Date: Mon, 11 May 1998 14:51:38 -0500
Reply-To: "Tverdek, Edward" <etverdek@SPSS.COM>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Tverdek, Edward" <etverdek@SPSS.COM>
Subject: Re: Capturing Coefficients
As of SPSS 7.5, the OUTFILE command in REGRESSION will
save an SPSS matrix file that includes unstandardized
regression coefficients and related variance estimates.
This file won't, as far as I know, include the standardized
Beta values, but it might be of some value to you anyway.
The file would be in an SPSS .sav format by default, but
if you're running production-type jobs to extract different
estimates now and put them in a separate database, you should
be able to approximate this in SPSS for Windows. You might,
for example, follow the REGRESSION specifications with a GET
FILE to call up the OUTFILE matrix you've just created and
save it immediately into a spreadsheet or database format
with SAVE TRANSLATE. There is probably a more elegant way of
doing this with the Script utility, though that's well outside
of anything I can advise on.
I hope I've understood your question and that this is useful.
SPSS Technical Support
>From: Hector E. Maletta [SMTP:email@example.com]
>Sent: Monday, May 11, 1998 2:21 PM
>Subject: Re: Capturing Coefficients
>James D. Munro wrote:
>> Here at GSBC, we do a lot of analysis that use the beta coefficients from
>> regression outputs. We have this program
>> that can scan a regression model....pick up the coefficents and constant
>> database. This database allows us to do
>> further analysis on the coefficients. This program runs on our mainframe
>> reads each line of the regression output -
>> which can be rather slow when we're dealling with multiple models with
>> different cuts of data.
>> I'm wondering if there are better ways to capture the coefficient data as
>> are trying to move away from the mainframe
>To rephrase and generanlize the question: Is there any way for SPSS to
>put regression coefficients and/or other suchlike statistical output
>into spreadsheet-like files for further analysis? Not picking them
>manuallyt one by one from the output window, but automatically when you
>have to run a multitude of similar analyses. Also, I often find myself
>needing a graph that shows the relative strength of different
>(standardized) regression coefficients from one or various regression
>runs: SPSS seems unable to satisfy these needs.
>Any help would be warmly acknowledged.
>Universidad del Salvador
>Buenos Aires, Argentina