Pith. sign in

REVIEW 2 major objections 5 minor 63 references

Proximal Mediation Analysis with Unmeasured Treatment-Induced Confounding

T0 review · 2 major / 5 minor · reviewed 2026-07-12 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Observed proxies can identify interventional mediation effects even when treatment-induced confounders are never measured.

desk verdict Solid, first-of-its-kind proximal identification for interventional mediation under unmeasured treatment-induced confounding; theory and sims check out, usual proximal caveats apply. read the letter →

arxiv 2607.02901 v1 pith:CSWBPASX submitted 2026-07-03 stat.ME

Xiaoying Zhang , Jiawei Shan , Wei Li This is my paper
classification stat.ME MSC 62D2062G0562P10
keywords proximalcausalinferenceinterventionalmediationeffectstreatment-inducedconfoundingbridgefunctionsmultiplerobustnessdebiasedmachinelearningefficientinfluencefunction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

When a treatment affects later confounders of the mediator-outcome relationship, the usual natural direct and indirect effects are typically unidentifiable. Interventional effects remain a usable alternative, but only if those intermediate confounders can be fully observed. This paper shows that two carefully chosen proxy variables for the unmeasured intermediate confounder restore nonparametric identification of the interventional direct and indirect effects. Four distinct observed-data formulas are derived, each relying on different combinations of bridge functions and conditional densities. From the efficient influence function the authors build a multiply robust estimator that stays consistent if any one of four model sets is correct, and a debiased machine-learning version that retains root-n rates under slower nuisance estimation. Simulations confirm the robustness property; an application to racial disparities in life satisfaction recovers a modest negative indirect effect operating through discrimination.

What carries the argument

Outcome and mediation confounding bridge functions: solutions h_a and q_a to the Fredholm integral equations that recover the latent conditional expectations of the outcome and of the reciprocal mediator density given the unmeasured intermediate confounder; these bridges convert the four proximal identification formulas into observable functionals and enter the efficient influence function.

What would settle it

In a simulation or experiment where the true unmeasured intermediate confounder is known and the completeness condition is deliberately violated (e.g., by making the proxies independent of the confounder), check whether any of the four identification formulas recovers the true interventional effect; systematic bias would falsify the claim.

Watch

Extended reading notes

Core claim

Under standard consistency, positivity and latent ignorability assumptions, plus two proxy variables satisfying conditional independence and completeness, the interventional parameter ψ_{a,a'}=E{Y(a,G(a'))} is nonparametrically identified by any of four functionals of the observed data. The corresponding efficient influence function yields a multiply robust, locally efficient estimator that remains consistent whenever at least one of four nested nuisance models is correct.

Load-bearing premise

The proxies must carry enough information about the unmeasured intermediate confounder that any non-zero function of that confounder produces a non-zero conditional expectation given the proxies (completeness); if this fails the bridge equations need not have solutions that recover the latent quantities.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper develops proximal identification and estimation for interventional direct and indirect effects when treatment-induced confounders of the mediator–outcome relationship are unmeasured. Under consistency, positivity, latent conditional ignorability, proxy independence (Assumption 4), and completeness (Assumption 5), Theorems 1–2 and Corollary 1 show that ψ_{a,a′}=E{Y(a,G(a′))} is identified by any of four observed-data functionals that involve an outcome bridge h_a solving the Fredholm equation (3) or a mediation bridge q_a solving (4). Theorem 3 derives the efficient influence function under the semiparametric model that only assumes existence of h_a; Theorem 4 establishes multiple robustness of the corresponding parametric estimator under the union of four model classes; and Theorem 5 gives root-n asymptotic normality and local efficiency for a cross-fit debiased machine-learning estimator that estimates the bridges by minimax learning. Simulations confirm multiple robustness (Table 1, Figure 2) and DML rate behavior (Table 2); an HRS application illustrates racial disparities in life satisfaction mediated by discrimination.

