ABSTRACT
For any particular correlation matrix, with n given, Kaiser and Rice proposed a measure of sampling adequacy (MSA) indicating how near R-1 is to a diagonal matrix. As MSA approaches unity, the correlation matrix becomes more and more suitable for common-factor analysis, and Kaiser and Rice suggest that if MSA <.5 the correlation matrix is unacceptable for factor-analytic purposes. A four-factor solution gave a better fit, but we had to retain factor d of the F-matrix with only the one highest loading,.259, higher than.167, and eigenvalue only.180. Ahmavaara makes the comparison graphically, by plotting pairs of corresponding loadings on the criterion factor matrix and the other transformed factor matrix, we can just as well use the coefficients of congruence to obtain numerical comparisons. The Procrustes transformation is applied to the unrotated factor matrix Fx of the group to be transformed, with Vu, the reference-vector-structure matrix of the criterion group, taken as the hypothesis matrix.
