Statistical Foundations of LLM-based A/B Testing: A Surrogacy Framework for Human Causal Inference

• URL: http://arxiv.org/abs/2606.17165v3
• Date: Tue, 23 Jun 2026 14:53:08 GMT
• Title: Statistical Foundations of LLM-based A/B Testing: A Surrogacy Framework for Human Causal Inference
• Abstract summary: We study when a treatment effect estimated on large language models (LLMs) can recover the effect for the human population of interest.<n>We develop a statistical framework that adapts surrogate endpoint theory to LLMs.<n>A central takeaway is that A/B testing on LLM responses is correct only by assumption, whereas A/B testing on humans is correct by design.
• Score: 0.509780930114934
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Organizations and researchers show increasing interest in using large language models (LLMs) in place of human participants in A/B tests, in the hope of experimenting faster and at lower cost. We study when a treatment effect estimated on LLM outcomes can recover the effect for the human population of interest. Distributional equivalence between LLM and human outcomes would make any standard estimator valid but is unrealistic. We therefore develop a statistical framework that adapts surrogate endpoint theory to LLMs, showing that calibrating LLM outcomes to human outcomes identifies the average treatment effect under surrogacy and comparability conditions that are jointly weaker than distributional equivalence. We present a falsification test for surrogacy and a bound on the worst-case bias from limited overlap between the LLM and human samples. We further show that the stochasticity inherent to LLMs can weaken surrogacy for identification while also introducing bias and variance during estimation, but that using an average over multiple LLM draws per unit as the surrogate mitigates these issues. Simulations validate the results, and an empirical application to the Upworthy Research Archive dataset shows that raw LLM outputs recover only 39% of the human treatment effect while nonparametric calibration closes the gap. A central takeaway is that A/B testing on LLM responses is correct only by assumption, whereas A/B testing on humans is correct by design, and that the required assumptions are hardest to justify precisely where LLMs promise the greatest benefit. We discuss the choice of LLM, prompting, and temperature as design variables, the compounded challenge posed by long-term outcomes, and how to size human pilot studies for validation.

