Closed ChristinaSchmidt1 closed 1 year ago
As it is now
Yeah, I had made some small changes there and I think its fine for now especially since we report the percentage to the user. So I will close this issue (Please reopen it if I missed something).
You said that the saphiro test is done for each metabolite and we report the percentage of normally distributed variables. Now we need to decide at what percentage of not normally distributed variables we should recommend the non parametric test. For now its a majority vote, meaning that we recommend t-test if normally distributed variables are equal or more than 50%.
I think we should just give a warning if we choose one test (e.g. t-test), but the shapiro test shows that >=50% of metabolites is not made for the test (e.g. not normally distributed. Otherwise we just give a message() stating the percentage of metabolites falling into each distribution.