Target-Weighted Neyman Allocation: Experimental Design for Heterogeneous Treatment Effects under Population Shift

Hoang Dang; Luan Pham; Minh Nguyen

Target-Weighted Neyman Allocation: Experimental Design for Heterogeneous Treatment Effects under Population Shift

T0 review · 0 major / 5 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 This paper proves the closed-form optimal allocation for target-weighted group treatment effects under population shift: within each group, treat at the Neyman fraction σ₁ₖ/(σ₁ₖ+σ₀ₖ), and across groups, allocate sample proportional to…

desk verdict Correct, well-scoped closed-form allocation for target-weighted GATE precision; the transportability assumption is the main caveat but does not undermine the result. read the letter →

arxiv 2608.06512 v1 pith:ZRFIHGS5 submitted 2026-08-06 cs.LG stat.ME

Hoang Dang , Luan Pham , Minh Nguyen This is my paper · ORCID

classification cs.LGstat.ME

keywords experimentaldesignheterogeneoustreatmenteffectspopulationshiftgeneralizationNeymanallocationstratifiedrandomizedexperimentstwo-stagedeployment-weightedprecision

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 asks how to split a fixed experimental budget across pre-specified groups and treatment arms when the experiment will guide decisions in a deployment population whose group composition differs from the experimental one. It shows that the design minimizing target-weighted mean squared error of group average treatment effects has a closed form: within each group, assign treatment in the Neyman proportion σ₁ₖ/(σ₁ₖ+σ₀ₖ); across groups, allocate sample proportional to √qₖ(σ₁ₖ+σ₀ₖ), the geometric compromise between deployment importance and measurement difficulty. A two-stage procedure that first runs a small pilot to estimate group–arm variances then plugs them in provably converges to this oracle as the pilot grows, and the rule degrades gracefully when deployment weights are misspecified. This matters because standard designs that allocate by experimental proportions, deployment proportions, or variance alone can lose 30–60% relative precision exactly when deployment importance and statistical difficulty coincide.

What carries the argument

The central object is the target-weighted group average treatment effect risk B_Q(ρ,e) = Σₖ (qₖ/ρₖ)(σ₁ₖ²/eₖ + σ₀ₖ²/(1−eₖ)), the leading-order variance of the stratified difference-in-means estimator for fixed groups, weighted by deployment shares qₖ. Minimizing it factorizes: the within-group Neyman treatment fraction e*ₖ = σ₁ₖ/(σ₁ₖ+σ₀ₖ) is obtained first by Cauchy–Schwarz on each group, leaving the group-allocation problem Σₖ qₖsₖ²/ρₖ whose minimizer ρ*ₖ ∝ √qₖsₖ, with sₖ = σ₁ₖ+σ₀ₖ, is the equality case of the Engel form of Cauchy–Schwarz. All extensions—the weight-robust average/worst-case/ignorance designs, the plug-in regret bound, and the saddle-point proofs—reuse the same factorization.

What would settle it

Evaluate the oracle formula against numerical grid-search minimization of B_Q(ρ,e) on thousands of random (q,σ) draws; any mismatch falsifies Theorem 1, and observing in a deployment population a group whose treatment effect differs from the experimental estimate would falsify the transportability premise the design rests on.

Watch

Extended reading notes

Core claim

The paper's central discovery is a complete solution to a design problem: minimize B_Q(ρ,e) = Σₖ (qₖ/ρₖ)(σ₁ₖ²/eₖ + σ₀ₖ²/(1−eₖ)) over group shares ρ and treatment probabilities e, the leading-order variance of the stratified difference-in-means estimator for a fixed vector of group effects weighted by deployment shares. Two Cauchy–Schwarz steps give the unique minimizer: e*ₖ = σ₁ₖ/(σ₁ₖ+σ₀ₖ) and ρ*ₖ = √qₖ(σ₁ₖ+σ₀ₖ)/Σₗ √qₗ(σ₁ₗ+σ₀ₗ). The result specializes classical Neyman allocation to a target-weighted GATE loss, introducing the √qₖ factor that balances deployment importance against statistical difficulty. The paper further shows that the plug-in two-stage rule recovers the oracle as pilot variance estimates stabilize, that deployment-weight uncertainty is handled by mean, least-favorable, or equalizing weights with closed forms, and that the proportional gains are largest when deployment importance and variance difficulty are aligned.

