
In version 9, use rotate. See [MV] factor postestimation for more details.
Title | Correlations between factors after oblique rotation | |
Authors |
Gene Fisher, Department of Sociology, University of Massachusetts Kenneth Higbee, StataCorp |
After factor analysis (the factor command), you can obtain several matrices using the matrix get command (see [P] matrix get). These matrices include
Ld factor loadings Ev eigenvalues Psi uniqueness Co correlation matrix SD standard deviations Mean means
In addition after rotation (the rotate command), you can obtain
L factor loadings after rotation
The correlation matrix of the rotated factors can be computed as
Phi = T*T'
where
T = inverse(L'*L) * L'*A L = the factor loadings matrix after rotation A = the factor loadings matrix before rotation
So to obtain the correlation matrix of the rotated factors you
. factor ... . rotate ... . mat L = get(L) . mat T = syminv(L'*L)*L'*get(Ld) . mat Phi = T*T' . mat list Phi