Date:         Sun, 26 Sep 2004 13:51:31 +0100
Reply-To:     Constantina Stamou <c_stamou@yahoo.co.uk>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Constantina Stamou <c_stamou@yahoo.co.uk>
Subject:      pooled data method based on Factor Analysis
In-Reply-To:  <20040926121434.BTOW22385.out006.verizon.net@frankjgallo>
Content-Type: text/plain; charset=iso-8859-1

Hello all,

I have nine independent variables measuring attitude
which I have reduced to two factors, 'positive' and
'negative'. First of all, one of the 'positive'
variables crept in the negative factor, however it has
a high negative loading. also common sense says it
should not be there because it distorts the
interpretability of the factor. Could it be due to the
fact that my sample is quite small, 34 cases?

Also, attitude was measured using 5-point Likert-type
scales. I intend to perform regression analysis.
Should I use the derived factors or can I add the
values of the 'positive' let's say factor, and use
that in regression? Is this what is known as pooled
data method? As an idea, it probably does not make
sense, however I came across a paper, and by the way
it was written, that was the impression I was given. I
am probably wrong somewhere in my thinking, so is it
possible for someone to help me?

Thank you in advance.

Tina





___________________________________________________________ALL-NEW Yahoo! Messenger - all new features - even more fun!  http://uk.messenger.yahoo.com