Significance. The work fills a genuine gap: existing proximal mediation methods address unmeasured pre-treatment confounding, while existing interventional-effect methods require treatment-induced confounders to be fully observed. Extending proximal causal inference to treatment-induced U, obtaining four distinct identification formulas, and delivering a multiply robust, locally efficient, and DML-compatible estimator is a substantial methodological contribution. The moment reformulation of the mediation bridge (Proposition 1) that avoids estimating f(M|W,A,L,X) is technically useful, and the simulation design cleanly isolates the multiple-robustness claim. Completeness is the standard, untestable price of proximal methods and is stated transparently; it does not undermine the internal coherence of the results.

major comments (2)
  1. The main text defers all proofs of Theorems 1–5 and Proposition 1 to a supplement that is not included in the submitted manuscript. For a paper whose central claims are nonparametric identification via Fredholm inversion and an EIF-based multiply robust estimator, the absence of the technical arguments prevents independent verification of the key steps (especially the passage from the integral equations (3)–(4) to the latent conditional expectations, the derivation of the EIF under surjectivity of S_a, and the second-order remainder analysis for Theorem 5). The supplement should be supplied and the main-text statements cross-checked against it before acceptance.
  2. Section 5 / Table 3: the parametric multiply-robust estimates IIE_mr and IDE_mr exhibit extremely large and unstable bootstrap standard errors (e.g., SE 1.252 and 2.092 under C1), rendering the corresponding confidence intervals non-informative, while the DML estimates remain stable. The paper correctly notes this as evidence of a complex data-generating process, but the application is presented as an illustration of the proposed methodology. Either a clearer discussion of when the parametric MR estimator is expected to be unreliable in finite samples, or a more carefully chosen real-data example in which at least one of the four model classes is plausibly well-specified, would strengthen the empirical claim.
minor comments (5)
  1. Page 5, line after (1): typographical error “conrresponds” should be “corresponds”.
  2. Figure 1 caption and surrounding text: L and X are omitted “for simplicity,” but the figure is the only graphical summary of the proxy conditions; a brief note that the same conditional independences hold after conditioning on L,X would help readers.
  3. Section 3.2: the choice of user-specified functions d(Z,L,M,X) and g(W,L,M,X) for the parametric estimating equations is left largely to Remark 1; a short practical recommendation or default (e.g., the score of a linear/logistic working model) would improve reproducibility.
  4. Table 1 caption: bias is reported “in units of 10^{-2}”; stating this once in the table note is fine, but the same convention should be applied consistently to Table 2 for readability.
  5. References: a few recent proximal-mediation and interventional-effect papers (e.g., on time-varying settings) are cited; ensuring the most closely related concurrent arXiv preprints are acknowledged would be useful for the final version.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: identification of interventional effects from proxy bridge equations is a standard proximal derivation, not tautological.

full rationale

The paper derives nonparametric identification of ψ_{a,a′}=E{Y(a,G(a′))} from stated Assumptions 1–5 (consistency, positivity, latent ignorability, proxy conditional independences, and completeness) plus existence of solutions to the Fredholm equations (3) and (4). Theorems 1–2 and Corollary 1 recover four observed-data functionals ψ₁–ψ₄ by inverting those equations under completeness; the target is not defined in terms of the bridge functions, nor are the bridges fitted to the target and then re-labeled as predictions. The EIF (Theorem 3), multiple robustness (Theorem 4), and DML rates (Theorem 5) are standard semiparametric constructions under the model M_sp that assumes existence of h_a. Self-citations (Miao et al., Tchetgen Tchetgen et al., Dukes et al., Ghassami et al., etc.) supply background proximal machinery and are not load-bearing uniqueness theorems that force the present estimands. Completeness is an untestable relevance condition, not a circular definition. No step reduces Eq. X to Eq. Y by construction or renames a fitted input as a first-principles result. Score 0 is therefore warranted.

