Date:         Tue, 6 May 2008 08:48:01 -0500
Reply-To:     Robin R High <rhigh@UNMC.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Robin R High <rhigh@UNMC.EDU>
Subject:      Re: PROC MIXED
Comments: To: Emanuele.Castano@GMAIL.COM
In-Reply-To:  <d22f4555-4520-4374-a656-14058c5a3f95@w4g2000prd.googlegroups.com>
Content-Type: text/plain; charset="US-ASCII"

proc mixed produces residual values for each observation, but they are
not saved in a file. how can i eliminate observations with high
residuals ?

emanuele


You can apply the outp= or outpm= options; the first is adjusted for
random effects, the second is the means only, so you have two types of
residuals to consider, which are placed in the output file.
Since all the original data is kept in the new files, you can apply
decision rules to outliers on them.

proc mixed data=dt ;
class grp trt;
model y = trt / outp=prd(keep=grp trt y pred resid)
                outpm=prdm(keep=grp trt y pred resid);
random grp / solution;
lsmeans trt;
run;

proc print data=prd; run;

proc print data=prdm; run;


Robin High
University of Nebraska Medical Center