When SPSS performs a principal components analysis using a direct oblimin rotation it produces three matrices of loadings. A component matrix of initial unrotated loadings, a structure matrix of item-factor correlations and a pattern matrix of standardised regression coefficients between each factor and each item. For reporting purposes http://lists.asu.edu/cgi-bin/wa?A2=ind0312&L=aera-d&T=0&F=&S=&P=7624 suggests using the pattern matrix, in this situation, as it partials out the effects of other items on each factor.
Revision 1 as of 2006-08-03 13:10:30
Clear message
location: FAQ / patternMatrix