Causal Inference and Diagrams

Causal inference is the science of attributing a particular outcome (or effect) to one or more particular causes. In addition to concluding that there is an association between two variables, causal inference implies that the effect is the direct result of a measurable cause. In medical research, the cause is often an intervention or treatment, and the outcome is often a disease or complication. Outcomes from those receiving the intervention, perhaps a particular drug, are often compared with those of a control group. When the difference in outcomes between the experimental and control groups is attributed to the intervention, causal inference is being made.

Causal inference is made most cleanly in a randomized, blinded study. However, even in a non-randomized setting, some degree of qualified causal inference may be possible. This depends on the extent of thorough understanding of the relationships involved, careful design, and data collection and analysis. Causal inference relationships can be visualized and clarified using causal diagrams—modern tools that use arrows to visualize the purported relationships between causal variables, outcome variables, and confounding variables in both randomized and nonrandomized studies.