Related papers

• Human-AI Collaboration in Requirements Engineering: Evidence of the Negative Effect of LLMs on Requirements Inspection [41.425568895513806]
  Large language models (LLMs) have stimulated interest in their potential to support requirements engineering tasks.<n>We aim to investigate the impact of LLM support on human-performed RI.
  arXiv  Detail & Related papers  (2026-08-21T16:58:48Z)
• Using Large Language Models as Low-Cost Statistical Estimators for Human-Response Data [0.282023281338763]
  We show that pretrained large language models induce risk-equivalent estimators of conditional expectations under squared loss.<n>We provide finite-sample concentration bounds and a calibration protocol with explicit decision rules.
  arXiv  Detail & Related papers  (2026-06-29T14:34:28Z)
• The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study [56.649987197786096]
  Large language models (LLMs) show potential as simulators of human behavior.<n>LLMs can induce unintended shifts in latent user attributes, causing user drift.<n>We formalize the confounding or selection bias that can arise due to user drift.
  arXiv  Detail & Related papers  (2026-05-20T06:09:41Z)
• PREBA: Surgical Duration Prediction via PCA-Weighted Retrieval-Augmented LLMs and Bayesian Averaging Aggregation [51.96735866702332]
  PREBA is a retrieval-augmented framework that integrates PCA-weighted retrieval and Bayesian averaging aggregation.<n> PREBA significantly improves performance-for instance, reducing MAE by up to 40% and raising R2 from -0.13 to 0.62 over zero-shot inference.
  arXiv  Detail & Related papers  (2026-02-27T07:19:23Z)
• This human study did not involve human subjects: Validating LLM simulations as behavioral evidence [15.56427716190418]
  Heuristic approaches seek to establish that simulated and observed human behavior are interchangeable.<n> statistical calibration combines auxiliary human data with statistical adjustments to account for discrepancies between observed and simulated responses.
  arXiv  Detail & Related papers  (2026-02-17T18:18:38Z)
• IV Co-Scientist: Multi-Agent LLM Framework for Causal Instrumental Variable Discovery [61.15184885636171]
  In the presence of confounding between an endogenous variable and the outcome, instrumental variables (IVs) are used to isolate the causal effect of the endogenous variable.<n>We investigate whether large language models (LLMs) can aid in this task.<n>We introduce IV Co-Scientist, a multi-agent system that proposes, critiques, and refines IVs for a given treatment-outcome pair.
  arXiv  Detail & Related papers  (2026-02-08T12:28:29Z)
• Can Finetuing LLMs on Small Human Samples Increase Heterogeneity, Alignment, and Belief-Action Coherence? [9.310571879281186]
  Large language models (LLMs) can serve as substitutes for human participants in survey and experimental research.<n>LLMs often fail to align with real human behavior, exhibiting limited diversity, systematic misalignment for minority subgroups, insufficient within-group variance, and discrepancies between stated beliefs and actions.<n>This study examines whether fine-tuning on a small subset of human survey data, such as that obtainable from a pilot study, can mitigate these issues and yield realistic simulated outcomes.
  arXiv  Detail & Related papers  (2025-11-26T09:50:42Z)
• Predicting Effects, Missing Distributions: Evaluating LLMs as Human Behavior Simulators in Operations Management [11.302500716500893]
  LLMs are emerging tools for simulating human behavior in business, economics, and social science.<n>This paper evaluates how well LLMs replicate human behavior in operations management.
  arXiv  Detail & Related papers  (2025-09-30T20:20:58Z)
• Mitigating Spurious Correlations in LLMs via Causality-Aware Post-Training [57.03005244917803]
  Large language models (LLMs) often fail on out-of-distribution (OOD) samples due to spurious correlations acquired during pre-training.<n>Here, we aim to mitigate such spurious correlations through causality-aware post-training (CAPT)<n> Experiments on the formal causal inference benchmark CLadder and the logical reasoning dataset PrOntoQA show that 3B-scale language models fine-tuned with CAPT can outperform both traditional SFT and larger LLMs on in-distribution (ID) and OOD tasks.
  arXiv  Detail & Related papers  (2025-06-11T06:30:28Z)
• Preference Leakage: A Contamination Problem in LLM-as-a-judge [69.96778498636071]
  Large Language Models (LLMs) as judges and LLM-based data synthesis have emerged as two fundamental LLM-driven data annotation methods.<n>In this work, we expose preference leakage, a contamination problem in LLM-as-a-judge caused by the relatedness between the synthetic data generators and LLM-based evaluators.
  arXiv  Detail & Related papers  (2025-02-03T17:13:03Z)
• Are Large Language Models Good Statisticians? [10.42853117200315]
  StatQA is a new benchmark designed for statistical analysis tasks. We show that even state-of-the-art models such as GPT-4o achieve a best performance of only 64.83%. While open-source LLMs show limited capability, those fine-tuned ones exhibit marked improvements.
  arXiv  Detail & Related papers  (2024-06-12T02:23:51Z)
• The Challenge of Using LLMs to Simulate Human Behavior: A Causal Inference Perspective [0.27624021966289597]
  Large Language Models (LLMs) have shown impressive potential to simulate human behavior.<n>We identify a fundamental challenge in using them to simulate experiments.<n>When LLM-simulated subjects are blind to the experimental design, variations in treatment systematically affect unspecified variables.
  arXiv  Detail & Related papers  (2023-12-24T16:32:35Z)
• Do LLMs exhibit human-like response biases? A case study in survey design [66.1850490474361]
  We investigate the extent to which large language models (LLMs) reflect human response biases, if at all. We design a dataset and framework to evaluate whether LLMs exhibit human-like response biases in survey questionnaires. Our comprehensive evaluation of nine models shows that popular open and commercial LLMs generally fail to reflect human-like behavior.
  arXiv  Detail & Related papers  (2023-11-07T15:40:43Z)
• Localized Debiased Machine Learning: Efficient Inference on Quantile Treatment Effects and Beyond [69.83813153444115]
  We consider an efficient estimating equation for the (local) quantile treatment effect ((L)QTE) in causal inference. Debiased machine learning (DML) is a data-splitting approach to estimating high-dimensional nuisances. We propose localized debiased machine learning (LDML), which avoids this burdensome step.
  arXiv  Detail & Related papers  (2019-12-30T14:42:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.