Can limits for Equivalence TOST use ratios instead of differences between the test and reference groups?

[#1]

Minitab allows the lower and upper equivalence limits for TOST to be set as differences (either balanced or unbalanced, e.g. -123 and +345) or as ratios (either balanced or unbalanced, e.g. 0.8 and 1.2 if you wanted an equivalence margin of +/- 20%).

I don't see this option for JMP and wondered if it was available and how to access it. 

Thanks.

[#2]

JMP does too, its just not listed called the same thing. 

Try using the distribution platform, and Test Equivalence. 

You will have to run the test for the upper and lower limit separately.

Make sure alpha is set correctly for you use case

[#3]

Thanks for the suggestion!

[#4]

The Distribution Platform => Test Equivalence is only applicable if the ratio of interest is derived from paired data. For example,

Then one can do equivalence test on the ratio using Distribution Platform => Test Equivalence, where ratio (A/B) is the Y.

I am not aware JMP 16 can do equivalence test for ratios directly if the ratio is derived from unpaired variables. There are other ways around it, eg. by log transformation, but it isn't always appropriate. If I am wrong, please let me know! Thanks!

[#5]

I wonder if you are asking about counts and rates. JMP 17 added equivalence tests for risk difference and relative risk based on rates derived from counts. That might suit your data. I am not sure.