Bounding Knowledge Decay From Agnostic Temporal Generalization

Causal generalization is essential to contemporary political science practice. We argue that recent methodological advances in causal generalization pay insufficient attention to issues which arise from generalization over time. For assumptions of varying degrees of strictness, we derive novel statistical bounds of the growing uncertainty of a given causal estimate into the future. We derive these bounds using the Wasserstein divergence which allows us to weaken assumptions of positivity which are not typically met in practice. In an empirical example, we demonstrate that actual variation in treatment effects over time tends to dominate reported statistical uncertainty. Once implicit and untenable assumptions about covariate distribution and conditional treatment effects are made explicit and relaxed, descriptive and causal knowledge are both essential for temporal causal generalization.