R help - Nov 2007 - Hottelings T2-test for multivariate lingitudinal data

Dear R-users
I've simulated a longitudinal multivariate normal data set from which
I've simulated missing-patterns such as MCAR MAR and a simple kind of
non-MAR. I've imputated the values so I now have 'complete' data
sets. I'm
trying to perform a T2-test as done in the multivariate case under th
enormal assumption. Is there something I've to think about when performing
this test on a longitudinal multivariate data set when dependencies between
the time-dependent variables are present. Any help will much appreciated.

Thanks in advance

Mauricio Malfert

	[[alternative HTML version deleted]]