Scale invariance and the factor analysis of correlation matrices

摘要: The traditional treatment of estimators in factor analysis as scale free versus not is compared to the results that can be obtained on invariance. A theorem given effect we obtain scale-invariant parameters any model prescribes structure a correlation matrix, rather than covariance by maximum likelihood and generalized least squares both metric sample matrix variances. An examination made conditions under which may properly neglect scaling connecting describing with fitted matrix.