Defining and identifying local average treatment effects.

The article discusses effect estimators on average treatment effects (ATE) that depend on key assumptions for validity. Topics covered include the alternative of instrumental variables (IV) and their conditions of validity and targeted effects, the estimates of the local ATE (LATE), and the derivation of the IV estimator. Also noted are the uses of IV methods without confounding assumptions, and on quantifying effects other than the complier average causal effect (CACE).