Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Bayesian Double Machine Learning for Causal Inference

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Bayesian Double Machine Learning recovers the causal effect of a treatment from the covariance of reduced-form errors, avoiding regularization-induced confounding while matching the large-sample guarantees of frequentist double machine…

desk verdict A clean and useful re-parameterization of DML with honest simulations, but the headline BvM/efficiency claims apply to a different posterior than the one implemented. read the letter →

arxiv 2508.12688 v1 pith:LYFT24IM submitted 2025-08-18 econ.EM

Francis J. DiTraglia , Laura Liu This is my paper · ORCID
classification econ.EM
keywords causalinferenceregularization-inducedconfoundingBayesianmethodsdoublemachinelearningpartiallylinearmodelBernstein-vonMisestheoremhigh-dimensionalregressionshrinkagepriors
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

The paper proposes and defends a fully Bayesian estimator, BDML, for the causal effect of a treatment $D$ on an outcome $Y$ in a partially linear model with many controls $X$: instead of shrinking the coefficient on $X$ in the structural regression $Y = \alpha D + X'\beta + \varepsilon$, it fits a bivariate reduced-form regression of $(Y,D)$ on $X$ and reads $\alpha$ from the error covariance, $\alpha = \sigma_{UV}/\sigma_V^2$. The authors argue this sidesteps regularization-induced confounding, the bias that arises when regularized estimators only partially adjust for controls, because independent shrinkage priors on the reduced-form coefficients do not imply a dogmatic prior of zero selection bias. They claim BDML has asymptotic bias of order $p^2/n^2$ rather than $p/n$, is $\sqrt{n}$-consistent when $p = o(n^{3/4})$, and satisfies a Bernstein–von Mises theorem that turns posterior credible sets into valid frequentist confidence intervals with semiparametric efficiency. If these claims hold, applied researchers get a Bayesian method that matches frequentist double machine learning asymptotically while giving better finite-sample uncertainty quantification. Simulations with $n=200$, $p=100$ show BDML with hierarchical shrinkage achieving near-nominal coverage, the lowest RMSE among competitors, and the shortest intervals among well-calibrated methods.

What carries the argument

The load-bearing object is the reduced-form bivariate regression of $(Y,D)$ on $X$, a pair of regression equations whose errors are allowed to correlate (a seemingly unrelated regression). Its error covariance is $\Sigma = \begin{pmatrix} \sigma_U^2 & \sigma_{UV} \\ \sigma_{UV} & \sigma_V^2 \end{pmatrix}$, and the causal effect is the ratio transform $\alpha = \sigma_{UV}/\sigma_V^2$ applied to posterior draws of $\Sigma$. This transform converts shrinkage from a problem about structural coefficients into a covariance-estimation problem: the reduced-form likelihood does not factorize, so independent priors on $\delta$ and $\gamma$ do not concentrate the prior on zero selection bias. Asymptotically, the posterior for $\Sigma$ is approximately Inverse-Wishart with an exponential tilting term $\exp(\mathrm{tr}(\tilde C_n \Sigma)/2)$; the tilting matrix encodes the shrinkage-induced shift that produces the $p^2/n^2$ bias, and the Bernstein–von Mises theorem verifies that the induced posterior of $\alpha$ is asymptotically normal around the true value.

What would settle it

Set known error variances, sub-Gaussian controls, and $p$ just below $\sqrt{n}$ (e.g., $n=400$, $p=19$), simulate many datasets with $\gamma^{*\prime}\beta^* \neq 0$, and measure the frequentist coverage of BDML's 95% credible interval; if coverage is systematically below nominal, the asserted Bernstein–von Mises theorem for the empirical $L^2$ norm is not valid. Separately, at $n=400,p=200$ versus $n=800,p=565$, compare BDML and naive ridge empirical bias: the paper predicts BDML bias decays like $p^2/n^2$ and the naive bias like $p/n$.

