When Can Digital Personas Reliably Approximate Human Survey Findings?

TL;DR

This study evaluates when digital personas generated by LLMs can reliably mimic human survey responses, highlighting their strengths in stable domains and limitations in individual prediction and complex respondent structures.

Contribution

It systematically compares multiple persona architectures and LLMs, providing practical guidelines for their use in survey research based on response variability and question type.

Findings

Digital personas align well with stable attribute responses.

Performance varies with question variability and respondent heterogeneity.

Retrieval-augmented models show notable improvements.

Abstract

Digital personas powered by Large Language Models (LLMs) are increasingly proposed as substitutes for human survey respondents, yet it remains unclear when they can reliably approximate human survey findings. We answer this question using the LISS panel, constructing personas from respondents' background variables and pre-2023 survey histories, then testing them against the same respondents' held-out post-cutoff answers. Across four persona architectures, three LLMs, and two prediction tasks, we assess performance at the question, respondent, distributional, equity, and clustering levels. Digital personas improve alignment with human response distributions, especially in domains tied to stable attributes and values, but remain limited for individual prediction and fail to recover multivariate respondent structure. Retrieval-augmented architectures provide the clearest gains, but performance depends more on human response structure than on model choice: personas perform best for low-variability questions and common respondent patterns, and worst for subjective, heterogeneous, or rare responses. Our results provide practical guidance on when digital personas could be appropriate for survey research and when human validation remains necessary.