Use GRF estimate average treatment effects when treatment variable is continuous

[gnay032100-blip]

Dear Authors,

Hello! I would like to ask whether the Conditional Average Treatment Effect (CATE) we estimate can be interpreted as the change in the outcome variable for each one-unit increase in the treatment variable, given the covariates, when the treatment variable is continuous.

Thank you for answering my question！

[erikcs]

Hi @gnay032100-blip, Sure, it's estimating a (conditional) average partial effect. From the docstring:

When W is continuous, we effectively estimate an average partial effect Cov[Y, W | X = x] / Var[W | X = x], and interpret it as a treatment effect given unconfoundedness.

[gnay032100-blip]

Hi @gnay032100-blip, Sure, it's estimating a (conditional) average partial effect. From the docstring:

When W is continuous, we effectively estimate an average partial effect Cov[Y, W | X = x] / Var[W | X = x], and interpret it as a treatment effect given unconfoundedness.

Thank you very much, it really helps！ @erikcs