adjustment method for contrast(combine = TRUE)

[gevro]

Is it possible to specify the adjust method in contrast(combine=TRUE) for the combination across the emmeans list? The documentation says that rbind is used to execute combine = TRUE, but the adjust parameter of contrast appears to be ignored and even though I set it to "tukey" the result says it used "sidak". The cross.adjust parameter doesn't work either.

Strangely, when I set adjust = "mvt", the final result says that "mvt" was used. But I would like to use "tukey" both for the standard p-value adjustment, and for the combine=TRUE combination step.

Thanks

[rvlenth]

You have to use these functions the way they are documented. In this case,

• When you combine objects into one group, there are no longer any groups to cross-adjust. Hence anything you put in cross.adjust is immaterial.
• The Tukey adjustment is not allowed except for pairwise comparisons of one set of means. That's because for other families of tests, the statistical assumptions are incorrect. So if you ask for tukey, it gets changed to something that is allowed, namely sidak. This is documented in the FAQs vignette and elsewhere.