Split plot
Looking at the effect of 3 different crates and 2 types of lamps on behavior of pigs. There are 3 barns each with 20 crates whose types are randomly determined and never changed. The 20 crates are arranged in 2 rows of 10 with some columns being closer to the door. There will be 8 replicates of this experiment although the crate locations will never be changed.
Currently the data is being collected for the first replicate.
Advice
Look at residuals from a simple analysis of the first replicate
Fit a model that simply includes crates, lamps, and their interaction. Then look at the residuals to see if patterns emerge in terms of barns, rows, columns, etc.
Ask to move the crates into an alpha-lattice design
Apparently moving the crates is hard. But a much better design could be used for these data to block the data better.
blog comments powered by Disqus