[Paper] Revising Context, Shifting Simulated Stance: Auditing LLM-Based Stance Simulation in Online Discussions

Source: arXiv - 2606.06443v1

Overview

Large language models are increasingly used to simulate social media users and infer how individuals may respond to online discussions. However, it remains unclear whether these simulations reflect precise user‑specific beliefs or whether they are highly sensitive to semantically independent changes in conversational contexts.

In this work, we study counterfactual context revision as a framework for auditing LLM‑based stance simulation. Given an original online conversation, we first infer a target user’s stance toward a specific topic. We then apply controlled revision strategies to the conversational context and simulate the user’s stance again under the revised context.

We compare text‑only revision strategies with a multimodal one that incorporates meme‑based context and evaluate two main effectiveness metrics:

• average directional stance shift
• stance transition rate

The results reveal effective and robust stance transitions in both text‑only and multimodal strategies across different polarization‑preference mechanisms. Our study contributes an evaluation framework for understanding the context sensitivity of LLM‑based stance simulation and highlights both the promise and risk of using LLMs to simulate online opinion dynamics.

Key Contributions

• cs.CL
• cs.MM
• cs.SI

Methodology

Please refer to the full paper for detailed methodology.

Practical Implications

This research contributes to the advancement of cs.CL.

Authors

• Xinnong Zhang
• Wanting Shan
• Hanjia Lyu
• Zhongyu Wei
• Jiebo Luo

Paper Information

• arXiv ID: 2606.06443v1
• Categories: cs.CL, cs.MM, cs.SI
• Published: June 4, 2026
• PDF: Download PDF