In the description of the challenges you write "the treatment and outcome may depend on shared confounding features". I took that to mean that ignorability is satisfied in all the datasets. That is $$\(Y(0), Y(1) \perp W \mid X\)$$ where X are the supplied covariates, W is the binary treatment indicator, and Y(0) and Y(1) are the corresponding potential outcomes. Is that a correct interpretation of your statement?
Created by Jennifer Hill jlhillny Hi Jennifer,
Yes, as mentioned in the data description there are no hidden confounders, and each counter-factual outcome has a model that is based on the data.
As a results, given the data X, each counter-factual outcome is independent of the treatment assignment.