I’m trying to obtain marginal means from a model fitted with brms. I know I can do so easily using the emmeans package. But I can also do this by getting the draws manually, and then calculate the averages. The problem is that I’m getting similar means, but rather different variances in the posterio…

I think the difference is that emmeans is giving you the result with random effects coefficients zeroed out, so those are the means conditional on the “typical” car. You should get the same thing if you do this: emm3 <- data_grid(mtcars2, vs) %>% add_fitted_draws(m1, re_formula = ~ 0) The other…

Marginal means using emmeans or manual calculation through add_fitted_draws don't agree

Interfaces brms

bgoodri November 17, 2018, 9:17pm 2

If you search for emmeans in the search magnifying glass, there are some other threads that have worked through similar problems.

Topic		Replies	Views
Trouble with {brms}/{emmeans} integration brms	0	288	April 1, 2024
Computation of contrasts using emmeans based on brms input in R brms emmeans	1	384	February 3, 2024
Integrating emmeans with brms brms emmeans	8	2353	October 27, 2020
A parameterization-agnostic way to get posterior samples of marginal means in {brms} Interfaces brms	1	416	June 2, 2023
What grid does passing a brms model to emmeans use when a predictor grid is not provided? brms posterior-predictive , brms , emmeans	3	1070	July 16, 2022

Marginal means using emmeans or manual calculation through add_fitted_draws don't agree

Related topics