|
Abstract:
|
The second half of the session focuses on designs and methods that relax the assumption that we have measured all confounding covariates. Examples include instrumental variables and regression discontinuity designs. We motivate each with a real-world application, discuss the plausibility of the assumptions, and describe relevant software. We conclude with a brief discussion of further avenues for exploration, including sensitivity analysis, modes of inference, and group-structured data.
|
