covariate in repeated measures

covariate in repeated measures

Post by Denni » Thu, 26 Jun 2003 22:53:01



Hi all

My question refers to using of covariates in repeated measures mixed design.

I have 5 tasks, -- 5 repeated measures (T1.. T5), one between subject
variable.
Now the tricky part.

For every task there's another variable (C1..C5, 5 in total) which I suspect
have influence on the results and it meets the conditions of a covariate. I
would like to test possible influence of every covariate on the
correspondent task.

The questions are:
how does SPSS treat the covariates, mapping every covariate to all repeated
measures on in a different way?
how can I be sure that each covariate relates to the correspondent task in
the SPSS analysis (T1-C1. etc)?

Thanks,

Dennis A.
http://www.epayresearch.com

 
 
 

covariate in repeated measures

Post by wuzz » Fri, 27 Jun 2003 10:31:12



> Hi all

> My question refers to using of covariates in repeated measures mixed design.

> I have 5 tasks, -- 5 repeated measures (T1.. T5), one between subject
> variable.
> Now the tricky part.

> For every task there's another variable (C1..C5, 5 in total) which I suspect
> have influence on the results and it meets the conditions of a covariate. I
> would like to test possible influence of every covariate on the
> correspondent task.

> The questions are:
> how does SPSS treat the covariates, mapping every covariate to all repeated
> measures on in a different way?
> how can I be sure that each covariate relates to the correspondent task in
> the SPSS analysis (T1-C1. etc)?

> Thanks,

> Dennis A.
> http://www.epayresearch.com

the covariates cannot be time-dependant afaik.. Maybe it is possible
to set this up manually by blocking by subject though..

 
 
 

covariate in repeated measures

Post by Denni » Fri, 27 Jun 2003 20:59:12


Quote:> the covariates cannot be time-dependant afaik.. Maybe it is possible
> to set this up manually by blocking by subject though..

I could do one-way ANOVA tests one by one with each task , but the whole
idea is to run repetitive measures, and now this covariates is on the way.

Dennis

 
 
 

covariate in repeated measures

Post by Bruce Weave » Fri, 27 Jun 2003 22:57:37



>>the covariates cannot be time-dependant afaik.. Maybe it is possible
>>to set this up manually by blocking by subject though..

> I could do one-way ANOVA tests one by one with each task , but the whole
> idea is to run repetitive measures, and now this covariates is on the way.

> Dennis

You could always restructure your data file from WIDE to
LONG (i.e., from one row per S to multiple rows per S), and
use GLM UNIANOVA (with S included as a random factor).  This
will allow you to include your time-dependent covariates.  A
word of caution though:  With this approach, you have to
specify on the /DESIGN line ALL of the terms to be included
in the model, including all of the error terms.  Actually,
all but the final error term, which will be the residual.

I have an example of how to perform a one-factor repeated
measures ANOVA with a LONG file here:

http://www.angelfire.com/wv/bwhomedir/spss/repmeas_ANOVA_with_long_fi...

Cheers,
Bruce
--
Bruce Weaver

www.angelfire.com/wv/bwhomedir/

 
 
 

covariate in repeated measures

Post by John Reec » Sat, 28 Jun 2003 10:58:10


As I read it, you're talking about "varying covariates" here. The good ol'
MANOVA command permits fixed or varying covariates with repeated measures,
but you have to use syntax.

Try the following:
WS = repeated measure
BS = Between-subjects factor
COV = covariate

Assume that the between-subjects factor has two levels

MANOVA ws1 ws2 ws3 ws4 ws5 BY bs(1,2) WITH cov1 cov2 cov3 cov4 cov5
/WSFACTOR task(5).

Does that accomplish what you want?
--
-------------------------
John Reece
RMIT University, Melbourne, Australia

-------------------------


Quote:> Hi all

> My question refers to using of covariates in repeated measures mixed
design.

> I have 5 tasks, -- 5 repeated measures (T1.. T5), one between subject
> variable.
> Now the tricky part.

> For every task there's another variable (C1..C5, 5 in total) which I
suspect
> have influence on the results and it meets the conditions of a covariate.
I
> would like to test possible influence of every covariate on the
> correspondent task.

> The questions are:
> how does SPSS treat the covariates, mapping every covariate to all
repeated
> measures on in a different way?
> how can I be sure that each covariate relates to the correspondent task in
> the SPSS analysis (T1-C1. etc)?

> Thanks,

> Dennis A.
> http://www.epayresearch.com

 
 
 

1. manova repeated measures + covariates

Hi all,

I did a manova with repeated measures : faktor time (3) and
with sex (0/1) and group (0/1) as covariates.
In this design time, sex and group are not signifikant (p = about
.55).
Although the single correlations between sex and group and the 3 times
are r = about .0.

If I use sex and group as normal factors instead, Time ist signifikant
(.001),
sex and Group not signifikant. This is it, what I expect.

Does really sex and group explain so much variance, that time remains
not
signifikant (as in the first example)? Or did I make some mistake with
the covariates
design? Which I am afraid I did, but I don’t know what is wrong.

Thanks
Paula

2. Membership functions

3. question on a repeated measure with a covariate

4. Old CD Rom and new sound card

5. repeated measure with covariates problem

6. Folders encrypted on drive

7. Repeated measures with varying covariates

8. Messanger Service POPUPs

9. Covariate in GLM repeated measure

10. Repeated measures when time is not the repeated measure

11. posthoc tests with repeated measures: general linear model--repeated measures

12. Mixed design ANOVA w covariate: GLM UNIANOVA vs Rep Measures

13. what is assumption for repeated measures analysis using proc GLM, MIXED, GENMOD?