Estimated Marignal Means of MMRM models

Generic method to run emmeans::emmeans on all MMRM package objects using Kenward-Rogers degrees of freedom calculation. Please see emmeans::emmeans for further documentation.

Usage

mmrm_emmeans(
  obj,
  specs = NULL,
  group = NULL,
  cov.reduce = FALSE,
  nesting = NULL,
  ...
)

Arguments

obj: a fitted MMRM model
specs: A character vector specifying the names of the predictors over which EMMs are desired. See emmeans::emmeans for more details. If omitted, group must be specified, and emmeans and contrasts will be calculated across groups at each time point.
group: the group over which to calculate emmeans and contrasts. If specs is not specified, group must be specified.
cov.reduce: logical indicating whether to average over a continuous variable (usually time points) when calculating emmeans and contrasts. By default, FALSE

Value

see emmeans::emmeans

Details

Support for Kenward-Rogers degrees of freedom calculation is made possible with i) a custom emm_basis function and ii) addition of support for glsObject in the pbkrtest package (currently in a development version gkane26/pbkrtest@nlme).

Kenward-Rogers d.f. calculation is memory and computation time intensive. By default, running this method on a large dataset will throw an error to warn users. To bypass this error, please set the pbkrtest.limit argument to a number larger than the number of observations in the model.

Examples

if (FALSE) {
mmrm_emmeans(
  object = mmrm_object,
  specs = pairwise ~ time | treatment,
  mode = "kenward",
  pbkrtest.limit = emmeans::get_emm_option("pbkrtest.limit")
)
}