Load-bearing premise

The design assumes that group-specific treatment effects transport unchanged from the experimental population to the deployment population; if a group's effect differs in deployment, the allocation optimizes precision for the wrong target quantities, and this transportability premise is untestable from experimental data alone.

Editorial extensions

If this is right

If the oracle rule is correct, a designer who knows or can estimate deployment shares and group–arm variances should allocate sample proportionally to √qₖ(σ₁ₖ+σ₀ₖ) and split treatment within each group by the Neyman fraction, with deployment-important and hard-to-measure groups receiving more units.
The two-stage plug-in procedure converges to the oracle allocation as pilot size grows, with design regret O_p(n₀^{-1/2}) in general and O_p(n₀^{-1}) at an interior optimum, so a few dozen observations per group–arm cell capture most of the attainable gain.
Misspecifying deployment weights cannot cost more than discarding them entirely: with uniform input weights the rule reduces to the q-free HHK plug-in, and for any partial misspecification the √q margin remains strictly beneficial.
Under complete ignorance of the deployment composition, the equalizing rule ρₖ ∝ (σ₁ₖ+σ₀ₖ)² keeps target-weighted risk constant across all compositions, and the minimax rule traces the lower envelope of worst-case risk.
The cross-group allocation margin composes with within-group stratification: applying a stratification-tree refinement on TWNA's composition performs no worse than on uniform composition in the paper's benchmarks, with the gain coming from the cross-group margin.

Reading between the lines

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

A natural extension is to relax the transportability assumption itself, for instance by estimating group-specific drift between experimental and deployment populations from auxiliary observational data and inserting those shifted estimands into the same closed-form allocation; the √qₖ mechanism would survive, but the weights would target shifted quantities.
The same Cauchy–Schwarz allocation logic should apply to other deployment-weighted objectives, such as weighted policy regret or weighted average treatment effects, where the √qₖ factor would be modified by the objective's curvature; the paper's weight-robust Proposition 1 already provides the template for convex ambiguity sets.
The equalizer rule ρₖ ∝ sₖ² suggests a testable design principle: when the deployment mix is genuinely unknown, allocating by squared total variance is a no-regret choice that an always-on platform with fixed traffic composition could benchmark against post-hoc correction.
The plug-in's tail fragility in spike-mixture settings, with coefficient of variation near 6 in the aligned platform cell, indicates that practical deployments should pair the plug-in rule with a robust scale estimator whenever pilot cells can miss rare high-variance events; the paper's robust variant is one such fix, and other shrinkage or Bayesian estimators are worth experimenting with.

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.

Desk Editor's Note

read the letter

Here's my read. The central result is Theorem 1: under composition shift, the target-weighted GATE risk is minimized by the within-group Neyman split and cross-group shares proportional to sqrt(q_k)(sigma1k+sigma0k). That derivation is correct, two Cauchy-Schwarz steps with no fitted constants. The paper is also honest about the lineage: Cytrynbaum already had the ATE version and Shi-Lin had q-linear rules; the sqrt(q) weighting is the genuine new element. That makes it a useful extension rather than a paradigm shift.

What the paper does well: the plug-in procedure and regret bound are standard but carefully done, with explicit positivity assumptions, clipping, and an allocation floor. The empirical work is unusually thorough: a simulation grid over budget and pilot size, an end-to-end check with realized sampling rather than first-order draws, real-covariate benchmarks on IHDP and LaLonde, and a battery of stress tests including heavy tails and contaminated pilots. The results behave the way the theory says, and the paired comparisons isolate where each margin contributes. The 13-44% effective sample-size savings is a fair way to state the gain.

Soft spots. The transportability condition—group-specific treatment effects must be the same in P and Q—is the load-bearing assumption. If it fails, the target estimands are not identified from the experimental sample and TWNA optimizes precision for the wrong quantities. The paper states this explicitly, but every simulation and benchmark generates data under that assumption, so the failure mode is never exercised. That is a genuine limitation, though shared by the transportability literature rather than a flaw in the math. Minor: the repository is referenced but no link is provided and no code is shipped, so the reproducibility appendix is weaker than it looks. The learned-partition proxy baseline also carries a known composition bias, which makes it a soft comparison.

