# Finetuning LLMs for Human Behavior Prediction in Social Science Experiments

arXiv: https://arxiv.org/abs/2509.05830 · HTML: https://mlsanity.com/paper/2509.05830/

## Key facts

- Cited by 39 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 41 references that are in the map.
- First posted on arXiv 6 Sep 2025.

## Builds on (most cited first, 21 of 21)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [SimPO: Simple Preference Optimization with a Reference-Free Reward](https://mlsanity.com/paper/2405.14734/) (2024)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [Out of One, Many: Using Language Models to Simulate Human Samples](https://mlsanity.com/paper/2209.06899/) (2022)
- [LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals](https://mlsanity.com/paper/2411.10109/) (2024)
- [Large language models that replace human participants can harmfully misportray and flatten identity groups](https://mlsanity.com/paper/2402.01908/) (2024)
- [Training Socially Aligned Language Models on Simulated Social Interactions](https://mlsanity.com/paper/2305.16960/) (2023)
- [Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinions](https://mlsanity.com/paper/2502.16761/) (2025)
- [Virtual Personas for Language Models via an Anthology of Backstories](https://mlsanity.com/paper/2407.06576/) (2024)
- [1.4 Million Open-Source Distilled Reasoning Dataset to Empower Large Language Model Training](https://mlsanity.com/paper/2503.19633/) (2025)
- [AI-Augmented Surveys: Leveraging Large Language Models and Surveys for Opinion Prediction](https://mlsanity.com/paper/2305.09620/) (2023)
- [Centaur: a foundation model of human cognition](https://mlsanity.com/paper/2410.20268/) (2024)
- [Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptions](https://mlsanity.com/paper/2502.20897/) (2025)
- [Language Models Trained on Media Diets Can Predict Public Opinion](https://mlsanity.com/paper/2303.16779/) (2023)
- [Take Caution in Using LLMs as Human Surrogates: Scylla Ex Machina](https://mlsanity.com/paper/2410.19599/) (2024)
- [Using Reinforcement Learning to Train Large Language Models to Explain Human Decisions](https://mlsanity.com/paper/2505.11614/) (2025)
- [Be.FM: Open Foundation Models for Human Behavior](https://mlsanity.com/paper/2505.23058/) (2025)
- [From ChatGPT to DeepSeek: Can LLMs Simulate Humanity?](https://mlsanity.com/paper/2502.18210/) (2025)

