Dear All
I have to do ANOVAs on a strip plot or strip-split-plot as in D.C.
Montgomery's Design and Analysis of Experiments.
The design has 3 blocks or strips with 3 treatments going down the block
and 6 varieties, with each of the varieties across the block through all of
the treatments.
aov model is:
strip.aov <- aov(height ~ variety*treatment
+ Error(Block+Block:treatment+Block:variety+
Block:treatment:variety), data = strip)
This gives the correct degrees of freedom and SS etc.
If I use model.tables(strip.aov) the means for the main effects are correct
but the there are only 6 variety:treatment interaction means which are all
the same and the value of the grand mean. I can live with that problem.
My problem is moving to lme. The fixed component is the same. Despite all
tries at for coding the random component for a single measurement in lme
hangs or gives a non convergence error or other error message.
Any help with the code for the random factor for a single measurement and
then for 3 repeated measurements would be appreciated.
I am using splus 6 release 2 on win98SE.
Regards
Duncan Mackay
Department of Agronomy and Soil Science
University of New England
ARMIDALE N.S.W. 2351
Email: dmackay@metz.une.edu.au
Ph 02 67729794
|