Who should read this: anyone designing stratified experiments for target-weighted GATEs with credible deployment weights and pilot variance estimates. The paper is honest about the gap between what it delivers and the harder problem of non-compositional shift. It deserves a serious referee and, after the caveat is moved forward and the code is shared, publication.

Referee Report

0 major / 5 minor

Summary. This paper proposes TWNA, a two-stage stratified experimental design for estimating group average treatment effects (GATEs) when the deployment population differs from the experimental population in group composition. The oracle design minimizes the leading term of the target-weighted GATE risk and takes the closed form e*_k = σ1k/(σ1k + σ0k) and ρ*_k ∝ √q_k(σ1k + σ0k), with the usual Neyman treatment split within each group. The plug-in rule replaces the oracle variances with pilot estimates, and the paper proves consistency and an O_p(n0^{-1/2}) regret bound (with an O_p(n0^{-1}) refinement at an interior optimum). It also extends the design to deployment-weight uncertainty via average-case, minimax, and full-ignorance rules, and introduces a winsorized 'robust TWNA' variant for small or heavy-tailed pilot cells. The empirical evaluation includes a full simulation grid, an end-to-end validation with realized assignment, a misspecified-weight study, and three covariate-rich benchmarks (calibrated synthetic, IHDP, LaLonde), all showing gains for TWNA over uniform, deployment-only, variance-only, and ATE-optimal designs.

Significance. The central derivation is correct and complete: Theorem 1 is proved by two tight Cauchy–Schwarz steps, the regret proof in Appendix A is careful about clipping and floor constraints, and Proposition 1 gives a clean minimax treatment of weight uncertainty. The oracle allocation is parameter-free given the variances and weights, and the plug-in rule is explicitly evaluated against oracle and external baselines. The empirical work is unusually extensive: a full budget-by-pilot grid, paired tests, machine-reproducible code, and stress tests outside the theorem's clean conditions. The main limitation is that the target estimands are identified only under the explicitly stated composition-shift assumption (E_Q{Y(1)−Y(0)|S=k} = E_P{Y(1)−Y(0)|S=k}), which is untestable from the experimental data alone; this is a scope condition rather than an internal inconsistency, and it does not affect the validity of the derivations conditional on that assumption.

minor comments (5)

[Method, estimand definition] The estimands τ_k are defined without a subscript indicating the distribution; since the paper later uses E_Q{Y(1)−Y(0)|S=k} = E_P{Y(1)−Y(0)|S=k}, please introduce them explicitly under the deployment population Q and add a sentence in the Discussion noting that this transportability condition is untestable from the pilot and final experimental data alone.
[Table 1 caption] In the 'Variance heterog.', 'Aligned shift+var', and 'Anti-aligned' rows, the entry 'σ σ' is ambiguous; define σ as the common value σ1k = σ0k and define σ rev as its reversal, so the reader does not have to consult Table 5 to parse the main-text table.
[Related Work, footnote 1] The footnote describing Cytrynbaum's v2 and v3 is confusing and appears to reference two different arXiv versions with different titles and objectives; please verify the citation and clarify whether the description applies to the published or arXiv version actually cited.
[Robustness to Deployment-Weight Uncertainty] The values 42.87, 48.86, 43.68, and 1.88% are quoted without stating their units or the exact objective (B_q(ρ,e⋆)) used in the exact frontier; please define these quantities in the main text or move the numerical discussion entirely to Appendix B.
[Appendix D] The repository is described by file names only; include a URL or explicit data-availability statement so the code and data can be located by readers.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the oracle allocation is derived from a closed-form objective by a two-line Cauchy–Schwarz argument, and the plug-in rule is evaluated against external baselines and a true-variance oracle.

