An Empirical Comparison of Factor Scoring Methods

Abstract

Various procedures have been used to estimate factor scores and the present study empirically compared the differential effectiveness of factor scoring by unit weights, by fractional weights using the factor loading, and by fractional weights estimated by multiple regression. The results indicated that these methods of scoring factors are roughly equivalent insofar as the intercorrelations among factor scores or reliabilities are concerned. However, estimation of factor scoring weights by multiple regression was the best method for maintaining orthogonality among the factors and appeared more sensitive in detecting group differences when factor scores derived by this method were used as dependent variables.