Ikenna,
I apologize that I do not have the expertise to properly understand the report you provided. So, may I ask a few questions?
How many varieties of yam were planted?
Was each variety planted in each of three locations?
If each was planted in a location, how was the planting distributed? (A non random planting my not recognize a particularly fertile spot near a less fertile spot.)
What does 'number of observations' mean? Does it relate to the number of plants? Or the frequency in which the plants were measured?
How tightly were other variables controlled?
I believe I read about sampling theory in such cases, written by W. Edwards Deming, possibly in one of his books on sampling theory, or perhaps one of his articles.
Dr. Deming warns us against erroneous inference in many cases. If the data collected in the study is purely enumerative, then no causal inference can be made.
So, is the question, 'which yam grew best in all 3 locations?' Or does the study also plan to establish 'why a yam grew better in those locations?' The first case is enumerative, the second is analytical.
I'm curious what decision might be made if it were found that each yam strain did best in a different location? Would each strain then be used in it's favored location?
Again, this is not my area of expertise, so please forgive any ignorance I've exhibited here. I would love to learn more about your study.
-Terry
------------------------------
Terry Rosen
BoulderTEC High School
------------------------------
Original Message:
Sent: 03-01-2019 10:23
From: Ikenna Nnabue
Subject: AMMI statistical model
Dear all,
I am actually collaborating on a research work in my Institute with sole objective of conducting a stability analysis of yam genotype yield across three different locations. AMMI model is the only statistical technique that comes to mind and I want to know the yam genotype that is stable in terms of yield across the three locations. I have attached here a file 'AMMI R output' for your view and input.
Questions:
1) how can I interpret the output with respect to the objective of the research
2) is the output enough to achieve my objective.
Lastly, with deep sense of appreciation, i want to sincerely thank all that contributed to resolving my earlier challenge on principal component regression.
Best Wishes
------------------------------
Ikenna Nnabue
Research Officer
National Root Crops Research Institute.
------------------------------