full rationale

The central derivation is self-contained and parameter-free. Theorem 1 minimizes B_Q(ρ,e) = Σ_k q_k V_k(e_k)/ρ_k over e_k and ρ_k; the proof first applies Cauchy–Schwarz to V_k(e_k) ≥ (σ_{1k}+σ_{0k})^2 with equality at e*_k = σ_{1k}/(σ_{1k}+σ_{0k}), then applies the same inequality to Σ_k q_k s_k^2/ρ_k ≥ (Σ_k √ q_k s_k)^2 with equality at ρ*_k ∝ √ q_k s_k. Nothing is fitted to the data used for evaluation: the oracle uses true variances, the plug-in uses pilot estimates, and the simulations and benchmarks draw fresh final-stage samples. Theorem 2 is a standard consistency/continuity transfer from Assumption 2, not a renaming of the target. Proposition 1 reuses the same reduction with q treated as uncertain, and its three cases follow from convex minimization and Sion's minimax theorem. The comparison designs (HHK, Shi–Lin, Tabord-Meehan, uniform, deployment-only, variance-only, oracle) are external or explicitly defined benchmarks, and the paper does not invoke any self-citation as load-bearing evidence. The composition-shift transportability condition is an untestable scope assumption and a correctness risk, but it is not circular: the paper states it openly, and the allocation rule optimizes precision for the estimands that are identified under that assumption. No prediction reduces by construction to its input, and no fitted parameter is renamed as a prediction. Overall circularity score 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central oracle result has no free parameters; all free parameters listed are tuning constants for the plug-in and robust variants. The key domain assumptions are composition-shift transportability of group effects, positivity and bounded variance inputs, and pilot variance consistency. No new entities are postulated.

free parameters (5)

Winsorization radius c = 3
Fixed robustness constant in robust TWNA; chosen by hand, not fitted to the data.
Shrinkage weight lambda = 5
Fixed robustness constant in robust TWNA; chosen by hand, not fitted to the data.
Scale floor = 0.05
Minimum allowed pilot scale estimate in all plug-in rules; enforces the bounded-variance input condition of Assumption 1.
Allocation floor rho_min = 0.01
Preserves positivity of group sample proportions; Assumption 1 requires it to lie below the oracle minimum share.
Treatment clipping epsilon = 0.05
Bounds treatment probabilities away from 0 and 1 in the plug-in rule; also used in Assumption 1.

assumptions (4)

domain assumption Group-specific effects transport from the experimental population to the deployment population (composition shift only): E_Q{Y(1)-Y(0)|S=k}=E_P{Y(1)-Y(0)|S=k}.
This identification assumption makes the target estimands tau_k equal to experimental effects; if it fails, the design optimizes precision for the wrong quantities. Stated in the Method section.
domain assumption Positivity and bounded variance inputs: c_sigma <= sigma_ak <= C_sigma, sigma_1k/(sigma_1k+sigma_0k) in [epsilon,1-epsilon], and rho_min < min_k rho*_k.
Assumption 1; needed for the oracle to be feasible and for the regret bound. The lower bound on rho_min is an ad hoc condition on the unknown truth.
domain assumption Pilot variance consistency: max_{a,k} |hat_sigma_ak - sigma_ak| -> 0 and = O_p(n_0^{-1/2}) for rate statements.
Assumption 2; standard consistency of cell standard deviations from a balanced pilot, but stated as an assumption rather than derived.
standard math Cauchy-Schwarz (Engel form) and Sion's minimax theorem.
Used in the proofs of Theorem 1 and Proposition 1.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Target-Weighted Neyman Allocation: Experimental Design for Heterogeneous Treatment Effects under Population Shift." pith.science (2026). https://pith.science/paper/ZRFIHGS5

@misc{pith2026260806512,
  author       = {Pith},
  title        = {Pith review of: Target-Weighted Neyman Allocation: Experimental Design for Heterogeneous Treatment Effects under Population Shift},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZRFIHGS5}},
  note         = {Machine review of arXiv:2608.06512}
}

read the original abstract

Randomized experiments are often run in one population to guide decisions in another. Allocating by experimental proportions wastes budget on groups that rarely appear in deployment, whereas allocating by deployment proportions under-samples groups that are hard to measure precisely. We propose \textbf{TWNA} (Target-Weighted Neyman Allocation), a two-stage stratified design that uses pilot estimates of group--arm outcome variances to allocate final-stage sample sizes and treatment probabilities for target-weighted group average treatment effect (GATE) precision. The oracle rule has a closed form and balances deployment importance with statistical difficulty; the plug-in rule recovers it as pilot variance estimates stabilize. We also extend TWNA to handle uncertainty about deployment composition, remaining robust whether the target mix is roughly known or entirely unknown. Finally, we distinguish this weight robustness from a pilot-robust variant for skewed, rare-event, or contaminated outcomes. Simulations and real-covariate benchmarks show the largest gains when groups are both deployment-important and difficult to measure.