Watch

Extended reading notes

Core claim

The central discovery is that the causal parameter of a partially linear model is identified by the covariance of the errors of two reduced-form regressions, and that a Bayesian posterior over that covariance yields a posterior over the causal effect directly. Writing $Y_i = X_i'\delta + U_i$ and $D_i = X_i'\gamma + V_i$, the structural assumption $\mathrm{Cov}(\varepsilon_i, V_i)=0$ implies $\mathrm{Cov}(U_i,V_i)=\alpha \mathrm{Var}(V_i)$, hence $\alpha = \sigma_{UV}/\sigma_V^2$. This reparameterization changes what shrinkage does: priors on $\delta$ and $\gamma$ induce a dependent prior on $\beta = \delta - \alpha\gamma$, so the model is not Bayesian-ignorable and the prior on selection bias stays non-degenerate as $p$ grows. The paper shows the posterior mean of $\alpha$ is consistent, $\sqrt{n}$-consistent under $p = o(n^{3/4})$, has asymptotic bias of order $p^2/n^2$, and shares a common asymptotic variance with frequentist double machine learning and the naive estimator. Under a Bernstein–von Mises theorem stated for known error variances, the posterior of $\alpha$ is asymptotically normal, which makes BDML credible intervals asymptotically valid confidence intervals and establishes semiparametric efficiency.

Load-bearing premise

Everything about Bayesian credible intervals being valid frequentist confidence intervals depends on an asserted but unproved adaptation of an existing Bernstein–von Mises theorem to the empirical $L^2$ norm under known error variances, so if that adaptation fails the inference guarantee collapses.

Editorial extensions

If this is right

  • BDML is $\sqrt{n}$-consistent whenever $p = o(n^{3/4})$, while the naive ridge-style estimator requires $p/\sqrt{n} \to 0$; for growth of $p$ between $\sqrt{n}$ and $n^{3/4}$, the naive estimator is not $\sqrt{n}$-consistent and BDML is.
  • BDML and FDML share an asymptotic bias of order $p^2/n^2$ and the same asymptotic variance, strictly beating the naive estimator's $p/n$ bias; consequently BDML dominates on asymptotic mean squared error.
  • Under the Bernstein–von Mises conditions, BDML's equal-tailed credible sets are asymptotically valid frequentist confidence intervals and the estimator attains semiparametric efficiency, so Bayesian inference can be reported with frequentist guarantees.
  • In the paper's simulation study, BDML with hierarchical shrinkage achieves coverage near 0.94, the lowest RMSE among all seven methods, and the shortest average interval width among methods with near-nominal coverage, across all values of the structural error variance $\sigma_\varepsilon$ studied.
  • The generative formulation allows hierarchical priors to adapt shrinkage to the data; the hierarchical version, which lets the variances of $\delta$ and $\gamma$ differ, is the best-performing method in the simulations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial extension: the identification $\alpha = \sigma_{UV}/\sigma_V^2$ suggests a general template for Bayesian double debiasing—any causal parameter expressible as a ratio of reduced-form covariances, such as an instrumental-variables local average treatment effect, could be handled by the same covariance-posterior device. The paper does not develop this extension.
  • The bias calculations imply a practical calibration rule: prior precisions should be proportional to $p$ to keep shrinkage bias at order $p^2/n^2$. A reader could test whether BDML's advantage over the naive estimator grows with $p/n$ beyond the $1/2$ ratio used in the simulations.
  • The selection-bias result yields a diagnostic applicable before estimation: on a given dataset, compute how concentrated the naive prior is on zero selection bias relative to the BDML prior; the choice of method should matter most in designs with strong confounding, a prediction that could be checked by reweighting the simulation design.
  • The Bernstein–von Mises proposition assumes known error variances $\sigma_\varepsilon^2$ and $\sigma_V^2$; until a version that integrates out these variances is proved, the semiparametric efficiency claim should be read as conditional on that assumption.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes Bayesian Double Machine Learning (BDML) for the partially linear model with high-dimensional controls. Instead of a structural regression of Y on (D, X), the method specifies a bivariate reduced-form regression of (Y, D) on X and recovers the causal parameter as the posterior ratio α = Σ12/Σ22 from the error covariance matrix. The authors argue that this reparameterization avoids regularization-induced confounding, that the implied prior on selection bias is non-dogmatic, and that the resulting estimator matches frequentist double machine learning in asymptotic bias (p²/n² versus p/n for a naive ridge/Bayes estimator) while improving finite-sample RMSE and coverage. The paper also claims a Bernstein-von Mises theorem, asymptotic normality, semiparametric efficiency, and robustness to misspecification, and it reports a simulation study comparing seven estimators.

