Pith. sign in

REVIEW 2 major objections 5 minor 69 references

Quantile treatment effects can be estimated consistently and with √n normality by combining a cumulative probability model with the efficient influence function, without prespecifying an outcome transformation.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Quantile treatment effects can be estimated doubly robustly with a cumulative probability model, giving reliable distributional comparisons for skewed outcomes.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection Useful CPM-based doubly robust QTE estimator with strong simulations; central √n-normality claim is asserted, not proven, for the diverging-dimension CPM plug-in. the 2 major comments →

arxiv 2607.27633 v1 pith:IDPXUD6H submitted 2026-07-30 stat.ME

Doubly Robust Estimators of Quantile Treatment Effects With Semiparametric Cumulative Probability Models

Hao Wu , Chun Li , Bryan E. Shepherd This is my paper
classification stat.ME MSC 62G0562G2062D20
keywords quantile treatment effectdoubly robust estimationcumulative probability modelsemiparametric transformation modelefficient influence functionprobability treatment effectskewed outcomesdetection limits
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

The paper claims that quantile treatment effects—differences between the p-th quantiles of potential outcomes under treatment and control—can be estimated consistently, with √n normality, by combining a rank-based semiparametric outcome model with the efficient influence function for the marginal quantile. The proposal is doubly robust: the point estimator stays consistent if either the propensity score model or the outcome regression model is correctly specified. Because the outcome model is a cumulative probability model, no outcome transformation or normality assumption is needed, and skewed or detection-limited outcomes can be analyzed on their original scale. The paper also reports that EIF-based variance estimators are not doubly robust, whereas sandwich and bootstrap variance estimators are, and demonstrates the method on an HIV cohort with skewed CD4 counts and viral load subject to detection limits.

Core claim

Under the standard causal identification assumptions, the paper derives two estimators of a marginal potential-outcome quantile q_{a,p}: an inverse-CDF estimator that builds a doubly robust marginal CDF from its efficient influence function and inverts it, and a direct estimator that solves the efficient influence function estimating equation for the quantile. Both are claimed to be √n-consistent, asymptotically normal, and doubly robust when the outcome model is a cumulative probability model. The same framework yields probability treatment effects and conditional effects. In simulations, these estimators match AIPW/TMLE applied to the correctly transformed outcome, without needing to know

What carries the argument

The cumulative probability model (CPM) is the central working model: it writes the conditional CDF of the outcome as F_ε(α(y) − β^T X), with a prespecified link F_ε and an unspecified monotone intercept function α(y) estimated nonparametrically from the outcome ranks. This makes the outcome-regression component invariant to monotone outcome transformations and capable of handling mixed discrete-continuous outcomes. The other load-bearing piece is the efficient influence function for the marginal CDF F_{Y_a}(y), which combines inverse probability weighting with an outcome-regression correction; the paper's estimators either invert this function to get quantiles or solve the corresponding quan

Load-bearing premise

The asymptotic result assumes that the cumulative probability model's nonsmooth, growing-dimensional maximum-likelihood fit behaves like a well-behaved plug-in nuisance estimator for the quantile estimating equation—the paper states 'standard regularity conditions' rather than proving them, so if that assumption fails, the claimed √n normality and variance formulas would not hold.

