I am presenting the results of a study and analysis that I recently performed
later this summer.  I will need to present the theoretical details of
the statistical model, and I have not been able to find the
information.  The design is repeated-measures (30 measurements per
subject on each of 3 medications).  The independent variable is
medication, though again, each subject undergoes each treatment for a
length of time, and is measured on two dependent variables at
different time points.  I think this is called doubly-MANOVA.  I would
specifically like (i) the model (ii) the MANOVA table and (iii) any
references found that are helpful in coming up with (i) and (ii) and
(iv) a brief explanation.  Thanks.

---

Request for Question Clarification by pafalafa-ga on 17 Jun 2005 16:32 PDT

I beleive you're referring to a two factor analysis of variance,
commonly known as a 2-way ANOVA.  Have a look at this link:


http://davidmlane.com/hyperstat/B111146.html


and let me know if you think this is the sort of test you need.

Thanks,

pafalafa-ga

---

Request for Question Clarification by pafalafa-ga on 17 Jun 2005 19:38 PDT

On second thought, just ignore my above remark...it's been too long!  

paf

---

Clarification of Question by argus8-ga on 18 Jun 2005 12:22 PDT

Thanks, will disregard, but will also clarify in case it helps other
researchers - it is not two-way anova, because 1) there is one factor,
and the other factor is time, as it is repeated measures, so it can't
be treated as a separate factor 2) the dependent variable is actually
a linear comination of TWO variables, hence Manova type model. 
Thanks, and let me know if you are able to find anything else.

---

Clarification of Question by argus8-ga on 01 Jul 2005 10:56 PDT

I do have the data and have already run the analysis - what I need to
do at this point is make a MANOVA table and write out the model.  I am
having difficulty doing this as I only have a MANOVA model without
repeated measures and I don't understand it well enough to adjust (for
example, my standard errors are no longer random, right?).  So I need
only the theory behind the test - what is spss doing
mathematically....

---

Request for Question Clarification by hedgie-ga on 12 Jul 2005 04:23 PDT

Is this question still active?

If you sre still looking for help, please go through this
http://ibgwww.colorado.edu/~carey/p7291dir/handouts/manova1.pdf
and indicate where you are getting lost.

---

Clarification of Question by argus8-ga on 12 Jul 2005 09:56 PDT

the question is still active.  i will go through the handout you
suggest this week and indicate my confusion.  thanks.

I think what you describe is sometimes called 'doubly multivariate designs'.

In the SPSS package the procedure MANOVA permits analyses of repeated
measures designs in which the dependent variables represent
measurements of several variables for the different levels of the
within-subjects factors.

Your within subject factors are Medication (with 3 levels) and Time
(with 30 levels). You have a multivariate (dimension 2) dependent
measure.

Have you thought about using Generalized Linear Models (GLM) or
Generalized Estimating Equations (GEE)?  That would handle repeated
nature of your data.

do you have the data, eviews will work fine to complete your task. Or
do you have any simulated data? If not let me know I can get some of
my students to collect this data, or simulate it for you.
Ed.