Hi, I have data from a toxicology study where I see a hormesis trend in the data and I want to make sure my hormesis model is appropriate. My model is driven by one value in the highest treatment concentration but I think the value is likely real. To test for outliers, I used a rosnerTest() from the EnvStats package on the principal component values. The outlier is ID# 20-3 in the attached data. I tried to look at Cook's distance but it does not work with my hormesis model output. I am using a 5 parameter Brain-Cousens hormesis model with the drc() package in R. Does anyone know if outliers need to be removed in this type of model and if there are other assumptions that need to be met? If assumptions must be met, how do you test for those in R with this type of model?

My code for the model with the attached data is:

library(drc)

hm.m2

More Suzanne C. Ball's questions See All
Similar questions and discussions