What would settle it

Simulate a continuous outcome with a heavy-tailed error (so the CPM's link is correctly specified but the support is unbounded), keep both nuisance models correct, and compare the empirical sampling distribution of √n(q̂a,p − qa,p) with the normal distribution implied by the sandwich variance at increasing sample sizes. If the distribution does not converge or the 95% sandwich intervals miss more than simulation noise would allow, the claimed √n normality fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • A practitioner can estimate QTE at any quantile level for skewed, bounded, or detection-limited outcomes without first choosing a log, Box-Cox, or other transformation; one of the two nuisance models being correct is enough for consistency.
  • The two implementations—inverse-CDF and direct estimating equation—are asymptotically equivalent and produce nearly identical point estimates, so either can be used in practice.
  • For confidence intervals, the sandwich and nonparametric bootstrap remain near nominal coverage under nuisance-model misspecification, while the EIF-based variance can overcover; inference should therefore rely on the former.
  • The HIV application shows that the method can reveal tail effects—such as a larger CD4 improvement at the 75th percentile—that a mean treatment effect understates, and can estimate the probability of viral suppression at the detection limit directly.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: because the inverse-CDF estimator produces a full estimate of each potential-outcome CDF, the same machinery could be repurposed for other distributional summaries (tail ratios, inequality indices, or threshold probabilities) with the same double-robustness guarantee, provided those summaries are continuous functionals of the CDF.
  • Beyond the paper: the EIF-variance failure under outcome-model misspecification is probably not specific to quantiles; it likely applies to any doubly robust estimator built from the same EIF template, so the paper's recommendation to use sandwich or bootstrap variance is a sensible default for that whole class.
  • Beyond the paper: the main remaining modeling choice is the link function of the CPM. The simulations show insensitivity to moderate link misspecification, but letting the transformation itself be estimated data-adaptively is a natural extension; whether double robustness is preserved at slower convergence rates is an open, testable question.
  • Beyond the paper: at extreme quantiles in sparse tails, the weighted interpolation used to invert the estimated CDF could dominate the estimator's finite-sample behavior; a simulation study varying the interpolation and the tail heaviness would clarify when the inverse-CDF version should be preferred over the direct version.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper proposes CPM-based doubly robust estimators of marginal quantile treatment effects (QTE) for binary treatments. Two strategies are developed: an inverse-CDF approach that first estimates the marginal CDF of each potential outcome using the efficient influence function (EIF) and then interpolates the quantile, and a direct approach that solves the EIF estimating equation for the quantile itself. The outcome-regression nuisance is modeled with a cumulative probability model (CPM), a rank-based semiparametric linear transformation model, while the propensity score is modeled separately. The paper claims √n-consistency and asymptotic normality under standard regularity conditions, and recommends sandwich or bootstrap variance estimators because EIF-based variance estimators are shown not to be doubly robust. Extensions to probability treatment effects and conditional QTEs are given. The methods are evaluated in an extensive simulation study and applied to an HIV cohort comparing INSTI- versus non-INSTI-based regimens on CD4 count and viral load.

Significance. If the theoretical claims can be rigorously established, the paper makes a useful contribution: it provides a flexible outcome-regression working model for QTE estimation that avoids prespecifying an outcome transformation, handles skewed and detection-limited outcomes, and preserves double robustness of point estimation under misspecification of one nuisance model. The simulation study is broad and the code is publicly available. The paper is also honest about a nontrivial limitation—EIF-based variance estimation is not doubly robust in this setting. However, the central asymptotic normality claim is currently asserted rather than proved for the CPM NPMLE plug-in, whose nuisance dimension grows with sample size, and this is load-bearing for the inference procedures that the paper recommends.

