I am trying to see if response surfaces from yield trials conducted in the same location differ from year to year. Thus I need some way of testing this hypothesis. Has anyone had any experience in this?
Try analysis of variance considering year as a class variable, and check whether the year effect is significant or not.
The year effect can be significant if yields are lower one year than another. This may not answer the question of whether trends in variability change from year to year.
I don't have any experience, but you could try to construct annual confidence intervals for the different terms of your model for the response surface. Then you could check not only if the surfaces vary from year to year. You'll find out which terms are most affected. Very valuable information. Enrique
Thanks Enrique. I find different terms are used in the model from year to year which is one indication that the response surfaces are different. I am at a loss as to whether the magnitude of the common terms change. I guess I could look at the % of the variability accounted for by each term as one indicator. WOuld that be acceptable in a scientific journal though? Typically I find 8 terms in the model. I placed a maximum limit of 8 terms.
How would I place confidence interval on a term in the model? Is there an error term associated with each term? I haven't noticed one. Thanks for any help.
- Badges
- Science topic
- Similar topics
- Agricultural Science
- Agronomy