I have recently began setting up greenhouse studies. We organize our treatments in a latin-square design and take into account all of the variation in our model. The main variation in our model is the row/column that each plant is in. Each row and column has variation due to the large amount of variation occuring in the greenhouse ie: one spot may be getting more light than another, one spot may be cooler than anothe etc.
Because of all this variation, we do not run power analyses on our study; instead we just look at historical p-values to determine what our sample size should be in upcomming studies. In non-greenhouse studies, i use the Sample Size and Power calculator with K sample means; but i've been told not to do this due to our mixed models approach to GH Studies.
Given this mixed model approach to a Greenhouse study scenario, is it actually possible to perfrom an effecteive power analysis in JMP?