STAT 8200 -
of Experiments for Research Workers
Lab 6 -
Suppose that in the redwing flaxseed example, the experiment was conducted in the same
way as described in lecture, with the exception that in each of the six plots per block,
two samples of flaxseed were taken and the percent oil content was measured on each
sample. That is, the design and data are as follows:
this example, we have 2 pseudo-replicates in each of the 24 plots. There are two
ways to analyze such data. The methods are equivalent for inference on the block and
treatment effects and means. However, in the first method we do not bother to quantify
variability due to pseudo-replicates (which is different from error variability) and in the
second method we do quantify this pseudo-replicate variability.
Copy the file redwing3.sas from the public directory to your home directory, run it, and
examine the output.
Since the experimental unit is the plot, we have 2 measurements per exper-
imental unit. We can collapse the two measurements on each plot to one by taking their