major comments (2)
  1. [§3.3 and Supplementary S2.3] The central asymptotic normality claim is not proven. The stacked estimating equations treat θ=(ψ^T, ξ^T, q_1,p, q_0,p)^T as fixed-dimensional and cite Stefanski and Boos M-estimation, but the CPM parameter ξ includes intercepts α_1,...,α_{J_n-1} whose number grows with the number of distinct outcomes (Section 2). Standard fixed-dimension M-estimation does not apply. The paper does not show that estimation error in the CPM NPMLE is first-order negligible for the quantile estimating equation, nor does it provide the stochastic equicontinuity or differentiability-in-q conditions needed for the non-smooth indicator I(Y≤q). Consequently the displayed Σ=A^{-1}B(A^{-1})^T is unjustified: A_{ξξ} is an observed information matrix of growing dimension and A_{1ξ} sums over a growing number of intercept terms. Simulations cannot substitute for this proof; the claim that the estimators are √n-consis
  2. [Abstract and §4.3.3, Table 2] The statement that the empirical sandwich estimator and the nonparametric bootstrap 'provide doubly robust variance estimation' with stable finite-sample performance under nuisance-model misspecification is overstated as written. Table 2 shows that when both nuisance models are misspecified, coverage is 13.8% for the empirical-SD interval, 7.8% for the bootstrap percentile interval, 10.9% for Sandwich-I, and 12.8% for Sandwich-S. While this is a consequence of point-estimator inconsistency, the text should explicitly condition all variance-double-robustness claims on at least one nuisance model being correctly specified and should distinguish variance estimation from valid confidence-interval coverage. The manuscript's own limitation statements in §4.3.3 should be reflected in the abstract and discussion.
minor comments (5)
  1. [§3.3] The displayed asymptotic variance for q̂_{a,p} uses φ_{q_{a,p}}, which includes the density f_{Y_a}(q_{a,p}) in its definition, but the estimating equation in (11) drops this multiplicative constant. The notation should be reconciled to avoid confusion about which quantity the displayed variance refers to.
  2. [S2.3.1 and Table S3.10] Sandwich-S depends on a smoothing approximation with bandwidth choices k=1.06σ̂ n^{-1/5} and bw.SJ. The supplementary table shows coverage as low as 81.9% for QTE(0.10) and 86.3% for QTE(0.90) in the correctly specified setting. A brief sensitivity analysis or guidance on bandwidth selection for extreme quantiles would strengthen the recommendation of Sandwich-I.
  3. [Table 1] The OR-CPM-icdf rows in the 'Misspecified OR' columns duplicate the values from the 'Correct OR' columns because this estimator does not use the propensity score. A footnote explaining that OR-CPM-icdf is not expected to be doubly robust would help readers interpret the table.
  4. [References] The reference 'Koenker (2005)' lacks initials and is incomplete; the entry should be brought into the journal style. There are also minor typographical issues such as '√nconsistent' in §3.3.
  5. [§5] In the viral-load application, the phrase 'the estimated probability of achieving viral suppression at 6 months was 0.87 under an INSTI-based regimen' could be misread as a conditional probability. Since the estimand is the marginal PTE, clarify that these are marginal potential-outcome probabilities under the identified causal assumptions.

Circularity Check

0 steps flagged

No circularity found; the main weakness is a proof gap, not a circular reduction.

full rationale

Walking the claimed derivation chain, the proposed estimators are defined as solutions to EIF estimating equations (Eq. 9 and Eq. 11), with the EIFs taken from standard semiparametric theory (Hahn 1998; Kennedy 2016; Zhang et al. 2012; Cheng and Li 2025). The double-robustness claim follows from the standard EIF mean-zero property under either correctly specified nuisance model, and the paper derives this explicitly in S2.3: E[phi_1,p] = F_{Y1}(q_1,p) - p under either correctly specified propensity score or outcome regression. The CPM is introduced as a flexible working model for the conditional CDF; the consistency of its NPMLE is cited to Li et al. (2023), a theorem about CPMs whose stated assumptions do not include the QTE target, so that citation is external evidence rather than a self-definitional input. The quantile estimators do not fit a parameter to the target quantile and then call it a prediction; they solve estimating equations for q after estimating nuisance functions. No uniqueness theorem from the authors is used to force the estimator, no ansatz is smuggled in via self-citation, and no known result is merely renamed. The genuine limitation is flagged in Section 3.3 and S2.3: the paper invokes "standard regularity conditions for M-estimation (Stefanski and Boos 2002)" for a stacked parameter vector including a CPM NPMLE with J_n intercepts and non-smooth indicators I(Y <= q), without proving those conditions. This is an unproved regularity/proof gap and a correctness risk, not a circular reduction: the claim is unsupported rather than equivalent to its inputs. Consistently, the score is 0.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

