|
|
Hello,
I have been struggling with trying to decompose a 3-way interaction
effect. I have looked through the archives and have gotten about 2/3 of
the way through but am now stuck. I would be grateful for any advice you
could give with regards to my syntax -- for some reason, my syntax to
decompose a 2(within) x 2(within) x 4(between) doesn't work. I just get
these dots where statistics should be.
To clarify, I have a 2(within, "parent") x 2(within, "advice") x
2(between, "sex") x 4(between, "class") design. The interaction is of
parent*advice*class. I was told that even though sex has no effects, I
should leave it in the /LMATRIX since it is part of the original
analysis model.
I've gotten through some of the analyses (was able to find interaction
effects at each level of the within-subject variable), but I am
specifically having trouble with the /LMATRIX and /MMATRIX for the
decomposing interactions at each between-group variable, i.e.
parent*advice at class1
parent*advice at class2
parent*advice at class3
parent*advice at class4
If I've been doing this correctly, the syntax should look something like
this:
/LMATRIX 'interaction parent*advice at Class=1'
Intercept 1 CLASS 1 0 0 0 SEX*CLASS .5 -.5 0 0 .5 -.5 0 0
/LMATRIX 'interaction parent*advice at Class=2'
Intercept 1 CLASS 0 1 0 0 SEX*CLASS 0 .5 -.5 0 0 .5 -.5 0
/LMATRIX 'interaction parent*advice at Class=3'
Intercept 1 CLASS 0 0 1 0 SEX*CLASS 0 .5 -.5 0 0 .5 -.5 0
/LMATRIX 'interaction parent*advice at Class=4'
Intercept 1 CLASS 0 0 0 1 SEX*CLASS 0 0 .5 -.5 0 0 .5 -.5
/MMATRIX pooled parent*advice differences
all 1 0 -1 0
all 0 1 0 -1
all 1 -1 0 0
all 0 0 -1 -1
(note the inclusion of sex*class, because it is in the original model,
even though sex is not in the interaction).
For some reason, when I run the above syntax all I get are output tables
with little dots in them. I have searched high and low to see if there's
anything I am doing wrong, but not as far as I know.
I was also told that one didn't need to include the last interaction
because of redundancy, but I never quite understood that. Could anyone
give a brief response?
I would be grateful if anyone could offer some tips!
Many thanks,
Bonnie Poon Zahl
--
Bonnie Poon Zahl
Psychology and Religion Research Group
Faculty of Divinity, West Road
Cambridge CB3 9BS, U.K.
|
Locked
