Hello,
A plant breeding customer wants to use JMP to evaluate/compare designs in order to reduce either #locations, #control seeds varieties, #repeats on control seeds, df on intra & inter-plots.
At year 1 they typically test ~600-1,000 varieties splitted in ~20-30 protocols regrouping ~20-30 varieties each. On each protocol a test-variety is tested once while control-varieties are tested 2-4 repeats to assess error and permits local geographical correction based on vertical or horizontal location of the plot within the field.
There is 10 locations (fields), 20% of Control seeds, 0.8 repeat by location for on-test varieties (meaning test-varieties are tested on 8 locations across 10 locations) compared to 2-4 repeats for each control-variety.
Using BIBD I’m not able to test > 500 treatments (varieties). When using 500 varieties I have no opportunity to reduce block size below 499 so this do not fulfill the 0.8 repeat requirement.
Using Custom design I can set 600 levels for variety. Setting 8 repeats, block size of 480 will lead to 10 blocks and 4800 tests. Unfortunately, on each block there is more than 1 repeat (up to 3-6) for selected test-varieties. Is there a way to force JMP to run each test-variety only once per block?
Note: I was thinking adding the 20% of controls on a second step by distributing them homogeneously on each block. Randomization (on X or Y axis) could ultimately be achieved on a 3rd step.
Thanks for the help.