Assumptions & free parameters 2 free parameters · 8 assumptions · 2 invented entities

The central claim rests on standard causal assumptions (consistency, positivity, latent ignorability), proxy conditional independences, completeness for invertibility of conditional-expectation operators, existence of outcome/mediation bridge functions solving Fredholm equations of the first kind, and (for local efficiency only) surjectivity of those operators. No numerical free parameters are fitted into the identification theory; simulation and application use standard parametric/ML nuisance fits that are not part of the identification claim. Invented entities are the two bridge functions and the four identification functionals, which are mathematical constructs rather than physical objects.

free parameters (2)
  • Regularization parameters λ^h_Q, λ^h_H, λ^q_H, λ^q_Q in minimax bridge estimation
    Chosen by the analyst for the DML bridge estimators in §3.3; affect finite-sample bridge fits though not the population identification claim.
  • Parametric bridge coefficients β_a, θ_a and nuisance regression coefficients
    Fitted by estimating equations / MLE in the parametric MR estimator; required for implementation but not free constants in the identification theorems.
assumptions (8)
  • domain assumption Consistency of potential outcomes/mediators (Assumption 1)
    Standard causal consistency; §2.1.
  • domain assumption Positivity of treatment, intermediate confounders, and mediator (Assumption 2)
    Standard overlap; §2.1.
  • domain assumption Latent conditional ignorability: Y(a,m)⊥A|X, M(a)⊥A|X, Y(a,m)⊥M|A,U,L,X (Assumption 3)
    Core no-unmeasured-confounding structure with U latent intermediate; §2.1.
  • domain assumption Proxy conditions Z⊥Y|A,U,L,M,X and W⊥(M,Z)|A,U,L,X (Assumption 4)
    Defines valid mediator- and outcome-inducing proxies; Figure 1, §2.2.
  • ad hoc to paper Completeness of U given (Z,...) and given (W,...) (Assumption 5)
    Standard in proximal CI but strong and untestable; required for Theorems 1–2.
  • ad hoc to paper Existence of outcome bridge h_a solving E(Y|Z,A=a,L,M,X)=E{h_a(W,L,M,X)|Z,A=a,L,M,X} (eq. 3)
    Fredholm first-kind solvability; dual completeness Assumption 6 plus regularity in S2.
  • ad hoc to paper Existence of mediation bridge q_a solving 1/f(M|W,A=a,L,X)=E{q_a(Z,L,M,X)|W,A=a,L,M,X} (eq. 4)
    Second identification route; dual completeness Assumption 7.
  • ad hoc to paper Surjectivity of conditional expectation operator S_a (Assumption 8)
    Used only for local efficiency of the EIF under M_sp; authors note not needed for identification/estimation.
invented entities (2)
  • Outcome confounding bridge function h_a(W,L,M,X)
    purpose: Maps observed proxies to recover E(Y|U,A,L,M,X) without observing U
    Standard proximal-CI construct adapted to treatment-induced U; no independent physical evidence beyond the integral equation.
  • Mediation confounding bridge function q_a(Z,L,M,X)
    purpose: Recovers inverse mediator density given U via proxies for the fully weighted identification formula ψ₄
    Paper-specific bridge for the mediation density inverse problem; existence assumed via completeness.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Proximal Mediation Analysis with Unmeasured Treatment-Induced Confounding." pith.science (2026). https://pith.science/paper/CSWBPASX

@misc{pith2026260702901,
  author       = {Pith},
  title        = {Pith review of: Proximal Mediation Analysis with Unmeasured Treatment-Induced Confounding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CSWBPASX}},
  note         = {Machine review of arXiv:2607.02901}
}
read the original abstract

Mediation analysis provides a central framework for elucidating causal mechanisms, yet its application is often impeded by treatment-induced confounding, under which the widely used natural mediation effects are generally unidentifiable. Interventional effects have been proposed as an alternative when these confounders are observable; however, identifying and estimating interventional effects remains challenging when confounders are unmeasured. In this paper, we address this issue by using observed variables as proxies for unmeasured treatment-induced confounders. We establish four proximal identification results and develop a multiply robust, semiparametric locally efficient estimator that accommodates flexible machine learning methods for nuisance parameter estimation. The proposed approach is illustrated through simulation studies and a real-data application evaluating racial disparities in life satisfaction mediated by discrimination.

Figures

Figures reproduced from arXiv: 2607.02901 by the authors.

