Social Learning with Selective Sampling

2026-08-17

AI Summary

This study addresses social learning bias arising from selective sampling by proposing a robust learning framework based on Bayesian agents. By explicitly modeling the selective sampling mechanism, this work reveals that the endogenous observation network inherently contains critical information for bias correction. Results demonstrate that Bayesian agents can effectively leverage this endogenous information to eliminate selection bias and achieve asymptotic learning, even under non-expansive observation conditions. Theoretically establishing the pivotal role of endogenous network structure in overcoming data selection bias, this research offers a novel paradigm for robust social learning in complex environments.

Abstract

This paper studies how robust social learning is when sampling is selective, i.e., some types of actions are more likely to be sampled by successors. We show that Bayesian agents can achieve asymptotic learning despite non-expanding observations, because the endogenous observation network itself carries information and agents have ways to undo the selection bias.

Problem

• Social Learning
• Selective Sampling
• Robustness
• Selection Bias

Innovation

• Social Learning
• Selective Sampling
• Bayesian Agents
• Endogenous Observation Network
• Selection Bias