Extending IMs to causal inference and differential privacy

Develop inferential model formulations and theory for causal inference frameworks (e.g., potential outcomes, structural causal models) and for privacy-preserving inference under differential privacy, and establish corresponding validity guarantees.

Background

The paper suggests that IMs could contribute to modern domains such as causal inference and differential privacy, especially given their generalization beyond model-based settings.

The conclusion flags this as an area where technical details remain to be worked out, inviting methodological development and rigorous validation.