Abstract
Two experiments were performed to investigate the critical determinants of group performance on a multiple-solution anagrams task. In both experiments, subjects were required to make five-or-more-letter words out of a set of twelve letters. In the first experiment, the subjects were male American college students who worked either alone or in two-member groups or in three-member groups. While two-member groups were found to produce more words than and not to differ on errors from individuals, three-member groups did in no way differ from two-member groups. In the case of both two-member and three-member groups, nominal groups (formed by pooling the work of individuals working alone) were found to produce more errors than real groups, but there was no difference on words. In the second experiment, with female American college students as subjects, two-member nominal groups, quasi groups (made up of members working for a common goal but not allowed to communicate with each other), and real groups were compared with reference to two task conditions (set and nonset). Five words were presented to the subjects in the set condition (unlike in the nonset condition) to channelize their problem-solving activity in particular directions. The groups did not differ on words in any task condition. Real and quasi groups made fewer errors than nominal groups in the nonset condition but not in the set condition.