pith. machine review for the scientific record. sign in

arxiv: 2605.05752 · v1 · submitted 2026-05-07 · 📊 stat.ME · stat.AP

Recognition: unknown

Generative AI-Based Monte Carlo Simulation for Method Evaluation Using Synthetic Multilevel Data

Youmi Suk , Chenguang Pan , Weixuan Xiao
Authors on Pith no claims yet

Pith reviewed 2026-05-08 08:05 UTC · model grok-4.3

classification 📊 stat.ME stat.AP
keywords generative AIMonte Carlo simulationsynthetic datamultilevel datamethod evaluationdiffusion modelsGANsstatistical methods
0
0 comments X

The pith

Generative AI trained on real multilevel data produces synthetic versions for Monte Carlo simulations that evaluate statistical methods under realistic conditions.

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

The paper introduces a framework that trains generative AI models on actual multilevel datasets to create synthetic data preserving key structures, then uses that data for Monte Carlo simulations to test quantitative methods. Traditional simulations often depend on arbitrary or simplified scenarios that fail to capture the complexities of real data, leading to evaluations that may not hold up in practice. The proposed six-stage workflow covers specifying the method and real data, training modified diffusion models and GANs, assessing synthetic data quality at within- and between-levels, running tailored simulations, evaluating performance or parameter recovery, and checking robustness. Demonstrations with empirical social science data and multilevel models illustrate how this produces evaluations more aligned with actual applications. If the synthetic data maintains sufficient fidelity, method assessments become more reliable for guiding real-world use.

Core claim

The paper proposes a six-stage workflow for AI-based Monte Carlo simulation studies: select a quantitative method and real multilevel data, train generative AI models with targeted modifications to diffusion models and GANs, evaluate the synthetic data for within-table and between-table fidelity, design and conduct simulations, assess the method's predictive performance or parameter recovery, and verify robustness. Using empirical multilevel data and multilevel modeling methods, the workflow shows that this produces evaluations more accurate and applicable to real settings than those based on arbitrary simulated scenarios.

What carries the argument

The six-stage workflow that trains modified diffusion models and GANs on real multilevel data to generate synthetic datasets, followed by systematic fidelity checks and tailored simulation designs for method evaluation.

If this is right

  • Evaluations of quantitative methods reflect data structures from actual applications instead of researcher-chosen arbitrary setups.
  • Simulation designs need to vary depending on whether the objective is predictive performance or accurate parameter recovery.
  • A quality evaluation framework checks fidelity at both individual and group levels to ensure synthetic data usability.
  • Robustness checks at the workflow's end confirm the stability of method performance estimates.
  • This process supports more honest assessments that better indicate how methods will behave with real data.

Where Pith is reading between the lines

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

  • The framework could be extended to generate synthetic versions of longitudinal or spatial data structures for method testing in those domains.
  • Combining AI-generated data with limited real samples might create hybrid evaluation designs that improve efficiency.
  • Different generative models may produce varying simulation outcomes, suggesting comparisons across model types to refine the approach.
  • Public repositories of representative real datasets could serve as bases for standardized synthetic benchmarks across statistical methods.

Load-bearing premise

Generative AI models with targeted modifications can create synthetic multilevel data whose structures and properties are close enough to real data that simulation results generalize beyond the synthetic cases.

What would settle it

If applying the same quantitative methods to the original real dataset yields substantially different performance rankings or accuracy metrics than those from the AI-generated synthetic simulations, the claim that the synthetic data supports valid evaluations would not hold.

read the original abstract

The role of AI-generated synthetic data has recently been expanded to support realistic Monte Carlo simulations. However, guidance is limited on generating data with multilevel structures and designing simulations based on such data. This study proposes a general framework for AI-based simulation studies to evaluate the predictive performance and parameter recovery of quantitative methods, specifically using multilevel data commonly observed in the social sciences. Our proposed six-stage workflow consists of (i) specifying a method and real data, (ii) training Generative AI with real data, (iii) assessing synthetic data quality, (iv) designing and conducting simulations, (v) evaluating method performance, and (vi) checking robustness. To enhance fidelity in multilevel data generation, we also introduce targeted modifications to diffusion models and Generative Adversarial Networks (GANs). Furthermore, we develop a systematic quality evaluation framework that assesses both within-table and between-table fidelity, and discuss how AI-based simulation designs should differ depending on whether the simulation's objective is predictive performance or parameter recovery. Finally, using empirical multilevel data and multilevel modeling methods, we demonstrate the utility of the proposed AI-based simulation framework. This approach leads to more accurate and honest evaluations of quantitative methods in the real world, unlike traditional simulation studies based on arbitrary simulated scenarios.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The paper proposes a six-stage workflow for AI-based Monte Carlo simulations to evaluate quantitative methods using synthetic multilevel data: (i) specify method and real data, (ii) train generative AI (with targeted modifications to diffusion models and GANs), (iii) assess synthetic data quality via a within-/between-table fidelity framework, (iv) design and run simulations, (v) evaluate method performance, and (vi) check robustness. It demonstrates the approach on empirical multilevel data and multilevel models, claiming this yields more accurate and honest method evaluations than traditional simulations based on arbitrary data-generating processes.

