Causation is a distinct concept from association and more important than association inepidemiologic studies.This paper proposes the concept of uniform non-confounding for causal distributioneffects over multiple cova...Causation is a distinct concept from association and more important than association inepidemiologic studies.This paper proposes the concept of uniform non-confounding for causal distributioneffects over multiple covariates,and gives the sufficient conditions for uniform non-confoundingover a covariate set C including confounders or non-confounders,and also shows the conditions forconditionally non-confounding in the subpopulations.All these conditions can be tested by observeddata.展开更多
基金supported by the Natural Science Foundation China under Grant Nos. 10801019 and 10726037
文摘Causation is a distinct concept from association and more important than association inepidemiologic studies.This paper proposes the concept of uniform non-confounding for causal distributioneffects over multiple covariates,and gives the sufficient conditions for uniform non-confoundingover a covariate set C including confounders or non-confounders,and also shows the conditions forconditionally non-confounding in the subpopulations.All these conditions can be tested by observeddata.
