Do the particular research purposes suggest the need for homogeneous, factorially pure scales? If so, the best selection is the PAQ, which provides M and F scales that are each unidimensional. Indeed, they are so factorially pure that Spence frequently (e.g- 1985) refers to them as instrumental and expressive scales, respectively. For Spence (see PAQ section) this unidimensionality is a major general advantage of her scales over those that are more heterogeneous. However, factorial impurity is not necessarily damning to a scale's validity. Bem, for example, is unconcerned over the common finding that her scales contain several factors. To state her position simply: Bem (e.g., 1979) holds that