Alternative Measures of Direct and Indirect Effects

Abstract

There are a number of measures of direct and indirect effects in the literature on causality. These are suitable in some cases and unsuitable in others. We describe a case where the existing measures are unsuitable and propose new suitable ones. We also show that the new measures can partially handle unmeasured treatment-outcome confounding, and bound long-term effects by combining experimental and observational data. We also introduce the concepts of indirect benefit and harm (i.e., through a mediator), and use our new measure to quantify them.

Cite

Text

Peña. "Alternative Measures of Direct and Indirect Effects." Proceedings of The 12th International Conference on Probabilistic Graphical Models, 2024.

Markdown

[Peña. "Alternative Measures of Direct and Indirect Effects." Proceedings of The 12th International Conference on Probabilistic Graphical Models, 2024.](https://mlanthology.org/pgm/2024/pena2024pgm-alternative/)

BibTeX

@inproceedings{pena2024pgm-alternative,
  title     = {{Alternative Measures of Direct and Indirect Effects}},
  author    = {Peña, Jose M.},
  booktitle = {Proceedings of The 12th International Conference on Probabilistic Graphical Models},
  year      = {2024},
  pages     = {1-19},
  volume    = {246},
  url       = {https://mlanthology.org/pgm/2024/pena2024pgm-alternative/}
}