DML Binary Outcome and binary exposure and return treatment effect as logOdd

[ilovemane]

Hi,

I am doing a simulation study related to healthcare data, where I have binary outcome and binary/multiple discrete group exposure. I am trying to implement DML to estimate the bias to the true treatment effect. But my treatment effect is in the form of logodd of the treatment. I wonder how I can use DML to return the treatment effect in logOdd form instead of probability differences.

thanks,

Mike

[SvenKlaassen]

Hi @ilovemane,

sorry for the late reply. Currently DoubleML does not estimate effects in logg odd form, but there will be an implmentation of a logit model in the future.

Best,
Sven

[ilovemane]

Hi @SvenKlaassen,

Thanks for reply.