split plot unbalanced number of replicates

[#1]

hello,

I'm working on a large breeding experiment testing the effect of drought on different genotypes. We set the main plot to irrigation level ( 2 levels), and the sub plot is the genotype ( 300 genotypes), 280 of them replicate in all three blocks and 20 are only found in two of the block. is there a way to estimate main effect of this kind of experimental design on JMP?

thanks,

Roy    

[#2]

JMP can model using Restricted Maximum Likelihood (REML), which means that you should be able to model unbalanced split plot situations like this.

[#3]

I want to expand just a bit on @Phil_Kay's reply. We use linear regression to analyze such data. The older, manual estimation methods required balance in order to be simple for manual estimation is not required for regression. The imbalance will affect the standard errors and therefore the confidence interval lengths, but it will not bias your estimates.