The central claim does not introduce ad hoc free parameters or invented entities. It relies on standard causal identification assumptions, prior asymptotic theory for CPM NPMLE, and M-estimation regularity conditions. The only tuning choices (e.g., bandwidth in Sandwich-S) are not central to the recommended approach.

axioms (4)
  • domain assumption Causal identification assumptions: consistency, unconfoundedness, positivity, and no interference.
    Stated in Section 3.1; needed to identify FYa(y) from observed data. Unconfoundedness is untestable in the observational HIV application.
  • domain assumption CPM NPMLE is consistent and asymptotically normal under the conditions cited from Li et al. (2023), including bounded responses.
    Invoked when plugging CPM estimates into the EIF; the paper does not verify these conditions for its own stacked estimating equations.
  • domain assumption M-estimation regularity conditions (Stefanski and Boos 2002) hold for the stacked estimating equations with non-smooth indicator functions.
    Used in Section 3.3 and S2.3 to justify sandwich variance estimation; the non-smoothness of I(Y≤q) makes this nontrivial.
  • domain assumption The potential-outcome density fYa(qa,p) is positive at the target quantiles.
    Required for the EIF of a quantile to be well defined; implicitly assumed throughout Section 3.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Doubly Robust Estimators of Quantile Treatment Effects With Semiparametric Cumulative Probability Models." pith.science (2026). https://pith.science/paper/IDPXUD6H

@misc{pith2026260727633,
  author       = {Pith},
  title        = {Pith review of: Doubly Robust Estimators of Quantile Treatment Effects With Semiparametric Cumulative Probability Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IDPXUD6H}},
  note         = {Machine review of arXiv:2607.27633}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The causal inference literature has traditionally focused on estimating the mean of the potential outcome, whereas evaluating how a treatment affects the entire outcome distribution can provide additional information in biomedical research. Quantile treatment effect (QTE) captures such distributional differences, particularly when outcomes are skewed. However, existing approaches for estimating QTE make distributional assumptions about the outcome and are thus sensitive to model misspecification. Motivated by an HIV study with skewed outcomes, one of which is subject to detection limits, we propose a doubly robust framework for estimating QTE based on the cumulative probability model (CPM), which is a rank-based, semiparametric linear transformation model. We develop two CPM-based estimation strategies: (1) an inverse-cumulative distribution function (CDF) approach that first estimates the marginal CDF of potential outcomes using the efficient influence function (EIF) and then obtains marginal quantiles via weighted quantile interpolation by inverting the distribution, and (2) a direct approach that solves the EIF of potential marginal quantiles. The proposed estimators are doubly robust and asymptotically normal. We further extend the framework to probability treatment effects (PTEs) and their conditional counterparts. For statistical inference, we investigate several variance estimation procedures, including EIF-based estimators, sandwich estimators, and the nonparametric bootstrap. Simulation studies illustrate that the empirical sandwich estimator and the nonparametric bootstrap provide doubly robust variance estimation with stable finite-sample performance under nuisance model misspecification. The proposed methods are evaluated through extensive Monte Carlo simulations and illustrated using an HIV data application.

Figures

Figures reproduced from arXiv: 2607.27633 by Bryan E. Shepherd, Chun Li, Hao Wu.

Figure 1
Figure 1. Figure 1: Estimated marginal and conditional distributional treatment effects of INSTI-based [PITH_FULL_IMAGE:figures/full_fig_p033_1.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

69 extracted references · 2 linked inside Pith

  1. [1]

    Kang, Joseph D. Y. and Schafer, Joseph L. , title =. Statistical Science , year =

  2. [2]

    American journal of epidemiology , volume=

    Doubly robust estimation of causal effects , author=. American journal of epidemiology , volume=. 2011 , publisher=

    2011

  3. [3]

    and Zivich, Paul N

    Shook-Sa, Bonnie E. and Zivich, Paul N. and Lee, Chanhwa and Xue, Keyi and Ross, Rachael K. and Edwards, Jessie K. and et al. , title =. Biometrics , year =

