Post hoc and effect size

[Isabela_Castro]

Hi everyone,

I’m an ecologist and after a while I was able to fit a good hierarquical model with brms.

I’ve seen several ways to do post hoc analysis with a brms fit. I used the emmeans package to do comparisons, which gave me reasonable results. However, I don’t feel confident enough to use it because I don’t know if this method is appropriate for Bayesian posterior.
I thought about using Cohen’s d, but I also don’t know if that is the best approach since I have so many groups (5). It would be a shame to use the wrong post hoc method for a really good and robust Bayesian fit.

Ps: I need to use something similar to effect size, which is more common in ecology.

Any thoughts on that?

[jsocolar]

Can you share more about your model and what inference you want to make with it?

[Isabela_Castro]

Sure!
I have 5 treatments (4 are with nutrient addition and one is the control area with nothing) where I collected plant traits. The model is:
trait ~ treatment * species + (1|site)

Basically, I want to know if the nutrient addition treatments differs from the control. I don’t need to compare treatments between them, only they with the control area.

Here I have the mean and upper and lower limits form the conditional effects

Here is the comparisons made with the emmeans package.