pith. sign in

arxiv: 2605.16615 · v1 · pith:FKKLCWMQnew · submitted 2026-05-15 · 💻 cs.LG

Learning What Evaluators Value: A Reliable Approach to Modeling Evaluator Preferences

Madeline Celi Kitch , Nihar B. Shah This is my paper

Pith reviewed 2026-05-20 20:00 UTC · model grok-4.3

classification 💻 cs.LG
keywords evaluator preferencespreference learningmodel mismatchrobust algorithmsnon-decreasing functionshuman evaluationLLM evaluation
0
0 comments X

The pith

An algorithm learns any coordinate-wise non-decreasing evaluator preference without losing performance under linear cases.

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

The paper focuses on learning how evaluators combine multiple criteria into overall judgments in settings like admissions or medical risk assessment. Common linear models for these preferences often mismatch reality and cause learning failures. The authors adopt only the minimal assumption that preferences are coordinate-wise non-decreasing and introduce a robust algorithm for learning under this condition. They prove the algorithm recovers any such preference function and matches the performance of linear methods exactly when linearity holds. Experiments with synthetic data and real human and LLM evaluations confirm the approach works in practice.

Core claim

We present an algorithm for learning evaluator preferences under the sole assumption that the preference function is coordinate-wise non-decreasing. We prove that the algorithm can learn any such preference function and that it incurs no performance penalty relative to linear-assumption methods when preferences are in fact linear. We also characterize the severity of model mismatch under common stronger assumptions and validate the method on synthetic simulations and real-world data involving both human and LLM evaluators.

What carries the argument

A robust learning algorithm that operates directly on coordinate-wise non-decreasing preference functions and avoids degradation from model mismatch.

Load-bearing premise

The preference function is coordinate-wise non-decreasing.

What would settle it

A collection of evaluation examples where the true preference decreases along at least one coordinate and the algorithm's recovered model produces substantially worse predictions than a correctly specified alternative.

Figures

Figures reproduced from arXiv: 2605.16615 by Madeline Celi Kitch, Nihar B. Shah.

Figure 1
Figure 1. Figure 1: An illustration of the evaluation settings we consider. Evaluators provide criteria evaluations [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An illustration of the problem we address in this paper. There are many evaluators, each of whom [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Simulation results for synthetic preferences. Error bars are computed using the standard error of [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Results of our algorithm relative to linear regression on Tripadvisor rating data. Error bars [PITH_FULL_IMAGE:figures/full_fig_p017_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Impact of increasing different criteria scores on the overall rating of the learned preference function [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Consistency and preference alignment for LLM (and human) reviewers on the ICLR dataset. [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Impact of increasing presentation on the overall paper rating. We omit estimates for all criterion [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Impact of increasing specific criterion scores on the overall paper rating. We omit estimates for [PITH_FULL_IMAGE:figures/full_fig_p041_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Review template for Astrobee evaluations. [PITH_FULL_IMAGE:figures/full_fig_p042_9.png] view at source ↗
read the original abstract

In many applications, human and LLM evaluators use assessments of relevant criteria to create an overall evaluation for an item or individual. For example, in admissions, committees assess candidates on attributes such as test scores, GPA, and research experience to evaluate their overall fit for the program. Another example arises in medical care where clinicians use patient reports of symptoms to consider preliminary diagnoses and assess risks. Each setting involves mapping multiple criteria to an overall evaluation -- a process that reflects the evaluator's underlying preferences. We focus on the fundamental question of learning these preferences. Many applications of this problem make specific modeling assumptions on evaluator preferences that may be substantially violated in the real world. We make the minimal assumption that the preference function is coordinate-wise non-decreasing, which is reasonable in a large number of evaluation settings. We theoretically characterize the severity of model mismatch for many common assumptions and show that it can lead to significant issues for learning evaluator preferences and other important downstream tasks. We then present an algorithm for learning evaluators' preferences that is robust to model mismatch. We prove theoretically that our algorithm can learn any preference function without sacrificing performance when the linearity assumption holds. Evaluations of our algorithm with synthetic simulations and real-world data confirm its ability to learn preferences robustly and illustrate key aspects of LLM and human preferences.

