The Need for a Socially-Grounded Persona Framework for User Simulation
This work addresses the need for more realistic user simulations in AI, offering a novel framework that could enhance applications like chatbots or social AI, though it is incremental in improving persona construction methods.
The paper tackled the problem of synthetic personas for social simulation by introducing SCOPE, a socially grounded framework built from detailed sociopsychological data, finding that demographic-only personas explain only ~1.5% of variance in human response similarity and that adding sociopsychological facets improves behavioral prediction and reduces bias.
Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes or summaries. We revisit persona creation by introducing SCOPE, a socially grounded framework for persona construction and evaluation, built from a 141-item, two-hour sociopsychological protocol collected from 124 U.S.-based participants. Across seven models, we find that demographic-only personas are a structural bottleneck: demographics explain only ~1.5% of variance in human response similarity. Adding sociopsychological facets improves behavioral prediction and reduces over-accentuation, and non-demographic personas based on values and identity achieve strong alignment with substantially lower bias. These trends generalize to SimBench (441 aligned questions), where SCOPE personas outperform default prompting and NVIDIA Nemotron personas, and SCOPE augmentation improves Nemotron-based personas. Our results indicate that persona quality depends on sociopsychological structure rather than demographic templates or summaries.