3,884 followers
8,123 followers
@vk_wilde You can’t introduce collider bias by adjusting for a baseline (pre-randomisation ) variable, so we’re good there. If anyone reading is interested in covariate adjustment: https://t.co/G3owwTcx5g
8,123 followers
3,884 followers
RT @andersgranholm: 3) If a baseline covariate selected based on clinical knowledge turns out to not be predictive of the outcome, this has…