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

2 major / 2 minor

Summary. The manuscript proposes learning evaluator preferences under the sole modeling assumption that the preference function is coordinate-wise non-decreasing. It derives mismatch bounds showing that common parametric families (e.g., linear) can produce large errors, introduces a robust algorithm, proves that the algorithm recovers arbitrary coordinate-wise non-decreasing functions while exactly matching linear-model performance when the ground truth is linear, and validates the approach on synthetic data plus real-world human and LLM evaluation tasks.

Significance. If the theoretical results hold, the work is significant for preference modeling in admissions, medical risk assessment, and LLM alignment, because it supplies both a quantitative characterization of mismatch severity and a no-sacrifice guarantee that removes the usual trade-off between robustness and efficiency under correct linearity. The combination of explicit minimal assumption, mismatch analysis, and dual synthetic/real-world evaluation strengthens its practical relevance.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (theoretical results): the central claim that the algorithm recovers any coordinate-wise non-decreasing preference function without sacrificing linear performance is load-bearing; the manuscript states the result but the provided text does not contain the full derivation steps or the precise conditions on the evaluation oracle and noise model needed to verify the no-sacrifice property.
  2. [§3.2] §3.2 (mismatch bounds): the claim that mismatch 'can lead to significant issues for learning evaluator preferences and other important downstream tasks' is asserted after deriving bounds for common families, yet no concrete numerical example or downstream-task simulation is given to show when the bound exceeds a practically relevant threshold.
minor comments (2)
  1. [§2] Notation for the preference function f and the coordinate-wise non-decreasing property should be introduced with a formal definition in §2 rather than deferred to the algorithm section.
  2. [Experiments] The real-world experiment section would benefit from an explicit table listing the number of evaluators, number of criteria, and total items evaluated to support reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback on our manuscript. We address each major comment below, indicating the revisions we will make to improve clarity and completeness.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (theoretical results): the central claim that the algorithm recovers any coordinate-wise non-decreasing preference function without sacrificing linear performance is load-bearing; the manuscript states the result but the provided text does not contain the full derivation steps or the precise conditions on the evaluation oracle and noise model needed to verify the no-sacrifice property.

    Authors: We agree that the full derivation steps and precise conditions on the evaluation oracle and noise model are necessary to allow verification of the central no-sacrifice claim. In the revised manuscript we will expand the theoretical section (and add an appendix if space is limited) to include the complete proof of recovery for arbitrary coordinate-wise non-decreasing functions together with the exact assumptions on the oracle (e.g., access model and query type) and the noise model (e.g., bounded or sub-Gaussian noise) under which the algorithm matches linear performance when the ground truth is linear. revision: yes

  2. Referee: [§3.2] §3.2 (mismatch bounds): the claim that mismatch 'can lead to significant issues for learning evaluator preferences and other important downstream tasks' is asserted after deriving bounds for common families, yet no concrete numerical example or downstream-task simulation is given to show when the bound exceeds a practically relevant threshold.

    Authors: We acknowledge that a concrete numerical illustration would help readers assess when the derived mismatch bounds become practically consequential. We will add a short numerical example and/or simulation in the revised §3.2 (or a new subsection) that instantiates the bounds for representative parameter values and shows the resulting error in a downstream task such as ranking or selection, thereby demonstrating when the bound exceeds a relevant threshold. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper states the coordinate-wise non-decreasing assumption explicitly as the minimal modeling choice, derives mismatch bounds for common parametric families, and supplies both a general algorithm and a separate no-sacrifice proof that the algorithm matches linear-model performance exactly when the true function is linear. These steps are self-contained against the stated assumption and do not reduce by construction to fitted parameters, self-citations, or renamed inputs. Independent synthetic simulations and real-world data evaluations are supplied as external checks. No load-bearing step matches any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Only the abstract is available, so the ledger reflects the single explicit modeling assumption stated; no free parameters or invented entities are mentioned.

axioms (1)
  • domain assumption The preference function is coordinate-wise non-decreasing.
    Presented as the minimal assumption reasonable for many evaluation settings.

pith-pipeline@v0.9.0 · 5758 in / 1070 out tokens · 23338 ms · 2026-05-20T20:00:19.276339+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

113 extracted references · 113 canonical work pages · 21 internal anchors

  1. [1]

    Bridging Human and LLM Judgments : Understanding and Narrowing the Gap , December 2025