  4. [4]

    Epidemiology , volume=

    The consistency statement in causal inference: a definition or an assumption? , author=. Epidemiology , volume=. 2009 , publisher=

    2009

  5. [5]

    Causal Inference: What If , publisher =

    Hern. Causal Inference: What If , publisher =

  6. [6]

    Journal of the American statistical Association , volume=

    Estimation of regression coefficients when some regressors are not always observed , author=. Journal of the American statistical Association , volume=. 1994 , publisher=

    1994

  7. [7]

    American Journal of Epidemiology , volume=

    Defining and identifying average treatment effects , author=. American Journal of Epidemiology , volume=. 2023 , publisher=

    2023

  8. [8]

    Statistics in medicine , volume=

    Modeling continuous response variables using ordinal regression , author=. Statistics in medicine , volume=. 2017 , publisher=

    2017

  9. [9]

    Biometrika , volume=

    Efficient estimation of semiparametric transformation models for counting processes , author=. Biometrika , volume=. 2006 , publisher=

    2006

  10. [10]

    R package version , pages=

    Regression modeling strategies , author=. R package version , pages=. 2022 , publisher=

    2022

  11. [11]

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

    Maximum likelihood estimation in semiparametric regression models with censored data , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2007 , publisher=

    2007

  12. [12]

    Mathematics , volume=

    Asymptotic properties for cumulative probability models for continuous outcomes , author=. Mathematics , volume=. 2023 , publisher=

    2023

  13. [13]

    2025 , note =

    rms: Regression Modeling Strategies , author =. 2025 , note =

    2025

  14. [14]

    Statistics in medicine , volume=

    An empirical comparison of two novel transformation models , author=. Statistics in medicine , volume=. 2020 , publisher=

    2020

  15. [15]

    Statistics in medicine , volume=

    Model-assisted analyses of longitudinal, ordinal outcomes with absorbing states , author=. Statistics in medicine , volume=. 2022 , publisher=

    2022

  16. [16]

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

    An analysis of transformations , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1964 , publisher=

    1964

  17. [17]

    Journal of the American Statistical Association , volume=

    Addressing Multiple Detection Limits with Semiparametric Cumulative Probability Models , author=. Journal of the American Statistical Association , volume=. 2024 , publisher=

    2024

  18. [18]

    Biometrics , volume=

    Analyzing clustered continuous response variables with ordinal regression models , author=. Biometrics , volume=. 2023 , publisher=

    2023

  19. [19]

    Statistics in medicine , volume=

    Semiparametric linear transformation models: Effect measures, estimators, and applications , author=. Statistics in medicine , volume=. 2019 , publisher=

    2019

  20. [20]

    Journal of the Royal Statistical Society: Series B (Methodological) , volume=

    Regression models for ordinal data , author=. Journal of the Royal Statistical Society: Series B (Methodological) , volume=. 1980 , publisher=

    1980

  21. [21]

    Econometrica: Journal of the Econometric Society , pages=

    The asymptotic variance of semiparametric estimators , author=. Econometrica: Journal of the Econometric Society , pages=. 1994 , publisher=

    1994

  22. [22]

    Essay on principles , author=

    On the application of probability theory to agricultural experiments. Essay on principles , author=. Ann. Agricultural Sciences , pages=

  23. [23]

    , author=

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

    1974

  24. [24]

    Econometrica , pages=

    On the role of the propensity score in efficient semiparametric estimation of average treatment effects , author=. Econometrica , pages=. 1998 , publisher=

    1998

  25. [25]

    2011 , publisher=

    Targeted learning: causal inference for observational and experimental data , author=. 2011 , publisher=

    2011

  26. [26]

    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

  27. [27]

    Econometrica , volume=

    Efficient semiparametric estimation of quantile treatment effects , author=. Econometrica , volume=. 2007 , publisher=

