Hi @mjskay @nelsjohnson and everyone. I found emmeans interesting. It is great that it supports brms and rstanarm. I have three questions:
(1) How could I get the constrasts expressed as differences in probabilties comparing all levels to the baseline rather than odds ratio when I set type = "response"? This is assuming that the above model is logistic.
(2) Is there in emmeans a function doing what make_conditions and conditions do in brms's marginal_effects? [Update: a solution to this issue using tidybayes is discussed on Add vignette for calculating average marginal effects and/or average predictive comparisons · Issue #139 · mjskay/tidybayes · GitHub].
(3) Is it possible in emmeans to marginalize over rather than have the results ... averaged over the levels of: the other predictors.
I have opened an issue (How to marginalize rather than condition on variables to make the output of brms marginal_effects literal AME, MER, and MEM · Issue #552 · paul-buerkner/brms · GitHub) to address these questions following this discussion and wanted to know if emmeans has these functionalities implemented already to get the marginal effects proposed for brms.
Thank you in advance.