    Polo, Felipe Maia and Wang, Xinhe and Yurochkin, Mikhail and Xu, Gongjun and Banerjee, Moulinath and Sun, Yuekai , month = dec, year =. Bridging. doi:10.48550/arXiv.2508.12792 , abstract =

    work page doi:10.48550/arxiv.2508.12792

  2. [2]

    HD - Eval : Aligning Large Language Model Evaluators Through Hierarchical Criteria Decomposition , February 2024

    Liu, Yuxuan and Yang, Tianchi and Huang, Shaohan and Zhang, Zihan and Huang, Haizhen and Wei, Furu and Deng, Weiwei and Sun, Feng and Zhang, Qi , month = feb, year =. doi:10.48550/arXiv.2402.15754 , abstract =

    work page doi:10.48550/arxiv.2402.15754

  3. [3]

    2024 , note =

    A n00b’s. 2024 , note =

    work page 2024

  4. [4]

    NPJ Mental Health Research , author =

    Development of the treatment prediction model in the artificial intelligence in depression – medication enhancement study , volume =. NPJ Mental Health Research , author =. 2025 , pages =. doi:10.1038/s44184-025-00136-8 , abstract =

    work page doi:10.1038/s44184-025-00136-8 2025

  5. [5]

    Journal of the American Statistical Association , author =

    Estimating. Journal of the American Statistical Association , author =. 2012 , pages =. doi:10.1080/01621459.2012.695674 , abstract =

    work page doi:10.1080/01621459.2012.695674 2012

  6. [6]

    ACM Trans

    Evaluating collaborative filtering recommender systems , volume =. ACM Trans. Inf. Syst. , author =. 2004 , pages =. doi:10.1145/963770.963772 , abstract =

    work page doi:10.1145/963770.963772 2004

  7. [7]

    Academic Medicine , author =

    Artificial. Academic Medicine , author =. 2023 , pages =. doi:10.1097/ACM.0000000000005202 , abstract =

    work page doi:10.1097/acm.0000000000005202 2023

  8. [8]

    James and Huang, Melody and Imai, Kosuke and Jiang, Zhichao and Shin, Sooahn , month = oct, year =

    Ben-Michael, Eli and Greiner, D. James and Huang, Melody and Imai, Kosuke and Jiang, Zhichao and Shin, Sooahn , month = oct, year =. Does. doi:10.48550/arXiv.2403.12108 , abstract =

    work page doi:10.48550/arxiv.2403.12108

  9. [9]

    and Su, Buxin and Collina, Natalie and Deng, Zhun and Su, Weijie , month = jan, year =

    Wen, Garrett G. and Su, Buxin and Collina, Natalie and Deng, Zhun and Su, Weijie , month = jan, year =. Recommending. doi:10.48550/arXiv.2601.15249 , abstract =

    work page doi:10.48550/arxiv.2601.15249

  10. [10]

    Su, Buxin and Collina, Natalie and Wen, Garrett and Li, Didong and Cho, Kyunghyun and Fan, Jianqing and Zhao, Bingxin and Su, Weijie , month = nov, year =. How to. doi:10.48550/arXiv.2510.02143 , abstract =

    work page doi:10.48550/arxiv.2510.02143

  11. [11]

    Handbook of

    Brandt, Felix and Conitzer, Vincent and Endriss, Ulle and Lang, Jerome and Procaccia, Ariel D , year =. Handbook of

    work page

  12. [12]

    van der and Dudoit, Sandrine and Laan, Mark J

    Vaart, Aad W. van der and Dudoit, Sandrine and Laan, Mark J. van der , year =. Oracle inequalities for multi-fold cross validation , volume =. Statistics & Risk Modeling , publisher =

    work page

  13. [13]

    Biometrika , author =

    Rank. Biometrika , author =

    work page

  14. [14]

    Grattafiori, Aaron and Dubey, Abhimanyu and Jauhri, Abhinav and Pandey, Abhinav and Kadian, Abhishek and Al-Dahle, Ahmad and Letman, Aiesha and Mathur, Akhil and Schelten, Alan and Vaughan, Alex and Yang, Amy and Fan, Angela and Goyal, Anirudh and Hartshorn, Anthony and Yang, Aobo and Mitra, Archi and Sravankumar, Archie and Korenev, Artem and Hinsvark, A...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2407.21783

  15. [15]