Significance. If the synthetic multilevel data achieves fidelity sufficient for simulation results to generalize, the framework could improve realism in social-science method evaluations by replacing arbitrary scenarios with data-driven structures, potentially leading to more reliable assessments of parameter recovery and predictive performance.

major comments (3)
  1. [Demonstration] Demonstration section (final empirical example): only internal quality metrics and a single example are reported; the manuscript provides no quantitative side-by-side comparison of method performance (e.g., bias, coverage, or predictive accuracy) under the AI-generated synthetic data versus conventional arbitrary DGPs when both are benchmarked against the same real multilevel data. This directly undermines the central claim that the workflow produces more accurate evaluations.
  2. [Workflow stages ii-iii] Stages (ii)–(iii) and fidelity framework: targeted modifications to GANs/diffusion models are introduced for multilevel structures, yet no external validation is shown quantifying how closely within-cluster and between-cluster statistics (variances, covariances, ICCs) match the real data, nor how deviations affect downstream Monte Carlo results on parameter recovery. The claim that results generalize therefore rests on untested fidelity.
  3. [Abstract and simulation-design discussion] Abstract and § on simulation design: the distinction between predictive-performance and parameter-recovery objectives is discussed, but the demonstration does not report separate error metrics or sensitivity analyses showing that the AI-based design improves upon arbitrary simulations for either objective when validated externally.
minor comments (1)
  1. [Workflow description] A flowchart or table summarizing the six-stage workflow and the within-/between-table fidelity criteria would improve readability.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments, which identify key areas where additional evidence can strengthen the manuscript's central claims. We address each major comment below and commit to revisions that directly respond to the concerns raised.

read point-by-point responses

  1. Referee: [Demonstration] Demonstration section (final empirical example): only internal quality metrics and a single example are reported; the manuscript provides no quantitative side-by-side comparison of method performance (e.g., bias, coverage, or predictive accuracy) under the AI-generated synthetic data versus conventional arbitrary DGPs when both are benchmarked against the same real multilevel data. This directly undermines the central claim that the workflow produces more accurate evaluations.

    Authors: We agree that a direct quantitative comparison is required to substantiate the claim that the AI-based workflow yields more accurate evaluations than traditional arbitrary DGPs. The current demonstration focuses on illustrating the workflow and reporting internal fidelity metrics. In the revised manuscript, we will add side-by-side comparisons of method performance metrics—including bias, coverage, and predictive accuracy—obtained under AI-generated synthetic data versus conventional arbitrary data-generating processes, with both benchmarked against the same real multilevel data. This addition will provide the empirical support needed for the claim. revision: yes

  2. Referee: [Workflow stages ii-iii] Stages (ii)–(iii) and fidelity framework: targeted modifications to GANs/diffusion models are introduced for multilevel structures, yet no external validation is shown quantifying how closely within-cluster and between-cluster statistics (variances, covariances, ICCs) match the real data, nor how deviations affect downstream Monte Carlo results on parameter recovery. The claim that results generalize therefore rests on untested fidelity.

    Authors: The referee correctly notes the absence of external validation for the fidelity framework. While the within-/between-table approach offers systematic internal checks, we will revise the manuscript to include explicit quantitative comparisons of within-cluster and between-cluster statistics (variances, covariances, and ICCs) between the synthetic data and the original real data. We will also add analyses examining how observed deviations in these statistics influence downstream Monte Carlo results on parameter recovery, thereby providing evidence for the generalizability of the framework. revision: yes

  3. Referee: [Abstract and simulation-design discussion] Abstract and § on simulation design: the distinction between predictive-performance and parameter-recovery objectives is discussed, but the demonstration does not report separate error metrics or sensitivity analyses showing that the AI-based design improves upon arbitrary simulations for either objective when validated externally.