Figure 1
Figure 1. The causal diagram with unmeasured treatment-induced confounder [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Box plots of bias of parametric estimators. First row corresponds to the case where [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

63 extracted references · 4 linked inside Pith

  1. [1]

    Handbook of Econometrics , volume=

    Linear inverse problems in structural econometrics estimation based on spectral decomposition and regularization , author=. Handbook of Econometrics , volume=. 2007 , publisher=

    2007

  2. [2]

    Statistical Science , volume =

    Identification, Inference and Sensitivity Analysis for Causal Mediation Effects , author =. Statistical Science , volume =. doi:10.1214/10-STS321 , url =

    doi:10.1214/10-sts321

  3. [3]

    Econometrica , volume =

    Maximum Likelihood Estimation of Misspecified Models , author =. Econometrica , volume =. doi:10.2307/1912526 , url =. 1912526 , eprinttype =

    doi:10.2307/1912526

  4. [4]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Non-parametric inference about mean functionals of non-ignorable non-response data without identifying the joint distribution , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2023 , publisher=

    2023

  5. [5]

    Statistical Science , volume=

    An Introduction to Proximal Causal Inference , author=. Statistical Science , volume=. 2024 , doi=

    2024

  6. [6]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Multiply robust causal inference with double-negative control adjustment for categorical unmeasured confounding , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2020 , publisher=

    2020

  7. [7]

    Epidemiology , volume=

    Estimation of direct causal effects , author=. Epidemiology , volume=. 2006 , publisher=

    2006

  8. [8]

    The International Journal of Biostatistics , volume=

    Direct Effect Models , author=. The International Journal of Biostatistics , volume=. 2008 , publisher=

    2008

Show all 63 references
  1. [9]

    Essay on principles

    On the application of probability theory to agricultural experiments. Essay on principles. Section 9. , author=. Annals of Agricultural Sciences , volume=

  2. [10]

    , author=

    Estimating causal effects of treatments in randomized and nonrandomized studies. , author=. Journal of Educational Psychology , volume=. 1974 , publisher=

    1974

  3. [11]

    Epidemiology , volume=

    Identifiability and exchangeability for direct and indirect effects , author=. Epidemiology , volume=. 1992 , publisher=

    1992

  4. [12]

    Proceedings of the Seventeenth Conference on Uncertainty and Artificial Intelligence , pages=

    Direct and indirect effects , author=. Proceedings of the Seventeenth Conference on Uncertainty and Artificial Intelligence , pages=

  5. [13]

    , author=

    A general approach to causal mediation analysis. , author=. Psychological Methods , volume=. 2010 , publisher=

    2010

  6. [14]

    Annals of Statistics , volume=

    Semiparametric theory for causal mediation analysis: efficiency bounds, multiple robustness, and sensitivity analysis , author=. Annals of Statistics , volume=

  7. [15]

    Epidemiology , volume=

    Cognitive aging in black and white Americans: cognition, cognitive decline, and incidence of Alzheimer disease dementia , author=. Epidemiology , volume=. 2018 , publisher=

    2018

  8. [16]

    Journal of Aging and Health , volume=

    Social determinants of racial disparities in cognitive functioning in later life in Canada , author=. Journal of Aging and Health , volume=. 2020 , publisher=

    2020

  9. [17]

    , author=

    Socioeconomic and psychosocial mechanisms underlying racial/ethnic disparities in cognition among older adults. , author=. Neuropsychology , volume=. 2021 , publisher=

    2021

  10. [18]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Robust inference on population indirect causal effects: the generalized front door criterion , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2020 , publisher=

    2020

  11. [19]

    Biometrika , volume=

    Proximal mediation analysis , author=. Biometrika , volume=. 2023 , publisher=

    2023

  12. [20]

    Biometrics , volume=

    Using instrumental variables to address unmeasured confounding in causal mediation analysis , author=. Biometrics , volume=. 2024 , publisher=

    2024

  13. [21]

    Biometrika , volume=

    Causal inference with hidden mediators , author=. Biometrika , volume=. 2025 , publisher=