Significance. If the advertised properties hold, BDML is a valuable and simple contribution: it gives applied researchers a fully Bayesian, conjugate-implementable procedure that avoids the tuning and cross-fitting sensitivities of frequentist DML while retaining its first-order guarantees. The core parameterization is clean, the posterior algebra in Section 4 is transparent, the distinction between naive shrinkage priors and reduced-form priors is well explained, and the simulation design is informative and clearly reported. The paper also gives explicit, falsifiable asymptotic rate claims (p/n versus p²/n²) that are directly comparable to the DML literature. However, the most consequential advertised guarantees—frequentist validity of credible intervals and semiparametric efficiency—are not actually proven for the algorithm as implemented, because the Bernstein-von Mises theorem is stated for a different posterior (known error variances) than the one used in Algorithm 1 and in the simulations.

major comments (3)
  1. [Section 5.4, Proposition 7] Proposition 7 assumes that σ²*ε and σ²*V are known and places a prior directly on α, whereas Algorithm 1 and the Section 6 simulations estimate the full covariance matrix Σ (via inverse-Wishart or LKJ/Cauchy priors) and form α as the ratio Σ12/Σ22. No theorem in the paper derives the posterior distribution of that ratio or shows it has the normal limit stated in Proposition 7. The proof sketch only says that Walker's assumptions are verified, and Remark 7 asserts without demonstration that Walker's proofs can be adapted from sup-norm to empirical L2 neighborhoods. The appendix verifies some empirical-process bounds, but it does not bridge the gap between the known-variance posterior and the unknown-Σ ratio posterior. Since the abstract's claims of asymptotic normality, semiparametric efficiency, and 'valid frequentist confidence intervals' rest on this proposition, the central frequentist guarantee is not established for the method as proposed.
  2. [Section 5.3, Propositions 4–5 and Appendix A] The derivation of the posterior bias of BDML relies on several uncontrolled approximations: the expansion |I−A|≈1−tr(A), a first-order Picard iteration for the posterior mode, and the identification of the posterior mean of Σ with the posterior mode plus a term of order O(1/n). Proposition 5 then asserts the p²/n² bias claim. These steps are heuristic rather than rigorous, and the claimed order of the bias, which is a headline result, is therefore not a theorem. The authors should either provide a complete proof with explicit remainders under Assumptions 2–5 or clearly state Proposition 5 as a heuristic expansion.
  3. [Section 5.4, Remark 7 and proof of Proposition 7] Even taking the known-variance setup at face value, the adaptation of Walker (2025) to the empirical L2 norm is not established. The sentence 'All proofs in Walker (2024) can be adapted' is not a proof, and the paper does not state the precise conditions on the prior, the function class, or the metric that would make Walker's theorems apply verbatim. Given that this adaptation is load-bearing for the efficiency claim, the authors should provide a self-contained statement of which Walker assumptions hold and where the empirical-L2 modification enters each verification.
minor comments (6)
  1. [Section 3, paragraph after Eq. (11)] The sentence beginning 'unless we specify a prior in which β and γ' is incomplete and garbled: it reads 'Treating σ2ε as known' mid-sentence. The intended logical point about Bayesian ignorability should be restated.
  2. [Abstract and Section 6] The abstract promises 'a number of empirical examples drawn from the recent literature,' but Section 6 contains only simulations. Either add the empirical applications or remove the promise from the abstract.
  3. [Equation (14)] In the second display of Eq. (14), 'Var(Ui) = Var(εi + αVi) = Var(Ui) + α²Var(Vi)' should read 'Var(εi) + α²Var(Vi)'.
  4. [Remark 2 and Appendix A] Remark 2 says that 'if C̃n is negative definite, then X follows a multivariate Generalized Inverse Gaussian distribution,' but the subject should be Σ, not the data matrix X.
  5. [References and proof of Proposition 7] The paper cites Walker (2025) in the introduction and references, but the proof of Proposition 7 and Remark 7 repeatedly refer to 'Walker (2024)'. The citation should be made consistent.
  6. [Figure 1 and Table 1] There are typographical errors in the captions: 'T able 1' in the table caption and 'labled' in the figure caption.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BDML's target is a direct reparameterization of the reduced-form covariance, and the BvM theorem rests on an external source.

