Let’s say you have a basic mediation model, but your mediator M you wish to model with a Gaussian family and your outcome Y with a cumulative probit family. Something like this in brms: bf_med <- bf(M ~ X) + gaussian() bf_out <- bf(Y ~ M + X) + cumulative(probit) m1 <- brm(bf_med + bf_out + set_res…

I could be mistaken, but I’m pretty sure this issue only arises if the mediator is not Gaussian. So I don’t think you even need to rescale the mediator to be standard normal. To put it another way, the mediated pathway from X to the latent scale of Y is still linearly from X to M and then from M to…

Mediation when using Gaussian and Cumulative probit models

Modeling

andrjohns June 16, 2022, 7:16pm 7

A similar question came up previously and I put together an answer for it here, hope it helps!

4 Likes

Topic		Replies	Views
Mediation question using brms brms	11	3569	February 12, 2024
BRMS longitudinal mediation analysis brms rstan , panel-data	4	261	February 13, 2024
Serial medition with 2 mediators using brms brms rstan , specification	11	1413	July 22, 2020
Brms: Mediation analysis or not? brms	6	1497	December 7, 2019
BRMS for PCA and latent variable modelling to sequential mediation modelling? brms techniques , specification , brms	1	237	January 4, 2024

Mediation when using Gaussian and Cumulative probit models

Related Topics