Open bwiernik opened 2 years ago
Hmmm but we don't compute that at any point because we estimate CIs with the ncp method. Would this be easy to implement for all 7 effect sizes ({pooled, non pooled, paired} * {d, g} + delta)?
Yeah, I can put together some formulas
bump
A good starting point: @MatthewBJane's https://matthewbjane.quarto.pub/effect-size-and-confidence-intervals-guide/
Saved as an attribute, extractable with parameters::standard_error()
.
It would be nice to include an option to add the SE or variance in
cohen_d()
, etc. to facilitate use of effectsize with meta-analyses.