VARIMAX AND MAXPLANE ROTATIONAL METHODS UNDER DIFFERENT CONDITIONS OF SAMPLING ERROR AND HIERARCHICAL STRUCTURE

The importance of hierarchical factor structure is well recognized in the analysis of both personality and ability factors. The notion of hierarchical factor structure arises logically from the fact that group factors may be almost infinitely subdivisible. Thus a complete factor analysis of a correlation matrix often will imply an hierarchical factor structure, with relatively specific factors descending from more general ones. The common methods for determining hierarchical s.tructure require the original factor solution to be rotated to oblique, rather than to orthogonal, simple structure. Yet the use of oblique factor rotation poses well-known problems. Relaxing the requirement of orthogonality may allow factor axes to be unduly influenced by chance variation, resulting in spuriously high factor loadings. Further, analytical methods for the oblique case, unlike analytical methods for the orthogonal case, are still in a developmental stage. Thus the bulk of currenk factor studies rely on Varimax rotation and emerge with orthogonal fac- tors, and the possibility of hierarchical factor ordering is little explored.