Fig. 1From: The alarming problems of confounding equivalence using logistic regression models in the perspective of causal diagramsFour typical causal diagrams with various confounding paths from simple to complex for the target causal path X→Y. a contains only one confounding path (X←Z→T→Y). b contains two confounding paths (X←Z→T→Y, X←W→Y). Two confounding paths (X←Z→T→Y, X←W→V→Y) that have another node (V) are included in (c). d has three confounding paths (X←W→Y, X←Z→W→Y and X←W←T→Y). X and Y indicates exposure and outcome respectively. T, Z, W and V are all confounders that can be observed. {c 0, c 1, c 2, c 3, c 4, c 5} are the effect parameters. For example, the effect of Z on T is c 0 Back to article page