Glossary Entry

Selection Bias

Bias from the treated and untreated groups differing even without treatment, formally E[Y(0) given treated] minus E[Y(0) given untreated]; the main reason naive comparisons mislead.

Statistics Data

Also called: survivorship bias

Seed source: Causal Inference - The Mixtape, ch. 4

Selection bias is the counterfactual gap between groups: the people who chose (or were chosen for) a treatment would have had different outcomes anyway. It needs no villain; optimal, well-informed decisions create it, because people act on information the analyst does not see.

Its most famous costume is survivorship bias, analyzing only the units that made it into the sample, such as the WWII bombers that returned. In the naive-gap decomposition it is the term randomization deletes and every observational method tries to remove by assumption.