Derivation and Usefulness of Tight Symbolic Causal Bounds
Erin Evelyn Gabriel, PhD
Associate Professor in Biostatistics, University of Copenhagen
Where: Hybrid Event | 2001 McGill College, Room 1140; Zoom
A causal query (estimand) will commonly not be identifiable from observed data without the assumption of no unmeasured confounders, in which case no estimator of the query can be contrived without further assumptions or additional measured variables. However, it may still be possible to derive bounds on the query in terms of the distribution of observed variables. Bounds, numeric or symbolic, can often be more valuable than a statistical estimator derived under implausible assumptions. Symbolic bounds, however, provide a measure of uncertainty and information loss due to the lack of an identifiable estimand even in the absence of data. We develop and describe a general approach for the computation of symbolic bounds and characterize a class of settings in which our method is guaranteed to provide tight valid bounds. This expands the known settings in which tight causal bounds are solutions to linear programs to include multicategorical variables, cross-world nested counterfactuals queries, and missing data problems. We demonstrate the usefulness of symbolic bounds and the estimates from them both conceptually for the planning of studies and in several real data settings, including the Danish Mask Study.
Erin is an Associate Professor of Biostatistics at the Section of Biostatistics in the Department of Public Health at the University of Copenhagen. Her research focuses on methods for causal inference and surrogate evaluation, and design, testing, and estimation in emulated and randomized clinical trials. Although her primary area of application is infectious diseases, she has recently started working more in cancer and chronic illness. Erin is currently funded by the Novo Nordisk Foundation. Website: https://eegabriel.github.io/