Relationship of causal effects in a causal chain and related inference (Q2386545)

Criteria of identifiability of causal effects in causal chains are discussed for discrete variables. Using Stone's potential causal functions the authors develop new such criteria which can be considered as extensions of the Pearl's front-door criterion. Their advantage is that they only assume the set of risk factors to the response, without assuming a completely constructed causal diagram. The developed technique can be applied to eliminate the confounding bias through controlling for intermediate variable.