full rationale

The central BDML construction is not circular. Equation (15) defines alpha as sigma_UV / sigma^2_V, which is an algebraic identity following from the reduced-form covariance parameterization in (13)-(14); the posterior for alpha is then obtained by transforming draws from the posterior of Sigma under standard conditionally conjugate priors. No fitted constants are fed back into the derivation, and the simulation priors (LKJ(4), Cauchy(0,2.5), inverse-gamma hyperpriors) are fixed defaults rather than tuned to the test data. The Bernstein-von Mises result in Proposition 7 cites Walker (2025), an independent external theorem, and the proof verifies Walker's assumptions rather than importing a result from the authors' own prior work. The known-variance assumption in Proposition 7 versus the unknown-Sigma implementation in Algorithm 1, together with Remark 7's asserted adaptation from sup-norm to empirical L2 neighborhoods, is a real gap between theorem and procedure, but it is a correctness and robustness concern, not circularity: the claimed posterior limiting distribution is not shown to be identical to an input by construction. Likewise, the abstract's promise of empirical examples is not fulfilled by Section 6, which contains only simulations, but this is a completeness issue. No step in the paper reduces a claimed prediction to its own inputs, so the appropriate circularity score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on prior hyperparameters and high-dimensional regularity conditions, but no new physical or conceptual entities are introduced, and no constants are fitted to data. The main unproved ingredient is the asserted adaptation of Walker's BvM theorem to the empirical L2 norm.

free parameters (4)
  • Prior precisions τδ, τγ, λ (theory)
    In the asymptotic analysis (Assumption 5), the prior precisions for the reduced-form coefficients and the ridge penalty are assumed to grow at rate p while being o(n); these rate conditions are needed for the bias and BvM results. They are not estimated from data.
  • Inverse-Wishart hyperparameters ν0, Σ0
    The conditionally conjugate prior in (19) uses user-specified ν0 and scale Σ0; Proposition 5's expansion includes a Σ0/n term, so the posterior mean depends on this choice.
  • BDML-Basic priors: Normal(0,5²) on δ and γ; LKJ(4); Cauchy(0,2.5) on scales = chosen
    In the simulation, these constants are fixed by the authors; they affect the finite-sample results but are not fit to the data.
  • BDML-Hier hyperpriors: Inverse-Gamma(2,2) on σ²_δ and σ²_γ
    Used for hierarchical shrinkage in the simulation; chosen by convention.
assumptions (4)
  • ad hoc to paper Walker (2025) BvM theorem for the partially linear model can be extended to the empirical L2 norm and to the linear reduced-form setup with unbounded X.
    Used in Proposition 7 and its proof; the paper states 'All proofs in Walker (2024) can be adapted to the empirical L2-norm' without providing the adaptation.
  • domain assumption The true reduced-form coefficients δ* and γ* are square-summable sequences in ℓ2 and the spectral distribution of ΣX converges nicely (Assumptions 3 and 4).
    These high-dimensional regularity conditions are required for the asymptotic bias and BvM theorems; they are standard in the DML literature but are not derived.
  • domain assumption The prior precisions satisfy τδ, τγ, λ = o(n) and ≍ p (Assumption 5).
    These rate restrictions shape the prior; they are needed for the p/n and p²/n² bias orders but are not implied by the model.
  • standard math Standard Bayesian updating with a normal likelihood and conjugate priors yields the posterior used in Algorithm 1.
    Foundation of the method; well-established but unproved in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian Double Machine Learning for Causal Inference." pith.science (2026). https://pith.science/paper/LYFT24IM

@misc{pith2026250812688,
  author       = {Pith},
  title        = {Pith review of: Bayesian Double Machine Learning for Causal Inference},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LYFT24IM}},
  note         = {Machine review of arXiv:2508.12688}
}
read the original abstract

This paper proposes a simple, novel, and fully-Bayesian approach for causal inference in partially linear models with high-dimensional control variables. Off-the-shelf machine learning methods can introduce biases in the causal parameter known as regularization-induced confounding. To address this, we propose a Bayesian Double Machine Learning (BDML) method, which modifies a standard Bayesian multivariate regression model and recovers the causal effect of interest from the reduced-form covariance matrix. Our BDML is related to the burgeoning frequentist literature on DML while addressing its limitations in finite-sample inference. Moreover, the BDML is based on a fully generative probability model in the DML context, adhering to the likelihood principle. We show that in high dimensional setups the naive estimator implicitly assumes no selection on observables--unlike our BDML. The BDML exhibits lower asymptotic bias and achieves asymptotic normality and semiparametric efficiency as established by a Bernstein-von Mises theorem, thereby ensuring robustness to misspecification. In simulations, our BDML achieves lower RMSE, better frequentist coverage, and shorter confidence interval width than alternatives from the literature, both Bayesian and frequentist.

