Re: PCA and factor analysis: when to use which

[Eric Bohlman]

Ken Reed <[EMAIL PROTECTED]> wrote:
> What is the basis for deciding when to use principal components analysis and
> when to use factor analysis. Could anyone describe a problem that
> illustrates the difference?

PCA is simply a reparameterization of your data, sort of analogous to 
taking the Fourier transform of a time series.  It retains all the 
properties of your data; it simply lets you look at them from a different 
perspective.

FA, OTOH, involves assuming that your data can be described by a very 
specific kind of linear model and then fitting such a model to your data.  
Like all models, it will be wrong, but it might be useful.  By doing FA, 
you're choosing to discard some information from your data in the hopes 
that what remains will be interpretable.  You're blurring some of the 
trees in order to get a better idea of the shape of the forest.



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

RE: PCA and factor analysis: when to use which

[Dale Glaser]

references you may want to look at that address this:

Gorsuch, R. L.  (1990).  Common factor analysis versus component analysis:
Some well and little known facts.  Multivariate Behavioral Research, 25(1),
33-39.

Snook, S. C., & Gorsuch, R. L.  (1989).  Component analysis versus common
factor analysis: A Monte Carlo study.  Psychological Bulletin, 106(1),
148-154.

Velicer, W. F., & Jackson, D. N.  (1990).  Component analysis versus common
factor analysis: Some issues in selecting an appropriate procedure.
Multivariate Behavioral Research, 25(1), 1-28.

for some the bottom line may be if your intention is to maximize
variance then PCA may be appropriate.whereas if estimation of the common
factor variance + uniqueness is the primary objective, then factor analysis
may be warranted.  I have heard many opinions about this issue, and even
though it is emphasized that just relying on a generic default (e.g., PCA
with varimax) may belie the researcher's primary intention, more times than
not I have found the result and my ultimate interpretation to be very
similar regardless if I used PCA or factor analysis...



Dale N. Glaser,Ph.D.
Senior Statistician
Pacific Science & Engineering Group
6310 Greenwich Drive; Suite 200
San Diego, CA 92122
Phone: (858) 535-1661
Fax: (858) 535-1665
e-mail: [EMAIL PROTECTED]

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Ken Reed
Sent: Wednesday, April 18, 2001 3:34 PM
To: [EMAIL PROTECTED]
Subject: PCA and factor analysis: when to use which


What is the basis for deciding when to use principal components analysis and
when to use factor analysis. Could anyone describe a problem that
illustrates the difference?



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

PCA and factor analysis: when to use which

[Ken Reed]

What is the basis for deciding when to use principal components analysis and
when to use factor analysis. Could anyone describe a problem that
illustrates the difference?



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=