New Tips from the RCS Stats Team

June 5, 2018

Imagine conducting a randomized controlled trial with noncompliance (participants assigned to the treatment group can choose to avoid treatment, and participants assigned to the control group are able to seek out treatment). If the experimenter observes the ultimate treatment status for all participants, then she can use random assignment as an instrumental variable to measure the causal effect of treatment. Miguel Hernán posted an interesting thread about the subtleties of this approach on Twitter.

twitter 1twitter 2twitter 3twitter 4

See also: Stats Tips