Permutation test for randomized complete block design
Data
Treatments:
- 1 plant
- 5 plants
- 10 plants
Response is number of eggs in each treatment across 32 blocks
Question
Can we do a permutation test to determine whether there is an effect of treatment?
Analysis
For a blocked design, the permutation test should permute treatments within a block.
Previously used a negative binomial regression model.
Advice
Concerns
How is the experiment run? Are all 32 blocks run simultaneously? Is there only one cage and it is reused 32 times. How many butterflies are put in each cage?
Permutation test
Permute treatments within a block
Responsible variable
There is a decision to be made about whether the response is the number of eggs (in a treatment unit) or the number of eggs per plant (in a treatment unit).
blog comments powered by Disqus