Inference and influence of network structure using snapshot social behavior without network data.

Area of Science:

• Social network analysis
• Computational social science
• Behavioral economics

Background:

• Population behaviors such as voting and vaccination are influenced by social network structures, which are often unobserved.
• Existing methods struggle to infer network properties from limited, snapshot-based behavioral data.

Purpose of the Study:

• To develop a novel method for jointly inferring social network structure and human behavior using only snapshot population-level behavioral data.
• To demonstrate the model's utility in predicting behavior and interpreting population homophily.

Main Methods:

• Utilizes a parsimonious few-parameter model, specifically a geometric sociodemographic network model.
• Incorporates a spin-based model to represent behavior dynamics.
• Applies the joint inference method to real-world datasets, including the European Union referendum and London mayoral elections.

Main Results:

• Successfully inferred behavior-specific network structures from snapshot behavioral datasets.
• Demonstrated the model's capability for both prediction and interpretation of population homophilic inclinations.
• Established a framework linking socioeconomic factors like inequality and social preferences to behavioral outcomes.

Conclusions:

• The developed method provides a powerful tool for understanding the interplay between social networks and population behavior.
• The findings offer insights into how network structures and social preferences shape collective actions.
• The research highlights potential network-sensitive policy interventions to mitigate societal polarization.