Date: Fri, 11 Jul 1997 16:54:17 EST
Reply-To: peter homel <HOMEL@SACC.HSCBKLYN.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: peter homel <HOMEL@SACC.HSCBKLYN.EDU>
Subject: answer to transform in repeated measures MANOVA
Hi -- this is an update and answer to my own question about how to
use a transform statement in a repeated measures MANOVA procedure in
SPSS. I needed to have a way of getting polynomial contrasts on a
repeated measure where the repeats were not of equal intervel (3, 4.5,
7.5, 10.5, 12)
The answer is that you don't use MANOVA. Instead you use the GLM
procedure in SPSS 7.5 (assuming you have access to it and use the
contrast statement as follows:
GLM deg3 to deg12
/wsfactors=degrees 5 polynomial(3,4.5,7.5,10.5,12)
/print = parameter.
The Advanced Statistics Manual for 7.5 on page 381 insists that you need to
use integer values for the metrics (i.e., 30,45,75,105,120) but I get the
same results using the actual values.
By the way, speaking of the Advanced statistics manual for 7.5, it
appears to contain a contradiction (dare I say error?) regarding the
commands for a repeated measures MANOVA. On page 461 under the Syntax Rules,
the manual states:
Do not use the TRANSFORM subcommand with the WSFACTORS
subcommand because WSFACTORS automatically causes an othonormal transformation
of the dependent variables.
[The WSFACTORS subcommand is of course essential for running the repeated
measures procedure in the first place.]
On page 464, the manual goes on to talk about CONTRAST for WSFACTORS:
To perform non-orthogonal contrasts, you must use the TRANSFORM subcommand
instead of CONTRAST. The TRANSFORM subcommand is discussed in MANOVA:
Is it just me or do these two statements sort of contradict each other?
These two statements have been in the Advanced manual since version 5,
by the by.
If the error is with me, then I aplogize in advance (advanced?).
PETER HOMEL PHD
HEALTH SCIENCE CENTER BROOKLYN
STATE UNIVERSITY OF NEW YORK
450 CLARKSON AVENUE BOX 7
BROOKLYN, NY 11203-2098
TEL: (718) 270-7424
FAX: (718) 270-7461
MOTTO: STATISTICS DON'T LIE!(PEOPLE DO!)