    2025

  14. [22]

    arXiv preprint arXiv:2504.11848 , year=

    Proximal Inference on Population Intervention Indirect Effect , author=. arXiv preprint arXiv:2504.11848 , year=

    arXiv

  15. [23]

    Epidemiology , volume=

    Effect decomposition in the presence of an exposure-induced mediator-outcome confounder , author=. Epidemiology , volume=. 2014 , publisher=

    2014

  16. [24]

    Journal of the American Statistical Association , volume=

    Identification, semiparametric efficiency, and quadruply robust estimation in mediation analysis with treatment-induced confounding , author=. Journal of the American Statistical Association , volume=. 2023 , publisher=

    2023

  17. [25]

    Epidemiology , volume=

    Identification of natural direct effects when a confounder of the mediator is directly affected by exposure , author=. Epidemiology , volume=. 2014 , publisher=

    2014

  18. [26]

    Biometrics , volume=

    Efficient and flexible estimation of natural direct and indirect effects under intermediate confounding and monotonicity constraints , author=. Biometrics , volume=. 2023 , publisher=

    2023

  19. [27]

    Causality and Psychopathology: Finding the Determinants of Disorders and their Cures , volume=

    Alternative graphical causal models and the identification of direct effects , author=. Causality and Psychopathology: Finding the Determinants of Disorders and their Cures , volume=. 2010 , publisher=

    2010

  20. [28]

    Epidemiology , volume=

    Bounds for pure direct effect , author=. Epidemiology , volume=. 2014 , publisher=

    2014

  21. [29]

    arXiv preprint arXiv:1509.01652 , year=

    On partial identification of the pure direct effect , author=. arXiv preprint arXiv:1509.01652 , year=

    arXiv

  22. [30]

    Proceedings of the 19th International Joint Conference on Artificial Intelligence , pages=

    Identifiability of path-specific effects , author=. Proceedings of the 19th International Joint Conference on Artificial Intelligence , pages=

  23. [31]

    Biometrika , volume=

    On semiparametric estimation of a path-specific effect in the presence of mediator-outcome confounding , author=. Biometrika , volume=. 2020 , publisher=

    2020

  24. [32]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Mediation analysis with time varying exposures and mediators , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2017 , publisher=

    2017

  25. [33]

    Biometrika , volume=

    Nonparametric efficient causal mediation with intermediate confounders , author=. Biometrika , volume=. 2021 , publisher=

    2021

  26. [34]

    Epidemiology , volume=

    Interventional effects for mediation analysis with multiple mediators , author=. Epidemiology , volume=. 2017 , publisher=

    2017

  27. [35]

    arXiv preprint arXiv:2311.16793 , year=

    Mediation pathway selection with unmeasured mediator-outcome confounding , author=. arXiv preprint arXiv:2311.16793 , year=

  28. [36]

    Biometrika , volume=

    Identifying causal effects with proxy variables of an unmeasured confounder , author=. Biometrika , volume=. 2018 , publisher=

    2018

  29. [37]

    arXiv preprint arXiv:2009.10982 , year=

    An introduction to proximal causal learning , author=. arXiv preprint arXiv:2009.10982 , year=

    2009 arXiv

  30. [38]

    Journal of the American Statistical Association , volume=

    Semiparametric proximal causal inference , author=. Journal of the American Statistical Association , volume=. 2024 , publisher=

    2024

  31. [39]

    Econometrica , volume=

    Instrumental variable estimation of nonparametric models , author=. Econometrica , volume=. 2003 , publisher=

    2003

  32. [40]

    1989 , publisher=

    Linear Integral Equations , author=. 1989 , publisher=

    1989

  33. [41]

    1993 , publisher=

    Efficient and Adaptive Estimation for Semiparametric Models , author=. 1993 , publisher=

    1993

  34. [42]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Proximal causal inference for complex longitudinal studies , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2023 , publisher=

    2023

  35. [43]

    Fonctions de r

    Sklar, A , booktitle=. Fonctions de r

  36. [44]