    OpenAI and Achiam, Josh and Adler, Steven and Agarwal, Sandhini and Ahmad, Lama and Akkaya, Ilge and Aleman, Florencia Leoni and Almeida, Diogo and Altenschmidt, Janko and Altman, Sam and Anadkat, Shyamal and Avila, Red and Babuschkin, Igor and Balaji, Suchir and Balcom, Valerie and Baltescu, Paul and Bao, Haiming and Bavarian, Mohammad and Belgum, Jeff a...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.08774

  16. [16]

    Yu, Sungduk and Luo, Man and Madasu, Avinash and Lal, Vasudev and Howard, Phillip , month = feb, year =. Is. doi:10.48550/arXiv.2502.19614 , abstract =

    work page doi:10.48550/arxiv.2502.19614

  17. [17]

    A Voting-Based System for Ethical Decision Making

    Noothigattu, Ritesh and Gaikwad, Snehalkumar 'Neil' S. and Awad, Edmond and Dsouza, Sohan and Rahwan, Iyad and Ravikumar, Pradeep and Procaccia, Ariel D. , month = dec, year =. A. doi:10.48550/arXiv.1709.06692 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1709.06692

  18. [18]

    Econometrica , author =

    A. Econometrica , author =

    work page

  19. [19]

    Arrow, Kenneth , year =. Social

    work page

  20. [20]

    Journal of Machine Learning Research , author =

    Simple,. Journal of Machine Learning Research , author =

    work page

  21. [21]

    Iterative ranking from pair-wise comparisons , url =

    work page

  22. [22]

    Feeling the Bern: Adaptive Estimators for Bernoulli Probabilities of Pairwise Comparisons

    Shah, Nihar B. and Balakrishnan, Sivaraman and Wainwright, Martin J. , month = mar, year =. Feeling the. doi:10.48550/arXiv.1603.06881 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1603.06881

  23. [23]

    Stretching the Effectiveness of MLE from Accuracy to Bias for Pairwise Comparisons

    Wang, Jingyan and Shah, Nihar B. and Ravi, R. , month = jun, year =. Stretching the. doi:10.48550/arXiv.1906.04066 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1906.04066 1906

  24. [24]

    ICLR , author =

    Gene. ICLR , author =

    work page

  25. [25]

    Lecture notes on:

    Wu, Yihong , month = jan, year =. Lecture notes on:

    work page

  26. [26]

    A studentized permutation test for the nonparametric

    Konietschke, Frank and Pauly, Markus , month = jan, year =. A studentized permutation test for the nonparametric. Electronic Journal of Statistics , publisher =. doi:10.1214/12-EJS714 , abstract =

    work page doi:10.1214/12-ejs714

  27. [27]

    The Annals of Statistics , author =

    Exact and asymptotically robust permutation tests , volume =. The Annals of Statistics , author =. 2013 , note =. doi:10.1214/13-AOS1090 , abstract =

    work page doi:10.1214/13-aos1090 2013

  28. [28]

    doi:10.48550/ARXIV.2510.08867 , abstract =

    Sahu, Gaurav and Larochelle, Hugo and Charlin, Laurent and Pal, Christopher , year =. doi:10.48550/ARXIV.2510.08867 , abstract =

    work page doi:10.48550/arxiv.2510.08867

  29. [29]

    Fast Causal Inference with Non-Random Missingness by Test-Wise Deletion

    Strobl, Eric V. and Visweswaran, Shyam and Spirtes, Peter L. , month = may, year =. Fast. doi:10.48550/arXiv.1705.09031 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1705.09031

  30. [30]

    On the identifiability and estimation of functional causal models in the presence of outcome-dependent selection , isbn =

    Zhang, Kun and Zhang, Jiji and Huang, Biwei and Schölkopf, Bernhard and Glymour, Clark , month = jun, year =. On the identifiability and estimation of functional causal models in the presence of outcome-dependent selection , isbn =. Proceedings of the

    work page

  31. [31]

    Applied Informatics , author =

    Causal discovery and inference: concepts and recent methodological advances , volume =. Applied Informatics , author =. 2016 , pages =. doi:10.1186/s40535-016-0018-x , abstract =

    work page doi:10.1186/s40535-016-0018-x 2016

  32. [32]

    Tu, Ruibo and Zhang, Kun and Ackermann, Paul and Bertilson, Bo Christer and Glymour, Clark and Kjellström, Hedvig and Zhang, Cheng , month = jul, year =. Causal. doi:10.48550/arXiv.1807.04010 , abstract =

    work page doi:10.48550/arxiv.1807.04010

  33. [33]

