GAMj mixed model - post-hoc tests

Everything related to the development of modules in jamovi
Post Reply
lauvos
Posts: 1
Joined: Fri Jul 01, 2022 6:41 am

GAMj mixed model - post-hoc tests

Post by lauvos »

Hi!
Thanks for developing jamovi! At a first glance it looks very logical and intuitive! I am looking forward to using it in a course I will be teaching soon.

Here's my question. In GAMj > Mixed Model, I was wondering if it is possible to only do a selected number of post-hoc comparisons. In R I would use the emmeans function like so:

Code: Select all

m2 <- lmer(Plaque ~ Treatment*Slice + (1|Mouse), data=mouse.data)
emmeans(m2, pairwise ~ Treatment | Slice, adjust = "Bonferroni")
which outputs

Slice = 1:
contrast estimate SE df t.ratio p.value
PBS - BRI 4.420 1.19 30 3.721 0.0008

Slice = 2:
etc.


The output from Jamovi is similar to running

Code: Select all

emmeans(m2, pairwise ~ Treatment:Slice, adjust = "Bonferroni")
which outputs

$contrasts
contrast estimate SE df t.ratio p.value
PBS 1 - BRI 1 4.420 1.19 30 3.721 0.0368
PBS 1 - PBS 2 8.968 1.19 24 7.550 <.0001
PBS 1 - BRI 2 8.808 1.19 30 7.415 <.0001
PBS 1 - PBS 3 9.190 1.19 24 7.736 <.0001
PBS 1 - BRI 3 8.691 1.19 30 7.316 <.0001
PBS 1 - PBS 4 6.285 1.19 24 5.291 0.0009
PBS 1 - BRI 4 6.093 1.19 30 5.129 0.0007
PBS 1 - PBS 5 6.671 1.19 24 5.616 0.0004
PBS 1 - BRI 5 9.628 1.19 30 8.105 <.0001
BRI 1 - PBS 2 4.548 1.19 30 3.829 0.0274
BRI 1 - BRI 2 4.388 1.19 24 3.694 0.0511
BRI 1 - PBS 3 4.770 1.19 30 4.016 0.0165
BRI 1 - BRI 3 4.271 1.19 24 3.596 0.0654
BRI 1 - PBS 4 1.865 1.19 30 1.570 1.0000
BRI 1 - BRI 4 1.673 1.19 24 1.408 1.0000
BRI 1 - PBS 5 2.251 1.19 30 1.895 1.0000
BRI 1 - BRI 5 5.208 1.19 24 4.385 0.0089

PBS 2 - BRI 2 -0.160 1.19 30 -0.135 1.0000
etc.

Since the red comparisons are irrelevant, I would like these to be left out for the post-hoc analysis and also not correct of these with the Bonferroni correction. Is this currently possible in jamovi?

Many thanks,
Laura
User avatar
mcfanda@gmail.com
Posts: 457
Joined: Thu Mar 23, 2017 9:24 pm

Re: GAMj mixed model - post-hoc tests

Post by mcfanda@gmail.com »

Hi
that is not possible at the moment. With a GUI, it is not simple to select just a few posthoc tests. We can think about something for future releases
Post Reply