ABSTRACT
The centroid method of factor extraction was developed by Thurstone as a practical method for use with the desk calculator. It is an approximation to the principal factor method that, though more demanding of computation time, does extract the maximum amount of variance with each succeeding factor. The centroid method only approximates this ideal. With electronic computers, the additional computation time required for doing a principal factor solution, instead of a centroid solution, does not represent an unreasonable expenditure of resources for most factor analytic problems. Consequently, the principal factor method in recent years has replaced the centroid method as the most widely used method of factor extraction. With the principal factor method, however, it is still necessary to estimate the communalities. Because there is no clear solution to this problem, a certain indeterminacy is introduced by using estimated communalities because the solution can be affected substantially by the values chosen. It is possible, however, to obtain a solution without using any communality estimates by using the minimum residual method of factor extraction (Comrey, 1962a; Comrey & Ahumada, 1964, 1965). This method is described later in this chapter.
