TRANSFORMATION OF FACTORS BY ARTIFICIAL PERSONAL PROBABILITY FUNCTIONS

Use of judgmental and heuristic principles has led to development of an automatized computer method for transformation (rotation) of factors. In this method the notion of an attribute vector being “in an hyperplane” is formalized in an artificial personal probability function and the hyperplane location is defined by a weighted least squares principle using the personal probabilities as weights. The method has yielded superior results in numerous trials with real and Monte Carlo data for positive manifold situations. The method has not been studied as thoroughly in the more general situation involving both positive and negative transformed factor loadings but appears to yield good results.