emmeans for scale in location-scale meta-analysis

[FBartos]

Classification:

Enhancement Suggestion

Summary

Enable emmeans prediction of scale for location scale meta-analytic models. Unless I missed an argument in the documentation, it does not seam to be possible to use emmprep -> emmeans to produce predicted scale across different levels etc via rma.ls model. Adding a possible argument to emmprep, e.g., what = c("location", "scale"") would be a possible solution.

Notes

I saw that I can use the predict function with the newscale argument. It can fulfil all tasks but the emmeans interface is much friendlier.

Cheers,
Frantisek

[wviechtb]

Suggestion noted. Not entirely trivial though because with location-scale models, even predict() has to do some extra work to distinguish various cases (only newmods specified, newmods and newscale both specified, only newscale specified) and the behavior of the function changes accordingly.