Figures

Figures reproduced from arXiv: 2508.12688 by the authors.

Figure 1
Figure 1. This figure presents results for the estimators described in section 6 and simulation from (20)–(21) over a grid of values for σε, the error variance in the structural equation given in (4). The panel labled RMSE gives the root mean-squared error of the estimator for α under each method, while those labled Coverage and Avg. Width give the coverage probability and average width of corresponding (nominal) 95% confiden… view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Decision Theoretic Subgroup Detection With Bayesian Machine Learning

    stat.ME 2025-09 conditional novelty 6.0 of 10

    A new BRAIDS utility interpolates between risk-seeking and risk-averse subgroup detection, and regularized Bayesian models can keep nominal coverage for subgroup effects without sample splitting.

Reference graph

Works this paper leans on

35 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    , author Chernozhukov, V

    author Ahrens, A. , author Chernozhukov, V. , author Hansen, C. , author Kozbur, D. , author Schaffer, M. , author Wiemann, T. , year 2025 . title An Introduction to Double / Debiased Machine Learning . http://arxiv.org/abs/2504.08324, :10.48550/arXiv.2504.08324. note arXiv:2504.08324 [econ]

    work page doi:10.48550/arxiv.2504.08324 2025

  2. [2]

    , author Durvasula, M

    author Alsan, M. , author Durvasula, M. , author Gupta, H. , author Schwartzstein, J. , author Williams, H. , year 2023 . title Representation and extrapolation: Evidence from clinical trials* . journal The Quarterly Journal of Economics volume 139 , pages 575--635 . https://doi.org/10.1093/qje/qjad036, :10.1093/qje/qjad036, http://arxiv.org/abs/https://a...

    work page doi:10.1093/qje/qjad036 2023

  3. [3]

    , author Frandsen, B

    author Angrist, J.D. , author Frandsen, B. , year 2022 . title Machine Labor . journal Journal of Labor Economics volume 40 , pages S97--S140 . https://www.journals.uchicago.edu/doi/10.1086/717933, :10.1086/717933

    work page doi:10.1086/717933 2022

  4. [4]

    , author Papadogeorgou, G

    author Antonelli, J. , author Papadogeorgou, G. , author Dominici, F. , year 2022 . title Causal inference in high dimensions: A marriage between Bayesian modeling and good frequentist properties . journal Biometrics volume 78 , pages 100--114 . https://academic.oup.com/biometrics/article/78/1/100-114/7459997, :10.1111/biom.13417

    doi:10.1111/biom.13417 2022

  5. [5]

    , author Coutts, A

    author Armand, A. , author Coutts, A. , author Vicente, P.C. , author Vilela, I. , year 2020 . title Does information break the political resource curse? experimental evidence from mozambique . journal American Economic Review volume 110 , pages 3431–53 . https://www.aeaweb.org/articles?id=10.1257/aer.20190842, :10.1257/aer.20190842

    work page doi:10.1257/aer.20190842 2020

  6. [6]

    , author Bach, P

    author Bach, P. , author Bach, P. , author De, U.H. , author Chernozhukov, V. , year 2024 . title Hyperparameter Tuning for Causal Inference with Double Machine Learning : A Simulation Study . journal Proceedings of Machine Learning Research volume 236 , pages 1065--1117

    work page 2024

  7. [7]

    , author Chernozhukov, V

    author Belloni, A. , author Chernozhukov, V. , author Hansen, C. , year 2014 . title Inference on Treatment Effects after Selection among High - Dimensional Controls . journal The Review of Economic Studies volume 81 , pages 608--650 . https://academic.oup.com/restud/article-lookup/doi/10.1093/restud/rdt044, :10.1093/restud/rdt044

    doi:10.1093/restud/rdt044 2014

  8. [8]

    , author Liu, R

    author Breunig, C. , author Liu, R. , author Yu, Z. , year 2024 . title Semiparametric Bayesian Difference -in- Differences https://christophbreunig.com/wp-content/uploads/2024/12/bayes_did.pdf

    work page 2024

Show all 35 references
  1. [9]

    , author Liu, R

    author Breunig, C. , author Liu, R. , author Yu, Z. , year 2025 . title Double Robust Bayesian Inference on Average Treatment Effects . journal Econometrica volume 93 , pages 539--568 . https://www.econometricsociety.org/doi/10.3982/ECTA21442, :10.3982/ECTA21442

    2025 doi

  2. [10]

    , author Chetverikov, D

    author Chernozhukov, V. , author Chetverikov, D. , author Demirer, M. , author Duflo, E. , author Hansen, C. , author Newey, W. , author Robins, J. , year 2018 . title Double/debiased machine learning for treatment and structural parameters . journal The Econometrics Journal v...

