A1216
Title: Bayesian inference for causal effects under interference with a partially observed diffusion process on networks
Authors: Fei Fang - Yale University (United States) [presenting]
Laura Forastiere - Yale University (United States)
Edoardo Airoldi - Fox School of Business, Temple University (United States)
Amir Ghasemianlangroodi - Yale University (United States)
Abstract: Behaviours are likely to spread in a connected population, and the presence of a behavioural intervention may boost this spread. The setting is considered where it is observed at baseline, the set of treated units, and at baseline and follow-up the social network and the prevalence of behaviours. To investigate the problem, a network-based diffusion model is assumed, including the network susceptible-infected-susceptible (SIS) model and network susceptible-infected (SI) model, formulated as a continuous-time Markov process. A Bayesian data augmentation procedure is developed to impute over time the behavioural change as a result of diffusion from social ties or as a result of the intervention for the treated. Based on the estimated parameters, an imputation method is used to evaluate the causal effects of hypothetical treatment allocations with different rates and network-based strategies. Under simplified network models, closed forms were also derived for the expected effect of increasing the treatment rate under different baseline behaviour prevalence and network structures. The proposed method is applied to a factorial randomized experiment delivering a behavioural intervention in villages in Honduras under different treatment rates and strategies. This data allows us to compare adoption rates under a hypothetical strategy imputed in one arm with the actual adoption rates observed in the arm assigned to that strategy.