    1997 , publisher=

    Multivariate Models and Multivariate Dependence Concepts (1st ed.) , author=. 1997 , publisher=

    1997

  37. [45]

    2006 , publisher=

    An Introduction to Copulas , author=. 2006 , publisher=

    2006

  38. [46]

    2010 , publisher=

    Copula Theory and its Applications , author=. 2010 , publisher=

    2010

  39. [47]

    The Econometrics Journal , volume=

    Double/debiased machine learning for treatment and structural parameters , author=. The Econometrics Journal , volume=. 2018 , publisher=

    2018

  40. [48]

    Handbook of Statistical Methods for Precision Medicine , pages=

    Semiparametric doubly robust targeted double machine learning: a review , author=. Handbook of Statistical Methods for Precision Medicine , pages=. 2024 , publisher=

    2024

  41. [49]

    arXiv preprint arXiv:2103.14029 , year=

    Causal inference under unmeasured confounding with negative controls: A minimax learning approach , author=. arXiv preprint arXiv:2103.14029 , year=

    arXiv

  42. [50]

    International Conference on Artificial Intelligence and Statistics , pages=

    Minimax kernel machine learning for a class of doubly robust functionals with application to proximal causal inference , author=. International Conference on Artificial Intelligence and Statistics , pages=

  43. [51]

    Alzheimer's & Dementia , volume=

    Racial disparities in cognitive function & mild cognitive impairment: Effects of income, education, and discrimination in the Midlife in the United States (MIDUS) study , author=. Alzheimer's & Dementia , volume=. 2024 , publisher=

    2024

  44. [52]

    Advances in Neural Information Processing Systems , volume=

    Minimax estimation of conditional moment models , author=. Advances in Neural Information Processing Systems , volume=

  45. [53]

    Journal of the American Statistical Association , volume=

    Quantifying an adherence path-specific effect of antiretroviral therapy in the Nigeria PEPFAR program , author=. Journal of the American Statistical Association , volume=. 2017 , publisher=

    2017

  46. [54]

    Statistical Causal Inferences and their Applications in Public Health Research , pages=

    Semiparametric theory and empirical processes in causal inference , author=. Statistical Causal Inferences and their Applications in Public Health Research , pages=. 2016 , publisher=

    2016

  47. [55]

    The Annals of Statistics , pages=

    On asymptotically efficient estimation in semiparametric models , author=. The Annals of Statistics , pages=. 1986 , publisher=

    1986

  48. [56]

    1996 , publisher=

    Pattern Recognition and Neural Networks , author=. 1996 , publisher=

    1996

  49. [57]

    Journal of Personality Assessment , volume=

    The satisfaction with life scale , author=. Journal of Personality Assessment , volume=. 1985 , publisher=

    1985

  50. [58]

    , author=

    Ethnic inequality between Arabs and Jews in Israel in global life satisfaction: A social determinants examination among young adults. , author=. American Journal of Orthopsychiatry , year=

  51. [59]

    Quality of Life Research , volume=

    The relative contributions of race/ethnicity, socioeconomic status, health, and social relationships to life satisfaction in the United States , author=. Quality of Life Research , volume=. 2009 , publisher=

    2009

  52. [60]

    Journal of Health Psychology , volume=

    Racial differences in physical and mental health: Socio-economic status, stress and discrimination , author=. Journal of Health Psychology , volume=. 1997 , publisher=

    1997

  53. [61]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Non-parametric methods for doubly robust estimation of continuous treatment effects , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2017 , publisher=

    2017

  54. [62]

    The international journal of biostatistics , volume=

    Locally efficient estimation of marginal treatment effects when outcomes are correlated: is the prize worth the chase? , author=. The international journal of biostatistics , volume=. 2014 , publisher=

    2014

  55. [63]

    Targeted learning: causal inference for observational and experimental data , pages=

    Cross-validated targeted minimum-loss-based estimation , author=. Targeted learning: causal inference for observational and experimental data , pages=. 2011 , publisher=

    2011

Pith tools

Reviewed July 12, 2026 · model on record in the stance chip above.