What is your purpose in doing the factor analysis? If it is just to reduce the number of variables (as is typically the case with principal components) then myvar would go on both factors. If you are trying to determine latent constructs then I would not use principal components. Principal Axes or maximum likelihood would be more approapriate.
Dr. Paul R. Swank,
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center-Houston
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of antonio ml
Sent: Tuesday, March 01, 2011 4:14 AM
To: [hidden email]
Subject: Factor Analysis, doubts with grouping variables into factors.
Hi list.
When doing Factor Analysis, (principal component analysis) I've got a satisfactory solution , not perfect but good enough for our (preliminary) purposes.
But we're having some doubts with grouping variables into factors.
The thing is that one variable, let's call it 'myvar', load with 4.1 in factor nº3 (which explains a 32%), and with 0.391 in factor nº10 (which explains a ridiculous 3.2%).
Also happnes that Factor nº10 only has one variable (with a loading of 0.772) and seems more logical, more readable to asociate 'myvar' with the factor #10 and not with the factor #nº3, so by the way we would have all factor with more or two variables, what looks nicer.
Can I do it?,
further than this example can I associate variables in diferents factors if they look more logical (to name it for example) and have some loadings (similar ones) in those new factors even when new load is lower?
One more question. How should I compute factor scores ? the same way?
Thank you very much.
Antonio.
Locked