    Authors: We acknowledge that the demonstration does not yet provide separate error metrics or external sensitivity analyses for the two simulation objectives. The distinction is outlined in the simulation-design section, but to demonstrate improvement, the revised version will expand the empirical example to report distinct metrics for predictive performance and parameter recovery. Sensitivity analyses comparing the AI-based design to arbitrary simulations, externally validated against the real data, will also be included for each objective. revision: yes

Circularity Check

0 steps flagged

No circularity: methodological framework with independent demonstration

full rationale

The paper proposes a six-stage workflow for generating synthetic multilevel data via modified GANs/diffusion models and using it for Monte Carlo method evaluation. No mathematical derivation, prediction, or result is presented that reduces by construction to fitted parameters, self-definitions, or self-citations. The quality assessment (within-/between-table fidelity) is a separate step from the simulation outcomes, and the empirical demonstration applies the workflow to real data without claiming that performance metrics are forced by the training process itself. No load-bearing self-citations or uniqueness theorems are invoked. The central claim—that AI-based simulations yield more realistic evaluations than arbitrary DGPs—is a methodological assertion supported by the framework design and example, not by circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that AI-generated synthetic data can faithfully replicate multilevel structures without introducing artifacts that distort method evaluations. No free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption Generative AI models trained on real multilevel data can produce synthetic versions that preserve both within-cluster and between-cluster relationships sufficiently for simulation purposes.
    Invoked in the training and quality assessment stages of the proposed workflow.

pith-pipeline@v0.9.0 · 5523 in / 1235 out tokens · 64884 ms · 2026-05-08T08:05:06.922394+00:00 · methodology

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

64 extracted references · 32 canonical work pages · 1 internal anchor

  1. [1]

    Suk, Youmi and Pan, Chenguang and Yang, Ke , year=. Using. doi:10.3102/10769986251397559 , journal=

    work page doi:10.3102/10769986251397559

  2. [2]

    arXiv , year=

    Sequential models in the synthetic data vault , author=. arXiv , year=

  3. [3]

    2024 , eprint=

    Structured Evaluation of Synthetic Tabular Data , author=. 2024 , eprint=

    2024

  4. [4]

    Speech Recognition: Papers Presented at the 1974

    Allen Newell , title =. Speech Recognition: Papers Presented at the 1974. 1975 , publisher =

    1974

  5. [5]

    2024 , note =

    Synthetic Data Metrics , author =. 2024 , note =

    2024

  6. [6]

    1988 , publisher =

    Cohen, Jacob , title =. 1988 , publisher =

    1988

  7. [7]

    Biological Reviews , year =

    Nakagawa, Shinichi and Schielzeth, Holger , title =. Biological Reviews , year =

  8. [8]

    arXiv preprint arXiv:2411.17672 , year=

    Synthetic data generation with llm for improved depression prediction , author=. arXiv preprint arXiv:2411.17672 , year=

    work page arXiv

  9. [9]

    International Conference on Machine Learning , pages=

    How faithful is your synthetic data? sample-level metrics for evaluating and auditing generative models , author=. International Conference on Machine Learning , pages=. 2022 , organization=

    2022

  10. [10]

    Frontiers in Digital Health , volume=

    Comprehensive evaluation framework for synthetic tabular data in health: fidelity, utility and privacy analysis of generative models with and without privacy guarantees , author=. Frontiers in Digital Health , volume=. 2025 , publisher=

    2025

  11. [11]

    2024 , organization =

    Wei Pang and Masoumeh Shafieinejad and Lucy Liu and Stephanie Hazlewood and Xi He , booktitle =. 2024 , organization =

    2024

  12. [12]

    Advances in Neural Information Processing Systems , doi=

    Denoising diffusion probabilistic models , author=. Advances in Neural Information Processing Systems , doi=

  13. [13]

    2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA) , pages=

    The synthetic data vault , author=. 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA) , pages=. 2016 , organization=

    2016

  14. [14]

    Mostly harmless simulations? Using Monte Carlo studies for estimator selection , volume=

