I recently ponder about the way to interpret the factor correlation matrix in an EFA (e.g. delivered in SPSS output). I realized that, despite the factor correlation matrix showing a negative correlation between two factors, these two factors may be positively related when I just calculate the mean and then run bivariate calculations. Also, the sign in the factor correlation matrix seems to be dependent on potential other factors in the analyses. For instance, while I am especially interested in the relation of two factors, I originally had 4 factors in the analysis. There, the correlation between the two factors of interest was negative. However, when I removed the other two factors, the correlation became positive.

I now wonder whether it actually makes a lot of sense to try to interpret the factor correlation matrix the way I did. Or what does it tell one, in the first place?

Thanks a lot!