Hello all,
I have a query about factor analysis. For some time I have been under the
impression that if you
are looking to reduce your data down from a large number of variables, to a
managable number of factors (perhaps to use the output for a further
procedure like multiple regression), that PCA was the most appropriate. On
the other hand if your are merely trying to derive meaningful factors and do
not intend to use the output further, that maximum likelihood was
appropriate.
I'd appreciate opinions on whether this is a useful distinction to make.
For example, I currently have a set of 20 variables from which I want to
derive factors. On other occasions, I have many more variables and I
perform a FA and then use the output for multiple regression.
Any thoughts?
thanks
Dewi
(This query has previously been posted to Stat-L)
|