Date:         Sat, 19 Oct 2002 23:17:59 +0100
Reply-To:     Jeremy Miles <jnvm1@york.ac.uk>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Jeremy Miles <jnvm1@york.ac.uk>
Subject:      Re: New Variables from Factor Analysis
Comments: To: R Cole <recole73@insightbb.com>
In-Reply-To:  <000001c277b5$21777ce0$0200a8c0@NewDad>
Content-Type: text/plain; charset="us-ascii"; format=flowed

When you have the new variables these are factors, so you lose the original
information in the items.  The reliability that you calculated is not of
the scales, but of the scales put together into one scale.  (I am guessing
that you did an orthogonal rotation, so the factors are uncorrelated.)

Jeremy


At 16:18 19/10/02 -0500, R Cole wrote:
>I recently completed a Factor Analysis on 25 Variables with 60 cases in
>the data set.  Cronbach's alpha for the variables was greater than .75.
>The data was from a 5 response Likert scale in a survey.  The results
>reduced to 7 factors.  I am planning on using the new variables in a
>correlation matrix with 4 factors from another analysis.  In the initial
>analysis, I used the SCORES > Save as Variables option to create then
>new variables for use in my correlation matrix (Regression Option).
>Everything seemed to run properly, and the new variables were created.
>However, I was advised to conduct a reliability analysis on the new
>variables (Cronbach's alpha).  When I run the analysis on these scores I
>come up with all zeros for values in the reliability analysis.
>
>I feel I am doing something fundamentally wrong.  Can anyone offer any
>suggestions?
>
>Thanks,
>Richard