Brunner-Munzel test in moretests package?

Everything related to the development of modules in jamovi

by UWde » Tue Feb 16, 2021 10:26 am

Hi guys,

the Brunner-Munzel test (described as a generalized Wilcoxon Test) and the Fligner-Policello test (as a modified Mann-Whitney test) are non-parametric alternatives to the widely used MW test for situations in which the assumptions underlying the Mann-Whitney test are not met. While the MW test assumes equal variances and a comparable shape of the distributions in both samples, the Fligner-Policello test does not require equal variances and only assumes the distributions in both samples to be symmetric. The Brunner-Munzel allows for unequal variances and is also less strict regarding the distributions of the samples.

Both are available as R functions (brunner.munzel.test from the brunnermunzel package and fp.test from the npsm package). However, they are not widely implemented in user-friendly statistical packages. Thus, implementing them in Jamovi might help to further set Jamovi apart from other statistical packages. They would possibly best fit into the moretests module, which in turn might become more comprehensive by addition of these two tests.

I would guess that implementation should be quite straightforward for any experienced developers, as the R syntax of both tests looks quite similar to their non-parametric cousin, the Mann-Whitney test. Unfortunately, I have no experience at all with Jamovi development and I lack the time required to become adequately acquainted with it. So I can only present this as a friendly proposal to the developers, specifically those of the moretests package :).

Thanks for your attention & all the best,

Posts: 1
Joined: Tue Feb 16, 2021 9:45 am

Return to Module development