Date: Sat, 13 Nov 1999 17:48:10 +0100
Reply-To: hmaletta@overnet.com.ar
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Hector E. Maletta" <hmaletta@OVERNET.COM.AR>
Subject: Re: Reading Correlation matrix
Content-Type: text/plain; charset=us-ascii
I don't know about the newer version you use (I'm still stuck in version
7.5). But regression analysis (and most of the analyses based in the
regression model) is, essentially, invariant to the unit of measurement
of variables. You may assume that all means are 0 and all standard
deviations are 1, which is in effect a very common standardization, and
go ahead with it. Of course, the resulting regression coefficients will
be standardized Beta coefficients, not ordinary b coefficients.
On second thought, you may wish some particular statistical result that
necessitates the absolute value of means and st.dev., but even in that
case using the standardized values would also do (you may always convert
them to their original units of measurement afterwards, if needed).
Hector Maletta
Universidad del Salvador
Buenos Aires, Argentina
Yow Wu B Wu wrote:
>
> I created a syntax file to read correlation matrix without
> providing means and sd for each variable. It ran in older version of
> SPSS. I have run it using examples from Pedhazur Multiple Regression in
> Behavioral Research. But the newer version SPSS won't run it without
> means and sd. Are there any ways to solve this problem?
> I have asked this question before but did not get answers.
> Please send the solution directly to me because I am not sure I am on the
> list.
> Thanks.
>
> ==========================================================================
> Yow-Wu Bill Wu
> Phone: (716) 829-3207
> School of Nursing Fax: (716) 829-2021
> SUNY-Buffalo e-mail: nurwu@acsu.buffalo.edu
> 820 Kimball Tower
> Buffalo, NY 14214
> ==========================================================================
| 
