What shifts threshold distributions in social contagions?

Abstract

Individual thresholds in social contagions capture what fraction of others must adopt a new product or behavior before an individual adopts it. Yet in a given choice context and population, we rarely know the empirical threshold distribution or what explains its heterogeneity. This limits the ability to predict diffusion dynamics and design effective network interventions. We address this by measuring thresholds with an incentivized online experiment in which participants make adoption decisions for products that vary in attractiveness and payoff uncertainty. The results show that product characteristics shift threshold distributions: lower attractiveness and higher uncertainty increase thresholds. Individual characteristics, however, remain essential for explaining threshold heterogeneity. We then use data-driven simulations to show that both product characteristics and individual characteristics moderate the effect of long ties in social contagions. In sum, the threshold distribution is the behavioral microfoundation of social contagion models and identifying empirical correlates of thresholds is crucial for understanding how network structure affects social contagions.