Alternative paradigms
- Includes approaches that go beyond or around classical assumptions of causality, such as interventions, counterfactuals, or acyclic graphs.
- These may handle feedbacks, unmeasured confounding, or time-varying treatment in complex systems.
- Useful when other assumptions (e.g., no unobserved confounding) are violated or untestable.
Overlap: Closely linked with Causal Discovery (when handling complex graphs), Ecology-Guided Modelling (when incorporating feedbacks), and Independent Detection (when causal claims are exploratory or limited)
Key feature: Loosens or shifts causal framework from the Potential Outcome and Structural Causal Model paradigms
Usage: For domain-driven causal reasoning, weakly identified systems, or model robustness