how to return the treatment effects as log-odd

[ilovemane]

I'm working on a simulation study with binary outcomes and binary treatment and would like to estimate treatment effects on the log-odds scale (rather than risk difference).

I’ve been using LinearDML, but it appears to return effects in probability/risk difference scale. Is there any way to adapt LinearDML, or should I use DRLearner with a logistic final model to get valid causal effects on the log-odds scale? I really want to run the simulation study on dml.

Thank you so much — this is an amazing package and I appreciate your work on it!