Date:         Fri, 25 Feb 2005 11:13:04 -0500
Reply-To:     Art@DrKendall.org
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Art Kendall <Arthur.Kendall@verizon.net>
Subject:      Re: Syntax for PFA
Comments: To: "Frank J. Gallo" <fjgallo@verizon.net>
In-Reply-To:  <0ICH00HVZ3UZ5ZL0@vms040.mailsrvcs.net>
Content-type: text/plain; charset=ISO-8859-1; format=flowed

PFA is not an specification argument in SPSS.

a specification like
  /DIAGONAL =  .56 .5 3*.7 .61 3*.2
would allow you to specify the max correlation that a variable has.
if you specify /EXTRACTION = PAF (aka PA2)
  /DIAGONAL = DEFAULT  would use the conventional squared multiple
correlation as the communality estimate.

With no specification of /EXTRACTION  the default is principal component
which uses ones in the diagonal and tries to account for the total variance
With specifications other than PAF, I don't know whether communality
estimates are applicable.

Art
Art@DrKendall.org
Social Research Consultants
University Park, MD  USA
(301) 864-5570



Frank J. Gallo wrote:

>Hello All,
>
>
>
>I am running a PFA on 25 variables. I would like to include in my syntax a
>command line that tells spss to use the largest absolute correlation for a
>variable with any other variable as the communality estimate for the
>variable. SAS uses a command line PRIORS=MAX. Any help with this syntax is
>greatly appreciated.
>
>
>
>Thanks,
>
>Frank
>
>
>
>