A0890
Title: Identifying the effect of persuasion
Authors: Sokbae Lee - Columbia University (United States) [presenting]
Sung Jae Jun - Penn State University (United States)
Abstract: A commonly used measure of persuasion is examined, whose precise interpretation has been obscure in the literature. Using the potential outcome framework, the causal persuasion rate is defined by a proper conditional probability of taking the action of interest with a persuasive message conditional on not taking action without the notification. Identification is then formally studied under empirically relevant data scenarios. The commonly adopted measure generally that does not estimate is shown, but the causal rate of persuasion is often overstated. Several new parameters of interest are discussed, and practical methods for causal inference are provided.