    Advani, Arun and Kitagawa, Toru and Słoczyński, Tymon , year=. Mostly harmless simulations? Using Monte Carlo studies for estimator selection , volume=. Journal of Applied Econometrics , publisher=. doi:10.1002/jae.2724 , number=

    work page doi:10.1002/jae.2724

  15. [15]

    Huang, Francis L. , year=. Multilevel Modeling and Ordinary Least Squares Regression: How Comparable Are They? , volume=. The Journal of Experimental Education , publisher=. doi:10.1080/00220973.2016.1277339 , number=

    work page doi:10.1080/00220973.2016.1277339 2016

  16. [16]

    Prediction in Multilevel Models , volume=

    Afshartous, David and de Leeuw, Jan , year=. Prediction in Multilevel Models , volume=. Journal of Educational and Behavioral Statistics , publisher=. doi:10.3102/10769986030002109 , number=

    work page doi:10.3102/10769986030002109

  17. [17]

    and Metzger, Jonas and Munro, Evan , year=

    Athey, Susan and Imbens, Guido W. and Metzger, Jonas and Munro, Evan , year=. Using. Journal of Econometrics , publisher=. doi:10.1016/j.jeconom.2020.09.013 , number=

    work page doi:10.1016/j.jeconom.2020.09.013 2020

  18. [18]

    Advances in Neural Information Processing Systems , page=

    Improved training of wasserstein gans , author=. Advances in Neural Information Processing Systems , page=

  19. [19]

    RandomForestsGLS: An R package for Random Forests for dependent data , volume=

    Saha, Arkajyoti and Basu, Sumanta and Datta, Abhirup , year=. RandomForestsGLS: An R package for Random Forests for dependent data , volume=. Journal of Open Source Software , publisher=. doi:10.21105/joss.03780 , number=

    work page doi:10.21105/joss.03780

  20. [20]

    2012 , publisher =

    Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling , author =. 2012 , publisher =

    2012

  21. [21]

    2002 , publisher =

    Hierarchical Linear Models: Applications and Data Analysis Methods , author =. 2002 , publisher =

    2002

  22. [22]

    Advances in Neural Information Processing Systems , year=

    Generative adversarial nets , author=. Advances in Neural Information Processing Systems , year=

  23. [23]

    Modeling tabular data using conditional

    Xu, Lei and Skoularidou, Maria and Cuesta-Infante, Alfredo and Veeramachaneni, Kalyan , journal=. Modeling tabular data using conditional. doi:https://dl.acm.org/doi/10.5555/3454287.3454946 , pages =

    work page doi:10.5555/3454287.3454946

  24. [24]

    2016 , address=

    Deep Learning , author=. 2016 , address=

    2016

  25. [25]

    Goyal, Mandeep and Mahmoud, Qusay H. , year=. A Systematic Review of Synthetic Data Generation Techniques Using Generative AI , volume=. Electronics , publisher=. doi:10.3390/electronics13173509 , number=

    work page doi:10.3390/electronics13173509

  26. [26]

    and Dervovic, Danial and Mahfouz, Mahmoud and Tillman, Robert E

    Assefa, Samuel A. and Dervovic, Danial and Mahfouz, Mahmoud and Tillman, Robert E. and Reddy, Prashant and Veloso, Manuela , year=. Generating synthetic data in finance: opportunities, challenges and pitfalls , url=. doi:10.1145/3383455.3422554 , booktitle=

    work page doi:10.1145/3383455.3422554

  27. [27]

    Evaluating Classifiers Trained on Differentially Private Synthetic Health Data , url=

    Movahedi, Parisa and Nieminen, Valtteri and Perez, Ileana Montoya and Pahikkala, Tapio and Airola, Antti , year=. Evaluating Classifiers Trained on Differentially Private Synthetic Health Data , url=. doi:10.1109/cbms58004.2023.00313 , booktitle=

    work page doi:10.1109/cbms58004.2023.00313 2023

  28. [28]

    Advances in Neural Information Processing Systems , volume=

    Language models are few-shot learners , author=. Advances in Neural Information Processing Systems , volume=. 2020 , doi=

    2020

  29. [29]

    International Conference on Learning Representations , year=

    Auto-encoding variational bayes , author=. International Conference on Learning Representations , year=

  30. [30]

    Proceedings of the 34th International Conference on Machine Learning , pages =

    Martin Arjovsky and Soumith Chintala and L. Proceedings of the 34th International Conference on Machine Learning , pages =. 2017 , editor =

