Annals of Statistics

Extended conditional independence and applications in causal inference

Abstract

The goal of this paper is to integrate the notions of stochastic conditional independence and variation conditional independence under a more general notion of extended conditional independence. We show that under appropriate assumptions the calculus that applies for the two cases separately (axioms of a separoid) still applies for the extended case. These results provide a rigorous basis for a wide range of statistical concepts, including ancillarity and sufficiency, and, in particular, the Decision Theoretic framework for statistical causality, which uses the language and calculus of conditional independence in order to express causal properties and make causal inferences.

Article information

Source
Ann. Statist., Volume 45, Number 6 (2017), 2618-2653.

Dates
Revised: December 2016
First available in Project Euclid: 15 December 2017

https://projecteuclid.org/euclid.aos/1513328585

Digital Object Identifier
doi:10.1214/16-AOS1537

Mathematical Reviews number (MathSciNet)
MR3737904

Zentralblatt MATH identifier
06838145

Subjects
Primary: 62A99: None of the above, but in this section
Secondary: 60A05: Axioms; other general questions

Citation

Constantinou, Panayiota; Dawid, A. Philip. Extended conditional independence and applications in causal inference. Ann. Statist. 45 (2017), no. 6, 2618--2653. doi:10.1214/16-AOS1537. https://projecteuclid.org/euclid.aos/1513328585