    2018 doi

  3. [11]

    , author Contreras-Loya, D

    author Chioda, L. , author Contreras-Loya, D. , author Gertler, P. , author Carney, D. , year 2021 . title Making Entrepreneurs: Returns to Training Youth in Hard Versus Soft Business Skills . type Working Paper number 28845 . National Bureau of Economic Research. http://www.n...

    2021 doi

  4. [12]

    , author Courtemanche, C

    author Cotti, C. , author Courtemanche, C. , author Liang, Y. , author Maclean, J.C. , author Nesson, E. , author Sabia, J.J. , year 2025 . title The effect of e-cigarette flavor bans on tobacco use . journal Journal of Health Economics volume 102 , pages 103013 . https://www....

    2025

  5. [13]

    , year 2024

    author Delfino, A. , year 2024 . title Breaking gender barriers: Experimental evidence on men in pink-collar jobs . journal American Economic Review volume 114 , pages 1816–53 . https://www.aeaweb.org/articles?id=10.1257/aer.20220582, :10.1257/aer.20220582

    2024 doi

  6. [14]

    , author Jacobs, J

    author Dube, A. , author Jacobs, J. , author Naidu, S. , author Suri, S. , year 2020 . title Monopsony in online labor markets . journal American Economic Review: Insights volume 2 , pages 33–46 . https://www.aeaweb.org/articles?id=10.1257/aeri.20180150, :10.1257/aeri.20180150

    2020 doi

  7. [15]

    , author Dupas, P

    author Duflo, E. , author Dupas, P. , author Kremer, M. , year 2021 . title The Impact of Free Secondary Education: Experimental Evidence from Ghana . type NBER Working Papers number 28937 . National Bureau of Economic Research, Inc. https://ideas.repec.org/p/nbr/nberwo/28937....

    2021

  8. [16]

    , author Moakler, R

    author Gordon, B.R. , author Moakler, R. , author Zettelmeyer, F. , year 2023 . title Close enough? a large-scale exploration of non-experimental approaches to advertising measurement . journal Marketing Science volume 42 , pages 768--793 . https://doi.org/10.1287/mksc.2022.14...

    2023

  9. [17]

    , author Carvalho, C.M

    author Hahn, P.R. , author Carvalho, C.M. , author Puelz, D. , author He, J. , year 2018 . title Regularization and Confounding in Linear Regression for Treatment Effect Estimation . journal Bayesian Analysis volume 13 . https://projecteuclid.org/journals/bayesian-analysis/vol...

    2018 doi

  10. [18]

    , author Murray, J.S

    author Hahn, P.R. , author Murray, J.S. , author Carvalho, C.M. , year 2020 . title Bayesian Regression Tree Models for Causal Inference : Regularization , Confounding , and Heterogeneous Effects (with Discussion ) . journal Bayesian Analysis volume 15 . https://projecteuclid....

    2020 doi

  11. [19]

    , author Lopes, H.F

    author Heckman, J.J. , author Lopes, H.F. , author Piatek, R. , year 2014 . title Treatment Effects : A Bayesian Perspective . journal Econometric Reviews volume 33 , pages 36--67 . http://www.tandfonline.com/doi/abs/10.1080/07474938.2013.807103, :10.1080/07474938.2013.807103

    2014

  12. [20]

    , author Kelley, E.M

    author Hussam, R. , author Kelley, E.M. , author Lane, G. , author Zahra, F. , year 2022 . title The psychosocial value of employment: Evidence from a refugee camp . journal American Economic Review volume 112 , pages 3694–3724 . https://www.aeaweb.org/articles?id=10.1257/aer....

    2022 doi

  13. [21]

    , author Müller, U.K

    author Kolesár, M. , author Müller, U.K. , author Roelsgaard, S.T. , year 2025 . title The Fragility of Sparsity . http://arxiv.org/abs/2311.02299, :10.48550/arXiv.2311.02299. note arXiv:2311.02299 [econ]

    2025 doi arXiv

  14. [22]

    , author Ding, P

    author Li, F. , author Ding, P. , author Mealli, F. , year 2023 . title Bayesian causal inference: a critical review . journal Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences volume 381 , pages 20220153 . https://royalsocietyp...

