Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It only takes a minute to sign up.
Sign up to join this community
In the 'pairs' function when doing pairwise comparisons after emmeans, tukey is set as default for adjustment of p-values. But what type of tukey is used, is it tukey-kramer? How can I know this? If its not, is it possible to change to that? To account for unbalanced data.
dflm1 <- lmer(formula= df[,i] ~ color +
(1|id), data=df)
emm <- emmeans(dflm1,"color")
emm_df <- data.frame(emm)
e <- pairs(emm, adjust="tukey")
# tukey-kramer? Or how to change to that?
This means it obtains P values from the pstudent() function in R. For unbalanced data (or in general, for unequal SEs or non-spherical correlation structure), this is only an approximate correction for multiplicity.
In the case of a one-way layout with independent samples, that is indeed the Tukey-Kramer method.
For mixed models, factorial experiments, etc., it depends on what you mean by "Tukey-Kramer." If by that you mean using the formulas that apply to a oneway layout, it isn't doing that because the SEs are obtained from the model, which isn't the standard one-way model. But if you interpret "Tukey-Kramer" to mean using the Tukey adjustment method in a case where it doesn't exactly apply, then it is doing that.