    Challenges, experiments, and computational solutions in peer review , volume =. Commun. ACM , author =. 2022 , pages =. doi:10.1145/3528086 , abstract =

    work page doi:10.1145/3528086 2022

  34. [35]

    Chatterjee, Sabyasachi and Guntuboyina, Adityanand and Sen, Bodhisattva , month = dec, year =. On. doi:10.48550/arXiv.1311.3765 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1311.3765

  35. [36]

    Multivariate extensions of isotonic regression and total variation denoising via entire monotonicity and Hardy - Krause variation, June 2020

    Fang, Billy and Guntuboyina, Adityanand and Sen, Bodhisattva , month = jun, year =. Multivariate extensions of isotonic regression and total variation denoising via entire monotonicity and. doi:10.48550/arXiv.1903.01395 , abstract =

    work page doi:10.48550/arxiv.1903.01395 1903

  36. [37]

    Lloyd, E. H. , year =. Least-. Biometrika , publisher =. doi:10.2307/2332466 , number =

    work page doi:10.2307/2332466

  37. [38]

    doi:10.48550/arXiv.2512.10895 , abstract =

    Ding, Lijie and Thomson, Janell and Taylor, Jon and Do, Changwoo , month = dec, year =. doi:10.48550/arXiv.2512.10895 , abstract =

    work page doi:10.48550/arxiv.2512.10895

  38. [39]

    , year =

    Manski, Charles F. , year =. Anatomy of the. The Journal of Human Resources , publisher =. doi:10.2307/145818 , abstract =

    work page doi:10.2307/145818

  39. [40]

    ProPublica , author =

    An. ProPublica , author =

    work page

  40. [41]

    Justia , month = may, year =

    Bail. Justia , month = may, year =

    work page

  41. [42]

    and Barber, Rina Foygel and Willett, Rebecca , month = apr, year =

    Soloff, Jake A. and Barber, Rina Foygel and Willett, Rebecca , month = apr, year =. Building a stable classifier with the inflated argmax , url =. doi:10.48550/arXiv.2405.14064 , abstract =

    work page doi:10.48550/arxiv.2405.14064

  42. [43]

    On the bias, risk and consistency of sample means in multi-armed bandits , url =

    Shin, Jaehyeok and Ramdas, Aaditya and Rinaldo, Alessandro , month = apr, year =. On the bias, risk and consistency of sample means in multi-armed bandits , url =. doi:10.48550/arXiv.1902.00746 , abstract =

    work page doi:10.48550/arxiv.1902.00746 1902

  43. [44]

    How much does your data exploration overfit?

    Russo, Daniel and Zou, James , month = oct, year =. How much does your data exploration overfit?. doi:10.48550/arXiv.1511.05219 , abstract =

    work page doi:10.48550/arxiv.1511.05219

  44. [45]

    Asymptotics of cross-validation , volume =

    Austern, Morgane and Zhou, Wenda , month = nov, year =. Asymptotics of cross-validation , volume =. Annales de l'Institut Henri Poincaré, Probabilités et Statistiques , publisher =. doi:10.1214/24-AIHP1488 , abstract =

    work page doi:10.1214/24-aihp1488

  45. [46]

    Non-stochastic Best Arm Identification and Hyperparameter Optimization

    Jamieson, Kevin and Talwalkar, Ameet , month = feb, year =. Non-stochastic. doi:10.48550/arXiv.1502.07943 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1502.07943

  46. [47]

    Gabillon, Victor and Ghavamzadeh, Mohammad and Lazaric, Alessandro and Bubeck, Sébastien , year =. Multi-. Advances in

    work page

  47. [48]

    Locally minimax optimal confidence sets for the best model , url =

    Kim, Ilmun and Ramdas, Aaditya , month = sep, year =. Locally minimax optimal confidence sets for the best model , url =. doi:10.48550/arXiv.2503.21639 , abstract =

    work page doi:10.48550/arxiv.2503.21639

  48. [49]

    Winners with

    Zhang, Tianyu and Lee, Hao and Lei, Jing , month = jan, year =. Winners with. doi:10.48550/arXiv.2408.02060 , abstract =

    work page doi:10.48550/arxiv.2408.02060

  49. [53]

