The Estimation of Factor Indeterminacy

Abstract

The degree of indeterminacy of the factor score estimates is usually measured by R2, the sample multiple correlation between the factor scores and the original variables. However, this value, and any of its linear transformations, is biased and, thus, can lead to erroneous conclusion regarding the nature of the results. The magnitude of this bias is illustrated, and the use of an adjusted version of R2 is recommended. Finally, some guidelines for designing factor analytical studies in which factor scores are to be estimated are offered.