First, I would guess that 8 factors is too many.
Second, be careful about principal components analysis vs. factor
analysis....they aren't the same. PCA is a data reduction technique.
FA is a search for latent factors Results are often similar, but
needn't be, and the underlying question is different.
Third as to your specific question, you need to think about what you
mean by "how well". The factors are extracted based on the 36
behaviors. But no list of behaviors describes a person perfectly, and
the factors can't do as well as the questions. Do you mean how well do
the factors relate to the questions? Also, each of the 106 people will
rate people differently. So, you could get a mean on the factor scores,
but is that really what you want?
Are the factors interpretable? (If not, start over)
HTH
Peter
Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
www.peterflom.com
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)
>>> [EMAIL PROTECTED] 3/30/2004 11:04:18 AM >>>
Hi everyone!
I am learning how to use some advanced statistical tools, and I'm
encountering
some problems I'm not able to resolve on my own. I really hope that
someone
will be
enough patient to help me:)
here is my problem:
I administered a questionnaire to 106 employees which asked them to
describe
their project leaders rating a list of 36 behaviors on a "1 to 9"
scale.
On the collected data, I have conducted a Principal Component Analysis
with
a Varimax rotation. Eight factors have been extracted which account for
the
75% of the total variance.
Now, I would like to measure each factor dependently on how well
it describes the project leader of these employees. How can I do that?
Thank you very much!
Gianluca Bosco
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
