Beyond Static Snapshots: Dynamic Modeling and Forecasting of Group-Level Value Evolution with Large Language Models

TL;DR

The paper tackles the problem of modeling dynamic, group-level values rather than static snapshots by introducing a two-stage framework: Value Trajectory Prediction (VTP) and Event-Aware Prediction (EAP). It builds a multi-wave, group-stratified dataset from the World Values Survey for China and the U.S., prompting LLMs with longitudinal context and historical trajectories, and aligns external events to value dimensions via a Value-Driven Major Event Matching mechanism. Empirical results across five open-source LLM families show that EAP consistently improves predictive accuracy over Vanilla and VTP, with notable cross-country heterogeneity—U.S. groups exhibit higher volatility and younger cohorts are more event-sensitive. The work advances dynamic social simulation by enabling interpretable, event-informed forecasts of group-value evolution and provides a valuable dataset and methodology for social scientists and policymakers to anticipate societal shifts.

Abstract

Social simulation is critical for mining complex social dynamics and supporting data-driven decision making. LLM-based methods have emerged as powerful tools for this task by leveraging human-like social questionnaire responses to model group behaviors. Existing LLM-based approaches predominantly focus on group-level values at discrete time points, treating them as static snapshots rather than dynamic processes. However, group-level values are not fixed but shaped by long-term social changes. Modeling their dynamics is thus crucial for accurate social evolution prediction--a key challenge in both data mining and social science. This problem remains underexplored due to limited longitudinal data, group heterogeneity, and intricate historical event impacts. To bridge this gap, we propose a novel framework for group-level dynamic social simulation by integrating historical value trajectories into LLM-based human response modeling. We select China and the U.S. as representative contexts, conducting stratified simulations across four core sociodemographic dimensions (gender, age, education, income). Using the World Values Survey, we construct a multi-wave, group-level longitudinal dataset to capture historical value evolution, and then propose the first event-based prediction method for this task, unifying social events, current value states, and group attributes into a single framework. Evaluations across five LLM families show substantial gains: a maximum 30.88\% improvement on seen questions and 33.97\% on unseen questions over the Vanilla baseline. We further find notable cross-group heterogeneity: U.S. groups are more volatile than Chinese groups, and younger groups in both countries are more sensitive to external changes. These findings advance LLM-based social simulation and provide new insights for social scientists to understand and predict social value changes.

Figures (12)

• Figure 1: A example of two social groups in the United States from the World Values Survey (WVS), reflecting the dynamic evolution of social values.
• Figure 2: The overview of our proposed framework.
• Figure 3: Experimental analysis of varying numbers of selected events for China and the United States.
• Figure 4: Changes in the 23 U.S. social groups from wave 6 to wave 7 by comparing predicted and observed values.
• Figure 5: Changes in the 28 China social groups from wave 6 to wave 7 by comparing predicted and observed values.