Amit Sharma

The algorithm shuts down the backdoor path by permuting the treatment variable data. Since the treatment is randomized, it cannot be affected by the confounders. This paper on negative controls...

That's a good idea @MichaelMarien . Feel free to start a PR on the print statements. On the p-values, @Jorawar-Singh can you share a minimum working example to bring up...

@siddhya In your example, $E[charges|smoker=1]-E[charges|smoker=0]$ is already high. However, we cannot conclude anything causal since there might be other variables that explain this correlation. So then you considered three observed...

The intention here is to select a default value of strength of unobserved confounding. For continuous Y or T, the correlation of U with Y or T denotes strength of...

Okay, I will work on adding that parameter. Great question on the default flip threshold. The code assumes an additive error model for generation of T. For example, consider `t=[0.1w0...

Wow, thanks for these thoughtful notes, @EgorKraevTransferwise. We've discussed how to translate the `effect_modifiers` parameters to `X` in CATE estimators with the EconML team, and it is still an ongoing...

Really sorry for the late reply, somehow I missed the notification for this. You have a good point about not including all variables for the treatment propensity model. Based on...

That's a good point. I was thinking of `treatment_modifiers` as any variable that affects the treatment (and the user would like to include in their treatment model like a propensity...

yes, you can do that. Just add the variable `age` as both a common_cause and as a effect modifier.

Can you give a minimum working example where you receive this error?