CoChat
Sign Up
Cited by 621Advanced Causal Inference TechniquesStatistical Methods and InferenceBayesian Modeling and Causal Inference

Causal inference and the data-fusion problem

Open access
Bareinboim, Elias, Pearl, Judea·2016·Proceedings of the National Academy of Sciences
DOI 10.1073/pnas.1510507113PMC PMC4941504
Open this paper in CoChat — free accountOpen PDF

Abstract

We review concepts, principles, and tools that unify current approaches to causal analysis and attend to new challenges presented by big data. In particular, we address the problem of data fusion—piecing together multiple datasets collected under heterogeneous conditions (i.e., different populations, regimes, and sampling methods) to obtain valid answers to queries of interest. The availability of multiple heterogeneous datasets presents new opportunities to big data analysts, because the knowledge that can be acquired from combined data would not be possible from any individual source alone. However, the biases that emerge in heterogeneous environments require new analytical tools. Some of these biases, including confounding, sampling selection, and cross-population biases, have been addressed in isolation, largely in restricted parametric models. We here present a general, nonparametric framework for handling these biases and, ultimately, a theoretical solution to the problem of data fusion in causal inference tasks.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1073/pnas.1510507113

Cites

12 of 12 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • A causal model of the data generating process can be used to identify regret-bounded policies following a distributional shift (sufficiency), with more accurate models allowing lower regret policies to be found.

    “It is known that a causal model of the data generating process can be used to identify regret-bounded policies following a distributional shift (sufficiency), with more accurate models allowing lower regret policies to be found.”

    From Robust agents learn causal world models (2024) · cites this paper as Bareinboim & Pearl, 2016

    The paper’s “Summary Result 1 (Identification in Policy Evaluation)” describes identifying policy questions, and “Transportability and the Problem of Data Fusion” discusses estimating causal effects in populations with different distributions. These passages support using causal models to evaluate interventions after a distributional shift, but they do not establish regret bounds for policies or say that more accurate models yield lower-regret policies. Page numbers were not available in our copy. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 42

