ABSTRACT
The Promax transformation tends merely to "close up" the orthogonal hyperplanes to oblique positions. If the orthogonal hyperplanes are not closely parallel to the best oblique hyperplanes, the Promax procedure applies to them little or no "spin" about the origin in order to attain the optimum oblique positions. Though the weighted-varimax rotation improves the over-all parallelism somewhat, as compared to the unweighted normal-varimax rotation, the hyperplanes of an orthogonal factor matrix cannot be closely parallel to those of the final oblique factor matrix unless the latter form an almost-symmetric structure, which they seldom do. Powering the elements of a varimax or weighted-varimax factor matrix is not a rotational procedure, and the numerical reduction of the loadings by powering may be considerable. A variant of these procedures is the direct oblimin solution that gives directly the primary factors. In order to use the Optres procedure we must know in advance, for each factor, the salient variables and the nonsalient variables.
