I have a question about emmeans and mixed effect model. I fit a complex model using lmer() with the following variables:
A: a binary categorical predictor, between-subject
B: a binary categorical predictor, within-subject
C: a categorical predictor with 3 levels, within-subject
The model is as follows:
library(lme4) model= lmer(d ~ A * B * C + (1|subject), data = ddata, REML = FALSE)
The results show a significant interaction between the predictors. I need to know the pairwise comparisons so I use emmeans as follow:
library(emmeans) emmeans(model, pairwise ~ A * B * C)
emmeans use the Tukey method for the pairwise comparisons. A reviewer of my paper does not like Tukey method as she said:
"Tukey’s HSD assumes all measures are independent from each other and is not appropriate for repeated measures. Please consider performing a different test even if that changes the results."
My questions are:
Is the claim that Tukey in this case is not appropriate? if no can you cite any source?
if yes what is the alternative method?