Previously, the reference grid that is used to compute the estimated marginal means grew exponentially with the number of predictors in the model. As a consequence, the reference grid could grow so big that the emmeans analysis threw an error.
This commit marks all predictors that are not of interest for the emmeans plot/table as so-called "nuisance parameters", so that the reference grid averages over all these parameters and thus remains small in size, even when the number of parameters grow.
Previously, the reference grid that is used to compute the estimated marginal means grew exponentially with the number of predictors in the model. As a consequence, the reference grid could grow so big that the emmeans analysis threw an error.
This commit marks all predictors that are not of interest for the emmeans plot/table as so-called "nuisance parameters", so that the reference grid averages over all these parameters and thus remains small in size, even when the number of parameters grow.
See https://cran.r-project.org/web/packages/emmeans/vignettes/messy-data.html
Closes jamovi/jamovi#1259