rvlenth/emmeans

standardization of slopes

samyogita-hardikar opened this issue · 2 comments

I recently reported emtrends slopes in a paper, and was asked to provide standardized estimates.
If I want to be quick, does it make sense to standardize them in the same way I would an "estimate" from linear model output?
standardized slope = unstandardized slope * stdev(predictor) / stdev(outcome)?
Thank you in advance!

That seems like one possibility. Another is that a slope cand be regarded as an "effect", in which case maybe they want an effect size, which you could get from the eff_size() function, standardizing just by the response SD.. Maybe it's best to ask them to clarify; or give your answer and say it's not completely clear that is what was requested.

Makes sense. Thank you!