Multivariate analysis with dependent variables and replicates

#1
Dear All,

I have been stuck in this issue for 2nd day now.

I have a dataset of 13 depended variables (consequitive (not repeated!) measurment ("profiles" from one batch of product) and there are 3 batches (replicates) and within each batch there are 10 containers. So I am dealing with intra and intra batch variability of each of 13 dependent variables.

I have to acquire mean, variance and covariance matrix to perform a Monte Carlo simulation
to acquire 1000 simulated profiles (that is 10 lots with 100 containers).

Is it OK to perform MANOVA (class=batch, var1-13=batch) and acquire - correlation and covariance structure to be used in simulation.

I get around, if there is only 1 batch (since no MANOVA is needed and mean, covariance structure (Cov=corr*std*std) is straightforward, howewer, when interbatch variability is introduced I get lost.

Could, I show with MANOVA that batches are not significantly different and proceed as if I had one batch? On the other hand, if I failed to to this, I would still be stuck at the begining.

Could someone point me in right direction.

BEst regards,

BEfreak
 
Last edited: