JASP equivalence testing non-normal data

BDF

I want to examine the equivalence of two language versions of a questionnaire. Inherently to the nature of the questionnaire, the data are non-normal. Can I still use the Baysian Equivalence testing for independent samples t-tests? Or should I stick with the t-test for independent samples / option Mann-Witney U test?

Thanks in advance,

Brechje

EJ

With the posterior distribution from the Mann-Whitney in hand, it should be possible to construct an equivalence test from that. But we have not yet implemented this. You could check whether it matters by executing the Mann-Whitney and comparing it to the standard t-test; if the results are relatively similar it seems safe to use the standard t-equivalence test.

Cheers,

E.J.