    Ramsey, J. B. , year =. Tests for. Journal of the Royal Statistical Society. Series B (Methodological) , publisher =

    work page

  50. [54]

    HotelRec : a Novel Very Large - Scale Hotel Recommendation Dataset , February 2020

    Antognini, Diego and Faltings, Boi , month = feb, year =. doi:10.48550/arXiv.2002.06854 , abstract =

    work page doi:10.48550/arxiv.2002.06854 2002

  51. [55]

    Regret in

    Aldridge, Irene , month = dec, year =. Regret in. doi:10.2139/ssrn.5987415 , abstract =

    work page doi:10.2139/ssrn.5987415

  52. [56]

    Empirical

    Hsu, Chih-Wei and Kveton, Branislav and Meshi, Ofer and Mladenov, Martin and Szepesvari, Csaba , month = jun, year =. Empirical. doi:10.48550/arXiv.1904.02664 , abstract =

    work page doi:10.48550/arxiv.1904.02664 1904

  53. [57]

    Random Design A nalysis of Ridge Regression

    Random. Foundations of Computational Mathematics , author =. 2014 , pages =. doi:10.1007/s10208-014-9192-1 , abstract =

    work page doi:10.1007/s10208-014-9192-1 2014

  54. [58]

    Alan and Roy, Jason A

    Sauer, Brian and Brookhart, M. Alan and Roy, Jason A. and VanderWeele, Tyler J. , month = jan, year =. Covariate. Developing a

    work page

  55. [59]

    , year =

    Kahneman, Daniel and Sibony, Olivier and Sunstein, Cass R. , year =. Noise:

    work page

  56. [60]

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

    Isotonic. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2021 , note =. doi:10.1111/rssb.12450 , abstract =

    work page doi:10.1111/rssb.12450 2021

  57. [61]

    , year =

    Han, Qiyang and Wang, Tengyao and Chatterjee, Sabyasachi and Samworth, Richard J. , year =. Isotonic. The Annals of Statistics , publisher =

    work page

  58. [62]

    , month = aug, year =

    Berker, Ratip Emin and Armstrong, Ben and Conitzer, Vincent and Shah, Nihar B. , month = aug, year =. Designing. doi:10.48550/arXiv.2508.17177 , abstract =

    work page doi:10.48550/arxiv.2508.17177

  59. [63]

    Inside Higher Ed , author =

    ‘. Inside Higher Ed , author =. 2022 , keywords =

    work page 2022

  60. [64]

    US News & World Report , author =

    How. US News & World Report , author =

    work page

  61. [65]

    Psychological Medicine , author =

    Comparison of different scoring methods based on latent variable models of the. Psychological Medicine , author =. 2022 , keywords =. doi:10.1017/S0033291721000131 , abstract =

    work page doi:10.1017/s0033291721000131 2022

  62. [66]

    General Hospital Psychiatry , author =

    A diagnostic meta-analysis of the. General Hospital Psychiatry , author =. 2015 , keywords =. doi:10.1016/j.genhosppsych.2014.09.009 , abstract =

    work page doi:10.1016/j.genhosppsych.2014.09.009 2015

  63. [67]

    Journal of Affective Disorders , author =

    The. Journal of Affective Disorders , author =. 2018 , keywords =. doi:10.1016/j.jad.2018.02.045 , abstract =

    work page doi:10.1016/j.jad.2018.02.045 2018

  64. [68]

    Kim, Sunhae and Lee, Hye-Kyung and Lee, Kounseok , month = jan, year =. Which. International Journal of Environmental Research and Public Health , publisher =. doi:10.3390/ijerph18073339 , abstract =

    work page doi:10.3390/ijerph18073339

  65. [70]

    Minimax rates of estimation for high-dimensional linear regression over $\ell_q$-balls

    Raskutti, Garvesh and Wainwright, Martin J. and Yu, Bin , month = oct, year =. Minimax rates of estimation for high-dimensional linear regression over \. doi:10.48550/arXiv.0910.2042 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.0910.2042 2042

  66. [73]

    Breaking the $1/\sqrt{n}$ Barrier: Faster Rates for Permutation-based Models in Polynomial Time

    Mao, Cheng and Pananjady, Ashwin and Wainwright, Martin J. , month = jun, year =. Breaking the \ 1/. doi:10.48550/arXiv.1802.09963 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1802.09963

  67. [74]

