ASA Connect

Dear fellow,

I conduct a EFA and obtain a pattern matrix through SPSS.

For one of the component, all the factor loading are with negative sign. May I know the interpretation?

Any official references to explain this?

Thanks so much!

Kelvin

dear Man-Fung Lo,

                the explanation is fairly simple. SPSS, as well as SAS and many other Stat Packs, use to start from a principal components solution for the extraction of the initial factor pattern. Because the coreelation matrix is simmetric, all eigenvalues are positive but this doesn't mean that factors have their directions concordant with the variables that load them. Often the factor ha an opposite direction and this leads to negative loadings. Why this happens for all variables is due to the problem in hand that makes things behave as they do in your case.

Ulderico.  

------------------------------
Ulderico Santarelli
Business Consultant
------------------------------

You can multiply everything by minus one just as you can for any variable.

Assuming that you have more than one factor, did you use factor rotation?  If so, orthogonal or oblique?  If orthogonal, what evidence do you have that the factors should be orthogonal?

------------------------------
James Frane
Self-Employed
------------------------------