Abstract
Rowe (1988) and Rowe, Nix, and Tepper (1986) compared single-sex and mixed-sex mathematics classes within a coeducational school in one of the few studies to use a true experimental design in which students were randomly assigned to class-types, and they claimed that the study provided strong support for single-sex classes, particularly for girls. However a critical reanalysis of the results indicated that the randomly assigned class-type intervention had relatively little effect on mathematics achievement or attitudes, and that at least some of the significant effects favoured mixed-sex classes. Furthermore the limited support for single-sex classes seemed to be stronger for boys than for girls. Whereas unanticipated difficulties in the research design complicate interpretations, claims that students randomly assigned to single-sex classes did better than students randomly assigned to mixed-sex classes are unwarranted.