Dice Question Streamline Icon: https://streamlinehq.com

Combining causal discovery and effect estimation across populations with covariate mismatch

Develop methods to jointly perform causal discovery and causal effect estimation across multiple populations in data fusion settings, including cases with covariate mismatch and differing variable sets, to enable effective transport and integration of causal knowledge.

Information Square Streamline Icon: https://streamlinehq.com

Background

The paper highlights challenges in fusing information from multiple studies and contexts when causal structure is not fully known and variable sets overlap only partially. While causal data fusion and joint causal inference frameworks provide partial tools, end-to-end solutions that combine discovery and estimation across populations with covariate mismatch are lacking.

The authors explicitly call out the need to integrate causal discovery and effect estimation effectively in these settings, noting this remains unresolved.

References

Another open question is how to combine causal discovery and causal effect estimation across populations. Recent work focused on observational data and experimental data, but how to do this effectively across different populations and with covariate mismatch remains an open question.

— Challenges in Statistics: A Dozen Challenges in Causality and Causal Inference (2508.17099 - Cinelli et al., 23 Aug 2025) in Section: Aggregation and Synthesis of Causal Knowledge — Generalizing effect estimates when (some) causal relations are unknown