    2017

  31. [31]

    Metropolis, Nicholas and Ulam, S. , year=. The Monte Carlo Method , volume=. Journal of the American Statistical Association , publisher=. doi:10.1080/01621459.1949.10483310 , number=

    work page doi:10.1080/01621459.1949.10483310 1949

  32. [32]

    Thomopoulos, Nick T. , year=. Essentials of Monte Carlo Simulation: Statistical Methods for Building Simulation Models , ISBN=. doi:10.1007/978-1-4614-6022-0 , publisher=

    work page doi:10.1007/978-1-4614-6022-0

  33. [33]

    VLDB Endowment , doi=

    Data synthesis based on generative adversarial networks , author=. VLDB Endowment , doi=

  34. [34]

    Conditional Generative Adversarial Nets

    Conditional Generative Adversarial Nets , author=. arXiv preprint arXiv:1411.1784 , year=

    work page internal anchor Pith review arXiv

  35. [35]

    and Felsovalyi, A

    Fan, X. and Felsovalyi, A. and Sivo, S., and Keenan, S. , year=

  36. [36]

    Machine learning for synthetic data generation: a review.arXiv preprint arXiv:2302.04062, 2023

    Machine Learning for Synthetic Data Generation: A Review , author =. arXiv preprint arXiv:2302.04062 , year =

    work page arXiv

  37. [37]

    Proceedings of the 40th International Conference on Machine Learning (ICML) , pages=

    TabDDPM: Modelling Tabular Data with Diffusion Models , author=. Proceedings of the 40th International Conference on Machine Learning (ICML) , pages=. 2023 , publisher=

    2023

  38. [38]

    Chen , booktitle=

    Zilong Zhao and Aditya Kunar and Robert Birke and Lydia Y. Chen , booktitle=. 2021 , editor=

    2021

  39. [39]

    Advances in Neural Information Processing Systems (NeurIPS) 33 , pages=

    VAEM: a Deep Generative Model for Heterogeneous Mixed Type Data , author=. Advances in Neural Information Processing Systems (NeurIPS) 33 , pages=. 2020 , url=

    2020

  40. [40]

    Markowitz , title =

    Victor M. Markowitz , title =. Proceedings of the 17th International Conference on Very Large Data Bases (VLDB) , year =

  41. [41]

    The Elements of Statistical Learning

    Hastie, Trevor and Tibshirani, Robert and Friedman, Jerome , year=. The Elements of Statistical Learning: Data Mining, Inference, and Prediction , ISBN=. doi:10.1007/978-0-387-84858-7 , journal=

    work page doi:10.1007/978-0-387-84858-7

  42. [42]

    2006 , publisher=

    Pattern recognition and machine learning , author=. 2006 , publisher=

    2006

  43. [43]

    and White, Ian R

    Morris, Tim P. and White, Ian R. and Crowther, Michael J. , year=. Using simulation studies to evaluate statistical methods , volume=. Statistics in Medicine , publisher=. doi:10.1002/sim.8086 , number=

    work page doi:10.1002/sim.8086

  44. [44]

    A comprehensive survey of synthetic tabular data generation.arXiv preprint arXiv:2504.16506, 2025

    A Comprehensive Survey of Synthetic Tabular Data Generation , author=. arXiv preprint arXiv:2504.16506 , year=

    work page arXiv

  45. [45]

    Statistics in medicine , volume=

    Multiple imputation using chained equations: issues and guidance for practice , author=. Statistics in medicine , volume=. 2011 , publisher=

    2011

  46. [46]

    & Walker, S

    Fitting Linear Mixed-Effects Models Using lme4 , volume=. Journal of Statistical Software , author=. 2015 , pages=. doi:10.18637/jss.v067.i01 , number=

    work page doi:10.18637/jss.v067.i01 2015

  47. [47]

    Journal of Machine Learning Research , year =

    Fabio Sigrist , title =. Journal of Machine Learning Research , year =

  48. [48]

    , year =

    Hox, Joop J. , year =. Multilevel. Classification,. doi:10.1007/978-3-642-72087-1_17 , language =

    work page doi:10.1007/978-3-642-72087-1_17

  49. [49]

    Research in Higher Education , author =

    Multi-. Research in Higher Education , author =. 2009 , pages =. doi:10.1007/s11162-009-9121-3 , language =

    work page doi:10.1007/s11162-009-9121-3 2009

  50. [50]

    When can group level clustering be ignored?