    2007

  28. [28]

    Journal of Statistical Planning and Inference , volume=

    Efficient estimation of quantiles in missing data models , author=. Journal of Statistical Planning and Inference , volume=. 2017 , publisher=

    2017

  29. [29]

    Biometrika , volume=

    Inverting estimating equations for causal inference on quantiles , author=. Biometrika , volume=. 2025 , publisher=

    2025

  30. [30]

    Biometrics , volume=

    Causal inference on quantiles with an obstetric application , author=. Biometrics , volume=. 2012 , publisher=

    2012

  31. [31]

    2018 , publisher=

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

    2018

  32. [32]

    Journal of Statistical Software , volume=

    tmle: an R package for targeted maximum likelihood estimation , author=. Journal of Statistical Software , volume=

  33. [33]

    JAMA network open , volume=

    Use of quantile treatment effects analysis to describe antidepressant response in randomized clinical trials submitted to the US food and drug administration: a secondary analysis of pooled trial data , author=. JAMA network open , volume=. 2023 , publisher=

    2023

  34. [34]

    BMC Infectious Diseases , volume=

    Application of quantile mixed-effects model in modeling CD4 count from HIV-infected patients in KwaZulu-Natal South Africa , author=. BMC Infectious Diseases , volume=. 2022 , publisher=

    2022

  35. [35]

    Nature Medicine , pages=

    Genetic and molecular landscape of comorbidities in people living with HIV , author=. Nature Medicine , pages=. 2025 , publisher=

    2025

  36. [36]

    JAMA Internal Medicine , volume=

    Concerns With Analysis in Study of Tecovirimat for Mpox Among People With HIV , author=. JAMA Internal Medicine , volume=. 2024 , publisher=

    2024

  37. [37]

    Aids , volume=

    Heterogeneity in the costs of medical care among people living with HIV/AIDS in the United States , author=. Aids , volume=. 2019 , publisher=

    2019

  38. [38]

    Scientific Reports , volume=

    Combination of machine learning and Raman spectroscopy for prediction of drug release in targeted drug delivery formulations , author=. Scientific Reports , volume=. 2025 , publisher=

    2025

  39. [39]

    The annals of statistics , pages=

    Empirical probability plots and statistical inference for nonlinear models in the two-sample case , author=. The annals of statistics , pages=. 1974 , publisher=

    1974

  40. [40]

    Nonparametrics

    Statistical methods based on ranks , author=. Nonparametrics. San Francisco, CA, Holden-Day , volume=. 1975 , publisher=

    1975

  41. [41]

    Econometrica , volume=

    Identification and inference in nonlinear difference-in-differences models , author=. Econometrica , volume=. 2006 , publisher=

    2006

  42. [42]

    Review of Economics and Statistics , volume=

    Recovering distributions in difference-in-differences models: A comparison of selective and comprehensive schooling , author=. Review of Economics and Statistics , volume=. 2011 , publisher=

    2011

  43. [43]

    Econometrica , volume=

    Average and quantile effects in nonseparable panel models , author=. Econometrica , volume=. 2013 , publisher=

    2013

  44. [44]

    Econometrica , volume=

    Efficient estimation of average treatment effects using the estimated propensity score , author=. Econometrica , volume=. 2003 , publisher=

    2003

  45. [45]

    38) , author=

    Quantile regression (Econometric Society monographs; no. 38) , author=. 2005 , publisher=

    2005

  46. [46]

    Econometrica , volume=

    Unconditional quantile regressions , author=. Econometrica , volume=. 2009 , publisher=

    2009

  47. [47]

    European Sociological Review , volume=

    Quantile regression estimands and models: revisiting the motherhood wage penalty debate , author=. European Sociological Review , volume=. 2023 , publisher=

    2023

  48. [48]

    Journal of the American Statistical Association , volume=

    Adjusting for nonignorable drop-out using semiparametric nonresponse models , author=. Journal of the American Statistical Association , volume=. 1999 , publisher=

