I have several questions about what SPSS Factor is doing during Varimax rotation. Can you support my assertion that factor loadings are always in correlation form (rather than covariance form) during SPSS's Varimax rotation, regardless of whether the relationships matrix was a correlation or covariance matrix?
Is Kaiser normalization of the factor loadings applied by default during Varimax rotation?
Are covariance or correlation loadings rotated when Kaiser normalization is turned off?
Resolving the problem
Kaiser normalization is the default during rotation. To avoid the implementation of Kaiser normalization, you would need to run FACTOR from a syntax window/file with the keyword NOKAISER (caps not necessary) in the /CRITERIA subcommand.
Rotation is based on the standardized (rescaled if the covariance matrix was used for extraction) loadings, regardless of whether the covariance or correlation matrix was analyzed or whether Kaiser normalization is applied. If Kaiser normalization was not suppressed by the NOKAISER keyword, it is applied to the standardized loadings. The rotated loadings are transformed by the variable SDs to provide the "Raw Component" table of the rotated factor/component matrix if the covariance matrix was analyzed.