Abstract
This study attempts to resolve the discrepancies between recent factor analytic studies of the Bern Sex-Role Inventory (BSRI). It examines the factor structure of the BSRI as a function of sample type (male students, female students, non-student male, non-student female) and method of analysis (orthogonal versus oblique rotation). Using the maximum likelihood method with an orthogonal (varimax) rotation, few differences were found among the four samples and hence a combined solution reporting the loadings for the first five factors is presented. Variations among the four separate samples are also discussed. The first two factors were found generally to correspond to the masculinity and feminity scales of the BSRI with the next three being labelled Sex, Independence and Negativity respectively. An oblique rotation in the total sample confirmed this basic structure and also supported Bem's conception of masculinity and femininity as separate independent dimensions. It is clear, however, that a refinement of the BSRI is necessary and one such possibility is suggested.