# How ti specify contrasts for lmer model

I am using lmer in R to check the effect of treatment and a group on some result (y). In my data (dat), there are 3 categorical variables and a continuous response variable: 1 id 1 treatment variable with 3 categories: ctrl, treat1, treat2 1 group variable with categories a, b, c And a continuous variable y.

I fit a lmer model:

library(lmer) library(emmeans)

Mod=lmer(y ~ Treatment*group +(1 | id), data=dat).

At first, i want to compare the change from control for the treat1, and treat2. (treat1 vs ctrl, treat2 vs ctrl). I know how to do this using trt.vs.ctrl contrasts in the emmeans package.

emmeans(Mod, trt.vs.ctrl~treat, data=dat)

At second, for each treatment (treat1 and treat2) i want to compare the change from control between the groups a and b, b and c, a and c.

For this latter, i want to set the contrasts in R with the emmeans package but don't know how to do it.

I would be grateful to anyone who will help me to find a way to solve this question. Thanks you in advance

To keep things straight, I suggest doing the contrasts and EMMs in separate steps.

emm <- emmeans(Mod, ~ Treatment * group)
con <- contrast(emm, "trt.vs.ctrl", by = "group")
con   # show the contrasts for each group
contrast(con, "pairwise", by = "contrast")


You can do the contrasts in one step, but it's a little more confusing:

contrast(emm, interaction = c("trt.vs.ctrl", "pairwise"))


For more information and examples, see vignette("interactions", "emmeans")

• Thank you a lot! Your answer helped me a lot Jan 14, 2021 at 23:06