Factor Score Estimation
Scott L. Hershberger · Encyclopedia of Statistics in Behavioral Science · 2005
Abstract Factors scores are measures of principal components or common factors. Under the principal components model, the factor scores are uniquely determined; under the common factor model, they are not. In the latter situation, the factor scores are indeterminate, potentially having an infinite number of solution sets, and thus their true values can only be estimated. Three methods of factor score estimation are discussed: (a) regression, (b) Bartlett's method, and (c) Anderson and Rubin's method. Although the properties and values of the factor score estimates produced by the three methods differ, the estimates are in general highly correlated.