Figures

Figures reproduced from arXiv: 2608.06512 by the authors.

**Figure 2.** Fixed-total-budget pilot frontier. Relative target [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

**Figure 3.** shows that the main ranking is largely preserved: deployment-aware rules remain tightly clustered, with robust TWNA slightly ahead of vanilla TWNA and both remaining close to the oracle. The takeaway is not that heavy tails overturn the allocation principle, but that the deployment-targeted design remains competitive and that pilot regularization can provide a modest stability gain when pilot cells are noisy. Scali… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Robustness to larger fixed-group partitions. Val [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]

**Figure 5.** Figure 5: Exact objective geometry over deployment con [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 10 canonical work pages

[1]

Optimalityofmatched-pairdesignsinrandom- izedcontrolledtrials.AmericanEconomicReview,112(12): 3911–3940

Bai,Y.2022. Optimalityofmatched-pairdesignsinrandom- izedcontrolledtrials.AmericanEconomicReview,112(12): 3911–3940. Method Micro Elig. Biom. Counts Latency Capped Contam. Uniform 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Deploy-only 0.481 0.985 0.602 0.497 0.795 0.955 0.666 Variance-only 0.595 0.968 0.697 0.565 0.800 0.893 1.402 Uniform Neyman 0.713 0.99...

work page 2022
[10]

Wei, W.; Ma, X.; and Wang, J

Enhancing Exter- nal Validity of Experiments with Ongoing Sampling.arXiv preprint arXiv:2502.18253. Wei, W.; Ma, X.; and Wang, J

work page arXiv
[11]

Xie, H.; and Aurisset, J

Efficient targeted learning of heteroge- neous treatment effects for multiple subgroups.Biometrics, 79(3): 1934–1946. Xie, H.; and Aurisset, J

work page 1934
[14]

Optimal Estimation of Generalized Average Treatment Effects using Kernel Optimal Matching

Kallus,N.;andSantacatterina,M.2019. Optimalestimation ofgeneralizedaveragetreatmenteffectsusingkerneloptimal matching.arXiv preprint arXiv:1908.04748. Kasy, M.; and Sautmann, A

work page Pith review arXiv 2019
[250]

Bethel, J

Bakshy,E.;Eckles,D.;andBernstein,M.S.2014.Designing anddeployingonlinefieldexperiments.InProceedingsofthe 23rdinternationalconferenceonWorldwideweb,283–292. Bethel, J

work page 2014
[2011]

Hu,Y.;Zhu,H.;Brunskil,E.;andWager,S.2024

Adaptive ex- perimental design using the propensity score.Journal of Business & Economic Statistics, 29(1): 96–108. Hu,Y.;Zhu,H.;Brunskil,E.;andWager,S.2024. Minimax- regret sample selection in randomized experiments. InPro- ceedings of the 25th ACM Conference on Economics and Computation, 1209–1235. Imai, K.; and Li, M. L

work page 2024
[2016]

Optimal Treatment Allocations Accounting for Population Differences

Improving the sensitivity of online controlled experiments: Case studies at netflix. In Proceedings of the 22nd ACM SIGKDD International Con- ferenceonKnowledgeDiscoveryandDataMining,645–654. Zhang,W.;Zhang,Z.;andLiu,A.2025. OptimalTreatment Allocations Accounting for Population Differences.arXiv preprint arXiv:2505.15944

work page Pith review arXiv 2025
[2018]

Technical report, National Bureau of Economic Research

Generic machine learning inference on hetero- geneous treatment effects in randomized experiments, with an application to immunization in India. Technical report, National Bureau of Economic Research. Cochran, W. G. 1977.Sampling techniques. john wiley & sons. Cole, S. R.; and Stuart, E. A

work page 1977

Show all 12 references

[2021]

InInternational Conference on Artificial Intelligence and Statistics, 2539–2547

Designing transportable experiments under s-admissability. InInternational Conference on Artificial Intelligence and Statistics, 2539–2547. PMLR. Robertson, S. E.; Steingrimsson, J. A.; Joyce, N. R.; Stuart, E.A.;andDahabreh,I.J.2024. Estimatingsubgroupeffects in generalizabil...

