Confirmatory Factor Analysis via Noniterative Estimation: A Fast, Inexpensive Method

Abstract

A new method of estimating the parameters of confirmatory and exploratory factor analytic models is developed. It is based on generalized least squares estimation that is conditioned on a zero residual estimate of part of the model covariance matrix. The estimates can be obtained from explicit matrix formulas that include solutions to linear equations under linear equality and inequality constraints, and thus the method is noniterative and inexpensive to implement. Examples are provided, and extensions to general structural equation models are discussed.