[R] Repeated Measures MANOVA in R

[Peter Dalgaard]

Chuck Cleland <ccleland at optonline.net> writes:

> A. Bolu Ajiboye wrote:
> > Can R do a repeated measures MANOVA and tell what dimensionality the statistical variance occupies?
> > 
> > I have been using MATLAB and SPSS to do my statistics.  MATLAB can do ANOVAs and MANOVAs.  When it performs a MANOVA, it returns a
> > parameter d that estimates the dimensionality in which the means lie.  It also returns a vector of p-values, where each p_n tests
> > the null hypothesis that the mean vectors lie in an n-1 dimensional space (0-D space implies same vector, 1-D space implies scaled
> > vectors that point in the same direction, etc...).  However, MATLAB does not do repeated measures MANOVA.  SPSS can do repeated
> > measures MANOVA but it does not return this dimension output.  Hence, I'm trying to find an environment that will allow me to do
> > repeated measures MANOVA and determine the dimensionality of the space, before I spend several weeks trying to learn it.
> > 
> > I know the dimensionality parameter is based upon the eigenvalues of the ratio of the different SSCP (sum of squares and cross
> > products) matrices, but a) I'm not sure how to calculate the SSCP matrices for repeated measures MANOVA, and b) once I get these
> > eigenvalues and convert them a Pillai-Bartlett or Wilk's-Lambda value, I don’t know how to convert to an f-statistic.
> > 
> > Does anybody know how to do this or has repeated measures MANOVA in R (while returning the dimensionality parameter)?  Thanks in
> > advance for your help.
> 
>   I don't know about the dimensionality parameter, but have a look at
> anova.mlm().  In particular, you might want to run example(anova.mlm)
> and study the output.  Also, you might look at SSD() and
> summary.manova() in package stats.

Yep. You may also want to look at 

http://www.pubhealth.ku.dk/bs/publikationer/rr-06-11.pdf/

(which is under revision for R News)

Re. dimensionality, it is not clear to me what exactly Matlab might be
doing and how (and I can't easily find it on the net, or so it seems).  

-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907