Read the Paper, Write the Code: Agentic Reproduction of Social-Science Results

TL;DR

This paper presents an agentic system that reproduces social science results from papers' methods and data alone, analyzing discrepancies to improve reproducibility with various LLMs.

Contribution

It introduces a novel approach for reproducing results solely from methods descriptions and data, without access to original code or results.

Findings

Agents can largely recover published results

Performance varies significantly across models and papers

Root causes of failures include agent errors and paper underspecification

Abstract

Recent work has used LLM agents to reproduce empirical social science results with access to both the data and code. We broaden this scope by asking: Can they reproduce results given only a paper's methods description and original data? We develop an agentic reproduction system that extracts structured methods descriptions from papers, runs reimplementations under strict information isolation -- agents never see the original code, results, or paper -- and enables deterministic, cell-level comparison of reproduced outputs to the original results. An error attribution step traces discrepancies through the system chain to identify root causes. Evaluating four agent scaffolds and four LLMs on 48 papers with human-verified reproducibility, we find that agents can largely recover published results, but performance varies substantially between models, scaffolds, and papers. Root cause analysis reveals that failures stem both from agent errors and from underspecification in the papers themselves.