Choose Language Hide Translation Bar
Level II

How do I properly identify the experimental unit? Treatment by covariate interaction

I am trying to analyse an experiment that involved growing plants at different densities.  I used several different plant genotypes that each experienced the low and high density treatments in a completely randomised design (plots of genotype x were assigned to either high or low density).  In addition to measuring the fruit yield of plants (response variable), I also have a measure of plant size for each genotype, measured in a different environment (apart from this experiment). I want to use this plant size measure as a covariate in the analysis of the density experiment, and I'm particulalry interested in testing for an interaction between the covariate and the density treatment.  


The data are set up like this mock example:


genotypedensity treatmentcovariateresponse (yield)










I am using the Fit Model platform, and I have so far fit a model with the effects 'density treatment', 'covariate', and their interaction. But I think I need something else to account for the covariate being a measure of the genotype as a whole -- i.e., the covariate is at the level of the genotype, whereas the treatment is applied at the plot level.  This is somewhat like a split-plot design with two levels of treatments, but, in this case, the higher-level factor is a continuous variable (measure of size) that I am treating as a covariate. 


Could someone please help by identifying the proper model to be fit?

Article Labels

    There are no labels assigned to this post.