| Date: | Thu, 10 Nov 2005 17:53:13 -0600 |
|---|
| Reply-To: | Robin High <robinh@UNLSERVE.UNL.EDU> |
|---|
| Sender: | "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU> |
|---|
| From: | Robin High <robinh@UNLSERVE.UNL.EDU> |
|---|
| Subject: | Re: Interaction in ZIP model |
|---|
|
| In-Reply-To: | <f00d287f0511101023m3acbf4c6te03dbe17523a553d@mail.gmail.com> |
|---|
| Content-Type: | TEXT/PLAIN; charset=US-ASCII |
|---|
> I want to change one error in the posting, all the covariates for
> interaction are nominal variables.
> Sorry for this and thanks in advance for your suggestions.
> Best regards,
> Tony
>
> On 11/10/05, Tony Yang <tonyyangsxz@gmail.com> wrote:
> >
> > Dear Listers;
> > I am trying to fit a ZIP model, and want to include one interaction term
> > in the model, while the interacted covariates are both dummy variables, then
> > how should I do for this kind of case? It seems the proc nlmixed do not have
> > CLASS statement?
> > Any idea will be highly appreciated, thanks in advance.
> >
Tony,
One way to get dummy coded nominal data and interactions with continuous
covariates to enter into NLMIXED is to make a new dataset with GLMMOD, as
shown below. If there are other variables (such as id) or pure numeric
data you can include them on the model statement as well.
ODS OUTPUT designpoints=dsn;
ODS EXCLUDE designpoints;
PROC GLMMOD DATA=one;
CLASS group ;
MODEL y = subject group x group*x;
run;
proc print data=dsn; run;
Robin High
Univ. of Oregon
| 
