Date:         Fri, 10 Jun 2005 10:28:03 +0200
Reply-To:     Marta García-Granero
              <biostatistics@terra.es>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Marta García-Granero
              <biostatistics@terra.es>
Organization: Asesoría Bioestadística
Subject:      Re: NOMREG or CATREG, DISCRIM or Multinomial Logistic?
In-Reply-To:  <1dbd3134050609095455a559b5@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

Hi Nico,

NP> I haven't been able to find much in the help menus regarding NOMREG. Then, I
NP> found CATREG. It appears that CATREG may be best with categorical predictors
NP> and an ordinal (ranked) criterion. yes? Is NOMREG better for a nominal
NP> criterion? If so, does anyone have a reference or pointer to where I can
NP> learn more about it int he help system.

NP> Then, I started to think about Multinomial Logistic, and began to one how
NP> one chooses between this and DISCRIM. Any thoughts, experiences, or
NP> references on this question?

Multinomial Logistic Regression is NOMREG.

If I remember well, DISCRIM requires continuous and normally
distributed predictors (also, multivariate homogeneity of variances is
required). See:

http://www2.chass.ncsu.edu/garson/pa765/discrim.htm

On the other hand, logistic regression methods can be used with
categorical predictors.

I insist you take a look at log-linear methods (GENLOG). Agresti's
book (Categorical Data Analysis. New York: Wiley. )has several
chapters about them, with worked examples. Also, you can take a look
at:

http://www2.chass.ncsu.edu/garson/pa765/logit.htm

Also, HILOGLINEAR, can be of help.

--
Best regards,
 Marta                            mailto:biostatistics@terra.es