Multiple Comparisons in Repeated Measures MANOVA

[Lisa-39]

Hi there,

I ran a repeated measures manova (two repeated dependent variables and three
between subjects variables) and am now interested in the multiple
comparisons between the independent, between subjects variables and the
dependent variables.

As far as I can tell, SPSS provides these types of tests, but they are
performed on the AVERAGE of the dependent variables, not each individually.
I'm interested in the relationships between the independent variables and
the individual dependent variables. What I'm really looking for is a
multiple comparisons table like that provided in a regular MANOVA output
(without the repeated measures).

I don't think I can 'get there from here' - is it appropriate for me to go
ahead and run a regular MANOVA to obtain these multiple comparison tests?

Or, should I be doing something else within the repeated measures MANOVA
procedure to obtain this information?

Thanks much,
Lisa

[Maguin, Eugene]

All,

Somebody has hypothesized that A but not B will predict Y. Y is
multicategory ordinal and A and B are dichotomous. I am a little stumped
about how to structure an analysis(es) for this. I think of this as a
multivariate test in that A and B jointly predict Y. But would two
univariate tests be logically equivalent? I think not but am I missing
something?

A predicts Y is easy. B will not predict Y is not easy. I understand that
this statement could describe three possible situations: a) the coefficient
for B is 0.0, b) the coefficient for B is significantly less than the
coefficient for A, and c) the coefficient for B is not significant.
Situation a) implies, I think, a test for equality. Somebody on the list
sent me a Pysch Bulletin article about eqivalence testing for t-tests.
Can/how would one construct an equivalence test in this situation. Situation
b) implies a test of coefficients. Is such thing possible for ordinal
logistic? How about for normal regression? Situation c) is simply accepting
but not proving the null hypothesis.

Anybody have any thoughts about something like this?

Thanks, Gene Maguin

[Maguin, Eugene]

All,

I have data like
Id x1 x2 x3 y1 y2 y3
10  5  6  7  0  2  3
11  4  5  6  1  1  2

I want to go to
Id index x  y
10   1   5  0
10   1   6  2
10   1   7  3
11   1   4  1
11   1   5  1
11   1   6  1

My reading of the varstocases command is that it won't work for this sort of
input data structure. If I only wanted x1-x3 or y1-y3, it'd work fine. Can
someone with more experience comment?

Thanks, Gene Maguin

[Melissa Ives]

Gene,

It seems to me that after running the VarstoCases command, you could
simply select where index=1 and you will have the file you list below.

Melissa

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Gene Maguin
Sent: Monday, August 27, 2007 10:26 AM
To: [hidden email]
Subject: [SPSSX-L] Varstocases question

All,

I have data like
Id x1 x2 x3 y1 y2 y3
10  5  6  7  0  2  3
11  4  5  6  1  1  2

I want to go to
Id index x  y
10   1   5  0
10   1   6  2
10   1   7  3
11   1   4  1
11   1   5  1
11   1   6  1

My reading of the varstocases command is that it won't work for this
sort of input data structure. If I only wanted x1-x3 or y1-y3, it'd work
fine. Can someone with more experience comment?

Thanks, Gene Maguin


PRIVILEGED AND CONFIDENTIAL INFORMATION
This transmittal and any attachments may contain PRIVILEGED AND
CONFIDENTIAL information and is intended only for the use of the
addressee. If you are not the designated recipient, or an employee
or agent authorized to deliver such transmittals to the designated
recipient, you are hereby notified that any dissemination,
copying or publication of this transmittal is strictly prohibited. If
you have received this transmittal in error, please notify us
immediately by replying to the sender and delete this copy from your
system. You may also call us at [phone] for assistance.

[Marta Garcia-Granero]

Hi Gene:

You wrote:

> I have data like
> Id x1 x2 x3 y1 y2 y3
> 10  5  6  7  0  2  3
> 11  4  5  6  1  1  2
>
> I want to go to
> Id index x  y
> 10   1   5  0
> 10   1   6  2
> 10   1   7  3
> 11   1   4  1
> 11   1   5  1
> 11   1   6  1
>
> My reading of the varstocases command is that it won't work for this sort of
> input data structure. If I only wanted x1-x3 or y1-y3, it'd work fine. Can
> someone with more experience comment?
>

I write ;)

Varstocases will indeed work OK.

DATA LIST LIST /Id x1 x2 x3 y1 y2 y3 (7 F8).
BEGIN DATA
10  5  6  7  0  2  3
11  4  5  6  1  1  2
END DATA.
COMPUTE index=1.
FORMAT index(F8).

VARSTOCASES
 /MAKE x FROM x1 x2 x3
 /MAKE y FROM y1 y2 y3
 /KEEP =  Id index.
LIST.

Regards,
Marta