Date: Thu, 17 Feb 2005 13:54:20 +1100
Reply-To: Patrick Flanagan <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Patrick Flanagan <email@example.com>
Subject: Re: simple effects in GLM
Content-Type: text/plain; charset="us-ascii" ; format="flowed"
What you want to do is available through Syntax.
e.g. add the compare and adj keywords to the commands generated in
the GUI and pasted into a syntax window:
/EMMEANS = TABLES(group*time) compare(group) adj(bonf)
This will compare the groups at each of the three times. Using a
bonferroni adjustment for multiple comparisons. Other possible
adjustments are Sidak and none.
here's the full syntax file:
time3 time2 time1 BY group
/WSFACTOR = time 3
/METHOD = SSTYPE(3)
/EMMEANS = TABLES(group)
/EMMEANS = TABLES(time)
/EMMEANS = TABLES(group*time) compare (group) adjust(bonferroni)
/PRINT = DESCRIPTIVE
/CRITERIA = ALPHA(.05)
/WSDESIGN = time
/DESIGN = group .
At 6:03 PM -0800 16/2/05, Parise, Carol A. wrote:
>I've run a mixed design anova using GLM. I have one repeated factor
>(time-3 levels) and one between subjects factor (group-2 levels). I
>have a statistically significant time x group interaction. Now, i
>want to know at what levels of time are the 2 groups different. I
>recall that when i ran MANOVA, i would have done something like this:
>/design = time1 within group
>/design=time2 within group
>/design=time3 within group
>and the result would be anova tables for each of these tests. a
>statistically significant F would mean that the 2 groups were
>different at the given time.
>i cannot seem to accomplish this in GLM. the error seems simply to
>be syntax. anyone know the correct way to tell spss to do this?
>thanks for your always valuable insights.