04-21-2018 07:01 PM - edited 04-21-2018 07:04 PM
Hello, I am a beginner at SAS, I have done an experiment where I wanted to discover the amount of repeated observations per experimental unity is needed to correctly access the trait: distribution of bean root. The design chosen was randomized complete block.
The factors are: block (2); generations (F2 and F5) and number of plants evaluated per experimental unity (2 4 6 or 8).
The plants were evaluated by opening trenches near plants and observing the presence or absence of roots at 60 5x5 cm squares per plant. It was given 1 for the square that presented at least 1 root and 0 if no root was visible.
So I want to know what is the best number of plants that I need to evaluate per experimental unit at each generations to have a reliable answer, what tests should I perform?
I am using SAS 9.4.
04-23-2018 03:54 PM
People here can help you run a SAS procedure but not so much with choosing your statistical analysis path.
If you can do some more research and determine that you need to run an XYZQ test, someone here will probably give you the ins and outs of PROC XYZQ.
04-23-2018 03:58 PM - edited 04-23-2018 04:00 PM
Please read the third example of PROC GLMPOWER, this comes pretty close to what you are asking
By the way, "reliable" is not the right word to use here. You want to know how many data points you need to collect to be able to detect differences of a certain amount with X% confidence.