THE STABILITY OF G ACROSS DIFFERENT METHODS OF ESTIMATION

Multiple methods were used to estimate g (general cognitive ability) from a representative multiple-aptitude test battery. These methods included unrotated principal components, unrotated principal factors, and hierarchical factor analysis. Several variants of the hierarchical factor analyses were used ranging from three to eight factors. Fourteen estimates of g were made and computed in the normative sample for the test. The correlations of these estimates were high, ranging from .930 to .999. For this test, all other multiple-aptitude batteries, and any other set of variables that displays positive manifold, it is argued that the methods are equivalent. This is not due to similarity of factoring techniques, but rather to the positive intercorrelations of the variables as demonstrated by Wilks (1938).