2024
[2022]

Tabord-Meehan, M

Strategy to select most efficient RCT samples based on observational data.arXiv preprint arXiv:2211.04672. Tabord-Meehan, M

arXiv
[2023]

Dahabreh, I

Optimal stratification of survey ex- periments.arXiv preprint arXiv:2111.08157v2. Dahabreh, I. J.; Robertson, S. E.; Tchetgen, E. J.; Stuart, E. A.; and Hernán, M. A

arXiv
[2025]

Imbens,G.;andXu,Y.2024.Lalonde(1986)afternearlyfour decades:Lessonslearned.arXivpreprintarXiv:2406.00827,

Statistical inference for het- erogeneous treatment effects discovered by generic machine learning in randomized experiments.Journal of Business & Economic Statistics, 43(1): 256–268. Imbens,G.;andXu,Y.2024.Lalonde(1986)afternearlyfour decades:Lessonslearned.arXivpreprintarXiv...

1986 arXiv

Pith tools

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

[1] [1]

Optimalityofmatched-pairdesignsinrandom- izedcontrolledtrials.AmericanEconomicReview,112(12): 3911–3940

Bai,Y.2022. Optimalityofmatched-pairdesignsinrandom- izedcontrolledtrials.AmericanEconomicReview,112(12): 3911–3940. Method Micro Elig. Biom. Counts Latency Capped Contam. Uniform 1.000 1.000 1.000 1.000 1.000 1.000 1.000 Deploy-only 0.481 0.985 0.602 0.497 0.795 0.955 0.666 Variance-only 0.595 0.968 0.697 0.565 0.800 0.893 1.402 Uniform Neyman 0.713 0.99...

work page 2022

[2] [10]

Wei, W.; Ma, X.; and Wang, J

Enhancing Exter- nal Validity of Experiments with Ongoing Sampling.arXiv preprint arXiv:2502.18253. Wei, W.; Ma, X.; and Wang, J

work page arXiv

[3] [11]

Xie, H.; and Aurisset, J

Efficient targeted learning of heteroge- neous treatment effects for multiple subgroups.Biometrics, 79(3): 1934–1946. Xie, H.; and Aurisset, J

work page 1934

[4] [14]

Optimal Estimation of Generalized Average Treatment Effects using Kernel Optimal Matching

Kallus,N.;andSantacatterina,M.2019. Optimalestimation ofgeneralizedaveragetreatmenteffectsusingkerneloptimal matching.arXiv preprint arXiv:1908.04748. Kasy, M.; and Sautmann, A

work page Pith review arXiv 2019

[5] [250]

Bethel, J

Bakshy,E.;Eckles,D.;andBernstein,M.S.2014.Designing anddeployingonlinefieldexperiments.InProceedingsofthe 23rdinternationalconferenceonWorldwideweb,283–292. Bethel, J

work page 2014

[6] [2011]

Hu,Y.;Zhu,H.;Brunskil,E.;andWager,S.2024

Adaptive ex- perimental design using the propensity score.Journal of Business & Economic Statistics, 29(1): 96–108. Hu,Y.;Zhu,H.;Brunskil,E.;andWager,S.2024. Minimax- regret sample selection in randomized experiments. InPro- ceedings of the 25th ACM Conference on Economics and Computation, 1209–1235. Imai, K.; and Li, M. L

work page 2024

[7] [2016]

Optimal Treatment Allocations Accounting for Population Differences

Improving the sensitivity of online controlled experiments: Case studies at netflix. In Proceedings of the 22nd ACM SIGKDD International Con- ferenceonKnowledgeDiscoveryandDataMining,645–654. Zhang,W.;Zhang,Z.;andLiu,A.2025. OptimalTreatment Allocations Accounting for Population Differences.arXiv preprint arXiv:2505.15944

work page Pith review arXiv 2025

[8] [2018]

Technical report, National Bureau of Economic Research

Generic machine learning inference on hetero- geneous treatment effects in randomized experiments, with an application to immunization in India. Technical report, National Bureau of Economic Research. Cochran, W. G. 1977.Sampling techniques. john wiley & sons. Cole, S. R.; and Stuart, E. A

work page 1977