Date:         Wed, 15 Jul 2009 13:29:43 -0400
Reply-To:     Kevin Viel <citam.sasl@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Kevin Viel <citam.sasl@GMAIL.COM>
Subject:      Re: Question about genmod output

On Wed, 15 Jul 2009 02:26:55 -0700, xinrong <xleiuiuc@GMAIL.COM> wrote:

>I am using proc genmod.
>I am confused how to infer my result.
                      ^^^^^

Interpret, not infer :)

>
>The Null model(model A) has better QICu than model B. However, the
>output of model B does show that both explain variables are
>significant.
>
>In this case, can I say the dependant variable is postive/negitive
>assoicated with C1 and C2?
>
>Thank you for your help...
>
>*******A**************
>Proc Genmod data=tb;
>class id;
>model  d=/type3;
>repeated id;
>run;
>quit;
>
>output:
>QICu: 3280
>
>*******B**************
>Proc Genmod data=tb;
>class id;
>model  d=c1+c2/type3;

>repeated id;
>run;
>quit;
>
>output:
>QICu: 3282
>
>Source   DF     Chi-Square    Pr > ChiSq
>c1          1       40.0      <.0001
>c2          1      43.50      <.0001


This is not enough data.  The very first thing to do when you are
comparing (nested) models is to BE SURE that you have the same number of
observations.  IF c1 or c2 have missing values when d is not missing, then
you will not have comparable models.  I am not sure that this is the
source of the apparent issue.

Also, I assume in the second run c1+c2 is a typo...

HTH,

Kevin

HTH,

Kevin