A0975
Title: Accounting for outcome spillover for causal inference with continuous spatiotemporal processes
Authors: Conor Kresin - University of Otago (New Zealand)
Duncan Clark - Williams College (United States) [presenting]
Martin Hazelton - University of Otago (New Zealand)
Abstract: The purpose is to demonstrate that, in a highly general parametric setting, causal inference with observational spatiotemporal data in the presence of arbitrary outcome spillover is feasible. A general framework is constructed for defining outcomes and associated causal estimands in continuous space-time, and it is demonstrated that estimation is achievable with a likelihood-based approach via stochastic expectation maximization. Leveraging results from point process theory, conditions necessary for estimation and subsequent inference are demonstrated. The proposed framework accommodates observational and experimental data, random and non-random treatment mechanisms, a general class of model specifications including those that allow for interaction between points, and general observation windows. The focus is pertinent to applications as diverse as epidemiology and finance, enabling previously impossible causal inference on rich continuous spatiotemporal data.
