The Annals of Statistics

Geometry of the faithfulness assumption in causal inference

Abstract

Many algorithms for inferring causality rely heavily on the faithfulness assumption. The main justification for imposing this assumption is that the set of unfaithful distributions has Lebesgue measure zero, since it can be seen as a collection of hypersurfaces in a hypercube. However, due to sampling error the faithfulness condition alone is not sufficient for statistical estimation, and strong-faithfulness has been proposed and assumed to achieve uniform or high-dimensional consistency. In contrast to the plain faithfulness assumption, the set of distributions that is not strong-faithful has nonzero Lebesgue measure and in fact, can be surprisingly large as we show in this paper. We study the strong-faithfulness condition from a geometric and combinatorial point of view and give upper and lower bounds on the Lebesgue measure of strong-faithful distributions for various classes of directed acyclic graphs. Our results imply fundamental limitations for the PC-algorithm and potentially also for other algorithms based on partial correlation testing in the Gaussian case.

Article information

Source
Ann. Statist., Volume 41, Number 2 (2013), 436-463.

Dates
First available in Project Euclid: 16 April 2013

https://projecteuclid.org/euclid.aos/1366138197

Digital Object Identifier
doi:10.1214/12-AOS1080

Mathematical Reviews number (MathSciNet)
MR3099109

Zentralblatt MATH identifier
1267.62068

Citation

Uhler, Caroline; Raskutti, Garvesh; Bühlmann, Peter; Yu, Bin. Geometry of the faithfulness assumption in causal inference. Ann. Statist. 41 (2013), no. 2, 436--463. doi:10.1214/12-AOS1080. https://projecteuclid.org/euclid.aos/1366138197

References

• [1] A. Becker, D. G. and Meek, C. (2000). Perfect tree-like Markovian distributions. In Proceedings of the 16th Conference on Uncertainty in Artificial Intelligence 19–23. Morgan Kaufmann, San Francisco, CA.
• [2] Adler, R. J. and Taylor, J. E. (2007). Random Fields and Geometry. Springer, New York.
• [3] Chickering, D. M. (2003). Optimal structure identification with greedy search: Computational learning theory. J. Mach. Learn. Res. 3 507–554.
• [4] Guth, L. (2009). Minimax problems related to cup powers and Steenrod squares. Geom. Funct. Anal. 18 1917–1987.
• [5] Kalisch, M. and Bühlmann, P. (2007). Estimating high-dimensional directed acyclic graphs with the PC-algorithm. J. Mach. Learn. Res. 8 613–636.
• [6] Kalisch, M., Mächler, M., Colombo, D., Maathuis, M. H. and Bühlmann, P. (2011). Causal inference using graphical models with the R package pcalg. Journal of Statistical Software 47 1–26.
• [7] Lin, S., Uhler, C., Sturmfels, B. and Bühlmann, P. (2012). Hypersurfaces and their singularities in partial correlation testing. Unpublished manuscript. Available at arXiv:1209.0285.
• [8] Maathuis, M. H., Kalisch, M. and Bühlmann, P. (2009). Estimating high-dimensional intervention effects from observational data. Ann. Statist. 37 3133–3164.
• [9] Pearl, J. (2000). Causality: Models, Reasoning, and Inference. Cambridge Univ. Press, Cambridge.
• [10] Ponstein, J. (1966). Self-avoiding paths and the adjacency matrix of a graph. SIAM J. Appl. Math. 14 600–609.
• [11] Robins, J. M., Scheines, R., Spirtes, P. and Wasserman, L. (2003). Uniform consistency in causal inference. Biometrika 90 491–515.
• [12] Spirtes, P., Glymour, C. and Scheines, R. (2001). Causation, Prediction and Search, 2nd ed. MIT Press, Cambridge.
• [13] Spirtes, P. and Zhang, J. (2012). A uniformly consistent estimator of causal effects under the $k$-triangle-faithfulness assumption. Unpublished manuscript.
• [14] Sullivant, S., Talaska, K. and Draisma, J. (2010). Trek separation for Gaussian graphical models. Ann. Statist. 38 1665–1685.
• [15] van de Geer, S. and Bühlmann, P. (2013). $\ell_0$-penalized maximum likelihood for sparse directed acyclic graphs. Ann. Statist. To appear.
• [16] Zhang, J. and Spirtes, P. (2003). Strong faithfulness and uniform consistency in causal inference. In Proceedings of the 19th Conference on Uncertainty in Artificial Intelligence 632–639. Morgan Kaufmann, San Francisco, CA.
• [17] Zhang, J. and Spirtes, P. (2008). Detection of unfaithfulness and robust causal inference. Minds and Machines 18 239–271.