  1. Pearl J, Glymour M, Jewell NP (2016) Causal Inference in Statistics: A Primer (Wiley, New York).
  2. Greenland S, Lash T (2008) Bias Analysis in Modern Epidemiology, eds Rothman K, Greenland S, Lash T (Lippincott Williams & Wilkins, Philadelphia), 3rd Ed, pp 345-380.
  3. Halpern J (1998) Axiomatizing causal reasoning. Uncertainty in Artificial Intelligence, eds Cooper G, Moral S (Morgan Kaufmann, San Francisco), pp 202-210.
  4. Balke A, Pearl J (1995) Counterfactuals and policy analysis in structural models. Uncertainty in Artificial Intelligence 11, eds Besnard P, Hanks S (Morgan Kaufmann, San Francisco), pp 11-18.
  5. Pearl J (1988) Probabilistic Reasoning in Intelligent Systems (Morgan Kaufmann, San Mateo, CA).
  6. Rosenbaum P, Rubin D (1983) The central role of propensity score in observational studies for causal effects. Biometrika 70(1):41-55.
  7. Hotz VJ, Imbens G, Mortimer JH (2005) Predicting the efficacy of future train- ing programs using past experiences at other locations. J Econom 125(1-2): 241-270.
  8. Spirtes P, Glymour C, Scheines R (2000) Causation, Prediction, and Search (MIT Press, Cambridge, MA), 2nd Ed.
  9. Huang Y, Valtorta M (2006) Pearl's calculus of intervention is complete. Proceedings of the Twenty-Second Conference on Uncertainty in Artificial Intelligence, eds
  10. Dechter R, Richardson T (AUAI Press, Corvallis, OR), pp 217-224.
  11. Shpitser I, Pearl J (2006) Identification of conditional interventional distributions. Proceedings of the Twenty-Second Conference on Uncertainty in Artificial In- telligence, eds Dechter R, Richardson T (AUAI Press, Corvallis, OR), pp 437-444.
  12. Pearl J (1993) Comment: Graphical models, causality, and intervention. Stat Sci 8(3): 266-269.
  13. Shpitser I, VanderWeele T, Robins J (2010) On the validity of covariate adjustment for estimating causal effects. Proceedings of the Twenty-Sixth Conference on Uncertainty in Artificial Intelligence (AUAI, Corvallis, OR), pp 527-536.
  14. Pearl J, Paz A (2014) Confounding equivalence in causal equivalence. J Causal Inference 2(1):77-93.
  15. Perkovic E, Textor MK, Maathuis M (2015) A complete adjustment criterion. Pro- ceedings of the Thirty-First Conference on Uncertainty in Artificial Intelligence, eds
  16. Meila M, Heskes T (AUAI Press, Corvallis, OR), pp 682-691.
  17. van der Zander B, Liskiewicz M, Textor J (2014) Constructing Separators and Adjustment Sets in Ancestral Graphs (AUAI Press, Corvallis, OR), pp 907-916.
  18. Bareinboim E, Pearl J (2012) Causal inference by surrogate experiments: z-identifi- ability. Proceedings of the Twenty-Eighth Conference on Uncertainty in Artificial In- telligence, eds de Freitas N, Murphy K (AUAI Press, Corvalis, OR), pp 113-120.
  19. Chen B, Pearl J (2014) Graphical tools for linear structural equation modeling. Tech- nical Report R-432 (Department of Computer Science, University of California, Los Angeles). Psychometrika, in press. Available at ftp.cs.ucla.edu/pub/stat_ser/r432.pdf.
  20. Tian J, Pearl J (2002) A general identification condition for causal effects. Proceedings of the Eighteenth National Conference on Artificial Intelligence, eds Dechter R, Kearns M, Sutton R (AAAI Press/MIT Press, Menlo Park, CA), pp 567-573.
  21. Cooper G (1995) Causal discovery from data in the presence of selection bias. Pro- ceedings of the Fifth International Workshop on Artificial Intelligence and Statistics, eds Fisher D, Lenz H-J (Artificial Intelligence and Statistics, Fort Lauderdale, FL), pp 140-150.
  22. Cortes C, Mohri M, Riley M, Rostamizadeh A (2008) Sample Selection Bias Correction Theory (Springer, Berlin/Heidelberg, Germany), pp 38-53.
  23. Whittemore A (1978) Collapsibility of multidimensional contingency tables. J R Stat Soc B 40(3):328-340.
  24. Kuroki M, Cai Z (2006) On recovering a population covariance matrix in the presence of selection bias. Biometrika 93(3):601-611.
  25. Mefford J, Witte JS (2012) The covariate's dilemma. PLoS Genet 8(11):e1003096.
  26. Angrist JD (1997) Conditional independence in sample selection models. Econ Lett 54(2):103-112.
  27. Glymour M, Greenland S (2008) Causal Diagrams in Modern Epidemiology, eds Rothman K, Greenland S, Lash T (Lippincott Williams & Wilkins, Philadelphia), 3rd Ed, pp 183-209.
  28. Greenland S, Pearl J (2011) Adjustments and their consequences -collapsibility analysis using graphical models. Int Stat Rev 79(3):401-426.
  29. Didelez V, Kreiner S, Keiding N (2010) Graphical models for inference under outcome- dependent sampling. Stat Sci 25(3):368-387.
  30. Bareinboim E, Tian J, Pearl J (2014) Recovering from selection bias in causal and statistical inference. Proceedings of the Twenty-Eighth National Conference on Ar- tificial Intelligence, eds Brodley C, Stone P (AAAI Press, Menlo Park, CA), pp 2410-2416.
  31. Bareinboim E, Tian J (2015) Recovering causal effects from selection bias. Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence, eds Koenig S, Bonet B (AAAI Press, Menlo Park, CA), pp 3475-3481.
  32. Bareinboim E, Pearl J (2012) Controlling selection bias in causal inference. Proceedings of the Fifteenth International Conference on Artificial Intelligence and Statistics, eds Lawrence N, Girolami M (Journal of Machine Learning Research Workshop and Conference Proceedings, La Palma, Canary Islands), Vol 22, pp 100-108.
  33. Cole SR, Stuart EA (2010) Generalizing evidence from randomized clinical trials to target populations: The ACTG 320 trial. Am J Epidemiol 172(1):107-115.
  34. Tipton E, et al. (2014) Sample selection in randomized experiments: A new method using propensity score stratified sampling. J Res Educ Eff 7(1):114-135.
  35. Pearl J (2015) Generalizing experimental findings. J Causal Inference 3(2):259-266.
  36. Pearl J, Bareinboim E (2014) External validity: From do-calculus to transportability across populations. Stat Sci 29(4):579-595.
  37. Manski C (2007) Identification for Prediction and Decision (Harvard Univ Press, Cambridge, MA).
  38. Höfler M, Gloster AT, Hoyer J (2010) Causal effects in psychotherapy: Counterfactuals counteract overgeneralization. Psychother Res 20(6):668-679.
  39. Bareinboim E, Pearl J (2012) Transportability of causal effects: Completeness results. Proceedings of The Twenty-Sixth Conference on Artificial Intelligence, eds Hoffmann J, Selman B (AAAI Press, Menlo Park, CA), pp 698-704.
  40. Bareinboim E, Pearl J (2013) A general algorithm for deciding transportability of experimental results. J Causal Inference 1(1):107-134.
  41. Bareinboim E, Pearl J (2013) Meta-transportability of causal effects: A formal ap- proach. Proceedings of the Sixteenth International Conference on Artificial In- telligence and Statistics, eds Carvalho C, Ravikumar P (Journal of Machine Learning Research Workshop and Conference Proceedings, Scottsdale, AZ), Vol 31, pp 135-143.
  42. Bareinboim E, Pearl J (2014) Transportability from multiple environments with limited experiments: Completeness results. Advances in Neural Information Processing Systems 27 (NIPS 2014), eds Ghahramani Z, Welling M, Cortes C, Lawrence N, Weinberger K. Available at https://papers.nips.cc/book/advances-in-neural- information-processing-systems-27-2014.

Cited by

Sorted by citations

Authors
Bareinboim, E., & Pearl, J.
Affiliations
University of California, Los Angeles (US)
Venue
Proceedings of the National Academy of Sciences
Publisher
National Academy of Sciences
Published
2016-07-05
Volume
113
Issue
27
Pages
7345-7352
ISSN
  • 0027-8424 (ISSN-L)
  • 1091-6490
DOI
10.1073/pnas.1510507113
PMC
PMC4941504
OpenAlex
W2466989778
Cited by
621
References
12
Citations by year
2026: 63 · 2025: 59 · 2024: 63 · 2023: 90 · 2022: 73 · 2021: 96 · 2020: 62 · 2019: 55 · 2018: 33 · 2017: 20 · 2016: 7
FWCI
38.84
Type
journal article
Language
en
Open access
Locations
  • Proceedings of the National Academy of Sciences (publishedVersion)
  • PubMed Central (publishedVersion)
  • PubMed (publishedVersion)
  • eScholarship (California Digital Library) (submittedVersion)
Topics
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Inference
  • Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Modeling and Causal Inference
Keywords
causal inference, data fusion, data heterogeneity, big data, confounding

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review