观察性研究中的因果推断方法二30分钟.ppt

Back-door path: A path that connects X to Y is a back-door path from X to Y if it has an arrowhead pointing to X. e.g. X ←U1→Y; X ←U2→Y. Front-door path: A path that connects X to Y is a front-door path from X to Y if it has an non-direct causal path arrowhead emanating from X. e.g. X → T → C → Y. Blocked path Unblocked path: A path is blocked if it has one or more colliders; otherwise it is unblocked. e.g. U1 → X ← U2. Conditioning: Examining the distribution of one variable within levels of another by regression adjustment, stratification, restriction(Subgroup analysis), or caused by Sample selection, Attrition, censoring, nonresponse. * Causal diagram (DAG)------ backdoor path frontdoor path Target path: X → Y Causal path: A directed path from X node to Y is one that can be traced through a sequence of single headed arrows, always entering an arrow through the tail and leaving through the head; X → T → C → Y. . X Y Causal diagram—— Causal Directed Acyclic Graphs (DAG) Causal diagram ------ causal paths, confounding paths and colliding paths. All DAGs can be constructed from just three elements—causal paths, confounding paths and colliding paths—the very elements that give rise to all associations via causation, confounding and collider variable. causal path confounding path colliding paths Sources of Association Between Two Variables A B Causal diagram—— Causal Directed Acyclic Graphs (DAG) Causal diagram ------ causal paths, confounding paths and colliding paths. causal path confounding path colliding paths Sources of Bias in Estimating the Causal Effect of A on B Causal diagram—— Causal Directed Acyclic Graphs (DAG) Let X, Y, Z, and W be arbitrary disjoint sets of nodes in a causal DAG G. : the graph obtained by deleting from G all arrows pointing to nodes in X. : the graph obtained by deleting from G all arrows emerging from nodes in X. : represent the deletion of both incoming and