    Low Permutation-rank Matrices: Structural Properties and Noisy Completion

    Shah, Nihar B. and Balakrishnan, Sivaraman and Wainwright, Martin J. , month = sep, year =. Low. doi:10.48550/arXiv.1709.00127 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1709.00127

  68. [75]

    Active Ranking from Pairwise Comparisons and when Parametric Assumptions Don't Help

    Heckel, Reinhard and Shah, Nihar B. and Ramchandran, Kannan and Wainwright, Martin J. , month = sep, year =. Active. doi:10.48550/arXiv.1606.08842 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1606.08842

  69. [76]

    Stochastically Transitive Models for Pairwise Comparisons: Statistical and Computational Issues

    Shah, Nihar B. and Balakrishnan, Sivaraman and Guntuboyina, Adityanand and Wainwright, Martin J. , month = sep, year =. Stochastically. doi:10.48550/arXiv.1510.05610 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1510.05610

  70. [77]

    Journal of Inequalities and Applications , author =

    Rank-one perturbation bounds for singular values of arbitrary matrices , volume =. Journal of Inequalities and Applications , author =. 2019 , keywords =. doi:10.1186/s13660-019-2089-4 , abstract =

    work page doi:10.1186/s13660-019-2089-4 2019

  71. [78]

    and Toubia, Olivier and Evgeniou, Theodoros and Befurt, Rene and Dzyabura, Daria , month = jun, year =

    Hauser, John R. and Toubia, Olivier and Evgeniou, Theodoros and Befurt, Rene and Dzyabura, Daria , month = jun, year =. Disjunctions of. Journal of Marketing Research , publisher =. doi:10.1509/jmkr.47.3.485 , abstract =

    work page doi:10.1509/jmkr.47.3.485

  72. [79]

    and Verma, Rohit , month = feb, year =

    Zhang, Jie J. and Verma, Rohit , month = feb, year =. What

    work page

  73. [80]

    Proceedings of the AAAI Conference on Artificial Intelligence , author =

    Non-. Proceedings of the AAAI Conference on Artificial Intelligence , author =. 2019 , pages =. doi:10.1609/aaai.v33i01.33014304 , abstract =

    work page doi:10.1609/aaai.v33i01.33014304 2019

  74. [82]

    Tourism Management , author =

    Importance-performance analysis:. Tourism Management , author =. 2015 , keywords =. doi:10.1016/j.tourman.2014.10.022 , abstract =

    work page doi:10.1016/j.tourman.2014.10.022 2015

  75. [83]

    Measuring Business Excellence , author =

    Importance-performance analysis of service attributes and its impact on decision making in the mobile telecommunication industry , volume =. Measuring Business Excellence , author =. 2009 , pages =. doi:10.1108/13683040910943072 , abstract =

    work page doi:10.1108/13683040910943072 2009

  76. [84]

    , month = mar, year =

    Deng, Jinyang and Pierskalla, Chad D. , month = mar, year =. Linking. Sustainability , publisher =. doi:10.3390/su10030704 , abstract =

    work page doi:10.3390/su10030704

  77. [86]

    , editor =

    Dawes, Robyn M. , editor =. The robust beauty of improper linear models in decision making , isbn =. Judgment under. 1982 , pages =. doi:10.1017/CBO9780511809477.029 , abstract =

    work page doi:10.1017/cbo9780511809477.029 1982

  78. [87]

    On matrix estimation under monotonicity constraints

    Chatterjee, Sabyasachi and Guntuboyina, Adityanand and Sen, Bodhisattva , month = nov, year =. On matrix estimation under monotonicity constraints , url =. doi:10.48550/arXiv.1506.03430 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1506.03430

  79. [88]

    and Chin, Elizabeth T

    Huynh, Benjamin Q. and Chin, Elizabeth T. and Koenecke, Allison and Ouyang, Derek and Ho, Daniel E. and Kiang, Mathew V. and Rehkopf, David H. , month = feb, year =. Mitigating allocative tradeoffs and harms in an environmental justice data tool , volume =. Nature Machine Intelligence , publisher =. doi:10.1038/s42256-024-00793-y , abstract =

    work page doi:10.1038/s42256-024-00793-y

  80. [90]

    Advances in Neural Information Processing Systems , author =

    Learning. Advances in Neural Information Processing Systems , author =. 2024 , pages =

    work page 2024

Showing first 80 references.