    Clarke, Philippa , month = aug, year =. When can group level clustering be ignored?. Journal of Epidemiology & Community Health , publisher =. doi:10.1136/jech.2007.060798 , language =

    work page doi:10.1136/jech.2007.060798 2007

  51. [51]

    General Linear Model Journal , author =

    A. General Linear Model Journal , author =. 2002 , keywords =

    2002

  52. [52]

    Lai, Mark H. C. and Kwok, Oi-man , month = jul, year =. Examining the. The Journal of Experimental Education , publisher =. doi:10.1080/00220973.2014.907229 , number =

    work page doi:10.1080/00220973.2014.907229 2014

  53. [53]

    NCES 2011-328

    High School Longitudinal Study of 2009 (HSLS: 09): Base-Year Data File Documentation. NCES 2011-328. , author=. National Center for Education Statistics , year=

    2009

  54. [54]

    Learning and Individual Differences , author =

    The longitudinal influences of peers, parents, motivation, and mathematics course-taking on high school math achievement , volume =. Learning and Individual Differences , author =. 2016 , pages =. doi:10.1016/j.lindif.2016.07.012 , urldate =

    work page doi:10.1016/j.lindif.2016.07.012 2016

  55. [55]

    Sense of

    Jang, Sung Tae , month = mar, year =. Sense of. Teachers College Record , publisher =. doi:10.1177/01614681231173019 , number =

    work page doi:10.1177/01614681231173019

  56. [56]

    and Eccles, Jacquelynne S

    Jiang, Su and Simpkins, Sandra D. and Eccles, Jacquelynne S. , year =. Individuals’ math and science motivation and their subsequent. Developmental Psychology , publisher =. doi:10.1037/dev0001110 , number =

    work page doi:10.1037/dev0001110

  57. [57]

    PsyArXiv , author =

    Learning feasible optimal treatment regimes for personalized decision-making , url =. PsyArXiv , author =. doi:10.31234/osf.io/arp48_v1 , urldate =

    work page doi:10.31234/osf.io/arp48_v1

  58. [58]

    Science Education , author =

    Visiting. Science Education , author =. 2014 , pages =. doi:10.1002/sce.21116 , language =

    work page doi:10.1002/sce.21116 2014

  59. [59]

    Educational Studies in Mathematics , author =

    A systematic review of factors associated with high schoolers’ algebra achievement according to. Educational Studies in Mathematics , author =. 2022 , keywords =. doi:10.1007/s10649-021-10130-4 , language =

    work page doi:10.1007/s10649-021-10130-4 2022

  60. [60]

    Examining the

    Pan, Chenguang and Zhang, Zhou , year =. Examining the

  61. [61]

    Laurent Perron and Frédéric Didier , organization =

  62. [62]

    Zhijing Jin, Yuen Chen, Felix Leeb, Luigi Gresele, Ojasv Kamal, Zhiheng Lyu, Kevin Blin, Fernando Gonzalez Adauto, Max Kleiman-Weiner, Mrinmaya Sachan, and Bernhard Schölkopf

    Hill, Jennifer L. , month = jan, year =. Bayesian Nonparametric Modeling for Causal Inference , volume =. Journal of Computational and Graphical Statistics , publisher =. doi:10.1198/jcgs.2010.08162 , number =

    work page doi:10.1198/jcgs.2010.08162 2010

  63. [63]

    Synth-Validation: Selecting the Best Causal Inference Method for a Given Dataset , shorttitle =

    Schuler, Alejandro and Jung, Ken and Tibshirani, Robert and Hastie, Trevor and Shah, Nigam , month = oct, year =. Synth-Validation: Selecting the Best Causal Inference Method for a Given Dataset , shorttitle =. doi:10.48550/arXiv.1711.00083 , urldate =

    work page doi:10.48550/arxiv.1711.00083

  64. [64]

    and Culpepper, Steven Andrew , month = sep, year =

    Aguinis, Herman and Gottfredson, Ryan K. and Culpepper, Steven Andrew , month = sep, year =. Best-Practice Recommendations for Estimating Cross-Level Interaction Effects Using Multilevel Modeling , volume =. Journal of Management , publisher =. doi:10.1177/0149206313478188 , language =

    work page doi:10.1177/0149206313478188