Excited to share our new preprint, BeliefSim, where we ask: Can LLMs simulate demographic differences in misinformation susceptibility without reducing people to surface-level demographic labels? We find that beliefs provide a stronger prior than demographics alone. 🧠 (1/n)