Congratulations to graduate student Naoki Egami, who won the 2019 Gosnell Prize for “Identification of Causal Diffusion Effects using Stationary Causal Directed Acyclic Graphs.”

According to the Gosnell Award Committee: "Egami’s paper brings fresh insights to the study of how ideas and behaviors diffuse across people, governments, or countries. Causal diffusion effects are difficult to study, however, due to common biases such as contextual confounding and homophily bias. Egami’s strongest contribution is to show that if diffusion is stationary over time, then these types of omitted variable biases can be detected by leveraging lagged dependent variables. The stationarity assumption requires only stability in the presence or absence of causal effects, which is weaker than an assumption about stability in the direction or magnitude of the effects. Egami productively leverages directed acyclic graphs to develop this placebo test, and, under stronger assumptions, develops a bias-corrected estimator that will remove the detected unmeasured confounding from the causal diffusion estimates. Given the interest in studying diffusion, these tools have the potential to become an integral part of the applied methodological toolkit in political science and beyond."