    2023

  15. [23]

    , year 2023

    author Linero, A.R. , year 2023 . title In Nonparametric and High - Dimensional Models , Bayesian Ignorability is an Informative Prior . journal Journal of the American Statistical Association , pages 1--14 https://www.tandfonline.com/doi/full/10.1080/01621459.2023.2278202, :1...

    2023

  16. [24]

    , author Antonelli, J.L

    author Linero, A.R. , author Antonelli, J.L. , year 2023 . title The how and why of Bayesian nonparametric causal inference . journal WIREs Computational Statistics volume 15 , pages e1583 . https://wires.onlinelibrary.wiley.com/doi/10.1002/wics.1583, :10.1002/wics.1583

    2023 doi

  17. [25]

    , author Graham, D.J

    author Luo, Y. , author Graham, D.J. , author McCoy, E.J. , year 2023 . title Semiparametric Bayesian doubly robust causal estimation . journal Journal of Statistical Planning and Inference volume 225 , pages 171--187 . https://linkinghub.elsevier.com/retrieve/pii/S03783758220...

    2023 doi

  18. [26]

    , author Van Der Vaart, A

    author Ray, K. , author Van Der Vaart, A. , year 2020 . title Semiparametric Bayesian causal inference . journal The Annals of Statistics volume 48 . https://projecteuclid.org/journals/annals-of-statistics/volume-48/issue-5/Semiparametric-Bayesian-causal-inference/10.1214/19-A...

    2020 doi

  19. [27]

    , author Belzile, L.R

    author Saarela, O. , author Belzile, L.R. , author Stephens, D.A. , year 2016 . title A Bayesian view of doubly robust causal inference . journal Biometrika volume 103 , pages 667--681 . https://academic.oup.com/biomet/article-lookup/doi/10.1093/biomet/asw025, :10.1093/biomet/asw025

    2016 doi

  20. [28]

    , author Xiu, D

    author Shen, Z. , author Xiu, D. , year 2024 . title Can Machines Learn Weak Signals ? type Technical Report . https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4748784

    2024

  21. [29]

    , year 2012

    author Sims, C.A. , year 2012 . title ROBINS - WASSERMAN , ROUND N

    2012

  22. [30]

    , year 2025

    author Walker, C.D. , year 2025 . title Parametrization, Prior Independence , and the Semiparametric Bernstein -von Mises Theorem for the Partially Linear Model . http://arxiv.org/abs/2306.03816, :10.48550/arXiv.2306.03816. note arXiv:2306.03816 [math]

    2025 doi arXiv

  23. [31]

    , author Zhu, Y

    author Wüthrich, K. , author Zhu, Y. , year 2023 . title Omitted Variable Bias of Lasso - Based Inference Methods : A Finite Sample Analysis . journal Review of Economics and Statistics volume 105 , pages 982--997 . https://direct.mit.edu/rest/article/105/4/982/107662/Omitted-...

    2023 doi

  24. [32]

    , author Chuang, H.C

    author Yang, J.C. , author Chuang, H.C. , author Kuan, C.M. , year 2020 . title Double machine learning with gradient boosting and its application to the big n audit quality effect . journal Journal of Econometrics volume 216 , pages 268--283 . https://www.sciencedirect.com/sc...

    2020 doi

  25. [33]

    , year 1971

    author Zellner, A. , year 1971 . title An Introduction to Bayesian Inference in Econometrics . publisher John Wiley & Sons , address New York

    1971

  26. [34]

    , author Naughton, B.P

    author Zhang, Y.D. , author Naughton, B.P. , author Bondell, H.D. , author Reich, B.J. , year 2022 . title Bayesian Regression Using a Prior on the Model Fit : The R2 - D2 Shrinkage Prior . journal Journal of the American Statistical Association volume 117 , pages 862--874 . h...

    2022

  27. [35]

    , year 2016

    author Zigler, C.M. , year 2016 . title The Central Role of Bayes ’ Theorem for Joint Estimation of Causal Effects and Propensity Scores . journal The American Statistician volume 70 , pages 47--54 . https://www.tandfonline.com/doi/full/10.1080/00031305.2015.1111260, :10.1080/...

    2016 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.