Date:         Thu, 28 Oct 2004 10:47:41 -0400
Reply-To:     Michael Doherty <mdoherty@NERI.ORG>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Michael Doherty <mdoherty@NERI.ORG>
Subject:      GLM MANOVA to Mixed

I'm trying to get around the issue with SAS in that the post hocs run for a
MANOVA are univariate post hocs -- not post hocs based on the MANOVA.

So, my question is, can I get post hocs based on the MANOVA?  If not, can I
convert to proc mixed, and if so how?

If I need to convert to proc mixed, can somebody show me how to do so using
the following code?

I have two subscales, X and Y, at time 1 and time 2.  So, naturally, the
variables are X1, X2, Y1 and Y2.  I'm controlling for age, group and
education level.

proc glm data = XXX; title '4 group 2 subscales age & edu';
 class intgrp;
  model X1 X2 Y1 Y2
     = group age education/nouni;
 manova h=group
     M= X2 - X1, Y2 - Y1/
     SUMMARY;
 lsmeans group;
run;