Dear all,

I am working on survey design. I performed the factor analysis with the varimax rotation, which returned nine factors for it.

My question is: can I reduce some of the factors that contain just 3 or 4 items by merging them together?

Thank you

------------------------------
faiz abdelhafid
------------------------------

Attachment(s)

factoranalysis.pdf   992 KB 1 version

The factors are orthogonal. Why would one merge uncorrelated dimensions? Well, one may, explicitly declaring "I'm going to make a battery construct, i.e. statistically not warranted".

But a better solution would be to extract less factors.