    1999

  49. [49]

    Biometrics , volume=

    Doubly robust estimation in missing data and causal inference models , author=. Biometrics , volume=. 2005 , publisher=

    2005

  50. [50]

    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

  51. [51]

    2016 , note=

    TMLE: as I (dimly) understand it , author=. 2016 , note=

    2016

  52. [52]

    Statistics in Medicine , volume=

    Inverse probability of treatment weighting with generalized linear outcome models for doubly robust estimation , author=. Statistics in Medicine , volume=. 2024 , publisher=

    2024

  53. [53]

    The International Journal of Biostatistics , volume=

    Targeted maximum likelihood learning , author=. The International Journal of Biostatistics , volume=. 2006 , doi=

    2006

  54. [54]

    Available at SSRN , year=

    Doubly-Robust Quantile Treatment Effects with Staggered Interventions , author=. Available at SSRN , year=

  55. [55]

    Available at SSRN 4884358 , year=

    Doubly-Robust Quantile Treatment Effect Estimation with Panel Data: A Difference-in-Difference Approach , author=. Available at SSRN 4884358 , year=

  56. [56]

    arXiv preprint arXiv:2307.01049 , year=

    Doubly robust estimation of direct and indirect quantile treatment effects with machine learning , author=. arXiv preprint arXiv:2307.01049 , year=

    Pith/arXiv arXiv

  57. [57]

    The American Statistician , volume=

    The calculus of M-estimation , author=. The American Statistician , volume=. 2002 , publisher=

    2002

  58. [58]

    1997 , publisher=

    Bootstrap methods and their application , author=. 1997 , publisher=

    1997

  59. [59]

    Biometrika , volume=

    Semiparametric counterfactual density estimation , author=. Biometrika , volume=. 2023 , publisher=

    2023

  60. [60]

    Biometrika , volume=

    Estimation of the probability of an event as a function of several independent variables , author=. Biometrika , volume=. 1967 , publisher=

    1967

  61. [61]

    Journal of the Royal Statistical Society: Series B (Methodological) , volume=

    A reliable data-based bandwidth selection method for kernel density estimation , author=. Journal of the Royal Statistical Society: Series B (Methodological) , volume=. 1991 , publisher=

    1991

  62. [62]

    arXiv preprint arXiv:2511.17907 , year=

    Why Is the Double-Robust Estimator for Causal Inference Not Doubly Robust for Variance Estimation? , author=. arXiv preprint arXiv:2511.17907 , year=

    arXiv

  63. [63]

    Asymptotic Methods , author=

    Improving Variance and Confidence Interval Estimation in Small-Sample Propensity Score Analyses: Bootstrap vs. Asymptotic Methods , author=. arXiv preprint arXiv:2511.10911 , year=

    Pith/arXiv arXiv

  64. [64]

    2018 , publisher=

    Density estimation for statistics and data analysis , author=. 2018 , publisher=

    2018

  65. [65]

    2015 , publisher=

    Multivariate Density Estimation: Theory, Practice, and Visualization , author=. 2015 , publisher=

    2015

  66. [66]

    2006 , publisher=

    Semiparametric theory and missing data , author=. 2006 , publisher=

    2006

  67. [67]

    Journal of the American Statistical Association , volume=

    Causal inference with general treatment regimes: Generalizing the propensity score , author=. Journal of the American Statistical Association , volume=. 2004 , publisher=

    2004

  68. [68]

    Journal of the Royal Statistical Society Series A: Statistics in Society , volume=

    Evaluating continuous training programmes by using the generalized propensity score , author=. Journal of the Royal Statistical Society Series A: Statistics in Society , volume=. 2012 , publisher=

    2012

  69. [69]

    BMC Medical Research Methodology , volume=

    Distributional regression in clinical trials: treatment effects on parameters other than the mean , author=. BMC Medical Research Methodology , volume=. 2022 , publisher=

    2022

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.