I'm new to non-parametric statistics and want to do a post hoc analysis of some data after a Kruskal-Wallis test. I am using the implementation of the Conover-Iman test available in the scikit-posthocs package in Python.

Inspecting the source code the test seems to be employing a two-tailed t-test (in the compare_conover method). Hence, my question is, if I anticipate a specific difference between two groups of data, should I divide the corresponding output of the Conover-Iman test by 2 to compensate for the t-test being two-tailed?


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.