[ Français ]

Causal adjustment strategies usually rely on knowledge of the (presumed) DAG structure underlying the data generation. However, the assumption that the presumed DAG itself is correct is strong, and can been relaxed to allow for graphical structures with more uncertainty about directionality to be proposed -- for example, Complete Partially Directed Acyclic Graphs (CPDAGs) allow the direction of relationships between nodes in the graph to be unknown. Such structures have thus far not been studied in great detail with respect to their implications for statistical procedures. This theme will investigate these more general structures and how they alter the practitioner's approach to causal adjustment.