The Meaning and Strategic Use of Factor Analysis

Abstract

The general surveys of multivariate analysis methods in other chapters in this and the first edition give some idea of the relation of factor analysis to other methods in a mathematical sense. However, factor analysis differs also in a wide, experimental, strategic sense, from, for example, both multiple-correlation linear equations, and discriminant functions, in not arbitrarily choosing a criterion variable or criterion group, but in arriving at a reduced number of abstract variables and a weighting of observed variables according to structural indications in the data itself. It is thus a means of creating concepts, not merely of employing them or checking their fit to new data, though the new methods of Maxwell, Lawley, and Jöreskog also fit it for hypothesis testing.