Date: Fri, 8 Mar 2002 13:54:35 -0500
Reply-To: "Burleson,Joseph A." <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Burleson,Joseph A." <email@example.com>
Subject: Re: Syntax for Categorical Variables
Content-Type: text/plain; charset="iso-8859-1"
Rob S: I spotted this problem on the previous (10.0?) version of SPSS that
was on my institutions server. The 10.1 that we have now does not have the
problem, nor did previous versions before 10.0. The GLM and MANOVA contrasts
were correct, as advertised, however, even though they had the exact same
In fact, I ran all the contrasts in LOGISTIC and found the following
contrasts to actually be sometimes the same (4 of them), sometimes different
(3 of them):
SIMPLE 1st versus 2nd group (of 3, for example)
REPEATED 1st versus 3rd group (of 3, for example)
I never saw any mention of this, and should have shared it with the group,
but when my version changed, I forgot about it. See if you find anything
similar. If not, try to plot through some benchmark problem that you know
the answer to, to see what your version produces. I did not have the luxury
of using two different versions of SPSS to compare.
From: Stanfield, Robert [mailto:firstname.lastname@example.org]
Sent: Friday, March 08, 2002 1:12 PM
Subject: Syntax for Categorical Variables
I would like to use deviation coding for the IVs in my logistic regression
procedure. I'm using the following syntax:
/CATEGORICAL = new_used, spread_c, mp_cat, prdf_cat, lt_cat.
/CONTRAST (new_used) = DEVIATION
/CONTRAST (spread_c) = DEVIATION
/CONTRAST (mp_cat) = DEVIATION
/CONTRAST (prdf_cat) = DEVIATION
/CONTRAST (lt_cat) = DEVIATION
I keep getting Indicator coding instead. What do I have wrong.
Thanks very much,