Dear R-List,
I have carried out a factor analysis using fa (psych) with nfactors=2,
rotation="oblimin" and fm="pa". Now I have to report both
pattern AND
structure matrix. As I have understood R-Documentation, this can be obtained
by setting the "oblique.scores" argument TRUE (structure matrix) or
FALSE
(pattern matrix), respectively. However both produced the same loading
matrix even though my factors correlate to .65 (shouldn't this have a
different effect on pattern matrix than on structure matrix?).
In order to find a reproducible example, I tried again using the
"Harman74.cor" data mentioned in the fa (psych) example. Again the fa
output
produced equal results with "oblique.scores" = TRUE/FALSE. Finally I
have
run the analysis with my initial data using SPSS and have found a
substantial difference between structure and pattern matrix AND my fa output
resemble the SPSS' pattern matrix more than the structure matrix (I thought
fa was using structure matrix?).
This might be very basic misunderstanding but how can I find pattern and
structure matrices using fa, and why is the fa result more like the pattern
matrix in SPSS than the structure matrix?
I would be very thankful for any hint.
Best wishes
Alain
[[alternative HTML version deleted]]
