Inference in experiments conditional on observed imbalances in covariates

Per Johansson, Mattias Nordin

Double blind randomized controlled trials are traditionally seen as the gold standard for causal inferences as the difference-in-means estimator is an unbiased estimator of the ave…