Causal Identification with Interference

Adapt the backdoor criterion to find valid adjustment sets in a DAG with explicit interference pathways.

Derive the interventional distribution P(Y|do(T)) for the triadic system using do-calculus rules adapted for interference.

Determine the identifiability of direct treatment effects versus indirect spillover effects on each member's satisfaction.

Apply the front-door criterion to identify a mediated effect when a mechanism (e.g., sense of autonomy) is fully observed and isolated.

Assess identifiability when key confounders or mediators in the triadic model are unmeasured.

Show how Berkson's paradox can arise from conditioning on a collider, such as the triad's joint decision to remain in the study.