to work co-operatively, and one in which the children were told that their contribution to the task would be assessed on an individual basis.

Performance measures indicated a disadvantage for mixed pairs of children, and a general improvement in performance for pairs instructed to co-operate. These instructions had least effect upon the performance of the mixed pairs, however, and limited effect upon pairs of girls. While girls tend to co-operate whether instructed to do so or not, mixed pairs showed little evidence of co-operative working. Pairs of boys showed the greatest gains when organized in co-operative relationships because they are effectively changing working styles, both on measures of keyboard activity and on measures of problem-solving effectiveness.