I have used the maximally selected log rank test to dichotomize a continuous variable (high vs low). I was curious if the maximally selected log rank test could be generalized to make 3 groups instead of two (high vs med vs low)?
I am using r and have been using the maxstat package but didn't see any scenario where 3 groups were created. https://cran.r-project.org/web/packages/maxstat/vignettes/maxstat.pdf
I understand categorizing a continuous variable can lead to a loss in power and information, and I prefer to not have answers that surround this topic.
