Some interaction models, mainly between two categorical moderators, may fail when metafor drops 'redundant predictors'. This creates a disconnect between data and model object such that emmeans cannot create a reference grid. Should add more checks in for this.
For now, a very simple solution, is to create a new categorical moderator that interacts the two moderators of interest and fit this new model. Keeping as character string will mean that factor levels don't cause issues.
Some interaction models, mainly between two categorical moderators, may fail when
metafor
drops 'redundant predictors'. This creates a disconnect between data and model object such thatemmeans
cannot create a reference grid. Should add more checks in for this.For now, a very simple solution, is to create a new categorical moderator that interacts the two moderators of interest and fit this new model. Keeping as character string will mean that factor levels don't cause issues.