References

• [1] Barndorff-Nielsen, O. (1978). Information and Exponential Families in Statistical Theory. Wiley, Chichester.
• [2] Basu, D. (1964). Recovery of ancillary information. Sankhya A 26 3–16.
• [3] Berzuini, C. R., Dawid, A. P. and Didelez, V. (2012). Assessing dynamic treatment strategies. In Causality: Statistical Perspectives and Applications (C. R. Berzuini, A. P. Dawid and L. Bernardinelli, eds.) Chapter 8 85–100. Wiley, Chichester.
• [4] Billingsley, P. (1995). Probability and Measure, 3rd ed. Wiley, New York.
• [5] Borkar, V. S. (1995). Probability Theory: An Advanced Course, 1st ed. Springer, New York.
• [6] Casella, G. and Berger, R. L. (2001). Statistical Inference, 2nd ed. Duxbury Press.
• [7] Constantinou, P. (2013). Conditional independence and applications in statistical causality. PhD Dissertation, Univ. Cambridge.
• [8] Constantinou, P. and Dawid, A. P. (2017). Supplement to “Extended conditional independence and applications in causal inference.” DOI:10.1214/16-AOS1537SUPP.
• [9] Cowell, R. G., Dawid, A. P., Lauritzen, S. L. and Spiegelhalter, D. J. (2007). Probabilistic Networks and Expert Systems: Exact Computational Methods for Bayesian Networks, 2nd ed. Springer, Berlin.
• [10] Dawid, A. P. (1975). On the concepts of sufficiency and ancillarity in the presence of nuisance parameters. J. R. Stat. Soc., B 37 248–258.
• [11] Dawid, A. P. (1976). Properties of diagnostic data distributions. Biometrics 32 647–658.
• [12] Dawid, A. P. (1979). Conditional independence in statistical theory. J. R. Stat. Soc., B 41 1–31.
• [13] Dawid, A. P. (1979). Some misleading arguments involving conditional independence. J. R. Stat. Soc., B 41 249–252.
• [14] Dawid, A. P. (1980). A Bayesian look at nuisance parameters. In Bayesian Statistics (Valencia, 1979) (J. M. Bernardo, M. H. DeGroot, D. V. Lindley and A. F. M. Smith, eds.) 167–203. Univ. Press, Valencia. With discussion.
• [15] Dawid, A. P. (1980). Conditional independence for statistical operations. Ann. Statist. 8 598–617.
• [16] Dawid, A. P. (1985). Invariance and independence in multivariate distribution theory. J. Multivariate Anal. 17 304–315.
• [17] Dawid, A. P. (1998). Conditional independence. In Encyclopedia of Statistical Sciences, Vol. 2 (S. Kotz, C. B. Read and D. L. Banks, eds.) 146–155. Wiley-Interscience.
• [18] Dawid, A. P. (2001). Separoids: A mathematical framework for conditional independence and irrelevance. Ann. Math. Artif. Intell. 32 335–372.
• [19] Dawid, A. P. (2001). Some variations on variation independence. In Proceedings of Artificial Intelligence and Statistics 2001 (T. Jaakkola and T. Richardson, eds.) 187–191.
• [20] Dawid, A. P. (2002). Influence diagrams for causal modelling and inference. Int. Stat. Rev. 70 161–189.
• [21] Dawid, A. P. (2003). Causal inference using influence diagrams: The problem of partial compliance. In Highly Structured Stochastic Systems (P. J. Green, N. L. Hjort and S. Richardson, eds.). Oxford Statist. Sci. Ser. 27 45–81. Oxford Univ. Press, Oxford. With part A by Elja Arjas and part B by James M. Robins.
• [22] Dawid, A. P. (2012). The decision-theoretic approach to causal inference. In Causality: Statistical Perspectives and Applications (C. R. Berzuini, A. P. Dawid and L. Bernardinelli, eds.) Chapter 4 25–42. Wiley, Chichester.
• [23] Dawid, A. P. (2015). Statistical causality from a decision-theoretic perspective. Annual Review of Statistics and Its Application 2 273–303.
• [24] Dawid, A. P. and Constantinou, P. (2014). A formal treatment of sequential ignorability. Stat. Biosci. 6 166–188.
• [25] Dawid, A. P. and Dickey, J. M. (1977). Likelihood and Bayesian inference from selectively reported data. J. Amer. Statist. Assoc. 72 845–850.
• [26] Dawid, A. P. and Didelez, V. (2010). Identifying the consequences of dynamic treatment strategies: A decision-theoretic overview. Stat. Surv. 4 184–231.
• [27] Dawid, A. P. and Lauritzen, S. L. (1993). Hyper Markov laws in the statistical analysis of decomposable graphical models. Ann. Statist. 21 1272–1317. [Correction: Ann. Statist. 23 (1995) 1864.]
• [28] Dawid, A. P. and Mortera, J. (1996). Coherent analysis of forensic identification evidence. J. R. Stat. Soc. Ser. B. Stat. Methodol. 58 425–443.
• [29] Dawid, A. P. and Mortera, J. (1998). Forensic identification with imperfect evidence. Biometrika 85 835–849. [Correction: Biometrika 86 (1999) 974.]
• [30] Dawid, A. P. and Studený, M. (1999). Conditional products: An alternative approach to conditional independence. In Artificial Intelligence and Statistics 99 (D. Heckerman and J. Whittaker, eds.) 32–40. Morgan Kaufmann.
• [31] Didelez, V., Dawid, A. P. and Geneletti, S. (2006). Direct and indirect effects of sequential treatments. In Proceedings of the 22nd Annual Conference on Uncertainty in Artificial Intelligence (R. Dechter and T. S. Richardson, eds.) 138–146. AUAI Press, Arlington, VA.
• [32] Durrett, R. (2013). Probability: Theory and Examples, 4th ed. Cambridge Univ. Press.
• [33] Fisher, R. A. (1925). Theory of statistical estimation. Math. Proc. Cambridge Philos. Soc. 22 700–725.
• [34] Geiger, D., Verma, T. and Pearl, J. (1990). Identifying independence in Bayesian networks. Networks 20 507–534.
• [35] Geneletti, S. and Dawid, A. P. (2011). Defining and identifying the effect of treatment on the treated. In Causality in the Sciences (P. M. Illari, F. Russo and J. Williamson, eds.) 728–749.
• [36] Ghosh, M., Reid, N. and Fraser, D. A. S. (2010). Ancillary statistics: A review. Statist. Sinica 20 1309–1332.
• [37] Guo, H. and Dawid, A. P. (2010). Sufficient covariates and linear propensity analysis. In Proceedings of the Thirteenth International Workshop on Artificial Intelligence and Statistics (AISTATS) 2010 (Y. W. Teh and D. M. Titterington, eds.). Journal of Machine Learning Research Workshop and Conference Proceedings 9 281–288.
• [38] Guo, H., Dawid, A. P. and Berzuini, G. M. (2016). Sufficient covariate, propensity variable and doubly robust estimation. In Statistical Causal Inferences and Their Applications in Public Health Research. (H. He, P. Wu and D.-G. Chen, eds.). Springer. To appear.
• [39] Halmos, P. R. and Savage, L. J. (1949). Application of the Radon–Nikodym theorem to the theory of sufficient statistics. Ann. Math. Stat. 1 225–241.
• [40] Kingman, J. F. C. (1993). Poisson Processes, 2nd ed. Oxford Studies in Probability 3. Oxford Univ. Press, New York.
• [41] Kolmogoroff, A. (1942). Sur l’estimation statistique des paramètres de la loi de Gauss. Bull. Acad. Sci. URSS. Sér. Math. [Izvestia Akad. Nauk SSSR] 6 3–32.
• [42] Lauritzen, S. L., Dawid, A. P., Larsen, B. N. and Leimer, H.-G. (1990). Independence properties of directed Markov fields. Networks 20 491–505.
• [43] Pearl, J. (1988). Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference, 1st ed. Morgan Kaufmann, San Mateo, CA.
• [44] Pearl, J. (2000). Causality: Models, Reasoning, and Inference. Cambridge Univ. Press, Cambridge.
• [45] Pearl, J. and Paz, A. (1986). Graphoids: Graph-based logic for reasoning about relevance relations or when would $x$ tell you more about $y$ if you already know $z$? In Advances in Artificial Intelligence II, Seventh European Conference on Artificial Intelligence, ECAI 86 (J. B. H. du Boulay, D. Hogg and L. Steels, eds.) 357–363.
• [46] Resnick, S. I. (2014). A Probability Path, 1st ed. Birkhäuser.
• [47] Robins, J. (1986). A new approach to causal inference in mortality studies with a sustained exposure period—application to control of the healthy worker survivor effect. Math. Modelling 7 1393–1512.
• [48] Robins, J. (1987). Addendum to “A new approach to causal inference in mortality studies with sustained exposure periods—Application to control of the healthy worker survivor effect”. Comput. Math. Appl. 14 923–945.
• [49] Robins, J. (1989). The analysis of randomized and nonrandomized AIDS treatment trials using a new approach to causal inference in longitudinal studies. In Health Service Research Methodology: A Focus on AIDS (L. Sechrest, H. Freeman and A. Mulley, eds.) 113–159.
• [50] Spohn, W. (1994). On the properties of conditional independence. In Patrick Suppes: Scientific Philosopher (P. Humphreys, ed.). Synthese Lib. 233 173–196. Kluwer Acad. Publ., Dordrecht.

Supplemental materials

• Some Proofs. Supplementary material, comprising proofs of Lemma 2.2, Theorem 2.4, Proposition 2.5, Proposition 2.6, Theorem 2.7, Proposition 3.1, Theorem 4.2 and Theorem 4.3, is available online.