Just for completeness, here is the solution that I have got working:
/TEST 'Group[1]Change v Group[2]Change' group*time -1 0 0 0 0 0 1 1 0 0 0
0 0 -1 0 0 0 0 0 0 0
Kylie.
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of
Kylie Lange
Sent: Tuesday, 29 April 2008 2:45 pm
To:
[hidden email]
Subject: TEST contrasts in MIXED
Hi all,
I have repeated measures data analysed as a MIXED model - time (7 levels),
group
(3 levels) and their interaction, all fixed:
MIXED
depvar BY group time
/FIXED = group time time*group | SSTYPE(3)
/METHOD = REML
/PRINT = SOLUTION
/REPEATED = time | SUBJECT(StudyID) COVTYPE(AR1)
/EMMEANS = TABLES(time*group) COMPARE(time).
We use the EMMEANS subcommand above to get tests of time1 vs time7 for each
group, which is one of the main hypotheses of interest. Another set of
specific
hypotheses that we wish to test for is between group pairwise tests of the
change from time1 to time7. Ie:
(Group1Time1 - Group1Time7) vs (Group2Time1 - Group2Time7)
(Group1Time1 - Group1Time7) vs (Group3Time1 - Group3Time7)
(Group2Time1 - Group2Time7) vs (Group3Time1 - Group3Time7)
I believe this should be possible using the /TEST subcommand, but all
examples I
have seen have been contrasts between specific cells of an interaction, not
the
difference between two cells. Can anyone suggest how to form these contrasts
for
this model?
Thanks,
Kylie.
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
