Skip to content
Back to archive

[Submitted] Reconstructing Consumer Reasoning via Persona-Based LLM Agents

Published:

Research Status & Connection

This manuscript was submitted to an academic journal on July 11, 2026. It extends the persona-based LLM agent framework developed in my Master’s Thesis to examine heterogeneous consumer decision pathways.


Research Overview

This study examines how consumers form judgments about emerging mobility services. It develops a theory-guided framework in which persona-based LLM agents reconstruct alternative decision pathways involving intuitive judgment, structured deliberation, and information from experts and peers.

The analysis indicates that service adoption and intended behavioral change may reflect different decision processes, and that the pathway most closely aligned with observed responses can vary across consumers.


Methodology

The framework combines behavioral theory with persona-based LLM agent simulation:

  1. Theory-Guided Path Reconstruction: Compares alternative reasoning pathways by integrating dual-process theory, the Theory of Planned Behavior, and different interpersonal information channels.
  2. Respondent-Level Evaluation: Represents survey respondents as persona-constrained agents and evaluates the reconstructed decisions against observed responses, followed by an analysis of individual-level pathway differences.

Rather than treating LLM agents as substitutes for human respondents, this research uses them as analytical proxies for comparing theoretically specified consumer decision processes.

For the foundational persona-based simulation framework, please refer to the Master’s Thesis Archive.


Share this research:

Next
Simulating Consumer Decisions via Persona-based LLM Agents (Master's Thesis)