## Built on by (most cited first, 30 of 39)

- [HumanLM: Simulating Users with State Alignment Beats Response Imitation](https://mlsanity.com/paper/2603.03303/) (2026)
- [This human study did not involve human subjects: Validating LLM simulations as behavioral evidence](https://mlsanity.com/paper/2602.15785/) (2026)
- [The Need for a Socially-Grounded Persona Framework for User Simulation](https://mlsanity.com/paper/2601.07110/) (2026)
- [OdysSim: Building Foundation Models for Human Behavior Simulation](https://mlsanity.com/paper/2606.14199/) (2026)
- [When Can Digital Personas Reliably Approximate Human Survey Findings?](https://mlsanity.com/paper/2605.10659/) (2026)
- [How Well Can LLM Agents Simulate End-User Security and Privacy Attitudes and Behaviors?](https://mlsanity.com/paper/2602.18464/) (2026)
- [Beyond the Mean: Three-Axis Fidelity for Aligning LLM-Based Survey Simulators from Small Pilot Data](https://mlsanity.com/paper/2606.28963/) (2026)
- [The Judgment-Consequence Gap: LLM Moral Reasoning in Healthcare Decisions](https://mlsanity.com/paper/2608.05583/) (2026)
- [HumanStudy-Bench: Towards AI Agent Design for Participant Simulation](https://mlsanity.com/paper/2602.00685/) (2026)
- [SYN-DIGITS: A Synthetic Control Framework for Calibrated Digital Twin Simulation](https://mlsanity.com/paper/2604.07513/) (2026)
- [The Collapse of Heterogeneity in Silicon Philosophers](https://mlsanity.com/paper/2604.23575/) (2026)
- [Synthetic Personalities: How Well Can LLMs Mimic Individual Respondents Using Socio-Economic Microdata?](https://mlsanity.com/paper/2606.04592/) (2026)
- [How Well Do Large Language Models Capture Human Personality?](https://mlsanity.com/paper/2606.18263/) (2026)
- [Mitigating LLM-based p-Hacking by Preregistering for the Next LLM](https://mlsanity.com/paper/2606.27687/) (2026)
- [Graph-Based Alternatives to LLMs for Human Simulation](https://mlsanity.com/paper/2511.02135/) (2025)
- [BehaviorBench: Benchmarking Foundation Models for Behavioral Science Tasks](https://mlsanity.com/paper/2606.24162/) (2026)
- [Item-Mean Surrogates: Why Richer Persona Data Fail to Improve LLMs as Human Surrogates](https://mlsanity.com/paper/2608.29455/) (2026)
- [Data-Driven Persona-Conditioned Agents for A/B Test Simulation](https://mlsanity.com/paper/2609.01038/) (2026)
- [When Can LLM Digital Twins Reduce Human Measurement? From Behavioral Fidelity to Statistical Substitutability](https://mlsanity.com/paper/2609.07987/) (2026)
- [Verifiable Social Reasoning for LLM Assistants](https://mlsanity.com/paper/2609.17496/) (2026)
- [The Limits of Simulated Societies: How Post-Training and Survey Fine-Tuning Erase Cross-Cultural Variance](https://mlsanity.com/paper/2609.25760/) (2026)
- [InterviewSim: A Scalable Framework for Interview-Grounded Personality Simulation](https://mlsanity.com/paper/2602.20294/) (2026)
- [Validated Hypotheses as a Lens for Human-Likeness Evaluation in AI Agents](https://mlsanity.com/paper/2605.15473/) (2026)
- [Predicting Causal Effects from Natural Language Queries using Structured Representations](https://mlsanity.com/paper/2605.29631/) (2026)
- [RedditPersona: A Modular Framework for Community-Conditioned LLM Adaptation from Reddit](https://mlsanity.com/paper/2606.06027/) (2026)
- [Multimodal Large Language Models as Synthetic Participants in Video-Based Studies: An Evaluation](https://mlsanity.com/paper/2606.07541/) (2026)
- [Posterior Twins: Distributional Behavioral Simulation for Enterprise Decisions](https://mlsanity.com/paper/2606.16415/) (2026)
- [Will Scaling Improve Social Simulation with LLMs?](https://mlsanity.com/paper/2607.02464/) (2026)
- [SafeStep: AI-powered Travel Assistance for Elderly People with Frailty or Dementia](https://mlsanity.com/paper/2607.21156/) (2026)
- [Mind the Gaps: Mixture-of-Minds for Human Simulation](https://mlsanity.com/paper/2608.06115/) (2026)

## Abstract (arXiv)

Large language models (LLMs) offer a powerful opportunity to simulate the results of social science experiments. In this work, we demonstrate that finetuning LLMs directly on individual-level responses from past experiments meaningfully improves the accuracy of such simulations across diverse social science domains. We construct SocSci210 via an automatic pipeline, a dataset comprising 2.9 million responses from 400,491 participants in 210 open-source social science experiments. Through finetuning, we achieve multiple levels of generalization. In completely unseen studies, our strongest model, Socrates-Qwen-14B, produces predictions that are 26% more aligned with distributions of human responses to diverse outcome questions under varying conditions relative to its base model (Qwen2.5-14B), outperforming GPT-4o by 13%. By finetuning on a subset of conditions in a study, generalization to new unseen conditions is particularly robust, improving by 71%. Since SocSci210 contains rich demographic information, we reduce demographic parity difference, a measure of bias, by 10.6% through finetuning. Because social sciences routinely generate rich, topic-specific datasets, our findings indicate that finetuning on such data could enable more accurate simulations for experimental hypothesis screening. We release our data, models and finetuning code at stanfordhci.github.io/socrates.