Hello Folks,
My data is not following a normal distribution and variance is also not homogeneous. Briefly, I spiked the soil with 6 different concentrations of a pesticide and concentration of pesticide was measured in different plant tissues. The experiment was contacted in duplicates.
I would really appreciate your suggestions
Thanks
Naveen
It's not clear to me where the block comes in in your design.
A permutation test may be a good option. It does not assume the distribution of the underlying data.
The coin package in R can handle up to one-way design with blocks, which I think fits your design.
The following link has an example of a permutation test analogous to a one-way anova. You could add a blocking variable in the model formula.
Note that in this case the dependent variable is a ordered factor (ordinal) variable. You would want to treat your dependent variable as a numeric (continuous) variable.
You will probably not be able to interpret the permutation test as a test of means. Instead, you would simply say that the response of the treatments differ or not.
http://rcompanion.org/handbook/K_02.html
Other ideas...
If you plot [plant] vs. [applied], what do the data look like?
If the relationship is simply monotonic, a Spearman or Kendall regression might be the simplest test.
Is it something you could model with, say a linear plateau model?
http://rcompanion.org/handbook/I_11.html#_Toc459810570
- Badges
- Science topic
- Similar topics
- Agriculture
- Soil Science