I am using the equivalence::tost method in R. It takes a parameter epsilon, but I am unsure of how to calculate this when I require a test for non-zero Difference%. Currently, I do this by using the formula = Mean(Control) * Difference% which works for the Python ttost method, but since we're moving our stats to R, this is an issue.

Looking at the code for tost method, the epsilon is compared against the confidence interval, but it seems incorrect with the above formula.

Can someone please help figure out a formula for epsilon based on difference% from Control?

Update: Seems I'm asking the wrong question. My data only contains positive values. Looking at the code for equivalence::tost method, epsilon and -epsilon are compared against the confidence interval. But the confidence interval only contains positive values, so the -epsilon doesn't make much sense.

I am wondering what epsilon is looking for. The documentation says it is referring to the region of similarity, but not much more help...


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.