JMA Journal

Table 1. Observed Data in a Hypothetical Cohort Study.

Figure 1. A causal diagram of an exposure E and an outcome Y.

Table 2. Underlying Hypothetical Data of the Hypothetical Cohort Study in Terms of Response Types^a.

Figure 2. Causation and association in the population of interest. Causation is defined by a contrast of risks in the entire population under two potential exposure values, whereas association is defined by a contrast of risks in subsets of the population determined by the subjects’ actual exposure value. Adapted from Hernán and Robins ⁽¹⁹⁾ with permission from the authors.

Table 3. Necessary and Sufficient Conditions for No Confounding.

Figure 3. Three types of sufficient causes for Y. I consider a binary exposure E and a binary outcome Y.

Table 4. Underlying Hypothetical Data of the Hypothetical Cohort Study in Terms of Risk Status Types^a.

Figure 4. Logical flows of the counterfactual model and the sufficient cause model.

Table 5. Correspondence between Response Types and Risk Status Types under a Binary Exposure and a Binary Outcome^a.

Table 6. The Relationship between Exchangeability and Covariate Balance for No Confounding^a.

Table 7. Correspondence between Response Types, Risk Status Types, and Sequence Types under a Binary Exposure and a Binary Outcome^a.

Figure 5. A causal diagram of an exposure E, a mediator M, and an outcome Y.

Figure 6. Three types of sufficient causes for M. I consider a binary exposure E and a binary mediator M. Under the assumption of sufficient cause positive monotonicity of E, I consider only A₁ and A₂E.

Figure 7. Nine types of sufficient causes for Y. I consider a binary exposure E, a binary mediator M, and a binary outcome Y. Under the assumption of sufficient cause positive monotonicity of E and M, I consider only B₁, B₂E, B₃M, and B₆EM.

Table 8. Correspondence between Potential Outcomes and Sufficient Causes under the Assumption of Sufficient Cause Positive Monotonicity in the Context of Mediation^a.

Figure 8. Mediation and mechanism in a causal diagram with a sufficient causation structure. I consider a binary exposure E, a binary mediator M, and a binary outcome Y, under the assumption of sufficient cause positive monotonicity of E and M. See Figures 6 and 7.

Table 9. The Nine Bradford Hill Viewpoints.